From pgsql-performance-owner@postgresql.org Fri Jul 1 05:57:14 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4AB2F529E7 for ; Fri, 1 Jul 2005 05:57:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42700-01 for ; Fri, 1 Jul 2005 08:57:06 +0000 (GMT) Received: from zigo.dhs.org (ua-83-227-204-174.cust.bredbandsbolaget.se [83.227.204.174]) by svr1.postgresql.org (Postfix) with ESMTP id 8C9FD529DA for ; Fri, 1 Jul 2005 05:57:03 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id A923C8467; Fri, 1 Jul 2005 10:57:02 +0200 (CEST) Date: Fri, 1 Jul 2005 10:57:02 +0200 (CEST) From: Dennis Bjorklund To: John Mendenhall Cc: pgsql-performance@postgresql.org Subject: Re: ported application having performance issues In-Reply-To: <20050630235821.GA2163@calvin.surfutopia.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.331 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/1 X-Sequence-Number: 13242 On Thu, 30 Jun 2005, John Mendenhall wrote: > Our setting for effective_cache_size is 2048. > > random_page_cost = 4, effective_cache_size = 2048 time approximately 4500ms > random_page_cost = 3, effective_cache_size = 2048 time approximately 1050ms > random_page_cost = 3, effective_cache_size = 4096 time approximately 1025ms > > The decrease of random_page_cost to 3 caused the plan > to work properly, using the lead_requests table as a > join starting point and using the contacts index. The effective_cache_size still looks small. As a rule of tumb you might want effective_cache_size to be something like 1/2 or 2/3 of your total memory. I don't know how much you had, but effective_cache_size = 4096 is only 32M. shared_buffers and effective_cache_size is normally the two most important settings in my experience. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Fri Jul 1 10:33:23 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7BC9D52A42 for ; Fri, 1 Jul 2005 10:33:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 03201-06 for ; Fri, 1 Jul 2005 13:33:10 +0000 (GMT) Received: from outmail.freedom2surf.net (outmail1.freedom2surf.net [194.106.33.237]) by svr1.postgresql.org (Postfix) with ESMTP id EE76B529A6 for ; Fri, 1 Jul 2005 10:33:06 -0300 (ADT) Received: from sam.lan.samason.me.uk (sam.samason.me.uk [83.67.41.231]) by outmail.freedom2surf.net (8.12.10/8.12.10) with ESMTP id j61DX9ch017854 for ; Fri, 1 Jul 2005 14:33:09 +0100 Received: by sam.lan.samason.me.uk (Postfix, from userid 1000) id 4827F63745; Fri, 1 Jul 2005 14:33:05 +0100 (BST) Date: Fri, 1 Jul 2005 14:33:05 +0100 From: Sam Mason To: pgsql-performance@postgresql.org Subject: planner picking more expensive plan Message-ID: <20050701133305.GB2623@sam.lan.samason.me.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.8i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/2 X-Sequence-Number: 13243 Hi, I've just been referred here after a conversion on IRC and everybody seemed to think I've stumbled upon some strangeness. The planner (in PG version 8.0.2) is choosing what it thinks is a more expensive plan. I've got a table of animals (about 3M rows) and their movements (about 16M rows), and I'm trying to execute this query: SELECT a.birthlocnid, m.locnid FROM animals a LEFT JOIN movements m ON (a.animalid = m.animalid AND m.mtypeid=0) LIMIT 10; If I have "work_mem" set to something small (1000) it uses this plan: QUERY PLAN Limit (cost=0.00..202.52 rows=10 width=8) (actual time=0.221..0.600 rows=10 loops=1) -> Merge Left Join (cost=0.00..66888828.30 rows=3302780 width=8) (actual time=0.211..0.576 rows=10 loops=1) Merge Cond: ("outer".animalid = "inner".animalid) -> Index Scan using animals_pkey on animals a (cost=0.00..10198983.91 rows=3302780 width=8) (actual time=0.112..0.276 rows=10 loops=1) -> Index Scan using movement_animal on movements m (cost=0.00..56642740.73 rows=3107737 width=8) (actual time=0.088..0.235 rows=10 loops=1) Filter: (mtypeid = 0) Total runtime: 0.413 ms But if I increase "work_mem" to 10000 it uses this plan: QUERY PLAN Limit (cost=565969.42..566141.09 rows=10 width=8) (actual time=27769.047..27769.246 rows=10 loops=1) -> Merge Right Join (cost=565969.42..57264070.77 rows=3302780 width=8) (actual time=27769.043..27769.228 rows=10 loops=1) Merge Cond: ("outer".animalid = "inner".animalid) -> Index Scan using movement_animal on movements m (cost=0.00..56642740.73 rows=3107737 width=8) (actual time=0.022..0.154 rows=10 loops=1) Filter: (mtypeid = 0) -> Sort (cost=565969.42..574226.37 rows=3302780 width=8) (actual time=27768.991..27769.001 rows=10 loops=1) Sort Key: a.animalid -> Seq Scan on animals a (cost=0.00..77086.80 rows=3302780 width=8) (actual time=0.039..5620.651 rows=3303418 loops=1) Total runtime: 27851.097 ms I've tried playing with the statistics as people suggested on IRC but to no effect. There was some discussion about why it would be doing this, but nothing obvious came out of it. SHOW ALL output is at the end of this mail but it should be pretty standard apart from: shared_buffers = 10000 work_mem = 8192 max_connections = 100 effective_cache_size = 10000 Hope that's enough information to be useful. Thanks. Sam name | setting --------------------------------+-------------------------------- add_missing_from | on archive_command | /home/postgres/pgarchive "%p" australian_timezones | off authentication_timeout | 60 bgwriter_delay | 200 bgwriter_maxpages | 100 bgwriter_percent | 1 block_size | 8192 check_function_bodies | on checkpoint_segments | 3 checkpoint_timeout | 300 checkpoint_warning | 30 client_encoding | SQL_ASCII client_min_messages | notice commit_delay | 0 commit_siblings | 5 config_file | /home/pgdata/postgresql.conf cpu_index_tuple_cost | 0.001 cpu_operator_cost | 0.0025 cpu_tuple_cost | 0.01 custom_variable_classes | unset data_directory | /home/pgdata DateStyle | ISO, MDY db_user_namespace | off deadlock_timeout | 1000 debug_pretty_print | off debug_print_parse | off debug_print_plan | off debug_print_rewritten | off debug_shared_buffers | 0 default_statistics_target | 10 default_tablespace | unset default_transaction_isolation | read committed default_transaction_read_only | off default_with_oids | on dynamic_library_path | $libdir effective_cache_size | 10000 enable_hashagg | on enable_hashjoin | on enable_indexscan | on enable_mergejoin | on enable_nestloop | on enable_seqscan | off enable_sort | on enable_tidscan | on explain_pretty_print | on external_pid_file | unset extra_float_digits | 0 from_collapse_limit | 8 fsync | on geqo | on geqo_effort | 5 geqo_generations | 0 geqo_pool_size | 0 geqo_selection_bias | 2 geqo_threshold | 12 hba_file | /home/pgdata/pg_hba.conf ident_file | /home/pgdata/pg_ident.conf integer_datetimes | off join_collapse_limit | 8 krb_server_keyfile | unset lc_collate | C lc_ctype | C lc_messages | C lc_monetary | C lc_numeric | C lc_time | C listen_addresses | * log_connections | on log_destination | stderr log_directory | pg_log log_disconnections | off log_duration | off log_error_verbosity | default log_executor_stats | off log_filename | postgresql-%Y-%m-%d_%H%M%S.log log_hostname | off log_line_prefix | %t %u log_min_duration_statement | -1 log_min_error_statement | panic log_min_messages | notice log_parser_stats | off log_planner_stats | off log_rotation_age | 1440 log_rotation_size | 10240 log_statement | all log_statement_stats | off log_truncate_on_rotation | off maintenance_work_mem | 256000 max_connections | 100 max_files_per_process | 1000 max_fsm_pages | 20000 max_fsm_relations | 1000 max_function_args | 32 max_identifier_length | 63 max_index_keys | 32 max_locks_per_transaction | 64 max_stack_depth | 2048 password_encryption | on port | 5432 pre_auth_delay | 0 preload_libraries | unset random_page_cost | 4 redirect_stderr | off regex_flavor | advanced rendezvous_name | unset search_path | $user,public server_encoding | SQL_ASCII server_version | 8.0.2 shared_buffers | 1000 silent_mode | off sql_inheritance | on ssl | off statement_timeout | 0 stats_block_level | off stats_command_string | off stats_reset_on_server_start | on stats_row_level | off stats_start_collector | on superuser_reserved_connections | 2 syslog_facility | LOCAL0 syslog_ident | postgres TimeZone | GMT trace_notify | off transaction_isolation | read committed transaction_read_only | off transform_null_equals | off unix_socket_directory | unset unix_socket_group | unset unix_socket_permissions | 511 vacuum_cost_delay | 0 vacuum_cost_limit | 200 vacuum_cost_page_dirty | 20 vacuum_cost_page_hit | 1 vacuum_cost_page_miss | 10 wal_buffers | 8 wal_sync_method | fdatasync work_mem | 128000 zero_damaged_pages | off From pgsql-performance-owner@postgresql.org Fri Jul 1 11:23:00 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A5BCD5287E for ; Fri, 1 Jul 2005 11:22:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 13951-02 for ; Fri, 1 Jul 2005 14:22:48 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id CC1A0529A6 for ; Fri, 1 Jul 2005 11:22:45 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j61EMoua028889; Fri, 1 Jul 2005 10:22:50 -0400 (EDT) To: Sam Mason Cc: pgsql-performance@postgresql.org Subject: Re: planner picking more expensive plan In-reply-to: <20050701133305.GB2623@sam.lan.samason.me.uk> References: <20050701133305.GB2623@sam.lan.samason.me.uk> Comments: In-reply-to Sam Mason message dated "Fri, 01 Jul 2005 14:33:05 +0100" Date: Fri, 01 Jul 2005 10:22:50 -0400 Message-ID: <28888.1120227770@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/3 X-Sequence-Number: 13244 Sam Mason writes: > The planner (in PG version 8.0.2) is choosing what it thinks is a more > expensive plan. I fooled around trying to duplicate this behavior, without success. Can you create a self-contained test case? regards, tom lane From pgsql-performance-owner@postgresql.org Fri Jul 1 11:58:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0EE5552952 for ; Fri, 1 Jul 2005 11:58:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 23235-03 for ; Fri, 1 Jul 2005 14:58:45 +0000 (GMT) Received: from colo.samason.me.uk (unknown [69.55.228.22]) by svr1.postgresql.org (Postfix) with ESMTP id 172855292D for ; Fri, 1 Jul 2005 11:58:44 -0300 (ADT) Received: from colo.samason.me.uk (colo [69.55.228.22]) by colo.samason.me.uk (8.12.11/8.12.11) with ESMTP id j61Ewn2L051935 for ; Fri, 1 Jul 2005 15:58:49 +0100 (BST) (envelope-from sam@colo.samason.me.uk) Received: (from sam@localhost) by colo.samason.me.uk (8.12.11/8.12.11/Submit) id j61EwnaS051895 for pgsql-performance@postgresql.org; Fri, 1 Jul 2005 15:58:49 +0100 (BST) (envelope-from sam) Date: Fri, 1 Jul 2005 15:58:48 +0100 From: Sam Mason To: pgsql-performance@postgresql.org Subject: Re: planner picking more expensive plan Message-ID: <20050701145848.GW62747@colo.samason.me.uk> Mail-Followup-To: Sam Mason , pgsql-performance@postgresql.org References: <20050701133305.GB2623@sam.lan.samason.me.uk> <28888.1120227770@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <28888.1120227770@sss.pgh.pa.us> User-Agent: Mutt/1.4.2.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/4 X-Sequence-Number: 13245 Tom Lane wrote: >I fooled around trying to duplicate this behavior, without success. >Can you create a self-contained test case? I'll try and see if I can put something together, it's probably going to be early next week though. I wont be able to give you our data, so I'll be a bit of a headscratching exercise generating something that'll provoke the same behaviour. Not sure if it'll help, but here's what the database schema looks like at the moment: Table "public.animals" Column | Type | Modifiers -------------+-----------------------+----------- animalid | integer | not null sex | character(1) | not null dob | date | not null birthlocnid | integer | breedid | character varying(8) | eartag_1 | character varying(20) | eartag_2 | character varying(20) | eartag_3 | character varying(20) | Indexes: "animals_pkey" primary key, btree (animalid) "animal_birthlocn" btree (birthlocnid) "animal_breed" btree (breedid) "animal_eartag" btree (eartag_1) Check constraints: "animal_sex" CHECK (sex = 'M'::bpchar OR sex = 'F'::bpchar) Table "public.movements" Column | Type | Modifiers ----------+---------+----------- locnid | integer | not null animalid | integer | not null movedate | date | not null mtypeid | integer | not null Indexes: "movement_animal" btree (animalid) "movement_location" btree (locnid) "movement_movedate" btree (movedate) "movement_movetype" btree (mtypeid) Foreign-key constraints: "movement_location" FOREIGN KEY (locnid) REFERENCES locations(locnid) "movement_animal" FOREIGN KEY (animalid) REFERENCES animals(animalid) "movement_type" FOREIGN KEY (mtypeid) REFERENCES k_movement_type(mtypeid) Table "public.locations" Column | Type | Modifiers --------+-----------------------+----------- locnid | integer | not null ptype | character varying(8) | ltype | character varying(8) | not null cph | character varying(20) | unk | integer | Indexes: "locations_pkey" primary key, btree (locnid) "location_cph" btree (cph) "location_ltype" btree (ltype) "location_ptype" btree (ptype) Foreign-key constraints: "location_ptype" FOREIGN KEY (ptype) REFERENCES k_premise_type(ptypeid) "location_ltype" FOREIGN KEY (ltype) REFERENCES k_location_type(ltypeid) As I said, animals contains about 3M rows, movements about 16M rows and locations about 80K rows. There are about 3 to 8 rows for each and every animal in the movements table, with at most one entry of mtypeid=0 for each animal (95% of the animals have an entry). Not sure if that's going to help making some demo data. It's just that it took quite a while loading it all here, so coming up with some code to make demo data may take a while. Thanks! Sam From pgsql-performance-owner@postgresql.org Fri Jul 1 12:18:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6931F5292D for ; Fri, 1 Jul 2005 12:18:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 27596-01 for ; Fri, 1 Jul 2005 15:18:17 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id 3AB295287E for ; Fri, 1 Jul 2005 12:18:15 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050701151814m92008hu1ne>; Fri, 1 Jul 2005 15:18:14 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id ED170560B3; Fri, 1 Jul 2005 10:18:12 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 9BD4655FF5; Fri, 1 Jul 2005 10:17:54 -0500 (CDT) Message-ID: <42C55EA0.1000404@arbash-meinel.com> Date: Fri, 01 Jul 2005 10:17:52 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Sam Mason Cc: pgsql-performance@postgresql.org Subject: Re: planner picking more expensive plan References: <20050701133305.GB2623@sam.lan.samason.me.uk> In-Reply-To: <20050701133305.GB2623@sam.lan.samason.me.uk> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig57E9EE20779CE456D1DF6438" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.049 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/5 X-Sequence-Number: 13246 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig57E9EE20779CE456D1DF6438 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sam Mason wrote: >Hi, > >I've just been referred here after a conversion on IRC and everybody >seemed to think I've stumbled upon some strangeness. > >The planner (in PG version 8.0.2) is choosing what it thinks is a more >expensive plan. I've got a table of animals (about 3M rows) and their >movements (about 16M rows), and I'm trying to execute this query: > > SELECT a.birthlocnid, m.locnid > FROM animals a > LEFT JOIN movements m ON (a.animalid = m.animalid AND m.mtypeid=0) > LIMIT 10; > > > Why are you using LIMIT without having an ORDER BY? What are actually trying to get out of this query? Is it just trying to determine where the 'home' locations are? It just seems like this query isn't very useful. As it doesn't restrict by animal id, and it just gets 10 randomly selected animals where m.mtypeid=0. And why a LEFT JOIN instead of a normal join? Anyway, the general constraints you are applying seem kind of confusing. What happens if you change the plan to: SELECT a.birthlocnid, m.locnid FROM animals a LEFT JOIN movements m ON (a.animalid = m.animalid AND m.mtypeid=0) ORDER BY a.animalid LIMIT 10; I would guess that this would help the planner realize it should try to use an index, since it can realize that it wants only a few rows by a.animalid in order. Though I also recognize that you aren't returning a.animalid so you don't really know which animals you are returning. I get the feeling you are trying to ask something like "do animals stay at their birth location", or at least "how are animals moving around". I don't know what m.typeid = 0 means, but I'm guessing it is something like where their home is. Anyway, I would say you need to put a little bit more restriction in, so the planner can figure out how to get only 10 rows. John =:-> >If I have "work_mem" set to something small (1000) it uses this plan: > > QUERY PLAN > > Limit (cost=0.00..202.52 rows=10 width=8) (actual time=0.221..0.600 rows=10 loops=1) > -> Merge Left Join (cost=0.00..66888828.30 rows=3302780 width=8) (actual time=0.211..0.576 rows=10 loops=1) > Merge Cond: ("outer".animalid = "inner".animalid) > -> Index Scan using animals_pkey on animals a (cost=0.00..10198983.91 rows=3302780 width=8) (actual time=0.112..0.276 rows=10 loops=1) > -> Index Scan using movement_animal on movements m (cost=0.00..56642740.73 rows=3107737 width=8) (actual time=0.088..0.235 rows=10 loops=1) > Filter: (mtypeid = 0) > Total runtime: 0.413 ms > >But if I increase "work_mem" to 10000 it uses this plan: > > QUERY PLAN > > Limit (cost=565969.42..566141.09 rows=10 width=8) (actual time=27769.047..27769.246 rows=10 loops=1) > -> Merge Right Join (cost=565969.42..57264070.77 rows=3302780 width=8) (actual time=27769.043..27769.228 rows=10 loops=1) > Merge Cond: ("outer".animalid = "inner".animalid) > -> Index Scan using movement_animal on movements m (cost=0.00..56642740.73 rows=3107737 width=8) (actual time=0.022..0.154 rows=10 loops=1) > Filter: (mtypeid = 0) > -> Sort (cost=565969.42..574226.37 rows=3302780 width=8) (actual time=27768.991..27769.001 rows=10 loops=1) > Sort Key: a.animalid > -> Seq Scan on animals a (cost=0.00..77086.80 rows=3302780 width=8) (actual time=0.039..5620.651 rows=3303418 loops=1) > Total runtime: 27851.097 ms > > >I've tried playing with the statistics as people suggested on IRC but to >no effect. There was some discussion about why it would be doing this, >but nothing obvious came out of it. > >SHOW ALL output is at the end of this mail but it should be pretty >standard apart from: > > shared_buffers = 10000 > work_mem = 8192 > max_connections = 100 > effective_cache_size = 10000 > >Hope that's enough information to be useful. > >Thanks. > > Sam > --------------enig57E9EE20779CE456D1DF6438 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCxV6iJdeBCYSNAAMRAme8AJwKdr9g+MxhH4aCrz95aKrrBr66fgCfboU2 0T3tjYDEPciwHkTLwJBmA8Q= =lbpU -----END PGP SIGNATURE----- --------------enig57E9EE20779CE456D1DF6438-- From pgsql-performance-owner@postgresql.org Fri Jul 1 12:58:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 802475297E for ; Fri, 1 Jul 2005 12:58:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 35693-03 for ; Fri, 1 Jul 2005 15:58:35 +0000 (GMT) Received: from colo.samason.me.uk (unknown [69.55.228.22]) by svr1.postgresql.org (Postfix) with ESMTP id 87C15529F3 for ; Fri, 1 Jul 2005 12:58:30 -0300 (ADT) Received: from colo.samason.me.uk (colo [69.55.228.22]) by colo.samason.me.uk (8.12.11/8.12.11) with ESMTP id j61FwUj5099838 for ; Fri, 1 Jul 2005 16:58:30 +0100 (BST) (envelope-from sam@colo.samason.me.uk) Received: (from sam@localhost) by colo.samason.me.uk (8.12.11/8.12.11/Submit) id j61FwUik099832 for pgsql-performance@postgresql.org; Fri, 1 Jul 2005 16:58:30 +0100 (BST) (envelope-from sam) Date: Fri, 1 Jul 2005 16:58:30 +0100 From: Sam Mason To: pgsql-performance@postgresql.org Subject: Re: planner picking more expensive plan Message-ID: <20050701155829.GX62747@colo.samason.me.uk> Mail-Followup-To: Sam Mason , pgsql-performance@postgresql.org References: <20050701133305.GB2623@sam.lan.samason.me.uk> <42C55EA0.1000404@arbash-meinel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42C55EA0.1000404@arbash-meinel.com> User-Agent: Mutt/1.4.2.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/6 X-Sequence-Number: 13247 John A Meinel wrote: >Why are you using LIMIT without having an ORDER BY? I'm just exploring the data, trying to figure out what it's like. >It just seems like this query isn't very useful. As it doesn't restrict >by animal id, and it just gets 10 randomly selected animals where >m.mtypeid=0. Yup, that's the point. Check to see if the animals were born where they say they were. The data's come from an external source and I'm just trying to figure out how good it is before I do too much with it >And why a LEFT JOIN instead of a normal join? I'm not sure if some animals will have missing data! >Anyway, the general constraints you are applying seem kind of confusing. This was a slightly cut down query in an attempt to reduce general confusion -- I guess I failed. Sorry! >I would guess that this would help the planner realize it should try to >use an index, since it can realize that it wants only a few rows by >a.animalid in order. This seems to work the appropiate magic. It always seems to prefer index scans now. The real point of asking this question orignally was to find out why the planner was choosing a more expensive plan over a cheaper one. When I discovered this orignally I was disabling seqscan and then it picked the correct version. The actual work_mem didn't change when I did this, it just picked the correct plan. I discovered the work_mem parameter fiddle later. I think I forgot to mention that in the original email though! Sam From pgsql-performance-owner@postgresql.org Fri Jul 1 21:52:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 91E8B52801 for ; Fri, 1 Jul 2005 21:52:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44621-05 for ; Sat, 2 Jul 2005 00:52:37 +0000 (GMT) Received: from calvin.surfutopia.net (calvin.surfutopia.net [67.120.245.34]) by svr1.postgresql.org (Postfix) with ESMTP id 3143452812 for ; Fri, 1 Jul 2005 21:52:36 -0300 (ADT) Received: by calvin.surfutopia.net (Postfix, from userid 1000) id 9D8DCF24B; Fri, 1 Jul 2005 17:52:37 -0700 (PDT) Date: Fri, 1 Jul 2005 17:52:37 -0700 From: John Mendenhall To: Dennis Bjorklund Cc: pgsql-performance@postgresql.org Subject: Re: ported application having performance issues Message-ID: <20050702005237.GA26087@calvin.surfutopia.net> References: <20050630235821.GA2163@calvin.surfutopia.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/7 X-Sequence-Number: 13248 Dennis, On Fri, 01 Jul 2005, Dennis Bjorklund wrote: > On Thu, 30 Jun 2005, John Mendenhall wrote: > > > Our setting for effective_cache_size is 2048. > > > > random_page_cost = 4, effective_cache_size = 2048 time approximately 4500ms > > random_page_cost = 3, effective_cache_size = 2048 time approximately 1050ms > > random_page_cost = 3, effective_cache_size = 4096 time approximately 1025ms > > The effective_cache_size still looks small. As a rule of tumb you might > want effective_cache_size to be something like 1/2 or 2/3 of your total > memory. I don't know how much you had, but effective_cache_size = 4096 is > only 32M. > > shared_buffers and effective_cache_size is normally the two most important > settings in my experience. I have increased the effective_cache_size to 16384 (128M). I have kept random_page_cost at 3 for now. This appears to give me the performance I need at this time. In the future, we'll look at other methods of increasing the performance. Thank you all for all your suggestions. JohnM -- John Mendenhall john@surfutopia.net surf utopia internet services From pgsql-performance-owner@postgresql.org Fri Jul 1 22:59:45 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9018352968 for ; Fri, 1 Jul 2005 22:59:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 53668-08 for ; Sat, 2 Jul 2005 01:59:42 +0000 (GMT) Received: from mail.baymountain.com (mail.baymountain.com [8.7.96.19]) by svr1.postgresql.org (Postfix) with SMTP id 8274C528C1 for ; Fri, 1 Jul 2005 22:59:40 -0300 (ADT) Received: (qmail 2314 invoked by uid 504); 2 Jul 2005 01:59:43 -0000 Received: from emil@baymountain.com by mail.baymountain.com by uid 501 with qmail-scanner-1.16 (spamassassin: 2.53. Clear:SA:0(0.0/7.0):. Processed in 0.246783 secs); 02 Jul 2005 01:59:43 -0000 Received: from unknown (HELO briggspack.com) (emil@briggspack.com@24.211.148.77) by mail.baymountain.com with SMTP; 2 Jul 2005 01:59:43 -0000 From: Emil Briggs Reply-To: emil@baymountain.com Organization: Baymountain, Inc. To: pgsql-performance@postgresql.org Subject: Planner constants for RAM resident databases Date: Fri, 1 Jul 2005 21:59:38 -0400 User-Agent: KMail/1.7.1 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507012159.39041.emil@baymountain.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.038 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/8 X-Sequence-Number: 13249 I'm working with an application where the database is entirely resident in RAM (the server is a quad opteron with 16GBytes of memory). It's a web application and handles a high volume of queries. The planner seems to be generating poor plans for some of our queries which I can fix by raising cpu_tuple_cost. I have seen some other comments in the archives saying that this is a bad idea but is that necessarily the case when the database is entirely resident in RAM? Emil From pgsql-performance-owner@postgresql.org Fri Jul 1 23:08:38 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9B8D152800 for ; Fri, 1 Jul 2005 23:08:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 59482-03 for ; Sat, 2 Jul 2005 02:08:28 +0000 (GMT) Received: from gp.word-to-the-wise.com (gp.word-to-the-wise.com [64.71.176.18]) by svr1.postgresql.org (Postfix) with ESMTP id 595D952801 for ; Fri, 1 Jul 2005 23:08:25 -0300 (ADT) Received: by gp.word-to-the-wise.com (Postfix, from userid 500) id 34B778FC08E; Fri, 1 Jul 2005 19:08:23 -0700 (PDT) Date: Fri, 1 Jul 2005 19:08:23 -0700 From: Steve Atkins To: pgsql-performance@postgresql.org Subject: Re: Planner constants for RAM resident databases Message-ID: <20050702020823.GA32264@gp.word-to-the-wise.com> References: <200507012159.39041.emil@baymountain.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200507012159.39041.emil@baymountain.com> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.84 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_POST X-Spam-Level: X-Archive-Number: 200507/9 X-Sequence-Number: 13250 On Fri, Jul 01, 2005 at 09:59:38PM -0400, Emil Briggs wrote: > I'm working with an application where the database is entirely resident in RAM > (the server is a quad opteron with 16GBytes of memory). It's a web > application and handles a high volume of queries. The planner seems to be > generating poor plans for some of our queries which I can fix by raising > cpu_tuple_cost. I have seen some other comments in the archives saying that > this is a bad idea but is that necessarily the case when the database is > entirely resident in RAM? If I'm understanding correctly that'll mostly increase the estimated cost of handling a row relative to a sequential page fetch, which sure sounds like it'll push plans in the right direction, but it doesn't sound like the right knob to twiddle. What do you have random_page_cost set to? Cheers, Steve From pgsql-performance-owner@postgresql.org Fri Jul 1 23:41:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6D47252812 for ; Fri, 1 Jul 2005 23:40:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65548-03 for ; Sat, 2 Jul 2005 02:40:50 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 59F3D52801 for ; Fri, 1 Jul 2005 23:40:49 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050702024053m9100ngas9e>; Sat, 2 Jul 2005 02:40:53 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id F0097560B3; Fri, 1 Jul 2005 21:40:52 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 9A77C56070; Fri, 1 Jul 2005 21:40:49 -0500 (CDT) Message-ID: <42C5FEB1.6090601@arbash-meinel.com> Date: Fri, 01 Jul 2005 21:40:49 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: emil@baymountain.com Cc: pgsql-performance@postgresql.org Subject: Re: Planner constants for RAM resident databases References: <200507012159.39041.emil@baymountain.com> In-Reply-To: <200507012159.39041.emil@baymountain.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig870DC8A41B021865FE229B70" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.048 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/10 X-Sequence-Number: 13251 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig870DC8A41B021865FE229B70 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Emil Briggs wrote: >I'm working with an application where the database is entirely resident in RAM >(the server is a quad opteron with 16GBytes of memory). It's a web >application and handles a high volume of queries. The planner seems to be >generating poor plans for some of our queries which I can fix by raising >cpu_tuple_cost. I have seen some other comments in the archives saying that >this is a bad idea but is that necessarily the case when the database is >entirely resident in RAM? > >Emil > > > Generally, the key knob to twiddle when everything fits in RAM is random_page_cost. If you truly have everything in RAM you could set it almost to 1. 1 means that it costs exactly the same to go randomly through the data then it does to go sequential. I would guess that even in RAM it is faster to go sequential (since you still have to page and deal with L1/L2/L3 cache, etc). But the default random_page_cost of 4 is probably too high for you. John =:-> --------------enig870DC8A41B021865FE229B70 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCxf6xJdeBCYSNAAMRArOgAKCo02YFzzjRTAOqW113BryLR2l59QCdEb10 WwGMl/V15c6lANMuXPQkN1U= =nHp/ -----END PGP SIGNATURE----- --------------enig870DC8A41B021865FE229B70-- From pgsql-performance-owner@postgresql.org Sat Jul 2 00:15:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A5E3752896 for ; Sat, 2 Jul 2005 00:15:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76535-06 for ; Sat, 2 Jul 2005 03:15:01 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id E2C8352801 for ; Sat, 2 Jul 2005 00:14:58 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050702031454m92008hqove>; Sat, 2 Jul 2005 03:14:54 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id CBD85560B3; Fri, 1 Jul 2005 22:14:53 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 70BFA56070; Fri, 1 Jul 2005 22:14:50 -0500 (CDT) Message-ID: <42C606AA.8060406@arbash-meinel.com> Date: Fri, 01 Jul 2005 22:14:50 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: emil@baymountain.com, Postgresql Performance Subject: Re: Planner constants for RAM resident databases References: <200507012159.39041.emil@baymountain.com> <42C5FFB9.9050302@arbash-meinel.com> <200507012311.37942.emil@baymountain.com> In-Reply-To: <200507012311.37942.emil@baymountain.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig6532213C64F4D9578C5667D0" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.048 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/11 X-Sequence-Number: 13252 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig6532213C64F4D9578C5667D0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Emil Briggs wrote: >>I just mentioned random_page_cost, but you should also tune >>effective_cache_size, since that is effectively most of your RAM. It >>depends what else is going on in the system, but setting it as high as >>say 12-14GB is probably reasonable if it is a dedicated machine. With >>random_page_cost 1.5-2, and higher effective_cache_size, you should be >>doing pretty well. >>John >>=:-> >> >> > >I tried playing around with these and they had no effect. It seems the only >thing that makes a difference is cpu_tuple_cost. > > > I'm surprised. I know cpu_tuple_cost can effect it as well, but usually the recommended way to get indexed scans is the above two parameters. When you do "explain analyze" of a query that you have difficulties with, how are the planner's estimates. Are the estimated number of rows about equal to the actual number of rows? If the planner is mis-estimating, there is a whole different set of tuning to do to help it estimate correctly. John =:-> PS> Use reply-all so that your comments go to the list. --------------enig6532213C64F4D9578C5667D0 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCxgaqJdeBCYSNAAMRAv9IAJ9KadS2PKPBL2jSFouJJcaG86+A8QCdGKdz Ivx8YqiiBA+DUJ14kL+P3Ww= =a9/R -----END PGP SIGNATURE----- --------------enig6532213C64F4D9578C5667D0-- From pgsql-performance-owner@postgresql.org Sat Jul 2 09:09:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5B635529F5 for ; Sat, 2 Jul 2005 09:09:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94221-07 for ; Sat, 2 Jul 2005 12:09:03 +0000 (GMT) Received: from tomts36-srv.bellnexxia.net (tomts36-srv.bellnexxia.net [209.226.175.93]) by svr1.postgresql.org (Postfix) with ESMTP id 5365F528B8 for ; Sat, 2 Jul 2005 09:09:02 -0300 (ADT) Received: from [192.168.1.99] ([64.229.154.120]) by tomts36-srv.bellnexxia.net (InterMail vM.5.01.06.10 201-253-122-130-110-20040306) with ESMTP id <20050702120903.HXMU16985.tomts36-srv.bellnexxia.net@[192.168.1.99]> for ; Sat, 2 Jul 2005 08:09:03 -0400 Message-ID: <42C61537.3080601@alteeve.com> Date: Sat, 02 Jul 2005 00:16:55 -0400 From: Madison Kelly User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Debian/1.7.8-1 X-Accept-Language: en-us, en, ja MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: B-Tree index not being used Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.105 tagged_above=0 required=5 tests=AWL, DATE_IN_PAST_06_12 X-Spam-Level: X-Archive-Number: 200507/13 X-Sequence-Number: 13254 Hi all, I have gone back to my index problem from a while ago where I am trying to do an update with a regex on the WHERE column. If I specifiy a constant the index is used so that much I know is working. I've been reading the 7.4 docs and I saw that a B-Tree index *should* but used when the regex is anchored to the start. This is from 11.2 of the docs; It says "The optimizer can also use a B-tree indexfor queries involving pattern matching operators LIKE, ILIKE, ~, and ~*, if, the pattern is anchored to the beginning of the string." In my case that is what I will always do. Specifically, this is a backup program I am using the DB for. The table I am working on stores all the file and directory information for a given partition. When the user toggles the checkbox for a given directory (to indicate that they do or do not what that directory backed up) I make a call to the DB telling it to change that column to given state. When the user toggle a directory I want to propgate that change to all sub directories and all files within those directories. The way I do this is: UPDATE file_info_11 SET file_backup='t' WHERE file_parent_dir~'^/foo/bar'; Which basically is just to say "change every directory and file with this parent directory and all sub directories to the new backup state". From what I gather this query should have used the index. Here is what I am actually getting though: tle-bu=> EXPLAIN ANALYZE UPDATE file_info_11 SET file_backup='t' WHERE file_parent_dir~'^/'; QUERY PLAN ---------------------------------------------------------------------------------------------------------------------- Seq Scan on file_info_11 (cost=0.00..13484.23 rows=1 width=183) (actual time=13.560..22040.603 rows=336039 loops=1) Filter: (file_parent_dir ~ '^/'::text) Total runtime: 514099.565 ms (3 rows) Now if I define a static directory the index IS used: tle-bu=> EXPLAIN ANALYZE UPDATE file_info_11 SET file_backup='t' WHERE file_parent_dir='/'; QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------------------- Index Scan using file_info_11_update_idx on file_info_11 (cost=0.00..109.69 rows=66 width=183) (actual time=22.828..62.020 rows=3 loops=1) Index Cond: (file_parent_dir = '/'::text) Total runtime: 88.334 ms (3 rows) Here is the table and index schemas: tle-bu=> \d file_info_11; \d file_info_11_update_idx; Table "public.file_info_11" Column | Type | Modifiers ----------------------+----------------------+----------------------------------------- file_group_name | text | file_group_uid | bigint | not null file_mod_time | bigint | not null file_name | text | not null file_parent_dir | text | not null file_perm | text | not null file_size | bigint | not null file_type | character varying(2) | not null default 'f'::character varying file_user_name | text | file_user_uid | bigint | not null file_backup | boolean | not null default true file_display | boolean | not null default false file_restore_display | boolean | not null default false file_restore | boolean | not null default false Indexes: "file_info_11_display_idx" btree (file_type, file_parent_dir, file_name) "file_info_11_update_idx" btree (file_parent_dir) Index "public.file_info_11_update_idx" Column | Type -----------------+------ file_parent_dir | text btree, for table "public.file_info_11" Can anyone see why the index might not be being used? I know that 'tsearch2' would probably work but it seems like way more than I need (because I will never be searching the middle of a string). Thanks for any advice/help/pointers! Madison -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Madison Kelly (Digimer) TLE-BU, The Linux Experience; Back Up http://tle-bu.thelinuxexperience.com -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- From pgsql-performance-owner@postgresql.org Sat Jul 2 02:24:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 19C695288E for ; Sat, 2 Jul 2005 02:24:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 03144-08 for ; Sat, 2 Jul 2005 05:24:33 +0000 (GMT) Received: from zigo.dhs.org (ua-83-227-204-174.cust.bredbandsbolaget.se [83.227.204.174]) by svr1.postgresql.org (Postfix) with ESMTP id 66CF952867 for ; Sat, 2 Jul 2005 02:24:30 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 7B6998467; Sat, 2 Jul 2005 07:24:26 +0200 (CEST) Date: Sat, 2 Jul 2005 07:24:26 +0200 (CEST) From: Dennis Bjorklund To: Sam Mason Cc: pgsql-performance@postgresql.org Subject: Re: planner picking more expensive plan In-Reply-To: <20050701133305.GB2623@sam.lan.samason.me.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.331 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/12 X-Sequence-Number: 13253 On Fri, 1 Jul 2005, Sam Mason wrote: The key thing with the query that Sam have is that if you turn off seqscan you get the first plan that run in 0.4ms and if seqscan is on the runtime is 27851ms. There are 100 way to make it select the seq scan, including rewriting the query to something more useful, tweaking different parameters and so on. The interesting part is that pg give the fast plan a cost of 202 and the slow a cost of 566141, but still it chooses the slow query unless seqscan is turned off (or some other tweak with the same effect). It know very well that the plan with the index scan will be much faster, it just don't manage to generate it unless you force it to. It makes you wonder if pg throws away some plans too early in the planning phase. > Limit (cost=0.00..202.52 rows=10 width=8) (actual time=0.221..0.600 rows=10 loops=1) > -> Merge Left Join (cost=0.00..66888828.30 rows=3302780 width=8) (actual time=0.211..0.576 rows=10 loops=1) > Merge Cond: ("outer".animalid = "inner".animalid) > -> Index Scan using animals_pkey on animals a (cost=0.00..10198983.91 rows=3302780 width=8) (actual time=0.112..0.276 rows=10 loops=1) > -> Index Scan using movement_animal on movements m (cost=0.00..56642740.73 rows=3107737 width=8) (actual time=0.088..0.235 rows=10 loops=1) > Filter: (mtypeid = 0) > Total runtime: 0.413 ms > > Limit (cost=565969.42..566141.09 rows=10 width=8) (actual time=27769.047..27769.246 rows=10 loops=1) > -> Merge Right Join (cost=565969.42..57264070.77 rows=3302780 width=8) (actual time=27769.043..27769.228 rows=10 loops=1) > Merge Cond: ("outer".animalid = "inner".animalid) > -> Index Scan using movement_animal on movements m (cost=0.00..56642740.73 rows=3107737 width=8) (actual time=0.022..0.154 rows=10 loops=1) > Filter: (mtypeid = 0) > -> Sort (cost=565969.42..574226.37 rows=3302780 width=8) (actual time=27768.991..27769.001 rows=10 loops=1) > Sort Key: a.animalid > -> Seq Scan on animals a (cost=0.00..77086.80 rows=3302780 width=8) (actual time=0.039..5620.651 rows=3303418 loops=1) > Total runtime: 27851.097 ms Another thing to notice is that if one remove the Limit node then the situation is reversed and the plan that pg choose (with the Limit node) is the one with the lowest cost. The startup cost is however very high so combining that Merge Join with a Limit will of course produce something slow compared to the upper plan where the startup cost is 0.0. A stand alone test case would be nice, but even without the above plans are interesting. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Sat Jul 2 10:44:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 75A29529C2 for ; Sat, 2 Jul 2005 10:44:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18485-04 for ; Sat, 2 Jul 2005 13:44:10 +0000 (GMT) Received: from mail.baymountain.com (mail.baymountain.com [8.7.96.19]) by svr1.postgresql.org (Postfix) with SMTP id C2014529D8 for ; Sat, 2 Jul 2005 10:44:08 -0300 (ADT) Received: (qmail 23726 invoked by uid 504); 2 Jul 2005 13:44:11 -0000 Received: from emil@baymountain.com by mail.baymountain.com by uid 501 with qmail-scanner-1.16 (spamassassin: 2.53. Clear:SA:0(0.0/7.0):. Processed in 0.245709 secs); 02 Jul 2005 13:44:11 -0000 Received: from unknown (HELO briggspack.com) (emil@briggspack.com@24.211.148.77) by mail.baymountain.com with SMTP; 2 Jul 2005 13:44:11 -0000 From: Emil Briggs Reply-To: emil@baymountain.com Organization: Baymountain, Inc. To: John A Meinel Subject: Re: Planner constants for RAM resident databases Date: Sat, 2 Jul 2005 09:44:07 -0400 User-Agent: KMail/1.7.1 Cc: Postgresql Performance References: <200507012159.39041.emil@baymountain.com> <200507012311.37942.emil@baymountain.com> <42C606AA.8060406@arbash-meinel.com> In-Reply-To: <42C606AA.8060406@arbash-meinel.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507020944.07339.emil@baymountain.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.038 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/14 X-Sequence-Number: 13255 > When you do "explain analyze" of a query that you have difficulties > with, how are the planner's estimates. Are the estimated number of rows > about equal to the actual number of rows? Some of them are pretty far off. For example -> Merge Left Join (cost=9707.71..13993.52 rows=1276 width=161) (actual time=164.423..361.477 rows=49 loops=1) I tried setting enable_merge_joins to off and that made the query about three times faster. It's using a hash join instead. From pgsql-performance-owner@postgresql.org Sat Jul 2 10:54:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4D2B8529C2 for ; Sat, 2 Jul 2005 10:54:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20937-01 for ; Sat, 2 Jul 2005 13:54:37 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E726C5299F for ; Sat, 2 Jul 2005 10:54:36 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j62DseIu025043; Sat, 2 Jul 2005 09:54:40 -0400 (EDT) To: Madison Kelly Cc: pgsql-performance@postgresql.org Subject: Re: B-Tree index not being used In-reply-to: <42C61537.3080601@alteeve.com> References: <42C61537.3080601@alteeve.com> Comments: In-reply-to Madison Kelly message dated "Sat, 02 Jul 2005 00:16:55 -0400" Date: Sat, 02 Jul 2005 09:54:40 -0400 Message-ID: <25042.1120312480@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/15 X-Sequence-Number: 13256 Madison Kelly writes: > Can anyone see why the index might not be being used? You didn't initdb in 'C' locale. You can either re-initdb, or create a specialized index with a non-default operator class to support LIKE. See the documentation. regards, tom lane From pgsql-performance-owner@postgresql.org Sat Jul 2 12:20:37 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DC2B85293E for ; Sat, 2 Jul 2005 12:20:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34842-07 for ; Sat, 2 Jul 2005 15:20:31 +0000 (GMT) Received: from tomts20-srv.bellnexxia.net (tomts20-srv.bellnexxia.net [209.226.175.74]) by svr1.postgresql.org (Postfix) with ESMTP id 5847D52923 for ; Sat, 2 Jul 2005 12:20:30 -0300 (ADT) Received: from [192.168.1.99] ([64.228.1.109]) by tomts20-srv.bellnexxia.net (InterMail vM.5.01.06.10 201-253-122-130-110-20040306) with ESMTP id <20050702152029.UYZG19894.tomts20-srv.bellnexxia.net@[192.168.1.99]>; Sat, 2 Jul 2005 11:20:29 -0400 Message-ID: <42C6B0B7.5010006@alteeve.com> Date: Sat, 02 Jul 2005 11:20:23 -0400 From: Madison Kelly User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Debian/1.7.8-1 X-Accept-Language: en-us, en, ja MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: B-Tree index not being used References: <42C61537.3080601@alteeve.com> <25042.1120312480@sss.pgh.pa.us> In-Reply-To: <25042.1120312480@sss.pgh.pa.us> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/16 X-Sequence-Number: 13257 Tom Lane wrote: > Madison Kelly writes: > >> Can anyone see why the index might not be being used? > > > You didn't initdb in 'C' locale. You can either re-initdb, > or create a specialized index with a non-default operator class > to support LIKE. See the documentation. > > regards, tom lane I'll look into the non-default op class. I want to keep anything that tweaks the DB in my code so that a user doesn't need to modify anything on their system. Thanks! Madison -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Madison Kelly (Digimer) TLE-BU, The Linux Experience; Back Up http://tle-bu.thelinuxexperience.com -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- From pgsql-performance-owner@postgresql.org Sun Jul 3 20:29:40 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B00B652834 for ; Sun, 3 Jul 2005 20:29:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44749-09 for ; Sun, 3 Jul 2005 23:29:26 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 0ED6D5282C for ; Sun, 3 Jul 2005 20:29:24 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j63MSURN008589 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 00:28:30 +0200 Date: Mon, 4 Jul 2005 00:28:30 +0200 From: Enrico Weigelt To: pgsql-performance@postgresql.org Subject: Re: investigating slow queries through pg_stat_activity Message-ID: <20050703222830.GA2056@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: pgsql-performance@postgresql.org References: <0B5C29FA-E949-447E-AEFE-A806E238F1C2@drivefaster.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <0B5C29FA-E949-447E-AEFE-A806E238F1C2@drivefaster.net> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/17 X-Sequence-Number: 13258 * Dan Harris wrote: Hi, > I've got some queries generated by my application that will, for some > reason, run forever until I kill the pid. Yet, when I run the > queries manually to check them out, they usually work fine. If you can change your application, you could try to encapsulate the queries into views - this makes logging and tracking down problems much easier. cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Sun Jul 3 20:46:45 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D109C52846 for ; Sun, 3 Jul 2005 20:46:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54414-07 for ; Sun, 3 Jul 2005 23:46:34 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 473C452809 for ; Sun, 3 Jul 2005 20:46:32 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j63Mjckt011851 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 00:45:38 +0200 Date: Mon, 4 Jul 2005 00:45:37 +0200 From: Enrico Weigelt To: postgresql performance list Subject: plain inserts and deletes very slow Message-ID: <20050703224537.GA9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/18 X-Sequence-Number: 13259 Hi folks, my application reads and writes some table quite often (multiple times per second). these tables are quite small (not more than 20 tuples), but the operations take quite a long time (>300 ms!). The query operations are just include text matching (=) and date comparison (<,>). I wasn't yet able to track down, if all these queries take sucha long time or just sometimes. When running them manually or trying explain, evrything's fast. Probably there could be some side effects with other concurrent quries. Could anyone give me advice ? cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Sun Jul 3 20:58:10 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 36D1952867 for ; Sun, 3 Jul 2005 20:58:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 56234-07 for ; Sun, 3 Jul 2005 23:58:03 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 810E852834 for ; Sun, 3 Jul 2005 20:58:01 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j63Mv7RF013155 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 00:57:07 +0200 Date: Mon, 4 Jul 2005 00:57:07 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050703225707.GB9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050703224537.GA9777@nibiru.borg.metux.de> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/20 X-Sequence-Number: 13261 * Enrico Weigelt wrote: forgot to mention: + linux-2.6.9 + postgres-7.4.6 + intel celeron 2ghz + intel ultra ata controller + 768mb ram cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Sun Jul 3 20:58:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 99A8652896 for ; Sun, 3 Jul 2005 20:58:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62330-02 for ; Sun, 3 Jul 2005 23:58:54 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 2C33152882 for ; Sun, 3 Jul 2005 20:58:52 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j63MvxA3013173 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 00:57:59 +0200 Date: Mon, 4 Jul 2005 00:57:59 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050703225759.GC9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703235424.GA16122@uio.no> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050703235424.GA16122@uio.no> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/21 X-Sequence-Number: 13262 * Steinar H. Gunderson wrote: > On Mon, Jul 04, 2005 at 12:45:37AM +0200, Enrico Weigelt wrote: > > my application reads and writes some table quite often > > (multiple times per second). these tables are quite small > > (not more than 20 tuples), but the operations take quite a > > long time (>300 ms!). > > Are you VACUUMing often enough? I've just VACUUM'ed multiple times, so it's perhaps not the problem. cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Sun Jul 3 20:54:32 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 008285284D for ; Sun, 3 Jul 2005 20:54:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 53729-08 for ; Sun, 3 Jul 2005 23:54:27 +0000 (GMT) Received: from cassarossa.samfundet.no (cassarossa.samfundet.no [129.241.93.19]) by svr1.postgresql.org (Postfix) with ESMTP id 8D84A52846 for ; Sun, 3 Jul 2005 20:54:26 -0300 (ADT) Received: from trofast.ipv6.sesse.net ([2001:700:300:dc03:20e:cff:fe36:a766] helo=trofast.sesse.net) by cassarossa.samfundet.no with esmtp (Exim 4.50) id 1DpEI5-0001uT-B7 for pgsql-performance@postgresql.org; Mon, 04 Jul 2005 01:54:26 +0200 Received: from sesse by trofast.sesse.net with local (Exim 3.36 #1 (Debian)) id 1DpEI4-0004DV-00 for ; Mon, 04 Jul 2005 01:54:24 +0200 Date: Mon, 4 Jul 2005 01:54:24 +0200 From: "Steinar H. Gunderson" To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050703235424.GA16122@uio.no> Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20050703224537.GA9777@nibiru.borg.metux.de> X-Operating-System: Linux 2.6.11.8 on a i686 X-Message-Flag: Outlook? --> http://www.mozilla.org/products/thunderbird/ User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.019 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/19 X-Sequence-Number: 13260 On Mon, Jul 04, 2005 at 12:45:37AM +0200, Enrico Weigelt wrote: > my application reads and writes some table quite often > (multiple times per second). these tables are quite small > (not more than 20 tuples), but the operations take quite a > long time (>300 ms!). Are you VACUUMing often enough? /* Steinar */ -- Homepage: http://www.sesse.net/ From pgsql-performance-owner@postgresql.org Sun Jul 3 22:03:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EF14C52809 for ; Sun, 3 Jul 2005 22:03:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76170-01 for ; Mon, 4 Jul 2005 01:02:53 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 3202D528AD for ; Sun, 3 Jul 2005 22:02:51 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j6401vQF024275 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 02:01:57 +0200 Date: Mon, 4 Jul 2005 02:01:57 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050704000157.GD9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42C887DC.40308@telogis.com> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/23 X-Sequence-Number: 13264 * David Mitchell wrote: > Did you vacuum full? > > When you do lots of inserts and deletes, dead tuples get left behind. > When you vacuum, postgres will reuse those dead tuples, but if you don't > vacuum for a long time these tuples will build up lots. Even when you > vacuum in this case, the dead tuples are still there, although they are > marked for reuse. Vacuuming full actually removes the dead tuples. I'm doing a VACUUM ANALYZE every 6 hours. vacuum'ing manually doesnt seem to have any effect on that. cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Sun Jul 3 22:18:54 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 38360528B1 for ; Sun, 3 Jul 2005 22:18:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76159-03 for ; Mon, 4 Jul 2005 01:18:43 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 5C83A528A9 for ; Sun, 3 Jul 2005 22:18:41 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j640HlXN027352 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 02:17:47 +0200 Date: Mon, 4 Jul 2005 02:17:47 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050704001747.GE9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> <42C88D17.4080004@telogis.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42C88D17.4080004@telogis.com> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/25 X-Sequence-Number: 13266 * David Mitchell wrote: > Perhaps if you are doing a lot of inserts and deletes, vacuuming every 6 > minutes would be closer to your mark. Try vacuuming every 15 minutes for > a start and see how that affects things (you will have to do a vacuum > full to get the tables back into shape after them slowing down as they > have). hmm. I've just done vacuum full at the moment on these tables, but it doesnt seem to change anything :( cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Sun Jul 3 21:50:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8BA5A528B4 for ; Sun, 3 Jul 2005 21:50:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71240-09 for ; Mon, 4 Jul 2005 00:50:43 +0000 (GMT) Received: from smtp004.bizmail.sc5.yahoo.com (smtp004.bizmail.sc5.yahoo.com [66.163.175.81]) by svr1.postgresql.org (Postfix) with SMTP id A957B528AD for ; Sun, 3 Jul 2005 21:50:41 -0300 (ADT) Received: (qmail 41774 invoked from network); 4 Jul 2005 00:50:44 -0000 Received: from unknown (HELO ?192.168.3.41?) (david.mitchell@telogis.com@203.98.10.169 with plain) by smtp004.bizmail.sc5.yahoo.com with SMTP; 4 Jul 2005 00:50:44 -0000 Message-ID: <42C887DC.40308@telogis.com> Date: Mon, 04 Jul 2005 12:50:36 +1200 From: David Mitchell User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: weigelt@metux.de Cc: postgresql performance list Subject: Re: plain inserts and deletes very slow References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> In-Reply-To: <20050703225707.GB9777@nibiru.borg.metux.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/22 X-Sequence-Number: 13263 Did you vacuum full? When you do lots of inserts and deletes, dead tuples get left behind. When you vacuum, postgres will reuse those dead tuples, but if you don't vacuum for a long time these tuples will build up lots. Even when you vacuum in this case, the dead tuples are still there, although they are marked for reuse. Vacuuming full actually removes the dead tuples. If you vacuum (normal) regularly, then the number of dead tuples will stay down, as they are regularly marked for reuse. David Enrico Weigelt wrote: > * Enrico Weigelt wrote: > > forgot to mention: > > + linux-2.6.9 > + postgres-7.4.6 > + intel celeron 2ghz > + intel ultra ata controller > + 768mb ram > > > cu -- David Mitchell Software Engineer Telogis NOTICE: This message (including any attachments) contains CONFIDENTIAL INFORMATION intended for a specific individual and purpose, and is protected by law. If you are not the intended recipient, you should delete this message and are hereby notified that any disclosure, copying, or distribution of this message, or the taking of any action based on it, is strictly prohibited. From pgsql-performance-owner@postgresql.org Sun Jul 3 22:13:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 88F38528AD for ; Sun, 3 Jul 2005 22:13:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74199-10 for ; Mon, 4 Jul 2005 01:13:00 +0000 (GMT) Received: from smtp005.bizmail.sc5.yahoo.com (smtp005.bizmail.sc5.yahoo.com [66.163.175.82]) by svr1.postgresql.org (Postfix) with SMTP id 357CF52828 for ; Sun, 3 Jul 2005 22:12:59 -0300 (ADT) Received: (qmail 60666 invoked from network); 4 Jul 2005 01:13:02 -0000 Received: from unknown (HELO ?192.168.3.41?) (david.mitchell@telogis.com@203.98.10.169 with plain) by smtp005.bizmail.sc5.yahoo.com with SMTP; 4 Jul 2005 01:13:02 -0000 Message-ID: <42C88D17.4080004@telogis.com> Date: Mon, 04 Jul 2005 13:12:55 +1200 From: David Mitchell User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: weigelt@metux.de Cc: postgresql performance list Subject: Re: plain inserts and deletes very slow References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> In-Reply-To: <20050704000157.GD9777@nibiru.borg.metux.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/24 X-Sequence-Number: 13265 Perhaps if you are doing a lot of inserts and deletes, vacuuming every 6 minutes would be closer to your mark. Try vacuuming every 15 minutes for a start and see how that affects things (you will have to do a vacuum full to get the tables back into shape after them slowing down as they have). David Enrico Weigelt wrote: > * David Mitchell wrote: > >>Did you vacuum full? >> >>When you do lots of inserts and deletes, dead tuples get left behind. >>When you vacuum, postgres will reuse those dead tuples, but if you don't >>vacuum for a long time these tuples will build up lots. Even when you >>vacuum in this case, the dead tuples are still there, although they are >>marked for reuse. Vacuuming full actually removes the dead tuples. > > > I'm doing a VACUUM ANALYZE every 6 hours. > > vacuum'ing manually doesnt seem to have any effect on that. > > > cu -- David Mitchell Software Engineer Telogis NOTICE: This message (including any attachments) contains CONFIDENTIAL INFORMATION intended for a specific individual and purpose, and is protected by law. If you are not the intended recipient, you should delete this message and are hereby notified that any disclosure, copying, or distribution of this message, or the taking of any action based on it, is strictly prohibited. From pgsql-performance-owner@postgresql.org Mon Jul 4 00:09:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 40381528A6 for ; Mon, 4 Jul 2005 00:09:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 01272-06 for ; Mon, 4 Jul 2005 03:09:46 +0000 (GMT) Received: from mr2.surnet.cl (smtp2.surnet.cl [216.155.73.163]) by svr1.postgresql.org (Postfix) with ESMTP id 816665288E for ; Mon, 4 Jul 2005 00:09:46 -0300 (ADT) Received: from smtp2.surnet.cl (216.155.73.169) by mr2.surnet.cl (7.0.031.3) id 4259ADFA012BF3FC for pgsql-performance@postgresql.org; Sun, 3 Jul 2005 23:09:39 -0400 Received: from smtp2.surnet.cl (mr2.surnet.cl []) by mr2.surnet.cl ([216.155.73.169]); Mon, 04 Jul 2005 03:09:39 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp2.surnet.cl (7.0.031.3) id 4259ADE300B65A19 for pgsql-performance@postgresql.org; Sun, 3 Jul 2005 23:09:39 -0400 Received: from alvh.no-ip.org (216.155.79.182) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF6000224778 for pgsql-performance@postgresql.org; Sun, 3 Jul 2005 23:09:45 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 86DCDC2DC70; Sun, 3 Jul 2005 23:10:04 -0400 (CLT) Resent-From: alvherre@surnet.cl Resent-Date: Sun, 3 Jul 2005 23:10:04 -0400 Resent-Message-ID: <20050704031004.GI24255@alvh.no-ip.org> Resent-To: pgsql-performance@postgresql.org Date: Sun, 3 Jul 2005 22:57:09 -0400 From: Alvaro Herrera To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050704025709.GA24255@alvh.no-ip.org> References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> <42C88D17.4080004@telogis.com> <20050704001747.GE9777@nibiru.borg.metux.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050704001747.GE9777@nibiru.borg.metux.de> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.418 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/26 X-Sequence-Number: 13267 On Mon, Jul 04, 2005 at 02:17:47AM +0200, Enrico Weigelt wrote: > * David Mitchell wrote: > > Perhaps if you are doing a lot of inserts and deletes, vacuuming every 6 > > minutes would be closer to your mark. Try vacuuming every 15 minutes for > > a start and see how that affects things (you will have to do a vacuum > > full to get the tables back into shape after them slowing down as they > > have). > > hmm. I've just done vacuum full at the moment on these tables, but it > doesnt seem to change anything :( Maybe you need a REINDEX, if you have indexes on that table. Try that, coupled with the frequent VACUUM suggestion. -- Alvaro Herrera () "World domination is proceeding according to plan" (Andrew Morton) From pgsql-performance-owner@postgresql.org Mon Jul 4 00:51:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 78F3152892 for ; Mon, 4 Jul 2005 00:51:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 08543-09 for ; Mon, 4 Jul 2005 03:51:41 +0000 (GMT) Received: from smtp008.bizmail.sc5.yahoo.com (smtp008.bizmail.sc5.yahoo.com [66.163.170.74]) by svr1.postgresql.org (Postfix) with SMTP id D7A8D52896 for ; Mon, 4 Jul 2005 00:51:36 -0300 (ADT) Received: (qmail 62853 invoked from network); 4 Jul 2005 03:51:35 -0000 Received: from unknown (HELO ?192.168.3.41?) (david.mitchell@telogis.com@203.98.10.169 with plain) by smtp008.bizmail.sc5.yahoo.com with SMTP; 4 Jul 2005 03:51:35 -0000 Message-ID: <42C8B23F.4000400@telogis.com> Date: Mon, 04 Jul 2005 15:51:27 +1200 From: David Mitchell User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: weigelt@metux.de Cc: postgresql performance list Subject: Re: plain inserts and deletes very slow References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> <42C88D17.4080004@telogis.com> <20050704001747.GE9777@nibiru.borg.metux.de> In-Reply-To: <20050704001747.GE9777@nibiru.borg.metux.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/27 X-Sequence-Number: 13268 Hmm, you said you don't experience this when executing the query manually. What adapter are you using to access postgres from your application? libpq, npgsql or something else? And what is your method for running the query 'manually'. Are you running it locally or from a remote machine or what? Regards David Enrico Weigelt wrote: > * David Mitchell wrote: > >>Perhaps if you are doing a lot of inserts and deletes, vacuuming every 6 >>minutes would be closer to your mark. Try vacuuming every 15 minutes for >>a start and see how that affects things (you will have to do a vacuum >>full to get the tables back into shape after them slowing down as they >>have). > > > hmm. I've just done vacuum full at the moment on these tables, but it > doesnt seem to change anything :( > > > cu -- David Mitchell Software Engineer Telogis From pgsql-performance-owner@postgresql.org Mon Jul 4 06:58:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EE3CA52809 for ; Mon, 4 Jul 2005 06:58:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79253-03 for ; Mon, 4 Jul 2005 09:58:39 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 959F85281B for ; Mon, 4 Jul 2005 06:58:27 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j648vTjn014161 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 10:57:29 +0200 Date: Mon, 4 Jul 2005 10:57:29 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050704085729.GF9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> <42C88D17.4080004@telogis.com> <20050704001747.GE9777@nibiru.borg.metux.de> <20050704025709.GA24255@alvh.no-ip.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050704025709.GA24255@alvh.no-ip.org> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/28 X-Sequence-Number: 13269 * Alvaro Herrera wrote: > On Mon, Jul 04, 2005 at 02:17:47AM +0200, Enrico Weigelt wrote: > > * David Mitchell wrote: > > > Perhaps if you are doing a lot of inserts and deletes, vacuuming every 6 > > > minutes would be closer to your mark. Try vacuuming every 15 minutes for > > > a start and see how that affects things (you will have to do a vacuum > > > full to get the tables back into shape after them slowing down as they > > > have). > > > > hmm. I've just done vacuum full at the moment on these tables, but it > > doesnt seem to change anything :( > > Maybe you need a REINDEX, if you have indexes on that table. Try that, > coupled with the frequent VACUUM suggestion. I've tried it, but it doesn't seem to help :( cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Mon Jul 4 07:00:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 20F1752838 for ; Mon, 4 Jul 2005 07:00:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79269-04 for ; Mon, 4 Jul 2005 10:00:09 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 954E45282B for ; Mon, 4 Jul 2005 07:00:04 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j648x4gp014262 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 10:59:04 +0200 Date: Mon, 4 Jul 2005 10:59:03 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050704085903.GG9777@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> <42C88D17.4080004@telogis.com> <20050704001747.GE9777@nibiru.borg.metux.de> <42C8B23F.4000400@telogis.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42C8B23F.4000400@telogis.com> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/29 X-Sequence-Number: 13270 * David Mitchell wrote: Hi, > Hmm, you said you don't experience this when executing the query > manually. What adapter are you using to access postgres from your > application? libpq, npgsql or something else? huh, its a delphi application ... (I didnt code it). > And what is your method for running the query 'manually'. Are you > running it locally or from a remote machine or what? using psql remotely - database and client machines are sitting on the same wire. cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Mon Jul 4 11:27:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9537F5286E for ; Mon, 4 Jul 2005 11:27:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 30152-01 for ; Mon, 4 Jul 2005 14:27:32 +0000 (GMT) Received: from mr1.surnet.cl (smtp1.surnet.cl [216.155.73.162]) by svr1.postgresql.org (Postfix) with ESMTP id D07D352806 for ; Mon, 4 Jul 2005 11:27:30 -0300 (ADT) Received: from smtp1.surnet.cl (216.155.73.168) by mr1.surnet.cl (7.0.031.3) id 42587EDE012D2108 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 10:26:59 -0400 Received: from smtp1.surnet.cl (mr1.surnet.cl []) by mr1.surnet.cl ([216.155.73.168]); Mon, 04 Jul 2005 14:26:58 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp1.surnet.cl (7.0.031.3) id 42587C6E00EA719F for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 10:26:58 -0400 Received: from alvh.no-ip.org (216.155.79.182) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF6000231915 for pgsql-performance@postgresql.org; Mon, 4 Jul 2005 10:27:34 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 49A8BC2DC70; Mon, 4 Jul 2005 10:27:55 -0400 (CLT) Date: Mon, 4 Jul 2005 10:27:55 -0400 From: Alvaro Herrera To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050704142755.GA12062@alvh.no-ip.org> References: <20050703224537.GA9777@nibiru.borg.metux.de> <20050703225707.GB9777@nibiru.borg.metux.de> <42C887DC.40308@telogis.com> <20050704000157.GD9777@nibiru.borg.metux.de> <42C88D17.4080004@telogis.com> <20050704001747.GE9777@nibiru.borg.metux.de> <20050704025709.GA24255@alvh.no-ip.org> <20050704085729.GF9777@nibiru.borg.metux.de> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <20050704085729.GF9777@nibiru.borg.metux.de> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.416 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/30 X-Sequence-Number: 13271 On Mon, Jul 04, 2005 at 10:57:29AM +0200, Enrico Weigelt wrote: > * Alvaro Herrera wrote: > > On Mon, Jul 04, 2005 at 02:17:47AM +0200, Enrico Weigelt wrote: > > > * David Mitchell wrote: > > > > Perhaps if you are doing a lot of inserts and deletes, vacuuming every 6 > > > > minutes would be closer to your mark. Try vacuuming every 15 minutes for > > > > a start and see how that affects things (you will have to do a vacuum > > > > full to get the tables back into shape after them slowing down as they > > > > have). > > > > > > hmm. I've just done vacuum full at the moment on these tables, but it > > > doesnt seem to change anything :( > > > > Maybe you need a REINDEX, if you have indexes on that table. Try that, > > coupled with the frequent VACUUM suggestion. > > I've tried it, but it doesn't seem to help :( So, lets back up a little. You have no table nor index bloat, because you reindexed and full-vacuumed. So where does the slowness come from? Can you post an example EXPLAIN ANALYZE of the queries in question? -- Alvaro Herrera () "El realista sabe lo que quiere; el idealista quiere lo que sabe" (An�nimo) From pgsql-performance-owner@postgresql.org Mon Jul 4 16:55:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id ED502528BB for ; Mon, 4 Jul 2005 16:55:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92239-01 for ; Mon, 4 Jul 2005 19:55:38 +0000 (GMT) Received: from relais.videotron.ca (relais.videotron.ca [24.201.245.36]) by svr1.postgresql.org (Postfix) with ESMTP id 071FF5289B for ; Mon, 4 Jul 2005 16:55:37 -0300 (ADT) Received: from [192.168.1.101] ([24.202.23.128]) by VL-MO-MR011.ip.videotron.ca (iPlanet Messaging Server 5.2 HotFix 1.21 (built Sep 8 2003)) with ESMTP id <0IJ4008UCCIHLF@VL-MO-MR011.ip.videotron.ca> for pgsql-performance@postgresql.org; Mon, 04 Jul 2005 15:51:53 -0400 (EDT) Date: Mon, 04 Jul 2005 15:57:49 -0400 From: David Gagnon Subject: Why the planner is not using the INDEX . In-reply-to: <20050630235821.GA2163@calvin.surfutopia.net> Cc: pgsql-performance@postgresql.org Message-id: <42C994BD.3000909@siunik.com> MIME-version: 1.0 Content-type: multipart/alternative; boundary="Boundary_(ID_UtyElAxQHbaJrkD3SgFi5w)" X-Accept-Language: en-us, en User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.169 tagged_above=0 required=5 tests=AWL, HTML_60_70, HTML_MESSAGE, HTML_TAG_EXIST_TBODY, HTML_TITLE_EMPTY X-Spam-Level: X-Archive-Number: 200507/31 X-Sequence-Number: 13272 This is a multi-part message in MIME format. --Boundary_(ID_UtyElAxQHbaJrkD3SgFi5w) Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7BIT Hi all, If you can just help my understanding the choice of the planner. Here is the Query: explain analyse SELECT IRNUM FROM IR INNER JOIN IT ON IT.ITIRNUM = ANY ('{1000, 2000}') AND IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M' Here is the Query plan: QUERY PLAN Hash Join (cost=1142.47..5581.75 rows=87 width=4) (actual time=125.000..203.000 rows=2 loops=1) Hash Cond: ("outer".itirnum = "inner".irnum) -> Seq Scan on it (cost=0.00..3093.45 rows=31646 width=9) (actual time=0.000..78.000 rows=2 loops=1) Filter: ((itirnum = ANY ('{1000,2000}'::integer[])) AND ((itypnum)::text = 'M'::text)) -> Hash (cost=1142.09..1142.09 rows=151 width=37) (actual time=125.000..125.000 rows=0 loops=1) -> Index Scan using ir_pk on ir (cost=0.00..1142.09 rows=151 width=37) (actual time=0.000..125.000 rows=2 loops=1) Index Cond: ((irypnum)::text = 'M'::text) Filter: (irnum = ANY ('{1000,2000}'::integer[])) Total runtime: 203.000 ms I don't understand why the planner do a Seq Scan (Seq Scan on table IT ..) instead of passing by the followin index: ALTER TABLE IT ADD CONSTRAINT IT_IR_FK foreign key (ITYPNUM,ITIRNUM) references IR (IRYPNUM, IRNUM) ON UPDATE CASCADE; I tried some stuff but I'm not able to change this behavior. The IT and IR table may be quite huge (from 20k to 1600k rows) so I think doing a SEQ SCAN is not a good idea.. am I wrong? Is this query plan is oki for you ? Thanks for your help. /David P.S.: I'm using postgresql 8.0.3 on windows and I change those setting in my postgresql.conf : shared_buffers = 12000 # min 16, at least max_connections*2, 8KB each work_mem = 15000 # min 64, size in KB --Boundary_(ID_UtyElAxQHbaJrkD3SgFi5w) Content-type: text/html; charset=ISO-8859-1 Content-transfer-encoding: 7BIT Hi all,

  If you can just help my understanding the choice of the planner. 

Here is the Query:
 explain analyse SELECT IRNUM FROM IR
        INNER JOIN IT ON  IT.ITIRNUM = ANY ('{1000, 2000}') AND IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM 
        WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M'

Here is the Query plan:

QUERY PLAN






Hash Join  (cost=1142.47..5581.75 rows=87 width=4) (actual time=125.000..203.000 rows=2 loops=1)
  Hash Cond: ("outer".itirnum = "inner".irnum)




  ->  Seq Scan on it  (cost=0.00..3093.45 rows=31646 width=9) (actual time=0.000..78.000 rows=2 loops=1)
        Filter: ((itirnum = ANY ('{1000,2000}'::integer[])) AND ((itypnum)::text = 'M'::text))

  ->  Hash  (cost=1142.09..1142.09 rows=151 width=37) (actual time=125.000..125.000 rows=0 loops=1)
        ->  Index Scan using ir_pk on ir  (cost=0.00..1142.09 rows=151 width=37) (actual time=0.000..125.000 rows=2 loops=1)
              Index Cond: ((irypnum)::text = 'M'::text)




              Filter: (irnum = ANY ('{1000,2000}'::integer[]))




Total runtime: 203.000 ms

















    I don't understand why the planner do a Seq Scan (Seq Scan on table IT ..) instead of passing by the followin index:
    ALTER TABLE IT ADD CONSTRAINT IT_IR_FK foreign key (ITYPNUM,ITIRNUM) references IR (IRYPNUM, IRNUM) ON UPDATE CASCADE;

I tried some stuff but I'm not able to change this behavior.  The IT and IR table may be quite huge (from 20k to 1600k rows) so I think doing a SEQ SCAN is not a good idea.. am I wrong?  Is this query plan is oki for you ?

Thanks for your help.

/David
 P.S.: I'm using postgresql 8.0.3 on windows and I change those setting in  my postgresql.conf :
shared_buffers = 12000        # min 16, at least max_connections*2, 8KB each
work_mem = 15000        # min 64, size in KB



--Boundary_(ID_UtyElAxQHbaJrkD3SgFi5w)-- From pgsql-performance-owner@postgresql.org Mon Jul 4 17:27:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1B53A52846 for ; Mon, 4 Jul 2005 17:27:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94578-06 for ; Mon, 4 Jul 2005 20:27:42 +0000 (GMT) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 3E24052808 for ; Mon, 4 Jul 2005 17:27:41 -0300 (ADT) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 12CB535578; Mon, 4 Jul 2005 13:27:41 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 10DAF354C0; Mon, 4 Jul 2005 13:27:41 -0700 (PDT) Date: Mon, 4 Jul 2005 13:27:41 -0700 (PDT) From: Stephan Szabo To: David Gagnon Cc: pgsql-performance@postgresql.org Subject: Re: Why the planner is not using the INDEX . In-Reply-To: <42C994BD.3000909@siunik.com> Message-ID: <20050704132446.O14579@megazone.bigpanda.com> References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/32 X-Sequence-Number: 13273 On Mon, 4 Jul 2005, David Gagnon wrote: > If you can just help my understanding the choice of the planner. > > Here is the Query: > explain analyse SELECT IRNUM FROM IR > INNER JOIN IT ON IT.ITIRNUM = ANY ('{1000, 2000}') AND > IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM > WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M' > > Here is the Query plan: > > QUERY PLAN > > Hash Join (cost=1142.47..5581.75 rows=87 width=4) (actual > time=125.000..203.000 rows=2 loops=1) > Hash Cond: ("outer".itirnum = "inner".irnum) > -> Seq Scan on it (cost=0.00..3093.45 rows=31646 width=9) (actual > time=0.000..78.000 rows=2 loops=1) > Filter: ((itirnum = ANY ('{1000,2000}'::integer[])) AND > ((itypnum)::text = 'M'::text)) > > -> Hash (cost=1142.09..1142.09 rows=151 width=37) (actual > time=125.000..125.000 rows=0 loops=1) > -> Index Scan using ir_pk on ir (cost=0.00..1142.09 rows=151 > width=37) (actual time=0.000..125.000 rows=2 loops=1) > Index Cond: ((irypnum)::text = 'M'::text) > Filter: (irnum = ANY ('{1000,2000}'::integer[])) > Total runtime: 203.000 ms > I don't understand why the planner do a Seq Scan (Seq Scan on table > IT ..) instead of passing by the followin index: > ALTER TABLE IT ADD CONSTRAINT IT_IR_FK foreign key (ITYPNUM,ITIRNUM) > references IR (IRYPNUM, IRNUM) ON UPDATE CASCADE; That doesn't create an index on IT. Primary keys (and unique constraints) create indexes, but not foreign keys. Did you also create an index on those fields? Also it looks like it's way overestimating the number of rows that condition would succeed for. You might consider raising the statistics targets on those columns and reanalyzing. From pgsql-performance-owner@postgresql.org Mon Jul 4 20:32:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 44E9B5282F for ; Mon, 4 Jul 2005 20:32:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 26360-05 for ; Mon, 4 Jul 2005 23:32:21 +0000 (GMT) Received: from mailhub2.une.edu.au (mailhub2.une.edu.au [129.180.1.142]) by svr1.postgresql.org (Postfix) with ESMTP id 5D5D052828 for ; Mon, 4 Jul 2005 20:32:19 -0300 (ADT) Received: from icarus.une.edu.au (icarus.une.edu.au [129.180.47.120]) by mailhub2.une.edu.au (Postfix) with ESMTP id EB53A7F8C; Tue, 5 Jul 2005 09:32:22 +1000 (EST) Received: from kgb (unknown [129.180.47.225]) by icarus.une.edu.au (Postfix) with SMTP id 663D0355562; Tue, 5 Jul 2005 09:32:22 +1000 (EST) Date: Tue, 05 Jul 2005 09:36:02 +1000 From: Klint Gore To: weigelt@metux.de Cc: postgresql performance list Subject: Re: plain inserts and deletes very slow In-Reply-To: <20050704085903.GG9777@nibiru.borg.metux.de> References: <42C8B23F.4000400@telogis.com> <20050704085903.GG9777@nibiru.borg.metux.de> Message-Id: <42C9C7E278.E5BFKG@129.180.47.120> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver 1.25.06 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/33 X-Sequence-Number: 13274 On Mon, 4 Jul 2005 10:59:03 +0200, Enrico Weigelt wrote: > * David Mitchell wrote: > > Hi, > > > Hmm, you said you don't experience this when executing the query > > manually. What adapter are you using to access postgres from your > > application? libpq, npgsql or something else? > > huh, its a delphi application ... (I didnt code it). Turn on statement logging. I've seen delphi interfaces do extra queries on system tables to find some structure information. The available interfaces for delphi that I know of are vitavoom's dbexpress (you should be able to find dbexppge.dll), zeos (you'll have to grep the executable), ODBC using ADO or bde, Or dot net. klint. +---------------------------------------+-----------------+ : Klint Gore : "Non rhyming : : EMail : kg@kgb.une.edu.au : slang - the : : Snail : A.B.R.I. : possibilities : : Mail University of New England : are useless" : : Armidale NSW 2351 Australia : L.J.J. : : Fax : +61 2 6772 5376 : : +---------------------------------------+-----------------+ From pgsql-performance-owner@postgresql.org Mon Jul 4 21:25:32 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 09268528C7 for ; Mon, 4 Jul 2005 21:25:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40363-01 for ; Tue, 5 Jul 2005 00:25:18 +0000 (GMT) Received: from relais.videotron.ca (relais.videotron.ca [24.201.245.36]) by svr1.postgresql.org (Postfix) with ESMTP id 3A5995288B for ; Mon, 4 Jul 2005 21:25:17 -0300 (ADT) Received: from [192.168.1.101] ([24.202.23.128]) by VL-MO-MR001.ip.videotron.ca (iPlanet Messaging Server 5.2 HotFix 1.21 (built Sep 8 2003)) with ESMTP id <0IJ400H2FP3VDT@VL-MO-MR001.ip.videotron.ca> for pgsql-performance@postgresql.org; Mon, 04 Jul 2005 20:23:55 -0400 (EDT) Date: Mon, 04 Jul 2005 20:29:50 -0400 From: David Gagnon Subject: Re: Why the planner is not using the INDEX . In-reply-to: <20050704132446.O14579@megazone.bigpanda.com> To: Stephan Szabo Cc: pgsql-performance@postgresql.org Message-id: <42C9D47E.7090103@siunik.com> MIME-version: 1.0 Content-type: multipart/alternative; boundary="Boundary_(ID_/5SMBe2bXvLrkUkYZhz9qA)" X-Accept-Language: en-us, en User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> <20050704132446.O14579@megazone.bigpanda.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.271 tagged_above=0 required=5 tests=AWL, HTML_20_30, HTML_MESSAGE X-Spam-Level: X-Archive-Number: 200507/34 X-Sequence-Number: 13275 This is a multi-part message in MIME format. --Boundary_(ID_/5SMBe2bXvLrkUkYZhz9qA) Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7BIT Thanks .. I miss that FK don't create indexed ... since Primary key implicitly does ... I'm a bit surprised of that behavior thought, since it means that if we delete a row from table A all tables (B,C,D) with FK pointing to this table (A) must be scanned. If there is no index on those tables it means we gone do all Sequantial scans. Than can cause significant performance problem!!!. Is there a reason why implicit index aren't created when FK are declared. I looked into the documentation and I haven't found a way to tell postgresql to automatically create an index when creating la FK. Does it means I need to manage it EXPLICITLY with create index statement ? Is there another way ? Thanks for you help that simple answer will solve a lot of performance problem I have ... /David >On Mon, 4 Jul 2005, David Gagnon wrote: > > > >> If you can just help my understanding the choice of the planner. >> >>Here is the Query: >> explain analyse SELECT IRNUM FROM IR >> INNER JOIN IT ON IT.ITIRNUM = ANY ('{1000, 2000}') AND >>IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM >> WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M' >> >>Here is the Query plan: >> >>QUERY PLAN >> >>Hash Join (cost=1142.47..5581.75 rows=87 width=4) (actual >>time=125.000..203.000 rows=2 loops=1) >> Hash Cond: ("outer".itirnum = "inner".irnum) >> -> Seq Scan on it (cost=0.00..3093.45 rows=31646 width=9) (actual >>time=0.000..78.000 rows=2 loops=1) >> Filter: ((itirnum = ANY ('{1000,2000}'::integer[])) AND >>((itypnum)::text = 'M'::text)) >> >> -> Hash (cost=1142.09..1142.09 rows=151 width=37) (actual >>time=125.000..125.000 rows=0 loops=1) >> -> Index Scan using ir_pk on ir (cost=0.00..1142.09 rows=151 >>width=37) (actual time=0.000..125.000 rows=2 loops=1) >> Index Cond: ((irypnum)::text = 'M'::text) >> Filter: (irnum = ANY ('{1000,2000}'::integer[])) >>Total runtime: 203.000 ms >> >> > > > >> I don't understand why the planner do a Seq Scan (Seq Scan on table >>IT ..) instead of passing by the followin index: >> ALTER TABLE IT ADD CONSTRAINT IT_IR_FK foreign key (ITYPNUM,ITIRNUM) >>references IR (IRYPNUM, IRNUM) ON UPDATE CASCADE; >> >> > >That doesn't create an index on IT. Primary keys (and unique constraints) >create indexes, but not foreign keys. Did you also create an index on >those fields? > >Also it looks like it's way overestimating the number of rows that >condition would succeed for. You might consider raising the statistics >targets on those columns and reanalyzing. > > > --Boundary_(ID_/5SMBe2bXvLrkUkYZhz9qA) Content-type: text/html; charset=ISO-8859-1 Content-transfer-encoding: 7BIT Thanks .. I miss that FK don't create indexed ...  since Primary key implicitly does ...

I'm a bit surprised of that behavior thought, since it means that if we delete a row from table A all tables (B,C,D) with FK pointing to this table (A) must be scanned. 
If there is no index on those tables it means we gone do all Sequantial scans. Than can cause significant performance problem!!!.

Is there a reason why implicit index aren't created when FK are declared.  I looked into the documentation and I haven't found a way to tell postgresql to automatically create an index when creating la FK.  Does it means I need to manage it EXPLICITLY with create index statement ?  Is there another way ?

Thanks for you help that simple answer will solve a lot of performance problem I have ...

/David


On Mon, 4 Jul 2005, David Gagnon wrote:

  
  If you can just help my understanding the choice of the planner.

Here is the Query:
 explain analyse SELECT IRNUM FROM IR
        INNER JOIN IT ON  IT.ITIRNUM = ANY ('{1000, 2000}') AND
IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM
        WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M'

Here is the Query plan:

QUERY PLAN

Hash Join  (cost=1142.47..5581.75 rows=87 width=4) (actual
time=125.000..203.000 rows=2 loops=1)
  Hash Cond: ("outer".itirnum = "inner".irnum)
  ->  Seq Scan on it  (cost=0.00..3093.45 rows=31646 width=9) (actual
time=0.000..78.000 rows=2 loops=1)
        Filter: ((itirnum = ANY ('{1000,2000}'::integer[])) AND
((itypnum)::text = 'M'::text))

  ->  Hash  (cost=1142.09..1142.09 rows=151 width=37) (actual
time=125.000..125.000 rows=0 loops=1)
        ->  Index Scan using ir_pk on ir  (cost=0.00..1142.09 rows=151
width=37) (actual time=0.000..125.000 rows=2 loops=1)
              Index Cond: ((irypnum)::text = 'M'::text)
              Filter: (irnum = ANY ('{1000,2000}'::integer[]))
Total runtime: 203.000 ms
    

  
    I don't understand why the planner do a Seq Scan (Seq Scan on table
IT ..) instead of passing by the followin index:
    ALTER TABLE IT ADD CONSTRAINT IT_IR_FK foreign key (ITYPNUM,ITIRNUM)
references IR (IRYPNUM, IRNUM) ON UPDATE CASCADE;
    

That doesn't create an index on IT.  Primary keys (and unique constraints)
create indexes, but not foreign keys.  Did you also create an index on
those fields?

Also it looks like it's way overestimating the number of rows that
condition would succeed for.  You might consider raising the statistics
targets on those columns and reanalyzing.
--Boundary_(ID_/5SMBe2bXvLrkUkYZhz9qA)-- From pgsql-performance-owner@postgresql.org Mon Jul 4 22:12:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E81DD5284F for ; Mon, 4 Jul 2005 22:12:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 47336-04 for ; Tue, 5 Jul 2005 01:12:46 +0000 (GMT) Received: from houston.familyhealth.com.au (houston.au.fhnetwork.com [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 82CC25282B for ; Mon, 4 Jul 2005 22:12:44 -0300 (ADT) Received: from houston.familyhealth.com.au (localhost [127.0.0.1]) by houston.familyhealth.com.au (Postfix) with ESMTP id 9A78724FDD; Tue, 5 Jul 2005 09:12:45 +0800 (WST) Received: from [127.0.0.1] (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (Postfix) with ESMTP id 9867E24FCE; Tue, 5 Jul 2005 09:12:45 +0800 (WST) Message-ID: <42C9DEBA.4000905@familyhealth.com.au> Date: Tue, 05 Jul 2005 09:13:30 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: David Gagnon Cc: Stephan Szabo , pgsql-performance@postgresql.org Subject: Re: Why the planner is not using the INDEX . References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> <20050704132446.O14579@megazone.bigpanda.com> <42C9D47E.7090103@siunik.com> In-Reply-To: <42C9D47E.7090103@siunik.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.063 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/35 X-Sequence-Number: 13276 > I'm a bit surprised of that behavior thought, since it means that if we > delete a row from table A all tables (B,C,D) with FK pointing to this > table (A) must be scanned. > If there is no index on those tables it means we gone do all Sequantial > scans. Than can cause significant performance problem!!!. Correct. > Is there a reason why implicit index aren't created when FK are > declared. Because it's not a requirement... > I looked into the documentation and I haven't found a way to > tell postgresql to automatically create an index when creating la FK. > Does it means I need to manage it EXPLICITLY with create index statement > ? Is there another way ? No other way - you need to explicitly create them. It's not that hard either to write a query to search the system catalogs for unindexed FK's. Chris From pgsql-performance-owner@postgresql.org Tue Jul 5 09:32:13 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3464A5281E for ; Tue, 5 Jul 2005 09:32:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 81210-05 for ; Tue, 5 Jul 2005 12:32:05 +0000 (GMT) Received: from wolff.to (wolff.to [66.93.197.194]) by svr1.postgresql.org (Postfix) with SMTP id 8E15352814 for ; Tue, 5 Jul 2005 09:32:03 -0300 (ADT) Received: (qmail 11854 invoked by uid 500); 5 Jul 2005 12:32:17 -0000 Date: Tue, 5 Jul 2005 07:32:17 -0500 From: Bruno Wolff III To: David Gagnon Cc: Stephan Szabo , pgsql-performance@postgresql.org Subject: Re: Why the planner is not using the INDEX . Message-ID: <20050705123217.GA11379@wolff.to> Mail-Followup-To: Bruno Wolff III , David Gagnon , Stephan Szabo , pgsql-performance@postgresql.org References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> <20050704132446.O14579@megazone.bigpanda.com> <42C9D47E.7090103@siunik.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42C9D47E.7090103@siunik.com> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.007 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/36 X-Sequence-Number: 13277 On Mon, Jul 04, 2005 at 20:29:50 -0400, David Gagnon wrote: > Thanks .. I miss that FK don't create indexed ... since Primary key > implicitly does ... > > I'm a bit surprised of that behavior thought, since it means that if we > delete a row from table A all tables (B,C,D) with FK pointing to this > table (A) must be scanned. But in some applications you don't ever do that, so you don't save anything by having the index for deletes but have to pay the cost to update it when modifying the referencing table. If you think an index will help in your case, just create one. From pgsql-performance-owner@postgresql.org Tue Jul 5 11:02:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F214852969 for ; Tue, 5 Jul 2005 11:02:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 02049-06 for ; Tue, 5 Jul 2005 14:02:03 +0000 (GMT) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id F0BBE52968 for ; Tue, 5 Jul 2005 11:02:02 -0300 (ADT) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 19C143541C; Tue, 5 Jul 2005 07:02:07 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 182F235408; Tue, 5 Jul 2005 07:02:07 -0700 (PDT) Date: Tue, 5 Jul 2005 07:02:07 -0700 (PDT) From: Stephan Szabo To: David Gagnon Cc: pgsql-performance@postgresql.org Subject: Re: Why the planner is not using the INDEX . In-Reply-To: <42C9D47E.7090103@siunik.com> Message-ID: <20050705065351.W63027@megazone.bigpanda.com> References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> <20050704132446.O14579@megazone.bigpanda.com> <42C9D47E.7090103@siunik.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/37 X-Sequence-Number: 13278 On Mon, 4 Jul 2005, David Gagnon wrote: > Thanks .. I miss that FK don't create indexed ... since Primary key > implicitly does ... > > I'm a bit surprised of that behavior thought, since it means that if we > delete a row from table A all tables (B,C,D) with FK pointing to this > table (A) must be scanned. > If there is no index on those tables it means we gone do all Sequantial > scans. Than can cause significant performance problem!!!. > > Is there a reason why implicit index aren't created when FK are > declared. I looked into the documentation and I haven't found a way to The reason is that it's not always useful to have an index for that purpose. You could either have low selectivity (in which case the index wouldn't be used) or low/batch changes to the referenced table (in which case the cost of maintaining the index may be greater than the value of having the index) or other such cases. In primary key and unique, we currently have no choice but to make an index because that's how the constraint is currently implemented. > tell postgresql to automatically create an index when creating la FK. > Does it means I need to manage it EXPLICITLY with create index statement > ? Yeah. >Is there another way ? Not that I can think of without changing the source. From pgsql-performance-owner@postgresql.org Tue Jul 5 13:10:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6D8E45287F for ; Tue, 5 Jul 2005 13:10:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31547-02 for ; Tue, 5 Jul 2005 16:09:58 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 8314152814 for ; Tue, 5 Jul 2005 13:09:57 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j65G9uL8023493; Tue, 5 Jul 2005 12:09:56 -0400 (EDT) To: David Gagnon Cc: pgsql-performance@postgresql.org Subject: Re: Why the planner is not using the INDEX . In-reply-to: <42C994BD.3000909@siunik.com> References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> Comments: In-reply-to David Gagnon message dated "Mon, 04 Jul 2005 15:57:49 -0400" Date: Tue, 05 Jul 2005 12:09:56 -0400 Message-ID: <23492.1120579796@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.109 tagged_above=0 required=5 tests=AWL, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/38 X-Sequence-Number: 13279 David Gagnon writes: > explain analyse SELECT IRNUM FROM IR > INNER JOIN IT ON IT.ITIRNUM = ANY ('{1000, 2000}') AND > IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM > WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M' Those =ANY constructs are not currently optimizable at all. You might get better results with "IT.ITIRNUM IN (1000, 2000)" etc. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 5 13:15:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DB2C752808 for ; Tue, 5 Jul 2005 13:15:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28549-09 for ; Tue, 5 Jul 2005 16:15:05 +0000 (GMT) Received: from jupiter.egsgroup.com (jupiter.egsgroup.com [217.199.182.198]) by svr1.postgresql.org (Postfix) with ESMTP id 495B252828 for ; Tue, 5 Jul 2005 13:15:02 -0300 (ADT) Received: by jupiter.egsgroup.com (Postfix, from userid 65) id A230F178D3; Tue, 5 Jul 2005 17:15:00 +0100 (BST) X-Scanned-By: AMaViS-ng at egsgroup.com Received: from [10.20.0.81] (unknown [217.150.127.117]) by jupiter.egsgroup.com (Postfix) with ESMTP id AA48D1789A; Tue, 5 Jul 2005 17:14:58 +0100 (BST) Message-ID: <42CAB1B6.50201@egsgroup.com> Date: Tue, 05 Jul 2005 17:13:42 +0100 From: Alexander Stanier User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Heavy virtual memory usage on production system Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/39 X-Sequence-Number: 13280 We are having terrible performance issues with a production instance of PostgreSQL version 7.4.5, but are struggling with which parameters in the postgresql.conf to change. Our database server is an Apple G5 (2 x 2GHz CPU, 2GB RAM). The operating system is Mac OS X 10.3. The database seems to fine to start with, but then as the load increases it seems to reach a threshold where the number of non-idle queries in pg_stat_activity grows heavily and we appear to get something similar to a motorway tail back with up to perhaps 140 queries awaiting processing. At the same time the virtual memory usage (reported by the OS) appears to grow heavily too (sometimes up to 50GB). The CPUs do not seems to be working overly hard nor do the disks and the memory monitor reports about 600MB of inactive memory. Once in this situation, the database never catches up with itself and the only way to put it back on an even keel is to stop the application and restart database. The system memory settings are: kern.sysv.shmmax: 536870912 kern.sysv.shmmin: 1 kern.sysv.shmmni: 4096 kern.sysv.shmseg: 4096 kern.sysv.shmall: 131072 We have unlimited the number of processes and open files for the user running PostgreSQL (therefore max 2048 processes and max 12288 open files). Non default postgresql parameters are: tcpip_socket = true max_connections = 500 unix_socket_directory = '/Local/PostgreSQL' shared_buffers = 8192 # min 16, at least max_connections*2, 8KB each sort_mem = 2048 # min 64, size in KB wal_buffers = 32 # min 4, 8KB each effective_cache_size = 100000 # typically 8KB each random_page_cost = 2 # units are one sequential page fetch cost log_min_error_statement = info # Values in order of increasing severity: log_duration = true log_pid = true log_statement = true log_timestamp = true stats_command_string = true although on the last restart I changed the following (since the current config clearly isn't working): shared_buffers = 16384 # min 16, at least max_connections*2, 8KB each effective_cache_size = 10000 # typically 8KB each We don't know whether these have helped yet - but we should get a good idea around 10am tomorrow morning. We currently have the application limited to a maximum of 40 concurrent connections to the database. Our application produces a fairly varied mix of queries, some quite complex and plenty of them. We seem to average about 400,000 queries per hour. At first I thought it might be one or two inefficient queries blocking the CPUs but the CPUs don't seem to be very stretched. My guess is that we have our postgresql memory settings wrong, however, the is lots of conflicting advice about what to set (from 1000 to 100000 shared buffers). Does this heavy use of VM and query tail back indicate which memory settings are wrong? Presumably if there are 140 queries in pg_stat_activity then postgresql will be trying to service all these queries at once? I also presume that if VM usage is high then we are paging a vast amount to disk. But I am not sure why. Has anyone seen this behaviour before and can anyone point me in the right direction? Regards, Alexander Stanier From pgsql-performance-owner@postgresql.org Tue Jul 5 13:55:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AE77B5286E for ; Tue, 5 Jul 2005 13:55:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 39778-01 for ; Tue, 5 Jul 2005 16:55:21 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 1CC935281B for ; Tue, 5 Jul 2005 13:55:17 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j65GtGNS023890; Tue, 5 Jul 2005 12:55:16 -0400 (EDT) To: Alexander Stanier Cc: pgsql-performance@postgresql.org Subject: Re: Heavy virtual memory usage on production system In-reply-to: <42CAB1B6.50201@egsgroup.com> References: <42CAB1B6.50201@egsgroup.com> Comments: In-reply-to Alexander Stanier message dated "Tue, 05 Jul 2005 17:13:42 +0100" Date: Tue, 05 Jul 2005 12:55:16 -0400 Message-ID: <23889.1120582516@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/40 X-Sequence-Number: 13281 Alexander Stanier writes: > The database seems to fine to start with, but then as the load increases > it seems to reach a threshold where the number of non-idle queries in > pg_stat_activity grows heavily and we appear to get something similar to > a motorway tail back with up to perhaps 140 queries awaiting processing. > At the same time the virtual memory usage (reported by the OS) appears > to grow heavily too (sometimes up to 50GB). The CPUs do not seems to be > working overly hard nor do the disks and the memory monitor reports > about 600MB of inactive memory. You shouldn't be putting a lot of credence in the virtual memory usage then, methinks. Some versions of top count the Postgres shared memory against *each* backend process, leading to a wildly inflated figure for total memory used. I'd suggest watching the output of "vmstat 1" (or local equivalent) to observe whether there's any significant amount of swapping going on; if not, excessive memory usage isn't the problem. Are you sure that the problem isn't at the level of some query taking an exclusive lock and then sitting on it? I would expect either CPU or disk bandwidth or both to be saturated if you were having a conventional resource limitation problem. Again, comparing vmstat readings during normal and slow response conditions would be instructive. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 5 14:49:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9D9175281B for ; Tue, 5 Jul 2005 14:49:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51716-02 for ; Tue, 5 Jul 2005 17:49:30 +0000 (GMT) Received: from relais.videotron.ca (relais.videotron.ca [24.201.245.36]) by svr1.postgresql.org (Postfix) with ESMTP id BFB7752814 for ; Tue, 5 Jul 2005 14:49:23 -0300 (ADT) Received: from [192.168.1.101] ([24.202.23.128]) by VL-MO-MR007.ip.videotron.ca (iPlanet Messaging Server 5.2 HotFix 1.21 (built Sep 8 2003)) with ESMTP id <0IJ600A931G2MU@VL-MO-MR007.ip.videotron.ca> for pgsql-performance@postgresql.org; Tue, 05 Jul 2005 13:48:02 -0400 (EDT) Date: Tue, 05 Jul 2005 13:53:59 -0400 From: David Gagnon Subject: Re: Why the planner is not using the INDEX . In-reply-to: <23492.1120579796@sss.pgh.pa.us> To: Tom Lane Cc: pgsql-performance@postgresql.org Message-id: <42CAC937.4080702@siunik.com> MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7BIT X-Accept-Language: en-us, en User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> <23492.1120579796@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.126 tagged_above=0 required=5 tests=AWL, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/41 X-Sequence-Number: 13282 Tom Lane wrote: >David Gagnon writes: > > >> explain analyse SELECT IRNUM FROM IR >> INNER JOIN IT ON IT.ITIRNUM = ANY ('{1000, 2000}') AND >>IT.ITYPNUM = 'M' AND IR.IRYPNUM = IT.ITYPNUM AND IR.IRNUM = IT.ITIRNUM >> WHERE IRNUM = ANY ('{1000, 2000}') and IRYPNUM = 'M' >> >> > >Those =ANY constructs are not currently optimizable at all. You might >get better results with "IT.ITIRNUM IN (1000, 2000)" etc. > > regards, tom lane > > > I already tried this construct. But the statement comes from a stored procedure where the {1000, 2000} is an array variable (requestIds). I tried to use IT.ITIRNUM IN (requestIds) or several other variant without success. Is there a way to make it work? Here is the statement the statement from the store procedure. Remenber requestIds is an array of int. FOR inventoryTransaction IN SELECT DISTINCT IRNUM, IRAENUM, IRSTATUT, IRSENS, IRSOURCE, IRDATE, IRQTE FROM IR WHERE IRNUM = ANY (requestIds) and IRYPNUM = companyId LOOP Thank for your help !!!! /David From pgsql-performance-owner@postgresql.org Tue Jul 5 18:31:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3EFDC52823 for ; Tue, 5 Jul 2005 18:31:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 05167-06 for ; Tue, 5 Jul 2005 21:31:14 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id C99FD52824 for ; Tue, 5 Jul 2005 18:31:13 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7574036; Tue, 05 Jul 2005 14:33:24 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: emil@baymountain.com Subject: Re: Planner constants for RAM resident databases Date: Tue, 5 Jul 2005 14:33:28 -0700 User-Agent: KMail/1.8 Cc: John A Meinel , Postgresql Performance References: <200507012159.39041.emil@baymountain.com> <42C606AA.8060406@arbash-meinel.com> <200507020944.07339.emil@baymountain.com> In-Reply-To: <200507020944.07339.emil@baymountain.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507051433.28971.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/42 X-Sequence-Number: 13283 Emil, > -> Merge Left Join (cost=9707.71..13993.52 rows=1276 width=161) > (actual time=164.423..361.477 rows=49 loops=1) That would indicate that you need to either increase your statistical sampling (SET STATISTICS) or your frequency of running ANALYZE, or both. -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Wed Jul 6 09:04:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 811F352838 for ; Wed, 6 Jul 2005 09:04:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68134-03 for ; Wed, 6 Jul 2005 12:04:48 +0000 (GMT) Received: from jupiter.egsgroup.com (jupiter.egsgroup.com [217.199.182.198]) by svr1.postgresql.org (Postfix) with ESMTP id 1975F52808 for ; Wed, 6 Jul 2005 09:04:47 -0300 (ADT) Received: by jupiter.egsgroup.com (Postfix, from userid 65) id AC90F1784D; Wed, 6 Jul 2005 13:04:49 +0100 (BST) X-Scanned-By: AMaViS-ng at egsgroup.com Received: from [10.20.0.81] (unknown [217.150.127.117]) by jupiter.egsgroup.com (Postfix) with ESMTP id 8E2ED643B; Wed, 6 Jul 2005 13:04:47 +0100 (BST) Message-ID: <42CBC8A0.4070901@egsgroup.com> Date: Wed, 06 Jul 2005 13:03:44 +0100 From: Alexander Stanier User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Heavy virtual memory usage on production system References: <42CAB1B6.50201@egsgroup.com> <23889.1120582516@sss.pgh.pa.us> In-Reply-To: <23889.1120582516@sss.pgh.pa.us> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/43 X-Sequence-Number: 13284 The problem happened again this morning and I took the chance to check out the locking situation. The number of locks increased dramatically up to over 1000, but they were all "AccessShareLocks" and all were granted. The odd "RowExclusiveLock" appeared but none persisted. On the basis that nothing seems to be waiting for a lock, I don't think it is a locking problem. I think the vast number of locks is symptom of the fact that the server is trying to service a vast number of requests. Eventually, the server seemed to catch up with itself - the CPU went up, the VM went down and the number of queries in pg_stat_activity reduced. The problem then occurred a second time and there seemed to be a lot of pageouts and pageins going on, but I was only looking at top so it was difficult to tell. I have now restarted with a statement_timeout of 2 mins to protect the server from poorly performing queries (fairly brutal - but it does at least stop the downward spiral). I have also reduced the sort_mem to 1024. I guess it could be that we simply need more memory in the server. I have got vmstat (vm_stat on Mac) running and I will watch the behaviour...... Regards, Alex Stanier. Tom Lane wrote: >Alexander Stanier writes: > > >>The database seems to fine to start with, but then as the load increases >>it seems to reach a threshold where the number of non-idle queries in >>pg_stat_activity grows heavily and we appear to get something similar to >>a motorway tail back with up to perhaps 140 queries awaiting processing. >>At the same time the virtual memory usage (reported by the OS) appears >>to grow heavily too (sometimes up to 50GB). The CPUs do not seems to be >>working overly hard nor do the disks and the memory monitor reports >>about 600MB of inactive memory. >> >> > >You shouldn't be putting a lot of credence in the virtual memory usage >then, methinks. Some versions of top count the Postgres shared memory >against *each* backend process, leading to a wildly inflated figure for >total memory used. I'd suggest watching the output of "vmstat 1" (or >local equivalent) to observe whether there's any significant amount of >swapping going on; if not, excessive memory usage isn't the problem. > >Are you sure that the problem isn't at the level of some query taking an >exclusive lock and then sitting on it? I would expect either CPU or >disk bandwidth or both to be saturated if you were having a conventional >resource limitation problem. Again, comparing vmstat readings during >normal and slow response conditions would be instructive. > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Wed Jul 6 11:15:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6A1F152809 for ; Wed, 6 Jul 2005 11:15:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 89436-06 for ; Wed, 6 Jul 2005 14:15:34 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E8B2B5280B for ; Wed, 6 Jul 2005 11:15:32 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j66EFXjV025887; Wed, 6 Jul 2005 10:15:33 -0400 (EDT) To: Alexander Stanier Cc: pgsql-performance@postgresql.org Subject: Re: Heavy virtual memory usage on production system In-reply-to: <42CBC8A0.4070901@egsgroup.com> References: <42CAB1B6.50201@egsgroup.com> <23889.1120582516@sss.pgh.pa.us> <42CBC8A0.4070901@egsgroup.com> Comments: In-reply-to Alexander Stanier message dated "Wed, 06 Jul 2005 13:03:44 +0100" Date: Wed, 06 Jul 2005 10:15:33 -0400 Message-ID: <25886.1120659333@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/44 X-Sequence-Number: 13285 Alexander Stanier writes: > The problem happened again this morning and I took the chance to check > out the locking situation. The number of locks increased dramatically up > to over 1000, but they were all "AccessShareLocks" and all were granted. > The odd "RowExclusiveLock" appeared but none persisted. On the basis > that nothing seems to be waiting for a lock, I don't think it is a > locking problem. Hmm. How many active processes were there, and how many locks per process? (A quick "SELECT pid, count(*) GROUP BY pid" query should give you this info next time.) We just recently got rid of some O(N^2) behavior in the lock manager for cases where a single backend holds many different locks. So if there's a single query acquiring a whole lot of locks, that could possibly have something to do with this. regards, tom lane From pgsql-performance-owner@postgresql.org Wed Jul 6 11:50:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D6F1D52993 for ; Wed, 6 Jul 2005 11:50:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95153-07 for ; Wed, 6 Jul 2005 14:50:42 +0000 (GMT) Received: from jupiter.egsgroup.com (jupiter.egsgroup.com [217.199.182.198]) by svr1.postgresql.org (Postfix) with ESMTP id D2CAD52968 for ; Wed, 6 Jul 2005 11:50:40 -0300 (ADT) Received: by jupiter.egsgroup.com (Postfix, from userid 65) id 5F08E178CC; Wed, 6 Jul 2005 15:50:45 +0100 (BST) X-Scanned-By: AMaViS-ng at egsgroup.com Received: from [10.20.0.81] (unknown [217.150.127.117]) by jupiter.egsgroup.com (Postfix) with ESMTP id 6B795177BD; Wed, 6 Jul 2005 15:50:43 +0100 (BST) Message-ID: <42CBEF84.8080007@egsgroup.com> Date: Wed, 06 Jul 2005 15:49:40 +0100 From: Alexander Stanier User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Heavy virtual memory usage on production system References: <42CAB1B6.50201@egsgroup.com> <23889.1120582516@sss.pgh.pa.us> <42CBC8A0.4070901@egsgroup.com> <25886.1120659333@sss.pgh.pa.us> In-Reply-To: <25886.1120659333@sss.pgh.pa.us> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/45 X-Sequence-Number: 13286 Looks as though there are several processes which are acquiring a load of locks: pid | count ------+------- 3193 | 2 3192 | 9 3191 | 7 3190 | 3 3189 | 2 3188 | 3 3187 | 3 3186 | 3 3185 | 3 3184 | 3 3183 | 3 3182 | 13 3181 | 3 3179 | 10 3175 | 13 3174 | 2 3173 | 10 2917 | 3 3153 | 8 3150 | 8 3149 | 8 3146 | 9 3145 | 8 3144 | 8 3143 | 9 3142 | 3 3141 | 10 3127 | 8 3125 | 13 3124 | 13 3121 | 8 3118 | 8 3114 | 8 3113 | 8 3110 | 8 3106 | 8 3104 | 9 3102 | 8 3100 | 13 2314 | 2 (40 rows) I guess it might be worth us getting this server up to PostgreSQL 8.0.3. At least we can then discount that as a problem. Regards, Alex Stanier. Tom Lane wrote: >Alexander Stanier writes: > > >>The problem happened again this morning and I took the chance to check >>out the locking situation. The number of locks increased dramatically up >>to over 1000, but they were all "AccessShareLocks" and all were granted. >>The odd "RowExclusiveLock" appeared but none persisted. On the basis >>that nothing seems to be waiting for a lock, I don't think it is a >>locking problem. >> >> > >Hmm. How many active processes were there, and how many locks per >process? (A quick "SELECT pid, count(*) GROUP BY pid" query should give >you this info next time.) We just recently got rid of some O(N^2) >behavior in the lock manager for cases where a single backend holds many >different locks. So if there's a single query acquiring a whole lot of >locks, that could possibly have something to do with this. > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Wed Jul 6 12:06:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EFD2452987 for ; Wed, 6 Jul 2005 12:06:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95867-10 for ; Wed, 6 Jul 2005 15:06:05 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 9818752984 for ; Wed, 6 Jul 2005 12:06:04 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j66F628w026527; Wed, 6 Jul 2005 11:06:02 -0400 (EDT) To: Alexander Stanier Cc: pgsql-performance@postgresql.org Subject: Re: Heavy virtual memory usage on production system In-reply-to: <42CBEF84.8080007@egsgroup.com> References: <42CAB1B6.50201@egsgroup.com> <23889.1120582516@sss.pgh.pa.us> <42CBC8A0.4070901@egsgroup.com> <25886.1120659333@sss.pgh.pa.us> <42CBEF84.8080007@egsgroup.com> Comments: In-reply-to Alexander Stanier message dated "Wed, 06 Jul 2005 15:49:40 +0100" Date: Wed, 06 Jul 2005 11:06:02 -0400 Message-ID: <26526.1120662362@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/46 X-Sequence-Number: 13287 Alexander Stanier writes: > Looks as though there are several processes which are acquiring a load > of locks: 13 locks isn't "a load". I was worried about scenarios in which a single process might take hundreds or thousands of locks; it doesn't look like you have that. regards, tom lane From pgsql-performance-owner@postgresql.org Wed Jul 6 14:15:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 817D2528B9 for ; Wed, 6 Jul 2005 14:15:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 30927-10 for ; Wed, 6 Jul 2005 17:15:49 +0000 (GMT) Received: from lon-mail-1.gradwell.net (lon-mail-1.gradwell.net [193.111.201.125]) by svr1.postgresql.org (Postfix) with ESMTP id DC1935289B for ; Wed, 6 Jul 2005 14:15:47 -0300 (ADT) Received: from www.gradwell.com ([193.111.200.100]) by lon-mail-1.gradwell.net with smtp (Gradwell gwh-smtpd 1.186) id 42cc11c3.f44.c7; Wed, 6 Jul 2005 18:15:47 +0100 (envelope-sender ) Received: from 217.45.209.171 (SquirrelMail authenticated user paul@pop3.oxton.com) by www.gradwell.com with HTTP; Wed, 6 Jul 2005 18:15:47 +0100 (BST) Message-ID: <3688.217.45.209.171.1120670147.squirrel@www.gradwell.com> Date: Wed, 6 Jul 2005 18:15:47 +0100 (BST) Subject: Data Warehousing Tuning From: "Paul Johnson" To: pgsql-performance@postgresql.org Reply-To: paul@oxton.com User-Agent: SquirrelMail/1.4.2 MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 Importance: Normal X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/47 X-Sequence-Number: 13288 Hi all, we have the following setup: - Sun V250 server - 2*1.3GHz Sparc IIIi CPU - 8GB RAM - 8*73GB SCSI drives - Solaris 10 - Postgres 8 Disks 0 and 1 are mirrored and contain the OS and the various software packages, disks 2-7 are configured as a 320GB concatenation mounted on /data, which is where load files and Postgres database and log files live. The box is used by a small number of developers doing solely Postgres-based data warehousing work. There are no end-users on the box, and we are aiming for the maximum IO throughput. Questions are as follows: 1) Should we have set the page size to 32MB when we compiled Postgres? We mainly do bulk loads using 'copy', full-table scans and large joins so this would seem sensible. Tables are typically 10 million rows at present. 2) What are the obvious changes to make to postgresql.conf? Things like shared_buffers, work_mem, maintenance_work_mem and checkpoint_segments seem like good candidates given our data warehousing workloads. 3) Ditto /etc/system? 4) We moved the pg_xlog files off /data/postgres (disks 2-7) and into /opt/pg_xlog (disks 0-1), but it seemed like performance decreased, so we moved them back again. Has anyone experienced real performance gains by moving the pg_xlog files? Thanks in anticipation, Paul. From pgsql-performance-owner@postgresql.org Wed Jul 6 14:58:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 33995528A6 for ; Wed, 6 Jul 2005 14:58:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 41475-08 for ; Wed, 6 Jul 2005 17:58:42 +0000 (GMT) Received: from pelego.atua.com.br (unknown [200.248.138.60]) by svr1.postgresql.org (Postfix) with ESMTP id A021952886 for ; Wed, 6 Jul 2005 14:58:41 -0300 (ADT) Received: from [10.0.0.171] (unknown [10.0.0.171]) by pelego.atua.com.br (Postfix) with ESMTP id 71D9313F79F for ; Wed, 6 Jul 2005 14:58:31 -0300 (BRT) Message-ID: <42CC1BC7.7060902@atua.com.br> Date: Wed, 06 Jul 2005 14:58:31 -0300 From: Alvaro Nunes Melo User-Agent: Debian Thunderbird 1.0 (X11/20050116) X-Accept-Language: en-us, en MIME-Version: 1.0 To: PostgreSQL - Performance Subject: Storing data and indexes in different disks Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.032 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/48 X-Sequence-Number: 13289 Hi, I'm about to recommend a server model to a client, and I've read a in many places (including in this list) that storing indexes in one disk and the rest of the database in other disk might increase the overall performance of the system in about 10%. Making this only by a symbolic link is enough, or there are any futher steps? We were thinking (for costs reasons) in use only 2 SCSI disks, without any level of RAID. Is this enough to achieve performance improvement mentioned above? Best regards, Alvaro From pgsql-performance-owner@postgresql.org Wed Jul 6 15:44:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2CD2C5293D for ; Wed, 6 Jul 2005 15:44:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51344-03 for ; Wed, 6 Jul 2005 18:44:22 +0000 (GMT) Received: from mail.aveo.aveopharma.com (67.109.105.227.ptr.us.xo.net [67.109.105.227]) by svr1.postgresql.org (Postfix) with SMTP id D28C4528E7 for ; Wed, 6 Jul 2005 15:44:21 -0300 (ADT) X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: Re: Storing data and indexes in different disks Date: Wed, 6 Jul 2005 14:46:03 -0400 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Storing data and indexes in different disks Thread-Index: AcWCVSB8qsP2PWPnToacjdApsfHK5gAA/3TA From: "Dmitri Bichko" To: "Alvaro Nunes Melo" , "PostgreSQL - Performance" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.064 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/49 X-Sequence-Number: 13290 I'd say it's a little early to worry about a 10% performance increase when you don't have any redundancy. You might want to consider using more, cheaper SATA disks - with more spindles you may very well get better performance in addition to redundancy. Anyway, here's an optimization project I just went through recently: the old database was running all on a SAN attached RAID5 partition; moved the indices to a local striped mirror set (faster disks too: 15K rpm), moved the WAL files to a separate local two disk mirror, and spent a lot of time tuning the config parameters (the old install was running the conservative defaults). All the hardware (apart from the additional disks) is the same. For some simple queries I saw as much as a 150X speedup, though that's certainly not typical of the performance improvement overall. Most of this is likely due to the memory settings, but faster disks certainly play a part. In any case, it's hard to say what would improve performance for you without knowing what kind of applications you are running and what sort of load they see. Dmitri -----Original Message----- From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of Alvaro Nunes Melo Sent: Wednesday, July 06, 2005 1:59 PM To: PostgreSQL - Performance Subject: [PERFORM] Storing data and indexes in different disks Hi, I'm about to recommend a server model to a client, and I've read a in=20 many places (including in this list) that storing indexes in one disk=20= and the rest of the database in other disk might increase the overall=20 performance of the system in about 10%. Making this only by a symbolic link is enough, or there are any futher=20= steps? We were thinking (for costs reasons) in use only 2 SCSI disks,=20 without any level of RAID. Is this enough to achieve performance=20 improvement mentioned above? Best regards, Alvaro ---------------------------(end of broadcast)--------------------------- TIP 7: don't forget to increase your free space map settings The information transmitted is intended only for the person or entity to = which it is addressed and may contain confidential and/or privileged mate= rial. Any review, retransmission, dissemination or other use of, or takin= g of any action in reliance upon, this information by persons or entities= other than the intended recipient is prohibited. If you received this in= error, please contact the sender and delete the material from any comput= er From pgsql-performance-owner@postgresql.org Wed Jul 6 16:49:37 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 302F6529A0 for ; Wed, 6 Jul 2005 16:49:36 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65543-01 for ; Wed, 6 Jul 2005 19:49:34 +0000 (GMT) Received: from netbox.unitech.com.ar (unknown [200.32.92.34]) by svr1.postgresql.org (Postfix) with ESMTP id C2B485299F for ; Wed, 6 Jul 2005 16:49:28 -0300 (ADT) Received: from dariop (dariop.unitech.com.ar [192.168.1.237]) by netbox.unitech.com.ar (8.11.6/8.11.6) with ESMTP id j66Kp0E27518 for ; Wed, 6 Jul 2005 16:51:00 -0400 Received: from 127.0.0.1 (AVG SMTP 7.0.323 [267.8.9]); Wed, 06 Jul 2005 16:49:21 -0300 From: "Dario" To: Subject: ALTER TABLE tabla ALTER COLUMN columna SET STATISTICS number Date: Wed, 6 Jul 2005 16:49:21 -0300 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4952.2800 Importance: Normal X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.174 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/50 X-Sequence-Number: 13291 �where is stored the value set by ALTER TABLE table_name ALTER COLUMN column_name SET STATISTICS = [1-1000]? I've set this to 1000, and I didn't remember in which column (doh!). Is there any table to look? (I did 'grep "set stat" $PGDATA/pg_log/*' and found it, but may be there is a better way) I couldn't find it in the docs neithr "googling" Greetings -------------------------------------- Long life, little spam and prosperity From pgsql-performance-owner@postgresql.org Wed Jul 6 17:08:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DA717529BA for ; Wed, 6 Jul 2005 17:08:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 70229-02 for ; Wed, 6 Jul 2005 20:07:56 +0000 (GMT) Received: from tigger.fuhr.org (tigger.fuhr.org [63.214.45.158]) by svr1.postgresql.org (Postfix) with ESMTP id A6564529B5 for ; Wed, 6 Jul 2005 17:07:54 -0300 (ADT) Received: from winnie.fuhr.org (winnie.fuhr.org [10.1.0.1]) by tigger.fuhr.org (8.13.3/8.13.3) with ESMTP id j66K7rA7096698 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Wed, 6 Jul 2005 14:07:55 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: from winnie.fuhr.org (localhost [127.0.0.1]) by winnie.fuhr.org (8.13.3/8.13.3) with ESMTP id j66K7q1p067069; Wed, 6 Jul 2005 14:07:52 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: (from mfuhr@localhost) by winnie.fuhr.org (8.13.3/8.13.3/Submit) id j66K7qg1067060; Wed, 6 Jul 2005 14:07:52 -0600 (MDT) (envelope-from mfuhr) Date: Wed, 6 Jul 2005 14:07:52 -0600 From: Michael Fuhr To: Dario Cc: pgsql-performance@postgresql.org Subject: Re: ALTER TABLE tabla ALTER COLUMN columna SET STATISTICS number Message-ID: <20050706200752.GA33295@winnie.fuhr.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.109 tagged_above=0 required=5 tests=AWL, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/51 X-Sequence-Number: 13292 On Wed, Jul 06, 2005 at 04:49:21PM -0300, Dario wrote: > where is stored the value set by ALTER TABLE table_name ALTER COLUMN > column_name SET STATISTICS = [1-1000]? pg_attribute.attstattarget Example query: SELECT attrelid::regclass, attname, attstattarget FROM pg_attribute WHERE attstattarget > 0; See the "System Catalogs" chapter in the documentation for more information. -- Michael Fuhr http://www.fuhr.org/~mfuhr/ From pgsql-performance-owner@postgresql.org Wed Jul 6 18:19:54 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9EB88529A4 for ; Wed, 6 Jul 2005 18:19:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 82655-04 for ; Wed, 6 Jul 2005 21:19:49 +0000 (GMT) Received: from dns.texnet.it (dns.texnet.it [217.19.150.6]) by svr1.postgresql.org (Postfix) with ESMTP id BD3B452998 for ; Wed, 6 Jul 2005 18:19:47 -0300 (ADT) Received: from pool-a-150.texnet.it ([217.19.150.150] helo=paros.rigacci.org) by dns.texnet.it with esmtp (Exim 3.35 #1 (Debian)) id 1DqHJ4-000541-00 for ; Wed, 06 Jul 2005 23:19:47 +0200 Received: from niccolo by paros.rigacci.org with local (Exim 4.50) id 1DqHJ4-0000tt-8b for pgsql-performance@postgresql.org; Wed, 06 Jul 2005 23:19:46 +0200 Date: Wed, 6 Jul 2005 23:19:46 +0200 To: pgsql-performance@postgresql.org Subject: CURSOR slowes down a WHERE clause 100 times? Message-ID: <20050706211946.GA3460@rigacci.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.9i From: Niccolo Rigacci X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/52 X-Sequence-Number: 13293 Hi to all, I have a performace problem with the following query: BEGIN; DECLARE mycursor BINARY CURSOR FOR SELECT toponimo, wpt FROM wpt_comuni_view WHERE ( wpt && setSRID('BOX3D(4.83 36, 20.16 47.5)'::BOX3D, 4326) ); FETCH ALL IN mycursor; END; I get the results in about 108 seconds (8060 rows). If I issue the SELECT alone (without the CURSOR) I get the same results in less than 1 second. The wpt_comuni_view is a VIEW of a 3 tables JOIN, and the "wpt" field is a PostGIS geometry column. The "&&" is the PostGIS "overlaps" operator. If I CURSOR SELECT from a temp table instead of the JOIN VIEW the query time 1 second. If I omit the WHERE clause the CURSOR fetches results in 1 second. Can the CURSOR on JOIN affects so heavly the WHERE clause? I suspect that - with the CURSOR - a sequential scan is performed on the entire data set for each fetched record... Any idea? This is the definition of the VIEW: CREATE VIEW wpt_comuni_view AS SELECT istat_wpt.oid, istat_wpt.id, istat_wpt.toponimo, istat_comuni.residenti, istat_wpt.wpt FROM istat_comuni JOIN istat_comuni2wpt USING (idprovincia, idcomune) JOIN istat_wpt ON (idwpt = id); Thank you for any hint. -- Niccolo Rigacci Firenze - Italy War against Iraq? Not in my name! From pgsql-performance-owner@postgresql.org Wed Jul 6 18:29:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 096DD5284F for ; Wed, 6 Jul 2005 18:29:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87165-02 for ; Wed, 6 Jul 2005 21:29:35 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id 8245E5285C for ; Wed, 6 Jul 2005 18:29:34 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050706212920m920060ieqe>; Wed, 6 Jul 2005 21:29:35 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 286DF56086; Wed, 6 Jul 2005 16:29:19 -0500 (CDT) Received: from [129.255.168.114] (host168-114.uihc.uiowa.edu [129.255.168.114]) by juju.arbash-meinel.com (Postfix) with ESMTP id E62835607F; Wed, 6 Jul 2005 16:29:07 -0500 (CDT) Message-ID: <42CC4D1C.9040606@arbash-meinel.com> Date: Wed, 06 Jul 2005 16:29:00 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Niccolo Rigacci , Postgresql Performance Subject: Re: CURSOR slowes down a WHERE clause 100 times? References: <20050706211946.GA3460@rigacci.org> In-Reply-To: <20050706211946.GA3460@rigacci.org> X-Enigmail-Version: 0.91.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig0BDC10F2B83E07547A560931" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/53 X-Sequence-Number: 13294 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig0BDC10F2B83E07547A560931 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Niccolo Rigacci wrote: >Hi to all, > >I have a performace problem with the following query: > > BEGIN; > DECLARE mycursor BINARY CURSOR FOR > SELECT > toponimo, > wpt > FROM wpt_comuni_view > WHERE ( > wpt && > setSRID('BOX3D(4.83 36, 20.16 47.5)'::BOX3D, 4326) > ); > FETCH ALL IN mycursor; > END; > >I get the results in about 108 seconds (8060 rows). > >If I issue the SELECT alone (without the CURSOR) I get the >same results in less than 1 second. > >The wpt_comuni_view is a VIEW of a 3 tables JOIN, and the "wpt" >field is a PostGIS geometry column. The "&&" is the PostGIS >"overlaps" operator. > >If I CURSOR SELECT from a temp table instead of the JOIN VIEW the >query time 1 second. > >If I omit the WHERE clause the CURSOR fetches results in 1 >second. > >Can the CURSOR on JOIN affects so heavly the WHERE clause? I >suspect that - with the CURSOR - a sequential scan is performed >on the entire data set for each fetched record... > >Any idea? > > What does it say if you do "EXPLAIN ANALYZE SELECT..." both with and without the cursor? It may not say much for the cursor, but I think you can explain analyze the fetch statements. It is my understanding that Cursors generally favor using an slow-startup style plan, which usually means using an index, because it expects that you won't actually want all of the data. A seqscan is not always slower, especially if you need to go through most of the data. Without an explain analyze it's hard to say what the planner is thinking and doing. >This is the definition of the VIEW: > > CREATE VIEW wpt_comuni_view AS > SELECT istat_wpt.oid, istat_wpt.id, istat_wpt.toponimo, > istat_comuni.residenti, istat_wpt.wpt > FROM istat_comuni > JOIN istat_comuni2wpt > USING (idprovincia, idcomune) > JOIN istat_wpt > ON (idwpt = id); > >Thank you for any hint. > > > You might also try comparing your CURSOR to a prepared statement. There are a few rare cases where preparing is worse than issuing the query directly, depending on your data layout. John =:-> --------------enig0BDC10F2B83E07547A560931 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCzE0fJdeBCYSNAAMRAk+eAJ96Sz+yQMX94j1ba77ojk021iA6IQCfYRvm ynbwpZw6yU9j96fu225RJvA= =DWez -----END PGP SIGNATURE----- --------------enig0BDC10F2B83E07547A560931-- From pgsql-performance-owner@postgresql.org Mon Jul 11 20:28:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 664EA529A3 for ; Wed, 6 Jul 2005 18:54:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91316-01 for ; Wed, 6 Jul 2005 21:54:11 +0000 (GMT) Received: from netbox.unitech.com.ar (unknown [200.32.92.34]) by svr1.postgresql.org (Postfix) with ESMTP id F1E1752995 for ; Wed, 6 Jul 2005 18:54:07 -0300 (ADT) Received: from dariop (dariop.unitech.com.ar [192.168.1.237]) by netbox.unitech.com.ar (8.11.6/8.11.6) with ESMTP id j66MtfE28801 for ; Wed, 6 Jul 2005 18:55:41 -0400 Received: from 127.0.0.1 (AVG SMTP 7.0.323 [267.8.9]); Wed, 06 Jul 2005 18:54:02 -0300 From: "Dario Pudlo" To: Subject: join and query planner Date: Wed, 6 Jul 2005 18:54:02 -0300 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4952.2800 In-Reply-To: Importance: Normal X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/111 X-Sequence-Number: 13352 (first at all, sorry for my english) Hi. - Does "left join" restrict the order in which the planner must join tables? I've read about join, but i'm not sure about left join... - If so: Can I avoid this behavior? I mean, make the planner resolve the query, using statistics (uniqueness, data distribution) rather than join order. My query looks like: SELECT ... FROM a, b, LEFT JOIN c ON (c.key = a.key) LEFT JOIN d on (d.key=a.key) WHERE (a.key = b.key) AND (b.column <= 100) b.column has a lot better selectivity, but planner insist on resolve first c.key = a.key. Of course, I could rewrite something like: SELECT ... FROM (SELECT ... FROM a,b LEFT JOIN d on (d.key=a.key) WHERE (b.column <= 100) ) as aa LEFT JOIN c ON (c.key = aa.key) but this is query is constructed by an application with a "multicolumn" filter. It's dynamic. It means that a user could choose to look for "c.column = 1000". And also, combinations of filters. So, I need the planner to choose the best plan... I've already change statistics, I clustered tables with cluster, ran vacuum analyze, changed work_mem, shared_buffers... Greetings. TIA. From pgsql-performance-owner@postgresql.org Wed Jul 6 19:05:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BBEFD529A2 for ; Wed, 6 Jul 2005 19:05:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91573-05 for ; Wed, 6 Jul 2005 22:04:53 +0000 (GMT) Received: from netbox.unitech.com.ar (unknown [200.32.92.34]) by svr1.postgresql.org (Postfix) with ESMTP id E5B8A52995 for ; Wed, 6 Jul 2005 19:04:52 -0300 (ADT) Received: from dariop (dariop.unitech.com.ar [192.168.1.237]) by netbox.unitech.com.ar (8.11.6/8.11.6) with ESMTP id j66N6QE28869 for ; Wed, 6 Jul 2005 19:06:26 -0400 Received: from 127.0.0.1 (AVG SMTP 7.0.323 [267.8.9]); Wed, 06 Jul 2005 19:04:48 -0300 From: "Dario" To: Subject: join and query planner Date: Wed, 6 Jul 2005 19:04:48 -0300 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4952.2800 In-Reply-To: Importance: Normal X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.131 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/54 X-Sequence-Number: 13295 (first at all, sorry for my english) Hi. - Does "left join" restrict the order in which the planner must join tables? I've read about join, but i'm not sure about left join... - If so: Can I avoid this behavior? I mean, make the planner resolve the query, using statistics (uniqueness, data distribution) rather than join order. My query looks like: SELECT ... FROM a, b, LEFT JOIN c ON (c.key = a.key) LEFT JOIN d on (d.key=a.key) WHERE (a.key = b.key) AND (b.column <= 100) b.column has a lot better selectivity, but planner insist on resolve first c.key = a.key. Of course, I could rewrite something like: SELECT ... FROM (SELECT ... FROM a,b LEFT JOIN d on (d.key=a.key) WHERE (b.column <= 100) ) as aa LEFT JOIN c ON (c.key = aa.key) but this is query is constructed by an application with a "multicolumn" filter. It's dynamic. It means that a user could choose to look for "c.column = 1000". And also, combinations of filters. So, I need the planner to choose the best plan... I've already change statistics, I clustered tables with cluster, ran vacuum analyze, changed work_mem, shared_buffers... Greetings. TIA. From pgsql-performance-owner@postgresql.org Wed Jul 6 19:34:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 13EFC529C8 for ; Wed, 6 Jul 2005 19:34:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 99288-02 for ; Wed, 6 Jul 2005 22:34:04 +0000 (GMT) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 940E6529C3 for ; Wed, 6 Jul 2005 19:34:03 -0300 (ADT) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 6BF3C353CD; Wed, 6 Jul 2005 15:34:03 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 6A7DC350DB; Wed, 6 Jul 2005 15:34:03 -0700 (PDT) Date: Wed, 6 Jul 2005 15:34:03 -0700 (PDT) From: Stephan Szabo To: Dario Cc: pgsql-performance@postgresql.org Subject: Re: join and query planner In-Reply-To: Message-ID: <20050706152750.V50953@megazone.bigpanda.com> References: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/55 X-Sequence-Number: 13296 On Wed, 6 Jul 2005, Dario wrote: > > (first at all, sorry for my english) > Hi. > - Does "left join" restrict the order in which the planner must join > tables? I've read about join, but i'm not sure about left join... Yes. Reordering the outer joins can change the results in some cases which would be invalid. Before we can change the ordering behavior, we really need to know under what conditions it is safe to do the reordering. From pgsql-performance-owner@postgresql.org Wed Jul 6 21:05:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 69DCA529A1 for ; Wed, 6 Jul 2005 21:05:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 15887-04 for ; Thu, 7 Jul 2005 00:05:15 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id AB6515299B for ; Wed, 6 Jul 2005 21:05:13 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Wed, 06 Jul 2005 20:05:01 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Wed, 6 Jul 2005 20:04:33 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: Data Warehousing Tuning Date: Wed, 6 Jul 2005 20:02:24 -0400 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Data Warehousing Tuning Thread-Index: AcWCToiHLbiY6PbdRHSRP/aj2gvOjQAJFodQ From: "Frank Wosczyna" To: paul@oxton.com, pgsql-performance@postgresql.org X-OriginalArrivalTime: 07 Jul 2005 00:04:33.0168 (UTC) FILETIME=[7433E100:01C58287] X-WSS-ID: 6ED2AE2728K2388301-07-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/56 X-Sequence-Number: 13297 Hi Paul, just some quick thoughts: =20 > -----Original Message----- > From: pgsql-performance-owner@postgresql.org=20 > [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of=20 > Paul Johnson > Sent: Wednesday, July 06, 2005 10:16 AM > To: pgsql-performance@postgresql.org > Subject: [PERFORM] Data Warehousing Tuning >=20 >=20 > Questions are as follows: >=20 > 1) Should we have set the page size to 32MB when we compiled Postgres? >=20 > We mainly do bulk loads using 'copy', full-table scans and=20 > large joins so this would seem sensible. Tables are typically=20 > 10 million rows at present. I would defer changing page size to the "fine-tuning" category, our experience with that has not produced substantial gains. Would focus on the other categories you mention below first. Also, for heavy use of COPY, you may consider using the latest release of Bizgres 0.6, which should speed loads: http://www.bizgres.org/pages.php?pg=3Ddownloads or http://www.greenplum.com/prod_download.html for compiled version. >=20 > 2) What are the obvious changes to make to postgresql.conf? >=20 > Things like shared_buffers, work_mem, maintenance_work_mem=20 > and checkpoint_segments seem like good candidates given our=20 > data warehousing workloads. You're on the right track, it depends on nature of queries (sorry for giving you the "consulting" answer on that one), but here are some PostgreSQL configurations to consider: a - Consider using separate disk partitions for transaction log, temp space and WAL. See separate note about WAL and directio in Solaris tuning note, link below. May put temp space on a separate partition, in anticipation of forthcoming changes which take advantage of this. b - Sizing temp space? Should be as large as the largest index. Set max speed read/write config: minimal journaling, use write-through cache on this. c - Might try increasing checkpoint segments (64?). More logs produces significant benefit. And turn checkpoint warnings on (Off by default). d - Sort mem and work mem - What queries are you running? Workmem used in aggregation/sorts. How many concurrent reports? For 3 complex queries, might try 256MB at work mem? e - You probably do this already, but always ANALYZE after loads. f - Maintenance work mem - used in vacuum, analyze, creating bulk indexes, bulk checking for keys. Might consider using 512 or 750? >=20 > 3) Ditto /etc/system? See http://www.bizgres.org/bb/viewtopic.php?t=3D6 for Solaris. >=20 > 4) We moved the pg_xlog files off /data/postgres (disks 2-7)=20 > and into /opt/pg_xlog (disks 0-1), but it seemed like=20 > performance decreased, so we moved them back again. >=20 > Has anyone experienced real performance gains by moving the=20 > pg_xlog files? >=20 Likely to help only with COPY. Feel free to contact me directly if you have any questions on my statements above. There is a Configurator in development which you might find helpful when it is complete: http://www.bizgres.org/pages.php?pg=3Ddevelopers%7Cprojects%7Cconfigurato= r Kind Regards, Frank Frank Wosczyna Systems Engineer +1 650 224 7374 =20 http://www.greenplum.com=20 GreenPlum, Inc. 1900 South Norfolk Street, Suite 224 San Mateo, California 94403 USA From pgsql-performance-owner@postgresql.org Wed Jul 6 21:37:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B102452835 for ; Wed, 6 Jul 2005 21:37:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 23028-02 for ; Thu, 7 Jul 2005 00:37:55 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 1094652808 for ; Wed, 6 Jul 2005 21:37:54 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7579489; Wed, 06 Jul 2005 17:40:05 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: paul@oxton.com Subject: Re: Data Warehousing Tuning Date: Wed, 6 Jul 2005 17:40:12 -0700 User-Agent: KMail/1.8 Cc: pgsql-performance@postgresql.org References: <3688.217.45.209.171.1120670147.squirrel@www.gradwell.com> In-Reply-To: <3688.217.45.209.171.1120670147.squirrel@www.gradwell.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507061740.12875.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/57 X-Sequence-Number: 13298 Paul, > Has anyone experienced real performance gains by moving the pg_xlog > files? Yes. Both for data load and on OLTP workloads, this increased write performance by as much as 15%. However, you need to configure the xlog drive correctly, you can't just move it to a new disk. Make sure the other disk is dedicated exclusively to the xlog, set it forcedirectio, and increase your checkpoint_segments to something like 128. -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Jul 7 04:42:05 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 606D252A15 for ; Thu, 7 Jul 2005 04:42:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34922-06 for ; Thu, 7 Jul 2005 07:41:54 +0000 (GMT) Received: from dns.texnet.it (dns.texnet.it [217.19.150.6]) by svr1.postgresql.org (Postfix) with ESMTP id B686852A0A for ; Thu, 7 Jul 2005 04:41:52 -0300 (ADT) Received: from pool-a-150.texnet.it ([217.19.150.150] helo=paros.rigacci.org) by dns.texnet.it with esmtp (Exim 3.35 #1 (Debian)) id 1DqR13-0000qJ-00; Thu, 07 Jul 2005 09:41:49 +0200 Received: from niccolo by paros.rigacci.org with local (Exim 4.50) id 1DqR11-0004HG-V5; Thu, 07 Jul 2005 09:41:47 +0200 Date: Thu, 7 Jul 2005 09:41:47 +0200 To: John A Meinel Cc: Postgresql Performance Subject: Re: CURSOR slowes down a WHERE clause 100 times? Message-ID: <20050707074147.GA16360@rigacci.org> References: <20050706211946.GA3460@rigacci.org> <42CC4D1C.9040606@arbash-meinel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42CC4D1C.9040606@arbash-meinel.com> User-Agent: Mutt/1.5.9i From: Niccolo Rigacci X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/58 X-Sequence-Number: 13299 > >Can the CURSOR on JOIN affects so heavly the WHERE clause? I > >suspect that - with the CURSOR - a sequential scan is performed > >on the entire data set for each fetched record... > > > >Any idea? > > > > > What does it say if you do "EXPLAIN ANALYZE SELECT..." both with and > without the cursor? > It may not say much for the cursor, but I think you can explain analyze > the fetch statements. How can I EXPLAIN ANALYZE a cursor like this? BEGIN; DECLARE mycursor BINARY CURSOR FOR SELECT ... FETCH ALL IN mycursor; END; I tried to put EXPLAIN ANALYZE in front of the SELECT and in front of the FETCH, but I got two "syntax error"... Thanks -- Niccolo Rigacci Firenze - Italy War against Iraq? Not in my name! From pgsql-performance-owner@postgresql.org Thu Jul 7 05:13:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6612C529F3 for ; Thu, 7 Jul 2005 05:13:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 41190-09 for ; Thu, 7 Jul 2005 08:13:27 +0000 (GMT) Received: from dns.texnet.it (dns.texnet.it [217.19.150.6]) by svr1.postgresql.org (Postfix) with ESMTP id 97F8C529BE for ; Thu, 7 Jul 2005 05:13:27 -0300 (ADT) Received: from pool-a-150.texnet.it ([217.19.150.150] helo=paros.rigacci.org) by dns.texnet.it with esmtp (Exim 3.35 #1 (Debian)) id 1DqRVg-00042R-00 for ; Thu, 07 Jul 2005 10:13:28 +0200 Received: from niccolo by paros.rigacci.org with local (Exim 4.50) id 1DqRVf-0004TL-VP for pgsql-performance@postgresql.org; Thu, 07 Jul 2005 10:13:27 +0200 Date: Thu, 7 Jul 2005 10:13:27 +0200 To: pgsql-performance@postgresql.org Subject: Re: CURSOR slowes down a WHERE clause 100 times? Message-ID: <20050707081327.GA17054@rigacci.org> References: <20050706211946.GA3460@rigacci.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050706211946.GA3460@rigacci.org> User-Agent: Mutt/1.5.9i From: Niccolo Rigacci X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.102 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/59 X-Sequence-Number: 13300 On Wed, Jul 06, 2005 at 11:19:46PM +0200, Niccolo Rigacci wrote: > > I have a performace problem with the following query: > > BEGIN; > DECLARE mycursor BINARY CURSOR FOR > SELECT > toponimo, > wpt > FROM wpt_comuni_view > WHERE ( > wpt && > setSRID('BOX3D(4.83 36, 20.16 47.5)'::BOX3D, 4326) > ); > FETCH ALL IN mycursor; > END; > > I get the results in about 108 seconds (8060 rows). > > If I issue the SELECT alone (without the CURSOR) I get the > same results in less than 1 second. By trial and error I discovered that adding an "ORDER BY toponimo" clause to the SELECT, boosts the CURSOR performances so that they are now equiparable to the SELECT alone. Is there some documentation on how an ORDER can affect the CURSOR in a different way than the SELECT? -- Niccolo Rigacci Firenze - Italy War against Iraq? Not in my name! From pgsql-performance-owner@postgresql.org Thu Jul 7 06:15:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4B6BF52986 for ; Thu, 7 Jul 2005 06:15:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51953-05 for ; Thu, 7 Jul 2005 09:15:30 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id 96E5C52930 for ; Thu, 7 Jul 2005 06:15:21 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id 2B7A040EF94; Thu, 7 Jul 2005 10:15:16 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id ADAD915ED6; Thu, 7 Jul 2005 10:14:53 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 20598-09; Thu, 7 Jul 2005 10:14:50 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 9232E15ED5; Thu, 7 Jul 2005 10:14:50 +0100 (BST) Message-ID: <42CCF28A.3010403@archonet.com> Date: Thu, 07 Jul 2005 10:14:50 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Niccolo Rigacci Cc: pgsql-performance@postgresql.org Subject: Re: CURSOR slowes down a WHERE clause 100 times? References: <20050706211946.GA3460@rigacci.org> <20050707081327.GA17054@rigacci.org> In-Reply-To: <20050707081327.GA17054@rigacci.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/60 X-Sequence-Number: 13301 Niccolo Rigacci wrote: >> >>I get the results in about 108 seconds (8060 rows). >> >>If I issue the SELECT alone (without the CURSOR) I get the >>same results in less than 1 second. > > > By trial and error I discovered that adding an "ORDER BY > toponimo" clause to the SELECT, boosts the CURSOR performances > so that they are now equiparable to the SELECT alone. > > Is there some documentation on how an ORDER can affect the > CURSOR in a different way than the SELECT? I think you're misunderstanding exactly what's happening here. If you ask for a cursor, PG assumes you aren't going to want all the results (or at least not straight away). After all, most people use them to work through results in comparatively small chunks, perhaps only ever fetching 1% of the total results. So - if you ask for a cursor, PG weights things to give you the first few rows as soon as possible, at the expense of fetching *all* rows quickly. If you're only going to fetch e.g. the first 20 rows this is exactly what you want. In your case, since you're immediately issuing FETCH ALL, you're not really using the cursor at all, but PG doesn't know that. So - the ORDER BY means PG has to sort all the results before returning the first row anyway. That probably means the plans with/without cursor are identical. Of course, all this assumes that your configuration settings are good and statistics adequate. To test that, try fetching just the first row from your cursor with/without the ORDER BY. Without should be quicker. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Jul 7 07:07:06 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4C46552996 for ; Thu, 7 Jul 2005 07:07:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62964-03 for ; Thu, 7 Jul 2005 10:06:57 +0000 (GMT) Received: from dns.texnet.it (dns.texnet.it [217.19.150.6]) by svr1.postgresql.org (Postfix) with ESMTP id 5C1F352816 for ; Thu, 7 Jul 2005 07:06:56 -0300 (ADT) Received: from pool-a-150.texnet.it ([217.19.150.150] helo=paros.rigacci.org) by dns.texnet.it with esmtp (Exim 3.35 #1 (Debian)) id 1DqTHX-0005yK-00 for ; Thu, 07 Jul 2005 12:06:59 +0200 Received: from niccolo by paros.rigacci.org with local (Exim 4.50) id 1DqTHW-00056m-G5 for pgsql-performance@postgresql.org; Thu, 07 Jul 2005 12:06:58 +0200 Date: Thu, 7 Jul 2005 12:06:58 +0200 To: pgsql-performance@postgresql.org Subject: Re: CURSOR slowes down a WHERE clause 100 times? Message-ID: <20050707100658.GA19237@rigacci.org> References: <20050706211946.GA3460@rigacci.org> <20050707081327.GA17054@rigacci.org> <42CCF28A.3010403@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42CCF28A.3010403@archonet.com> User-Agent: Mutt/1.5.9i From: Niccolo Rigacci X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.084 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/61 X-Sequence-Number: 13302 On Thu, Jul 07, 2005 at 10:14:50AM +0100, Richard Huxton wrote: > >By trial and error I discovered that adding an "ORDER BY > >toponimo" clause to the SELECT, boosts the CURSOR performances > >so that they are now equiparable to the SELECT alone. > I think you're misunderstanding exactly what's happening here. If you > ask for a cursor, PG assumes you aren't going to want all the results > (or at least not straight away). After all, most people use them to work > through results in comparatively small chunks, perhaps only ever > fetching 1% of the total results. This make finally sense! > In your case, since you're immediately issuing FETCH ALL, > you're not really using the cursor at all, but PG doesn't know > that. In fact, fetching only the first rows from the cursor, is rather quick! This demonstrates that the PG planner is smart. Not so smart are the MapServer and QGIS query builders, which use a CURSOR to FETCH ALL. I will investigate in this direction now. Thank you very much, your help was excellent! -- Niccolo Rigacci Firenze - Italy War against Iraq? Not in my name! From pgsql-performance-owner@postgresql.org Thu Jul 7 08:19:24 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0665952816 for ; Thu, 7 Jul 2005 08:16:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76379-06 for ; Thu, 7 Jul 2005 11:16:40 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.195]) by svr1.postgresql.org (Postfix) with ESMTP id 683F8529BF for ; Thu, 7 Jul 2005 08:16:36 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 69so178096wri for ; Thu, 07 Jul 2005 04:16:40 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; b=MqFLFn6b6HYDLVpxEECKfPkuwfc0FNn0Rp76D87fR/A4nuKAxWrNAVB8Uekw/UBSsS8hkvvwxaCCkG4s2cr1j5JcvfV6FBSukR76cTcliyjOO1pn26YgtzsLOjs89k1+dhgThsUiYDtHZHM5NXE3QrTKa1ttDUJlPEe+KE3vhEM= Received: by 10.54.29.10 with SMTP id c10mr615683wrc; Thu, 07 Jul 2005 04:16:39 -0700 (PDT) Received: from ?10.80.1.45? ([194.248.208.82]) by mx.gmail.com with ESMTP id d61sm1040970wra.2005.07.07.04.16.39; Thu, 07 Jul 2005 04:16:39 -0700 (PDT) Mime-Version: 1.0 (Apple Message framework v728) Content-Transfer-Encoding: 7bit Message-Id: <6F239216-7936-4E62-B708-46DC9171649C@gmail.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: Bendik Rognlien Johansen Subject: How to speed up delete Date: Thu, 7 Jul 2005 13:16:30 +0200 X-Mailer: Apple Mail (2.728) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.141 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/62 X-Sequence-Number: 13303 Hello, I was wondering if there is any way to speed up deletes on this table (see details below)? I am running few of these deletes (could become many more) inside a transaction and each one takes allmost a second to complete. Is it because of the foreign key constraint, or is it something else? Thanks! Table "public.contacts" Column | Type | Modifiers -------------+------------------------ +---------------------------------------------------------- id | integer | not null default nextval ('public.contacts_id_seq'::text) record | integer | type | integer | value | character varying(128) | description | character varying(255) | priority | integer | itescotype | integer | original | integer | Indexes: "contacts_pkey" PRIMARY KEY, btree (id) "contacts_record_idx" btree (record) Foreign-key constraints: "contacts_original_fkey" FOREIGN KEY (original) REFERENCES contacts(id) dev=# select count(id) from contacts; count -------- 984834 (1 row) dev=# explain analyze DELETE FROM contacts WHERE id = 985458; QUERY PLAN ------------------------------------------------------------------------ ------------------------------------------------ Index Scan using contacts_pkey on contacts (cost=0.00..3.01 rows=1 width=6) (actual time=0.043..0.049 rows=1 loops=1) Index Cond: (id = 985458) Total runtime: 840.481 ms (3 rows) From pgsql-performance-owner@postgresql.org Thu Jul 7 09:42:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AA97752815 for ; Thu, 7 Jul 2005 09:42:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93537-04 for ; Thu, 7 Jul 2005 12:42:06 +0000 (GMT) Received: from nproxy.gmail.com (nproxy.gmail.com [64.233.182.197]) by svr1.postgresql.org (Postfix) with ESMTP id 1B1B8528A6 for ; Thu, 7 Jul 2005 09:42:04 -0300 (ADT) Received: by nproxy.gmail.com with SMTP id o25so47377nfa for ; Thu, 07 Jul 2005 05:42:08 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=b9rqCUvtOM1OWWIL5dfvlnm6HBcH/w8CxJToeG14i60NpSLqaOGlSJUj0OklPN5djx2MSCWHECDQTWJVAhGGrJj1N0YzqQCe2Uip3CtOWAYVSOK67UyM1A3njLnXlmLQ0YhmvR/gohVY5bhN71u/FcCR209Nr1htluMCPctuy6w= Received: by 10.48.237.18 with SMTP id k18mr30206nfh; Thu, 07 Jul 2005 05:42:08 -0700 (PDT) Received: by 10.48.49.14 with HTTP; Thu, 7 Jul 2005 05:42:08 -0700 (PDT) Message-ID: <4b09a0c0507070542fd0d197@mail.gmail.com> Date: Thu, 7 Jul 2005 14:42:08 +0200 From: Jean-Max Reymond Reply-To: Jean-Max Reymond To: pgsql-performance@postgresql.org Subject: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.834 tagged_above=0 required=5 tests=AWL, MISSING_SUBJECT, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/63 X-Sequence-Number: 13304 Hi, These last two days, I have some troubles with a very strange phenomena: I have a 400 Mb database and a stored procedure written in perl which call 14 millions times spi_exec_query (thanks to Tom to fix the memory leak ;-) ). On my laptop whith Centrino 1.6 GHz, 512 Mb RAM, - it is solved in 1h50' for Linux 2.6 - it is solved in 1h37' for WXP Professionnal ( WXP better tan Linux ;-) ) On a Desktop with PIV 2.8 GHz,=20 - it is solved in 3h30 for W2K On a Desktop with PIV 1.8 GHz, two disks with data and index's on each disk - it is solved in 4h for W2K I test CPU, memory performance on my laptop and it seems that the performances are not perfect except for one single test: String sort. So, it seems that for my application (database in memory, 14 millions of very small requests), Centrino (aka Pentium M) has a build-in hardware to boost Postgres performance :-) Any experience to confirm this fact ? --=20 Jean-Max Reymond CKR Solutions Open Source Nice France http://www.ckr-solutions.com From pgsql-performance-owner@postgresql.org Thu Jul 7 09:49:14 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C974C5288B for ; Thu, 7 Jul 2005 09:49:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94289-06 for ; Thu, 7 Jul 2005 12:49:05 +0000 (GMT) Received: from nproxy.gmail.com (nproxy.gmail.com [64.233.182.202]) by svr1.postgresql.org (Postfix) with ESMTP id B019852A08 for ; Thu, 7 Jul 2005 09:49:01 -0300 (ADT) Received: by nproxy.gmail.com with SMTP id x4so49067nfb for ; Thu, 07 Jul 2005 05:49:05 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=ui/eiu5TFXZzCBbGwP/pUlZQj6YhaC0vyjgOmvU9cHJQd8HGPpZMHE9vHepkKb65fm1jXZiIK7+hLnqy/I853fDLksccuMfNrDxJLQx4M2ZYNlhzm5gyo/DI9Pn2z0kbkO4zH7Unf0IUDNbm6kuEr8eQKHxG07KKu8YSpz68Ul4= Received: by 10.48.237.18 with SMTP id k18mr30583nfh; Thu, 07 Jul 2005 05:49:05 -0700 (PDT) Received: by 10.48.49.14 with HTTP; Thu, 7 Jul 2005 05:49:05 -0700 (PDT) Message-ID: <4b09a0c05070705495df28801@mail.gmail.com> Date: Thu, 7 Jul 2005 14:49:05 +0200 From: Jean-Max Reymond Reply-To: Jean-Max Reymond To: pgsql-performance@postgresql.org Subject: Surprizing performances for Postgres on Centrino Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.08 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/64 X-Sequence-Number: 13305 Hi, These last two days, I have some troubles with a very strange phenomena: I have a 400 Mb database and a stored procedure written in perl which call 14 millions times spi_exec_query (thanks to Tom to fix the memory leak ;-) ). On my laptop whith Centrino 1.6 GHz, 512 Mb RAM, - it is solved in 1h50' for Linux 2.6 - it is solved in 1h37' for WXP Professionnal ( WXP better tan Linux ;-) ) On a Desktop with PIV 2.8 GHz, - it is solved in 3h30 for W2K On a Desktop with PIV 1.8 GHz, two disks with data and index's on each disk - it is solved in 4h for W2K I test CPU, memory performance on my laptop and it seems that the performances are not perfect except for one single test: String sort. So, it seems that for my application (database in memory, 14 millions of very small requests), Centrino (aka Pentium M) has a build-in hardware to boost Postgres performance :-) Any experience to confirm this fact ? Some tips to speed up Postgres on non-Centrino processors ? --=20 Jean-Max Reymond CKR Solutions Open Source Nice France http://www.ckr-solutions.com From pgsql-performance-owner@postgresql.org Thu Jul 7 09:55:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F363452923 for ; Thu, 7 Jul 2005 09:55:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95038-09 for ; Thu, 7 Jul 2005 12:55:30 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id DA3D652823 for ; Thu, 7 Jul 2005 09:55:26 -0300 (ADT) Received: (qmail 8571 invoked from network); 7 Jul 2005 14:55:56 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 7 Jul 2005 14:55:56 +0200 To: "Bendik Rognlien Johansen" , pgsql-performance@postgresql.org Subject: Re: How to speed up delete References: <6F239216-7936-4E62-B708-46DC9171649C@gmail.com> Message-ID: Date: Thu, 07 Jul 2005 14:55:28 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <6F239216-7936-4E62-B708-46DC9171649C@gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/65 X-Sequence-Number: 13306 On Thu, 07 Jul 2005 13:16:30 +0200, Bendik Rognlien Johansen wrote: > Hello, > I was wondering if there is any way to speed up deletes on this table > (see details below)? > I am running few of these deletes (could become many more) inside a > transaction and each one takes allmost a second to complete. > Is it because of the foreign key constraint, or is it something else? > > Thanks! Check your references : on delete, pg needs to find which rows to cascade-delete, or set null, or restrict, in the tables which reference this one. Also if this table references another I think it will lookup it too. Do you have indexes for all this ? From pgsql-performance-owner@postgresql.org Thu Jul 7 10:03:41 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 255DB528CE for ; Thu, 7 Jul 2005 10:03:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 99310-04 for ; Thu, 7 Jul 2005 13:03:20 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 44F785288B for ; Thu, 7 Jul 2005 10:03:18 -0300 (ADT) Received: (qmail 9038 invoked from network); 7 Jul 2005 15:03:49 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 7 Jul 2005 15:03:49 +0200 Date: Thu, 07 Jul 2005 15:03:22 +0200 To: "Jean-Max Reymond" , pgsql-performance@postgresql.org Subject: Re: References: <4b09a0c0507070542fd0d197@mail.gmail.com> From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-ID: In-Reply-To: <4b09a0c0507070542fd0d197@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/66 X-Sequence-Number: 13307 > So, it seems that for my application (database in memory, 14 millions > of very small requests), Centrino (aka Pentium M) has a build-in > hardware to boost Postgres performance :-) > Any experience to confirm this fact ? On my Centrino, Python flies. This might be due to the very large processor cache. Probably it is the same for perl. With two megabytes of cache, sorting things that fit into the cache should be a lot faster too. Maybe this explains it. Check this out : http://www.anandtech.com/linux/showdoc.aspx?i=2308&p=5 http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2129&p=11 Bonus for Opteron lovers : "The Dual Opteron 252's lead by 19% over the Quad Xeon 3.6 GHz 667MHz FSB" http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2397&p=12 From pgsql-performance-owner@postgresql.org Thu Jul 7 10:14:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2ABDC5298E for ; Thu, 7 Jul 2005 10:14:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 02547-03 for ; Thu, 7 Jul 2005 13:14:15 +0000 (GMT) Received: from tomts25-srv.bellnexxia.net (tomts25.bellnexxia.net [209.226.175.188]) by svr1.postgresql.org (Postfix) with ESMTP id BA9E4529EA for ; Thu, 7 Jul 2005 10:14:11 -0300 (ADT) Received: from [192.168.1.99] ([216.208.56.114]) by tomts25-srv.bellnexxia.net (InterMail vM.5.01.06.10 201-253-122-130-110-20040306) with ESMTP id <20050707131413.CUPQ27245.tomts25-srv.bellnexxia.net@[192.168.1.99]> for ; Thu, 7 Jul 2005 09:14:13 -0400 Message-ID: <42CD2A7A.10603@alteeve.com> Date: Thu, 07 Jul 2005 09:13:30 -0400 From: Madison Kelly User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Debian/1.7.8-1 X-Accept-Language: en-us, en, ja MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Need suggestion high-level suggestion on how to solve a performance problem Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/67 X-Sequence-Number: 13308 Hi all, I hope I am not asking too many questions. :) I have been trying to solve a performance problem in my program for a while now and, after getting an index to work which didn't speed things up enough, I am stumped. I am hoping someone here might have come across a similar issue and came up with a creative solution they wouldn't mind sharing. I am not looking for details, I expect to do my homework, I just need a pointer, suggestion or trick. The problem I have is that I am using pgSQL as a back end for my web-based *nix backup program. Part of the database stores info on every file and directory per partition. I use this information to build my directory tree. I have a test partition with ~325,000 files of which ~30,000 are directories. I have been able to get the performance up to a reasonable level for displaying the directory tree including expanding and contracting branches (~3-5sec). I do this by loading all the directory info into an array and a hash once and using them as needed instead of hitting the DB. The problem comes when the user toggles a directory branch's backup flag (a simple check box beside the directory name). If it's a directory near the end of a branch it is fast enough. If they toggle a single file it is nearly instant. However if they toggle say the root directory, so every file and directory below it needs to be updated, it can take 500-600sec to return. Obviously this is no good. What I need is a scheme for being able to say, essentially: UPDATE file_info_1 SET file_backup='t' WHERE file_parent_dir~'^/'; Faster. An index isn't enough because it needs to hit every entry anyway. I use perl to access the DB and generate the web pages. The file browser portion looks and acts like most file browsers (directory tree in the left frame with expanding and contracting directory branches and a list of files in a given directory on the right). It does not use any plug-ins like Java and that is important to me that it stays that way (I want it to be as simple as possible for the user to install). So far the only suggestion I've received is to keep a secondary 'delta' table to store just the request. Then on load get the existing data then check it against the delta table before creating the page. The biggest draw back for me with this is that currently I don't need to provide an 'Apply' button because a simple javascript call passes the request onto the perl script immediately. I really like the Mac-esque approach to keeping the UI as simple and straight forward as possible. So, a suggestion that doesn't require something like an 'Apply' button would be much appreciated. Thanks for any suggestions in advance! Madison PS - For what it's worth, this is the last hurdle for me to overcome before I can finally release my program as 'beta' after over 15 months of work! :) From pgsql-performance-owner@postgresql.org Thu Jul 7 10:48:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9249552923 for ; Thu, 7 Jul 2005 10:48:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 06089-10 for ; Thu, 7 Jul 2005 13:48:07 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.207]) by svr1.postgresql.org (Postfix) with ESMTP id 7DE8C529EA for ; Thu, 7 Jul 2005 10:48:01 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 8so87528nzo for ; Thu, 07 Jul 2005 06:48:06 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=KPGx8DJlLtmcLu9s/RtKbBpALPUTChZivZKTdWkhEe3qb0hg0rbka/wCIJqHZDTNGpuEBX4lrbw2itGSwObi8SDnbf0LXMU98SaLK8ikUOGltBQW5O3Dfa3YQXoATq6hBHFKrRY4F22hktDO87JvlEUzN2Cl0PMnBSf47VNbKJ8= Received: by 10.36.222.46 with SMTP id u46mr294300nzg; Thu, 07 Jul 2005 06:48:06 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Thu, 7 Jul 2005 06:48:06 -0700 (PDT) Message-ID: <758d5e7f05070706481ef034d4@mail.gmail.com> Date: Thu, 7 Jul 2005 15:48:06 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Jean-Max Reymond Subject: Re: Surprizing performances for Postgres on Centrino Cc: Postgresql Performance In-Reply-To: <4b09a0c05070705495df28801@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <4b09a0c05070705495df28801@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.613 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/68 X-Sequence-Number: 13309 On 7/7/05, Jean-Max Reymond wrote: > On my laptop whith Centrino 1.6 GHz, 512 Mb RAM, > - it is solved in 1h50' for Linux 2.6 > - it is solved in 1h37' for WXP Professionnal ( WXP better > tan Linux ;-) ) [...] > I test CPU, memory performance on my laptop and it seems that the > performances are not perfect except for one single test: String sort. Well, Pentium 4 is not the most efficient processor around (despite all the advertisiing and all the advanced hyper features). Sure it reaches high GHz rates, but that's not what matters the most. This is why AMD stopped giving GHz ratings and instead uses numbers which indicate how their processor relate to Pentium 4s. For instance AMD Athlon XP 1700+ is running at 1.45 GHz, but competes with Pentium 4 1.7 GHz. Same is with Intels Pentium-III line (which evolved into Pentium-M Centrino actually). Like AMD Athlon, Pentium-M is more efficient about its clockspeed than Pentium 4. In other words, you shouldn't compare Pentium 4 and Pentium-M clock-by-clock. Pentium 4 just needs more GHz to do same job as Pentium-M or Athlon. If you want to get some better (more technical) information, just google around for reviews and articles. There are plenty of them recently since Apple intends to use Pentium-M as their future platform, at least for notebooks. As for technical stuff, for instance look at: http://www.tomshardware.com/howto/20050621/37watt-pc-02.html What really is interesting is the performance difference between WXP and L26... Are you sure they use exactly the same config parameters (shared buffers) and have similar statistics (both VACUUM ANALYZEd recently)? Regards, Dawid From pgsql-performance-owner@postgresql.org Thu Jul 7 10:59:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D427E52A11 for ; Thu, 7 Jul 2005 10:59:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07513-08 for ; Thu, 7 Jul 2005 13:59:34 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 115B1528CE for ; Thu, 7 Jul 2005 10:59:33 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j67DxX2l007485; Thu, 7 Jul 2005 09:59:33 -0400 (EDT) To: Niccolo Rigacci Cc: John A Meinel , Postgresql Performance Subject: Re: CURSOR slowes down a WHERE clause 100 times? In-reply-to: <20050707074147.GA16360@rigacci.org> References: <20050706211946.GA3460@rigacci.org> <42CC4D1C.9040606@arbash-meinel.com> <20050707074147.GA16360@rigacci.org> Comments: In-reply-to Niccolo Rigacci message dated "Thu, 07 Jul 2005 09:41:47 +0200" Date: Thu, 07 Jul 2005 09:59:33 -0400 Message-ID: <7484.1120744773@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.109 tagged_above=0 required=5 tests=AWL, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/69 X-Sequence-Number: 13310 Niccolo Rigacci writes: > How can I EXPLAIN ANALYZE a cursor like this? > BEGIN; > DECLARE mycursor BINARY CURSOR FOR > SELECT ... > FETCH ALL IN mycursor; > END; > I tried to put EXPLAIN ANALYZE in front of the SELECT and in > front of the FETCH, but I got two "syntax error"... Just FYI, you can't EXPLAIN ANALYZE this, but you can EXPLAIN it: EXPLAIN DECLARE x CURSOR FOR ... so you can at least find out what the plan is. It might be cool to support EXPLAIN ANALYZE FETCH --- not sure what that would take. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 7 11:02:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 15C3352A1A for ; Thu, 7 Jul 2005 11:02:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 09802-09 for ; Thu, 7 Jul 2005 14:02:31 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id C152F52A0E for ; Thu, 7 Jul 2005 11:02:21 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j67E2PvJ007518; Thu, 7 Jul 2005 10:02:25 -0400 (EDT) To: Bendik Rognlien Johansen Cc: pgsql-performance@postgresql.org Subject: Re: How to speed up delete In-reply-to: <6F239216-7936-4E62-B708-46DC9171649C@gmail.com> References: <6F239216-7936-4E62-B708-46DC9171649C@gmail.com> Comments: In-reply-to Bendik Rognlien Johansen message dated "Thu, 07 Jul 2005 13:16:30 +0200" Date: Thu, 07 Jul 2005 10:02:25 -0400 Message-ID: <7517.1120744945@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/70 X-Sequence-Number: 13311 Bendik Rognlien Johansen writes: > I am running few of these deletes (could become many more) inside a > transaction and each one takes allmost a second to complete. > Is it because of the foreign key constraint, or is it something else? You need an index on "original" to support that FK efficiently. Check for references from other tables to this one, too. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 7 11:03:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B77AC52A16 for ; Thu, 7 Jul 2005 11:03:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 11537-07 for ; Thu, 7 Jul 2005 14:03:03 +0000 (GMT) Received: from cassarossa.samfundet.no (cassarossa.samfundet.no [129.241.93.19]) by svr1.postgresql.org (Postfix) with ESMTP id DBB5B52A03 for ; Thu, 7 Jul 2005 11:03:01 -0300 (ADT) Received: from trofast.sesse.net ([129.241.93.32]) by cassarossa.samfundet.no with esmtp (Exim 4.50) id 1DqWxx-0005IL-Md; Thu, 07 Jul 2005 16:03:03 +0200 Received: from sesse by trofast.sesse.net with local (Exim 3.36 #1 (Debian)) id 1DqWxw-0008BT-00; Thu, 07 Jul 2005 16:03:00 +0200 Date: Thu, 7 Jul 2005 16:03:00 +0200 From: "Steinar H. Gunderson" To: Dawid Kuroczko Cc: Jean-Max Reymond , Postgresql Performance Subject: Re: Surprizing performances for Postgres on Centrino Message-ID: <20050707140300.GA31348@uio.no> Mail-Followup-To: Dawid Kuroczko , Jean-Max Reymond , Postgresql Performance References: <4b09a0c05070705495df28801@mail.gmail.com> <758d5e7f05070706481ef034d4@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <758d5e7f05070706481ef034d4@mail.gmail.com> X-Operating-System: Linux 2.6.11.8 on a i686 X-Message-Flag: Outlook? --> http://www.mozilla.org/products/thunderbird/ User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.018 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/71 X-Sequence-Number: 13312 On Thu, Jul 07, 2005 at 03:48:06PM +0200, Dawid Kuroczko wrote: > This is why AMD stopped giving GHz ratings and instead uses numbers > which indicate how their processor relate to Pentium 4s. For instance > AMD Athlon XP 1700+ is running at 1.45 GHz, but competes with > Pentium 4 1.7 GHz. Actually, the XP ratings are _Athlon Thunderbird_ ratings, not P4 ratings. At least they were intended to be that originally :-) /* Steinar */ -- Homepage: http://www.sesse.net/ From pgsql-performance-owner@postgresql.org Thu Jul 7 11:09:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DE40852815 for ; Thu, 7 Jul 2005 11:09:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 10644-06 for ; Thu, 7 Jul 2005 14:09:32 +0000 (GMT) Received: from mr1.surnet.cl (smtp1.surnet.cl [216.155.73.162]) by svr1.postgresql.org (Postfix) with ESMTP id 847EC52A09 for ; Thu, 7 Jul 2005 11:09:32 -0300 (ADT) Received: from smtp1.surnet.cl (216.155.73.168) by mr1.surnet.cl (7.0.031.3) id 42587EDE013A5F14; Thu, 7 Jul 2005 10:08:58 -0400 Received: from smtp2.surnet.cl (smtp2.surnet.cl []) by mr1.surnet.cl ([216.155.73.168]); Thu, 07 Jul 2005 14:08:58 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp2.surnet.cl (7.0.031.3) id 4259ADE300C05169; Thu, 7 Jul 2005 10:09:23 -0400 Received: from alvh.no-ip.org (200.85.217.230) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60002B0D8F; Thu, 7 Jul 2005 10:09:34 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 2BBDFC2D425; Thu, 7 Jul 2005 10:09:34 -0400 (CLT) Date: Thu, 7 Jul 2005 10:09:34 -0400 From: Alvaro Herrera To: Jean-Max Reymond Cc: pgsql-performance@postgresql.org Subject: Re: Surprizing performances for Postgres on Centrino Message-ID: <20050707140934.GB6139@alvh.no-ip.org> References: <4b09a0c05070705495df28801@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4b09a0c05070705495df28801@mail.gmail.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.45 tagged_above=0 required=5 tests=DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/72 X-Sequence-Number: 13313 On Thu, Jul 07, 2005 at 02:49:05PM +0200, Jean-Max Reymond wrote: > Hi, > These last two days, I have some troubles with a very strange phenomena: > I have a 400 Mb database and a stored procedure written in perl which > call 14 millions times spi_exec_query (thanks to Tom to fix the memory > leak ;-) ). > On my laptop whith Centrino 1.6 GHz, 512 Mb RAM, > - it is solved in 1h50' for Linux 2.6 > - it is solved in 1h37' for WXP Professionnal ( WXP better > tan Linux ;-) ) > On a Desktop with PIV 2.8 GHz, > - it is solved in 3h30 for W2K > On a Desktop with PIV 1.8 GHz, two disks with data and index's on each disk > - it is solved in 4h for W2K Do you have the same locale settings on all of them? -- Alvaro Herrera () "We are who we choose to be", sang the goldfinch when the sun is high (Sandman) From pgsql-performance-owner@postgresql.org Thu Jul 7 11:23:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D278D52A0D for ; Thu, 7 Jul 2005 11:23:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 19647-01 for ; Thu, 7 Jul 2005 14:23:15 +0000 (GMT) Received: from nproxy.gmail.com (nproxy.gmail.com [64.233.182.207]) by svr1.postgresql.org (Postfix) with ESMTP id 5227352A09 for ; Thu, 7 Jul 2005 11:23:14 -0300 (ADT) Received: by nproxy.gmail.com with SMTP id x37so55488nfc for ; Thu, 07 Jul 2005 07:23:18 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=hdI12fkAqS5vX/2HshcWxWlma2P3YKyyTUkPjXi+GsB2x9343IN9SmYJCKf9W67YpwKp8NpEAxDkwatnaaKBEUnRSK6AigMoaUPVqN7ogOFdNaRTZSuaNL50r+vQ1wimI6UxVulfP56wScYQ5uwsPJBvHKMIf24PZoE0lc3vwkk= Received: by 10.48.3.15 with SMTP id 15mr35731nfc; Thu, 07 Jul 2005 07:23:18 -0700 (PDT) Received: by 10.48.49.14 with HTTP; Thu, 7 Jul 2005 07:23:18 -0700 (PDT) Message-ID: <4b09a0c050707072372e87ce6@mail.gmail.com> Date: Thu, 7 Jul 2005 16:23:18 +0200 From: Jean-Max Reymond Reply-To: Jean-Max Reymond To: pgsql-performance@postgresql.org Subject: Re: Surprizing performances for Postgres on Centrino In-Reply-To: <20050707140934.GB6139@alvh.no-ip.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <4b09a0c05070705495df28801@mail.gmail.com> <20050707140934.GB6139@alvh.no-ip.org> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.076 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/73 X-Sequence-Number: 13314 On 7/7/05, Alvaro Herrera wrote: >=20 > Do you have the same locale settings on all of them? >=20 interressant: UNICODE on the fast laptop SQL_ASCII on the slowest desktops. is UNICODE database faster than SQL_ASCII ? --=20 Jean-Max Reymond CKR Solutions Open Source Nice France http://www.ckr-solutions.com From pgsql-performance-owner@postgresql.org Thu Jul 7 11:55:23 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4C06A52A21 for ; Thu, 7 Jul 2005 11:55:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25592-02 for ; Thu, 7 Jul 2005 14:55:20 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id 0C52252A1B for ; Thu, 7 Jul 2005 11:55:18 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id AB9C940C4C3; Thu, 7 Jul 2005 15:55:11 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id 4FBB9FF73; Thu, 7 Jul 2005 15:54:02 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 03370-08; Thu, 7 Jul 2005 15:53:57 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 91927FF70; Thu, 7 Jul 2005 15:52:34 +0100 (BST) Message-ID: <42CD41B1.5050003@archonet.com> Date: Thu, 07 Jul 2005 15:52:33 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Jean-Max Reymond Cc: pgsql-performance@postgresql.org Subject: Re: Surprizing performances for Postgres on Centrino References: <4b09a0c05070705495df28801@mail.gmail.com> <20050707140934.GB6139@alvh.no-ip.org> <4b09a0c050707072372e87ce6@mail.gmail.com> In-Reply-To: <4b09a0c050707072372e87ce6@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/74 X-Sequence-Number: 13315 Jean-Max Reymond wrote: > On 7/7/05, Alvaro Herrera wrote: > > >>Do you have the same locale settings on all of them? >> > > > interressant: > UNICODE on the fast laptop > SQL_ASCII on the slowest desktops. > is UNICODE database faster than SQL_ASCII ? That's your encoding (character-set). Locale is something like "C" or "en_US" or "fr_FR". -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Jul 7 12:07:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 29D7E5288B for ; Thu, 7 Jul 2005 12:03:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 24868-07 for ; Thu, 7 Jul 2005 15:03:42 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.195]) by svr1.postgresql.org (Postfix) with ESMTP id 3AC4C52A16 for ; Thu, 7 Jul 2005 12:03:41 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 70so217248wra for ; Thu, 07 Jul 2005 08:03:40 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=Y4ZjKsR6QmfKsZc+jKPdd0ZvOMXX6mGyDP9MwsdFeqTDvLRaw4tiz+dn7vl+rVsirkHFgNxYjKpoH5nVgkHaBGbG3D6W5TYU1zbL6TLcYOzjqrOtPqNy1hq3FHuEOK0BReTHsEt8piIKEzqBlz8Z7L5yYW4dbPnK8GE9vnDDC0c= Received: by 10.54.26.4 with SMTP id 4mr832288wrz; Thu, 07 Jul 2005 08:03:40 -0700 (PDT) Received: from ?10.80.1.45? ([194.248.208.82]) by mx.gmail.com with ESMTP id 29sm1727248wrl.2005.07.07.08.03.39; Thu, 07 Jul 2005 08:03:39 -0700 (PDT) In-Reply-To: <7517.1120744945@sss.pgh.pa.us> References: <6F239216-7936-4E62-B708-46DC9171649C@gmail.com> <7517.1120744945@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v728) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Cc: Tom Lane Content-Transfer-Encoding: 7bit From: Bendik Rognlien Johansen Subject: Re: How to speed up delete Date: Thu, 7 Jul 2005 17:03:36 +0200 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.728) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.129 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/75 X-Sequence-Number: 13316 Thanks! That took care of it. On Jul 7, 2005, at 4:02 PM, Tom Lane wrote: > Bendik Rognlien Johansen writes: > >> I am running few of these deletes (could become many more) inside a >> transaction and each one takes allmost a second to complete. >> Is it because of the foreign key constraint, or is it something else? >> > > You need an index on "original" to support that FK efficiently. Check > for references from other tables to this one, too. > > regards, tom lane > From pgsql-performance-owner@postgresql.org Thu Jul 7 19:21:41 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2714452A03 for ; Thu, 7 Jul 2005 19:21:36 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 96208-02 for ; Thu, 7 Jul 2005 22:21:33 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id E2A5E52965 for ; Thu, 7 Jul 2005 19:21:30 -0300 (ADT) Received: (qmail 2507 invoked from network); 8 Jul 2005 00:21:59 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 8 Jul 2005 00:21:59 +0200 To: "Madison Kelly" , pgsql-performance@postgresql.org Subject: Re: Need suggestion high-level suggestion on how to solve a performance problem References: <42CD2A7A.10603@alteeve.com> Message-ID: Date: Fri, 08 Jul 2005 00:21:31 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <42CD2A7A.10603@alteeve.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/76 X-Sequence-Number: 13317 Hello, I once upon a time worked in a company doing backup software and I remember these problems, we had exactly the same ! The file tree was all into memory and everytime the user clicked on something it haaad to update everything. Being C++ it was very fast, but to backup a million files you needed a gig of RAM, which is... a problem let's say, when you think my linux laptop has about 400k files on it. So we rewrote the project entirely with the purpose of doing the million files thingy with the clunky Pentium 90 with 64 megabytes of RAM, and it worked. What I did was this : - use Berkeley DB Berkeley DB isn't a database like postgres, it's just a tree, but it's cool for managing trees. It's quite fast, uses key compression, etc. It has however a few drawbacks : - files tend to fragment a lot over time and it can't reindex or vacuum like postgres. You have to dump and reload. - the price of the licence to be able to embed it in your product and sell it is expensive, and if you want crash-proof, it's insanely expensive. - Even though it's a tree it has no idea what a parent is so you have to mess with that manually. We used a clever path encoding to keep all the paths inside the same directory close in the tree ; and separated database for dirs and files because we wanted the dirs to be in the cache, whereas we almost never touched the files. And... You can't make it if you update every node everytime the user clicks on something. You have to update 1 node. In your tree you have nodes. Give each node a state being one of these three : include, exclude, inherit When you fetch a node you also fetch all of its parents, and you propagate the state to know the state of the final node. If a node is in state 'inherit' it is like its parent, etc. So you have faster updates but slower selects. However, there is a bonus : if you check a directory as "include" and one of its subdirectory as "exclude", and the user adds files all over the place, the files added in the "included" directory will be automatically backed up and the ones in the 'ignored' directory will be automatically ignored, you have nothing to change. And it is not that slow because, if you think about it, suppose you have /var/www/mysite/blah with 20.000 files in it, in order to inherit the state of the parents on them you only have to fetch /var once, www once, etc. So if you propagate your inherited properties when doing a tree traversal it comes at no cost. IMHO it's the only solution. It can be done quite easily also, using ltree types and a little stored procedures, you can even make a view which gives the state of each element, computed by inheritance. Here's the secret : the user will select 100.000 files by clicking on a directory near root, but the user will NEVER look at 100.000 files. So you can make looking at files 10x slower if you can make including/excluding directories 100.000 times faster. Now you'll ask me, but how do I calculate the total size of the backup without looking at all the files ? when I click on a directory I don't know what files are in it and which will inherit and which will not. It's simple : you precompute it when you scan the disk for changed files. This is the only time you should do a complete tree exploration. On each directory we put a matrix [M]x[N], M and N being one of the three above state, containing the amount of stuff in the directory which would be in state M if the directory was in state N. This is very easy to compute when you scan for new files. Then when a directory changes state, you have to sum a few cells of that matrix to know how much more that adds to the backup. And you only look up 1 record. Is that helpful ? From pgsql-performance-owner@postgresql.org Thu Jul 7 20:29:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 81DA752A20 for ; Thu, 7 Jul 2005 20:29:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 21938-09 for ; Thu, 7 Jul 2005 23:29:32 +0000 (GMT) Received: from tomts5-srv.bellnexxia.net (tomts5.bellnexxia.net [209.226.175.25]) by svr1.postgresql.org (Postfix) with ESMTP id 179FB52815 for ; Thu, 7 Jul 2005 20:29:30 -0300 (ADT) Received: from [192.168.1.99] ([64.229.11.238]) by tomts5-srv.bellnexxia.net (InterMail vM.5.01.06.10 201-253-122-130-110-20040306) with ESMTP id <20050707232932.USOW26128.tomts5-srv.bellnexxia.net@[192.168.1.99]>; Thu, 7 Jul 2005 19:29:32 -0400 Message-ID: <42CDBAB2.9030406@alteeve.com> Date: Thu, 07 Jul 2005 19:28:50 -0400 From: Madison Kelly User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Debian/1.7.8-1 X-Accept-Language: en-us, en, ja MIME-Version: 1.0 To: PFC Cc: pgsql-performance@postgresql.org Subject: Re: Need suggestion high-level suggestion on how to solve References: <42CD2A7A.10603@alteeve.com> In-Reply-To: Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.01 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/77 X-Sequence-Number: 13318 PFC wrote: > > Hello, > I once upon a time worked in a company doing backup software and I > remember these problems, we had exactly the same ! Prety neat. :) > The file tree was all into memory and everytime the user clicked on > something it haaad to update everything. Being C++ it was very fast, > but to backup a million files you needed a gig of RAM, which is... a > problem let's say, when you think my linux laptop has about 400k files > on it. I want this to run on "average" systems (I'm developing it primarily on my modest P3 1GHz Thinkpad w/ 512MB RAM running Debian) so expecting that much free memory is not reasonable. As it is my test DB, with a realistic amount of data, is ~150MB. > So we rewrote the project entirely with the purpose of doing the > million files thingy with the clunky Pentium 90 with 64 megabytes of > RAM, and it worked. > What I did was this : > - use Berkeley DB > - the price of the licence to be able to embed it in your product > and sell it is expensive, and if you want crash-proof, it's insanely > expensive. This is the kicker right there; my program is released under the GPL so it's fee-free. I can't eat anything costly like that. As it is there is hundreds and hundreds of hours in this program that I am already hoping to recoup one day through support contracts. Adding commercial software I am afraid is not an option. > bonus : if you check a directory as "include" and one of its > subdirectory as "exclude", and the user adds files all over the place, > the files added in the "included" directory will be automatically > backed up and the ones in the 'ignored' directory will be automatically > ignored, you have nothing to change. > IMHO it's the only solution. Now *this* is an idea worth looking into. How I will implement it with my system I don't know yet but it's a new line of thinking. Wonderful! > Now you'll ask me, but how do I calculate the total size of the > backup without looking at all the files ? when I click on a directory I > don't know what files are in it and which will inherit and which will not. > > It's simple : you precompute it when you scan the disk for changed > files. This is the only time you should do a complete tree exploration. This is already what I do. When a user selects a partition they want to select files to backup or restore the partition is scanned. The scan looks at every file, directory and symlink and records it's size (on disk), it mtime, owner, group, etc. and records it to the database. I've got this scan/update running at ~1,500 files/second on my laptop. That was actually the first performance tuning I started with. :) With all the data in the DB the backup script can calculate rather intelligently where it wants to copy each directory to. > On each directory we put a matrix [M]x[N], M and N being one of the > three above state, containing the amount of stuff in the directory > which would be in state M if the directory was in state N. This is very > easy to compute when you scan for new files. Then when a directory > changes state, you have to sum a few cells of that matrix to know how > much more that adds to the backup. And you only look up 1 record. In my case what I do is calculate the size of all the files selected for backup in each directory, sort the directories from all sources by the total size of all their selected files and then start assigning the directories, largest to smallest to each of my available destination medias. If it runs out of destination space it backs up what it can and then waits a user-definable amount of time and then checks to see if any new destination media has been made available. If so it again tries to assign the files/directories that didn't fit. It will loop a user-definable number of times before giving up and warning the user that more destination space is needed for that backup job. > Is that helpful ? The three states (inhertied, backup, ignore) has definately caught my attention. Thank you very much for your idea and lengthy reply! Madison From pgsql-performance-owner@postgresql.org Thu Jul 7 20:58:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A412052A72 for ; Thu, 7 Jul 2005 20:58:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 24193-05 for ; Thu, 7 Jul 2005 23:58:44 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 35AEC52A6E for ; Thu, 7 Jul 2005 20:58:43 -0300 (ADT) Received: (qmail 7364 invoked from network); 8 Jul 2005 01:59:14 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 8 Jul 2005 01:59:14 +0200 To: "Madison Kelly" Cc: pgsql-performance@postgresql.org Subject: Re: Need suggestion high-level suggestion on how to solve a performance problem References: <42CD2A7A.10603@alteeve.com> <42CDBAB2.9030406@alteeve.com> Message-ID: From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Date: Fri, 08 Jul 2005 01:58:46 +0200 In-Reply-To: <42CDBAB2.9030406@alteeve.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/78 X-Sequence-Number: 13319 > This is the kicker right there; my program is released under the GPL > so it's fee-free. I can't eat anything costly like that. As it is there > is hundreds and hundreds of hours in this program that I am already > hoping to recoup one day through support contracts. Adding commercial > software I am afraid is not an option. If you open-source GPL then Berkeley is free for you. From pgsql-performance-owner@postgresql.org Thu Jul 7 22:02:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 04102529EE for ; Thu, 7 Jul 2005 22:02:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 38007-09 for ; Fri, 8 Jul 2005 01:02:10 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 924E9529E5 for ; Thu, 7 Jul 2005 22:02:09 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7584349; Thu, 07 Jul 2005 18:04:22 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Madison Kelly Subject: Re: Need suggestion high-level suggestion on how to solve a performance problem Date: Thu, 7 Jul 2005 18:04:28 -0700 User-Agent: KMail/1.8 Cc: pgsql-performance@postgresql.org References: <42CD2A7A.10603@alteeve.com> In-Reply-To: <42CD2A7A.10603@alteeve.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Message-Id: <200507071804.29189.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/79 X-Sequence-Number: 13320 Madison, > =A0 =A0The problem comes when the user toggles a directory branch's backup > flag (a simple check box beside the directory name). If it's a directory > near the end of a branch it is fast enough. If they toggle a single file > it is nearly instant. However if they toggle say the root directory, so > every file and directory below it needs to be updated, it can take > 500-600sec to return. Obviously this is no good. > > =A0 =A0What I need is a scheme for being able to say, essentially: > > UPDATE file_info_1 SET file_backup=3D't' WHERE file_parent_dir~'^/'; Well, from the sound of it the problem is not selecting the files to be=20 updated, it's updating them. =20 What I would do, personally, is *not* store an update flag for each file. = =20 Instead, I would store the update flag for the directory which was=20 selected. If users want to exclude certain files and subdirectories, I'd=20 also include a dont_update flag. When it's time to back up, you simply=20 check the tree for the most immediate update or don't update flag above=20 each file. =46or the table itself, I'd consider using ltree for the directory tree=20 structure. It has some nice features which makes it siginifcanly better=20 than using a delimited text field. =2D-=20 =2D-Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Jul 7 22:33:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EE0E352965 for ; Thu, 7 Jul 2005 22:33:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 47651-02 for ; Fri, 8 Jul 2005 01:32:56 +0000 (GMT) Received: from mail.iinet.net.au (mail-08.iinet.net.au [203.59.3.40]) by svr1.postgresql.org (Postfix) with SMTP id 55DBD52A08 for ; Thu, 7 Jul 2005 22:32:54 -0300 (ADT) Received: (qmail 19059 invoked from network); 8 Jul 2005 01:32:51 -0000 Received: from unknown (HELO localhost.localdomain) (203.217.37.199) by mail.iinet.net.au with SMTP; 8 Jul 2005 01:32:51 -0000 Received: from [127.0.0.1] (localhost.localdomain [127.0.0.1]) by localhost.localdomain (Postfix) with ESMTP id 6349D14359; Fri, 8 Jul 2005 11:32:46 +1000 (EST) Message-ID: <42CDD7B8.6050109@stuartbishop.net> Date: Fri, 08 Jul 2005 11:32:40 +1000 From: Stuart Bishop User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Mount database on RAM disk? X-Enigmail-Version: 0.90.2.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigEBFE01B41C91D1AD1C2DFB6D" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.137 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/80 X-Sequence-Number: 13321 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigEBFE01B41C91D1AD1C2DFB6D Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit I'm putting together a road map on how our systems can scale as our load increases. As part of this, I need to look into setting up some fast read only mirrors of our database. We should have more than enough RAM to fit everything into memory. I would like to find out if I could expect better performance by mounting the database from a RAM disk, or if I would be better off keeping that RAM free and increasing the effective_cache_size appropriately. I'd also be interested in knowing if this is dependant on whether I am running 7.4, 8.0 or 8.1. -- Stuart Bishop http://www.stuartbishop.net/ --------------enigEBFE01B41C91D1AD1C2DFB6D Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) iD8DBQFCzde9AfqZj7rGN0oRAi34AJ9PMD44LBvppopJLYz+BL19HWGb+wCgkmsO g0oTZpwKfU+yRowWeExNv+w= =3Du5 -----END PGP SIGNATURE----- --------------enigEBFE01B41C91D1AD1C2DFB6D-- From pgsql-performance-owner@postgresql.org Fri Jul 8 00:05:41 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 32B2252A37 for ; Fri, 8 Jul 2005 00:05:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65031-07 for ; Fri, 8 Jul 2005 03:05:30 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 5BC7E52A01 for ; Fri, 8 Jul 2005 00:05:29 -0300 (ADT) Received: from [192.168.1.100] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6835SoQ020507; Thu, 7 Jul 2005 20:05:29 -0700 Message-ID: <42CDED86.3000108@commandprompt.com> Date: Thu, 07 Jul 2005 20:05:42 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Stuart Bishop Cc: pgsql-performance@postgresql.org Subject: Re: Mount database on RAM disk? References: <42CDD7B8.6050109@stuartbishop.net> In-Reply-To: <42CDD7B8.6050109@stuartbishop.net> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/81 X-Sequence-Number: 13322 Stuart Bishop wrote: > I'm putting together a road map on how our systems can scale as our load > increases. As part of this, I need to look into setting up some fast read > only mirrors of our database. We should have more than enough RAM to fit > everything into memory. I would like to find out if I could expect better > performance by mounting the database from a RAM disk, or if I would be > better off keeping that RAM free and increasing the effective_cache_size > appropriately. In theory yes if you can fit the entire database onto a ram disk then you would see a performance benefit. Sincerely, Joshua D. Drake > > I'd also be interested in knowing if this is dependant on whether I am > running 7.4, 8.0 or 8.1. > > -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Fri Jul 8 01:04:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 48D38529C8 for ; Fri, 8 Jul 2005 01:04:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42325-04 for ; Fri, 8 Jul 2005 04:04:10 +0000 (GMT) Received: from floppy.pyrenet.fr (news.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7473C52858 for ; Fri, 8 Jul 2005 01:04:08 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id 240AB30B41; Fri, 8 Jul 2005 06:08:07 +0200 (MET DST) From: Chris Browne X-Newsgroups: pgsql.performance Subject: Re: Mount database on RAM disk? Date: Thu, 07 Jul 2005 23:30:09 -0400 Organization: cbbrowne Computing Inc Lines: 30 Message-ID: <60zmsykm9q.fsf@dba2.int.libertyrms.com> References: <42CDD7B8.6050109@stuartbishop.net> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Complaints-To: usenet@news.hub.org User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4.17 (Jumbo Shrimp, linux) Cancel-Lock: sha1:WI0k/qpXJjfTMWU2RHxH3a6FG8A= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.136 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/82 X-Sequence-Number: 13323 stuart@stuartbishop.net (Stuart Bishop) writes: > I'm putting together a road map on how our systems can scale as our > load increases. As part of this, I need to look into setting up some > fast read only mirrors of our database. We should have more than > enough RAM to fit everything into memory. I would like to find out > if I could expect better performance by mounting the database from a > RAM disk, or if I would be better off keeping that RAM free and > increasing the effective_cache_size appropriately. If you were willing to take on a not-inconsiderable risk, I'd think that storing WAL files on a RAMDISK would be likely to be the fastest improvement imaginable. If I could get and deploy some SSD (Solid State Disk) devices that would make this sort of thing *actually safe,* I'd expect that to be a pretty fabulous improvement, at least for write-heavy database activity. > I'd also be interested in knowing if this is dependant on whether I > am running 7.4, 8.0 or 8.1. Behaviour of all three could be somewhat different, as management of the shared cache has been in flux... -- (format nil "~S@~S" "cbbrowne" "acm.org") http://www.ntlug.org/~cbbrowne/sap.html Rules of the Evil Overlord #78. "I will not tell my Legions of Terror "And he must be taken alive!" The command will be: ``And try to take him alive if it is reasonably practical.''" From pgsql-performance-owner@postgresql.org Fri Jul 8 11:47:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 52674529D7 for ; Fri, 8 Jul 2005 11:47:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 45239-10 for ; Fri, 8 Jul 2005 14:47:36 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 9BD785281A for ; Fri, 8 Jul 2005 11:47:34 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j68DkdWK008133 for pgsql-performance@postgresql.org; Fri, 8 Jul 2005 15:46:39 +0200 Date: Fri, 8 Jul 2005 15:46:39 +0200 From: Enrico Weigelt To: postgresql performance list Subject: Re: plain inserts and deletes very slow Message-ID: <20050708134638.GA6368@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: postgresql performance list References: <42C8B23F.4000400@telogis.com> <20050704085903.GG9777@nibiru.borg.metux.de> <42C9C7E278.E5BFKG@129.180.47.120> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42C9C7E278.E5BFKG@129.180.47.120> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/84 X-Sequence-Number: 13325 * Klint Gore wrote: > Turn on statement logging. I've seen delphi interfaces do extra queries > on system tables to find some structure information. I'm already using statement logging of all queries taking longer than 200ms. It seems that only the INSERT takes such a time. The client is in fact written in delphi, and it sometimes seems to do strange things. For example we had the effect, that some new fields in some table were regularily NULL'ed. None of the triggers and rules inside the DB could do that (since there's no dynamic query stuff) and the delphi application is the only one writing directly to this table. cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Fri Jul 8 12:01:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 22BB652A79 for ; Fri, 8 Jul 2005 12:01:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51557-09 for ; Fri, 8 Jul 2005 15:01:28 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id E372E52A62 for ; Fri, 8 Jul 2005 12:01:26 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j68E0Pw8009760; Fri, 8 Jul 2005 16:00:25 +0200 Date: Fri, 8 Jul 2005 16:00:24 +0200 From: Enrico Weigelt To: pgsql-performance@postgresql.org Cc: Bruno Wolff III , mark durrant Subject: Re: Select performance vs. mssql Message-ID: <20050708140024.GB6368@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: pgsql-performance@postgresql.org, Bruno Wolff III , mark durrant References: <20050524153637.64525.qmail@web53101.mail.yahoo.com> <20050524172713.GC8129@wolff.to> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050524172713.GC8129@wolff.to> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/85 X-Sequence-Number: 13326 * Bruno Wolff III wrote: > This gets brought up a lot. The problem is that the index doesn't include > information about whether the current transaction can see the referenced > row. Putting this information in the index will add significant overhead > to every update and the opinion of the developers is that this would be > a net loss overall. wouldn't it work well to make this feature optionally for each index ? There could be some flag on the index (ie set at create time) which tells postgres whether to store mvcc information. cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Fri Jul 8 11:07:10 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6332E52A0E for ; Fri, 8 Jul 2005 11:07:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31283-06 for ; Fri, 8 Jul 2005 14:06:59 +0000 (GMT) Received: from gatekeeper.look.ca (gamma1.look.ca [207.136.80.8]) by svr1.postgresql.org (Postfix) with ESMTP id 023CC52815 for ; Fri, 8 Jul 2005 11:06:57 -0300 (ADT) Received: from 216-154-13-109.dsl.look.ca ([216.154.13.109]) by gatekeeper.look.ca with esmtp (Exim 4.20) id 1DqtVN-0006ld-2d; Fri, 08 Jul 2005 14:07:01 +0000 From: Rod Taylor To: Chris Browne Cc: pgsql-performance@postgresql.org In-Reply-To: <60zmsykm9q.fsf@dba2.int.libertyrms.com> References: <42CDD7B8.6050109@stuartbishop.net> <60zmsykm9q.fsf@dba2.int.libertyrms.com> Date: Fri, 08 Jul 2005 10:03:48 -0400 Message-Id: <1120831428.24708.228.camel@home> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 FreeBSD GNOME Team Port X-SA-Exim-Mail-From: pg@rbt.ca Subject: Re: Mount database on RAM disk? Content-Type: text/plain Content-Transfer-Encoding: 7bit X-SA-Exim-Version: 3.1 (built Tue Feb 24 05:09:27 GMT 2004) X-SA-Exim-Scanned: Yes X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.035 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/83 X-Sequence-Number: 13324 > If I could get and deploy some SSD (Solid State Disk) devices that > would make this sort of thing *actually safe,* I'd expect that to be a > pretty fabulous improvement, at least for write-heavy database > activity. Not nearly as much as you would expect. For the price of the SSD and a SCSI controller capable of keeping up to the SSD along with your regular storage with enough throughput to keep up to structure IO, you can purchase a pretty good mid-range SAN which will be just as capable and much more versatile. -- From pgsql-performance-owner@postgresql.org Fri Jul 8 12:44:45 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B620A52AD6 for ; Fri, 8 Jul 2005 12:44:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 27159-01 for ; Fri, 8 Jul 2005 15:44:42 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id 3E00452AD9 for ; Fri, 8 Jul 2005 12:44:37 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j68Ehbb2017967 for pgsql-performance@postgresql.org; Fri, 8 Jul 2005 16:43:37 +0200 Date: Fri, 8 Jul 2005 16:43:36 +0200 From: Enrico Weigelt To: pgsql-performance@postgresql.org Subject: Re: Need help to decide Mysql vs Postgres Message-ID: <20050708144336.GC6368@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: pgsql-performance@postgresql.org References: <001c01c56acb$bc9ba730$797ba8c0@jfradkin> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/86 X-Sequence-Number: 13327 * PFC wrote: > For Python it's the reverse : the MySQL driver is slow and dumb, > and the postgres driver (psycopg 2) is super fast, handles all quoting, > and knows about type conversions, it will automatically convert a > Python List into a postgres Array and do the right thing with quoting, > and it works both ways (ie you select a TEXT[] you get a list of > strings all parsed for you). It knows about all the postgres types (yes > even numeric <=> python Decimal) and you can even add your own types. > That's really cool, plus the developer is a friendly guy. Is there anything similar for java ? cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Fri Jul 8 13:37:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4E3A652ADD for ; Fri, 8 Jul 2005 13:37:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 98202-02 for ; Fri, 8 Jul 2005 16:37:46 +0000 (GMT) Received: from metux.de (seven.metux.de [193.16.1.1]) by svr1.postgresql.org (Postfix) with ESMTP id BF5B652A98 for ; Fri, 8 Jul 2005 13:37:41 -0300 (ADT) Received: (from weigelt@localhost) by metux.de (8.12.10/8.12.10) id j68FacK8027693 for pgsql-performance@postgresql.org; Fri, 8 Jul 2005 17:36:38 +0200 Date: Fri, 8 Jul 2005 17:36:38 +0200 From: Enrico Weigelt To: pgsql-performance@postgresql.org Subject: Re: Why the planner is not using the INDEX . Message-ID: <20050708153638.GD6368@nibiru.borg.metux.de> Reply-To: weigelt@metux.de Mail-Followup-To: pgsql-performance@postgresql.org References: <20050630222451.GA17628@calvin.surfutopia.net> <42C473BA.9090305@commandprompt.com> <20050630235821.GA2163@calvin.surfutopia.net> <42C994BD.3000909@siunik.com> <23492.1120579796@sss.pgh.pa.us> <42CAC937.4080702@siunik.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42CAC937.4080702@siunik.com> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.107 tagged_above=0 required=5 tests=AWL, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/90 X-Sequence-Number: 13331 * David Gagnon wrote: > FOR inventoryTransaction IN > SELECT DISTINCT IRNUM, IRAENUM, IRSTATUT, IRSENS, IRSOURCE, > IRDATE, IRQTE > FROM IR > WHERE IRNUM = ANY (requestIds) and IRYPNUM = companyId > LOOP hmm. you probably could create the query dynamically and then execute it. BTW: why isn't IN not usable with arrays ? cu -- --------------------------------------------------------------------- Enrico Weigelt == metux IT service phone: +49 36207 519931 www: http://www.metux.de/ fax: +49 36207 519932 email: contact@metux.de --------------------------------------------------------------------- Realtime Forex/Stock Exchange trading powered by postgresSQL :)) http://www.fxignal.net/ --------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Fri Jul 8 13:08:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2B46752AD7 for ; Fri, 8 Jul 2005 13:08:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91877-03 for ; Fri, 8 Jul 2005 16:08:22 +0000 (GMT) Received: from mir3-fs.mir3.com (www.mir3.com [216.74.11.46]) by svr1.postgresql.org (Postfix) with ESMTP id 9823B52AF6 for ; Fri, 8 Jul 2005 13:08:14 -0300 (ADT) Received: from archimedes.mirlogic.com ([172.16.2.68]) by mir3-fs.mir3.com with Microsoft SMTPSVC(5.0.2195.6713); Fri, 8 Jul 2005 09:11:03 -0700 Subject: Re: Need help to decide Mysql vs Postgres From: Mark Lewis To: weigelt@metux.de Cc: pgsql-performance@postgresql.org In-Reply-To: <20050708144336.GC6368@nibiru.borg.metux.de> References: <001c01c56acb$bc9ba730$797ba8c0@jfradkin> <20050708144336.GC6368@nibiru.borg.metux.de> Content-Type: text/plain Organization: MIR3, Inc. Date: Fri, 08 Jul 2005 09:08:10 -0700 Message-Id: <1120838891.3326.79.camel@archimedes.mirlogic.com> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-16) Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 08 Jul 2005 16:11:03.0109 (UTC) FILETIME=[A3506750:01C583D7] X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/87 X-Sequence-Number: 13328 On Fri, 2005-07-08 at 16:43 +0200, Enrico Weigelt wrote: > * PFC wrote: > > > > For Python it's the reverse : the MySQL driver is slow and dumb, > > and the postgres driver (psycopg 2) is super fast, handles all quoting, > > and knows about type conversions, it will automatically convert a > > Python List into a postgres Array and do the right thing with quoting, > > and it works both ways (ie you select a TEXT[] you get a list of > > strings all parsed for you). It knows about all the postgres types (yes > > even numeric <=> python Decimal) and you can even add your own types. > > That's really cool, plus the developer is a friendly guy. > > Is there anything similar for java ? > The postgres JDBC driver is very good-- refer to pgsql-jdbc mailing list or look at jdbc.postgresql.org. I've had only limited experience with the mysql jdbc driver, but it seemed servicable enough, if you can live with their licensing and feature set. From pgsql-performance-owner@postgresql.org Fri Jul 8 13:23:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EEF7452A72 for ; Fri, 8 Jul 2005 13:23:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94739-07 for ; Fri, 8 Jul 2005 16:23:04 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.202]) by svr1.postgresql.org (Postfix) with ESMTP id 78F7852A26 for ; Fri, 8 Jul 2005 13:23:03 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i31so456845wra for ; Fri, 08 Jul 2005 09:23:03 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=Dw3ZQKAKAna4HtrkGP05mcCztrJRoCd+RMZoZ4fYMmOYSSuskfqLdZemev7ZiROYBncrkE6ENsH4clIE7lAy6eANJItxrRVa0qtXKKLB6mOp7Ns5Ve/V+VV25yxbJ2te6xDO4fwc51u2oZxzH5uR4TODh1LFiSK4pWhO6vJPe/I= Received: by 10.54.16.77 with SMTP id 77mr300612wrp; Fri, 08 Jul 2005 09:22:38 -0700 (PDT) Received: by 10.54.30.27 with HTTP; Fri, 8 Jul 2005 09:22:38 -0700 (PDT) Message-ID: <5db591c00507080922236f9192@mail.gmail.com> Date: Fri, 8 Jul 2005 11:22:38 -0500 From: Moises Alberto Lindo Gutarra Reply-To: Moises Alberto Lindo Gutarra To: weigelt@metux.de, pgsql-performance@postgresql.org Subject: Re: Need help to decide Mysql vs Postgres In-Reply-To: <1120838891.3326.79.camel@archimedes.mirlogic.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <001c01c56acb$bc9ba730$797ba8c0@jfradkin> <20050708144336.GC6368@nibiru.borg.metux.de> <1120838891.3326.79.camel@archimedes.mirlogic.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.007 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP, RCVD_IN_BL_SPAMCOP_NET X-Spam-Level: * X-Archive-Number: 200507/88 X-Sequence-Number: 13329 Linux(Debian) + Java + PostgreSQL =3D Fastest 2005/7/8, Mark Lewis : > On Fri, 2005-07-08 at 16:43 +0200, Enrico Weigelt wrote: > > * PFC wrote: > > > > > > > For Python it's the reverse : the MySQL driver is slow and dumb, > > > and the postgres driver (psycopg 2) is super fast, handles all q= uoting, > > > and knows about type conversions, it will automatically convert a > > > Python List into a postgres Array and do the right thing with quotin= g, > > > and it works both ways (ie you select a TEXT[] you get a list of > > > strings all parsed for you). It knows about all the postgres types (= yes > > > even numeric <=3D> python Decimal) and you can even add your own typ= es. > > > That's really cool, plus the developer is a friendly guy. > > > > Is there anything similar for java ? > > >=20 > The postgres JDBC driver is very good-- refer to pgsql-jdbc mailing list > or look at jdbc.postgresql.org. I've had only limited experience with > the mysql jdbc driver, but it seemed servicable enough, if you can live > with their licensing and feature set. >=20 >=20 >=20 >=20 > ---------------------------(end of broadcast)--------------------------- > TIP 5: don't forget to increase your free space map settings >=20 --=20 Atte Moises Alberto Lindo Gutarra Consultor y Desarrollador Java / Open Source TUMI Solutions SAC Tel: +51.13481104 Cel: +51.197366260=20 MSN : mlindo@tumisolutions.com From pgsql-performance-owner@postgresql.org Fri Jul 8 13:25:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5E70152990 for ; Fri, 8 Jul 2005 13:25:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94717-09 for ; Fri, 8 Jul 2005 16:25:05 +0000 (GMT) Received: from mailhost.intellivid.com (mailhost.intellivid.com [64.32.200.11]) by svr1.postgresql.org (Postfix) with ESMTP id 0CEEE52A0E for ; Fri, 8 Jul 2005 13:25:03 -0300 (ADT) Received: from [192.168.2.68] (spectre.intellivid.com [192.168.2.68]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by newmail.intellivid.com (Postfix) with ESMTP id D68D4F1829D for ; Fri, 8 Jul 2005 12:25:02 -0400 (EDT) Subject: cost-based vacuum From: Ian Westmacott To: pgsql-performance@postgresql.org Content-Type: text/plain Organization: Intellivid Corp. Message-Id: <1120839902.20657.197.camel@spectre.intellivid.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Fri, 08 Jul 2005 12:25:02 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/89 X-Sequence-Number: 13330 I am beginning to look at Postgres 8, and am particularly interested in cost-based vacuum/analyze. I'm hoping someone can shed some light on the behavior I am seeing. Suppose there are three threads: writer_thread every 1/15 second do BEGIN TRANSACTION COPY table1 FROM stdin ... COPY tableN FROM stdin perform several UPDATEs, DELETEs and INSERTs COMMIT reader_thread every 1/15 second do BEGIN TRANSACTION SELECT FROM table1 ... ... SELECT FROM tableN ... COMMIT analyze_thread every 5 minutes do ANALYZE table1 ... ANALYZE tableN Now, Postgres 8.0.3 out-of-the-box (all default configs) on a particular piece of hardware runs the Postgres connection for writer_thread at about 15% CPU (meaningless, I know, but for comparison) and runs the Postgres connection for reader_thread at about 30% CPU. Latency for reader_thread seeing updates from writer_thread is well under 1/15s. Impact of analyze_thread is negligible. If I make the single configuration change of setting vacuum_cost_delay=1000, each iteration in analyze_thread takes much longer, of course. But what I also see is that the CPU usage of the connections for writer_thread and reader_thread spike up to well over 80% each (this is a dualie) and latency drops to 8-10s, during the ANALYZEs. I don't understand why this would be. I don't think there are any lock issues, and I don't see any obvious I/O issues. Am I missing something? Is there any way to get some insight into what those connections are doing? Thanks, --Ian From pgsql-performance-owner@postgresql.org Fri Jul 8 14:19:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DB96F52ADA for ; Fri, 8 Jul 2005 14:19:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44635-05 for ; Fri, 8 Jul 2005 17:19:49 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id C162152ABD for ; Fri, 8 Jul 2005 14:19:43 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7586968; Fri, 08 Jul 2005 10:21:54 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Re: Mount database on RAM disk? Date: Fri, 8 Jul 2005 10:22:01 -0700 User-Agent: KMail/1.8 Cc: Stuart Bishop References: <42CDD7B8.6050109@stuartbishop.net> In-Reply-To: <42CDD7B8.6050109@stuartbishop.net> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507081022.01680.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/91 X-Sequence-Number: 13332 Stuart, > I'm putting together a road map on how our systems can scale as our load > increases. As part of this, I need to look into setting up some fast > read only mirrors of our database. We should have more than enough RAM > to fit everything into memory. I would like to find out if I could > expect better performance by mounting the database from a RAM disk, or > if I would be better off keeping that RAM free and increasing the > effective_cache_size appropriately. If you're accessing a dedicated, read-only system with a database small enough to fit in RAM, it'll all be cached there anyway, at least on Linux and BSD. You won't be gaining anything by creating a ramdisk. BTW, effective_cache_size doesn't determine the amount of caching done. It just informs the planner about how much db is likely to be cached. The actual caching is up to the OS/filesystem. -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Jul 8 14:48:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3BF5752A99 for ; Fri, 8 Jul 2005 14:48:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58872-09 for ; Fri, 8 Jul 2005 17:48:24 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id DD48852A79 for ; Fri, 8 Jul 2005 14:48:23 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j68HmMi8024554; Fri, 8 Jul 2005 13:48:22 -0400 (EDT) To: Ian Westmacott Cc: pgsql-performance@postgresql.org Subject: Re: cost-based vacuum In-reply-to: <1120839902.20657.197.camel@spectre.intellivid.com> References: <1120839902.20657.197.camel@spectre.intellivid.com> Comments: In-reply-to Ian Westmacott message dated "Fri, 08 Jul 2005 12:25:02 -0400" Date: Fri, 08 Jul 2005 13:48:22 -0400 Message-ID: <24553.1120844902@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/92 X-Sequence-Number: 13333 Ian Westmacott writes: > If I make the single configuration change of setting > vacuum_cost_delay=1000, each iteration in analyze_thread takes > much longer, of course. But what I also see is that the CPU > usage of the connections for writer_thread and reader_thread > spike up to well over 80% each (this is a dualie) and latency > drops to 8-10s, during the ANALYZEs. [ scratches head... ] That doesn't make any sense at all. > I don't understand why this would be. I don't think there > are any lock issues, and I don't see any obvious I/O issues. > Am I missing something? Is there any way to get some > insight into what those connections are doing? Profiling maybe? Can you put together a self-contained test case that replicates this behavior, so other people could look? regards, tom lane From pgsql-performance-owner@postgresql.org Fri Jul 8 16:21:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4D3FD52AEA for ; Fri, 8 Jul 2005 16:21:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 80632-06 for ; Fri, 8 Jul 2005 19:21:23 +0000 (GMT) Received: from Herge.rcsinc.local (mail.rcsonline.com [205.217.85.91]) by svr1.postgresql.org (Postfix) with ESMTP id B46A25282F for ; Fri, 8 Jul 2005 16:21:21 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Subject: Re: Mount database on RAM disk? X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Date: Fri, 8 Jul 2005 15:21:21 -0400 Message-ID: <6EE64EF3AB31D5448D0007DD34EEB3415C2CFE@Herge.rcsinc.local> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Mount database on RAM disk? Thread-Index: AcWD4ZHX+xf/FouUSYSUAOgtXWoebQAD8weg From: "Merlin Moncure" To: Cc: "Stuart Bishop" , X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/93 X-Sequence-Number: 13334 > Stuart, >=20 > > I'm putting together a road map on how our systems can scale as our load > > increases. As part of this, I need to look into setting up some fast > > read only mirrors of our database. We should have more than enough RAM > > to fit everything into memory. I would like to find out if I could > > expect better performance by mounting the database from a RAM disk, or > > if I would be better off keeping that RAM free and increasing the > > effective_cache_size appropriately. >=20 > If you're accessing a dedicated, read-only system with a database small > enough to fit in RAM, it'll all be cached there anyway, at least on Linux > and BSD. You won't be gaining anything by creating a ramdisk. =20 ditto windows. =20 Files cached in memory are slower than reading straight from memory but not nearly enough to justify reserving memory for your use. In other words, your O/S is a machine with years and years of engineering designed best how to dole memory out to caching and various processes. Why second guess it? Merlin From pgsql-performance-owner@postgresql.org Fri Jul 8 18:47:41 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5615F52AF2 for ; Fri, 8 Jul 2005 18:47:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68888-09 for ; Fri, 8 Jul 2005 21:47:31 +0000 (GMT) Received: from mailservice.tudelft.nl (mailservice.tudelft.nl [130.161.131.5]) by svr1.postgresql.org (Postfix) with ESMTP id 65DA552AF7 for ; Fri, 8 Jul 2005 18:47:30 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by rav.antivirus (Postfix) with ESMTP id 3EB6E22F1A8 for ; Fri, 8 Jul 2005 23:47:32 +0200 (CEST) Received: from [145.94.90.156] (jochemd.tnw-s.tudelft.nl [145.94.90.156]) by mx1.tudelft.nl (Postfix) with ESMTP id 9595D22F16F for ; Fri, 8 Jul 2005 23:47:30 +0200 (CEST) Message-ID: <42CEF46F.4060308@oli.tudelft.nl> Date: Fri, 08 Jul 2005 23:47:27 +0200 From: Jochem van Dieten Organization: OnLine Internet User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Select performance vs. mssql References: <20050708140024.GB6368@nibiru.borg.metux.de> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at tudelft.nl X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/94 X-Sequence-Number: 13335 Enrico Weigelt wrote: > Bruno Wolff III wrote: >> >> This gets brought up a lot. The problem is that the index doesn't include >> information about whether the current transaction can see the referenced >> row. Putting this information in the index will add significant overhead >> to every update and the opinion of the developers is that this would be >> a net loss overall. > > wouldn't it work well to make this feature optionally for each > index ? There could be some flag on the index (ie set at create > time) which tells postgres whether to store mvcc information. There is no reason to assume it can't work. There is little reason to assume that it will be the best solution in many circumstances. There is a big reason why people are sceptical: there is no patch. The issue has been debated and beaten to death. People have formed their opinions and are unlikely to change their position. If you want to convince people, your best bet is to submit a patch and have OSDL measure the performance improvement. Jochem From pgsql-performance-owner@postgresql.org Fri Jul 8 20:57:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5D32C52AF8 for ; Fri, 8 Jul 2005 20:57:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07592-07 for ; Fri, 8 Jul 2005 23:57:06 +0000 (GMT) Received: from wsmailap01.firstam.com (outbound-smtp01.firstam.com [208.246.101.197]) by svr1.postgresql.org (Postfix) with ESMTP id DC7FA5282F for ; Fri, 8 Jul 2005 20:57:01 -0300 (ADT) Received: from 172.21.131.5 by wsmailap01.firstam.com with ESMTP ( Tumbleweed MMS SMTP Relay (MMS v5.6.3)); Fri, 08 Jul 2005 16:56:52 -0700 X-Server-Uuid: 2A3A6F04-3324-429D-9DE4-35238E3EA19C Received: from anammx01.ana.firstamdata.com ([192.168.173.35]) by famarp01.firstam.com (MOS 3.5.6-GR) with SMTP id EBG53698; Fri, 8 Jul 2005 16:56:54 -0700 (PDT) Received: from pisgana01sxch21.ana.firstamdata.com (Not Verified[192.168.173.71]) by mailgateway.firstam.com (post.office MTA v5.0 0924 ) with ESMTP id ; Fri, 08 Jul 2005 16:56:54 -0700 Received: from pisgana01sxch01.ana.firstamdata.com ([192.168.173.70]) by pisgana01sxch21.ana.firstamdata.com with Microsoft SMTPSVC(6.0.3790.211 ); Fri, 8 Jul 2005 16:56:54 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: How to revoke a password Date: Fri, 8 Jul 2005 16:56:53 -0700 Message-ID: Thread-Topic: How to revoke a password Thread-Index: AcWEGLckZtmr1XeqTb6IvNYHSkBD6w== From: "Bailey, Larry" To: pgsql-performance@postgresql.org X-OriginalArrivalTime: 08 Jul 2005 23:56:54.0221 (UTC) FILETIME=[B7799FD0:01C58418] X-WSS-ID: 6ED1CD4E1GS114755-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/95 X-Sequence-Number: 13336 I created a user with a password. That newly created user now have tables and indexes. I want to ALTER that user to exclude the password. How is this accomplished without dropping and recreating the users? Larry Bailey Sr. Oracle DBA First American Real Estate Solution (714) 701-3347 lbailey@firstam.com=20 ********************************************************************** This message contains confidential information intended only for the=20 use of the addressee(s) named above and may contain information that=20 is legally privileged. If you are not the addressee, or the person=20 responsible for delivering it to the addressee, you are hereby=20 notified that reading, disseminating, distributing or copying this=20 message is strictly prohibited. If you have received this message by=20 mistake, please immediately notify us by replying to the message and=20 delete the original message immediately thereafter. Thank you. FADLD Tag ********************************************************************** From pgsql-performance-owner@postgresql.org Fri Jul 8 21:09:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E06D152B00 for ; Fri, 8 Jul 2005 21:09:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 12985-01 for ; Sat, 9 Jul 2005 00:09:48 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 9477952B09 for ; Fri, 8 Jul 2005 21:09:48 -0300 (ADT) Received: from [192.168.1.100] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6909loQ001117; Fri, 8 Jul 2005 17:09:51 -0700 Message-ID: <42CF15CC.9040203@commandprompt.com> Date: Fri, 08 Jul 2005 17:09:48 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Bailey, Larry" Cc: pgsql-performance@postgresql.org Subject: Re: How to revoke a password References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/96 X-Sequence-Number: 13337 Bailey, Larry wrote: > I created a user with a password. That newly created user now have > tables and indexes. I want to ALTER that user to exclude the password. > How is this accomplished without dropping and recreating the users? Never tried to go backwards before but: alter user foo with encrypted password ''; But as I look at pg_shadow there is still a hash... You could do: update pg_shadow set passwd = '' where usename = 'foo'; Sincerely, Joshua D. Drake > > Larry Bailey > Sr. Oracle DBA > First American Real Estate Solution > (714) 701-3347 > lbailey@firstam.com > ********************************************************************** > This message contains confidential information intended only for the > use of the addressee(s) named above and may contain information that > is legally privileged. If you are not the addressee, or the person > responsible for delivering it to the addressee, you are hereby > notified that reading, disseminating, distributing or copying this > message is strictly prohibited. If you have received this message by > mistake, please immediately notify us by replying to the message and > delete the original message immediately thereafter. > > Thank you. FADLD Tag > ********************************************************************** > > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Fri Jul 8 21:16:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1959352A03 for ; Fri, 8 Jul 2005 21:16:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07822-09 for ; Sat, 9 Jul 2005 00:16:37 +0000 (GMT) Received: from wsmailap02.firstam.com (outbound-smtp02.firstam.com [208.246.101.198]) by svr1.postgresql.org (Postfix) with ESMTP id B45AF52B00 for ; Fri, 8 Jul 2005 21:16:33 -0300 (ADT) Received: from 172.21.131.6 by wsmailap01.firstam.com with ESMTP ( Tumbleweed MMS SMTP Relay (MMS v5.6.3)); Fri, 08 Jul 2005 17:16:25 -0700 X-Server-Uuid: 2A3A6F04-3324-429D-9DE4-35238E3EA19C Received: from anammx01.ana.firstamdata.com ([192.168.173.35]) by famarp02.firstam.com (MOS 3.5.6-GR) with SMTP id JCL60623; Fri, 8 Jul 2005 17:16:28 -0700 (PDT) Received: from pisgana01sxch21.ana.firstamdata.com (Not Verified[192.168.173.71]) by mailgateway.firstam.com (post.office MTA v5.0 0924 ) with ESMTP id ; Fri, 08 Jul 2005 17:16:28 -0700 Received: from pisgana01sxch01.ana.firstamdata.com ([192.168.173.70]) by pisgana01sxch21.ana.firstamdata.com with Microsoft SMTPSVC(6.0.3790.211 ); Fri, 8 Jul 2005 17:16:27 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: How to revoke a password Date: Fri, 8 Jul 2005 17:16:27 -0700 Message-ID: Thread-Topic: [PERFORM] How to revoke a password Thread-Index: AcWEGo2XOEuF2erwQ0KELy1SMwep4QAAM+Mg From: "Bailey, Larry" To: "Joshua D. Drake" Cc: pgsql-performance@postgresql.org X-OriginalArrivalTime: 09 Jul 2005 00:16:27.0589 (UTC) FILETIME=[72DB5F50:01C5841B] X-WSS-ID: 6ED1C8D31GS118751-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/97 X-Sequence-Number: 13338 Thanks but it is still prompting for a password.=20 Larry Bailey Sr. Oracle DBA First American Real Estate Solution (714) 701-3347 lbailey@firstam.com=20 -----Original Message----- From: Joshua D. Drake [mailto:jd@commandprompt.com]=20 Sent: Friday, July 08, 2005 5:10 PM To: Bailey, Larry Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] How to revoke a password Bailey, Larry wrote: > I created a user with a password. That newly created user now have=20 > tables and indexes. I want to ALTER that user to exclude the password. > How is this accomplished without dropping and recreating the users? Never tried to go backwards before but: alter user foo with encrypted password ''; But as I look at pg_shadow there is still a hash... You could do: update pg_shadow set passwd =3D '' where usename =3D 'foo'; Sincerely, Joshua D. Drake >=20 > Larry Bailey > Sr. Oracle DBA > First American Real Estate Solution > (714) 701-3347 > lbailey@firstam.com > ********************************************************************** > This message contains confidential information intended only for the=20 > use of the addressee(s) named above and may contain information that=20 > is legally privileged. If you are not the addressee, or the person=20 > responsible for delivering it to the addressee, you are hereby=20 > notified that reading, disseminating, distributing or copying this=20 > message is strictly prohibited. If you have received this message by=20 > mistake, please immediately notify us by replying to the message and=20 > delete the original message immediately thereafter. >=20 > Thank you. FADLD Tag > ********************************************************************** >=20 >=20 > ---------------------------(end of=20 > broadcast)--------------------------- > TIP 6: explain analyze is your friend -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Fri Jul 8 21:17:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F18CE52A03 for ; Fri, 8 Jul 2005 21:17:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 12844-10 for ; Sat, 9 Jul 2005 00:17:45 +0000 (GMT) Received: from mr2.surnet.cl (smtp2.surnet.cl [216.155.73.163]) by svr1.postgresql.org (Postfix) with ESMTP id 242DE5282F for ; Fri, 8 Jul 2005 21:17:44 -0300 (ADT) Received: from smtp2.surnet.cl (216.155.73.169) by mr2.surnet.cl (7.0.031.3) id 4259ADFA0141AF22; Fri, 8 Jul 2005 20:17:26 -0400 Received: from smtp2.surnet.cl (mr2.surnet.cl []) by mr2.surnet.cl ([216.155.73.169]); Sat, 09 Jul 2005 00:17:26 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp2.surnet.cl (7.0.031.3) id 4259ADE300C499B4; Fri, 8 Jul 2005 20:17:26 -0400 Received: from alvh.no-ip.org (216.155.85.95) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60002F08C0; Fri, 8 Jul 2005 20:17:38 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 82085C2DC70; Fri, 8 Jul 2005 20:17:43 -0400 (CLT) Date: Fri, 8 Jul 2005 20:17:43 -0400 From: Alvaro Herrera To: "Joshua D. Drake" Cc: "Bailey, Larry" , pgsql-performance@postgresql.org Subject: Re: How to revoke a password Message-ID: <20050709001743.GA10063@alvh.no-ip.org> References: <42CF15CC.9040203@commandprompt.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <42CF15CC.9040203@commandprompt.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.742 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/98 X-Sequence-Number: 13339 On Fri, Jul 08, 2005 at 05:09:48PM -0700, Joshua D. Drake wrote: > Bailey, Larry wrote: > >I created a user with a password. That newly created user now have > >tables and indexes. I want to ALTER that user to exclude the password. > >How is this accomplished without dropping and recreating the users? > > Never tried to go backwards before but: > > alter user foo with encrypted password ''; I think you use NULL as password to ALTER USER. -- Alvaro Herrera () "Y eso te lo doy firmado con mis l�grimas" (Fiebre del Loco) From pgsql-performance-owner@postgresql.org Fri Jul 8 21:56:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D119252A78 for ; Fri, 8 Jul 2005 21:56:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 17225-08 for ; Sat, 9 Jul 2005 00:56:05 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id B4DC2529C8 for ; Fri, 8 Jul 2005 21:56:04 -0300 (ADT) Received: from [192.168.1.100] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j690u4oQ004071; Fri, 8 Jul 2005 17:56:07 -0700 Message-ID: <42CF20A6.8010901@commandprompt.com> Date: Fri, 08 Jul 2005 17:56:06 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Bailey, Larry" Cc: pgsql-performance@postgresql.org Subject: Re: How to revoke a password References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/99 X-Sequence-Number: 13340 Bailey, Larry wrote: > Thanks but it is still prompting for a password. > Does your pg_hba.conf require a password? Sincerely, Joshua D. Drake > > Larry Bailey > Sr. Oracle DBA > First American Real Estate Solution > (714) 701-3347 > lbailey@firstam.com > -----Original Message----- > From: Joshua D. Drake [mailto:jd@commandprompt.com] > Sent: Friday, July 08, 2005 5:10 PM > To: Bailey, Larry > Cc: pgsql-performance@postgresql.org > Subject: Re: [PERFORM] How to revoke a password > > Bailey, Larry wrote: > >>I created a user with a password. That newly created user now have >>tables and indexes. I want to ALTER that user to exclude the password. >>How is this accomplished without dropping and recreating the users? > > > Never tried to go backwards before but: > > alter user foo with encrypted password ''; > > But as I look at pg_shadow there is still a hash... > > You could do: > > update pg_shadow set passwd = '' where usename = 'foo'; > > Sincerely, > > Joshua D. Drake > > > >>Larry Bailey >>Sr. Oracle DBA >>First American Real Estate Solution >>(714) 701-3347 >>lbailey@firstam.com >>********************************************************************** >>This message contains confidential information intended only for the >>use of the addressee(s) named above and may contain information that >>is legally privileged. If you are not the addressee, or the person >>responsible for delivering it to the addressee, you are hereby >>notified that reading, disseminating, distributing or copying this >>message is strictly prohibited. If you have received this message by >>mistake, please immediately notify us by replying to the message and >>delete the original message immediately thereafter. >> >>Thank you. FADLD Tag >>********************************************************************** >> >> >>---------------------------(end of >>broadcast)--------------------------- >>TIP 6: explain analyze is your friend > > > > -- > Your PostgreSQL solutions provider, Command Prompt, Inc. > 24x7 support - 1.800.492.2240, programming, and consulting Home of > PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit > http://www.commandprompt.com / http://www.postgresql.org > > > > ---------------------------(end of broadcast)--------------------------- > TIP 5: don't forget to increase your free space map settings -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Fri Jul 8 21:58:24 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9D9AD52AEF for ; Fri, 8 Jul 2005 21:58:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28175-04 for ; Sat, 9 Jul 2005 00:58:20 +0000 (GMT) Received: from tigger.fuhr.org (tigger.fuhr.org [63.214.45.158]) by svr1.postgresql.org (Postfix) with ESMTP id ED37852AEC for ; Fri, 8 Jul 2005 21:58:19 -0300 (ADT) Received: from winnie.fuhr.org (winnie.fuhr.org [10.1.0.1]) by tigger.fuhr.org (8.13.3/8.13.3) with ESMTP id j690wGg2099724 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Fri, 8 Jul 2005 18:58:18 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: from winnie.fuhr.org (localhost [127.0.0.1]) by winnie.fuhr.org (8.13.3/8.13.3) with ESMTP id j690wGfw011609; Fri, 8 Jul 2005 18:58:16 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: (from mfuhr@localhost) by winnie.fuhr.org (8.13.3/8.13.3/Submit) id j690wFu8011608; Fri, 8 Jul 2005 18:58:15 -0600 (MDT) (envelope-from mfuhr) Date: Fri, 8 Jul 2005 18:58:15 -0600 From: Michael Fuhr To: "Bailey, Larry" Cc: "Joshua D. Drake" , pgsql-performance@postgresql.org Subject: Re: How to revoke a password Message-ID: <20050709005815.GA11532@winnie.fuhr.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/100 X-Sequence-Number: 13341 On Fri, Jul 08, 2005 at 05:16:27PM -0700, Bailey, Larry wrote: > > Thanks but it is still prompting for a password. Let's back up a bit: what problem are you trying to solve? Do you want the user to be able to log in without entering a password? If so then see "Client Authentication" in the documentation: http://www.postgresql.org/docs/8.0/static/client-authentication.html If you're trying to do something else then please elaborate, as it's not clear what you mean by "I want to ALTER that user to exclude the password." -- Michael Fuhr http://www.fuhr.org/~mfuhr/ From pgsql-performance-owner@postgresql.org Mon Jul 11 20:28:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 323135282C for ; Sat, 9 Jul 2005 15:38:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 61806-02 for ; Sat, 9 Jul 2005 18:38:47 +0000 (GMT) Received: from nextmail.ru (unknown [83.222.5.149]) by svr1.postgresql.org (Postfix) with SMTP id 2EABF52820 for ; Sat, 9 Jul 2005 15:38:42 -0300 (ADT) Received: (qmail 57103 invoked from network); 9 Jul 2005 18:38:51 -0000 Received: from ppp85-140-125-53.pppoe.mtu-net.ru (HELO z) (85.140.125.53) by nextmail.ru with SMTP; 9 Jul 2005 18:38:51 -0000 From: "jobapply" To: Subject: Sorting on longer key is faster ? Date: Sat, 9 Jul 2005 22:38:40 +0400 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.6353 Thread-Index: AcWEtWWnpTyLkFRvR2eGsKrTi7ZHLA== X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 Message-Id: <20050709183842.2EABF52820@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=2.078 tagged_above=0 required=5 tests=DNS_FROM_RFC_ABUSE, MSGID_FROM_MTA_ID X-Spam-Level: ** X-Archive-Number: 200507/112 X-Sequence-Number: 13353 The 2 queries are almost same, but ORDER BY x||t is FASTER than ORDER BY x.. How can that be possible? Btw: x and x||t are same ordered phoeniks=> explain analyze SELECT * FROM test WHERE i<20 ORDER BY x || t; QUERY PLAN ---------------------------------------------------------------------------- ---------------------------------------------- Sort (cost=2282.65..2284.92 rows=907 width=946) (actual time=74.982..79.114 rows=950 loops=1) Sort Key: (x || t) -> Index Scan using i_i on test (cost=0.00..2238.09 rows=907 width=946) (actual time=0.077..51.015 rows=950 loops=1) Index Cond: (i < 20) Total runtime: 85.944 ms (5 rows) phoeniks=> explain analyze SELECT * FROM test WHERE i<20 ORDER BY x; QUERY PLAN ---------------------------------------------------------------------------- --------------------------------------------- Sort (cost=2280.38..2282.65 rows=907 width=946) (actual time=175.431..179.239 rows=950 loops=1) Sort Key: x -> Index Scan using i_i on test (cost=0.00..2235.82 rows=907 width=946) (actual time=0.024..5.378 rows=950 loops=1) Index Cond: (i < 20) Total runtime: 183.317 ms (5 rows) phoeniks=> \d+ test Table "public.test" Column | Type | Modifiers | Description --------+---------+-----------+------------- i | integer | | t | text | | x | text | | Indexes: "i_i" btree (i) "x_i" btree (xpath_string(x, 'data'::text)) "x_ii" btree (xpath_string(x, 'movie/characters/character'::text)) Has OIDs: no From pgsql-performance-owner@postgresql.org Sat Jul 9 18:48:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9616152B12 for ; Sat, 9 Jul 2005 18:48:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92895-08 for ; Sat, 9 Jul 2005 21:48:43 +0000 (GMT) Received: from smtp-out5.blueyonder.co.uk (smtp-out5.blueyonder.co.uk [195.188.213.8]) by svr1.postgresql.org (Postfix) with ESMTP id E979F52B20 for ; Sat, 9 Jul 2005 18:48:41 -0300 (ADT) Received: from [192.168.1.10] ([82.43.93.192]) by smtp-out5.blueyonder.co.uk with Microsoft SMTPSVC(5.0.2195.6713); Sat, 9 Jul 2005 22:49:26 +0100 In-Reply-To: <6EE64EF3AB31D5448D0007DD34EEB3415C2CFE@Herge.rcsinc.local> References: <6EE64EF3AB31D5448D0007DD34EEB3415C2CFE@Herge.rcsinc.local> Mime-Version: 1.0 (Apple Message framework v730) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <41DED96F-84A2-436A-86C7-B3C2A3210591@advfn.com> Cc: , "Stuart Bishop" , Content-Transfer-Encoding: 7bit From: Alex Stapleton Subject: Re: Mount database on RAM disk? Date: Sat, 9 Jul 2005 22:48:43 +0100 To: Merlin Moncure X-Mailer: Apple Mail (2.730) X-OriginalArrivalTime: 09 Jul 2005 21:49:26.0686 (UTC) FILETIME=[139A23E0:01C584D0] X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/101 X-Sequence-Number: 13342 On 8 Jul 2005, at 20:21, Merlin Moncure wrote: >> Stuart, >> >> >>> I'm putting together a road map on how our systems can scale as our >>> > load > >>> increases. As part of this, I need to look into setting up some fast >>> read only mirrors of our database. We should have more than enough >>> > RAM > >>> to fit everything into memory. I would like to find out if I could >>> expect better performance by mounting the database from a RAM disk, >>> > or > >>> if I would be better off keeping that RAM free and increasing the >>> effective_cache_size appropriately. >>> >> >> If you're accessing a dedicated, read-only system with a database >> > small > >> enough to fit in RAM, it'll all be cached there anyway, at least on >> > Linux > >> and BSD. You won't be gaining anything by creating a ramdisk. >> > > > > ditto windows. > > Files cached in memory are slower than reading straight from memory > but > not nearly enough to justify reserving memory for your use. In other > words, your O/S is a machine with years and years of engineering > designed best how to dole memory out to caching and various processes. > Why second guess it? Because sometimes it gets it wrong. The most brutal method is occasionally the most desirable. Even if it not the "right" way to do it. > Merlin > > ---------------------------(end of > broadcast)--------------------------- > TIP 5: don't forget to increase your free space map settings > From pgsql-performance-owner@postgresql.org Sun Jul 10 01:50:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DE2655293B for ; Sun, 10 Jul 2005 01:50:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 10229-01 for ; Sun, 10 Jul 2005 04:50:09 +0000 (GMT) Received: from floppy.pyrenet.fr (floppy.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2A5515281A for ; Sun, 10 Jul 2005 01:50:06 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id 1EBD430B42; Sun, 10 Jul 2005 06:54:25 +0200 (MET DST) From: Christopher Browne X-Newsgroups: pgsql.performance Subject: Re: Mount database on RAM disk? Date: Sun, 10 Jul 2005 00:54:37 -0400 Organization: cbbrowne Computing Inc Lines: 31 Message-ID: References: <6EE64EF3AB31D5448D0007DD34EEB3415C2CFE@Herge.rcsinc.local> <41DED96F-84A2-436A-86C7-B3C2A3210591@advfn.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@news.hub.org X-message-flag: Outlook is rather hackable, isn't it? X-Home-Page: http://www.cbbrowne.com/info/ X-Affero: http://svcs.affero.net/rm.php?r=cbbrowne User-Agent: Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Jumbo Shrimp, berkeley-unix) Cancel-Lock: sha1:ZaBaYpuXjz9ZoRWJrULZb8VPKhI= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.135 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/102 X-Sequence-Number: 13343 > On 8 Jul 2005, at 20:21, Merlin Moncure wrote: >> ditto windows. >> >> Files cached in memory are slower than reading straight from memory >> but not nearly enough to justify reserving memory for your use. In >> other words, your O/S is a machine with years and years of >> engineering designed best how to dole memory out to caching and >> various processes. Why second guess it? > > Because sometimes it gets it wrong. The most brutal method is > occasionally the most desirable. Even if it not the "right" way to do > it. The fact that cache allows reads to come from memory means that for read-oriented activity, you're generally going to be better off leaving RAM as "plain ordinary system memory" so that it can automatically be drawn into service as cache. Thus, the main reason to consider using a RAM-disk is the fact that update times are negligible as there is not the latency of a round-trip to the disk. That would encourage its use for write-heavy tables, with the STRONG caveat that a power outage could readily destroy the database :-(. -- let name="cbbrowne" and tld="acm.org" in String.concat "@" [name;tld];; http://cbbrowne.com/info/rdbms.html Rules of the Evil Overlord #153. "My Legions of Terror will be an equal-opportunity employer. Conversely, when it is prophesied that no man can defeat me, I will keep in mind the increasing number of non-traditional gender roles." From pgsql-performance-owner@postgresql.org Mon Jul 11 20:29:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 33BEB52B80 for ; Sun, 10 Jul 2005 19:07:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 10162-01 for ; Sun, 10 Jul 2005 22:07:20 +0000 (GMT) Received: from r163ip.btv.lv (r163ip.btv.lv [217.198.224.163]) by svr1.postgresql.org (Postfix) with ESMTP id BF84352ABB for ; Sun, 10 Jul 2005 19:07:18 -0300 (ADT) Received: from p4 (pIV.homenet [192.168.88.4]) by r163ip.btv.lv (8.12.9/8.12.6) with ESMTP id j6AM7Ep6073933 for ; Mon, 11 Jul 2005 01:07:18 +0300 (EEST) (envelope-from postgres@bilteks.com) From: "Alexander Kirpa" To: pgsql-performance@postgresql.org Date: Mon, 11 Jul 2005 01:07:14 +0300 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Subject: Re: Data Warehousing Tuning Message-ID: <42D1C642.16182.39B63C29@localhost> X-mailer: Pegasus Mail for Win32 (v3.12c) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/113 X-Sequence-Number: 13354 >- Sun V250 server >- 2*1.3GHz Sparc IIIi CPU >- 8GB RAM >- 8*73GB SCSI drives >- Solaris 10 >- Postgres 8 >4) We moved the pg_xlog files off /data/postgres (disks 2-7) and into >/opt/pg_xlog (disks 0-1), but it seemed like performance decreased, >so we moved them back again. You have saturated SCSI bus. 1x160GB/s SCSI too small for 8xHDD with 30-70MB/s Solutions: Replace CD/DVD/tape at top 2x5" slots on 2xHDD (320 SCSI), install PCI 64/66 SCSI 320 controller (or simple RAID1 controller for minimize saturation of PCI buses) and attach to 2xHDD. Move /opt/pg_xlog on this drives. Best regards, Alexander Kirpa From pgsql-performance-owner@postgresql.org Mon Jul 11 08:31:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9D0645283C for ; Mon, 11 Jul 2005 08:31:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 15805-03 for ; Mon, 11 Jul 2005 11:31:49 +0000 (GMT) Received: from cmailg1.svr.pol.co.uk (cmailg1.svr.pol.co.uk [195.92.195.171]) by svr1.postgresql.org (Postfix) with ESMTP id BC5975287F for ; Mon, 11 Jul 2005 08:31:43 -0300 (ADT) Received: from modem-2435.leopard.dialup.pol.co.uk ([217.135.153.131] helo=192.168.0.102) by cmailg1.svr.pol.co.uk with esmtp (Exim 4.41) id 1DrwVi-0005oz-LJ; Mon, 11 Jul 2005 12:31:43 +0100 Subject: Re: cost-based vacuum From: Simon Riggs To: Ian Westmacott Cc: pgsql-performance@postgresql.org In-Reply-To: <1120839902.20657.197.camel@spectre.intellivid.com> References: <1120839902.20657.197.camel@spectre.intellivid.com> Content-Type: text/plain Organization: 2nd Quadrant Date: Mon, 11 Jul 2005 12:31:44 +0100 Message-Id: <1121081504.3970.53.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.053 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/103 X-Sequence-Number: 13344 On Fri, 2005-07-08 at 12:25 -0400, Ian Westmacott wrote: > I am beginning to look at Postgres 8, and am particularly > interested in cost-based vacuum/analyze. I'm hoping someone > can shed some light on the behavior I am seeing. > > Suppose there are three threads: > > writer_thread > every 1/15 second do > BEGIN TRANSACTION > COPY table1 FROM stdin > ... > COPY tableN FROM stdin > perform several UPDATEs, DELETEs and INSERTs > COMMIT > > reader_thread > every 1/15 second do > BEGIN TRANSACTION > SELECT FROM table1 ... > ... > SELECT FROM tableN ... > COMMIT > > analyze_thread > every 5 minutes do > ANALYZE table1 > ... > ANALYZE tableN > > > Now, Postgres 8.0.3 out-of-the-box (all default configs) on a > particular piece of hardware runs the Postgres connection for > writer_thread at about 15% CPU (meaningless, I know, but for > comparison) and runs the Postgres connection for reader_thread > at about 30% CPU. Latency for reader_thread seeing updates > from writer_thread is well under 1/15s. Impact of > analyze_thread is negligible. > > If I make the single configuration change of setting > vacuum_cost_delay=1000, each iteration in analyze_thread takes > much longer, of course. But what I also see is that the CPU > usage of the connections for writer_thread and reader_thread > spike up to well over 80% each (this is a dualie) and latency > drops to 8-10s, during the ANALYZEs. > > I don't understand why this would be. I don't think there > are any lock issues, and I don't see any obvious I/O issues. > Am I missing something? Is there any way to get some > insight into what those connections are doing? The ANALYZE commands hold read locks on the tables you wish to write to. If you slow them down, you merely slow down your write transactions also, and then the read transactions that wait behind them. Every time the ANALYZE sleeps it wakes up the other transactions, which then realise they can't move because of locks and then wake up the ANALYZEs for another shot. The end result is that you introduce more context- switching, without any chance of doing more useful work while the ANALYZEs sleep. Don't use the vacuum_cost_delay in this situation. You might try setting it to 0 for the analyze_thread only. Sounds like you could speed things up by splitting everything into two sets of tables, with writer_thread1 and writer_thread2 etc. That way your 2 CPUs would be able to independently be able to get through more work without locking each other out. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Mon Jul 11 09:59:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2175252933 for ; Mon, 11 Jul 2005 09:59:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42328-07 for ; Mon, 11 Jul 2005 12:59:50 +0000 (GMT) Received: from web54702.mail.yahoo.com (web54702.mail.yahoo.com [68.142.225.200]) by svr1.postgresql.org (Postfix) with SMTP id A588252849 for ; Mon, 11 Jul 2005 09:59:48 -0300 (ADT) Received: (qmail 50725 invoked by uid 60001); 11 Jul 2005 12:59:51 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=uRsb9G+llvpPE1zh2c9FlYOExmxnS5yZrUCKxyS2/2HamwfodWLWxOQCfiU59baM32Y3Jw1AlrQn5dhBK7ZMZ8QVv0vN9aET3YI3HoNZ9MkS8PUfH8IKzbCw6dhpCpZbWvruBHd4PuOVOcND4l0k+qyWfBlFaPFfSmv/RGNWNPU= ; Message-ID: <20050711125951.50723.qmail@web54702.mail.yahoo.com> Received: from [200.91.204.38] by web54702.mail.yahoo.com via HTTP; Mon, 11 Jul 2005 07:59:51 CDT Date: Mon, 11 Jul 2005 07:59:51 -0500 (CDT) From: Alejandro Lemus Subject: Question To: pgsql-performance@postgresql.org MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.752 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, FROM_HAS_ULINE_NUMS X-Spam-Level: * X-Archive-Number: 200507/104 X-Sequence-Number: 13345 In the past week, one guy of Unix Group in Colombia say: "Postgrest in production is bat, if the power off in any time the datas is lost why this datas is in plain files. Postgrest no ssupport data bases with more 1 millon of records". Wath tell me in this respect?, is more best Informix as say Ing. Alejandro Lemus G. Radio Taxi Aeropuerto S.A. Avenida de las Am�ricas # 51 - 39 Bogot� - Colombia Tel: 571-4470694 / 571-4202600 Ext. 260 Fax: 571-2624070 email: alejandro_lemus2003@yahoo.com __________________________________________________ Correo Yahoo! Espacio para todos tus mensajes, antivirus y antispam �gratis! Reg�strate ya - http://correo.espanol.yahoo.com/ From pgsql-performance-owner@postgresql.org Mon Jul 11 10:07:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4EFDA5281F for ; Mon, 11 Jul 2005 10:07:48 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 43269-07 for ; Mon, 11 Jul 2005 13:07:43 +0000 (GMT) Received: from mailhost.intellivid.com (mailhost.intellivid.com [64.32.200.11]) by svr1.postgresql.org (Postfix) with ESMTP id AB5A852802 for ; Mon, 11 Jul 2005 10:07:43 -0300 (ADT) Received: from [192.168.2.68] (spectre.intellivid.com [192.168.2.68]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by newmail.intellivid.com (Postfix) with ESMTP id 8379EF1829C; Mon, 11 Jul 2005 09:07:46 -0400 (EDT) Subject: Re: cost-based vacuum From: Ian Westmacott To: Simon Riggs Cc: pgsql-performance@postgresql.org In-Reply-To: <1121081504.3970.53.camel@localhost.localdomain> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> Content-Type: text/plain Organization: Intellivid Corp. Message-Id: <1121087266.27427.35.camel@spectre.intellivid.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Mon, 11 Jul 2005 09:07:46 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/105 X-Sequence-Number: 13346 On Mon, 2005-07-11 at 07:31, Simon Riggs wrote: > The ANALYZE commands hold read locks on the tables you wish to write to. > If you slow them down, you merely slow down your write transactions > also, and then the read transactions that wait behind them. Every time > the ANALYZE sleeps it wakes up the other transactions, which then > realise they can't move because of locks and then wake up the ANALYZEs > for another shot. The end result is that you introduce more context- > switching, without any chance of doing more useful work while the > ANALYZEs sleep. Let me make sure I understand. ANALYZE acquires a read lock on the table, that it holds until the operation is complete (including any sleeps). That read lock blocks the extension of that table via COPY. Is that right? According to the 8.0 docs, ANALYZE acquires an ACCESS SHARE lock on the table, and that conflicts only with ACCESS EXCLUSIVE. Thats why I didn't think I had a lock issue, since I think COPY only needs ROW EXCLUSIVE. Or perhaps the transaction needs something more? Thanks, --Ian From pgsql-performance-owner@postgresql.org Mon Jul 11 10:29:01 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 564815281F for ; Mon, 11 Jul 2005 10:29:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 46965-05 for ; Mon, 11 Jul 2005 13:28:54 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 1EBF352809 for ; Mon, 11 Jul 2005 10:28:53 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6BDSqq1007753; Mon, 11 Jul 2005 09:28:52 -0400 (EDT) To: Simon Riggs Cc: Ian Westmacott , pgsql-performance@postgresql.org Subject: Re: cost-based vacuum In-reply-to: <1121081504.3970.53.camel@localhost.localdomain> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> Comments: In-reply-to Simon Riggs message dated "Mon, 11 Jul 2005 12:31:44 +0100" Date: Mon, 11 Jul 2005 09:28:52 -0400 Message-ID: <7752.1121088532@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/106 X-Sequence-Number: 13347 Simon Riggs writes: >> I don't understand why this would be. I don't think there >> are any lock issues, and I don't see any obvious I/O issues. > The ANALYZE commands hold read locks on the tables you wish to write to. Unless there were more commands that Ian didn't show us, he's not taking any locks that would conflict with ANALYZE. So I don't believe this is the explanation. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 11 10:30:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5AB0852809 for ; Mon, 11 Jul 2005 10:30:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 46440-08 for ; Mon, 11 Jul 2005 13:30:18 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id 543095297B for ; Mon, 11 Jul 2005 10:30:16 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id 38795414F46; Mon, 11 Jul 2005 14:30:10 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id 45F7515ED6; Mon, 11 Jul 2005 14:29:10 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 19615-06; Mon, 11 Jul 2005 14:29:07 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 6F6CA15ED5; Mon, 11 Jul 2005 14:29:07 +0100 (BST) Message-ID: <42D27423.1010808@archonet.com> Date: Mon, 11 Jul 2005 14:29:07 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Alejandro Lemus Cc: pgsql-performance@postgresql.org Subject: Re: Question References: <20050711125951.50723.qmail@web54702.mail.yahoo.com> In-Reply-To: <20050711125951.50723.qmail@web54702.mail.yahoo.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/107 X-Sequence-Number: 13348 Perhaps choose a better subject than "question" next time? Alejandro Lemus wrote: > In the past week, one guy of Unix Group in Colombia > say: "Postgrest in production is bat, if the power off > in any time the datas is lost Wrong. And it's called "PostgreSQL". > why this datas is in > plain files. Postgrest no ssupport data bases with > more 1 millon of records". Wrong. > Wath tell me in this respect?, is more best Informix > as say Your contact in the Unix Group in Columbia obviously talks on subjects where he knows little. Perhaps re-evaluate anything else you've heard from him. You can find details on PostgreSQL at http://www.postgresql.org/, including the manuals: http://www.postgresql.org/docs/8.0/static/index.html The FAQ: http://www.postgresql.org/docs/faq/ Spanish/Brazilian communities, which might prove useful http://www.postgresql.org/community/international PostgreSQL is licensed under the BSD licence, which means you can freely download or deploy it in a commercial setting if you desire. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Mon Jul 11 10:35:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CE7485296E for ; Mon, 11 Jul 2005 10:35:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 50519-04 for ; Mon, 11 Jul 2005 13:34:57 +0000 (GMT) Received: from mx-2.sollentuna.net (mx-2.sollentuna.net [195.84.163.199]) by svr1.postgresql.org (Postfix) with ESMTP id DBE0452977 for ; Mon, 11 Jul 2005 10:34:52 -0300 (ADT) Received: from ALGOL.sollentuna.se (janus.sollentuna.se [62.65.68.67]) by mx-2.sollentuna.net (Postfix) with ESMTP id 05F178F284; Mon, 11 Jul 2005 15:34:52 +0200 (CEST) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Subject: Re: Question Date: Mon, 11 Jul 2005 15:34:51 +0200 Message-ID: <6BCB9D8A16AC4241919521715F4D8BCE6C7756@algol.sollentuna.se> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Question Thread-Index: AcWGGPuxqXR4tdx6Sb+/b1jXaLAc8gAA/LLA From: "Magnus Hagander" To: "Alejandro Lemus" , X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.062 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/108 X-Sequence-Number: 13349 > In the past week, one guy of Unix Group in Colombia > say: "Postgrest in production is bat, if the power off in any=20 > time the datas is lost why this datas is in plain files.=20 > Postgrest no ssupport data bases with more 1 millon of records".=20 > Wath tell me in this respect?, is more best Informix as say=20 Both these statements are completely incorrect.=20 Unlike some other "database systems", PostgreSQL *does* survive power loss without any major problems. Assuming you use a metadata journailng filesystem, and don't run with non-battery-backed write-cache (but no db can survive that..) And having a million records is no problem at all. You may run into considerations when you're talking billions, but you can do that as well - it just takes a bit more knowledge before you can do it right. //Magnus From pgsql-performance-owner@postgresql.org Mon Jul 11 11:51:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 473765284D for ; Mon, 11 Jul 2005 11:51:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 66453-10 for ; Mon, 11 Jul 2005 14:51:24 +0000 (GMT) Received: from cmailg4.svr.pol.co.uk (cmailg4.svr.pol.co.uk [195.92.195.174]) by svr1.postgresql.org (Postfix) with ESMTP id F1C2852824 for ; Mon, 11 Jul 2005 11:51:23 -0300 (ADT) Received: from modem-1688.lion.dialup.pol.co.uk ([217.135.166.152] helo=192.168.0.102) by cmailg4.svr.pol.co.uk with esmtp (Exim 4.41) id 1Drzd0-0002CH-S4; Mon, 11 Jul 2005 15:51:27 +0100 Subject: Re: cost-based vacuum From: Simon Riggs To: Ian Westmacott Cc: pgsql-performance@postgresql.org In-Reply-To: <1121087266.27427.35.camel@spectre.intellivid.com> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> Content-Type: text/plain Organization: 2nd Quadrant Date: Mon, 11 Jul 2005 15:51:28 +0100 Message-Id: <1121093488.3970.70.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.053 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/109 X-Sequence-Number: 13350 On Mon, 2005-07-11 at 09:07 -0400, Ian Westmacott wrote: > On Mon, 2005-07-11 at 07:31, Simon Riggs wrote: > > The ANALYZE commands hold read locks on the tables you wish to write to. > > If you slow them down, you merely slow down your write transactions > > also, and then the read transactions that wait behind them. Every time > > the ANALYZE sleeps it wakes up the other transactions, which then > > realise they can't move because of locks and then wake up the ANALYZEs > > for another shot. The end result is that you introduce more context- > > switching, without any chance of doing more useful work while the > > ANALYZEs sleep. > > Let me make sure I understand. ANALYZE acquires a read > lock on the table, that it holds until the operation is > complete (including any sleeps). That read lock blocks > the extension of that table via COPY. Is that right? > > According to the 8.0 docs, ANALYZE acquires an ACCESS SHARE > lock on the table, and that conflicts only with ACCESS > EXCLUSIVE. Thats why I didn't think I had a lock issue, > since I think COPY only needs ROW EXCLUSIVE. Or perhaps > the transaction needs something more? The docs are correct, but don't show catalog and buffer locks. ...but on further reading of the code there are no catalog locks or buffer locks held across the sleep points. So, my explanation doesn't work as an explanation for the sleep/no sleep difference you have observed. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Mon Jul 11 20:26:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AD8BA528D5 for ; Mon, 11 Jul 2005 20:26:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71872-05 for ; Mon, 11 Jul 2005 23:26:25 +0000 (GMT) Received: from loki.globexplorer.com (loki.globexplorer.com [208.35.14.101]) by svr1.postgresql.org (Postfix) with ESMTP id 75BBC52824 for ; Mon, 11 Jul 2005 20:26:22 -0300 (ADT) X-MimeOLE: Produced By Microsoft Exchange V6.0.6603.0 content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: Re: Question Date: Mon, 11 Jul 2005 16:26:23 -0700 Message-ID: <71E37EF6B7DCC1499CEA0316A256832801D4C5BB@loki.wc.globexplorer.net> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Question Thread-Index: AcWGGO7BHh8jYKuKTOesRaRPQ7f+bwAWkIXQ From: "Gregory S. Williamson" To: "Alejandro Lemus" , X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/110 X-Sequence-Number: 13351 As a sometimes Informix and PostgreSQL DBA, I disagree with the = contentions below. We have many tables with 10s of millions of rows in = Postgres. We have had (alas) power issues with our lab on more than one = occasion and the afflicted servers have recovered like a champ, every = time. This person may not like postgres (or very much likes Informix), but he = shouldn't conjure up spurious reasons to support his/her prejudice. Informix is an excellent product, but it can be costly for web related = applications. PostgeSQL is also an excellent database. Each has = differences which may make the decision between the two of them clear. = But facts are necessary to have a real discussion. Greg WIlliamson DBA GlobeXplorer LLC -----Original Message----- From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org]On Behalf Of Alejandro Lemus Sent: Monday, July 11, 2005 6:00 AM To: pgsql-performance@postgresql.org Subject: [PERFORM] Question In the past week, one guy of Unix Group in Colombia say: "Postgrest in production is bat, if the power off in any time the datas is lost why this datas is in plain files. Postgrest no ssupport data bases with more 1 millon of records".=20 Wath tell me in this respect?, is more best Informix as say=20 Ing. Alejandro Lemus G. Radio Taxi Aeropuerto S.A. Avenida de las Am=E9ricas # 51 - 39 Bogot=E1 - Colombia Tel: 571-4470694 / 571-4202600 Ext. 260 Fax: 571-2624070 email: alejandro_lemus2003@yahoo.com __________________________________________________ Correo Yahoo! Espacio para todos tus mensajes, antivirus y antispam =A1gratis!=20 Reg=EDstrate ya - http://correo.espanol.yahoo.com/=20 ---------------------------(end of broadcast)--------------------------- TIP 6: explain analyze is your friend !DSPAM:42d26e2065882109568359! From pgsql-performance-owner@postgresql.org Mon Jul 11 20:39:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E4D1A52987 for ; Mon, 11 Jul 2005 20:39:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 73684-09 for ; Mon, 11 Jul 2005 23:39:22 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id E1BA852983 for ; Mon, 11 Jul 2005 20:39:17 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050711233920m9100ceqiae>; Mon, 11 Jul 2005 23:39:20 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id E37B855FEC; Mon, 11 Jul 2005 18:39:19 -0500 (CDT) Received: from [192.168.1.102] (209-181-211-15.cdrr.qwest.net [209.181.211.15]) by juju.arbash-meinel.com (Postfix) with ESMTP id 8050E55F98; Mon, 11 Jul 2005 18:39:10 -0500 (CDT) Message-ID: <42D3031C.8050907@arbash-meinel.com> Date: Mon, 11 Jul 2005 18:39:08 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dario Pudlo Cc: pgsql-performance@postgresql.org Subject: Re: join and query planner References: In-Reply-To: X-Enigmail-Version: 0.91.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigAC7C853A4D28D578E3FB0172" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/114 X-Sequence-Number: 13355 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigAC7C853A4D28D578E3FB0172 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Dario Pudlo wrote: > (first at all, sorry for my english) > Hi. > - Does "left join" restrict the order in which the planner must join > tables? I've read about join, but i'm not sure about left join... > - If so: Can I avoid this behavior? I mean, make the planner resolve the > query, using statistics (uniqueness, data distribution) rather than join > order. > > My query looks like: > SELECT ... > FROM a, b, > LEFT JOIN c ON (c.key = a.key) > LEFT JOIN d on (d.key=a.key) > WHERE (a.key = b.key) AND (b.column <= 100) > > b.column has a lot better selectivity, but planner insist on resolve > first c.key = a.key. > > Of course, I could rewrite something like: > SELECT ... > FROM > (SELECT ... > FROM a,b > LEFT JOIN d on (d.key=a.key) > WHERE (b.column <= 100) > ) > as aa > LEFT JOIN c ON (c.key = aa.key) > > but this is query is constructed by an application with a "multicolumn" > filter. It's dynamic. > It means that a user could choose to look for "c.column = 1000". And > also, combinations of filters. > > So, I need the planner to choose the best plan... Probably forcing the other join earlier could help: SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) ... I think the problem is that postgresql can't break JOIN syntax very easily. But you can make the JOIN earlier. John =:-> > > I've already change statistics, I clustered tables with cluster, ran vacuum > analyze, changed work_mem, shared_buffers... > > Greetings. TIA. > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster > --------------enigAC7C853A4D28D578E3FB0172 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC0wMcJdeBCYSNAAMRAs+mAJ4y/5cqJ6BshkSFmos3M/wCXuZf6gCeMiQB vLDYL89Rt1TCEGFFkOm9lhI= =QwbF -----END PGP SIGNATURE----- --------------enigAC7C853A4D28D578E3FB0172-- From pgsql-performance-owner@postgresql.org Mon Jul 11 20:42:23 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7C6885287D for ; Mon, 11 Jul 2005 20:42:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74943-05 for ; Mon, 11 Jul 2005 23:42:13 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id 57D0952998 for ; Mon, 11 Jul 2005 20:42:12 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050711234215m920060tite>; Mon, 11 Jul 2005 23:42:15 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id CB24C55FEC; Mon, 11 Jul 2005 18:42:14 -0500 (CDT) Received: from [192.168.1.102] (209-181-211-15.cdrr.qwest.net [209.181.211.15]) by juju.arbash-meinel.com (Postfix) with ESMTP id 3715E55F98; Mon, 11 Jul 2005 18:42:07 -0500 (CDT) Message-ID: <42D303CA.2020808@arbash-meinel.com> Date: Mon, 11 Jul 2005 18:42:02 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: jobapply Cc: pgsql-performance@postgresql.org Subject: Re: Sorting on longer key is faster ? References: <20050709183842.2EABF52820@svr1.postgresql.org> In-Reply-To: <20050709183842.2EABF52820@svr1.postgresql.org> X-Enigmail-Version: 0.91.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigB810D5620F6C038341525139" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.038 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/115 X-Sequence-Number: 13356 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigB810D5620F6C038341525139 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit jobapply wrote: > The 2 queries are almost same, but ORDER BY x||t is FASTER than ORDER BY x.. > > How can that be possible? > > Btw: x and x||t are same ordered > > phoeniks=> explain analyze SELECT * FROM test WHERE i<20 ORDER BY x || t; > QUERY PLAN > What types are x and t, I have the feeling "x || t" is actually a boolean, so it is only a True/False sort, while ORDER BY x has to do some sort of string comparison (which might actually be a locale depended comparison, and strcoll can be very slow on some locales) John =:-> > ---------------------------------------------------------------------------- > ---------------------------------------------- > Sort (cost=2282.65..2284.92 rows=907 width=946) (actual > time=74.982..79.114 rows=950 loops=1) > Sort Key: (x || t) > -> Index Scan using i_i on test (cost=0.00..2238.09 rows=907 width=946) > (actual time=0.077..51.015 rows=950 loops=1) > Index Cond: (i < 20) > Total runtime: 85.944 ms > (5 rows) > > phoeniks=> explain analyze SELECT * FROM test WHERE i<20 ORDER BY x; > QUERY PLAN > ---------------------------------------------------------------------------- > --------------------------------------------- > Sort (cost=2280.38..2282.65 rows=907 width=946) (actual > time=175.431..179.239 rows=950 loops=1) > Sort Key: x > -> Index Scan using i_i on test (cost=0.00..2235.82 rows=907 width=946) > (actual time=0.024..5.378 rows=950 loops=1) > Index Cond: (i < 20) > Total runtime: 183.317 ms > (5 rows) > > > > > > phoeniks=> \d+ test > Table "public.test" > Column | Type | Modifiers | Description > --------+---------+-----------+------------- > i | integer | | > t | text | | > x | text | | > Indexes: > "i_i" btree (i) > "x_i" btree (xpath_string(x, 'data'::text)) > "x_ii" btree (xpath_string(x, 'movie/characters/character'::text)) > Has OIDs: no > > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > --------------enigB810D5620F6C038341525139 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD4DBQFC0wPKJdeBCYSNAAMRAk14AJdsYOPYY3PHuBNilislLLiqOX11AKCC4Cki S4ZzdD/Y9WtZEsc0LA+xwA== =GwKr -----END PGP SIGNATURE----- --------------enigB810D5620F6C038341525139-- From pgsql-performance-owner@postgresql.org Mon Jul 11 20:47:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 26E725298D for ; Mon, 11 Jul 2005 20:47:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74667-09 for ; Mon, 11 Jul 2005 23:47:55 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 637D452987 for ; Mon, 11 Jul 2005 20:47:54 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6BNloXr012962; Mon, 11 Jul 2005 19:47:50 -0400 (EDT) To: "jobapply" Cc: pgsql-performance@postgresql.org Subject: Re: Sorting on longer key is faster ? In-reply-to: <20050709183842.2EABF52820@svr1.postgresql.org> References: <20050709183842.2EABF52820@svr1.postgresql.org> Comments: In-reply-to "jobapply" message dated "Sat, 09 Jul 2005 22:38:40 +0400" Date: Mon, 11 Jul 2005 19:47:50 -0400 Message-ID: <12961.1121125670@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/116 X-Sequence-Number: 13357 "jobapply" writes: > The 2 queries are almost same, but ORDER BY x||t is FASTER than ORDER BY x.. > How can that be possible? Hmm, how long are the x values? Is it possible many of them are TOASTed? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 11 22:48:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0841D52848 for ; Mon, 11 Jul 2005 22:48:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 97422-07 for ; Tue, 12 Jul 2005 01:48:49 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id E9ED8529A9 for ; Mon, 11 Jul 2005 22:48:47 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050712014852m920060ddae>; Tue, 12 Jul 2005 01:48:52 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id A86C255FEC; Mon, 11 Jul 2005 20:48:51 -0500 (CDT) Received: from [192.168.1.102] (209-181-211-15.cdrr.qwest.net [209.181.211.15]) by juju.arbash-meinel.com (Postfix) with ESMTP id D9C0755F98; Mon, 11 Jul 2005 20:48:46 -0500 (CDT) Message-ID: <42D3217C.20703@arbash-meinel.com> Date: Mon, 11 Jul 2005 20:48:44 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Chris Travers , Postgresql Performance Subject: Re: Sorting on longer key is faster ? References: <20050709183842.2EABF52820@svr1.postgresql.org> <42D303CA.2020808@arbash-meinel.com> <42D31613.5020608@metatrontech.com> In-Reply-To: <42D31613.5020608@metatrontech.com> X-Enigmail-Version: 0.91.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigEDC246B45A7210D61FBECEB2" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.038 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/117 X-Sequence-Number: 13358 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigEDC246B45A7210D61FBECEB2 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Chris Travers wrote: > John A Meinel wrote: > >> jobapply wrote: >> >> >>> The 2 queries are almost same, but ORDER BY x||t is FASTER than ORDER >>> BY x.. >>> >>> How can that be possible? >>> >>> Btw: x and x||t are same ordered >>> >>> phoeniks=> explain analyze SELECT * FROM test WHERE i<20 ORDER BY x >>> || t; >>> QUERY PLAN >>> >>> >> >> >> What types are x and t, I have the feeling "x || t" is actually a >> boolean, so it is only a True/False sort, while ORDER BY x has to do >> some sort of string comparison (which might actually be a locale >> depended comparison, and strcoll can be very slow on some locales) >> >> >> > Am I reading this that wrong? I would think that x || t would mean > "concatenate x and t." Sorry, I think you are right. I was getting my operators mixed up. > > This is interesting. I never through of writing a multicolumn sort this > way.... I'm also surprised that the sort is faster with a merge operation. Are you using UNICODE as the database format? I'm just wondering if it is doing something funny like casting it to an easier to sort type. > > Best Wishes, > Chris Travers > Metatron Technology Consulting PS> Don't forget to Reply All so that your messages go back to the list. --------------enigEDC246B45A7210D61FBECEB2 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC0yF8JdeBCYSNAAMRAuLoAKCQRpqH4VQrgX1WYAPsdIkhvTGsIgCcDIsH X59AVC17LTtLbpt8i6BHxaw= =L2PD -----END PGP SIGNATURE----- --------------enigEDC246B45A7210D61FBECEB2-- From pgsql-performance-owner@postgresql.org Mon Jul 11 22:52:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1A08B52977 for ; Mon, 11 Jul 2005 22:52:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 00888-03 for ; Tue, 12 Jul 2005 01:52:28 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 4082C52994 for ; Mon, 11 Jul 2005 22:52:27 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050712015231m9100cftf2e>; Tue, 12 Jul 2005 01:52:32 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 87E3755FEC; Mon, 11 Jul 2005 20:52:31 -0500 (CDT) Received: from [192.168.1.102] (209-181-211-15.cdrr.qwest.net [209.181.211.15]) by juju.arbash-meinel.com (Postfix) with ESMTP id C03F155F98; Mon, 11 Jul 2005 20:52:27 -0500 (CDT) Message-ID: <42D32259.2000705@arbash-meinel.com> Date: Mon, 11 Jul 2005 20:52:25 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: jobapply Cc: pgsql-performance@postgresql.org Subject: Re: Sorting on longer key is faster ? References: <20050709183842.2EABF52820@svr1.postgresql.org> In-Reply-To: <20050709183842.2EABF52820@svr1.postgresql.org> X-Enigmail-Version: 0.91.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigBEFD86D74D4F6EC79AD16D69" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.04 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/118 X-Sequence-Number: 13359 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigBEFD86D74D4F6EC79AD16D69 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit jobapply wrote: > The 2 queries are almost same, but ORDER BY x||t is FASTER than ORDER BY x.. > > How can that be possible? > > Btw: x and x||t are same ordered > > phoeniks=> explain analyze SELECT * FROM test WHERE i<20 ORDER BY x || t; > QUERY PLAN I also thought of another possibility. Are there a lot of similar entries in X? Meaning that the same value is repeated over and over? It is possible that the sort code has a weakness when sorting equal values. For instance, if it was doing a Hash aggregation, you would have the same hash repeated. (It isn't I'm just mentioning a case where it might affect something). If it is creating a tree representation, it might cause some sort of pathological worst-case behavior, where all entries keep adding to the same side of the tree, rather than being more balanced. I don't know the internals of postgresql sorting, but just some ideas. John =:-> --------------enigBEFD86D74D4F6EC79AD16D69 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC0yJZJdeBCYSNAAMRAqNvAJ9ISh6cVjxogvmOiwRfIl4qATHrEgCfetFN P5ittkgSVxFUPpcfR+AztqU= =MbUn -----END PGP SIGNATURE----- --------------enigBEFD86D74D4F6EC79AD16D69-- From pgsql-performance-owner@postgresql.org Tue Jul 12 18:23:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E17D352BA4 for ; Tue, 12 Jul 2005 18:15:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95953-03 for ; Tue, 12 Jul 2005 21:15:40 +0000 (GMT) Received: from svr2.postgresql.org (svr2.postgresql.org [65.19.161.25]) by svr1.postgresql.org (Postfix) with ESMTP id E9DCF52BD2 for ; Tue, 12 Jul 2005 18:07:47 -0300 (ADT) Received: from web33902.mail.mud.yahoo.com (web33902.mail.mud.yahoo.com [66.163.178.66]) by svr2.postgresql.org (Postfix) with SMTP id C91E1F1178 for ; Tue, 12 Jul 2005 07:19:54 +0100 (BST) Received: (qmail 37245 invoked by uid 60001); 12 Jul 2005 06:22:47 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=uL5maoKeM37Pkbe400pZ+/EUX/fumpdAJmRhtvmRL9BuZCujG1RTNMW9SeSa53aI1c9Ilfi7dhr3rNuTvUDKzgGp4Df/sOea9j2N5GGnxhuLeoP82evkGryU9K7GIcDeySPxnmz0aCR4c2FR1/xHTOR8DX77U5JuF5aSxjWqEts= ; Message-ID: <20050712062247.37243.qmail@web33902.mail.mud.yahoo.com> Received: from [67.106.37.5] by web33902.mail.mud.yahoo.com via HTTP; Mon, 11 Jul 2005 23:22:47 PDT Date: Mon, 11 Jul 2005 23:22:47 -0700 (PDT) From: Brent Henry Subject: General DB Tuning To: pgsql-performance@postgresql.org MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=2.174 tagged_above=0 required=5 tests=FORGED_YAHOO_RCVD X-Spam-Level: ** X-Archive-Number: 200507/125 X-Sequence-Number: 13366 Help! After recently migrating to Postgres 8, I've discovered to my horror that I can't determine which queries are poorly performing anymore because the logging has drastically changed and no longer shows durations for anything done through JDBC. So I'm desperately trying to do performance tuning on my servers and have no way to sort out which statements are the slowest. Does anyone have any suggestions? How do you determine what queries are behaving badly when you can't get durations out of the logs? I have a perl script that analyzes the output from Postgres 7 logs and it works great! But it relies on the duration being there. I did some searches on postgresql.org mailing lists and have seen a few people discussing this problem, but noone seems to be too worried about it. Is there a simple work-around? Sincerely, Brent ____________________________________________________ Sell on Yahoo! Auctions � no fees. Bid on great items. http://auctions.yahoo.com/ From pgsql-performance-owner@postgresql.org Tue Jul 12 09:58:24 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 48C8F5281F for ; Tue, 12 Jul 2005 09:58:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 59884-06 for ; Tue, 12 Jul 2005 12:58:03 +0000 (GMT) Received: from cmailg2.svr.pol.co.uk (cmailg2.svr.pol.co.uk [195.92.195.172]) by svr1.postgresql.org (Postfix) with ESMTP id A607A52D37 for ; Tue, 12 Jul 2005 04:45:41 -0300 (ADT) Received: from modem-3803.leopard.dialup.pol.co.uk ([217.135.158.219] helo=192.168.0.102) by cmailg2.svr.pol.co.uk with esmtp (Exim 4.41) id 1DsFSW-0005Ka-Cb; Tue, 12 Jul 2005 08:45:40 +0100 Subject: Re: cost-based vacuum From: Simon Riggs To: Ian Westmacott Cc: pgsql-performance@postgresql.org In-Reply-To: <1121093488.3970.70.camel@localhost.localdomain> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> Content-Type: text/plain Organization: 2nd Quadrant Date: Tue, 12 Jul 2005 08:45:38 +0100 Message-Id: <1121154338.3970.106.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.053 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/119 X-Sequence-Number: 13360 On Mon, 2005-07-11 at 15:51 +0100, Simon Riggs wrote: > On Mon, 2005-07-11 at 09:07 -0400, Ian Westmacott wrote: > > On Mon, 2005-07-11 at 07:31, Simon Riggs wrote: > > > The ANALYZE commands hold read locks on the tables you wish to write to. > > > If you slow them down, you merely slow down your write transactions > > > also, and then the read transactions that wait behind them. Every time > > > the ANALYZE sleeps it wakes up the other transactions, which then > > > realise they can't move because of locks and then wake up the ANALYZEs > > > for another shot. The end result is that you introduce more context- > > > switching, without any chance of doing more useful work while the > > > ANALYZEs sleep. > > > > Let me make sure I understand. ANALYZE acquires a read > > lock on the table, that it holds until the operation is > > complete (including any sleeps). That read lock blocks > > the extension of that table via COPY. Is that right? > > > > According to the 8.0 docs, ANALYZE acquires an ACCESS SHARE > > lock on the table, and that conflicts only with ACCESS > > EXCLUSIVE. Thats why I didn't think I had a lock issue, > > since I think COPY only needs ROW EXCLUSIVE. Or perhaps > > the transaction needs something more? > > The docs are correct, but don't show catalog and buffer locks. > > ...but on further reading of the code there are no catalog locks or > buffer locks held across the sleep points. So, my explanation doesn't > work as an explanation for the sleep/no sleep difference you have > observed. I've been through all the code now and can't find any resource that is held across a delay point. Nor any reason to believe that the vacuum cost accounting would slow anything down. Since vacuum_cost_delay is a userset parameter, you should be able to SET this solely for the analyze_thread. That way we will know with more certainty that it is the analyze_thread that is interfering. What is your default_statistics_target? Do you have other stats targets set? How long does ANALYZE take to run, with/without the vacuum_cost_delay? Thanks, Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Tue Jul 12 11:35:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 910F752946 for ; Tue, 12 Jul 2005 11:35:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 82789-02 for ; Tue, 12 Jul 2005 14:35:27 +0000 (GMT) Received: from wolff.to (wolff.to [66.93.197.194]) by svr1.postgresql.org (Postfix) with SMTP id 4515D528A6 for ; Tue, 12 Jul 2005 11:35:29 -0300 (ADT) Received: (qmail 2462 invoked by uid 500); 12 Jul 2005 14:35:52 -0000 Date: Tue, 12 Jul 2005 09:35:52 -0500 From: Bruno Wolff III To: Dario Pudlo Cc: pgsql-performance@postgresql.org Subject: Re: join and query planner Message-ID: <20050712143552.GA32711@wolff.to> Mail-Followup-To: Bruno Wolff III , Dario Pudlo , pgsql-performance@postgresql.org References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.007 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/120 X-Sequence-Number: 13361 On Wed, Jul 06, 2005 at 18:54:02 -0300, Dario Pudlo wrote: > (first at all, sorry for my english) > Hi. > - Does "left join" restrict the order in which the planner must join > tables? I've read about join, but i'm not sure about left join... The left join operator is not associative so in general the planner doesn't have much flexibility to reorder left (or right) joins. From pgsql-performance-owner@postgresql.org Tue Jul 12 13:22:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 165DA5294F for ; Tue, 12 Jul 2005 13:22:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 05612-06 for ; Tue, 12 Jul 2005 16:22:00 +0000 (GMT) Received: from adicia.telenet-ops.be (adicia.telenet-ops.be [195.130.132.56]) by svr1.postgresql.org (Postfix) with ESMTP id 4B56C52867 for ; Tue, 12 Jul 2005 13:21:56 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by adicia.telenet-ops.be (Postfix) with SMTP id B521D44180 for ; Tue, 12 Jul 2005 18:21:52 +0200 (MEST) Received: from [10.0.1.2] (d5152B1F3.access.telenet.be [81.82.177.243]) by adicia.telenet-ops.be (Postfix) with ESMTP id 2AB6244061 for ; Tue, 12 Jul 2005 18:21:52 +0200 (MEST) Mime-Version: 1.0 (Apple Message framework v622) To: pgsql-performance@postgresql.org Message-Id: Content-Type: multipart/mixed; boundary=Apple-Mail-36--1003286864 From: Yves Vindevogel Subject: Projecting currentdb to more users Date: Tue, 12 Jul 2005 18:21:57 +0200 X-Mailer: Apple Mail (2.622) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/121 X-Sequence-Number: 13362 --Apple-Mail-36--1003286864 Content-Type: multipart/alternative; boundary=Apple-Mail-37--1003286864 --Apple-Mail-37--1003286864 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed Hi, We have a couple of database that are identical (one for each customer). They are all relatively small, ranging from 100k records to 1m records. There's only one main table with some smaller tables, a lot of indexes=20= and some functions. I would like to make an estimation of the performance, the diskspace=20 and other related things, when we have database of for instance 10 million records or 100 million=20= records. Is there any math to be done on that ? Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements --Apple-Mail-37--1003286864 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 Hi, We have a couple of database that are identical (one for each customer). They are all relatively small, ranging from 100k records to 1m records. There's only one main table with some smaller tables, a lot of indexes and some functions. I would like to make an estimation of the performance, the diskspace and other related things,=20 when we have database of for instance 10 million records or 100 million records. Is there any math to be done on that ? Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements = --Apple-Mail-37--1003286864-- --Apple-Mail-36--1003286864 Content-Transfer-Encoding: base64 Content-Type: image/tiff; x-unix-mode=0666; name="Pasted Graphic 2.tiff" Content-Disposition: inline; filename="Pasted Graphic 2.tiff" TU0AKgAAFciAP6BP5/wWDQeEQmFQuGQ2HQ+FP5+v9rOh5P9IMBuP8zK9tP8fJNlP8QoNjP8FndhP 8MnyVjJHSMrqFnv9BL1vP9ett2v98PuJxChUOiUWGQOCUalUujOV4PZ/qBluR/mdVtJ/jVHst/ho 9MF/g08WABHJfv8AWa0WoAHFfWuzgg6sB/hY9ysUohjv8qqVqv9LMZxv9ouV40zEYmHUjFY2HPx+ RNtOl4Rhftt/phiRs6LC/iA5LewnJeWg46W22+02cAHOwa26bC0HPY7TZ7G2G9dv8EHFdP8Om9YP 8rKJov9UtFzv9Dzh/sRvu5/vZ9PzHdeF4zsUqgQRlOR5v9DMBwv8uqe/jZIVwNHu6B0/WAKHvWGr RAA1ra0G1c/vfgANrdraXrTQIAA6JW2TUrQNz/ja0oAjO4YAjSWS0DOWawjys4RkIk4Lj2sAWEUZ B/ieUJqH+PZdnAf5aG06R5nsfLtqW7UaoWZJwukwLyi4VRsn+F5HGYf8QJWBQ7JWAI5tYti2Dm1T bQHAq0De1Q2FwtA1Fof4AjVDIBDQV0vDQWK0DY/oADYWsrNQN81NOuDZtYOM6ycs4Cjq+Q9JWE5D pGJpOpqP5dMwW5sHUf53xlHCExupjqoIbh3Hu55vp6ch3Hof4wlMrAdEWYh/gmP1RtXK0DDi2rXy nVzXjpVs8NM1g1vuNU2gANEugGMBRwu4YADTNsvwyAA3v+0k6LRWLb2dBUpjk2I4NVOx/gYPC6BU QiwCgSsSlWZpxH+ahzsOaBzvCe7qoMpKlUghd1n2f5oHId5/lQaJ0n+RBf3GLpUGwf4iksrg0FVF ICzlXT9QBB0tWO/+FjlKVWtfijTNRBsvDbhrdP3NUAwY38KFfkc0PuNjRWLLz8ytiTUVdVlnjlA2 MWPAWRAAM8LAAMjhhuR5ir4ULjigT5pn+P5co2T6pn+YBup6dp6xoh6kHie95modNOFkah0H+Qhd pyKBNmcf68r2Bo9GGtdpwIAUotM1Q4QMNzUDZhuVZbLsvy7KkqZZCnAb4AFcQY1C3LW1Wb5lOa2t ZvUt78NNjcMN83wEOMDczvuTzXNQ3QEN01WRL2b8XBYEDuugQEFUYgkpIo5FjIJTmhfZkHG8J0nm fB/sifp/EQYLBioUKsBkRKVgWPLX7stA4VnaDa+tZlZel0j+gCMpWS8MRULQNeIDY/4zTIAIyFX0 GFybZnr+rZ8p2a1s6tUNfADXXI1Yb8jLX+MNQk+1Az7zWqwVa/FaJrHpvSLeAQOhdE/l7CSJwrAb BXDXH+PkyA/gUiPGaftASWW6JeDgnGAsCmLNvcYqlz5+FchnZMrqGbeE3JoYhAMAKbEvOGeigtaE CDZwJcgxtiML0uMgQYzllKt2GhtP/EaH6cgAm+fnCtVEVXFOKNU54AC1B/tyLfBx4YJBGwhAAG5m yUoGH/TgxlZ7nGXw3YdEpYTPWOvaSq/JVEX28rEDQsGKDoGPx+QYnE1URg2paZY4dy7EHTx9cc41 upqIfwkjSgJWyW0MgBdK9I1DGAAwnS8G8+4cGYyUWWbKMg/gSiOjQG+FMQ44qqlmgBNUSYYMohKW 2ULFZaR8NdMFucoz/w/ZcAAMzgH+j/AKGV8QAZkhmWNM0AJ/I5zCiJEGYjF1qoGmxI5MqwT7JVlG buLULJhPXlbK+WMt3HwHmIbF8xaA0n6mvPgN7EJhx9Vcax+hsX6wGZuldNDnTSgQDoloGohS6BYE 62gPQqyRhlFE2gIYkCRgdD4WcAUsnxvlYklKgUCn4upi29JkkpG/pbYgtKbr86TLOnbLCF066ZLP kGsI+7NY9rRjZHB+0ew4M5S0AKT4FA8l0BaIUk4XHkj/EYLwzAtVyj/GwRYf49B8D6H+OMdw9R/j DG8vcTgx1xhpFTBoHYilRgaDygQAobXCynVVSSVdRYTLJls5CcIaqX0zpw/Kms76c2DSnTsNNPaA TbpggqOSCg4MkQwcAPiBAzimJqI4Xg3R/itX0P8bY66xD7MiYkgg5h4vAF2NgdY/xNGCH+HQVMIQ WiCNVM2L6Ap4s3ShFixMubA2HnjcUulhab3GuIbGxVjLlxXZoawMKFgsCiaSOUd6nB6XbH+PofKN LtqcH7eMf5AyimQOtegf4/TIEFvYP8dyjTm2eV1KdmNgrlIKr/cOxF+LjmRldTZY88L/U6P/YtyF +b82PDAhkOQsyNkHu8jS76NB4YXH+PvDUG7v3lIFeu8mFbuj6q8PXE2HEaXqIEUEgxnh2T2YhD/B WBZcUunVjO59yMB3PsHc3BONLemxwaP8NwsTMEPvVeMieJKvXmHZk8f48h5GHyYP8e+V3g3tvMQs VY01FLCxjgTHk8b943yBTOVoJEiXJzPgae1zr+2HwYhnIuRzrjczwP8eY8zDmKy7l8NOYcx40zKW nNswizytg+kXHc/qS48x9o7QbNMhZ0yMdvPBG7wmNz/jB6WYs436uFmbHhbFrAFgjhl4Y/g5IvH+ 0EkYFw+NuAHqmoet9bWJwPnDHEfM55E0udjTNW7uZ+y9p6L+oNe3Bxtoa4lACzgQee2kQ5Jwo1RK QPQfK8xejaJ6FG6w/wEh1NZJ+L6ydoH+zfqTUK0dK7AzsY7Yem9jaA0Fu2wVf9eJObqW+TxuwbCS K4JEYVnh2O/vWQM5K+wtCehCCAP5YABybV035xCAJIJqnrTyFqB9H8e1Lu/OumM870MTp3MGn9J8 g5Zsxhk2ZURLfGmrBAEg5m/CISIf4jRekblaCgRoyS0BjWNZVw6uZ70hnsrkND/Fbn3kWWtmz1ID a5NxyLYJ1xujdI3iZTm9dkYy4/P9yFKIcFoDQsbtKW4nT2YbLrpr450V6zTGem7N+YzXVzDZ0PSl ddIb8GQVrlD9sQjy4FKHVy0ZDDeLHCPWuuD/68Y0Vg1LX8c0bQSBiAq9JrVv2pYwZkzq6mqxDi8y XJgBkjZGftMMdUgm5PKyCBnogBI72hYgZlgrD6WWgMzPZkxGdUzfIfIzsdb66PXr/Jxp+XwRGlOL GnuJm2b0l7vo/dJeDL4N7qwWN+qn5CqedNMATuzZmObaad1s4dBMl/6wmG/Pt1+oAAYk2hToiuQc 5FxyjyRoHqH0xYKE+Q8k+UIgH0KAH+HSHqXmGwHSIuEGF0I25eWE8A96/g2akwhIk8Puf25mpi3w /IeG/M8y0O10/WgM3cpUS8l0mS+ocOYgDQS0ASDUTOB6EaJWDkXyXwFyL+GqGkPKHQHMOkxIXmIO HAHAJyyuKgvGIIHgU2H+G0GuHKH+FwF6KwEAtAH+CUEcLOAiNC2QnKcOZSZCmO9ongNep2sA3Y2W nY/KwE9hDazc8w8Sug9mh6cAZ0/UAEC0FKH+AOCOEoJQBeD8H+A2A+DYYGBuD2H+D6DmFCH+F0Fq hDB+PKGcGaaSGsGmJyGOGGg0EuEgP0CqCOEOH+BIA+DQH+AgBSDrD+CAEcH+AICwFNA+pamupfDQ /G1E2asfDksJDesNBCv012x+5WDmQSseeiAaCGEiH+BEAmC0H+A8AcCyH+A4AXGqAqAOCsLqAQCw H+BMAuDIH+CJEWPMCpFKCaCEEASEBFESA2AaC2H+AoAOCqK6AVGqA8AYCvGcAiC9HmBlHZBglBDt F80LF7BNDdBHDg2U0JGI0lGEseWsAgB6EYH+BUAkC4H+BKAuDDGcArH+BCAnH+A8AeC6K7HjG7H4 AoANHtG1G4A1GxGkAfHkBAAjJOBCAqC/FMAuDAH+BOApJ8A0BYDy2aWtGEzI1G2dKRIUwDGDF9BP DooBISnVImB9IsBRJHI2AuDHK3HHI4DFK3K6BGArI8BAAlJ8A8AhJqAlJxJBK3LDLBK9LmBNKCH+ AwBZEZIGcjGM3zKVIRL7BEH8zVKfKojrKlMCQVIk2jKuH/KzH/I5K7MjLFMpMnI5HGBIAvG/I5J9 MvMrK5M/K7LrJ9LxL1DXL5KYp1L/MM80g2wABMzWAGjVDZIc/XDrIiNjKtKxK1MtNBN7MlN9HFJ5 M3J7NDONMnNHLvLzKNKnMS5dNPMAsGpgLKLoAQNswAH8EAGAXGB0EobQAwEEJOblOa1uzJIfNvF9 MXFVMbMfOPODOBOBMxM1K3M7PfPdNFLtNLOZOcfjIOgVPIfeAaD2VGBUEY6ECyFQL+awXWH+HCHa KgW+KoAqDsPu3Mt8rxMO+fOjPTNzMZN3MhPtN/M/PlOJPrPhPvHBPzOXL3KPKhP69Mv4SmYWWs4m QyDGFUGsH+GeHEOkHW204SIED6FmaSBqeZHmQ3Fikw7iPwZWkCSs40Yg0CwTQAxpPUAhPZN5RDS1 PjOHPpRRRFOTP1RZSo5W1skw6MDa6g5omYZW94AYDkN+BOD+NKCoE6SKHy1WA8EOL2Z8cuhGTU/f SlArAu95L2cmiO1xSrQ7PXQ/S/S1RJS9RFUlNBTDRXNPRavw3St25kTXBiQsADBfUFNO/fUC6K+A N+9fTI2SkOfG6YmY9MZWDEFS+0TInFUQSfPI1NQ9MdSzRPUnS5MzRLUdRPUqD02RRaeoseDdSaQn ScQi9G/kPu+e/o+E8UQUoGv/IXMKsOfqZu8O/eV2S89E7QZ7XG8w/edGQYxiTeoSB3FKBQAlRBV9 S3LnWDUjXpN7LCBOAnJ8AyBXKKV0Z7KPKO6i764rXI9wS8DJVq6TAw+hP5MFBJDi0Ofq0i86mQPv XMDRWlBi6QVy6iAYCUEuJIBDESBGA0DNI2AxLjRCAxS4BGAvH5I5J2BLZdZVV9ZtZqDKJIA+DUH+ AiB4ES2a4pDFDE+fVIkg84ZylyTUn7RdYjIZYhQy14fipY9SDTWhSjRg6UAEC+fZGWEfEMBCDbFM AtJ9MzI9LlM8BFZjJ5ZpNBMzLCBJLKH+A+A3ZSAiByEMmcC0FO7Q7eTVSW+ectBY9HQ2oFDTNXak Ns9fIa0hPPVUfqfq5i3OjohIAEC6fYAWB6ERFUBSDxLuAtZSA7JSA/JoH/LPH+BIAtK7bZG4BGAt JCApH+A+AhJPGnJOA0AfH+AiBIDoWuBzc6AICvD8PxUAWJWWj2sfWwsPP9cXWzKc/PGE0jPQvwiE 9kWsAeB+EVI2AaCpdQAbGjGnHkA4AbJPHhI0A2AZI0A6AdI1drG+A6AfG/HhGjfVI1fLfJGpfBGq BGAZHsAvKJA+j7eZencVNSlZGBelINchalStIqbTV7J5I8BDLtdrJPfLfSAZHkA2AXfPfXENJTfl JrK0BFdjMoBOArX7UsYhUxaledgRcZgVBLMTepVU17StSxXlMrZzbgAtLDgrdoAlGiBBK1dhZZLi AxK/OFORRVNNhbhu0PhhafgTW1gXilgbgRhzUbV/OOAwDPFMAxe/ZrPriXXnWJidP3hiuZgPiphk eGA8EUJGWEc3MNhtgdUXSvi5XxS1Z3Zhe/MzbfjNRRUrieVpeejtOg0OseDeNYQexAIEE4GcX2B4 EkRKAMDsNjXSTWcVVVjvi1jzh1WHM/j8AvkAAvkFlHkLjVjdjZF4pPk0nANKLKQIBCESL2D6F6XG XgvWqwHQMqEsGIRaL6RSAgD6bcQAbyNRBkLQDCWNhdNxkPj1V5h3i7N7lLlPlTmtUpjTTHak1tUE DHDyQMDWN+ggQIByEiJOEGOcGIG8Okq7CMXay2KMG8HaU4rcMGCIE8aSBcEWL2Ci4ccggbKXhxlD j3jPPdmxbdLnm3PxNJhZkO1CseseDUNKBIEMJWB8EqhCBqEuGgH+DiFiL+GXQcINAEIhl4McIkXc IICurWWuDulCNiDW85VVi3mplHmuH/j/oZM9odRTohkNNQnjYIc6QIAILmH+BgEOLOtabCUcIPpV qiIOHMKeOQGoJ6C+FaMwAuEJT474DYTfliJQrbgjmrj5OBoXkDobrTJ/LsArX+LQDI72nBlkJQLv C2FBpAEoGOMGGhqsvKH7nlqpqlnpsKKMHgJ+tgGQPKCmFIaSAwD8LOA4EGJWBiCdEFH1G/iPp0Av rXlRrboTdXLDdrI0BSBkD4H/n8JOAkD7o0EwhCECF+PKG6HceBsQRtsPtyOuq4/+22RcFYJGBkBK Ddp4AtI8A2AdHkA/H9uPZZrVp5lNp9OFMzK7iLJ9HhHlgrJ8BSA3FSEWECe+vNsVnkXdt4MVqnvR qiHmHkKgF+F0JqEMD8fECQB4D7uPI8AyAZGjubJ8BAAqClbqAng4AbfZdsKyBTuMDiDOE2H+FmFe JGHMU1vXwrvVwrwxCKagF3pADyDiFEH+CWCAEHJ+A3G4B+BmDmH+DQC7EEFGE4NKHYHUXvwxxqIj t3xtxyIaw0ImGwGus8HgUZx1yGavxxyJyPyRySKIKQFuFibQGOGEYFyVynypymKQCaB/xIDkDOE8 IQFSFEVGE+EsQIIWGEF4L+EwEeP6IOE4EoN2DOC0ErpCDLy6DYDAEyH+FiFU6EHyq6H+DqDUFAH+ G4GwOWIWw0IIEMD68HzsE0H+DgDIE4H+DaDD0cG8G4X2FgFS6EFGE2LAIWG0GsHMH+EoEWTaIOFn uEH+DKCzzkDjzoH+DJ1aH+FuFgbRsGIIEKD2fZyCrEIOGUGMI2E8EqNLz6XmDsDXEgDUC6EwH+DW C9zx1ZzkF4FuOOHUq0D8Dnb8DeDHwd0h0kEeEKQsKQCkCLFKDv2SIQD6DkFJENg2H/ygYFxkPCBu BQDkH+CwCXIsIOCiCJb2CQB0D+H+FEE0LoEYEGTOCgCGEKH/2uMrGfI8GEF7RyIWHwOoH+CEBptU EaEITOFCEyLOF0FoJr16aUDpb9fbH+GAF0RSz2KgCABjdACSB4ED3V3YH+C8CjbCFAEwLP006EG8 G2X35LbnK6B/5iH+GcGSs8FoFehCC0CYEavXAT5WRSDyDdxACKBtvwFwFkJqHEHAUVxeLACSB3HZ 2GNKEyEgP6EsEaNEKQCn3MH+DuDZEgIOEUEAZMDuDbxB7jFL5zbCDQC2Et2cC/zwIP8AH+BoBKDf 1iCxEEC4CfFgEqEYNFCeU4BWA2DSXIGgXGIWu+XmCQB35r2R0EEgEOQyIWEOD88GECDvVmCyCZIs DACpGaE0EkN+DP8GIQEGDyFUH/3LFL0Ue+HkHh18IMHPwoCqCRe54ILoCMBvEKC6Cl8CC18IIWFa FKL2DGCtEEIXE0MH2+H+EKD4fZ10fZ374Z7h7l7p7sIMEF9eRd6d1p1t4H4L/ohCDECsEkIQIAVC MiX+dTUoH++n0/X++4W/4hEHe7Xo/xQGTO/zQXEu/0Cd1S/0Yglg/3g73q/yWPUE/zsa1C/0IeVW /0GeFU/3G4Xa/0MfVY/1or2bQlbRUiiVpJpQ/y+U0jEYggzzOSkRUM/0KfJqfzop3+p1Aw3+53I7 n+UCJWak3m26H+OxYdX+Xiij6leVRY3+YKheaks1Yy3+VCPBEIe5zM5qOxbdICAADgEAAAMAAAAB ADgAAAEBAAMAAAABAEAAAAECAAMAAAAEAAAWdgEDAAMAAAABAAUAAAEGAAMAAAABAAIAAAERAAQA AAABAAAACAEVAAMAAAABAAQAAAEWAAQAAAABAAAAkgEXAAQAAAABAAAVwAEaAAUAAAABAAAWfgEb AAUAAAABAAAWhgEcAAMAAAABAAEAAAEoAAMAAAABAAIAAAFSAAMAAAABAAEAAAAAAAAACAAIAAgA CAAK/IAAACcQAAr8gAAAJxA= --Apple-Mail-36--1003286864 Content-Type: multipart/alternative; boundary=Apple-Mail-38--1003286863 --Apple-Mail-38--1003286863 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-38--1003286863 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-38--1003286863-- --Apple-Mail-36--1003286864-- From pgsql-performance-owner@postgresql.org Tue Jul 12 14:51:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E4B2352A2C for ; Tue, 12 Jul 2005 14:50:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25938-10 for ; Tue, 12 Jul 2005 17:50:53 +0000 (GMT) Received: from mailhost.intellivid.com (mailhost.intellivid.com [64.32.200.11]) by svr1.postgresql.org (Postfix) with ESMTP id 123EF5283C for ; Tue, 12 Jul 2005 14:50:52 -0300 (ADT) Received: from [192.168.2.68] (spectre.intellivid.com [192.168.2.68]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by newmail.intellivid.com (Postfix) with ESMTP id E7B73F1829E; Tue, 12 Jul 2005 13:50:51 -0400 (EDT) Subject: Re: cost-based vacuum From: Ian Westmacott To: Simon Riggs Cc: pgsql-performance@postgresql.org In-Reply-To: <1121154338.3970.106.camel@localhost.localdomain> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> <1121154338.3970.106.camel@localhost.localdomain> Content-Type: text/plain Organization: Intellivid Corp. Message-Id: <1121190651.10346.342.camel@spectre.intellivid.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Tue, 12 Jul 2005 13:50:51 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/122 X-Sequence-Number: 13363 On Tue, 2005-07-12 at 03:45, Simon Riggs wrote: > Since vacuum_cost_delay is a userset parameter, you should be able to > SET this solely for the analyze_thread. That way we will know with more > certainty that it is the analyze_thread that is interfering. That is what I have been doing. In fact, I have eliminated the reader_thread and analyze_thread. I just have the writer_thread running, and a psql connection with which I perform ANALYZE, for various vacuum_cost_* parameters. (I'm trying to extract a reproducible experiment) It appears not to matter whether it is one of the tables being written to that is ANALYZEd. I can ANALYZE an old, quiescent table, or a system table and see this effect. > What is your default_statistics_target? All other configs are default; default_statistics_target=10. > Do you have other stats targets set? No. The only thing slightly out of the ordinary with the tables is that they are created WITHOUT OIDS. Some indexes, but no primary keys. All columns NOT NULL. > How long does ANALYZE take to run, with/without the vacuum_cost_delay? Well, on one table with about 50K rows, it takes about 1/4s to ANALYZE with vacuum_cost_delay=0, and about 15s with vacuum_cost_delay=1000. Other things of note: - VACUUM has the same effect. If I VACUUM or ANALYZE the whole DB, the CPU spikes reset between tables. - vmstat reports blocks written drops as the CPU rises. Don't know if it is cause or effect yet. On a small test system, I'm writing about 1.5MB/s. After about 20s of cost-based ANALYZE, this drops under 0.5MB/s. - this is a dual Xeon. I have tried both with and without hyperthreading. I haven't tried to reproduce it elsewhere yet, but will. - Looking at oprofile reports for 10-minute runs of a database-wide VACUUM with vacuum_cost_delay=0 and 1000, shows the latter spending a lot of time in LWLockAcquire and LWLockRelease (20% each vs. 2%). Thanks, --Ian From pgsql-performance-owner@postgresql.org Tue Jul 12 15:01:05 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E723752A7B for ; Tue, 12 Jul 2005 15:01:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 29766-05 for ; Tue, 12 Jul 2005 18:00:50 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.198]) by svr1.postgresql.org (Postfix) with ESMTP id 9845752A2C for ; Tue, 12 Jul 2005 15:00:49 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 71so8059wra for ; Tue, 12 Jul 2005 11:00:50 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=HAJcu+o9+qAhSOaA1JM+cpPsdvf92FdqCF+Rs1PROlRtFquyWg1x46tuQOfTbyU1H5hzbdW8RKRpajP2piuzpZWHGkgkxmE7owzf+y0dt/kD5WLXUJBrOGKB2zm4n9dqrjgx2kuiQIru5doXEQyrPQ6AK2dBYpf4L0PPav95Ovs= Received: by 10.54.106.13 with SMTP id e13mr31902wrc; Tue, 12 Jul 2005 11:00:49 -0700 (PDT) Received: by 10.54.22.15 with HTTP; Tue, 12 Jul 2005 11:00:49 -0700 (PDT) Message-ID: Date: Tue, 12 Jul 2005 13:00:49 -0500 From: Matthew Nuzum Reply-To: newz@bearfruit.org To: Yves Vindevogel Subject: Re: Projecting currentdb to more users Cc: pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.985 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP, RCVD_IN_BL_SPAMCOP_NET X-Spam-Level: X-Archive-Number: 200507/123 X-Sequence-Number: 13364 On 7/12/05, Yves Vindevogel wrote: > Hi, >=20 > We have a couple of database that are identical (one for each customer). > They are all relatively small, ranging from 100k records to 1m records. > There's only one main table with some smaller tables, a lot of indexes > and some functions. >=20 > I would like to make an estimation of the performance, the diskspace > and other related things, > when we have database of for instance 10 million records or 100 million > records. >=20 > Is there any math to be done on that ? Its pretty easy to make a database run fast with only a few thousand records, or even a million records, however things start to slow down non-linearly when the database grows too big to fit in RAM. I'm not a guru, but my attempts to do this have not been very accurate. Maybe (just maybe) you could get an idea by disabling the OS cache on the file system(s) holding the database and then somehow fragmenting the drive severly (maybe by putting each table in it's own disk partition?!?) and measuring performance. On the positive side, there are a lot of wise people on this list who have +++ experience optimzing slow queries on big databases. So queries now that run in 20 ms but slow down to 7 seconds when your tables grow will likely benefit from optimizing. --=20 Matthew Nuzum www.bearfruit.org From pgsql-general-owner@postgresql.org Tue Jul 12 15:25:53 2005 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 21E6252A74 for ; Tue, 12 Jul 2005 15:25:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40449-01 for ; Tue, 12 Jul 2005 18:25:33 +0000 (GMT) Received: from imail02.thexchange.com (imail.arbinet.com [64.74.47.121]) by svr1.postgresql.org (Postfix) with ESMTP id 252DC52A1D for ; Tue, 12 Jul 2005 15:25:32 -0300 (ADT) Received: from imail02.thexchange.com (localhost [127.0.0.1]) by imail02.thexchange.com (8.12.10/8.12.2) with ESMTP id j6CIekBu008128 for ; Tue, 12 Jul 2005 18:40:46 GMT Received: from vamail01.TheXchange.com (mailbox.arbinet.com [64.74.47.120]) by imail02.thexchange.com (8.12.2/8.12.2) with SMTP id j6CIejdA008125; Tue, 12 Jul 2005 18:40:45 GMT X-MimeOLE: Produced By Microsoft Exchange V6.0.6603.0 content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: Re: [PERFORM] Projecting currentdb to more users Date: Tue, 12 Jul 2005 18:24:52 -0000 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Projecting currentdb to more users Thread-Index: AcWHC/l3wnTfUa+vQi6A9Zbm3m7FswAAudvA From: "Mohan, Ross" To: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.11 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/328 X-Sequence-Number: 80426 From AMD's suit against Intel. Perhaps relevant to some PG/AMD issues.=20 "...125. Intel has designed its compiler purposely to degrade = performance when a program is run on an AMD platform. To achieve this, Intel designed the compiler = to compile code along several alternate code paths. Some paths are executed when the = program runs on an Intel platform and others are executed when the program is operated on a = computer with an AMD microprocessor. (The choice of code path is determined when the program = is started, using a feature known as "CPUID" which identifies the computer's = microprocessor.) By design, the code paths were not created equally. If the program detects a "Genuine = Intel" microprocessor, it executes a fully optimized code path and operates with the maximum = efficiency. However, if the program detects an "Authentic AMD" microprocessor, it executes a = different code path that will degrade the program's performance or cause it to crash..." From pgsql-general-owner@postgresql.org Tue Jul 12 15:41:17 2005 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0358152A80 for ; Tue, 12 Jul 2005 15:41:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42972-04 for ; Tue, 12 Jul 2005 18:41:14 +0000 (GMT) Received: from koolancexeon.g2switchworks.com (mail.g2switchworks.com [63.87.162.25]) by svr1.postgresql.org (Postfix) with ESMTP id 7C51252A88 for ; Tue, 12 Jul 2005 15:41:14 -0300 (ADT) Received: mail.g2switchworks.com 10.10.1.8 from 10.10.1.37 10.10.1.37 via HTTP with MS-WebStorage 6.5.6944 Received: from state.g2switchworks.com by mail.g2switchworks.com; 12 Jul 2005 13:41:14 -0500 Subject: Re: [PERFORM] Projecting currentdb to more users From: Scott Marlowe To: "Mohan, Ross" Cc: pgsql-general@postgresql.org In-Reply-To: References: Content-Type: text/plain Content-Transfer-Encoding: 7bit Message-Id: <1121193674.8208.236.camel@state.g2switchworks.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 (1.4.6-2) Date: Tue, 12 Jul 2005 13:41:14 -0500 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/332 X-Sequence-Number: 80430 On Tue, 2005-07-12 at 13:24, Mohan, Ross wrote: > From AMD's suit against Intel. Perhaps relevant to some PG/AMD issues. > > "...125. Intel has designed its compiler purposely to degrade performance when a program > is run on an AMD platform. To achieve this, Intel designed the compiler to compile code > along several alternate code paths. Some paths are executed when the program runs on an Intel > platform and others are executed when the program is operated on a computer with an AMD > microprocessor. (The choice of code path is determined when the program is started, using a > feature known as "CPUID" which identifies the computer's microprocessor.) By design, the > code paths were not created equally. If the program detects a "Genuine Intel" microprocessor, > it executes a fully optimized code path and operates with the maximum efficiency. However, > if the program detects an "Authentic AMD" microprocessor, it executes a different code path > that will degrade the program's performance or cause it to crash..." Well, this is, right now, just AMD's supposition about Intel's behaviour, I'm not sure one way or the other if Intel IS doing this. Being a big, money hungry company, I wouldn't be surprised if they are, but I don't think it would affect postgresql for most people, since they would be using the gcc compiler. From pgsql-general-owner@postgresql.org Tue Jul 12 17:06:29 2005 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 48BB252968 for ; Tue, 12 Jul 2005 17:06:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68716-01 for ; Tue, 12 Jul 2005 20:06:14 +0000 (GMT) Received: from rutherford.zen.co.uk (rutherford.zen.co.uk [212.23.3.142]) by svr1.postgresql.org (Postfix) with ESMTP id 71E39529F1 for ; Tue, 12 Jul 2005 17:06:13 -0300 (ADT) Received: from [62.3.112.246] (helo=gweek.purplebat.com) by rutherford.zen.co.uk with esmtp (Exim 4.34) id 1DsR18-0003tP-7N; Tue, 12 Jul 2005 20:06:10 +0000 Received: from mrae by gweek.purplebat.com with local (Exim 3.36 #1 (Debian)) id 1DsR17-0006WK-00; Tue, 12 Jul 2005 21:06:09 +0100 Date: Tue, 12 Jul 2005 21:06:09 +0100 To: Scott Marlowe Cc: "Mohan, Ross" , pgsql-general@postgresql.org Subject: Re: [PERFORM] Projecting currentdb to more users Message-ID: <20050712200609.GA24958@purplebat.com> References: <1121193674.8208.236.camel@state.g2switchworks.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1121193674.8208.236.camel@state.g2switchworks.com> User-Agent: Mutt/1.5.9i From: Mark Rae X-Originating-Rutherford-IP: [62.3.112.246] X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.033 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/347 X-Sequence-Number: 80445 On Tue, Jul 12, 2005 at 01:41:14PM -0500, Scott Marlowe wrote: > On Tue, 2005-07-12 at 13:24, Mohan, Ross wrote: > > From AMD's suit against Intel. Perhaps relevant to some PG/AMD issues. > Well, this is, right now, just AMD's supposition about Intel's > behaviour, I'm not sure one way or the other if Intel IS doing this. I think its more a case of AMD now having solid evidence to back up the claims. This discovery, and that fact that you could get round it by toggling some flags, was being discussed on various HPC mailing lists around about the beginning of this year. -Mark From pgsql-general-owner@postgresql.org Tue Jul 12 17:11:45 2005 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D8FBB529BF for ; Tue, 12 Jul 2005 17:11:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 66866-06 for ; Tue, 12 Jul 2005 20:11:35 +0000 (GMT) Received: from koolancexeon.g2switchworks.com (mail.g2switchworks.com [63.87.162.25]) by svr1.postgresql.org (Postfix) with ESMTP id E768E52968 for ; Tue, 12 Jul 2005 17:11:34 -0300 (ADT) Received: mail.g2switchworks.com 10.10.1.8 from 10.10.1.37 10.10.1.37 via HTTP with MS-WebStorage 6.5.6944 Received: from state.g2switchworks.com by mail.g2switchworks.com; 12 Jul 2005 15:11:35 -0500 Subject: Re: [PERFORM] Projecting currentdb to more users From: Scott Marlowe To: Mark Rae Cc: "Mohan, Ross" , pgsql-general@postgresql.org In-Reply-To: <20050712200609.GA24958@purplebat.com> References: <1121193674.8208.236.camel@state.g2switchworks.com> <20050712200609.GA24958@purplebat.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Message-Id: <1121199095.8208.245.camel@state.g2switchworks.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 (1.4.6-2) Date: Tue, 12 Jul 2005 15:11:35 -0500 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/348 X-Sequence-Number: 80446 On Tue, 2005-07-12 at 15:06, Mark Rae wrote: > On Tue, Jul 12, 2005 at 01:41:14PM -0500, Scott Marlowe wrote: > > On Tue, 2005-07-12 at 13:24, Mohan, Ross wrote: > > > From AMD's suit against Intel. Perhaps relevant to some PG/AMD issues. > > Well, this is, right now, just AMD's supposition about Intel's > > behaviour, I'm not sure one way or the other if Intel IS doing this. > > I think its more a case of AMD now having solid evidence to back > up the claims. > > This discovery, and that fact that you could get round it by > toggling some flags, was being discussed on various HPC mailing > lists around about the beginning of this year. Wow! That's pretty fascinating. So, is the evidence pretty overwhelming that this was not simple incompetence, but real malice? I could see either one being a cause of this issue, and wouldn't really be surprised by either one. From pgsql-general-owner@postgresql.org Tue Jul 12 17:32:25 2005 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A5E04529A1 for ; Tue, 12 Jul 2005 17:32:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 77817-04 for ; Tue, 12 Jul 2005 20:32:15 +0000 (GMT) Received: from pythagoras.zen.co.uk (pythagoras.zen.co.uk [212.23.3.140]) by svr1.postgresql.org (Postfix) with ESMTP id B614D52968 for ; Tue, 12 Jul 2005 17:32:14 -0300 (ADT) Received: from [62.3.112.246] (helo=gweek.purplebat.com) by pythagoras.zen.co.uk with esmtp (Exim 4.30) id 1DsRQL-0002vw-So; Tue, 12 Jul 2005 20:32:13 +0000 Received: from mrae by gweek.purplebat.com with local (Exim 3.36 #1 (Debian)) id 1DsRQL-0006YZ-00; Tue, 12 Jul 2005 21:32:13 +0100 Date: Tue, 12 Jul 2005 21:32:13 +0100 To: Scott Marlowe Cc: "Mohan, Ross" , pgsql-general@postgresql.org Subject: Re: [PERFORM] Projecting currentdb to more users Message-ID: <20050712203213.GB24958@purplebat.com> References: <1121193674.8208.236.camel@state.g2switchworks.com> <20050712200609.GA24958@purplebat.com> <1121199095.8208.245.camel@state.g2switchworks.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1121199095.8208.245.camel@state.g2switchworks.com> User-Agent: Mutt/1.5.9i From: Mark Rae X-Originating-Pythagoras-IP: [62.3.112.246] X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.033 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/352 X-Sequence-Number: 80450 On Tue, Jul 12, 2005 at 03:11:35PM -0500, Scott Marlowe wrote: > On Tue, 2005-07-12 at 15:06, Mark Rae wrote: > > I think its more a case of AMD now having solid evidence to back > > up the claims. > > Wow! That's pretty fascinating. So, is the evidence pretty > overwhelming that this was not simple incompetence, but real malice? I suppose that depends on the exact nature of the 'check'. As far as I was aware it was more a case of 'I don't recognise this processor, so I'll do it the slow but safe way'. However from what AMD are claiming, it seems to be more of a 'Its an AMD processor so I'll be deliberately slow and buggy' Having said that, I have tried compiling PG with the intel compiler in the past, and haven't noticed any real difference. But in a database there isn't much scope for vectorization and pipelining compared with numerical code, which is where the Intel compiler makes the greatest difference. -Mark From pgsql-performance-owner@postgresql.org Tue Jul 12 18:54:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E478E52AFD for ; Tue, 12 Jul 2005 18:54:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07286-02 for ; Tue, 12 Jul 2005 21:54:41 +0000 (GMT) Received: from mail.jobflash.com (oncallweb2.jobflash.com [64.62.211.41]) by svr1.postgresql.org (Postfix) with ESMTP id 9974952ACC for ; Tue, 12 Jul 2005 18:54:40 -0300 (ADT) Received: from [192.168.2.22] (user-38lc18a.dialup.mindspring.com [209.86.5.10]) by mail.jobflash.com (Postfix) with ESMTP id 883DC420E4; Tue, 12 Jul 2005 14:54:41 -0700 (PDT) Message-ID: <42D43C17.90005@jobflash.com> Date: Tue, 12 Jul 2005 14:54:31 -0700 From: Tom Arthurs User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Brent Henry Cc: pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050712062247.37243.qmail@web33902.mail.mud.yahoo.com> In-Reply-To: <20050712062247.37243.qmail@web33902.mail.mud.yahoo.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/126 X-Sequence-Number: 13367 I have this in my postgresql.conf file and it works fine (set the min to whatever you want to log) log_min_duration_statement = 3000 # -1 is disabled, in milliseconds. Another setting that might get what you want: #log_duration = false uncomment and change to true. From the docs: (http://www.postgresql.org/docs/8.0/interactive/runtime-config.html) Causes the duration of every completed statement which satisfies log_statement to be logged. When using this option, if you are not using syslog, it is recommended that you log the PID or session ID using log_line_prefix so that you can link the statement to the duration using the process ID or session ID. The default is off. Only superusers can change this setting. Brent Henry wrote: > Help! After recently migrating to Postgres 8, I've > discovered to my horror that I can't determine which > queries are poorly performing anymore because the > logging has drastically changed and no longer shows > durations for anything done through JDBC. > > So I'm desperately trying to do performance tuning on > my servers and have no way to sort out which > statements are the slowest. > > Does anyone have any suggestions? How do you > determine what queries are behaving badly when you > can't get durations out of the logs? > > I have a perl script that analyzes the output from > Postgres 7 logs and it works great! But it relies on > the duration being there. > > I did some searches on postgresql.org mailing lists > and have seen a few people discussing this problem, > but noone seems to be too worried about it. Is there > a simple work-around? > > Sincerely, > > Brent > > > > ____________________________________________________ > Sell on Yahoo! Auctions � no fees. Bid on great items. > http://auctions.yahoo.com/ > > ---------------------------(end of broadcast)--------------------------- > TIP 5: don't forget to increase your free space map settings > > > From pgsql-performance-owner@postgresql.org Tue Jul 12 20:32:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DAB6D52AD2 for ; Tue, 12 Jul 2005 20:32:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 26593-08 for ; Tue, 12 Jul 2005 23:32:32 +0000 (GMT) Received: from web33901.mail.mud.yahoo.com (web33901.mail.mud.yahoo.com [66.163.178.65]) by svr1.postgresql.org (Postfix) with SMTP id 92CF252ACE for ; Tue, 12 Jul 2005 20:32:30 -0300 (ADT) Received: (qmail 5468 invoked by uid 60001); 12 Jul 2005 23:32:30 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=tRlxZBNqo0nFert/UevCiQarJFH9FX68N2hz1J4K4wwGvFVabe2P97L84LJ3Lpy/huvVlvVvZDpNtwaqqXUYq43On0+9PPwLBxBjVsa6rQKWKJ0CZiUvQjCiqxS+MY4X0Vw9mAnbfhe8r7q83JEW/vNVMM4cR6tqrOwfFEIvC/c= ; Message-ID: <20050712233230.5466.qmail@web33901.mail.mud.yahoo.com> Received: from [67.106.37.5] by web33901.mail.mud.yahoo.com via HTTP; Tue, 12 Jul 2005 16:32:30 PDT Date: Tue, 12 Jul 2005 16:32:30 -0700 (PDT) From: Brent Henry Subject: Re: General DB Tuning To: Tom Arthurs Cc: pgsql-performance@postgresql.org In-Reply-To: <42D43C17.90005@jobflash.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=2.174 tagged_above=0 required=5 tests=FORGED_YAHOO_RCVD X-Spam-Level: ** X-Archive-Number: 200507/127 X-Sequence-Number: 13368 Yes, that is exactly what I want to use! Unfortunately, it doesn't work if you access postgres through a JDBC connection. I don't know why. I found a posting from back in February which talks aobut this a little: http://archives.postgresql.org/pgsql-admin/2005-02/msg00055.php But I can't find anywhere where someone has fixed it. Am I the only one accessing postgres through JDBC? -Brent --- Tom Arthurs wrote: > I have this in my postgresql.conf file and it works > fine (set the min to > whatever you want to log) > log_min_duration_statement = 3000 # -1 is disabled, > in milliseconds. > > Another setting that might get what you want: > > #log_duration = false > > uncomment and change to true. > > From the docs: > (http://www.postgresql.org/docs/8.0/interactive/runtime-config.html) > > Causes the duration of every completed statement > which satisfies > log_statement to be logged. When using this option, > if you are not using > syslog, it is recommended that you log the PID or > session ID using > log_line_prefix so that you can link the statement > to the duration using > the process ID or session ID. The default is off. > Only superusers can > change this setting. > > Brent Henry wrote: > > Help! After recently migrating to Postgres 8, > I've > > discovered to my horror that I can't determine > which > > queries are poorly performing anymore because the > > logging has drastically changed and no longer > shows > > durations for anything done through JDBC. > > > > So I'm desperately trying to do performance tuning > on > > my servers and have no way to sort out which > > statements are the slowest. > > > > Does anyone have any suggestions? How do you > > determine what queries are behaving badly when you > > can't get durations out of the logs? > > > > I have a perl script that analyzes the output from > > Postgres 7 logs and it works great! But it relies > on > > the duration being there. > > > > I did some searches on postgresql.org mailing > lists > > and have seen a few people discussing this > problem, > > but noone seems to be too worried about it. Is > there > > a simple work-around? > > > > Sincerely, > > > > Brent > > > > > > > > > ____________________________________________________ > > Sell on Yahoo! Auctions � no fees. Bid on great > items. > > http://auctions.yahoo.com/ > > > > ---------------------------(end of > broadcast)--------------------------- > > TIP 5: don't forget to increase your free space > map settings > > > > > > > > ---------------------------(end of > broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From pgsql-performance-owner@postgresql.org Tue Jul 12 21:37:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D1A35529B3 for ; Tue, 12 Jul 2005 21:37:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42775-02 for ; Wed, 13 Jul 2005 00:36:53 +0000 (GMT) Received: from mail.jobflash.com (oncallweb2.jobflash.com [64.62.211.41]) by svr1.postgresql.org (Postfix) with ESMTP id 5436D52A11 for ; Tue, 12 Jul 2005 21:36:52 -0300 (ADT) Received: from [192.168.2.22] (user-38lc18a.dialup.mindspring.com [209.86.5.10]) by mail.jobflash.com (Postfix) with ESMTP id 278C8420E3; Tue, 12 Jul 2005 17:36:55 -0700 (PDT) Message-ID: <42D4621D.4060803@jobflash.com> Date: Tue, 12 Jul 2005 17:36:45 -0700 From: Tom Arthurs User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Brent Henry Cc: pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050712233230.5466.qmail@web33901.mail.mud.yahoo.com> In-Reply-To: <20050712233230.5466.qmail@web33901.mail.mud.yahoo.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/128 X-Sequence-Number: 13369 we are using jdbc -- the "log_min_duration_statement = 3000 " statement works fine for me. Looks like there's no other work around for the bug(?). Not sure since I have no interest in logging a million statements a day, I only want to see the poorly performing hits. Brent Henry wrote: > Yes, that is exactly what I want to use! > > Unfortunately, it doesn't work if you access postgres > through a JDBC connection. I don't know why. I found > a posting from back in February which talks aobut this > a little: > > http://archives.postgresql.org/pgsql-admin/2005-02/msg00055.php > > But I can't find anywhere where someone has fixed it. > Am I the only one accessing postgres through JDBC? > > -Brent > > > --- Tom Arthurs wrote: > > >>I have this in my postgresql.conf file and it works >>fine (set the min to >>whatever you want to log) >>log_min_duration_statement = 3000 # -1 is disabled, >>in milliseconds. >> >>Another setting that might get what you want: >> >>#log_duration = false >> >>uncomment and change to true. >> >> From the docs: >> > > (http://www.postgresql.org/docs/8.0/interactive/runtime-config.html) > >> Causes the duration of every completed statement >>which satisfies >>log_statement to be logged. When using this option, >>if you are not using >>syslog, it is recommended that you log the PID or >>session ID using >>log_line_prefix so that you can link the statement >>to the duration using >>the process ID or session ID. The default is off. >>Only superusers can >>change this setting. >> >>Brent Henry wrote: >> >>>Help! After recently migrating to Postgres 8, >> >>I've >> >>>discovered to my horror that I can't determine >> >>which >> >>>queries are poorly performing anymore because the >>>logging has drastically changed and no longer >> >>shows >> >>>durations for anything done through JDBC. >>> >>>So I'm desperately trying to do performance tuning >> >>on >> >>>my servers and have no way to sort out which >>>statements are the slowest. >>> >>>Does anyone have any suggestions? How do you >>>determine what queries are behaving badly when you >>>can't get durations out of the logs? >>> >>>I have a perl script that analyzes the output from >>>Postgres 7 logs and it works great! But it relies >> >>on >> >>>the duration being there. >>> >>>I did some searches on postgresql.org mailing >> >>lists >> >>>and have seen a few people discussing this >> >>problem, >> >>>but noone seems to be too worried about it. Is >> >>there >> >>>a simple work-around? >>> >>>Sincerely, >>> >>>Brent >>> >>> >>> >>> >> >>____________________________________________________ >> >>>Sell on Yahoo! Auctions � no fees. Bid on great >> >>items. >> >>>http://auctions.yahoo.com/ >>> >>>---------------------------(end of >> >>broadcast)--------------------------- >> >>>TIP 5: don't forget to increase your free space >> >>map settings >> >>> >>> >>---------------------------(end of >>broadcast)--------------------------- >>TIP 2: Don't 'kill -9' the postmaster >> > > > > __________________________________________________ > Do You Yahoo!? > Tired of spam? Yahoo! Mail has the best spam protection around > http://mail.yahoo.com > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend > > > From pgsql-performance-owner@postgresql.org Tue Jul 12 22:05:23 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E3987529E2 for ; Tue, 12 Jul 2005 22:05:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58266-06 for ; Wed, 13 Jul 2005 01:05:13 +0000 (GMT) Received: from lakermmtao09.cox.net (lakermmtao09.cox.net [68.230.240.30]) by svr1.postgresql.org (Postfix) with ESMTP id A46C752A0F for ; Tue, 12 Jul 2005 22:05:11 -0300 (ADT) Received: from [127.0.0.1] (really [68.3.22.74]) by lakermmtao09.cox.net (InterMail vM.6.01.04.00 201-2131-118-20041027) with ESMTP id <20050713010513.KULZ15825.lakermmtao09.cox.net@[127.0.0.1]> for ; Tue, 12 Jul 2005 21:05:13 -0400 Message-ID: <42D468C9.9070200@works4me.com> Date: Tue, 12 Jul 2005 18:05:13 -0700 From: Dennis User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050712233230.5466.qmail@web33901.mail.mud.yahoo.com> <42D4621D.4060803@jobflash.com> In-Reply-To: <42D4621D.4060803@jobflash.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit X-Antivirus: avast! (VPS 0528-2, 07/12/2005), Outbound message X-Antivirus-Status: Clean X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/129 X-Sequence-Number: 13370 Tom Arthurs wrote: > we are using jdbc -- the "log_min_duration_statement = 3000 " > statement works fine for me. Looks like there's no other work around > for the bug(?). Not sure since I have no interest in logging a > million statements a day, I only want to see the poorly performing hits. Doesn't it depend on what jdbc driver you are using? Dennis From pgsql-performance-owner@postgresql.org Tue Jul 12 22:14:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0B59552802 for ; Tue, 12 Jul 2005 22:14:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 57461-07 for ; Wed, 13 Jul 2005 01:14:37 +0000 (GMT) Received: from houston.familyhealth.com.au (houston.au.fhnetwork.com [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 0168752965 for ; Tue, 12 Jul 2005 22:14:33 -0300 (ADT) Received: from houston.familyhealth.com.au (localhost [127.0.0.1]) by houston.familyhealth.com.au (Postfix) with ESMTP id 97D5824FE1; Wed, 13 Jul 2005 09:14:33 +0800 (WST) Received: from [127.0.0.1] (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (Postfix) with ESMTP id 7E63124FE0; Wed, 13 Jul 2005 09:14:33 +0800 (WST) Message-ID: <42D46BC4.2010907@familyhealth.com.au> Date: Wed, 13 Jul 2005 09:17:56 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dennis Cc: pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050712233230.5466.qmail@web33901.mail.mud.yahoo.com> <42D4621D.4060803@jobflash.com> <42D468C9.9070200@works4me.com> In-Reply-To: <42D468C9.9070200@works4me.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.063 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/130 X-Sequence-Number: 13371 >> we are using jdbc -- the "log_min_duration_statement = 3000 " >> statement works fine for me. Looks like there's no other work around >> for the bug(?). Not sure since I have no interest in logging a >> million statements a day, I only want to see the poorly performing hits. > > Doesn't it depend on what jdbc driver you are using? It depends if he's using new-protocol prepared queries which don't get logged properly. Wasn't that fixed for 8.1 or something? Chris From pgsql-performance-owner@postgresql.org Tue Jul 12 22:30:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E81AA5281B for ; Tue, 12 Jul 2005 22:30:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 60646-10 for ; Wed, 13 Jul 2005 01:30:21 +0000 (GMT) Received: from mail.jobflash.com (oncallweb2.jobflash.com [64.62.211.41]) by svr1.postgresql.org (Postfix) with ESMTP id 216EC529C2 for ; Tue, 12 Jul 2005 22:30:20 -0300 (ADT) Received: from [192.168.2.22] (user-38lc18a.dialup.mindspring.com [209.86.5.10]) by mail.jobflash.com (Postfix) with ESMTP id 49003420CF; Tue, 12 Jul 2005 18:30:24 -0700 (PDT) Message-ID: <42D46EA6.605@jobflash.com> Date: Tue, 12 Jul 2005 18:30:14 -0700 From: Tom Arthurs User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dennis Cc: pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050712233230.5466.qmail@web33901.mail.mud.yahoo.com> <42D4621D.4060803@jobflash.com> <42D468C9.9070200@works4me.com> In-Reply-To: <42D468C9.9070200@works4me.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/131 X-Sequence-Number: 13372 hmm, yea maybe -- we are using the 7.4 driver with 8.0.x db. Dennis wrote: > Tom Arthurs wrote: > >> we are using jdbc -- the "log_min_duration_statement = 3000 " >> statement works fine for me. Looks like there's no other work around >> for the bug(?). Not sure since I have no interest in logging a >> million statements a day, I only want to see the poorly performing hits. > > > > Doesn't it depend on what jdbc driver you are using? > > Dennis > > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster > > > From pgsql-performance-owner@postgresql.org Tue Jul 12 22:36:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E101A5281B for ; Tue, 12 Jul 2005 22:36:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65765-01 for ; Wed, 13 Jul 2005 01:36:28 +0000 (GMT) Received: from web33911.mail.mud.yahoo.com (web33911.mail.mud.yahoo.com [66.163.178.75]) by svr1.postgresql.org (Postfix) with SMTP id 9EE1D52A16 for ; Tue, 12 Jul 2005 22:36:26 -0300 (ADT) Received: (qmail 18627 invoked by uid 60001); 13 Jul 2005 01:36:30 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=dlYn9uGbeNqkmptWPFwWrLks7RsgJNmSD1q5hMJaAa9dAPjqDOQUsuN4ftlFEkYMdYBqBHo1Qo5TSjAX5rdC1xkfm7deTdKCFT228Qn12Zt6zkziSUf0MgSpLdU/O07AAr695Evu0+Lo5uGlSehu5GTWoN3i35nQXgG8BM4/M4Y= ; Message-ID: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> Received: from [67.106.37.5] by web33911.mail.mud.yahoo.com via HTTP; Tue, 12 Jul 2005 18:36:30 PDT Date: Tue, 12 Jul 2005 18:36:30 -0700 (PDT) From: Brent Henry Subject: Re: General DB Tuning To: Christopher Kings-Lynne , Dennis , Tom Arthurs Cc: pgsql-performance@postgresql.org In-Reply-To: <42D46BC4.2010907@familyhealth.com.au> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=2.174 tagged_above=0 required=5 tests=FORGED_YAHOO_RCVD X-Spam-Level: ** X-Archive-Number: 200507/132 X-Sequence-Number: 13373 We are running Postgres 8.0.2 with the 8.0.2 jdbc driver. And yes we are using prepared statements. I've spent hours trying to get the 'log_min_duration_statement' and 'log_duration' options to work with no luck. I never get any duration from the statement. I also never see 'begin' or 'commit' in the log so I can't tell how long my batch commands are taking to commit to the DB. Is there a different kind of 'prepared' statements that we should be using in the driver to get logging to work properly? What is the 'new' protocol? Tom, what version are you using? Are you using prepared statements in JDBC? -Brent --- Christopher Kings-Lynne wrote: > >> we are using jdbc -- the > "log_min_duration_statement = 3000 " > >> statement works fine for me. Looks like there's > no other work around > >> for the bug(?). Not sure since I have no > interest in logging a > >> million statements a day, I only want to see the > poorly performing hits. > > > > Doesn't it depend on what jdbc driver you are > using? > > It depends if he's using new-protocol prepared > queries which don't get > logged properly. Wasn't that fixed for 8.1 or > something? > > Chris > > > ---------------------------(end of > broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > ____________________________________________________ Start your day with Yahoo! - make it your home page http://www.yahoo.com/r/hs From pgsql-performance-owner@postgresql.org Tue Jul 12 22:49:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 14BED52A08 for ; Tue, 12 Jul 2005 22:49:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 63838-08 for ; Wed, 13 Jul 2005 01:48:57 +0000 (GMT) Received: from houston.familyhealth.com.au (houston.au.fhnetwork.com [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id D600F529C2 for ; Tue, 12 Jul 2005 22:48:56 -0300 (ADT) Received: from houston.familyhealth.com.au (localhost [127.0.0.1]) by houston.familyhealth.com.au (Postfix) with ESMTP id 0EF3624FE2; Wed, 13 Jul 2005 09:48:57 +0800 (WST) Received: from [127.0.0.1] (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (Postfix) with ESMTP id E3E0524FE1; Wed, 13 Jul 2005 09:48:56 +0800 (WST) Message-ID: <42D473D4.7040500@familyhealth.com.au> Date: Wed, 13 Jul 2005 09:52:20 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Brent Henry Cc: Dennis , Tom Arthurs , pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> In-Reply-To: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.062 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/133 X-Sequence-Number: 13374 > Is there a different kind of 'prepared' statements > that we should be using in the driver to get logging > to work properly? What is the 'new' protocol? The 8.0.2 jdbc driver uses real prepared statements instead of faked ones. The problem is the new protocol (that the 8.0.2 driver users) has a bug where protocol-prepared queries don't get logged properly. I don't know if it's been fixed... Chris From pgsql-performance-owner@postgresql.org Tue Jul 12 22:53:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 292E352A1D for ; Tue, 12 Jul 2005 22:53:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 69050-03 for ; Wed, 13 Jul 2005 01:53:19 +0000 (GMT) Received: from mail.jobflash.com (oncallweb2.jobflash.com [64.62.211.41]) by svr1.postgresql.org (Postfix) with ESMTP id 14D9B529D5 for ; Tue, 12 Jul 2005 22:53:18 -0300 (ADT) Received: from [192.168.2.22] (user-38lc18a.dialup.mindspring.com [209.86.5.10]) by mail.jobflash.com (Postfix) with ESMTP id 87228420CF; Tue, 12 Jul 2005 18:53:21 -0700 (PDT) Message-ID: <42D47407.10301@jobflash.com> Date: Tue, 12 Jul 2005 18:53:11 -0700 From: Tom Arthurs User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Brent Henry Cc: Christopher Kings-Lynne , Dennis , pgsql-performance@postgresql.org Subject: Re: General DB Tuning References: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> In-Reply-To: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/134 X-Sequence-Number: 13375 Here's the answer for you from the jdbc list: > Alvin Hung wrote: > > >>> Currently, 8.0.2 / JDBC 8.0-310, log_min_duration_statement does not >>> work with JDBC. Nothing will get logged. This makes it very >>> difficult to tune a java application. Can you tell me when will this >>> be fixed? Thanks. > > > This is a server limitation: it does not handle logging of the V3 > extended query protocol very well. There's gradual progress being made > on it; you might want to search the pgsql-hackers and pgsql-patches > archives for details. ========================================================================================== We are using prepared statements, but we are using the 7.4 driver with the 8.0.3 server. I think it comes down to locally (on the client) prepared statements vs using server side prepared statments. I never got past this issue (changing the code is in our todo list, but pretty far down it) so I never noticed the logging issues.) I had a problem with prepared statements with the 8.x drivers -- here's what I got from the jdbc list when I asked the question: >>1. What changed between the driver versions that generate this error? > > > The driver started to use server-side prepared statements for > parameterization of queries (i.e. the driver translates ? to $n in the > main query string, and sends the actual parameter values out-of-band > from the query itself). One sideeffect of this is that parameters are > more strongly typed than in the 7.4.x versions where the driver would do > literal parameter substitution into the query string before sending it > to the backend. Also, you can use parameters in fewer places (they must > fit the backend's idea of where parameterizable expressions are allowed) > -- e.g. see the recent thread about "ORDER BY ?" changing behaviour with > the newer driver. > > >>> 2. What is the downside of continuing to use the 7.x version of the >>> driver -- or are there better alternatives (patch, new version, etc). I >>> am using build 311 of the driver. > > > Most active development happens on the 8.0 version; 7.4.x is maintained > for bugfixes but that's about it, you won't get the benefit of any > performance improvements or added features that go into 8.0. Also, the > 7.4.x driver won't necessarily work with servers >= 8.0. > > In the longer term, the 7.4.x version will eventually become unmaintained. So for the short term, you could downgrade your driver. Brent Henry wrote: > We are running Postgres 8.0.2 with the 8.0.2 jdbc > driver. And yes we are using prepared statements. > I've spent hours trying to get the > 'log_min_duration_statement' and 'log_duration' > options to work with no luck. I never get any > duration from the statement. I also never see 'begin' > or 'commit' in the log so I can't tell how long my > batch commands are taking to commit to the DB. > > Is there a different kind of 'prepared' statements > that we should be using in the driver to get logging > to work properly? What is the 'new' protocol? > > Tom, what version are you using? Are you using > prepared statements in JDBC? > > -Brent > > > --- Christopher Kings-Lynne > wrote: > > >>>>we are using jdbc -- the >> >>"log_min_duration_statement = 3000 " >> >>>>statement works fine for me. Looks like there's >> >>no other work around >> >>>>for the bug(?). Not sure since I have no >> >>interest in logging a >> >>>>million statements a day, I only want to see the >> >>poorly performing hits. >> >>>Doesn't it depend on what jdbc driver you are >> >>using? >> >>It depends if he's using new-protocol prepared >>queries which don't get >>logged properly. Wasn't that fixed for 8.1 or >>something? >> >>Chris >> >> >>---------------------------(end of >>broadcast)--------------------------- >>TIP 3: Have you checked our extensive FAQ? >> >> http://www.postgresql.org/docs/faq >> > > > > > > ____________________________________________________ > Start your day with Yahoo! - make it your home page > http://www.yahoo.com/r/hs > > > > From pgsql-performance-owner@postgresql.org Wed Jul 13 02:18:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 53D3E5284E for ; Wed, 13 Jul 2005 02:18:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 16591-05 for ; Wed, 13 Jul 2005 05:18:36 +0000 (GMT) Received: from nproxy.gmail.com (nproxy.gmail.com [64.233.182.197]) by svr1.postgresql.org (Postfix) with ESMTP id B315252BD9 for ; Wed, 13 Jul 2005 02:18:34 -0300 (ADT) Received: by nproxy.gmail.com with SMTP id x37so16816nfc for ; Tue, 12 Jul 2005 22:18:32 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=YFtvkHgUV2lPts5Hr4eWNCYh4BXqs8OBf6FB4Ywoh6lZNbTYeFRGjuKJhGdxbAzCXQKCUrjjaYMZvvMJKqhYIhZ3k1TwsOWlZX7t3lltn0PI3fOkkt8q8vUPqSif+8YF1TOutNygPVlUa9Q5sEqrSgwKgpwNXqWPDxRMT4l6UK8= Received: by 10.48.239.19 with SMTP id m19mr13102nfh; Tue, 12 Jul 2005 22:18:32 -0700 (PDT) Received: by 10.48.49.14 with HTTP; Tue, 12 Jul 2005 22:18:32 -0700 (PDT) Message-ID: <4b09a0c050712221813fa6d64@mail.gmail.com> Date: Wed, 13 Jul 2005 07:18:32 +0200 From: Jean-Max Reymond Reply-To: Jean-Max Reymond To: pgsql-performance@postgresql.org Subject: Re: Projecting currentdb to more users In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.072 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/135 X-Sequence-Number: 13376 2005/7/12, Mohan, Ross : > From AMD's suit against Intel. Perhaps relevant to some PG/AMD issues. Postgres is compiled with gnu compiler. Isn't it ? I don't know how much can Postgres benefit from an optimized Intel compiler= . --=20 Jean-Max Reymond CKR Solutions Open Source Nice France http://www.ckr-solutions.com From pgsql-performance-owner@postgresql.org Wed Jul 13 05:21:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3538C52BC1 for ; Wed, 13 Jul 2005 05:20:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58554-01 for ; Wed, 13 Jul 2005 08:20:54 +0000 (GMT) Received: from nproxy.gmail.com (nproxy.gmail.com [64.233.182.203]) by svr1.postgresql.org (Postfix) with ESMTP id 8D2D652862 for ; Wed, 13 Jul 2005 05:20:52 -0300 (ADT) Received: by nproxy.gmail.com with SMTP id n15so24528nfc for ; Wed, 13 Jul 2005 01:20:51 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=rnErMerhImKEqUJ/Pe9i+Nhhyr2f/GIxNOp1FUQVyQsz4052XNArVeK1PZv0HOi+oonaHPgiUVp03WlyFpaIv6SliEUzzSkqPSb9sbst+agUWcAS1fYCU7IHhrGoq8qL6phS5n4izai9gbQbRnHK+ThZ+2banSp4Hhk/koUkNSU= Received: by 10.48.237.12 with SMTP id k12mr17690nfh; Wed, 13 Jul 2005 01:20:51 -0700 (PDT) Received: by 10.48.49.14 with HTTP; Wed, 13 Jul 2005 01:20:51 -0700 (PDT) Message-ID: <4b09a0c05071301207d452ace@mail.gmail.com> Date: Wed, 13 Jul 2005 10:20:51 +0200 From: Jean-Max Reymond Reply-To: Jean-Max Reymond To: pgsql-performance@postgresql.org Subject: size of cache Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.069 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/136 X-Sequence-Number: 13377 with my application, it seems that size of cache has great effect: from 512 Kb of L2 cache to 1Mb boost performance with a factor 3 and 20% again from 1Mb L2 cache to 2Mb L2 cache. I don't understand why a 512Kb cache L2 is too small to fit the data's does it exist a tool to trace processor activity and confirm that processor is waiting for memory ? does it exist a tool to snapshot postgres activity and understand where we spend time and potentialy avoid the bottleneck ? thanks for your tips. --=20 Jean-Max Reymond CKR Solutions Open Source Nice France http://www.ckr-solutions.com From pgsql-performance-owner@postgresql.org Wed Jul 13 07:07:54 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D082252ABF for ; Wed, 13 Jul 2005 07:07:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 81805-06 for ; Wed, 13 Jul 2005 10:07:48 +0000 (GMT) Received: from pythagoras.zen.co.uk (pythagoras.zen.co.uk [212.23.3.140]) by svr1.postgresql.org (Postfix) with ESMTP id 0DC9252AB4 for ; Wed, 13 Jul 2005 07:07:46 -0300 (ADT) Received: from [62.3.112.246] (helo=gweek.purplebat.com) by pythagoras.zen.co.uk with esmtp (Exim 4.30) id 1Dse9V-0005RL-Fr; Wed, 13 Jul 2005 10:07:41 +0000 Received: from mrae by gweek.purplebat.com with local (Exim 3.36 #1 (Debian)) id 1Dse9U-0007HG-00; Wed, 13 Jul 2005 11:07:40 +0100 Date: Wed, 13 Jul 2005 11:07:40 +0100 To: Christopher Kings-Lynne Cc: Brent Henry , Dennis , Tom Arthurs , pgsql-performance@postgresql.org Subject: Re: General DB Tuning Message-ID: <20050713100740.GA27912@purplebat.com> References: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> <42D473D4.7040500@familyhealth.com.au> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="W/nzBZO5zC0uMSeA" Content-Disposition: inline In-Reply-To: <42D473D4.7040500@familyhealth.com.au> User-Agent: Mutt/1.5.9i From: Mark Rae X-Originating-Pythagoras-IP: [62.3.112.246] X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.033 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/138 X-Sequence-Number: 13379 --W/nzBZO5zC0uMSeA Content-Type: text/plain; charset=us-ascii Content-Disposition: inline On Wed, Jul 13, 2005 at 09:52:20AM +0800, Christopher Kings-Lynne wrote: > The 8.0.2 jdbc driver uses real prepared statements instead of faked > ones. The problem is the new protocol (that the 8.0.2 driver users) has > a bug where protocol-prepared queries don't get logged properly. > I don't know if it's been fixed... It's not in 8.0.3, but I was having the same problems with DBD::Pg so I backported some of it and also changed the code so that it listed the values of the bind parameters, so you get something like LOG: statement: SELECT sr.name,sr.seq_region_id, sr.length, 1 FROM seq_region sr WHERE sr.name = $1 AND sr.coord_system_id = $2 LOG: binding: "dbdpg_2" with 2 parameters LOG: bind "dbdpg_2" $1 = "20" LOG: bind "dbdpg_2" $2 = "1" LOG: statement: EXECUTE [PREPARE: SELECT sr.name,sr.seq_region_id, sr.length, 1 FROM seq_region sr WHERE sr.name = $1 AND sr.coord_system_id = $2] LOG: duration: 0.164 ms I've attached a patch in case anyone finds it useful. -Mark --W/nzBZO5zC0uMSeA Content-Type: text/plain; charset=us-ascii Content-Disposition: attachment; filename="timing.patch" *** postgresql-8.0.3/src/backend/tcop/postgres.c 2005-07-13 09:42:04.997669193 +0100 --- postgresql-8.0.3/src/backend/tcop/postgres.c 2005-07-13 09:34:24.618195580 +0100 *************** *** 1370,1375 **** --- 1370,1378 ---- else portal = CreatePortal(portal_name, false, false); + if (log_statement == LOGSTMT_ALL) + ereport(LOG, (errmsg("binding: \"%s\" with %d parameters", stmt_name, numParams))); + /* * Fetch parameters, if any, and store in the portal's memory context. * *************** *** 1428,1433 **** --- 1431,1439 ---- * grotty but is a big win when dealing with very * large parameter strings. */ + if (log_statement == LOGSTMT_ALL) + ereport(LOG, (errmsg("bind \"%s\" $%d = \"%s\"", stmt_name, i+1, pvalue))); + pbuf.data = (char *) pvalue; pbuf.maxlen = plength + 1; pbuf.len = plength; *************** *** 1578,1583 **** --- 1584,1593 ---- bool is_trans_exit = false; bool completed; char completionTag[COMPLETION_TAG_BUFSIZE]; + struct timeval start_t, stop_t; + bool save_log_duration = log_duration; + int save_log_min_duration_statement = log_min_duration_statement; + bool save_log_statement_stats = log_statement_stats; /* Adjust destination to tell printtup.c what to do */ dest = whereToSendOutput; *************** *** 1614,1619 **** --- 1624,1647 ---- set_ps_display(portal->commandTag); + /* + * We use save_log_* so "SET log_duration = true" and "SET + * log_min_duration_statement = true" don't report incorrect time + * because gettimeofday() wasn't called. Similarly, + * log_statement_stats has to be captured once. + */ + if (save_log_duration || save_log_min_duration_statement != -1) + gettimeofday(&start_t, NULL); + + if (save_log_statement_stats) + ResetUsage(); + + if (log_statement == LOGSTMT_ALL) + /* We have the portal, so output the source query. */ + ereport(LOG, + (errmsg("statement: EXECUTE %s [PREPARE: %s]", portal_name, + portal->sourceText ? portal->sourceText : ""))); + BeginCommand(portal->commandTag, dest); /* Check for transaction-control commands */ *************** *** 1708,1713 **** --- 1736,1785 ---- pq_putemptymessage('s'); } + /* + * Combine processing here as we need to calculate the query duration + * in both instances. + */ + if (save_log_duration || save_log_min_duration_statement != -1) + { + long usecs; + + gettimeofday(&stop_t, NULL); + if (stop_t.tv_usec < start_t.tv_usec) + { + stop_t.tv_sec--; + stop_t.tv_usec += 1000000; + } + usecs = (long) (stop_t.tv_sec - start_t.tv_sec) * 1000000 + + (long) (stop_t.tv_usec - start_t.tv_usec); + + /* Only print duration if we previously printed the statement. */ + if (log_statement == LOGSTMT_ALL && save_log_duration) + ereport(LOG, + (errmsg("duration: %ld.%03ld ms", + (long) ((stop_t.tv_sec - start_t.tv_sec) * 1000 + + (stop_t.tv_usec - start_t.tv_usec) / 1000), + (long) (stop_t.tv_usec - start_t.tv_usec) % 1000))); + + /* + * Output a duration_statement to the log if the query has + * exceeded the min duration, or if we are to print all durations. + */ + if (save_log_min_duration_statement == 0 || + (save_log_min_duration_statement > 0 && + usecs >= save_log_min_duration_statement * 1000)) + ereport(LOG, + (errmsg("duration: %ld.%03ld ms statement: EXECUTE %s [PREPARE: %s]", + (long) ((stop_t.tv_sec - start_t.tv_sec) * 1000 + + (stop_t.tv_usec - start_t.tv_usec) / 1000), + (long) (stop_t.tv_usec - start_t.tv_usec) % 1000, + portal_name, + portal->sourceText ? portal->sourceText : ""))); + } + + if (save_log_statement_stats) + ShowUsage("QUERY STATISTICS"); + debug_query_string = NULL; } --W/nzBZO5zC0uMSeA-- From pgsql-performance-owner@postgresql.org Wed Jul 13 06:58:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4D99852A22 for ; Wed, 13 Jul 2005 06:58:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79477-03 for ; Wed, 13 Jul 2005 09:58:01 +0000 (GMT) Received: from smtp2.univ-nantes.fr (Smtp2.univ-nantes.fr [193.52.82.19]) by svr1.postgresql.org (Postfix) with ESMTP id 32ADE529A9 for ; Wed, 13 Jul 2005 06:57:59 -0300 (ADT) Received: from localhost (debian [127.0.0.1]) by smtp2.univ-nantes.fr (Postfix) with ESMTP id B9B23381021 for ; Wed, 13 Jul 2005 11:58:00 +0200 (CEST) Received: from smtp2.univ-nantes.fr ([172.20.12.56]) by localhost (smtp2.univ-nantes.prive [172.20.12.56]) (amavisd-new, port 10024) with LMTP id 00670-01-47 for ; Wed, 13 Jul 2005 11:58:00 +0200 (CEST) Received: from [172.19.33.37] (unknown [172.19.33.37]) by smtp2.univ-nantes.fr (Postfix) with ESMTP id A69C4380E78 for ; Wed, 13 Jul 2005 11:58:00 +0200 (CEST) Message-ID: <42D4E836.9020607@univ-nantes.fr> Date: Wed, 13 Jul 2005 12:08:54 +0200 From: Nicolas Beaume User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: fr, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: large table vs multiple smal tables Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new-20030616-p7 (Debian) at smtp.univ-nantes.fr X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/137 X-Sequence-Number: 13378 Hello I have a large database with 4 large tables (each containing at least 200 000 rows, perhaps even 1 or 2 million) and i ask myself if it's better to split them into small tables (e.g tables of 2000 rows) to speed the access and the update of those tables (considering that i will have few update but a lot of reading). Do you think it would be efficient ? Nicolas, wondering if he hadn't be too greedy -- ------------------------------------------------------------------------- � soyez ce que vous voudriez avoir l'air d'�tre � Lewis Caroll From pgsql-performance-owner@postgresql.org Fri Jul 15 02:05:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A5C5E5294F for ; Wed, 13 Jul 2005 09:50:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18660-10 for ; Wed, 13 Jul 2005 12:50:44 +0000 (GMT) Received: from is.rice.edu (is.rice.edu [128.42.42.24]) by svr1.postgresql.org (Postfix) with ESMTP id 3D4D552856 for ; Wed, 13 Jul 2005 09:50:43 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by is.rice.edu (Postfix) with ESMTP id 205B741EDB; Wed, 13 Jul 2005 07:50:47 -0500 (CDT) Received: from is.rice.edu ([127.0.0.1]) by localhost (it.is.rice.edu [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 27809-01-81; Wed, 13 Jul 2005 07:50:39 -0500 (CDT) Received: by is.rice.edu (Postfix, from userid 18612) id BBB4B41EEB; Wed, 13 Jul 2005 07:33:05 -0500 (CDT) Date: Wed, 13 Jul 2005 07:33:05 -0500 From: Kenneth Marshall To: Nicolas Beaume Cc: pgsql-performance@postgresql.org Subject: Re: large table vs multiple smal tables Message-ID: <20050713123305.GB17225@it.is.rice.edu> References: <42D4E836.9020607@univ-nantes.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42D4E836.9020607@univ-nantes.fr> User-Agent: Mutt/1.4.2i X-Virus-Scanned: by amavis-2.2.1 at is.rice.edu X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/206 X-Sequence-Number: 13447 Nicolas, These sizes would not be considered large. I would leave them as single tables. Ken On Wed, Jul 13, 2005 at 12:08:54PM +0200, Nicolas Beaume wrote: > Hello > > I have a large database with 4 large tables (each containing at least > 200 000 rows, perhaps even 1 or 2 million) and i ask myself if it's > better to split them into small tables (e.g tables of 2000 rows) to > speed the access and the update of those tables (considering that i will > have few update but a lot of reading). > > Do you think it would be efficient ? > > Nicolas, wondering if he hadn't be too greedy > > -- > > ------------------------------------------------------------------------- > ? soyez ce que vous voudriez avoir l'air d'?tre ? Lewis Caroll > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Wed Jul 13 10:45:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CCC3A52A2C for ; Wed, 13 Jul 2005 10:45:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 33713-06 for ; Wed, 13 Jul 2005 13:44:58 +0000 (GMT) Received: from smtp2.univ-nantes.fr (Smtp2.univ-nantes.fr [193.52.82.19]) by svr1.postgresql.org (Postfix) with ESMTP id C7DB85298C for ; Wed, 13 Jul 2005 10:44:57 -0300 (ADT) Received: from localhost (debian [127.0.0.1]) by smtp2.univ-nantes.fr (Postfix) with ESMTP id C3D0C380F32 for ; Wed, 13 Jul 2005 15:44:58 +0200 (CEST) Received: from smtp2.univ-nantes.fr ([172.20.12.56]) by localhost (smtp2.univ-nantes.prive [172.20.12.56]) (amavisd-new, port 10024) with LMTP id 16863-01-97 for ; Wed, 13 Jul 2005 15:44:58 +0200 (CEST) Received: from [172.19.33.37] (unknown [172.19.33.37]) by smtp2.univ-nantes.fr (Postfix) with ESMTP id B362A380E85 for ; Wed, 13 Jul 2005 15:44:58 +0200 (CEST) Message-ID: <42D51D68.8000405@univ-nantes.fr> Date: Wed, 13 Jul 2005 15:55:52 +0200 From: Nicolas Beaume User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: fr, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: (pas de sujet) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new-20030616-p7 (Debian) at smtp.univ-nantes.fr X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/139 X-Sequence-Number: 13380 >Nicolas, > >These sizes would not be considered large. I would leave them >as single tables. > >Ken ok, i though it was large but i must confess i'm relatively new in the database word. thank you for the answer. Just another question : what is the maximal number of rows that can be contain in a cursor ? Nicolas, having a lot of things to learn -- ------------------------------------------------------------------------- � soyez ce que vous voudriez avoir l'air d'�tre � Lewis Caroll From pgsql-performance-owner@postgresql.org Wed Jul 13 12:56:12 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 953F152B80 for ; Wed, 13 Jul 2005 12:56:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 78975-03 for ; Wed, 13 Jul 2005 15:56:01 +0000 (GMT) Received: from cmailg3.svr.pol.co.uk (cmailg3.svr.pol.co.uk [195.92.195.173]) by svr1.postgresql.org (Postfix) with ESMTP id 1055052B16 for ; Wed, 13 Jul 2005 12:56:00 -0300 (ADT) Received: from modem-2867.leopard.dialup.pol.co.uk ([217.135.155.51] helo=192.168.0.102) by cmailg3.svr.pol.co.uk with esmtp (Exim 4.41) id 1DsjaX-000574-9Z; Wed, 13 Jul 2005 16:55:57 +0100 Subject: Re: cost-based vacuum From: Simon Riggs To: Ian Westmacott , Tom Lane Cc: pgsql-performance@postgresql.org In-Reply-To: <1121190651.10346.342.camel@spectre.intellivid.com> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> <1121154338.3970.106.camel@localhost.localdomain> <1121190651.10346.342.camel@spectre.intellivid.com> Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 13 Jul 2005 16:55:50 +0100 Message-Id: <1121270150.3970.256.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/140 X-Sequence-Number: 13381 On Tue, 2005-07-12 at 13:50 -0400, Ian Westmacott wrote: > It appears not to matter whether it is one of the tables > being written to that is ANALYZEd. I can ANALYZE an old, > quiescent table, or a system table and see this effect. Can you confirm that this effect is still seen even when the ANALYZE doesn't touch *any* of the tables being accessed? > - this is a dual Xeon. Is that Xeon MP then? > - Looking at oprofile reports for 10-minute runs of a > database-wide VACUUM with vacuum_cost_delay=0 and 1000, > shows the latter spending a lot of time in LWLockAcquire > and LWLockRelease (20% each vs. 2%). Is this associated with high context switching also? Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Wed Jul 13 13:08:13 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 35A1352ABE for ; Wed, 13 Jul 2005 13:08:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 83738-01 for ; Wed, 13 Jul 2005 16:07:56 +0000 (GMT) Received: from cmailm2.svr.pol.co.uk (cmailm2.svr.pol.co.uk [195.92.193.210]) by svr1.postgresql.org (Postfix) with ESMTP id 5ED1952A58 for ; Wed, 13 Jul 2005 13:07:54 -0300 (ADT) Received: from modem-2867.leopard.dialup.pol.co.uk ([217.135.155.51] helo=192.168.0.102) by cmailm2.svr.pol.co.uk with esmtp (Exim 4.41) id 1Dsjm6-0001VV-32; Wed, 13 Jul 2005 17:07:54 +0100 Subject: Re: size of cache From: Simon Riggs To: Jean-Max Reymond Cc: pgsql-performance@postgresql.org In-Reply-To: <4b09a0c05071301207d452ace@mail.gmail.com> References: <4b09a0c05071301207d452ace@mail.gmail.com> Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 13 Jul 2005 17:07:47 +0100 Message-Id: <1121270867.3970.266.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/141 X-Sequence-Number: 13382 On Wed, 2005-07-13 at 10:20 +0200, Jean-Max Reymond wrote: > with my application, it seems that size of cache has great effect: > from 512 Kb of L2 cache to 1Mb boost performance with a factor 3 and > 20% again from 1Mb L2 cache to 2Mb L2 cache. Memory request time is the main bottleneck in well tuned database systems, so your results could be reasonable. > I don't understand why a 512Kb cache L2 is too small to fit the data's > does it exist a tool to trace processor activity and confirm that > processor is waiting for memory ? You have both data and instruction cache on the CPU. It is likely it is the instruction cache that is too small to fit all of the code required for your application's workload mix. Use Intel VTune or similar to show the results you seek. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Wed Jul 13 13:11:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 01BDD52A08 for ; Wed, 13 Jul 2005 13:10:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 83642-03 for ; Wed, 13 Jul 2005 16:10:21 +0000 (GMT) Received: from cmailg1.svr.pol.co.uk (cmailg1.svr.pol.co.uk [195.92.195.171]) by svr1.postgresql.org (Postfix) with ESMTP id 8F7D452BCF for ; Wed, 13 Jul 2005 13:10:18 -0300 (ADT) Received: from modem-2867.leopard.dialup.pol.co.uk ([217.135.155.51] helo=192.168.0.102) by cmailg1.svr.pol.co.uk with esmtp (Exim 4.41) id 1DsjoL-0007Kq-2r; Wed, 13 Jul 2005 17:10:13 +0100 Subject: Re: General DB Tuning From: Simon Riggs To: Christopher Kings-Lynne Cc: Brent Henry , Dennis , Tom Arthurs , pgsql-performance@postgresql.org In-Reply-To: <42D473D4.7040500@familyhealth.com.au> References: <20050713013630.18625.qmail@web33911.mail.mud.yahoo.com> <42D473D4.7040500@familyhealth.com.au> Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 13 Jul 2005 17:10:06 +0100 Message-Id: <1121271006.3970.268.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/142 X-Sequence-Number: 13383 On Wed, 2005-07-13 at 09:52 +0800, Christopher Kings-Lynne wrote: > > Is there a different kind of 'prepared' statements > > that we should be using in the driver to get logging > > to work properly? What is the 'new' protocol? > > The 8.0.2 jdbc driver uses real prepared statements instead of faked > ones. The problem is the new protocol (that the 8.0.2 driver users) has > a bug where protocol-prepared queries don't get logged properly. > > I don't know if it's been fixed... Yes, there is a fix for this in 8.1 Brent has been sent the details. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Wed Jul 13 15:40:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9CAE752B9E for ; Wed, 13 Jul 2005 15:40:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31268-08 for ; Wed, 13 Jul 2005 18:40:37 +0000 (GMT) Received: from mailhost.intellivid.com (mailhost.intellivid.com [64.32.200.11]) by svr1.postgresql.org (Postfix) with ESMTP id 36DCE52B4E for ; Wed, 13 Jul 2005 15:40:36 -0300 (ADT) Received: from [192.168.2.68] (spectre.intellivid.com [192.168.2.68]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by newmail.intellivid.com (Postfix) with ESMTP id 8595FF1829F; Wed, 13 Jul 2005 14:40:36 -0400 (EDT) Subject: Re: cost-based vacuum From: Ian Westmacott To: Simon Riggs Cc: Tom Lane , pgsql-performance@postgresql.org In-Reply-To: <1121270150.3970.256.camel@localhost.localdomain> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> <1121154338.3970.106.camel@localhost.localdomain> <1121190651.10346.342.camel@spectre.intellivid.com> <1121270150.3970.256.camel@localhost.localdomain> Content-Type: text/plain Organization: Intellivid Corp. Message-Id: <1121280036.13208.8.camel@spectre.intellivid.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Wed, 13 Jul 2005 14:40:36 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/143 X-Sequence-Number: 13384 On Wed, 2005-07-13 at 11:55, Simon Riggs wrote: > On Tue, 2005-07-12 at 13:50 -0400, Ian Westmacott wrote: > > It appears not to matter whether it is one of the tables > > being written to that is ANALYZEd. I can ANALYZE an old, > > quiescent table, or a system table and see this effect. > > Can you confirm that this effect is still seen even when the ANALYZE > doesn't touch *any* of the tables being accessed? Yes. > > - this is a dual Xeon. > > Is that Xeon MP then? Yes. > > - Looking at oprofile reports for 10-minute runs of a > > database-wide VACUUM with vacuum_cost_delay=0 and 1000, > > shows the latter spending a lot of time in LWLockAcquire > > and LWLockRelease (20% each vs. 2%). > > Is this associated with high context switching also? Yes, it appears that context switches increase up to 4-5x during cost-based ANALYZE. --Ian From pgsql-performance-owner@postgresql.org Wed Jul 13 15:54:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2F3AD52A1A for ; Wed, 13 Jul 2005 15:54:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36640-02 for ; Wed, 13 Jul 2005 18:54:40 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id BF44E529CF for ; Wed, 13 Jul 2005 15:54:38 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id B09F46441D3 for ; Wed, 13 Jul 2005 12:54:15 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28395-03 for ; Wed, 13 Jul 2005 12:54:13 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id BA38A6441CB for ; Wed, 13 Jul 2005 12:54:13 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) Content-Transfer-Encoding: 7bit Message-Id: Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: Dan Harris Subject: Quad Opteron stuck in the mud Date: Wed, 13 Jul 2005 12:54:35 -0600 X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/144 X-Sequence-Number: 13385 Gurus, A table in one of my databases has just crossed the 30 million row mark and has begun to feel very sluggish for just about anything I do with it. I keep the entire database vacuumed regularly. And, as long as I'm not doing a sequential scan, things seem reasonably quick most of the time. I'm now thinking that my problem is IO because anything that involves heavy ( like a seq scan ) IO seems to slow to a crawl. Even if I am using indexed fields to grab a few thousand rows, then going to sequential scans it gets very very slow. I have also had the occurrence where queries will not finish for days ( I eventually have to kill them ). I was hoping to provide an explain analyze for them, but if they never finish... even the explain never finishes when I try that. For example, as I'm writing this, I am running an UPDATE statement that will affect a small part of the table, and is querying on an indexed boolean field. I have been waiting for over an hour and a half as I write this and it still hasn't finished. I'm thinking "I bet Tom, Simon or Josh wouldn't put up with this kind of wait time..", so I thought I would see if anyone here had some pointers. Maybe I have a really stupid setting in my conf file that is causing this. I really can't believe I am at the limits of this hardware, however. The query: update eventactivity set ftindex = false where ftindex = true; ( added the where clause because I don't want to alter where ftindex is null ) The table: Column | Type | Modifiers -------------+-----------------------------+----------- entrydate | timestamp without time zone | incidentid | character varying(40) | statustype | character varying(20) | unitid | character varying(20) | recordtext | character varying(255) | recordtext2 | character varying(255) | insertdate | timestamp without time zone | ftindex | boolean | Indexes: eventactivity1 btree (incidentid), eventactivity_entrydate_idx btree (entrydate), eventactivity_ftindex_idx btree (ftindex), eventactivity_oid_idx btree (oid) The hardware: 4 x 2.2GHz Opterons 12 GB of RAM 4x10k 73GB Ultra320 SCSI drives in RAID 0+1 1GB hardware cache memory on the RAID controller The OS: Fedora, kernel 2.6.6-1.435.2.3smp ( redhat stock kernel ) filesystem is mounted as ext2 ##### vmstat output ( as I am waiting for this to finish ): procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 5436 2823908 26140 9183704 0 1 2211 540 694 336 9 2 76 13 ##### iostat output ( as I am waiting for this to finish ): avg-cpu: %user %nice %sys %iowait %idle 9.19 0.00 2.19 13.08 75.53 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn cciss/c0d0 329.26 17686.03 4317.57 161788630 39496378 ##### This is a dedicated postgresql server, so maybe some of these settings are more liberal than they should be? relevant ( I hope ) postgresql.conf options are: shared_buffers = 50000 effective_cache_size = 1348000 random_page_cost = 3 work_mem = 512000 max_fsm_pages = 80000 log_min_duration_statement = 60000 fsync = true ( not sure if I'm daring enough to run without this ) wal_buffers = 1000 checkpoint_segments = 64 checkpoint_timeout = 3000 #---- FOR PG_AUTOVACUUM --# stats_command_string = true stats_row_level = true Thanks in advance, Dan From pgsql-performance-owner@postgresql.org Wed Jul 13 15:59:01 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A71CD52BC7 for ; Wed, 13 Jul 2005 15:58:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36499-05 for ; Wed, 13 Jul 2005 18:58:57 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E212E52B8E for ; Wed, 13 Jul 2005 15:58:56 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6DIwqHE020660; Wed, 13 Jul 2005 14:58:52 -0400 (EDT) To: Ian Westmacott Cc: Simon Riggs , pgsql-performance@postgresql.org Subject: Re: cost-based vacuum In-reply-to: <1121280036.13208.8.camel@spectre.intellivid.com> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> <1121154338.3970.106.camel@localhost.localdomain> <1121190651.10346.342.camel@spectre.intellivid.com> <1121270150.3970.256.camel@localhost.localdomain> <1121280036.13208.8.camel@spectre.intellivid.com> Comments: In-reply-to Ian Westmacott message dated "Wed, 13 Jul 2005 14:40:36 -0400" Date: Wed, 13 Jul 2005 14:58:52 -0400 Message-ID: <20659.1121281132@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/145 X-Sequence-Number: 13386 Ian Westmacott writes: > On Wed, 2005-07-13 at 11:55, Simon Riggs wrote: >> On Tue, 2005-07-12 at 13:50 -0400, Ian Westmacott wrote: >>> It appears not to matter whether it is one of the tables >>> being written to that is ANALYZEd. I can ANALYZE an old, >>> quiescent table, or a system table and see this effect. >> >> Can you confirm that this effect is still seen even when the ANALYZE >> doesn't touch *any* of the tables being accessed? > Yes. This really isn't making any sense at all. I took another look through the vacuum_delay_point() calls, and I can see a couple that are questionably placed: * the one in count_nondeletable_pages() is done while we are holding exclusive lock on the table; we might be better off not to delay there, so as not to block non-VACUUM processes longer than we have to. * the ones in hashbulkdelete and rtbulkdelete are done while holding various forms of exclusive locks on the index (this was formerly true of gistbulkdelete as well). Again it might be better not to delay. However, these certainly do not explain Ian's problem, because (a) these only apply to VACUUM, not ANALYZE; (b) they would only lock the table being VACUUMed, not other ones; (c) if these locks were to block the reader or writer thread, it'd manifest as blocking on a semaphore, not as a surge in LWLock thrashing. >> Is that Xeon MP then? > Yes. The LWLock activity is certainly suggestive of prior reports of excessive buffer manager lock contention, but it makes *no* sense that that would be higher with vacuum cost delay than without. I'd have expected the other way around. I'd really like to see a test case for this... regards, tom lane From pgsql-performance-owner@postgresql.org Wed Jul 13 16:09:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7E9ED52A1A for ; Wed, 13 Jul 2005 16:09:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40131-02 for ; Wed, 13 Jul 2005 19:09:40 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id C0778529CF for ; Wed, 13 Jul 2005 16:09:39 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id B833C6441FC for ; Wed, 13 Jul 2005 13:09:17 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28395-10 for ; Wed, 13 Jul 2005 13:09:15 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id C0D4D6441DD for ; Wed, 13 Jul 2005 13:09:15 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: References: Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <67AA7707-6473-4AF2-89EF-2C90A6AAA2AB@drivefaster.net> Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Quad Opteron stuck in the mud Date: Wed, 13 Jul 2005 13:09:37 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/146 X-Sequence-Number: 13387 So sorry, I forgot to mention I'm running version 8.0.1 Thanks From pgsql-performance-owner@postgresql.org Wed Jul 13 16:11:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CC72E52B9C for ; Wed, 13 Jul 2005 16:11:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36077-06 for ; Wed, 13 Jul 2005 19:11:46 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 6A91F52A2C for ; Wed, 13 Jul 2005 16:11:45 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050713191145m9100cf4v0e>; Wed, 13 Jul 2005 19:11:46 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id ADF9055FEC; Wed, 13 Jul 2005 14:11:45 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id CA49655F98; Wed, 13 Jul 2005 14:11:41 -0500 (CDT) Message-ID: <42D56772.6020408@arbash-meinel.com> Date: Wed, 13 Jul 2005 14:11:46 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: Quad Opteron stuck in the mud References: In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigD190497174A8400B62E3C137" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.047 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/147 X-Sequence-Number: 13388 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigD190497174A8400B62E3C137 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Dan Harris wrote: > Gurus, > > even the explain never > finishes when I try that. Just a short bit. If "EXPLAIN SELECT" doesn't return, there seems to be a very serious problem. Because I think EXPLAIN doesn't actually run the query, just has the query planner run. And the query planner shouldn't ever get heavily stuck. I might be wrong, but there may be something much more substantially wrong than slow i/o. John =:-> --------------enigD190497174A8400B62E3C137 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1WdyJdeBCYSNAAMRAo7GAJ4oLZc+JTpiwDJDJ5nzwYRbEAoocwCcDE+o V8XImXDydGv7uXMX+Fhc31I= =1gn4 -----END PGP SIGNATURE----- --------------enigD190497174A8400B62E3C137-- From pgsql-performance-owner@postgresql.org Wed Jul 13 16:15:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E849852A1A for ; Wed, 13 Jul 2005 16:13:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40438-02 for ; Wed, 13 Jul 2005 19:13:31 +0000 (GMT) Received: from lakermmtao06.cox.net (lakermmtao06.cox.net [68.230.240.33]) by svr1.postgresql.org (Postfix) with ESMTP id 2E2F052C79 for ; Wed, 13 Jul 2005 16:13:30 -0300 (ADT) Received: from [127.0.0.1] (really [68.3.22.74]) by lakermmtao06.cox.net (InterMail vM.6.01.04.00 201-2131-118-20041027) with ESMTP id <20050713191328.NCB2772.lakermmtao06.cox.net@[127.0.0.1]> for ; Wed, 13 Jul 2005 15:13:28 -0400 Message-ID: <42D567DB.6000100@works4me.com> Date: Wed, 13 Jul 2005 12:13:31 -0700 From: Dennis User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: performance problems ... 100 cpu utilization Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Antivirus: avast! (VPS 0528-3, 07/13/2005), Outbound message X-Antivirus-Status: Clean X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/148 X-Sequence-Number: 13389 Hi, I've got a java based web application that uses PostgreSQL 8.0.2. PostgreSQL runs on its own machine with RHEL 3, ia32e kernel, dual Xeon processor, 4 Gb ram. The web application runs on a seperate machine from the database. The application machine has three tomcat instances configured to use 64 database connections each using DBCP for pooling. Most of the data access is via Hibernate. The database itself is about 100 meg in size. We're perf testing the application with Loadrunner. At about 500 virtual users hitting the web application, the cpu utilization on the database server is at 100%, PostgreSQL is on its knees. The memory usage isn't bad, the I/O isn't bad, only the CPU seems to be maxed out. checking the status of connections at this point ( ps -eaf | grep "postgres:") where the CPU is maxed out I saw this: 127 idle 12 bind 38 parse 34 select Hibernate is used in the application and unfortunately this seems to cause queries not to get logged. (see http://archives.postgresql.org/pgsql-admin/2005-05/msg00241.php) I know there has been discussion about problems on Xeon MP systems. Is this what we are running into? Or is something else going on? Is there other information I can provide that might help determine what is going on? Here are the postgresql.conf settings: # The maximum number of connections. max_connections = 256 # Standard performance-related settings. shared_buffers = 16384 max_fsm_pages = 200000 max_fsm_relations = 10000 fsync = false wal_sync_method = fsync wal_buffers = 32 checkpoint_segments = 6 effective_cache_size = 38400 random_page_cost = 2 work_mem = 16384 maintenance_work_mem = 16384 # TODO - need to investigate these. commit_delay = 0 commit_siblings = 5 max_locks_per_transaction = 512 From pgsql-performance-owner@postgresql.org Wed Jul 13 16:16:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 67F2752BD6 for ; Wed, 13 Jul 2005 16:16:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40357-04 for ; Wed, 13 Jul 2005 19:16:28 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 14C4052B9C for ; Wed, 13 Jul 2005 16:16:28 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 572CD644146 for ; Wed, 13 Jul 2005 13:16:06 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28615-04 for ; Wed, 13 Jul 2005 13:16:03 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 187BF64411B for ; Wed, 13 Jul 2005 13:16:03 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <42D56772.6020408@arbash-meinel.com> References: <42D56772.6020408@arbash-meinel.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <862C6AD9-646B-4907-AB18-09F1AD08B5F5@drivefaster.net> Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Quad Opteron stuck in the mud Date: Wed, 13 Jul 2005 13:16:25 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/149 X-Sequence-Number: 13390 On Jul 13, 2005, at 1:11 PM, John A Meinel wrote: > > I might be wrong, but there may be something much more substantially > wrong than slow i/o. > John > Yes, I'm afraid of that too. I just don't know what tools I should use to figure that out. I have some 20 other databases on this system, same schema but varying sizes, and the small ones perform very well. It feels like there is an O(n) increase in wait time that has recently become very noticeable on the largest of them. -Dan From pgsql-performance-owner@postgresql.org Wed Jul 13 17:17:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5860952BDF for ; Wed, 13 Jul 2005 17:17:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54397-05 for ; Wed, 13 Jul 2005 20:17:04 +0000 (GMT) Received: from ns.snowman.net (ns.snowman.net [66.92.160.21]) by svr1.postgresql.org (Postfix) with ESMTP id 8AE8452BE5 for ; Wed, 13 Jul 2005 17:17:03 -0300 (ADT) Received: by ns.snowman.net (Postfix, from userid 1000) id 36C0F17AF4; Wed, 13 Jul 2005 16:17:15 -0400 (EDT) Date: Wed, 13 Jul 2005 16:17:15 -0400 From: Stephen Frost To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: Quad Opteron stuck in the mud Message-ID: <20050713201715.GI24207@ns.snowman.net> Mail-Followup-To: Dan Harris , pgsql-performance@postgresql.org References: <42D56772.6020408@arbash-meinel.com> <862C6AD9-646B-4907-AB18-09F1AD08B5F5@drivefaster.net> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="+fvCYAx0wXfEGTTZ" Content-Disposition: inline In-Reply-To: <862C6AD9-646B-4907-AB18-09F1AD08B5F5@drivefaster.net> X-Editor: Vim http://www.vim.org/ X-Info: http://www.snowman.net X-Operating-System: Linux/2.4.24ns.3.0 (i686) X-Uptime: 16:16:30 up 32 days, 12:36, 8 users, load average: 0.12, 0.08, 0.07 User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.003 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/150 X-Sequence-Number: 13391 --+fvCYAx0wXfEGTTZ Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable * Dan Harris (fbsd@drivefaster.net) wrote: > On Jul 13, 2005, at 1:11 PM, John A Meinel wrote: > >I might be wrong, but there may be something much more substantially > >wrong than slow i/o. >=20 > Yes, I'm afraid of that too. I just don't know what tools I should =20 > use to figure that out. I have some 20 other databases on this =20 > system, same schema but varying sizes, and the small ones perform =20 > very well. It feels like there is an O(n) increase in wait time that =20 > has recently become very noticeable on the largest of them. Could you come up w/ a test case that others could reproduce where explain isn't returning? I think that would be very useful towards solving at least that issue... Thanks, Stephen --+fvCYAx0wXfEGTTZ Content-Type: application/pgp-signature; name="signature.asc" Content-Description: Digital signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQFC1XbKrzgMPqB3kigRAtjJAJ4lRzAyya3SuMnzv6yczJPFvlqdjwCcDLMK ruMkTUhhdJNCw0hWvi6PGLE= =/e4k -----END PGP SIGNATURE----- --+fvCYAx0wXfEGTTZ-- From pgsql-performance-owner@postgresql.org Wed Jul 13 17:18:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7243552BD4 for ; Wed, 13 Jul 2005 17:18:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54607-01 for ; Wed, 13 Jul 2005 20:18:18 +0000 (GMT) Received: from mr2.surnet.cl (smtp2.surnet.cl [216.155.73.163]) by svr1.postgresql.org (Postfix) with ESMTP id 545F952B5B for ; Wed, 13 Jul 2005 17:18:16 -0300 (ADT) Received: from smtp2.surnet.cl (216.155.73.169) by mr2.surnet.cl (7.0.031.3) id 4259ADFA015441EB; Wed, 13 Jul 2005 16:17:57 -0400 Received: from smtp2.surnet.cl (mr2.surnet.cl []) by mr2.surnet.cl ([216.155.73.169]); Wed, 13 Jul 2005 20:17:57 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp2.surnet.cl (7.0.031.3) id 4259ADE300CFADA6; Wed, 13 Jul 2005 16:17:57 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF6000399CCB; Wed, 13 Jul 2005 16:18:13 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 925F2C370EE; Wed, 13 Jul 2005 16:18:17 -0400 (CLT) Date: Wed, 13 Jul 2005 16:18:17 -0400 From: Alvaro Herrera To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: Quad Opteron stuck in the mud Message-ID: <20050713201817.GA5450@alvh.no-ip.org> References: <42D56772.6020408@arbash-meinel.com> <862C6AD9-646B-4907-AB18-09F1AD08B5F5@drivefaster.net> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <862C6AD9-646B-4907-AB18-09F1AD08B5F5@drivefaster.net> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.378 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/151 X-Sequence-Number: 13392 On Wed, Jul 13, 2005 at 01:16:25PM -0600, Dan Harris wrote: > On Jul 13, 2005, at 1:11 PM, John A Meinel wrote: > > >I might be wrong, but there may be something much more substantially > >wrong than slow i/o. > > Yes, I'm afraid of that too. I just don't know what tools I should > use to figure that out. I have some 20 other databases on this > system, same schema but varying sizes, and the small ones perform > very well. It feels like there is an O(n) increase in wait time that > has recently become very noticeable on the largest of them. I'd guess it's stuck on some lock. Try that EXPLAIN, and when it blocks, watch the pg_locks view for locks not granted to the process executing the EXPLAIN. Then check what else is holding the locks. -- Alvaro Herrera () "La rebeld�a es la virtud original del hombre" (Arthur Schopenhauer) From pgsql-performance-owner@postgresql.org Wed Jul 13 17:20:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 05EB752A78 for ; Wed, 13 Jul 2005 17:20:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 55524-01 for ; Wed, 13 Jul 2005 20:20:26 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 1D28552C9C for ; Wed, 13 Jul 2005 17:20:25 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 21AB36441AA for ; Wed, 13 Jul 2005 14:20:04 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 29029-10 for ; Wed, 13 Jul 2005 14:20:02 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 22206644195 for ; Wed, 13 Jul 2005 14:20:02 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <20050713201715.GI24207@ns.snowman.net> References: <42D56772.6020408@arbash-meinel.com> <862C6AD9-646B-4907-AB18-09F1AD08B5F5@drivefaster.net> <20050713201715.GI24207@ns.snowman.net> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <7C8917ED-62B8-4D76-AC02-90CF04002C97@drivefaster.net> Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Quad Opteron stuck in the mud Date: Wed, 13 Jul 2005 14:20:24 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/152 X-Sequence-Number: 13393 On Jul 13, 2005, at 2:17 PM, Stephen Frost wrote: > > Could you come up w/ a test case that others could reproduce where > explain isn't returning? This was simply due to my n00bness :) I had always been doing explain analyze, instead of just explain. Next time one of these queries comes up, I will be sure to do the explain without analyze. FYI that update query I mentioned in the initial thread just finished after updating 8.3 million rows. -Dan From pgsql-performance-owner@postgresql.org Wed Jul 13 17:40:06 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id ACEA752C9C for ; Wed, 13 Jul 2005 17:40:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58063-09 for ; Wed, 13 Jul 2005 20:39:58 +0000 (GMT) Received: from cmailm1.svr.pol.co.uk (cmailm1.svr.pol.co.uk [195.92.193.18]) by svr1.postgresql.org (Postfix) with ESMTP id B58FC52B23 for ; Wed, 13 Jul 2005 17:39:56 -0300 (ADT) Received: from modem-875.llama.dialup.pol.co.uk ([217.135.179.107] helo=192.168.0.102) by cmailm1.svr.pol.co.uk with esmtp (Exim 4.41) id 1Dso1M-0000RA-5I; Wed, 13 Jul 2005 21:39:56 +0100 Subject: Re: cost-based vacuum From: Simon Riggs To: Tom Lane Cc: Ian Westmacott , pgsql-performance@postgresql.org In-Reply-To: <20659.1121281132@sss.pgh.pa.us> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> <1121154338.3970.106.camel@localhost.localdomain> <1121190651.10346.342.camel@spectre.intellivid.com> <1121270150.3970.256.camel@localhost.localdomain> <1121280036.13208.8.camel@spectre.intellivid.com> <20659.1121281132@sss.pgh.pa.us> Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 13 Jul 2005 21:39:48 +0100 Message-Id: <1121287188.3970.296.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/153 X-Sequence-Number: 13394 On Wed, 2005-07-13 at 14:58 -0400, Tom Lane wrote: > Ian Westmacott writes: > > On Wed, 2005-07-13 at 11:55, Simon Riggs wrote: > >> On Tue, 2005-07-12 at 13:50 -0400, Ian Westmacott wrote: > >>> It appears not to matter whether it is one of the tables > >>> being written to that is ANALYZEd. I can ANALYZE an old, > >>> quiescent table, or a system table and see this effect. > >> > >> Can you confirm that this effect is still seen even when the ANALYZE > >> doesn't touch *any* of the tables being accessed? > > > Yes. > > This really isn't making any sense at all. Agreed. I think all of this indicates that some wierdness (technical term) is happening at a different level in the computing stack. I think all of this points fairly strongly to it *not* being a PostgreSQL algorithm problem, i.e. if the code was executed by an idealised Knuth- like CPU then we would not get this problem. Plus, I have faith that if it was a problem in that "plane" then you or another would have uncovered it by now. > However, these certainly do not explain Ian's problem, because (a) these > only apply to VACUUM, not ANALYZE; (b) they would only lock the table > being VACUUMed, not other ones; (c) if these locks were to block the > reader or writer thread, it'd manifest as blocking on a semaphore, not > as a surge in LWLock thrashing. I've seen enough circumstantial evidence to connect the time spent inside LWLockAcquire/Release as being connected to the Semaphore ops within them, not the other aspects of the code. Months ago we discussed the problem of false sharing on closely packed arrays of shared variables because of the large cache line size of the Xeon MP. When last we touched on that thought, I focused on the thought that the LWLock array was too tightly packed for the predefined locks. What we didn't discuss (because I was too focused on the other array) was the PGPROC shared array is equally tightly packed, which could give problems on the semaphores in LWLock. Intel says fairly clearly that this would be an issue. > >> Is that Xeon MP then? > > > Yes. > > The LWLock activity is certainly suggestive of prior reports of > excessive buffer manager lock contention, but it makes *no* sense that > that would be higher with vacuum cost delay than without. I'd have > expected the other way around. > > I'd really like to see a test case for this... My feeling is that a "micro-architecture" test would be more likely to reveal some interesting information. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Wed Jul 13 17:49:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D854A52CAB for ; Wed, 13 Jul 2005 17:49:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58751-09 for ; Wed, 13 Jul 2005 20:49:31 +0000 (GMT) Received: from yertle.kcilink.com (yertle.kcilink.com [65.205.34.180]) by svr1.postgresql.org (Postfix) with ESMTP id A231E52B1A for ; Wed, 13 Jul 2005 17:49:30 -0300 (ADT) Received: from [192.168.7.103] (host-103.int.kcilink.com [192.168.7.103]) by yertle.kcilink.com (Postfix) with ESMTP id 5353FB80C for ; Wed, 13 Jul 2005 16:49:32 -0400 (EDT) Mime-Version: 1.0 (Apple Message framework v733) In-Reply-To: References: Content-Type: multipart/signed; micalg=sha1; boundary=Apple-Mail-7--900832296; protocol="application/pkcs7-signature" Message-Id: From: Vivek Khera Subject: Re: Quad Opteron stuck in the mud Date: Wed, 13 Jul 2005 16:49:31 -0400 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.733) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.023 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/154 X-Sequence-Number: 13395 --Apple-Mail-7--900832296 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed On Jul 13, 2005, at 2:54 PM, Dan Harris wrote: > 4 x 2.2GHz Opterons > 12 GB of RAM > 4x10k 73GB Ultra320 SCSI drives in RAID 0+1 > 1GB hardware cache memory on the RAID controller > if it is taking that long to update about 25% of your table, then you must be I/O bound. check I/o while you're running a big query. also, what RAID controller are you running? be sure you have the latest BIOS and drivers for it. on a pair of dual opterons, I can do large operations on tables with 100 million rows much faster than you seem to be able. I have MegaRAID 320-2x controllers with 15kRPM drives. Vivek Khera, Ph.D. +1-301-869-4449 x806 --Apple-Mail-7--900832296 Content-Transfer-Encoding: base64 Content-Type: application/pkcs7-signature; name=smime.p7s Content-Disposition: attachment; filename=smime.p7s MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIGhzCCAz8w ggKooAMCAQICAQ0wDQYJKoZIhvcNAQEFBQAwgdExCzAJBgNVBAYTAlpBMRUwEwYDVQQIEwxXZXN0 ZXJuIENhcGUxEjAQBgNVBAcTCUNhcGUgVG93bjEaMBgGA1UEChMRVGhhd3RlIENvbnN1bHRpbmcx KDAmBgNVBAsTH0NlcnRpZmljYXRpb24gU2VydmljZXMgRGl2aXNpb24xJDAiBgNVBAMTG1RoYXd0 ZSBQZXJzb25hbCBGcmVlbWFpbCBDQTErMCkGCSqGSIb3DQEJARYccGVyc29uYWwtZnJlZW1haWxA dGhhd3RlLmNvbTAeFw0wMzA3MTcwMDAwMDBaFw0xMzA3MTYyMzU5NTlaMGIxCzAJBgNVBAYTAlpB MSUwIwYDVQQKExxUaGF3dGUgQ29uc3VsdGluZyAoUHR5KSBMdGQuMSwwKgYDVQQDEyNUaGF3dGUg UGVyc29uYWwgRnJlZW1haWwgSXNzdWluZyBDQTCBnzANBgkqhkiG9w0BAQEFAAOBjQAwgYkCgYEA xKY8VXNV+065yplaHmjAdQRwnd/p/6Me7L3N9VvyGna9fww6YfK/Uc4B1OVQCjDXAmNaLIkVcI7d yfArhVqqP3FWy688Cwfn8R+RNiQqE88r1fOCdz0Dviv+uxg+B79AgAJk16emu59l0cUqVIUPSAR/ p7bRPGEEQB5kGXJgt/sCAwEAAaOBlDCBkTASBgNVHRMBAf8ECDAGAQH/AgEAMEMGA1UdHwQ8MDow OKA2oDSGMmh0dHA6Ly9jcmwudGhhd3RlLmNvbS9UaGF3dGVQZXJzb25hbEZyZWVtYWlsQ0EuY3Js MAsGA1UdDwQEAwIBBjApBgNVHREEIjAgpB4wHDEaMBgGA1UEAxMRUHJpdmF0ZUxhYmVsMi0xMzgw DQYJKoZIhvcNAQEFBQADgYEASIzRUIPqCy7MDaNmrGcPf6+svsIXoUOWlJ1/TCG4+DYfqi2fNi/A 9BxQIJNwPP2t4WFiw9k6GX6EsZkbAMUaC4J0niVQlGLH2ydxVyWN3amcOY6MIE9lX5Xa9/eH1sYI Tq726jTlEBpbNU1341YheILcIRk13iSx0x1G/11fZU8wggNAMIICqaADAgECAgMOah8wDQYJKoZI hvcNAQEEBQAwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkp IEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFpbCBJc3N1aW5nIENBMB4XDTA1 MDQwNTIwMzEzMloXDTA2MDQwNTIwMzEzMlowgYoxHzAdBgNVBAMTFlRoYXd0ZSBGcmVlbWFpbCBN ZW1iZXIxHjAcBgkqhkiG9w0BCQEWD3ZpdmVrQGtoZXJhLm9yZzEgMB4GCSqGSIb3DQEJARYRa2hl cmFAa2NpbGluay5jb20xJTAjBgkqhkiG9w0BCQEWFnZpdmVrQG1haWxlcm1haWxlci5jb20wggEi MA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQDdo7hywGcY0VvK6WqqXXV77MS/t/4X3WkCaCXo RSl2W58GP4P21hodPn7hlIxUoDOW7x9O+FbqTgE2Ejqr6yA00Mm90tGPFgjFjqPGAqg7xk6IDcv9 uTyMia/FKEHSIynM6zqokXY8JklvdbJOiByE/8VeyEXOANWiflo8o4+GHnhMKpA9982YTXUqeKU6 mMQVaLCBRjTDc7j2XkMC/UNcp2HMyDQdTqYVnhLxbvbLX8CNDBY/7OWFlB9evru46SpGWhe4lhv5 DSgE2RdCKvDytzxRDvP49L8V0TnFjAVeC1C1Pj0/KQsoL/AP4APplROiD4QaUhshQl28pXxJtfbl AgMBAAGjVzBVMEUGA1UdEQQ+MDyBD3ZpdmVrQGtoZXJhLm9yZ4ERa2hlcmFAa2NpbGluay5jb22B FnZpdmVrQG1haWxlcm1haWxlci5jb20wDAYDVR0TAQH/BAIwADANBgkqhkiG9w0BAQQFAAOBgQAr CWop3h28qPwofzLrkoT410J4d7Bqk6FLeVlKZfg/wXlS1MTqYMNcCm4x+JsJbjwsO0fb2elFIuGq 1razoSzPpgi89itydvUT0U0U/u+AkZA5rW4AptTpMZ70YW5u9wzkcvmifqZmcfbaaeGdZfruzUXZ 6qvdXDpNb3ZHeQw6PjGCAucwggLjAgEBMGkwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0 ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFp bCBJc3N1aW5nIENBAgMOah8wCQYFKw4DAhoFAKCCAVMwGAYJKoZIhvcNAQkDMQsGCSqGSIb3DQEH ATAcBgkqhkiG9w0BCQUxDxcNMDUwNzEzMjA0OTMyWjAjBgkqhkiG9w0BCQQxFgQUj+qsC8cfQ5/b y7Gb4h5SRjnjHdkweAYJKwYBBAGCNxAEMWswaTBiMQswCQYDVQQGEwJaQTElMCMGA1UEChMcVGhh d3RlIENvbnN1bHRpbmcgKFB0eSkgTHRkLjEsMCoGA1UEAxMjVGhhd3RlIFBlcnNvbmFsIEZyZWVt YWlsIElzc3VpbmcgQ0ECAw5qHzB6BgsqhkiG9w0BCRACCzFroGkwYjELMAkGA1UEBhMCWkExJTAj BgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJz b25hbCBGcmVlbWFpbCBJc3N1aW5nIENBAgMOah8wDQYJKoZIhvcNAQEBBQAEggEAidjVSGVuMKpN 5/9Xnr/27QlkNGUunR5NgosvXVUjYyWRFAPKUL0cM41gQQpNEvtxpamC4kDOwmH5B1ikabYR7Ego B6VBucu6OCAo8OjevutOBtCeahRtTW6hFeP/yDJyoNdvXEwDRuxkDQGvJ0zTiZZJqihKYEbN56cF 0/28b4j7WiggQM7OwrDRpBVlcKlbrtGijL55WJiUAgLylVdFh7yzkomVViwNvPUM2OofjVgUFMY7 olpGFMVeTvRl/5/neXw6xeN3uOT/i2rzOQRPZf60YKrpYSUvJCHyIb7CFzlbtnynPRhlIFcdtYuD EuehnXgghJNUJgdrB8W71g+8NwAAAAAAAA== --Apple-Mail-7--900832296-- From pgsql-performance-owner@postgresql.org Wed Jul 13 18:46:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C4591529C2 for ; Wed, 13 Jul 2005 18:46:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 72280-02 for ; Wed, 13 Jul 2005 21:46:09 +0000 (GMT) Received: from mailhost.intellivid.com (mailhost.intellivid.com [64.32.200.11]) by svr1.postgresql.org (Postfix) with ESMTP id 7690D52ADE for ; Wed, 13 Jul 2005 18:46:08 -0300 (ADT) Received: from [192.168.2.68] (spectre.intellivid.com [192.168.2.68]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by newmail.intellivid.com (Postfix) with ESMTP id C60D6F1829E; Wed, 13 Jul 2005 17:46:10 -0400 (EDT) Subject: Re: cost-based vacuum From: Ian Westmacott To: Simon Riggs Cc: Tom Lane , pgsql-performance@postgresql.org In-Reply-To: <1121287188.3970.296.camel@localhost.localdomain> References: <1120839902.20657.197.camel@spectre.intellivid.com> <1121081504.3970.53.camel@localhost.localdomain> <1121087266.27427.35.camel@spectre.intellivid.com> <1121093488.3970.70.camel@localhost.localdomain> <1121154338.3970.106.camel@localhost.localdomain> <1121190651.10346.342.camel@spectre.intellivid.com> <1121270150.3970.256.camel@localhost.localdomain> <1121280036.13208.8.camel@spectre.intellivid.com> <20659.1121281132@sss.pgh.pa.us> <1121287188.3970.296.camel@localhost.localdomain> Content-Type: text/plain Organization: Intellivid Corp. Message-Id: <1121291170.13208.25.camel@spectre.intellivid.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Wed, 13 Jul 2005 17:46:10 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/155 X-Sequence-Number: 13396 I can at least report that the problem does not seem to occur with Postgres 8.0.1 running on a dual Opteron. --Ian On Wed, 2005-07-13 at 16:39, Simon Riggs wrote: > On Wed, 2005-07-13 at 14:58 -0400, Tom Lane wrote: > > Ian Westmacott writes: > > > On Wed, 2005-07-13 at 11:55, Simon Riggs wrote: > > >> On Tue, 2005-07-12 at 13:50 -0400, Ian Westmacott wrote: > > >>> It appears not to matter whether it is one of the tables > > >>> being written to that is ANALYZEd. I can ANALYZE an old, > > >>> quiescent table, or a system table and see this effect. > > >> > > >> Can you confirm that this effect is still seen even when the ANALYZE > > >> doesn't touch *any* of the tables being accessed? > > > > > Yes. > > > > This really isn't making any sense at all. > > Agreed. I think all of this indicates that some wierdness (technical > term) is happening at a different level in the computing stack. I think > all of this points fairly strongly to it *not* being a PostgreSQL > algorithm problem, i.e. if the code was executed by an idealised Knuth- > like CPU then we would not get this problem. Plus, I have faith that if > it was a problem in that "plane" then you or another would have > uncovered it by now. > > > However, these certainly do not explain Ian's problem, because (a) these > > only apply to VACUUM, not ANALYZE; (b) they would only lock the table > > being VACUUMed, not other ones; (c) if these locks were to block the > > reader or writer thread, it'd manifest as blocking on a semaphore, not > > as a surge in LWLock thrashing. > > I've seen enough circumstantial evidence to connect the time spent > inside LWLockAcquire/Release as being connected to the Semaphore ops > within them, not the other aspects of the code. > > Months ago we discussed the problem of false sharing on closely packed > arrays of shared variables because of the large cache line size of the > Xeon MP. When last we touched on that thought, I focused on the thought > that the LWLock array was too tightly packed for the predefined locks. > What we didn't discuss (because I was too focused on the other array) > was the PGPROC shared array is equally tightly packed, which could give > problems on the semaphores in LWLock. > > Intel says fairly clearly that this would be an issue. > > > >> Is that Xeon MP then? > > > > > Yes. > > > > The LWLock activity is certainly suggestive of prior reports of > > excessive buffer manager lock contention, but it makes *no* sense that > > that would be higher with vacuum cost delay than without. I'd have > > expected the other way around. > > > > I'd really like to see a test case for this... > > My feeling is that a "micro-architecture" test would be more likely to > reveal some interesting information. > > Best Regards, Simon Riggs > > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend From pgsql-performance-owner@postgresql.org Fri Jul 15 02:08:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 64BA952D61 for ; Thu, 14 Jul 2005 23:08:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62506-04 for ; Fri, 15 Jul 2005 02:08:14 +0000 (GMT) Received: from hub.org (hub.org [200.46.204.220]) by svr1.postgresql.org (Postfix) with ESMTP id CAFD552D26 for ; Thu, 14 Jul 2005 23:08:14 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by hub.org (Postfix) with ESMTP id 2D99CA246A4 for ; Thu, 14 Jul 2005 23:08:14 -0300 (ADT) Received: from hub.org ([200.46.204.220]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62363-08 for ; Fri, 15 Jul 2005 02:08:13 +0000 (GMT) Received: from ganymede.hub.org (blk-224-176-51.eastlink.ca [24.224.176.51]) by hub.org (Postfix) with ESMTP id 831A6A246C4 for ; Thu, 14 Jul 2005 23:08:12 -0300 (ADT) Received: by ganymede.hub.org (Postfix, from userid 1000) id 707C33DEB3; Thu, 14 Jul 2005 23:08:23 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 66F3B3CEE4 for ; Thu, 14 Jul 2005 23:08:23 -0300 (ADT) X-Return-Path: X-Received: from localhost ([unix socket]) by ganymede.hub.org (Cyrus v2.2.12) with LMTPA; Wed, 13 Jul 2005 19:36:18 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 9C4B73846E for ; Wed, 13 Jul 2005 19:35:34 -0300 (ADT) X-Received: from mail.postgresql.org [200.46.204.71] by localhost with IMAP (fetchmail-6.2.5) for scrappy@localhost (single-drop); Wed, 13 Jul 2005 19:35:34 -0300 (ADT) X-Received: from svr1.postgresql.org ([unix socket]) by svr1.postgresql.org (Cyrus v2.2.12) with LMTPA; Wed, 13 Jul 2005 19:35:14 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EA83752C3B for ; Wed, 13 Jul 2005 19:35:13 -0300 (ADT) X-Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76830-09 for ; Wed, 13 Jul 2005 22:35:11 +0000 (GMT) X-Received: from nextmail.ru (unknown [83.222.5.149]) by svr1.postgresql.org (Postfix) with SMTP id D720952C2C for ; Wed, 13 Jul 2005 19:35:09 -0300 (ADT) X-Received: (qmail 44456 invoked from network); 13 Jul 2005 22:34:01 -0000 X-Received: from ppp85-140-125-114.pppoe.mtu-net.ru (HELO z) (85.140.125.114) by nextmail.ru with SMTP; 13 Jul 2005 22:34:01 -0000 From: "jobapply" To: Subject: Functional index is 5 times slower than the basic one Date: Thu, 14 Jul 2005 02:35:10 +0400 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.6353 Thread-Index: AcWH+yAUOw6G+W3wThu3dWh4z4ubKA== X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 Message-Id: <20050713223509.D720952C2C@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-DCC: : X-Spam-Pyzor: Reported 0 times. ReSent-Date: Thu, 14 Jul 2005 23:08:18 -0300 (ADT) Resent-From: "Marc G. Fournier" Resent-To: pgsql-performance@postgresql.org ReSent-Subject: Functional index is 5 times slower than the basic one ReSent-Message-ID: <20050714230818.L66818@ganymede.hub.org> X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.588 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/210 X-Sequence-Number: 13451 VACUUM FULL ANALYZE is performed right before tests. UPDATE test SET t = xpath_string(x, 'movie/rating'::text); is performed also to make selects equal. Xpath_string is IMMUTABLE. Table "public.test" Column | Type | Modifiers | Description --------+------------------+-----------+------------- i | integer | | t | text | | x | text | | d | double precision | | Indexes: "floatind" btree (d) "i_i" btree (i) CLUSTER "t_ind" btree (t) "t_x_ind" btree (t, xpath_string(x, 'data'::text)) "x_i" btree (xpath_string(x, 'data'::text)) "x_ii" btree (xpath_string(x, 'movie/characters/character'::text)) "x_iii" btree (xpath_string(x, 'movie/rating'::text)) Has OIDs: no explain analyze select count(*) from ( select * from test order by xpath_string(x, 'movie/rating'::text) limit 1000 offset 10 ) a; QUERY PLAN Aggregate (cost=342.37..342.37 rows=1 width=0) (actual time=403.580..403.584 rows=1 loops=1) -> Subquery Scan a (cost=3.27..339.87 rows=1000 width=0) (actual time=4.252..398.261 rows=1000 loops=1) -> Limit (cost=3.27..329.87 rows=1000 width=969) (actual time=4.242..389.557 rows=1000 loops=1) -> Index Scan using x_iii on test (cost=0.00..3266.00 rows=10000 width=969) (actual time=0.488..381.049 rows=1010 loops=1) Total runtime: 403.695 ms explain analyze select count(*) from ( select * from test order by t limit 1000 offset 10 ) a; QUERY PLAN Aggregate (cost=339.84..339.84 rows=1 width=0) (actual time=26.662..26.666 rows=1 loops=1) -> Subquery Scan a (cost=3.24..337.34 rows=1000 width=0) (actual time=0.228..22.416 rows=1000 loops=1) -> Limit (cost=3.24..327.34 rows=1000 width=969) (actual time=0.217..14.244 rows=1000 loops=1) -> Index Scan using t_ind on test (cost=0.00..3241.00 rows=10000 width=969) (actual time=0.099..6.371 rows=1010 loops=1) Total runtime: 26.749 ms From pgsql-performance-owner@postgresql.org Wed Jul 13 19:51:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AB12B52A00 for ; Wed, 13 Jul 2005 19:51:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 81225-09 for ; Wed, 13 Jul 2005 22:51:26 +0000 (GMT) Received: from cmailm4.svr.pol.co.uk (cmailm4.svr.pol.co.uk [195.92.193.211]) by svr1.postgresql.org (Postfix) with ESMTP id A1D0D52800 for ; Wed, 13 Jul 2005 19:51:26 -0300 (ADT) Received: from modem-430.llama.dialup.pol.co.uk ([217.135.177.174] helo=192.168.0.102) by cmailm4.svr.pol.co.uk with esmtp (Exim 4.41) id 1Dsq4a-0005eR-Bg; Wed, 13 Jul 2005 23:51:24 +0100 Subject: Re: Quad Opteron stuck in the mud From: Simon Riggs To: Dan Harris Cc: pgsql-performance@postgresql.org In-Reply-To: References: Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 13 Jul 2005 23:51:16 +0100 Message-Id: <1121295076.3970.381.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/156 X-Sequence-Number: 13397 On Wed, 2005-07-13 at 12:54 -0600, Dan Harris wrote: > For example, as I'm writing this, I am running an UPDATE statement > that will affect a small part of the table, and is querying on an > indexed boolean field. An indexed boolean field? Hopefully, ftindex is false for very few rows of the table? Try changing the ftindex to be a partial index, so only index the false values. Or don't index it at all. Split the table up into smaller pieces. Don't use an UPDATE statement. Keep a second table, and insert records into it when you would have updated previously. If a row is not found, you know that it has ftindex=true. That way, you'll never have row versions building up in the main table, which you'll still get even if you VACUUM. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Fri Jul 15 02:05:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0D87552D38 for ; Wed, 13 Jul 2005 23:53:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 39299-01 for ; Thu, 14 Jul 2005 02:53:21 +0000 (GMT) Received: from hub.org (hub.org [200.46.204.220]) by svr1.postgresql.org (Postfix) with ESMTP id 1026152D0C for ; Wed, 13 Jul 2005 23:53:21 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by hub.org (Postfix) with ESMTP id 83402A246B6 for ; Wed, 13 Jul 2005 23:53:20 -0300 (ADT) Received: from hub.org ([200.46.204.220]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 39255-01 for ; Thu, 14 Jul 2005 02:53:18 +0000 (GMT) Received: from ganymede.hub.org (blk-224-176-51.eastlink.ca [24.224.176.51]) by hub.org (Postfix) with ESMTP id DCB35A246AF for ; Wed, 13 Jul 2005 23:53:17 -0300 (ADT) Received: by ganymede.hub.org (Postfix, from userid 1000) id 2DB8434EC9; Wed, 13 Jul 2005 23:53:29 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 2807534584 for ; Wed, 13 Jul 2005 23:53:29 -0300 (ADT) X-Return-Path: X-Received: from localhost ([unix socket]) by ganymede.hub.org (Cyrus v2.2.12) with LMTPA; Wed, 13 Jul 2005 20:50:42 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 5427A3D65A for ; Wed, 13 Jul 2005 20:50:23 -0300 (ADT) X-Received: from mail.postgresql.org [200.46.204.71] by localhost with IMAP (fetchmail-6.2.5) for scrappy@localhost (single-drop); Wed, 13 Jul 2005 20:50:23 -0300 (ADT) X-Received: from svr1.postgresql.org ([unix socket]) by svr1.postgresql.org (Cyrus v2.2.12) with LMTPA; Wed, 13 Jul 2005 20:46:31 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B1FA452C58 for ; Wed, 13 Jul 2005 20:46:31 -0300 (ADT) X-Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95818-02 for ; Wed, 13 Jul 2005 23:46:23 +0000 (GMT) X-Received: from nextmail.ru (unknown [83.222.5.149]) by svr1.postgresql.org (Postfix) with SMTP id 302A952C5D for ; Wed, 13 Jul 2005 20:46:21 -0300 (ADT) X-Received: (qmail 24036 invoked from network); 13 Jul 2005 23:45:15 -0000 X-Received: from ppp85-140-125-114.pppoe.mtu-net.ru (HELO z) (85.140.125.114) by nextmail.ru with SMTP; 13 Jul 2005 23:45:15 -0000 From: "jobapply" To: Subject: Indexing Function called on VACUUM and sorting ? Date: Thu, 14 Jul 2005 03:46:23 +0400 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.6353 Thread-Index: AcWIBRPH0sxvB4uyS7CMZ6QrhZOEzw== X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 Message-Id: <20050713234621.302A952C5D@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-DCC: : X-Spam-Pyzor: Reported 0 times. ReSent-Date: Wed, 13 Jul 2005 23:53:21 -0300 (ADT) Resent-From: "Marc G. Fournier" Resent-To: pgsql-performance@postgresql.org ReSent-Subject: Indexing Function called on VACUUM and sorting ? ReSent-Message-ID: <20050713235321.A66818@ganymede.hub.org> X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.946 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/205 X-Sequence-Number: 13446 The question appeared because of strange issues with functional indexes. It seems they are recalculated even where it is obviously not needed. \d+ test: i | integer | | t | text | | x | text | | "i_i" btree (i) "x_i" btree (xpath_string(x, 'data'::text)) "x_ii" btree (xpath_string(x, 'movie/characters/character'::text)) "x_iii" btree (xpath_string(x, 'movie/rating'::text)) 1) When I run VACUUM FULL ANALYZE VERBOSE OR VACUUM ANALYZE After text INFO: analyzing "public.test" INFO: "test": scanned 733 of 733 pages, containing 10000 live rows and 0 dead rows; 3000 rows in sample, 10000 estimated total rows a lot of xpath_string calls occur. Does VACUUM rebuild indexes ? What for to recalculate that all? It makes VACUUMing very slow. Simple VACUUM call does not lead to such function calls. 2) When I do select * from test order by xpath_string(x, 'movie/rating'::text) limit 1000 offset 10; Planner uses index x_iii (as it should, ok here): Limit -> Index scan. But many of calls to xpath_string occur in execution time. Why ? Index is calculated already and everything is so immutable.. Please answer if you have any ideas.. Functional indexes seemed so great first, but now I uncover weird issues I can't understand.. From pgsql-performance-owner@postgresql.org Wed Jul 13 21:07:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 19D9752BC8 for ; Wed, 13 Jul 2005 21:07:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93839-10 for ; Thu, 14 Jul 2005 00:07:47 +0000 (GMT) Received: from intranet.squiz.net (extranet.squiz.net [203.222.154.144]) by svr1.postgresql.org (Postfix) with ESMTP id F14CF52A78 for ; Wed, 13 Jul 2005 21:07:45 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by intranet.squiz.net (Postfix) with ESMTP id AEA3337856; Thu, 14 Jul 2005 10:07:39 +1000 (EST) Received: from intranet.squiz.net ([127.0.0.1]) by localhost (intranet [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 01852-04; Thu, 14 Jul 2005 10:07:39 +1000 (EST) Received: from [192.168.0.116] (116.squiz.net [192.168.0.116]) by intranet.squiz.net (Postfix) with ESMTP id 53641376E0 for ; Thu, 14 Jul 2005 10:07:39 +1000 (EST) Message-ID: <42D5AC8A.9000003@squiz.net> Date: Thu, 14 Jul 2005 10:06:34 +1000 From: Marc McIntyre User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Slow Query Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at squiz.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/157 X-Sequence-Number: 13398 Hi, I'm having a problem with a query that performs a sequential scan on a table when it should be performing an index scan. The interesting thing is, when we dumped the database on another server, it performed an index scan on that server. The systems are running the same versions of postgres (7.4.8) and the problem persists after running an "ANALYZE VERBOSE" and after a "REINDEX TABLE sq_ast FORCE". The only difference that i can see is that the postgresql.conf files differ slightly, and the hardware is different. Note that the system performing the sequential scan is a Dual 2.8GHz Xeon, 4GB Ram, 300GB HDD. And the system performing an index scan is not as powerful. A copy of the postgresql.conf for the system performing the index scan can be found at http://beta.squiz.net/~mmcintyre/postgresql_squiz_uk.conf A copy of the postgresql.conf for the system performing the sequential scan can be found at http://beta.squiz.net/~mmcintyre/postgresql_future.conf The Query: SELECT a.assetid, a.short_name, a.type_code, a.status, l.linkid, l.link_type, l.sort_order, lt.num_kids, u.url, ap.path, CASE u.http WHEN '1' THEN 'http' WHEN '0' THEN 'https' END AS protocol FROM ((sq_ast a LEFT JOIN sq_ast_url u ON a.assetid = u.assetid) LEFT JOIN sq_ast_path ap ON a.assetid = ap.assetid),sq_ast_lnk l, sq_ast_lnk_tree lt WHERE a.assetid = l.minorid AND l.linkid = lt.linkid AND l.majorid = '2' AND l.link_type <= 2 ORDER BY sort_order The EXPLAIN ANALYZE from the system performing an sequential scan: QUERY PLAN Sort (cost=30079.79..30079.89 rows=42 width=113) (actual time=39889.989..39890.346 rows=260 loops=1) Sort Key: l.sort_order -> Nested Loop (cost=25638.02..30078.65 rows=42 width=113) (actual time=9056.336..39888.557 rows=260 loops=1) -> Merge Join (cost=25638.02..29736.01 rows=25 width=109) (actual time=9056.246..39389.359 rows=260 loops=1) Merge Cond: (("outer".assetid)::text = "inner"."?column5?") -> Merge Left Join (cost=25410.50..29132.82 rows=150816 width=97) (actual time=8378.176..38742.111 rows=150567 loops=1) Merge Cond: (("outer".assetid)::text = ("inner".assetid)::text) -> Merge Left Join (cost=25410.50..26165.14 rows=150816 width=83) (actual time=8378.130..9656.413 rows=150489 loops=1) Merge Cond: ("outer"."?column5?" = "inner"."?column4?") -> Sort (cost=25408.17..25785.21 rows=150816 width=48) (actual time=8377.733..8609.218 rows=150486 loops=1) Sort Key: (a.assetid)::text -> Seq Scan on sq_ast a (cost=0.00..12436.16 rows=150816 width=48) (actual time=0.011..5578.231 rows=151378 loops=1) -> Sort (cost=2.33..2.43 rows=37 width=43) (actual time=0.364..0.428 rows=37 loops=1) Sort Key: (u.assetid)::text -> Seq Scan on sq_ast_url u (cost=0.00..1.37 rows=37 width=43) (actual time=0.023..0.161 rows=37 loops=1) -> Index Scan using sq_ast_path_ast on sq_ast_path ap (cost=0.00..2016.98 rows=45893 width=23) (actual time=0.024..14041.571 rows=45812 loops=1) -> Sort (cost=227.52..227.58 rows=25 width=21) (actual time=131.838..132.314 rows=260 loops=1) Sort Key: (l.minorid)::text -> Index Scan using sq_ast_lnk_majorid on sq_ast_lnk l (cost=0.00..226.94 rows=25 width=21) (actual time=0.169..126.201 rows=260 loops=1) Index Cond: ((majorid)::text = '2'::text) Filter: (link_type <= 2) -> Index Scan using sq_ast_lnk_tree_linkid on sq_ast_lnk_tree lt (cost=0.00..13.66 rows=3 width=8) (actual time=1.539..1.900 rows=1 loops=260) Index Cond: ("outer".linkid = lt.linkid) Total runtime: 39930.395 ms The EXPLAIN ANALYZE from the system performing an index scan scan: Sort (cost=16873.64..16873.74 rows=40 width=113) (actual time=2169.905..2169.912 rows=13 loops=1) Sort Key: l.sort_order -> Nested Loop (cost=251.39..16872.58 rows=40 width=113) (actual time=45.724..2169.780 rows=13 loops=1) -> Merge Join (cost=251.39..16506.42 rows=32 width=109) (actual time=45.561..2169.012 rows=13 loops=1) Merge Cond: (("outer".assetid)::text = "inner"."?column5?") -> Merge Left Join (cost=2.33..15881.92 rows=149982 width=97) (actual time=0.530..1948.718 rows=138569 loops=1) Merge Cond: (("outer".assetid)::text = ("inner".assetid)::text) -> Merge Left Join (cost=2.33..13056.04 rows=149982 width=83) (actual time=0.406..953.781 rows=138491 loops=1) Merge Cond: (("outer".assetid)::text = "inner"."?column4?") -> Index Scan using sq_ast_pkey on sq_ast a (cost=0.00..14952.78 rows=149982 width=48) (actual time=0.154..388.872 rows=138488 loops=1) -> Sort (cost=2.33..2.43 rows=37 width=43) (actual time=0.235..0.264 rows=37 loops=1) Sort Key: (u.assetid)::text -> Seq Scan on sq_ast_url u (cost=0.00..1.37 rows=37 width=43) (actual time=0.036..0.103 rows=37 loops=1) -> Index Scan using sq_ast_path_ast on sq_ast_path ap (cost=0.00..1926.18 rows=42071 width=23) (actual time=0.110..105.918 rows=42661 loops=1) -> Sort (cost=249.05..249.14 rows=36 width=21) (actual time=0.310..0.324 rows=13 loops=1) Sort Key: (l.minorid)::text -> Index Scan using sq_ast_lnk_majorid on sq_ast_lnk l (cost=0.00..248.12 rows=36 width=21) (actual time=0.141..0.282 rows=13 loops=1) Index Cond: ((majorid)::text = '2'::text) Filter: (link_type <= 2) -> Index Scan using sq_ast_lnk_tree_linkid on sq_ast_lnk_tree lt (cost=0.00..11.41 rows=2 width=8) (actual time=0.043..0.045 rows=1 loops=13) Index Cond: ("outer".linkid = lt.linkid) Total runtime: 2170.165 ms (22 rows) THE DESC of the sq_ast table. future_v3_schema=# \d sq_ast Table "public.sq_ast" Column | Type | Modifiers -----------------------+-----------------------------+--------------------------------------------- assetid | character varying(15) | not null type_code | character varying(100) | not null version | character varying(20) | not null default '0.0.0'::character varying name | character varying(255) | not null default ''::character varying short_name | character varying(255) | not null default ''::character varying status | integer | not null default 1 languages | character varying(50) | not null default ''::character varying charset | character varying(50) | not null default ''::character varying force_secure | character(1) | not null default '0'::bpchar created | timestamp without time zone | not null created_userid | character varying(255) | not null updated | timestamp without time zone | not null updated_userid | character varying(255) | not null published | timestamp without time zone | published_userid | character varying(255) | status_changed | timestamp without time zone | status_changed_userid | character varying(255) | Indexes: "sq_asset_pkey" primary key, btree (assetid) "sq_ast_created" btree (created) "sq_ast_name" btree (name) "sq_ast_published" btree (published) "sq_ast_type_code" btree (type_code) "sq_ast_updated" btree (updated) Any ideas? -- Marc McIntyre MySource Matrix Lead Developer From pgsql-performance-owner@postgresql.org Fri Jul 15 02:08:37 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7BB6D52B1A for ; Wed, 13 Jul 2005 22:07:36 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 15054-01 for ; Thu, 14 Jul 2005 01:07:34 +0000 (GMT) Received: from hub.org (hub.org [200.46.204.220]) by svr1.postgresql.org (Postfix) with ESMTP id AA43752B27 for ; Wed, 13 Jul 2005 22:07:34 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by hub.org (Postfix) with ESMTP id 5431FA246A4 for ; Wed, 13 Jul 2005 22:07:34 -0300 (ADT) Received: from hub.org ([200.46.204.220]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 11806-08 for ; Thu, 14 Jul 2005 01:07:02 +0000 (GMT) Received: from ganymede.hub.org (blk-224-176-51.eastlink.ca [24.224.176.51]) by hub.org (Postfix) with ESMTP id 328A9A24653 for ; Wed, 13 Jul 2005 22:07:02 -0300 (ADT) Received: by ganymede.hub.org (Postfix, from userid 1000) id 423AF3C282; Wed, 13 Jul 2005 22:07:12 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 416D93C1F0 for ; Wed, 13 Jul 2005 22:07:12 -0300 (ADT) X-Return-Path: X-Received: from localhost ([unix socket]) by ganymede.hub.org (Cyrus v2.2.12) with LMTPA; Wed, 13 Jul 2005 21:25:47 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 84B4D3E9D7 for ; Wed, 13 Jul 2005 21:25:27 -0300 (ADT) X-Received: from mail.postgresql.org [200.46.204.71] by localhost with IMAP (fetchmail-6.2.5) for scrappy@localhost (single-drop); Wed, 13 Jul 2005 21:25:27 -0300 (ADT) X-Received: from svr1.postgresql.org ([unix socket]) by svr1.postgresql.org (Cyrus v2.2.12) with LMTPA; Wed, 13 Jul 2005 21:21:34 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C2E9B52B23 for ; Wed, 13 Jul 2005 21:21:34 -0300 (ADT) X-Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 02320-08 for ; Thu, 14 Jul 2005 00:21:33 +0000 (GMT) X-Received: from gghcwest.com (adsl-71-128-90-172.dsl.pltn13.pacbell.net [71.128.90.172]) by svr1.postgresql.org (Postfix) with ESMTP id E2C8452B1A for ; Wed, 13 Jul 2005 21:21:31 -0300 (ADT) X-Received: from toonses.gghcwest.com (toonses.gghcwest.com [192.168.168.115]) by gghcwest.com (8.12.10/8.12.9) with ESMTP id j6E0LVmd013717 for ; Wed, 13 Jul 2005 17:21:31 -0700 X-Received: from jwb by toonses.gghcwest.com with local (Exim 4.50) id 1DsrSZ-0005da-FH for pgsql-users@postgresql.org; Wed, 13 Jul 2005 17:20:15 -0700 Subject: JFS fastest filesystem for PostgreSQL? From: "Jeffrey W. Baker" To: pgsql-users@postgresql.org Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Wed, 13 Jul 2005 17:20:15 -0700 Message-Id: <1121300415.20950.33.camel@toonses.gghcwest.com> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-DCC: : X-Spam-Pyzor: Reported 0 times. ReSent-Date: Wed, 13 Jul 2005 22:07:04 -0300 (ADT) Resent-From: "Marc G. Fournier" Resent-To: pgsql-performance@postgresql.org ReSent-Subject: JFS fastest filesystem for PostgreSQL? ReSent-Message-ID: <20050713220704.V66818@ganymede.hub.org> X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/208 X-Sequence-Number: 13449 I just took delivery of a new system, and used the opportunity to benchmark postgresql 8.0 performance on various filesystems. The system in question runs Linux 2.6.12, has one CPU and 1GB of system memory, and 5 7200RPM SATA disks attached to an Areca hardware RAID controller having 128MB of cache. The caches are all write-back. I ran pgbench with a scale factor of 1000 and a total of 100,000 transactions per run. I varied the number of clients between 10 and 100. It appears from my test JFS is much faster than both ext3 and XFS for this workload. JFS and XFS were made with the mkfs defaults. ext3 was made with -T largefile4 and -E stride=32. The deadline scheduler was used for all runs (anticipatory scheduler is much worse). Here's the result, in transactions per second. ext3 jfs xfs ----------------------------- 10 Clients 55 81 68 100 Clients 61 100 64 ---------------------------- -jwb From pgsql-performance-owner@postgresql.org Thu Jul 14 00:21:38 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BB43552C5C for ; Thu, 14 Jul 2005 00:17:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 49237-03 for ; Thu, 14 Jul 2005 03:17:39 +0000 (GMT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 97FAD52824 for ; Thu, 14 Jul 2005 00:17:39 -0300 (ADT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.13.1/8.13.1) with ESMTP id j6E3HdaS050479 for ; Thu, 14 Jul 2005 03:17:39 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.13.1/8.13.1/Submit) id j6E3Do1x049325 for pgsql-performance@postgresql.org; Thu, 14 Jul 2005 03:13:50 GMT (envelope-from news) From: "Qingqing Zhou" X-Newsgroups: pgsql.performance Subject: Re: performance problems ... 100 cpu utilization Date: Thu, 14 Jul 2005 11:11:29 +0800 Organization: Hub.Org Networking Services Lines: 31 Message-ID: References: <42D567DB.6000100@works4me.com> Reply-To: "Qingqing Zhou" X-Complaints-To: usenet@news.hub.org X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 6.00.2800.1506 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1506 To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/158 X-Sequence-Number: 13399 "Dennis" writes > > checking the status of connections at this point ( ps -eaf | grep > "postgres:") where the CPU is maxed out I saw this: > > 127 idle > 12 bind > 38 parse > 34 select > Are you sure 100% CPU usage is solely contributed by Postgresql? Also, from the ps status you list, I can hardly see that's a problem because of problem you mentioned below. > > I know there has been discussion about problems on Xeon MP systems. Is > this what we are running into? Or is something else going on? Is there > other information I can provide that might help determine what is going on? > Here is a talk about Xeon-SMP spinlock contention problem: http://archives.postgresql.org/pgsql-performance/2005-05/msg00441.php Regards, Qingqing From pgsql-performance-owner@postgresql.org Thu Jul 14 01:28:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6822D52BFD for ; Thu, 14 Jul 2005 00:28:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 50081-09 for ; Thu, 14 Jul 2005 03:28:48 +0000 (GMT) Received: from lakermmtao11.cox.net (eastrmmtai02.cox.net [68.230.240.28]) by svr1.postgresql.org (Postfix) with ESMTP id D538752BF7 for ; Thu, 14 Jul 2005 00:28:44 -0300 (ADT) Received: from [127.0.0.1] (really [68.3.22.74]) by lakermmtao11.cox.net (InterMail vM.6.01.04.00 201-2131-118-20041027) with ESMTP id <20050714032840.KSYU28203.lakermmtao11.cox.net@[127.0.0.1]>; Wed, 13 Jul 2005 23:28:40 -0400 Message-ID: <42D5DBEB.4010503@works4me.com> Date: Wed, 13 Jul 2005 20:28:43 -0700 From: Dennis User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Qingqing Zhou Cc: pgsql-performance@postgresql.org Subject: Re: performance problems ... 100 cpu utilization References: <42D567DB.6000100@works4me.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Antivirus: avast! (VPS 0528-3, 07/13/2005), Outbound message X-Antivirus-Status: Clean X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/159 X-Sequence-Number: 13400 Qingqing Zhou wrote: >Are you sure 100% CPU usage is solely contributed by Postgresql? Also, from >the ps status you list, I can hardly see that's a problem because of problem >you mentioned below. > > The postgreSQL processes are what is taking up all the cpu. There aren't any other major applications on the machine. Its a dedicated database server, only for this application. It doesn't seem to make sense that PostgreSQL would be maxed out at this point. I think given the size of the box, it could do quite a bit better. So, what is going on? I don't know. Dennis From pgsql-performance-owner@postgresql.org Thu Jul 14 02:07:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9B9E152BF8 for ; Thu, 14 Jul 2005 02:07:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 70390-06 for ; Thu, 14 Jul 2005 05:07:47 +0000 (GMT) Received: from smtp003.bizmail.yahoo.com (smtp003.bizmail.yahoo.com [216.136.130.195]) by svr1.postgresql.org (Postfix) with SMTP id F2EBF52B06 for ; Thu, 14 Jul 2005 02:07:45 -0300 (ADT) Received: (qmail 4619 invoked from network); 14 Jul 2005 05:07:39 -0000 Received: from unknown (HELO ?192.168.3.41?) (david.mitchell@telogis.com@203.98.10.169 with plain) by smtp003.bizmail.yahoo.com with SMTP; 14 Jul 2005 05:07:39 -0000 Message-ID: <42D5F315.3090301@telogis.com> Date: Thu, 14 Jul 2005 17:07:33 +1200 From: David Mitchell User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dennis Cc: Qingqing Zhou , pgsql-performance@postgresql.org Subject: Re: performance problems ... 100 cpu utilization References: <42D567DB.6000100@works4me.com> <42D5DBEB.4010503@works4me.com> In-Reply-To: <42D5DBEB.4010503@works4me.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/160 X-Sequence-Number: 13401 What is the load average on this machine? Do you do many updates? If you do a lot of updates, perhaps you haven't vacuumed recently. We were seeing similar symptoms when we started load testing our stuff and it turned out we were vacuuming too infrequently. David Dennis wrote: > Qingqing Zhou wrote: > >> Are you sure 100% CPU usage is solely contributed by Postgresql? Also, >> from >> the ps status you list, I can hardly see that's a problem because of >> problem >> you mentioned below. >> >> > The postgreSQL processes are what is taking up all the cpu. There aren't > any other major applications on the machine. Its a dedicated database > server, only for this application. > > It doesn't seem to make sense that PostgreSQL would be maxed out at this > point. I think given the size of the box, it could do quite a bit > better. So, what is going on? I don't know. > > Dennis > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > > -- David Mitchell Software Engineer Telogis From pgsql-performance-owner@postgresql.org Thu Jul 14 02:08:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C8E0552C25 for ; Thu, 14 Jul 2005 02:08:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71721-09 for ; Thu, 14 Jul 2005 05:08:34 +0000 (GMT) Received: from ash25e.internode.on.net (ash25e.internode.on.net [203.16.214.182]) by svr1.postgresql.org (Postfix) with ESMTP id 75E3052BCA for ; Thu, 14 Jul 2005 02:08:32 -0300 (ADT) Received: from pluto.mirrabooka.com (ppp188-133.static.internode.on.net [150.101.188.133]) by ash25e.internode.on.net (8.12.9/8.12.6) with ESMTP id j6E58ULj089210 for ; Thu, 14 Jul 2005 14:38:31 +0930 (CST) (envelope-from alison@mirrabooka.com) Received: by pluto.mirrabooka.com (Postfix, from userid 506) id 6A4A92B11E; Thu, 14 Jul 2005 15:08:30 +1000 (EST) Date: Thu, 14 Jul 2005 15:08:30 +1000 To: pgsql-performance@postgresql.org Subject: lots of updates on small table Message-ID: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> User-Agent: nail 11.6 9/7/04 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: alison@mirrabooka.com (Alison Winters) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/161 X-Sequence-Number: 13402 Hi, Our application requires a number of processes to select and update rows from a very small (<10 rows) Postgres table on a regular and frequent basis. These processes often run for weeks at a time, but over the space of a few days we find that updates start getting painfully slow. We are running a full vacuum/analyze and reindex on the table every day, but the updates keep getting slower and slower until the processes are restarted. Restarting the processes isn't really a viable option in our 24/7 production environment, so we're trying to figure out what's causing the slow updates. The environment is as follows: Red Hat 9, kernel 2.4.20-8 PostgreSQL 7.3.2 ecpg 2.10.0 The processes are all compiled C programs accessing the database using ECPG. Does anyone have any thoughts on what might be happening here? Thanks Alison From pgsql-performance-owner@postgresql.org Thu Jul 14 02:58:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2B30F52C99 for ; Thu, 14 Jul 2005 02:58:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 84885-05 for ; Thu, 14 Jul 2005 05:58:27 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.205]) by svr1.postgresql.org (Postfix) with ESMTP id E60A152C64 for ; Thu, 14 Jul 2005 02:58:25 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id o1so191144nzf for ; Wed, 13 Jul 2005 22:58:26 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type; b=JzrDmm93bL/5YERxwfvHqY8pJR22+G4Yk+rsLrzGLscMSeHNjgO+1FE1ZE2NYX8RRYdYjTJFG5jb2d/Yrb+CbSoJz7dHuoZSLsPvHXUVdCiF8r3i3xXiMI5K7tjXtlL2GuHPVzPii1A305bqoKz0vvnKpV9GPCRiJ5RQHFrsu+0= Received: by 10.36.103.20 with SMTP id a20mr79672nzc; Wed, 13 Jul 2005 22:58:26 -0700 (PDT) Received: by 10.36.96.2 with HTTP; Wed, 13 Jul 2005 22:58:26 -0700 (PDT) Message-ID: <8d1c715f05071322582a27b996@mail.gmail.com> Date: Thu, 14 Jul 2005 10:58:26 +0500 From: Agha Asif Raza Reply-To: Agha Asif Raza To: pgsql-performance@postgresql.org Subject: Profiler for PostgreSQL Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_23643_11673196.1121320706415" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.32 tagged_above=0 required=5 tests=HTML_10_20, HTML_MESSAGE, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/162 X-Sequence-Number: 13403 ------=_Part_23643_11673196.1121320706415 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Is there any MS-SQL Server like 'Profiler' available for PostgreSQL? A=20 profiler is a tool that monitors the database server and outputs a detailed= =20 trace of all the transactions/queries that are executed on a database durin= g=20 a specified period of time. Kindly let me know if any of you knows of such = a=20 tool for PostgreSQL. Agha Asif Raza ------=_Part_23643_11673196.1121320706415 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline
Is there any MS-SQL Server like 'Profiler' available for PostgreSQL? A= profiler is a tool that monitors the database server and outputs a detaile= d trace of all the transactions/queries that are executed on a da= tabase during a specified period of time. Kindly let me know if any of= you knows of such a tool for PostgreSQL.
 
Agha Asif Raza
------=_Part_23643_11673196.1121320706415-- From pgsql-performance-owner@postgresql.org Thu Jul 14 03:13:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E64BD52D4E for ; Thu, 14 Jul 2005 03:12:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87897-07 for ; Thu, 14 Jul 2005 06:12:51 +0000 (GMT) Received: from stark.xeocode.com (stark.xeocode.com [216.58.44.227]) by svr1.postgresql.org (Postfix) with ESMTP id AE09652D47 for ; Thu, 14 Jul 2005 03:12:47 -0300 (ADT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1DswxS-0003ZU-00; Thu, 14 Jul 2005 02:12:30 -0400 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: Quad Opteron stuck in the mud References: In-Reply-To: From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 14 Jul 2005 02:12:30 -0400 Message-ID: <87oe95ykz5.fsf@stark.xeocode.com> Lines: 57 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/163 X-Sequence-Number: 13404 Dan Harris writes: > I keep the entire database vacuumed regularly. How often is "regularly"? We get frequent posts from people who think daily or every 4 hours is often enough. If the table is very busy you can need vacuums as often as every 15 minutes. Also, if you've done occasional massive batch updates like you describe here you may need a VACUUM FULL or alternatively a CLUSTER command to compact the table -- vacuum identifies the free space but if you've doubled the size of your table with a large update that's a lot more free space than you want hanging around waiting to be used. > For example, as I'm writing this, I am running an UPDATE statement that will > affect a small part of the table, and is querying on an indexed boolean field. ... > update eventactivity set ftindex = false where ftindex = true; ( added the > where clause because I don't want to alter where ftindex is null ) It's definitely worthwhile doing an "EXPLAIN UPDATE..." to see if this even used the index. It sounds like it did a sequential scan. Sequential scans during updates are especially painful. If there isn't free space lying around in the page where the updated record lies then another page has to be used or a new page added. If you're doing a massive update you can exhaust the free space available making the update have to go back and forth between the page being read and the end of the table where pages are being written. > ##### > > vmstat output ( as I am waiting for this to finish ): > procs -----------memory---------- ---swap-- -----io---- --system-- > ----cpu---- > r b swpd free buff cache si so bi bo in cs us sy id wa > 0 1 5436 2823908 26140 9183704 0 1 2211 540 694 336 9 2 76 13 [I assume you ran "vmstat 10" or some other interval and then waited for at least the second line? The first line outputted from vmstat is mostly meaningless] Um. That's a pretty meager i/o rate. Just over 2MB/s. The cpu is 76% idle which sounds fine but that could be one processor pegged at 100% while the others are idle. If this query is the only one running on the system then it would behave just like that. Is it possible you have some foreign keys referencing these records that you're updating? In which case every record being updated might be causing a full table scan on another table (or multiple other tables). If those tables are entirely in cache then it could cause these high cpu low i/o symptoms. Or are there any triggers on this table? -- greg From pgsql-performance-owner@postgresql.org Thu Jul 14 03:27:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 876AC52D3B for ; Thu, 14 Jul 2005 03:27:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91341-04 for ; Thu, 14 Jul 2005 06:27:33 +0000 (GMT) Received: from candle.pha.pa.us (candle.pha.pa.us [64.139.89.126]) by svr1.postgresql.org (Postfix) with ESMTP id 4FA6652C70 for ; Thu, 14 Jul 2005 03:27:32 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id j6E6RWv11806; Thu, 14 Jul 2005 02:27:32 -0400 (EDT) From: Bruce Momjian Message-Id: <200507140627.j6E6RWv11806@candle.pha.pa.us> Subject: Re: Profiler for PostgreSQL In-Reply-To: <8d1c715f05071322582a27b996@mail.gmail.com> To: Agha Asif Raza Date: Thu, 14 Jul 2005 02:27:32 -0400 (EDT) Cc: pgsql-performance@postgresql.org X-Mailer: ELM [version 2.4ME+ PL121 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/165 X-Sequence-Number: 13406 Agha Asif Raza wrote: > Is there any MS-SQL Server like 'Profiler' available for PostgreSQL? A > profiler is a tool that monitors the database server and outputs a detailed > trace of all the transactions/queries that are executed on a database during > a specified period of time. Kindly let me know if any of you knows of such a > tool for PostgreSQL. > Agha Asif Raza Sure see log_statement in postgresql.conf. There are a lot of settings in there to control what is logged. -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073 From pgsql-performance-owner@postgresql.org Thu Jul 14 03:28:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3705452D10 for ; Thu, 14 Jul 2005 03:28:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91792-03 for ; Thu, 14 Jul 2005 06:28:09 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 9A77652D4C for ; Thu, 14 Jul 2005 03:28:08 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 7F389644197 for ; Thu, 14 Jul 2005 00:27:45 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 01772-07 for ; Thu, 14 Jul 2005 00:27:43 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 8A9A6644195 for ; Thu, 14 Jul 2005 00:27:43 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <87oe95ykz5.fsf@stark.xeocode.com> References: <87oe95ykz5.fsf@stark.xeocode.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Quad Opteron stuck in the mud Date: Thu, 14 Jul 2005 00:28:05 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/166 X-Sequence-Number: 13407 On Jul 14, 2005, at 12:12 AM, Greg Stark wrote: > Dan Harris writes: > > >> I keep the entire database vacuumed regularly. >> > > How often is "regularly"? Well, once every day, but there aren't a ton of inserts or updates going on a daily basis. Maybe 1,000 total inserts? > > Also, if you've done occasional massive batch updates like you > describe here > you may need a VACUUM FULL or alternatively a CLUSTER command to > compact the > table -- vacuum identifies the free space but if you've doubled the > size of > your table with a large update that's a lot more free space than > you want > hanging around waiting to be used. > I have a feeling I'm going to need to do a cluster soon. I have done several mass deletes and reloads on it. > >> For example, as I'm writing this, I am running an UPDATE >> statement that will >> affect a small part of the table, and is querying on an indexed >> boolean field. >> > ... > >> update eventactivity set ftindex = false where ftindex = true; >> ( added the >> where clause because I don't want to alter where ftindex is null ) >> > > It's definitely worthwhile doing an "EXPLAIN UPDATE..." to see if > this even > used the index. It sounds like it did a sequential scan. > I tried that, and indeed it was using an index, although after reading Simon's post, I realize that was kind of dumb to have an index on a bool. I have since removed it. > Sequential scans during updates are especially painful. If there > isn't free > space lying around in the page where the updated record lies then > another page > has to be used or a new page added. If you're doing a massive > update you can > exhaust the free space available making the update have to go back > and forth > between the page being read and the end of the table where pages > are being > written. This is great info, thanks. > > >> ##### >> >> vmstat output ( as I am waiting for this to finish ): >> procs -----------memory---------- ---swap-- -----io---- --system-- >> ----cpu---- >> r b swpd free buff cache si so bi bo in >> cs us sy id wa >> 0 1 5436 2823908 26140 9183704 0 1 2211 540 694 >> 336 9 2 76 13 >> > > [I assume you ran "vmstat 10" or some other interval and then > waited for at > least the second line? The first line outputted from vmstat is mostly > meaningless] Yeah, this was at least 10 or so down the list ( the last one before ctrl-c ) > > Um. That's a pretty meager i/o rate. Just over 2MB/s. The cpu is > 76% idle > which sounds fine but that could be one processor pegged at 100% > while the > others are idle. If this query is the only one running on the > system then it > would behave just like that. Well, none of my processors had ever reached 100% until I changed to ext2 today ( read below for more info ) > > Is it possible you have some foreign keys referencing these records > that > you're updating? In which case every record being updated might be > causing a > full table scan on another table (or multiple other tables). If > those tables > are entirely in cache then it could cause these high cpu low i/o > symptoms. > No foreign keys or triggers. Ok, so I remounted this drive as ext2 shortly before sending my first email today. It wasn't enough time for me to notice the ABSOLUTELY HUGE difference in performance change. Ext3 must really be crappy for postgres, or at least is on this box. Now that it's ext2, this thing is flying like never before. My CPU utilization has skyrocketed, telling me that the disk IO was constraining it immensely. I always knew that it might be a little faster, but the box feels like it can "breathe" again and things that used to be IO intensive and run for an hour or more are now running in < 5 minutes. I'm a little worried about not having a journalized file system, but that performance difference will keep me from switching back ( at least to ext3! ). Maybe someday I will try XFS. I would be surprised if everyone who ran ext3 had this kind of problem, maybe it's specific to my kernel, raid controller, I don't know. But, this is amazing. It's like I have a new server. Thanks to everyone for their valuable input and a big thanks to all the dedicated pg developers on here who make this possible! -Dan From pgsql-performance-owner@postgresql.org Thu Jul 14 03:24:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F307D52D49 for ; Thu, 14 Jul 2005 03:24:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91092-04 for ; Thu, 14 Jul 2005 06:24:40 +0000 (GMT) Received: from houston.familyhealth.com.au (houston.au.fhnetwork.com [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 03BE452D46 for ; Thu, 14 Jul 2005 03:24:38 -0300 (ADT) Received: from houston.familyhealth.com.au (localhost [127.0.0.1]) by houston.familyhealth.com.au (Postfix) with ESMTP id 1A38A24FE0; Thu, 14 Jul 2005 14:24:31 +0800 (WST) Received: from [127.0.0.1] (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (Postfix) with ESMTP id 0118224FDD; Thu, 14 Jul 2005 14:24:31 +0800 (WST) Message-ID: <42D60664.90400@familyhealth.com.au> Date: Thu, 14 Jul 2005 14:29:56 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Agha Asif Raza Cc: pgsql-performance@postgresql.org Subject: Re: Profiler for PostgreSQL References: <8d1c715f05071322582a27b996@mail.gmail.com> In-Reply-To: <8d1c715f05071322582a27b996@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.062 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/164 X-Sequence-Number: 13405 Try turning on query logging and using the 'pqa' utility on pgfoundry.org. Chris Agha Asif Raza wrote: > Is there any MS-SQL Server like 'Profiler' available for PostgreSQL? A > profiler is a tool that monitors the database server and outputs a > detailed trace of all the transactions/queries that are executed on a > database during a specified period of time. Kindly let me know if any of > you knows of such a tool for PostgreSQL. > > Agha Asif Raza From pgsql-performance-owner@postgresql.org Thu Jul 14 03:33:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 07AE052D3D for ; Thu, 14 Jul 2005 03:33:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 90271-08 for ; Thu, 14 Jul 2005 06:33:42 +0000 (GMT) Received: from mail24.sea5.speakeasy.net (mail24.sea5.speakeasy.net [69.17.117.26]) by svr1.postgresql.org (Postfix) with ESMTP id 1555F52D4E for ; Thu, 14 Jul 2005 03:33:41 -0300 (ADT) Received: (qmail 7549 invoked from network); 14 Jul 2005 06:33:42 -0000 Received: from dsl081-060-184.sfo1.dsl.speakeasy.net (HELO noodles) ([64.81.60.184]) (envelope-sender ) by mail24.sea5.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 14 Jul 2005 06:33:41 -0000 Received: from jwb by noodles with local (Exim 3.36 #1 (Debian)) id 1DsxHx-0001lg-00 for ; Wed, 13 Jul 2005 23:33:41 -0700 Subject: JFS fastest filesystem for PostgreSQL? From: "Jeffrey W. Baker" To: pgsql-performance@postgresql.org Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Wed, 13 Jul 2005 23:33:41 -0700 Message-Id: <1121322821.6719.0.camel@noodles> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/167 X-Sequence-Number: 13408 [reposted due to delivery error -jwb] I just took delivery of a new system, and used the opportunity to benchmark postgresql 8.0 performance on various filesystems. The system in question runs Linux 2.6.12, has one CPU and 1GB of system memory, and 5 7200RPM SATA disks attached to an Areca hardware RAID controller having 128MB of cache. The caches are all write-back. I ran pgbench with a scale factor of 1000 and a total of 100,000 transactions per run. I varied the number of clients between 10 and 100. It appears from my test JFS is much faster than both ext3 and XFS for this workload. JFS and XFS were made with the mkfs defaults. ext3 was made with -T largefile4 and -E stride=32. The deadline scheduler was used for all runs (anticipatory scheduler is much worse). Here's the result, in transactions per second. ext3 jfs xfs ----------------------------- 10 Clients 55 81 68 100 Clients 61 100 64 ---------------------------- -jwb From pgsql-performance-owner@postgresql.org Fri Jul 15 02:08:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7E92A52CA2 for ; Thu, 14 Jul 2005 17:58:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94441-06 for ; Thu, 14 Jul 2005 20:58:09 +0000 (GMT) Received: from hub.org (hub.org [200.46.204.220]) by svr1.postgresql.org (Postfix) with ESMTP id 0DB1352CA0 for ; Thu, 14 Jul 2005 17:58:09 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by hub.org (Postfix) with ESMTP id 64975A24653 for ; Thu, 14 Jul 2005 17:58:08 -0300 (ADT) Received: from hub.org ([200.46.204.220]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94466-08 for ; Thu, 14 Jul 2005 20:57:59 +0000 (GMT) Received: from ganymede.hub.org (blk-224-176-51.eastlink.ca [24.224.176.51]) by hub.org (Postfix) with ESMTP id F02F0A246B6 for ; Thu, 14 Jul 2005 17:57:58 -0300 (ADT) Received: by ganymede.hub.org (Postfix, from userid 1000) id 402093DF45; Thu, 14 Jul 2005 17:58:04 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 3CF683DF2F for ; Thu, 14 Jul 2005 17:58:04 -0300 (ADT) X-Return-Path: X-Received: from localhost ([unix socket]) by ganymede.hub.org (Cyrus v2.2.12) with LMTPA; Thu, 14 Jul 2005 03:35:49 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id F172945997 for ; Thu, 14 Jul 2005 03:35:23 -0300 (ADT) X-Received: from mail.postgresql.org [200.46.204.71] by localhost with IMAP (fetchmail-6.2.5) for scrappy@localhost (single-drop); Thu, 14 Jul 2005 03:35:24 -0300 (ADT) X-Received: from svr1.postgresql.org ([unix socket]) by svr1.postgresql.org (Cyrus v2.2.12) with LMTPA; Thu, 14 Jul 2005 03:34:03 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C130652D49 for ; Thu, 14 Jul 2005 03:34:03 -0300 (ADT) X-Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 90271-10 for ; Thu, 14 Jul 2005 06:33:56 +0000 (GMT) X-Received: from nextmail.ru (unknown [83.222.5.149]) by svr1.postgresql.org (Postfix) with SMTP id CFE0552D10 for ; Thu, 14 Jul 2005 03:33:54 -0300 (ADT) X-Received: (qmail 63581 invoked from network); 14 Jul 2005 06:32:47 -0000 X-Received: from ppp85-140-125-114.pppoe.mtu-net.ru (HELO z) (85.140.125.114) by nextmail.ru with SMTP; 14 Jul 2005 06:32:47 -0000 From: "jobapply" To: Subject: Indexing Function called on VACUUM and sorting ? Date: Thu, 14 Jul 2005 10:33:53 +0400 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.6353 Thread-Index: AcWIBRPH0sxvB4uyS7CMZ6QrhZOEzw== X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 Message-Id: <20050714063354.CFE0552D10@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-DCC: : X-Spam-Pyzor: Reported 0 times. ReSent-Date: Thu, 14 Jul 2005 17:57:59 -0300 (ADT) Resent-From: "Marc G. Fournier" Resent-To: pgsql-performance@postgresql.org ReSent-Subject: Indexing Function called on VACUUM and sorting ? ReSent-Message-ID: <20050714175759.A66818@ganymede.hub.org> X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.647 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/209 X-Sequence-Number: 13450 It seems functional indexes are recalculated even where it is obviously not needed. \d+ test: i | integer | | t | text | | x | text | | "i_i" btree (i) "x_iii" btree (xpath_string(x, 'movie/rating'::text)) 1) When I run VACUUM FULL ANALYZE VERBOSE OR VACUUM ANALYZE a lot of xpath_string calls occur. Does VACUUM rebuild indexes ? What for to recalculate that all? It makes VACUUMing very slow. Simple VACUUM call does not lead to such function calls. 2) When I do select * from test order by xpath_string(x, 'movie/rating'::text) limit 1000 offset 10; Planner uses index x_iii (as it should, ok here): Limit -> Index scan. But many of calls to xpath_string occur in execution time. Why ? Index is calculated already and everything is so immutable.. Please answer if you have any ideas.. Functional indexes seemed so great first, but now I uncover weird issues I can't understand.. From pgsql-performance-owner@postgresql.org Thu Jul 14 03:54:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7BC2C52C2B for ; Thu, 14 Jul 2005 03:54:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 97294-07 for ; Thu, 14 Jul 2005 06:54:40 +0000 (GMT) Received: from lakermmtao06.cox.net (lakermmtao06.cox.net [68.230.240.33]) by svr1.postgresql.org (Postfix) with ESMTP id 4F0D152D10 for ; Thu, 14 Jul 2005 03:54:37 -0300 (ADT) Received: from [127.0.0.1] (really [68.3.22.74]) by lakermmtao06.cox.net (InterMail vM.6.01.04.00 201-2131-118-20041027) with ESMTP id <20050714065435.PEON2772.lakermmtao06.cox.net@[127.0.0.1]>; Thu, 14 Jul 2005 02:54:35 -0400 Message-ID: <42D60C2E.3020004@works4me.com> Date: Wed, 13 Jul 2005 23:54:38 -0700 From: Dennis User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: David Mitchell Cc: Qingqing Zhou , pgsql-performance@postgresql.org Subject: Re: performance problems ... 100 cpu utilization References: <42D567DB.6000100@works4me.com> <42D5DBEB.4010503@works4me.com> <42D5F315.3090301@telogis.com> In-Reply-To: <42D5F315.3090301@telogis.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Antivirus: avast! (VPS 0528-3, 07/13/2005), Outbound message X-Antivirus-Status: Clean X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/168 X-Sequence-Number: 13409 David Mitchell wrote: > What is the load average on this machine? Do you do many updates? If > you do a lot of updates, perhaps you haven't vacuumed recently. We > were seeing similar symptoms when we started load testing our stuff > and it turned out we were vacuuming too infrequently. The load average at the 100% utilization point was about 30! A vacuum analyze was done before the test was started. I believe there are many more selects than updates happening at any one time. Dennis From pgsql-performance-owner@postgresql.org Thu Jul 14 04:07:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C050A52C2B for ; Thu, 14 Jul 2005 04:07:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 00182-05 for ; Thu, 14 Jul 2005 07:06:58 +0000 (GMT) Received: from mail.aveo.aveopharma.com (67.109.105.227.ptr.us.xo.net [67.109.105.227]) by svr1.postgresql.org (Postfix) with SMTP id A9FD3529FA for ; Thu, 14 Jul 2005 04:06:54 -0300 (ADT) X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: Re: JFS fastest filesystem for PostgreSQL? Date: Thu, 14 Jul 2005 03:08:41 -0400 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] JFS fastest filesystem for PostgreSQL? Thread-Index: AcWIPqkoJ2ayb+NuQBS/NaTV7qnmNwAA7l1A From: "Dmitri Bichko" To: "Jeffrey W. Baker" , X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.062 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/169 X-Sequence-Number: 13410 I was wondering - have you had a chance to run the same benchmarks on ReiserFS (ideally both 3 and 4, with notail)? I'd be quite interested to see how it performs in this situation since it's my fs of choice for most things. Thanks, Dmitri -----Original Message----- From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of Jeffrey W. Baker Sent: Thursday, July 14, 2005 2:34 AM To: pgsql-performance@postgresql.org Subject: [PERFORM] JFS fastest filesystem for PostgreSQL? [reposted due to delivery error -jwb] I just took delivery of a new system, and used the opportunity to benchmark postgresql 8.0 performance on various filesystems. The system in question runs Linux 2.6.12, has one CPU and 1GB of system memory, and 5 7200RPM SATA disks attached to an Areca hardware RAID controller having 128MB of cache. The caches are all write-back. I ran pgbench with a scale factor of 1000 and a total of 100,000 transactions per run. I varied the number of clients between 10 and 100. It appears from my test JFS is much faster than both ext3 and XFS for this workload. JFS and XFS were made with the mkfs defaults. ext3 was made with -T largefile4 and -E stride=3D32. The deadline scheduler was used for all runs (anticipatory scheduler is much worse). Here's the result, in transactions per second. ext3 jfs xfs ----------------------------- 10 Clients 55 81 68 100 Clients 61 100 64 ---------------------------- -jwb ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match The information transmitted is intended only for the person or entity to = which it is addressed and may contain confidential and/or privileged mate= rial. Any review, retransmission, dissemination or other use of, or takin= g of any action in reliance upon, this information by persons or entities= other than the intended recipient is prohibited. If you received this in= error, please contact the sender and delete the material from any comput= er From pgsql-performance-owner@postgresql.org Fri Jul 15 02:06:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 40B1752C70 for ; Thu, 14 Jul 2005 04:19:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 01513-09 for ; Thu, 14 Jul 2005 07:19:38 +0000 (GMT) Received: from floppy.pyrenet.fr (news.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id B734252C2B for ; Thu, 14 Jul 2005 04:19:35 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id D494430B43; Thu, 14 Jul 2005 09:24:39 +0200 (MET DST) From: "Relaxin" X-Newsgroups: pgsql.performance Subject: PostgresSQL vs. Firebird Date: Thu, 14 Jul 2005 00:19:34 -0700 Organization: Hub.Org Networking Services Lines: 23 Message-ID: X-Complaints-To: usenet@news.hub.org X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 6.00.2900.2527 X-RFC2646: Format=Flowed; Original X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2527 To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.185 tagged_above=0 required=5 tests=PRIORITY_NO_NAME X-Spam-Level: * X-Archive-Number: 200507/207 X-Sequence-Number: 13448 Before I ask, I don't want to start a war. Can someone here give me an honest opinion of how PostgresSQL (PG) is better than Firebird on Windows? I've just recently started reading the Firebird NG and a poster over there has brought up some serious issues with Firebird, but they seem to not take the issues seriously. I first wanted to go with Firebird for 2 reasons... Very easy to configure and very easy to install. I assumed that the database worked ok, but I'm not so sure now. So, I've decided to give PG a try...I've downloaded it, but haven't installed it yet. So any provable information that you can provide as to why/how PG is better/faster/easier/reliable than Firebird would be greatly appreciated. Thanks From pgsql-performance-owner@postgresql.org Thu Jul 14 05:03:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9834E52D3B for ; Thu, 14 Jul 2005 05:03:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 11396-05 for ; Thu, 14 Jul 2005 08:03:12 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.204]) by svr1.postgresql.org (Postfix) with ESMTP id E494B52C6B for ; Thu, 14 Jul 2005 05:03:09 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id z3so193403nzf for ; Thu, 14 Jul 2005 01:03:10 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=L4TVgV9rP8fiocoQAv/EPG6AsQr9pqx58I0FQrnJz0FPVcf5AYBNDQTtmTnPk47Z61DS0npX6JiTvgR9fZixyZ/XbBJIjZWqH3pr0MXklhybA5dLO7Myfh37cmPGT1+G+TXNkfuiL9nFoxS/tv3h8kUxjXCSKD7rtch6IYr0KYY= Received: by 10.36.79.4 with SMTP id c4mr163872nzb; Thu, 14 Jul 2005 01:03:10 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Thu, 14 Jul 2005 01:03:10 -0700 (PDT) Message-ID: <758d5e7f05071401034665841c@mail.gmail.com> Date: Thu, 14 Jul 2005 10:03:10 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: "Jeffrey W. Baker" Subject: Re: JFS fastest filesystem for PostgreSQL? Cc: pgsql-performance@postgresql.org In-Reply-To: <1121322821.6719.0.camel@noodles> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1121322821.6719.0.camel@noodles> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.615 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/170 X-Sequence-Number: 13411 On 7/14/05, Jeffrey W. Baker wrote: > [reposted due to delivery error -jwb] >=20 > I just took delivery of a new system, and used the opportunity to > benchmark postgresql 8.0 performance on various filesystems. The system > in question runs Linux 2.6.12, has one CPU and 1GB of system memory, and > 5 7200RPM SATA disks attached to an Areca hardware RAID controller > having 128MB of cache. The caches are all write-back. >=20 > I ran pgbench with a scale factor of 1000 and a total of 100,000 > transactions per run. I varied the number of clients between 10 and > 100. It appears from my test JFS is much faster than both ext3 and XFS > for this workload. JFS and XFS were made with the mkfs defaults. ext3 > was made with -T largefile4 and -E stride=3D32. The deadline scheduler > was used for all runs (anticipatory scheduler is much worse). >=20 > Here's the result, in transactions per second. >=20 > ext3 jfs xfs > ----------------------------- > 10 Clients 55 81 68 > 100 Clients 61 100 64 > ---------------------------- If you still have a chance, could you do tests with other journaling options for ext3 (journal=3Dwriteback, journal=3Ddata)? And could you give figures about performace of other IO elevators? I mean, you wrote that anticipatory is much wore -- how much worse? :) Could you give numbers for deadline,anticipatory,cfq elevators? :) And, additionally would it be possible to give numbers for bonnie++ results? To see how does pgbench to bonnie++ relate? Regards, Dawid From pgsql-performance-owner@postgresql.org Thu Jul 14 05:05:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8844752D52 for ; Thu, 14 Jul 2005 05:05:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 11817-09 for ; Thu, 14 Jul 2005 08:05:25 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 60C7552D56 for ; Thu, 14 Jul 2005 05:05:24 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 39EE5644195 for ; Thu, 14 Jul 2005 02:05:03 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 02754-01 for ; Thu, 14 Jul 2005 02:05:01 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 00A57644191 for ; Thu, 14 Jul 2005 02:05:00 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) Content-Transfer-Encoding: 7bit Message-Id: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: Dan Harris Subject: slow joining very large table to smaller ones Date: Thu, 14 Jul 2005 02:05:23 -0600 X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/171 X-Sequence-Number: 13412 I'm trying to improve the speed of this query: explain select recordtext from eventactivity inner join ( select incidentid from k_r where id = 94 ) a using ( incidentid ) inner join ( select incidentid from k_b where id = 107 ) b using ( incidentid ); QUERY PLAN ------------------------------------------------------------------------ -------------------------------------- Merge Join (cost=2747.29..4249364.96 rows=11968693 width=35) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Merge Join (cost=1349.56..4230052.73 rows=4413563 width=117) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Index Scan using eventactivity1 on eventactivity (cost=0.00..4051200.28 rows=44519781 width=49) -> Sort (cost=1349.56..1350.85 rows=517 width=68) Sort Key: (k_b.incidentid)::text -> Index Scan using k_b_idx on k_b (cost=0.00..1326.26 rows=517 width=68) Index Cond: (id = 107) -> Sort (cost=1397.73..1399.09 rows=542 width=68) Sort Key: (k_r.incidentid)::text -> Index Scan using k_r_idx on k_r (cost=0.00..1373.12 rows=542 width=68) Index Cond: (id = 94) (13 rows) There are many millions of rows in eventactivity. There are a few ten-thousand rows in k_r and k_b. There is an index on 'incidentid' in all three tables. There should only be less than 100 rows matched in k_r and k_b total. That part on its own is very very fast. But, it should have those 100 or so incidentids extracted in under a second and then go into eventactivity AFTER doing that. At least, that's my intention to make this fast. Right now, it looks like pg is trying to sort the entire eventactivity table for the merge join which is taking several minutes to do. Can I rephrase this so that it does the searching through k_r and k_b FIRST and then go into eventactivity using the index on incidentid? It seems like that shouldn't be too hard to make fast but my SQL query skills are only average. Thanks -Dan From pgsql-performance-owner@postgresql.org Thu Jul 14 06:56:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7F0EC52C28 for ; Thu, 14 Jul 2005 06:56:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36973-10 for ; Thu, 14 Jul 2005 09:56:36 +0000 (GMT) Received: from mail0.rawbw.com (mail0.rawbw.com [198.144.192.41]) by svr1.postgresql.org (Postfix) with ESMTP id 90C2E52C16 for ; Thu, 14 Jul 2005 06:56:31 -0300 (ADT) Received: (from www@localhost) by mail0.rawbw.com (8.11.6p2/8.11.6) id j6E9uXF61147 for pgsql-performance@postgresql.org; Thu, 14 Jul 2005 02:56:33 -0700 (PDT) Received: from 198.144.203.173 ([198.144.203.173]) by webmail.rawbw.com (IMP) with HTTP for ; Thu, 14 Jul 2005 02:56:33 -0700 Message-ID: <1121334993.42d636d187a3b@webmail.rawbw.com> Date: Thu, 14 Jul 2005 02:56:33 -0700 From: mudfoot@rawbw.com To: pgsql-performance@postgresql.org Subject: Re: JFS fastest filesystem for PostgreSQL? References: <1121322821.6719.0.camel@noodles> In-Reply-To: <1121322821.6719.0.camel@noodles> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit User-Agent: Internet Messaging Program (IMP) 3.2.1 X-Originating-IP: 198.144.203.173 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.178 tagged_above=0 required=5 tests=AWL, NO_REAL_NAME X-Spam-Level: X-Archive-Number: 200507/172 X-Sequence-Number: 13413 Quoting "Jeffrey W. Baker" : > > Here's the result, in transactions per second. > > ext3 jfs xfs > ---------------------- ------- > 10 Clients 55 81 68 > 100 Clients 61 100 64 > ---------------------------- Was fsync true? And have you tried ext2? Legend has it that ext2 is the fastest thing going for synchronous writes (besides O_DIRECT or raw) because there's no journal. > > -jwb > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match > From pgsql-performance-owner@postgresql.org Thu Jul 14 07:30:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4FF1F5294D for ; Thu, 14 Jul 2005 07:30:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 45814-04 for ; Thu, 14 Jul 2005 10:30:45 +0000 (GMT) Received: from vms040pub.verizon.net (vms040pub.verizon.net [206.46.252.40]) by svr1.postgresql.org (Postfix) with ESMTP id 1C0A35280C for ; Thu, 14 Jul 2005 07:30:42 -0300 (ADT) Received: from osgiliath.mathom.us ([70.108.75.132]) by vms040.mailsrvcs.net (Sun Java System Messaging Server 6.2 HotFix 0.04 (built Dec 24 2004)) with ESMTPA id <0IJM0066N5770OC0@vms040.mailsrvcs.net> for pgsql-performance@postgresql.org; Thu, 14 Jul 2005 05:30:44 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by osgiliath.mathom.us (Postfix) with ESMTP id 99CE76005EA for ; Thu, 14 Jul 2005 06:30:42 -0400 (EDT) Received: from osgiliath.mathom.us ([127.0.0.1]) by localhost (osgiliath [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 10654-01-20 for ; Thu, 14 Jul 2005 06:30:41 -0400 (EDT) Received: by osgiliath.mathom.us (Postfix, from userid 1000) id 067B9600197; Thu, 14 Jul 2005 06:30:41 -0400 (EDT) Date: Thu, 14 Jul 2005 06:30:40 -0400 From: Michael Stone Subject: Re: JFS fastest filesystem for PostgreSQL? In-reply-to: <1121322821.6719.0.camel@noodles> To: pgsql-performance@postgresql.org Mail-Followup-To: pgsql-performance@postgresql.org Message-id: <20050714103040.GR19080@mathom.us> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-disposition: inline X-Pgp-Fingerprint: 53 FF 38 00 E7 DD 0A 9C 84 52 84 C5 EE DF 7C 88 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mathom.us References: <1121322821.6719.0.camel@noodles> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.011 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/173 X-Sequence-Number: 13414 Did you seperate the data & the transaction log? I've noticed less than optimal performance on xfs if the transaction log is on the xfs data partition, and it's silly to put the xlog on a journaled filesystem anyway. Try putting xlog on an ext2 for all the tests. Mike Stone From pgsql-performance-owner@postgresql.org Thu Jul 14 08:20:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 29F9452C34 for ; Thu, 14 Jul 2005 08:20:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58844-06 for ; Thu, 14 Jul 2005 11:20:53 +0000 (GMT) Received: from quasar.skima.is (quasar.skima.is [212.30.200.205]) by svr1.postgresql.org (Postfix) with ESMTP id 659A952A42 for ; Thu, 14 Jul 2005 08:20:49 -0300 (ADT) Received: from adsl7-104-140.du.simnet.is ([85.220.100.140] [85.220.100.140]) by quasar.skima.is with ESMTP; Thu, 14 Jul 2005 11:20:26 Z Subject: Re: Slow Query From: Ragnar =?ISO-8859-1?Q?Hafsta=F0?= To: Marc McIntyre Cc: pgsql-performance@postgresql.org In-Reply-To: <42D5AC8A.9000003@squiz.net> References: <42D5AC8A.9000003@squiz.net> Content-Type: text/plain Date: Thu, 14 Jul 2005 11:22:33 +0000 Message-Id: <1121340153.31360.22.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/174 X-Sequence-Number: 13415 On Thu, 2005-07-14 at 10:06 +1000, Marc McIntyre wrote: > I'm having a problem with a query that performs a sequential scan on a > table when it should be performing an index scan. The interesting thing > is, when we dumped the database on another server, it performed an index > scan on that server. ... > The EXPLAIN ANALYZE from the system performing an sequential scan: > > QUERY PLAN > Sort (cost=30079.79..30079.89 rows=42 width=113) (actual time=39889.989..39890.346 rows=260 loops=1) ... > The EXPLAIN ANALYZE from the system performing an index scan scan: > Sort (cost=16873.64..16873.74 rows=40 width=113) (actual time=2169.905..2169.912 rows=13 loops=1) looks like the first query is returning 260 rows, but the second one 13 this may not be your problem, but are you sure you are using the same query on the same data here ? gnari From pgsql-performance-owner@postgresql.org Thu Jul 14 09:18:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4F5DA52C5C for ; Thu, 14 Jul 2005 09:18:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 75524-02 for ; Thu, 14 Jul 2005 12:18:49 +0000 (GMT) Received: from p0f.net (p0f.net [193.77.154.190]) by svr1.postgresql.org (Postfix) with ESMTP id ACBF652C46 for ; Thu, 14 Jul 2005 09:18:47 -0300 (ADT) Received: from [10.10.105.30] ([193.77.165.58]) (authenticated bits=0) by p0f.net (8.13.3/8.13.3) with ESMTP id j6EC7Uo8014077 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Thu, 14 Jul 2005 14:07:32 +0200 Message-ID: <42D65778.7010500@p0f.net> Date: Thu, 14 Jul 2005 14:15:52 +0200 From: Grega Bremec Organization: P0F User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dawid Kuroczko Cc: "Jeffrey W. Baker" , pgsql-performance@postgresql.org Subject: Re: JFS fastest filesystem for PostgreSQL? References: <1121322821.6719.0.camel@noodles> <758d5e7f05071401034665841c@mail.gmail.com> In-Reply-To: <758d5e7f05071401034665841c@mail.gmail.com> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.429 tagged_above=0 required=5 tests=AWL, RCVD_IN_NJABL_DUL X-Spam-Level: * X-Archive-Number: 200507/175 X-Sequence-Number: 13416 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Dawid Kuroczko wrote: | | If you still have a chance, could you do tests with other journaling | options for ext3 (journal=writeback, journal=data)? And could you | give figures about performace of other IO elevators? I mean, you | wrote that anticipatory is much wore -- how much worse? :) Could | you give numbers for deadline,anticipatory,cfq elevators? :) | | And, additionally would it be possible to give numbers for bonnie++ | results? To see how does pgbench to bonnie++ relate? | Hello, list. I've been thinking on this one for a while - I'm not sure as to what ratio pgbench has with regard to stressing CPU vs. I/O. There is one thing that's definitely worth mentioning though: in the tests that I've been doing with bonnie++ and iozone at my former job, while building a distributed indexing engine, jfs was the one filesystem with the least strain on the CPU, which might be one of the deciding factors in making it look good for a particular workload. I'm afraid I don't have any concrete figures to offer as the material itself was classified. I can tell though that we've been comparing it with both ext2 and ext3, as well as xfs, and notably, xfs was the worst CPU hog of all. The CPU load difference between jfs and xfs was about 10% in favor of jfs in all random read/write tests, and the interesting thing is, jfs managed to shuffle around quite a lot of data: the mbps/cpu% ratio in xfs was much worse. As expected, there wasn't much difference in block transfer tests, but jfs was slightly winning in the area of CPU consumption and slightly lagging in the transfer rate field. What is a little bit concerning though, is the fact that some Linux distributors like SuSE have removed jfs support from their admin tooling due to technical problems with jfs (http://your-local-suse-mirror/.../suse/i386/9.3/docu/RELEASE-NOTES.en.html#14) I'm curious as to what this means - did they have problems integrating it into their toolchain or are there actual problems going on in jfs currently? Kind regards, - -- Grega Bremec gregab at p0f dot net -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFC1ld4fu4IwuB3+XoRAqEyAJ0TS9son+brhbQGtV7Cw7T8wa9W2gCfZ02/ dWm/E/Dc99TyKbxxl2tKaZc= =nvv3 -----END PGP SIGNATURE----- From pgsql-performance-owner@postgresql.org Thu Jul 14 10:21:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B1B5A52D31 for ; Thu, 14 Jul 2005 10:21:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86615-08 for ; Thu, 14 Jul 2005 13:20:56 +0000 (GMT) Received: from vms046pub.verizon.net (vms046pub.verizon.net [206.46.252.46]) by svr1.postgresql.org (Postfix) with ESMTP id DB9C352C34 for ; Thu, 14 Jul 2005 10:20:54 -0300 (ADT) Received: from osgiliath.mathom.us ([70.108.75.132]) by vms046.mailsrvcs.net (Sun Java System Messaging Server 6.2 HotFix 0.04 (built Dec 24 2004)) with ESMTPA id <0IJM005XOD2X5Z60@vms046.mailsrvcs.net> for pgsql-performance@postgresql.org; Thu, 14 Jul 2005 08:20:58 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by osgiliath.mathom.us (Postfix) with ESMTP id CB79F6005EA for ; Thu, 14 Jul 2005 09:20:56 -0400 (EDT) Received: from osgiliath.mathom.us ([127.0.0.1]) by localhost (osgiliath [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 14200-01-6 for ; Thu, 14 Jul 2005 09:20:56 -0400 (EDT) Received: by osgiliath.mathom.us (Postfix, from userid 1000) id B2E91600197; Thu, 14 Jul 2005 09:20:56 -0400 (EDT) Date: Thu, 14 Jul 2005 09:20:56 -0400 From: Michael Stone Subject: Re: JFS fastest filesystem for PostgreSQL? In-reply-to: <42D65778.7010500@p0f.net> To: pgsql-performance@postgresql.org Mail-Followup-To: pgsql-performance@postgresql.org Message-id: <20050714132056.GU19080@mathom.us> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-disposition: inline X-Pgp-Fingerprint: 53 FF 38 00 E7 DD 0A 9C 84 52 84 C5 EE DF 7C 88 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mathom.us References: <1121322821.6719.0.camel@noodles> <758d5e7f05071401034665841c@mail.gmail.com> <42D65778.7010500@p0f.net> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.01 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/176 X-Sequence-Number: 13417 On Thu, Jul 14, 2005 at 02:15:52PM +0200, Grega Bremec wrote: >I'm curious as to what this means - did they have problems integrating >it into their toolchain or are there actual problems going on in jfs >currently? I've found jfs to be the least stable linux filesystem and won't allow it anywhere near an important system. YMMV. Mike Stone From pgsql-performance-owner@postgresql.org Thu Jul 14 10:27:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B6E0152B9E for ; Thu, 14 Jul 2005 10:27:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87068-09 for ; Thu, 14 Jul 2005 13:27:20 +0000 (GMT) Received: from mail23.sea5.speakeasy.net (mail23.sea5.speakeasy.net [69.17.117.25]) by svr1.postgresql.org (Postfix) with ESMTP id 7B869528C6 for ; Thu, 14 Jul 2005 10:27:19 -0300 (ADT) Received: (qmail 4086 invoked from network); 14 Jul 2005 13:27:22 -0000 Received: from dsl081-060-184.sfo1.dsl.speakeasy.net (HELO noodles) ([64.81.60.184]) (envelope-sender ) by mail23.sea5.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 14 Jul 2005 13:27:22 -0000 Received: from jwb by noodles with local (Exim 3.36 #1 (Debian)) id 1Dt3kK-0002HH-00; Thu, 14 Jul 2005 06:27:24 -0700 Subject: Re: JFS fastest filesystem for PostgreSQL? From: "Jeffrey W. Baker" To: Dawid Kuroczko Cc: pgsql-performance@postgresql.org In-Reply-To: <758d5e7f05071401034665841c@mail.gmail.com> References: <1121322821.6719.0.camel@noodles> <758d5e7f05071401034665841c@mail.gmail.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Thu, 14 Jul 2005 06:27:24 -0700 Message-Id: <1121347644.8468.3.camel@noodles> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/177 X-Sequence-Number: 13418 On Thu, 2005-07-14 at 10:03 +0200, Dawid Kuroczko wrote: > On 7/14/05, Jeffrey W. Baker wrote: > > [reposted due to delivery error -jwb] > > > > I just took delivery of a new system, and used the opportunity to > > benchmark postgresql 8.0 performance on various filesystems. The system > > in question runs Linux 2.6.12, has one CPU and 1GB of system memory, and > > 5 7200RPM SATA disks attached to an Areca hardware RAID controller > > having 128MB of cache. The caches are all write-back. > > > > I ran pgbench with a scale factor of 1000 and a total of 100,000 > > transactions per run. I varied the number of clients between 10 and > > 100. It appears from my test JFS is much faster than both ext3 and XFS > > for this workload. JFS and XFS were made with the mkfs defaults. ext3 > > was made with -T largefile4 and -E stride=32. The deadline scheduler > > was used for all runs (anticipatory scheduler is much worse). > > > > Here's the result, in transactions per second. > > > > ext3 jfs xfs > > ----------------------------- > > 10 Clients 55 81 68 > > 100 Clients 61 100 64 > > ---------------------------- > > If you still have a chance, could you do tests with other journaling > options for ext3 (journal=writeback, journal=data)? And could you > give figures about performace of other IO elevators? I mean, you > wrote that anticipatory is much wore -- how much worse? :) Could > you give numbers for deadline,anticipatory,cfq elevators? :) > > And, additionally would it be possible to give numbers for bonnie++ > results? To see how does pgbench to bonnie++ relate? Phew, that's a lot of permutations. At 20-30 minutes per run, I'm thinking 5-8 hours or so. Still, for you dear readers, I'll somehow accomplish this tedious feat. As for Bonnie, JFS is a good 60-80% faster than ext3. See my message to ext3-users yesterday. Using bonnie++ with a 10GB fileset, in MB/s: ext3 jfs xfs Read 112 188 141 Write 97 157 167 Rewrite 51 71 60 -jwb From pgsql-performance-owner@postgresql.org Thu Jul 14 12:43:01 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6C21652A1E for ; Thu, 14 Jul 2005 12:42:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 22317-02 for ; Thu, 14 Jul 2005 15:42:47 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 91D7952D86 for ; Thu, 14 Jul 2005 12:42:46 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050714154245m9100k7u6ee>; Thu, 14 Jul 2005 15:42:45 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 842E555FFA; Thu, 14 Jul 2005 10:42:44 -0500 (CDT) Received: from [192.168.1.120] (65-100-20-12.cdrr.qwest.net [65.100.20.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id 4AA0055F99; Thu, 14 Jul 2005 10:42:35 -0500 (CDT) Message-ID: <42D687E5.7080202@arbash-meinel.com> Date: Thu, 14 Jul 2005 10:42:29 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> In-Reply-To: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigC9B6C8B1D8700835E847F8BE" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/178 X-Sequence-Number: 13419 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigC9B6C8B1D8700835E847F8BE Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Dan Harris wrote: > I'm trying to improve the speed of this query: > > explain select recordtext from eventactivity inner join ( select > incidentid from k_r where id = 94 ) a using ( incidentid ) inner join ( > select incidentid from k_b where id = 107 ) b using ( incidentid ); You might try giving it a little bit more freedom with: EXPLAIN ANALYZE SELECT recordtext FROM eventactivity, k_r, k_b WHERE eventactivity.incidentid = k_r.incidentid AND eventactivity.incidentid = k_b.incidentid AND k_r.id = 94 AND k_b.id = 107 -- AND k_r.incidentid = k_b.incidentid ; I'm pretty sure that would give identical results, just let the planner have a little bit more freedom about how it does it. Also the last line is commented out, because I think it is redundant. You might also try: EXPLAIN ANALYZE SELECT recordtext FROM eventactivity JOIN k_r USING (incidentid) JOIN k_b USING (incidentid) WHERE k_r.id = 94 AND k_b.id = 107 ; Also, if possible give us the EXPLAIN ANALYZE so that we know if the planner is making accurate estimates. (You might send an EXPLAIN while waiting for the EXPLAIN ANALYZE to finish) You can also try disabling merge joins, and see how that changes things. > QUERY PLAN > ------------------------------------------------------------------------ > -------------------------------------- > Merge Join (cost=2747.29..4249364.96 rows=11968693 width=35) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Merge Join (cost=1349.56..4230052.73 rows=4413563 width=117) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Index Scan using eventactivity1 on eventactivity > (cost=0.00..4051200.28 rows=44519781 width=49) > -> Sort (cost=1349.56..1350.85 rows=517 width=68) > Sort Key: (k_b.incidentid)::text > -> Index Scan using k_b_idx on k_b (cost=0.00..1326.26 > rows=517 width=68) > Index Cond: (id = 107) > -> Sort (cost=1397.73..1399.09 rows=542 width=68) > Sort Key: (k_r.incidentid)::text > -> Index Scan using k_r_idx on k_r (cost=0.00..1373.12 > rows=542 width=68) > Index Cond: (id = 94) > (13 rows) > > > There are many millions of rows in eventactivity. There are a few > ten-thousand rows in k_r and k_b. There is an index on 'incidentid' in > all three tables. There should only be less than 100 rows matched in > k_r and k_b total. That part on its own is very very fast. But, it > should have those 100 or so incidentids extracted in under a second and > then go into eventactivity AFTER doing that. At least, that's my > intention to make this fast. Well, postgres is estimating around 500 rows each, is that way off? Try just doing: EXPLAIN ANALYZE SELECT incidentid FROM k_b WHERE id = 107; EXPLAIN ANALYZE SELECT incidentid FROM k_r WHERE id = 94; And see if postgres estimates the number of rows properly. I assume you have recently VACUUM ANALYZEd, which means you might need to update the statistics target (ALTER TABLE k_b ALTER COLUMN incidientid SET STATISTICS 100) default is IIRC 10, ranges from 1-1000, higher is more accurate, but makes ANALYZE slower. > > Right now, it looks like pg is trying to sort the entire eventactivity > table for the merge join which is taking several minutes to do. Can I > rephrase this so that it does the searching through k_r and k_b FIRST > and then go into eventactivity using the index on incidentid? It seems > like that shouldn't be too hard to make fast but my SQL query skills > are only average. To me, it looks like it is doing an index scan (on k_b.id) through k_b first, sorting the results by incidentid, then merge joining that with eventactivity. I'm guessing you actually want it to merge k_b and k_r to get extra selectivity before joining against eventactivity. I think my alternate forms would let postgres realize this. But if not, you could try: EXPLAIN ANALYZE SELECT recordtext FROM eventactivity JOIN (SELECT incidentid FROM k_r JOIN k_b USING (incidentid) WHERE k_r.id = 94 AND k_b.id = 107) USING (incidentid); I don't know how selective your keys are, but one of these queries should probably structure it better for the planner. It depends a lot on how selective your query is. If you have 100M rows, the above query looks like it expects k_r to restrict it to 44M rows, and k_r + k_b down to 11M rows, which really should be a seq scan (> 10% of the rows = seq scan). But if you are saying the selectivity is mis-estimated it could be different. John =:-> > > Thanks > -Dan > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > --------------enigC9B6C8B1D8700835E847F8BE Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1ofoJdeBCYSNAAMRAvVVAKCwf0mWSGyfFRaY831W30LLDy0aNQCffkjr F3km+Kw4JmYlf8leYWOdya0= =JmgF -----END PGP SIGNATURE----- --------------enigC9B6C8B1D8700835E847F8BE-- From pgsql-performance-owner@postgresql.org Thu Jul 14 12:47:54 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 23F7552D86 for ; Thu, 14 Jul 2005 12:47:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 22829-04 for ; Thu, 14 Jul 2005 15:47:47 +0000 (GMT) Received: from mr1.surnet.cl (smtp1.surnet.cl [216.155.73.162]) by svr1.postgresql.org (Postfix) with ESMTP id 57CD452A1A for ; Thu, 14 Jul 2005 12:47:46 -0300 (ADT) Received: from smtp1.surnet.cl (216.155.73.168) by mr1.surnet.cl (7.0.031.3) id 42587EDE01566457; Thu, 14 Jul 2005 11:47:07 -0400 Received: from smtp1.surnet.cl (mr1.surnet.cl []) by mr1.surnet.cl ([216.155.73.168]); Thu, 14 Jul 2005 15:47:07 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp1.surnet.cl (7.0.031.3) id 42587C6E01070D56; Thu, 14 Jul 2005 11:47:07 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60003B9223; Thu, 14 Jul 2005 11:47:44 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 42D15C370EE; Thu, 14 Jul 2005 11:47:51 -0400 (CLT) Date: Thu, 14 Jul 2005 11:47:51 -0400 From: Alvaro Herrera To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: Quad Opteron stuck in the mud Message-ID: <20050714154750.GC19232@alvh.no-ip.org> References: <87oe95ykz5.fsf@stark.xeocode.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.377 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/179 X-Sequence-Number: 13420 On Thu, Jul 14, 2005 at 12:28:05AM -0600, Dan Harris wrote: > Ok, so I remounted this drive as ext2 shortly before sending my first > email today. It wasn't enough time for me to notice the ABSOLUTELY > HUGE difference in performance change. Ext3 must really be crappy > for postgres, or at least is on this box. Now that it's ext2, this > thing is flying like never before. My CPU utilization has > skyrocketed, telling me that the disk IO was constraining it immensely. Were you using the default journal settings for ext3? An interesting experiment would be to use the other journal options (particularly data=writeback). From the mount manpage: data=journal / data=ordered / data=writeback Specifies the journalling mode for file data. Metadata is always journaled. To use modes other than ordered on the root file system, pass the mode to the kernel as boot parameter, e.g. rootflags=data=journal. journal All data is committed into the journal prior to being written into the main file system. ordered This is the default mode. All data is forced directly out to the main file system prior to its metadata being committed to the journal. writeback Data ordering is not preserved - data may be written into the main file system after its metadata has been commit- ted to the journal. This is rumoured to be the highest- throughput option. It guarantees internal file system integrity, however it can allow old data to appear in files after a crash and journal recovery. -- Alvaro Herrera () Officer Krupke, what are we to do? Gee, officer Krupke, Krup you! (West Side Story, "Gee, Officer Krupke") From pgsql-performance-owner@postgresql.org Thu Jul 14 13:06:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 26E8352D49 for ; Thu, 14 Jul 2005 13:06:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25770-06 for ; Thu, 14 Jul 2005 16:06:00 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 2E9C252C2E for ; Thu, 14 Jul 2005 13:05:56 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 448D7644196 for ; Thu, 14 Jul 2005 10:05:32 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 07169-06 for ; Thu, 14 Jul 2005 10:05:29 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 9D08A64419F for ; Thu, 14 Jul 2005 10:05:29 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <20050714154750.GC19232@alvh.no-ip.org> References: <87oe95ykz5.fsf@stark.xeocode.com> <20050714154750.GC19232@alvh.no-ip.org> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Quad Opteron stuck in the mud Date: Thu, 14 Jul 2005 10:05:52 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/180 X-Sequence-Number: 13421 On Jul 14, 2005, at 9:47 AM, Alvaro Herrera wrote: > On Thu, Jul 14, 2005 at 12:28:05AM -0600, Dan Harris wrote: > >> . Ext3 must really be crappy >> for postgres, or at least is on this box. > > Were you using the default journal settings for ext3? Yes, I was. Next time I get a chance to reboot this box, I will try writeback and compare the benchmarks to my previous config. Thanks for the tip. From pgsql-performance-owner@postgresql.org Thu Jul 14 13:26:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DB88452D94 for ; Thu, 14 Jul 2005 13:21:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 29436-05 for ; Thu, 14 Jul 2005 16:21:12 +0000 (GMT) Received: from mr1.surnet.cl (smtp1.surnet.cl [216.155.73.162]) by svr1.postgresql.org (Postfix) with ESMTP id AC37A52D56 for ; Thu, 14 Jul 2005 13:21:11 -0300 (ADT) Received: from smtp1.surnet.cl (216.155.73.168) by mr1.surnet.cl (7.0.031.3) id 42587EDE015680B1; Thu, 14 Jul 2005 12:20:30 -0400 Received: from smtp2.surnet.cl (smtp2.surnet.cl []) by mr1.surnet.cl ([216.155.73.168]); Thu, 14 Jul 2005 16:20:30 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp2.surnet.cl (7.0.031.3) id 4259ADE300D1F941; Thu, 14 Jul 2005 12:20:50 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60003BA895; Thu, 14 Jul 2005 12:21:06 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id B2F58C370EE; Thu, 14 Jul 2005 12:21:14 -0400 (CLT) Date: Thu, 14 Jul 2005 12:21:14 -0400 From: Alvaro Herrera To: Alison Winters Cc: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table Message-ID: <20050714162114.GA19778@alvh.no-ip.org> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.378 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/181 X-Sequence-Number: 13422 On Thu, Jul 14, 2005 at 03:08:30PM +1000, Alison Winters wrote: > Hi, > > Our application requires a number of processes to select and update rows > from a very small (<10 rows) Postgres table on a regular and frequent > basis. These processes often run for weeks at a time, but over the > space of a few days we find that updates start getting painfully slow. > We are running a full vacuum/analyze and reindex on the table every day, Full vacuum, eh? I wonder if what you really need is very frequent non-full vacuum. Say, once in 15 minutes (exact rate depending on dead tuple rate.) -- Alvaro Herrera () "World domination is proceeding according to plan" (Andrew Morton) From pgsql-performance-owner@postgresql.org Thu Jul 14 13:41:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 92A9752B5B for ; Thu, 14 Jul 2005 13:41:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34498-06 for ; Thu, 14 Jul 2005 16:41:37 +0000 (GMT) Received: from twiddle.look.ca (beta1.look.ca [207.136.80.125]) by svr1.postgresql.org (Postfix) with ESMTP id 9609452AA1 for ; Thu, 14 Jul 2005 13:41:36 -0300 (ADT) Received: from 209-161-193-204.dsl.look.ca ([209.161.193.204]) by twiddle.look.ca with esmtp (Exim 4.20) id 1Dt6mE-0000fS-RR; Thu, 14 Jul 2005 16:41:34 +0000 From: Rod Taylor To: Alison Winters Cc: pgsql-performance@postgresql.org In-Reply-To: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> Date: Thu, 14 Jul 2005 12:37:42 -0400 Message-Id: <1121359062.780.31.camel@home> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 FreeBSD GNOME Team Port X-SA-Exim-Mail-From: pg@rbt.ca Subject: Re: lots of updates on small table Content-Type: text/plain Content-Transfer-Encoding: 7bit X-SA-Exim-Version: 3.1 (built Tue Feb 24 05:09:27 GMT 2004) X-SA-Exim-Scanned: Yes X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.041 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/182 X-Sequence-Number: 13423 On Thu, 2005-07-14 at 15:08 +1000, Alison Winters wrote: > Hi, > > Our application requires a number of processes to select and update rows > from a very small (<10 rows) Postgres table on a regular and frequent > basis. These processes often run for weeks at a time, but over the Are these long running transactions or is the process issuing many short transactions? If your transaction lasts a week, then a daily vacuum isn't really doing anything. I presume you also run ANALYZE in some shape or form periodically? > space of a few days we find that updates start getting painfully slow. > We are running a full vacuum/analyze and reindex on the table every day, If they're short transactions, run vacuum (not vacuum full) every 100 or so updates. This might even be once a minute. Analyze periodically as well. -- From pgsql-performance-owner@postgresql.org Thu Jul 14 15:27:01 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 78B5952C22 for ; Thu, 14 Jul 2005 15:26:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 63775-03 for ; Thu, 14 Jul 2005 18:26:54 +0000 (GMT) Received: from flake.decibel.org (flake.decibel.org [67.100.216.10]) by svr1.postgresql.org (Postfix) with ESMTP id 8556252C02 for ; Thu, 14 Jul 2005 15:26:53 -0300 (ADT) Received: by flake.decibel.org (Postfix, from userid 1001) id 3634015234; Thu, 14 Jul 2005 13:26:53 -0500 (CDT) Date: Thu, 14 Jul 2005 13:26:52 -0500 From: "Jim C. Nasby" To: Nicolas Beaume Cc: pgsql-performance@postgresql.org Subject: Re: large table vs multiple smal tables Message-ID: <20050714182651.GB92165@decibel.org> References: <42D4E836.9020607@univ-nantes.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42D4E836.9020607@univ-nantes.fr> X-Operating-System: FreeBSD 4.11-RELEASE-p10 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/183 X-Sequence-Number: 13424 On Wed, Jul 13, 2005 at 12:08:54PM +0200, Nicolas Beaume wrote: > Hello > > I have a large database with 4 large tables (each containing at least > 200 000 rows, perhaps even 1 or 2 million) and i ask myself if it's > better to split them into small tables (e.g tables of 2000 rows) to > speed the access and the update of those tables (considering that i will > have few update but a lot of reading). 2 million rows is nothing unless you're on a 486 or something. As for your other question, remember the first rule of performance tuning: don't tune unless you actually need to. -- Jim C. Nasby, Database Consultant decibel@decibel.org Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Thu Jul 14 15:29:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 987DF52C32 for ; Thu, 14 Jul 2005 15:29:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 60982-09 for ; Thu, 14 Jul 2005 18:29:56 +0000 (GMT) Received: from flake.decibel.org (flake.decibel.org [67.100.216.10]) by svr1.postgresql.org (Postfix) with ESMTP id CA42252C2D for ; Thu, 14 Jul 2005 15:29:54 -0300 (ADT) Received: by flake.decibel.org (Postfix, from userid 1001) id 22AC715234; Thu, 14 Jul 2005 13:29:48 -0500 (CDT) Date: Thu, 14 Jul 2005 13:29:48 -0500 From: "Jim C. Nasby" To: "Jeffrey W. Baker" Cc: pgsql-performance@postgresql.org Subject: Re: JFS fastest filesystem for PostgreSQL? Message-ID: <20050714182948.GC92165@decibel.org> References: <1121322821.6719.0.camel@noodles> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1121322821.6719.0.camel@noodles> X-Operating-System: FreeBSD 4.11-RELEASE-p10 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/184 X-Sequence-Number: 13425 On Wed, Jul 13, 2005 at 11:33:41PM -0700, Jeffrey W. Baker wrote: > [reposted due to delivery error -jwb] > > I just took delivery of a new system, and used the opportunity to > benchmark postgresql 8.0 performance on various filesystems. The system > in question runs Linux 2.6.12, has one CPU and 1GB of system memory, and > 5 7200RPM SATA disks attached to an Areca hardware RAID controller > having 128MB of cache. The caches are all write-back. > > I ran pgbench with a scale factor of 1000 and a total of 100,000 > transactions per run. I varied the number of clients between 10 and > 100. It appears from my test JFS is much faster than both ext3 and XFS > for this workload. JFS and XFS were made with the mkfs defaults. ext3 > was made with -T largefile4 and -E stride=32. The deadline scheduler > was used for all runs (anticipatory scheduler is much worse). > > Here's the result, in transactions per second. > > ext3 jfs xfs > ----------------------------- > 10 Clients 55 81 68 > 100 Clients 61 100 64 > ---------------------------- BTW, it'd be interesting to see how UFS on FreeBSD compared. -- Jim C. Nasby, Database Consultant decibel@decibel.org Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Thu Jul 14 15:37:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5E25652C22 for ; Thu, 14 Jul 2005 15:37:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65061-06 for ; Thu, 14 Jul 2005 18:37:14 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 755E252C02 for ; Thu, 14 Jul 2005 15:37:13 -0300 (ADT) Received: from [192.168.1.52] (fc1smp [66.93.38.87]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6EIbAoQ030071; Thu, 14 Jul 2005 11:37:10 -0700 Message-ID: <42D6B0D6.6000101@commandprompt.com> Date: Thu, 14 Jul 2005 11:37:10 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Jim C. Nasby" Cc: "Jeffrey W. Baker" , pgsql-performance@postgresql.org Subject: Re: JFS fastest filesystem for PostgreSQL? References: <1121322821.6719.0.camel@noodles> <20050714182948.GC92165@decibel.org> In-Reply-To: <20050714182948.GC92165@decibel.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.014 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/185 X-Sequence-Number: 13426 >>I ran pgbench with a scale factor of 1000 and a total of 100,000 >>transactions per run. I varied the number of clients between 10 and >>100. It appears from my test JFS is much faster than both ext3 and XFS >>for this workload. JFS and XFS were made with the mkfs defaults. ext3 >>was made with -T largefile4 and -E stride=32. The deadline scheduler >>was used for all runs (anticipatory scheduler is much worse). >> >>Here's the result, in transactions per second. >> >> ext3 jfs xfs >>----------------------------- >> 10 Clients 55 81 68 >>100 Clients 61 100 64 >>---------------------------- I would be curious as to what options were passed to jfs and xfs. Sincerely, Joshua D. Drake > > > BTW, it'd be interesting to see how UFS on FreeBSD compared. -- Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 PostgreSQL Replication, Consulting, Custom Programming, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ From pgsql-performance-owner@postgresql.org Thu Jul 14 17:47:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3F4AD52C41 for ; Thu, 14 Jul 2005 17:47:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93472-03 for ; Thu, 14 Jul 2005 20:47:10 +0000 (GMT) Received: from stark.xeocode.com (stark.xeocode.com [216.58.44.227]) by svr1.postgresql.org (Postfix) with ESMTP id B97BF52AA1 for ; Thu, 14 Jul 2005 17:47:08 -0300 (ADT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1DtAbn-0008IM-00; Thu, 14 Jul 2005 16:47:03 -0400 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: Quad Opteron stuck in the mud References: <87oe95ykz5.fsf@stark.xeocode.com> In-Reply-To: From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 14 Jul 2005 16:47:02 -0400 Message-ID: <87k6jtw1x5.fsf@stark.xeocode.com> Lines: 61 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/186 X-Sequence-Number: 13427 Dan Harris writes: > Well, once every day, but there aren't a ton of inserts or updates going on a > daily basis. Maybe 1,000 total inserts? It's actually deletes and updates that matter. not inserts. > I have a feeling I'm going to need to do a cluster soon. I have done several > mass deletes and reloads on it. CLUSTER effectively does a VACUUM FULL but takes a different approach and writes out a whole new table, which if there's lots of free space is faster than moving records around to compact the table. > I tried that, and indeed it was using an index, although after reading Simon's > post, I realize that was kind of dumb to have an index on a bool. I have since > removed it. If there are very few records (like well under 10%) with that column equal to false (or very few equal to true) then it's not necessarily useless. But probably more useful is a partial index on some other column. Something like CREATE INDEX ON pk WHERE flag = false; > No foreign keys or triggers. Note that I'm talking about foreign keys in *other* tables that refer to columns in this table. Every update on this table would have to scan those other tables looking for records referencing the updated rows. > Ok, so I remounted this drive as ext2 shortly before sending my first email > today. It wasn't enough time for me to notice the ABSOLUTELY HUGE difference > in performance change. Ext3 must really be crappy for postgres, or at least > is on this box. Now that it's ext2, this thing is flying like never before. > My CPU utilization has skyrocketed, telling me that the disk IO was > constraining it immensely. > > I always knew that it might be a little faster, but the box feels like it can > "breathe" again and things that used to be IO intensive and run for an hour or > more are now running in < 5 minutes. I'm a little worried about not having a > journalized file system, but that performance difference will keep me from > switching back ( at least to ext3! ). Maybe someday I will try XFS. @spock(Fascinating). I wonder if ext3 might be issuing IDE cache flushes on every fsync (to sync the journal) whereas ext2 might not be issuing any cache flushes at all. If the IDE cache is never being flushed then you'll see much better performance but run the risk of data loss in a power failure or hardware failure. (But not in the case of an OS crash, or at least no more than otherwise.) You could also try using the "-O journal_dev" option to put the ext3 journal on a separate device. -- greg From pgsql-performance-owner@postgresql.org Thu Jul 14 17:48:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 53BFC52C91 for ; Thu, 14 Jul 2005 17:48:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93152-06 for ; Thu, 14 Jul 2005 20:48:08 +0000 (GMT) Received: from cmailm2.svr.pol.co.uk (cmailm2.svr.pol.co.uk [195.92.193.210]) by svr1.postgresql.org (Postfix) with ESMTP id 504EE52C41 for ; Thu, 14 Jul 2005 17:48:07 -0300 (ADT) Received: from modem-1171.leopard.dialup.pol.co.uk ([217.135.148.147] helo=192.168.0.102) by cmailm2.svr.pol.co.uk with esmtp (Exim 4.41) id 1DtAcj-0002Wl-7O; Thu, 14 Jul 2005 21:48:01 +0100 Subject: Re: Profiler for PostgreSQL From: Simon Riggs To: Christopher Kings-Lynne Cc: Agha Asif Raza , pgsql-performance@postgresql.org In-Reply-To: <42D60664.90400@familyhealth.com.au> References: <8d1c715f05071322582a27b996@mail.gmail.com> <42D60664.90400@familyhealth.com.au> Content-Type: text/plain Organization: 2nd Quadrant Date: Thu, 14 Jul 2005 21:47:46 +0100 Message-Id: <1121374066.3970.450.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/187 X-Sequence-Number: 13428 On Thu, 2005-07-14 at 14:29 +0800, Christopher Kings-Lynne wrote: > Try turning on query logging and using the 'pqa' utility on pgfoundry.org. Have you got that to work for 8 ? pqa 1.5 doesn't even work with its own test file. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Thu Jul 14 19:30:10 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8E0A752BCA for ; Thu, 14 Jul 2005 19:30:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 13632-07 for ; Thu, 14 Jul 2005 22:30:05 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 721195287D for ; Thu, 14 Jul 2005 19:30:01 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 36E55644159 for ; Thu, 14 Jul 2005 16:29:39 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 10917-07 for ; Thu, 14 Jul 2005 16:29:36 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 02D4F644151 for ; Thu, 14 Jul 2005 16:29:35 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <42D687E5.7080202@arbash-meinel.com> References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: slow joining very large table to smaller ones Date: Thu, 14 Jul 2005 16:29:58 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/188 X-Sequence-Number: 13429 On Jul 14, 2005, at 9:42 AM, John A Meinel wrote: > > > You might try giving it a little bit more freedom with: > > EXPLAIN ANALYZE > SELECT recordtext FROM eventactivity, k_r, k_b > WHERE eventactivity.incidentid = k_r.incidentid > AND eventactivity.incidentid = k_b.incidentid > AND k_r.id = 94 > AND k_b.id = 107 > -- AND k_r.incidentid = k_b.incidentid > ; > > I'm pretty sure that would give identical results, just let the > planner > have a little bit more freedom about how it does it. > Also the last line is commented out, because I think it is redundant. > Ok, I tried this one. My ssh keeps getting cut off by a router somewhere between me and the server due to inactivity timeouts, so all I know is that both the select and explain analyze are taking over an hour to run. Here's the explain select for that one, since that's the best I can get. explain select recordtext from eventactivity,k_r,k_b where eventactivity.incidentid = k_r.incidentid and eventactivity.incidentid = k_b.incidentid and k_r.id = 94 and k_b.id = 107; QUERY PLAN ------------------------------------------------------------------------ -------------------------------------- Merge Join (cost=9624.61..4679590.52 rows=151009549 width=35) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Merge Join (cost=4766.92..4547684.26 rows=16072733 width=117) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Index Scan using eventactivity1 on eventactivity (cost=0.00..4186753.16 rows=46029271 width=49) -> Sort (cost=4766.92..4771.47 rows=1821 width=68) Sort Key: (k_b.incidentid)::text -> Index Scan using k_b_idx on k_b (cost=0.00..4668.31 rows=1821 width=68) Index Cond: (id = 107) -> Sort (cost=4857.69..4862.39 rows=1879 width=68) Sort Key: (k_r.incidentid)::text -> Index Scan using k_r_idx on k_r (cost=0.00..4755.52 rows=1879 width=68) Index Cond: (id = 94) (13 rows) > You might also try: > EXPLAIN ANALYZE > SELECT recordtext > FROM eventactivity JOIN k_r USING (incidentid) > JOIN k_b USING (incidentid) > WHERE k_r.id = 94 > AND k_b.id = 107 > ; > Similar results here. The query is taking at least an hour to finish. explain select recordtext from eventactivity join k_r using ( incidentid ) join k_b using (incidentid ) where k_r.id = 94 and k_b.id = 107; QUERY PLAN ------------------------------------------------------------------------ -------------------------------------- Merge Join (cost=9542.77..4672831.12 rows=148391132 width=35) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Merge Join (cost=4726.61..4542825.87 rows=15930238 width=117) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Index Scan using eventactivity1 on eventactivity (cost=0.00..4184145.43 rows=46000104 width=49) -> Sort (cost=4726.61..4731.13 rows=1806 width=68) Sort Key: (k_b.incidentid)::text -> Index Scan using k_b_idx on k_b (cost=0.00..4628.92 rows=1806 width=68) Index Cond: (id = 107) -> Sort (cost=4816.16..4820.82 rows=1863 width=68) Sort Key: (k_r.incidentid)::text -> Index Scan using k_r_idx on k_r (cost=0.00..4714.97 rows=1863 width=68) Index Cond: (id = 94) (13 rows) > Also, if possible give us the EXPLAIN ANALYZE so that we know if the > planner is making accurate estimates. (You might send an EXPLAIN while > waiting for the EXPLAIN ANALYZE to finish) > > You can also try disabling merge joins, and see how that changes > things. > Are there any negative sideaffects of doing this? > >> >> > > Well, postgres is estimating around 500 rows each, is that way off? > Try > just doing: > EXPLAIN ANALYZE SELECT incidentid FROM k_b WHERE id = 107; > EXPLAIN ANALYZE SELECT incidentid FROM k_r WHERE id = 94; > > And see if postgres estimates the number of rows properly. > > I assume you have recently VACUUM ANALYZEd, which means you might need > to update the statistics target (ALTER TABLE k_b ALTER COLUMN > incidientid SET STATISTICS 100) default is IIRC 10, ranges from > 1-1000, > higher is more accurate, but makes ANALYZE slower. > > >> >> Right now, it looks like pg is trying to sort the entire >> eventactivity >> table for the merge join which is taking several minutes to do. >> Can I >> rephrase this so that it does the searching through k_r and k_b >> FIRST >> and then go into eventactivity using the index on incidentid? It >> seems >> like that shouldn't be too hard to make fast but my SQL query skills >> are only average. >> > > To me, it looks like it is doing an index scan (on k_b.id) through k_b > first, sorting the results by incidentid, then merge joining that with > eventactivity. > > I'm guessing you actually want it to merge k_b and k_r to get extra > selectivity before joining against eventactivity. > I think my alternate forms would let postgres realize this. But if > not, > you could try: > > EXPLAIN ANALYZE > SELECT recordtext FROM eventactivity > JOIN (SELECT incidentid FROM k_r JOIN k_b USING (incidentid) > WHERE k_r.id = 94 AND k_b.id = 107) > USING (incidentid); > This one looks like the same plan as the others: explain select recordtext from eventactivity join ( select incidentid from k_r join k_b using (incidentid) where k_r.id = 94 and k_b.id = 107 ) a using (incidentid ); QUERY PLAN ------------------------------------------------------------------------ -------------------------------------- Merge Join (cost=9793.33..4693149.15 rows=156544758 width=35) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Merge Join (cost=4847.75..4557237.59 rows=16365843 width=117) Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") -> Index Scan using eventactivity1 on eventactivity (cost=0.00..4191691.79 rows=46084161 width=49) -> Sort (cost=4847.75..4852.38 rows=1852 width=68) Sort Key: (k_b.incidentid)::text -> Index Scan using k_b_idx on k_b (cost=0.00..4747.24 rows=1852 width=68) Index Cond: (id = 107) -> Sort (cost=4945.58..4950.36 rows=1913 width=68) Sort Key: (k_r.incidentid)::text -> Index Scan using k_r_idx on k_r (cost=0.00..4841.30 rows=1913 width=68) Index Cond: (id = 94) (13 rows) > I don't know how selective your keys are, but one of these queries > should probably structure it better for the planner. It depends a > lot on > how selective your query is. eventactivity currently has around 36 million rows in it. There should only be maybe 200-300 incidentids at most that will be matched with the combination of k_b and k_r. That's why I was thinking I could somehow get a list of just the incidentids that matched the id = 94 and id = 107 in k_b and k_r first. Then, I would only need to grab a few hundred out of 36 million rows from eventactivity. > If you have 100M rows, the above query looks like it expects k_r to > restrict it to 44M rows, and k_r + k_b down to 11M rows, which really > should be a seq scan (> 10% of the rows = seq scan). But if you are > saying the selectivity is mis-estimated it could be different. > Yeah, if I understand you correctly, I think the previous paragraph shows this is a significant misestimate. From pgsql-performance-owner@postgresql.org Thu Jul 14 19:46:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A694D5287D for ; Thu, 14 Jul 2005 19:46:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 19756-03 for ; Thu, 14 Jul 2005 22:46:23 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id 64DA652814 for ; Thu, 14 Jul 2005 19:46:22 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050714224623m920006dife>; Thu, 14 Jul 2005 22:46:23 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id AD53D55FF0; Thu, 14 Jul 2005 17:46:22 -0500 (CDT) Received: from [192.168.1.102] (65-100-20-12.cdrr.qwest.net [65.100.20.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id C534955FF0; Thu, 14 Jul 2005 17:46:16 -0500 (CDT) Message-ID: <42D6EB31.9040909@arbash-meinel.com> Date: Thu, 14 Jul 2005 17:46:09 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig934AD7077F7DA9AB52E83521" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.027 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/189 X-Sequence-Number: 13430 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig934AD7077F7DA9AB52E83521 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Dan Harris wrote: > > On Jul 14, 2005, at 9:42 AM, John A Meinel wrote: ... Did you try doing this to see how good the planners selectivity estimates are? >> Well, postgres is estimating around 500 rows each, is that way off? Try >> just doing: >> EXPLAIN ANALYZE SELECT incidentid FROM k_b WHERE id = 107; >> EXPLAIN ANALYZE SELECT incidentid FROM k_r WHERE id = 94; These should be fast queries. John =:-> --------------enig934AD7077F7DA9AB52E83521 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1usxJdeBCYSNAAMRAk5XAKCFggsfmcdSCEGUkeqRYx8M2+Q0mQCg2XEr dWjvGIfwMNGw4Szff32o9gM= =6zRN -----END PGP SIGNATURE----- --------------enig934AD7077F7DA9AB52E83521-- From pgsql-performance-owner@postgresql.org Thu Jul 14 19:47:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 049A552D82 for ; Thu, 14 Jul 2005 19:47:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18602-08 for ; Thu, 14 Jul 2005 22:47:49 +0000 (GMT) Received: from vms048pub.verizon.net (vms048pub.verizon.net [206.46.252.48]) by svr1.postgresql.org (Postfix) with ESMTP id 2741A52D73 for ; Thu, 14 Jul 2005 19:47:45 -0300 (ADT) Received: from osgiliath.mathom.us ([70.108.75.132]) by vms048.mailsrvcs.net (Sun Java System Messaging Server 6.2 HotFix 0.04 (built Dec 24 2004)) with ESMTPA id <0IJN00D2C3BNFKJ1@vms048.mailsrvcs.net> for pgsql-performance@postgresql.org; Thu, 14 Jul 2005 17:47:48 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by osgiliath.mathom.us (Postfix) with ESMTP id 72BCA6005EA; Thu, 14 Jul 2005 18:47:47 -0400 (EDT) Received: from osgiliath.mathom.us ([127.0.0.1]) by localhost (osgiliath [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 24705-01; Thu, 14 Jul 2005 18:47:47 -0400 (EDT) Received: by osgiliath.mathom.us (Postfix, from userid 1000) id 14ED3600197; Thu, 14 Jul 2005 18:47:47 -0400 (EDT) Date: Thu, 14 Jul 2005 18:47:46 -0400 From: Michael Stone Subject: Re: slow joining very large table to smaller ones In-reply-to: To: Dan Harris Cc: pgsql-performance@postgresql.org Mail-Followup-To: Dan Harris , pgsql-performance@postgresql.org Message-id: <20050714224746.GD19080@mathom.us> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-disposition: inline X-Pgp-Fingerprint: 53 FF 38 00 E7 DD 0A 9C 84 52 84 C5 EE DF 7C 88 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mathom.us References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/190 X-Sequence-Number: 13431 On Thu, Jul 14, 2005 at 04:29:58PM -0600, Dan Harris wrote: >Ok, I tried this one. My ssh keeps getting cut off by a router >somewhere between me and the server due to inactivity timeouts, so >all I know is that both the select and explain analyze are taking >over an hour to run. Try running the query as a script with nohup & redirect the output to a file. Mike Stone From pgsql-performance-owner@postgresql.org Thu Jul 14 20:08:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9875252C60 for ; Thu, 14 Jul 2005 20:08:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 21401-06 for ; Thu, 14 Jul 2005 23:08:42 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 14B2A52C5A for ; Thu, 14 Jul 2005 20:08:41 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6EN8hiU005546; Thu, 14 Jul 2005 19:08:43 -0400 (EDT) To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones In-reply-to: References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> Comments: In-reply-to Dan Harris message dated "Thu, 14 Jul 2005 16:29:58 -0600" Date: Thu, 14 Jul 2005 19:08:43 -0400 Message-ID: <5545.1121382523@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/191 X-Sequence-Number: 13432 Dan Harris writes: > Here's the explain select for that one, since > that's the best I can get. > explain select recordtext from eventactivity,k_r,k_b where > eventactivity.incidentid = k_r.incidentid and > eventactivity.incidentid = k_b.incidentid and k_r.id = 94 and k_b.id > = 107; > QUERY PLAN > ------------------------------------------------------------------------ > -------------------------------------- > Merge Join (cost=9624.61..4679590.52 rows=151009549 width=35) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Merge Join (cost=4766.92..4547684.26 rows=16072733 width=117) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Index Scan using eventactivity1 on eventactivity > (cost=0.00..4186753.16 rows=46029271 width=49) > -> Sort (cost=4766.92..4771.47 rows=1821 width=68) > Sort Key: (k_b.incidentid)::text > -> Index Scan using k_b_idx on k_b > (cost=0.00..4668.31 rows=1821 width=68) > Index Cond: (id = 107) > -> Sort (cost=4857.69..4862.39 rows=1879 width=68) > Sort Key: (k_r.incidentid)::text > -> Index Scan using k_r_idx on k_r (cost=0.00..4755.52 > rows=1879 width=68) > Index Cond: (id = 94) > (13 rows) There's something awfully fishy here. The 8.0 planner is definitely capable of figuring out that it ought to join the smaller relations first. As an example, using 8.0.3+ (CVS branch tip) I did regression=# create table eventactivity(incidentid varchar, recordtext text); CREATE TABLE regression=# create table k_r(incidentid varchar); CREATE TABLE regression=# create table k_b(incidentid varchar); CREATE TABLE regression=# explain select recordtext from eventactivity inner join (select incidentid from k_r) a using (incidentid) inner join (select incidentid from k_b) b using (incidentid); (Being too impatient to actually fill the eventactivity table with 36M rows of data, I just did some debugger magic to make the planner think that that was the table size...) The default plan looks like Merge Join (cost=16137814.70..36563453.23 rows=1361700000 width=32) Merge Cond: (("outer".incidentid)::text = "inner"."?column3?") -> Merge Join (cost=170.85..290.48 rows=7565 width=64) Merge Cond: ("outer"."?column2?" = "inner"."?column2?") -> Sort (cost=85.43..88.50 rows=1230 width=32) Sort Key: (k_r.incidentid)::text -> Seq Scan on k_r (cost=0.00..22.30 rows=1230 width=32) -> Sort (cost=85.43..88.50 rows=1230 width=32) Sort Key: (k_b.incidentid)::text -> Seq Scan on k_b (cost=0.00..22.30 rows=1230 width=32) -> Sort (cost=16137643.84..16227643.84 rows=36000000 width=64) Sort Key: (eventactivity.incidentid)::text -> Seq Scan on eventactivity (cost=0.00..1080000.00 rows=36000000 width=64) and if I "set enable_mergejoin TO 0;" I get Hash Join (cost=612.54..83761451.54 rows=1361700000 width=32) Hash Cond: (("outer".incidentid)::text = ("inner".incidentid)::text) -> Seq Scan on eventactivity (cost=0.00..1080000.00 rows=36000000 width=64) -> Hash (cost=504.62..504.62 rows=7565 width=64) -> Hash Join (cost=25.38..504.62 rows=7565 width=64) Hash Cond: (("outer".incidentid)::text = ("inner".incidentid)::text) -> Seq Scan on k_r (cost=0.00..22.30 rows=1230 width=32) -> Hash (cost=22.30..22.30 rows=1230 width=32) -> Seq Scan on k_b (cost=0.00..22.30 rows=1230 width=32) which is the plan I would judge Most Likely To Succeed based on what we know about Dan's problem. (The fact that the planner is estimating it as twice as expensive as the mergejoin comes from the fact that with no statistics about the join keys, the planner deliberately estimates hash join as expensive, because it can be pretty awful in the presence of many equal keys.) So the planner is certainly capable of finding the desired plan, even without any tweaking of the query text. This means that what we have is mainly a statistical problem. Have you ANALYZEd these tables recently? If so, may we see the pg_stats rows for incidentid in all three tables? regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 14 20:14:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B7EC552C6C for ; Thu, 14 Jul 2005 20:14:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 23979-04 for ; Thu, 14 Jul 2005 23:13:52 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 177DC52C5C for ; Thu, 14 Jul 2005 20:13:51 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050714231353m9100k91dpe>; Thu, 14 Jul 2005 23:13:54 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 514AF55FFC; Thu, 14 Jul 2005 18:13:53 -0500 (CDT) Received: from [192.168.1.102] (65-100-20-12.cdrr.qwest.net [65.100.20.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id C57A355FF0; Thu, 14 Jul 2005 18:13:38 -0500 (CDT) Message-ID: <42D6F177.3000002@arbash-meinel.com> Date: Thu, 14 Jul 2005 18:12:55 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigC423413175948A4F4F847397" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.028 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/192 X-Sequence-Number: 13433 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigC423413175948A4F4F847397 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Dan Harris wrote: > > On Jul 14, 2005, at 9:42 AM, John A Meinel wrote: > >> >> >> You might try giving it a little bit more freedom with: >> >> EXPLAIN ANALYZE >> SELECT recordtext FROM eventactivity, k_r, k_b >> WHERE eventactivity.incidentid = k_r.incidentid >> AND eventactivity.incidentid = k_b.incidentid >> AND k_r.id = 94 >> AND k_b.id = 107 >> -- AND k_r.incidentid = k_b.incidentid >> ; >> >> I'm pretty sure that would give identical results, just let the planner >> have a little bit more freedom about how it does it. >> Also the last line is commented out, because I think it is redundant. >> > > Ok, I tried this one. My ssh keeps getting cut off by a router > somewhere between me and the server due to inactivity timeouts, so all > I know is that both the select and explain analyze are taking over an > hour to run. Here's the explain select for that one, since that's the > best I can get. > > explain select recordtext from eventactivity,k_r,k_b where > eventactivity.incidentid = k_r.incidentid and eventactivity.incidentid > = k_b.incidentid and k_r.id = 94 and k_b.id = 107; > QUERY PLAN > ------------------------------------------------------------------------ > -------------------------------------- > Merge Join (cost=9624.61..4679590.52 rows=151009549 width=35) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Merge Join (cost=4766.92..4547684.26 rows=16072733 width=117) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Index Scan using eventactivity1 on eventactivity > (cost=0.00..4186753.16 rows=46029271 width=49) > -> Sort (cost=4766.92..4771.47 rows=1821 width=68) > Sort Key: (k_b.incidentid)::text > -> Index Scan using k_b_idx on k_b (cost=0.00..4668.31 > rows=1821 width=68) > Index Cond: (id = 107) > -> Sort (cost=4857.69..4862.39 rows=1879 width=68) > Sort Key: (k_r.incidentid)::text > -> Index Scan using k_r_idx on k_r (cost=0.00..4755.52 > rows=1879 width=68) > Index Cond: (id = 94) > (13 rows) > If anything, the estimations have gotten worse. As now it thinks there will be 1800 rows returned each, whereas you were thinking it would be more around 100. Since you didn't say, you did VACUUM ANALYZE recently, right? > ... >> >> You can also try disabling merge joins, and see how that changes things. >> > > Are there any negative sideaffects of doing this? If the planner is estimating things correctly, you want to give it the most flexibility of plans to pick from, because sometimes a merge join is faster (postgres doesn't pick things because it wants to go slower). The only reason for the disable flags is that sometimes the planner doesn't estimate correctly. Usually disabling a method is not the final solution, but a way to try out different methods, and see what happens to the results. Using: SET enable_mergejoin TO off; You can disable it just for the current session (not for the entire database). Which is the recommended way if you have a query that postgres is messing up on. (Usually it is correct elsewhere). >> >> Well, postgres is estimating around 500 rows each, is that way off? Try >> just doing: >> EXPLAIN ANALYZE SELECT incidentid FROM k_b WHERE id = 107; >> EXPLAIN ANALYZE SELECT incidentid FROM k_r WHERE id = 94; Once again, do this and post the results. We might just need to tweak your settings so that it estimates the number of rows correctly, and we don't need to do anything else. >> >> And see if postgres estimates the number of rows properly. >> >> I assume you have recently VACUUM ANALYZEd, which means you might need >> to update the statistics target (ALTER TABLE k_b ALTER COLUMN >> incidientid SET STATISTICS 100) default is IIRC 10, ranges from 1-1000, >> higher is more accurate, but makes ANALYZE slower. >> ... >> EXPLAIN ANALYZE >> SELECT recordtext FROM eventactivity >> JOIN (SELECT incidentid FROM k_r JOIN k_b USING (incidentid) >> WHERE k_r.id = 94 AND k_b.id = 107) >> USING (incidentid); >> > > This one looks like the same plan as the others: > > explain select recordtext from eventactivity join ( select incidentid > from k_r join k_b using (incidentid) where k_r.id = 94 and k_b.id = 107 > ) a using (incidentid ); Well, the planner is powerful enough to flatten nested selects. To make it less "intelligent" you can do: SET join_collapse_limit 1; or SET join_collapse_limit 0; Which should tell postgres to not try and get tricky with your query. Again, *usually* the planner knows better than you do. So again just do it to see what you get. The problem is that if you are only using EXPLAIN SELECT, you will probably get something which *looks* worse. Because if it looked better, the planner would have used it. That is why you really need the EXPLAIN ANALYZE, so that you can see where the planner is incorrect in it's estimates. > QUERY PLAN > ------------------------------------------------------------------------ > -------------------------------------- > Merge Join (cost=9793.33..4693149.15 rows=156544758 width=35) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Merge Join (cost=4847.75..4557237.59 rows=16365843 width=117) > Merge Cond: (("outer".incidentid)::text = "inner"."?column2?") > -> Index Scan using eventactivity1 on eventactivity > (cost=0.00..4191691.79 rows=46084161 width=49) > -> Sort (cost=4847.75..4852.38 rows=1852 width=68) > Sort Key: (k_b.incidentid)::text > -> Index Scan using k_b_idx on k_b (cost=0.00..4747.24 > rows=1852 width=68) > Index Cond: (id = 107) > -> Sort (cost=4945.58..4950.36 rows=1913 width=68) > Sort Key: (k_r.incidentid)::text > -> Index Scan using k_r_idx on k_r (cost=0.00..4841.30 > rows=1913 width=68) > Index Cond: (id = 94) > (13 rows) > What I don't understand is that the planner is actually estimating that joining against the new table is going to *increase* the number of returned rows. Because the final number of rows here is 156M while the initial join shows only 16M. And it also thinks that it will only grab 46M rows from eventactivity. If you have analyzed recently can you do: SELECT relname, reltuples FROM pg_class WHERE relname='eventactivity'; It is a cheaper form than "SELECT count(*) FROM eventactivity" to get an approximate estimate of the number of rows. But if it isn't too expensive, please also give the value from SELECT count(*) FROM eventactivity. Again, that helps us know if your tables are up-to-date. > > >> I don't know how selective your keys are, but one of these queries >> should probably structure it better for the planner. It depends a lot on >> how selective your query is. > > > eventactivity currently has around 36 million rows in it. There should > only be maybe 200-300 incidentids at most that will be matched with the > combination of k_b and k_r. That's why I was thinking I could somehow > get a list of just the incidentids that matched the id = 94 and id = > 107 in k_b and k_r first. Then, I would only need to grab a few hundred > out of 36 million rows from eventactivity. > Well, you can also try: SELECT count(*) FROM k_b JOIN k_r USING (incidentid) WHERE k_b.id=?? AND k_r.id=?? ; That will tell you how many rows they have in common. >> If you have 100M rows, the above query looks like it expects k_r to >> restrict it to 44M rows, and k_r + k_b down to 11M rows, which really >> should be a seq scan (> 10% of the rows = seq scan). But if you are >> saying the selectivity is mis-estimated it could be different. >> > > Yeah, if I understand you correctly, I think the previous paragraph > shows this is a significant misestimate. > Well, if you look at the latest plans, things have gone up from 44M to 156M, I don't know why it is worse, but it is getting there. John =:-> --------------enigC423413175948A4F4F847397 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1vF3JdeBCYSNAAMRAgXCAKCXgpSixITwsvr8NvXZKTcmXyprnQCdHwQR uTYwazpzK/ZyRRlx2+TLKxM= =7UD2 -----END PGP SIGNATURE----- --------------enigC423413175948A4F4F847397-- From pgsql-performance-owner@postgresql.org Thu Jul 14 20:30:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A457152B0D for ; Thu, 14 Jul 2005 20:30:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 24847-08 for ; Thu, 14 Jul 2005 23:30:05 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 8CA5752C63 for ; Thu, 14 Jul 2005 20:30:03 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6ENU2Zr005709; Thu, 14 Jul 2005 19:30:02 -0400 (EDT) To: John A Meinel Cc: Dan Harris , pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones In-reply-to: <42D6F177.3000002@arbash-meinel.com> References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> Comments: In-reply-to John A Meinel message dated "Thu, 14 Jul 2005 18:12:55 -0500" Date: Thu, 14 Jul 2005 19:30:02 -0400 Message-ID: <5708.1121383802@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/193 X-Sequence-Number: 13434 John A Meinel writes: > What I don't understand is that the planner is actually estimating that > joining against the new table is going to *increase* the number of > returned rows. It evidently thinks that incidentid in the k_r table is pretty nonunique. We really need to look at the statistics data to see what's going on. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 14 20:40:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1903152814 for ; Thu, 14 Jul 2005 20:40:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28735-10 for ; Thu, 14 Jul 2005 23:40:16 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 860C652C6B for ; Thu, 14 Jul 2005 20:40:12 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050714234011m9100k85k1e>; Thu, 14 Jul 2005 23:40:16 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id B51E355FFC; Thu, 14 Jul 2005 18:40:07 -0500 (CDT) Received: from [192.168.1.102] (65-100-20-12.cdrr.qwest.net [65.100.20.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id A941D55FF0; Thu, 14 Jul 2005 18:40:02 -0500 (CDT) Message-ID: <42D6F7CE.5010007@arbash-meinel.com> Date: Thu, 14 Jul 2005 18:39:58 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: Dan Harris , pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> <5708.1121383802@sss.pgh.pa.us> In-Reply-To: <5708.1121383802@sss.pgh.pa.us> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig20905F1FC17C14B2B7F6EFD9" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.03 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/194 X-Sequence-Number: 13435 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig20905F1FC17C14B2B7F6EFD9 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Tom Lane wrote: > John A Meinel writes: > >>What I don't understand is that the planner is actually estimating that >>joining against the new table is going to *increase* the number of >>returned rows. > > > It evidently thinks that incidentid in the k_r table is pretty > nonunique. We really need to look at the statistics data to > see what's going on. > > regards, tom lane > Okay, sure. What about doing this, then: EXPLAIN ANALYZE SELECT recordtext FROM eventactivity JOIN (SELECT DISTINCT incidentid FROM k_r JOIN k_b USING (incidentid) WHERE k_r.id = ?? AND k_b.id = ??) USING (incidentid) ; Since I assume that eventactivity is the only table with "recordtext", and that you don't get any columns from k_r and k_b, meaning it would be pointless to get duplicate incidentids. I may be misunderstanding what the query is trying to do, but depending on what is in k_r and k_b, is it possible to use a UNIQUE INDEX rather than just an index on incidentid? There is also the possibility of EXPLAIN ANALYZE SELECT recordtext FROM eventactivtity JOIN (SELECT incidentid FROM k_r WHERE k_r.id = ?? UNION SELECT incidentid FROM k_b WHERE k_b.id = ??) USING (incidentid) ; But both of these would mean that you don't actually want columns from k_r or k_b, just a unique list of incident ids. But first, I agree, we should make sure the pg_stats values are reasonable. John =:-> --------------enig20905F1FC17C14B2B7F6EFD9 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1vfPJdeBCYSNAAMRAgTWAJoDylT5gpCS6knbpQ0ztcboKlp/GACgsWSb zm/lpqNKxkD6fOywBwCNIls= =n7aS -----END PGP SIGNATURE----- --------------enig20905F1FC17C14B2B7F6EFD9-- From pgsql-performance-owner@postgresql.org Thu Jul 14 20:42:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E663D52B0D for ; Thu, 14 Jul 2005 20:42:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28191-06 for ; Thu, 14 Jul 2005 23:42:14 +0000 (GMT) Received: from ash25e.internode.on.net (ash25e.internode.on.net [203.16.214.182]) by svr1.postgresql.org (Postfix) with ESMTP id 09BDD52B5B for ; Thu, 14 Jul 2005 20:42:12 -0300 (ADT) Received: from pluto.mirrabooka.com (ppp188-133.static.internode.on.net [150.101.188.133]) by ash25e.internode.on.net (8.12.9/8.12.6) with ESMTP id j6ENgDEa061538 for ; Fri, 15 Jul 2005 09:12:14 +0930 (CST) (envelope-from alison@mirrabooka.com) Received: by pluto.mirrabooka.com (Postfix, from userid 506) id A6D522B11E; Fri, 15 Jul 2005 09:42:12 +1000 (EST) Date: Fri, 15 Jul 2005 09:42:12 +1000 To: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table Message-ID: <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> <20050714162114.GA19778@alvh.no-ip.org> In-Reply-To: <20050714162114.GA19778@alvh.no-ip.org> User-Agent: nail 11.6 9/7/04 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: alison@mirrabooka.com (Alison Winters) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/195 X-Sequence-Number: 13436 Hi, > > Our application requires a number of processes to select and update rows > > from a very small (<10 rows) Postgres table on a regular and frequent > > basis. These processes often run for weeks at a time, but over the > > space of a few days we find that updates start getting painfully slow. > > We are running a full vacuum/analyze and reindex on the table every day, > Full vacuum, eh? I wonder if what you really need is very frequent > non-full vacuum. Say, once in 15 minutes (exact rate depending on dead > tuple rate.) > Is there a difference between vacuum and vacuum full? Currently we have a cron job going every hour that does: VACUUM FULL VERBOSE ANALYZE plc_fldio REINDEX TABLE plc_fldio The most recent output was this: INFO: --Relation public.plc_fldio-- INFO: Pages 1221: Changed 3, reaped 256, Empty 0, New 0; Tup 108137: Vac 4176, Keep/VTL 108133/108133, UnUsed 19, MinLen 84, MaxLen 84; Re-using: Free/Avail. Space 445176/371836; EndEmpty/Avail. Pages 0/256. CPU 0.04s/0.14u sec elapsed 0.18 sec. INFO: Index plcpage_idx: Pages 315; Tuples 108137: Deleted 4176. CPU 0.03s/0.04u sec elapsed 0.14 sec. INFO: Rel plc_fldio: Pages: 1221 --> 1221; Tuple(s) moved: 0. CPU 0.03s/0.04u sec elapsed 0.36 sec. INFO: Analyzing public.plc_fldio VACUUM REINDEX We'll up it to every 15 minutes, but i don't know if that'll help because even with the current vacuuming the updates are still getting slower and slower over the course of several days. What really puzzles me is why restarting the processes fixes it. Does PostgreSQL keep some kind of backlog of transactions all for one database connection? Isn't it normal to have processes that keep a single database connection open for days at a time? Regarding the question another poster asked: all the transactions are very short. The table is essentially a database replacement for a shared memory segment - it contains a few rows of byte values that are constantly updated byte-at-a-time to communicate data between different industrial control processes. Thanks for the thoughts everyone, Alison From pgsql-performance-owner@postgresql.org Thu Jul 14 20:52:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 05A7252C34 for ; Thu, 14 Jul 2005 20:52:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 26328-07 for ; Thu, 14 Jul 2005 23:52:30 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id 4443E52C46 for ; Thu, 14 Jul 2005 20:52:29 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050714235232m920006unne>; Thu, 14 Jul 2005 23:52:32 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 6205F55FFD; Thu, 14 Jul 2005 18:52:32 -0500 (CDT) Received: from [192.168.1.102] (65-100-20-12.cdrr.qwest.net [65.100.20.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id 91A5955FFA; Thu, 14 Jul 2005 18:52:26 -0500 (CDT) Message-ID: <42D6FAB8.9030103@arbash-meinel.com> Date: Thu, 14 Jul 2005 18:52:24 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Alison Winters Cc: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> <20050714162114.GA19778@alvh.no-ip.org> <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> In-Reply-To: <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig379259FCDF523B5F31208C09" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.031 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/196 X-Sequence-Number: 13437 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig379259FCDF523B5F31208C09 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Alison Winters wrote: > Hi, > > >>>Our application requires a number of processes to select and update rows >>>from a very small (<10 rows) Postgres table on a regular and frequent >>>basis. These processes often run for weeks at a time, but over the >>>space of a few days we find that updates start getting painfully slow. >>>We are running a full vacuum/analyze and reindex on the table every day, >> >>Full vacuum, eh? I wonder if what you really need is very frequent >>non-full vacuum. Say, once in 15 minutes (exact rate depending on dead >>tuple rate.) >> > > Is there a difference between vacuum and vacuum full? Currently we have > a cron job going every hour that does: > > VACUUM FULL VERBOSE ANALYZE plc_fldio > REINDEX TABLE plc_fldio VACUUM FULL exclusively locks the table (so that nothing else can happen) and the compacts it as much as it can. You almost definitely want to only VACUUM every 15min, maybe VACUUM FULL 1/day. VACUUM FULL is more for when you haven't been VACUUMing often enough. Or have major changes to your table. Basically VACUUM marks rows as empty and available for reuse, VACUUM FULL removes empty space (but requires a full lock, because it is moving rows around). If anything, I would estimate that VACUUM FULL would be hurting your performance. But it may happen fast enough not to matter. > > The most recent output was this: > > INFO: --Relation public.plc_fldio-- > INFO: Pages 1221: Changed 3, reaped 256, Empty 0, New 0; Tup 108137: Vac 4176, Keep/VTL 108133/108133, UnUsed 19, MinLen 84, MaxLen 84; Re-using: Free/Avail. Space 445176/371836; EndEmpty/Avail. Pages 0/256. > CPU 0.04s/0.14u sec elapsed 0.18 sec. > INFO: Index plcpage_idx: Pages 315; Tuples 108137: Deleted 4176. > CPU 0.03s/0.04u sec elapsed 0.14 sec. > INFO: Rel plc_fldio: Pages: 1221 --> 1221; Tuple(s) moved: 0. > CPU 0.03s/0.04u sec elapsed 0.36 sec. > INFO: Analyzing public.plc_fldio > VACUUM > REINDEX > > We'll up it to every 15 minutes, but i don't know if that'll help > because even with the current vacuuming the updates are still getting > slower and slower over the course of several days. What really puzzles > me is why restarting the processes fixes it. Does PostgreSQL keep some > kind of backlog of transactions all for one database connection? Isn't > it normal to have processes that keep a single database connection open > for days at a time? I believe that certain locks are grabbed per session. Or at least there is some command that you can run, which you don't want to run in a maintained connection. (It might be VACUUM FULL, I don't remember which one it is). But the fact that your application works at all seems to be that it isn't acquiring any locks. I know VACUUM cannot clean up any rows that are visible in one of the transactions, I don't know if this includes active connections or not. > > Regarding the question another poster asked: all the transactions are > very short. The table is essentially a database replacement for a > shared memory segment - it contains a few rows of byte values that are > constantly updated byte-at-a-time to communicate data between different > industrial control processes. > > Thanks for the thoughts everyone, > > Alison > Is it possible to have some sort of timer that would recognize it has been connected for too long, drop the database connection, and reconnect? I don't know that it would solve anything, but it would be something you could try. John =:-> --------------enig379259FCDF523B5F31208C09 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1vq4JdeBCYSNAAMRAslFAJ9U+XYt88G/+8jsqdqaA0WlVh4IHACgrMba pOpfm6NZgTyXss6Vc3fr9JM= =FPqm -----END PGP SIGNATURE----- --------------enig379259FCDF523B5F31208C09-- From pgsql-performance-owner@postgresql.org Thu Jul 14 20:57:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1131952C3B for ; Thu, 14 Jul 2005 20:57:36 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28199-09 for ; Thu, 14 Jul 2005 23:57:26 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 1B46752BC7 for ; Thu, 14 Jul 2005 20:57:25 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6ENvRLR005894; Thu, 14 Jul 2005 19:57:27 -0400 (EDT) To: alison@mirrabooka.com (Alison Winters) Cc: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table In-reply-to: <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> <20050714162114.GA19778@alvh.no-ip.org> <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> Comments: In-reply-to alison@mirrabooka.com (Alison Winters) message dated "Fri, 15 Jul 2005 09:42:12 +1000" Date: Thu, 14 Jul 2005 19:57:27 -0400 Message-ID: <5893.1121385447@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/197 X-Sequence-Number: 13438 alison@mirrabooka.com (Alison Winters) writes: >>> Our application requires a number of processes to select and update rows >>> from a very small (<10 rows) Postgres table on a regular and frequent >>> basis. These processes often run for weeks at a time, but over the >>> space of a few days we find that updates start getting painfully slow. No wonder, considering that your "less than 10 rows" table contains something upwards of 100000 tuples: > INFO: --Relation public.plc_fldio-- > INFO: Pages 1221: Changed 3, reaped 256, Empty 0, New 0; Tup 108137: Vac 4176, Keep/VTL 108133/108133, UnUsed 19, MinLen 84, MaxLen 84; Re-using: Free/Avail. Space 445176/371836; EndEmpty/Avail. Pages 0/256. > CPU 0.04s/0.14u sec elapsed 0.18 sec. What you need to do is find out why VACUUM is unable to reclaim all those dead row versions. The reason is likely that some process is sitting on a open transaction for days at a time. > Isn't it normal to have processes that keep a single database > connection open for days at a time? Database connection, sure. Single transaction, no. > Regarding the question another poster asked: all the transactions are > very short. Somewhere you have one that isn't. Try watching the backends with ps, or look at the pg_stat_activity view if your version of PG has it, to see which sessions are staying "idle in transaction" indefinitely. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 14 21:12:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B0F8452C72 for ; Thu, 14 Jul 2005 21:12:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31032-10 for ; Fri, 15 Jul 2005 00:12:12 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 470F152C71 for ; Thu, 14 Jul 2005 21:12:10 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id A732D6441F1 for ; Thu, 14 Jul 2005 18:11:50 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 12146-06 for ; Thu, 14 Jul 2005 18:11:48 -0600 (MDT) Received: from [192.168.1.100] (c-67-162-134-190.hsd1.co.comcast.net [67.162.134.190]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id D5C956441ED for ; Thu, 14 Jul 2005 18:11:47 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <42D6F177.3000002@arbash-meinel.com> References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: slow joining very large table to smaller ones Date: Thu, 14 Jul 2005 18:12:07 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/198 X-Sequence-Number: 13439 On Jul 14, 2005, at 5:12 PM, John A Meinel wrote: > Dan Harris wrote: > > >>> >>> Well, postgres is estimating around 500 rows each, is that way >>> off? Try >>> just doing: >>> EXPLAIN ANALYZE SELECT incidentid FROM k_b WHERE id = 107; >>> EXPLAIN ANALYZE SELECT incidentid FROM k_r WHERE id = 94; >>> > > Once again, do this and post the results. We might just need to tweak > your settings so that it estimates the number of rows correctly, > and we > don't need to do anything else. > Ok, sorry I missed these the first time through: explain analyze select incidentid from k_b where id = 107; QUERY PLAN ------------------------------------------------------------------------ ------------------------------------------------ Index Scan using k_b_idx on k_b (cost=0.00..1926.03 rows=675 width=14) (actual time=0.042..298.394 rows=2493 loops=1) Index Cond: (id = 107) Total runtime: 299.103 ms select count(*) from k_b; count -------- 698350 ( sorry! I think I said this one only had tens of thousands in it ) explain analyze select incidentid from k_r where id = 94; QUERY PLAN ------------------------------------------------------------------------ ------------------------------------------------- Index Scan using k_r_idx on k_r (cost=0.00..2137.61 rows=757 width=14) (actual time=0.092..212.187 rows=10893 loops=1) Index Cond: (id = 94) Total runtime: 216.498 ms (3 rows) select count(*) from k_r; count -------- 671670 That one is quite a bit slower, yet it's the same table structure and same index as k_b, also it has fewer records. I did run VACUUM ANALYZE immediately before running these queries. It seems a lot better with the join_collapse set. > > \ > Well, the planner is powerful enough to flatten nested selects. To > make > it less "intelligent" you can do: > SET join_collapse_limit 1; > or > SET join_collapse_limit 0; > Which should tell postgres to not try and get tricky with your query. > Again, *usually* the planner knows better than you do. So again > just do > it to see what you get. > Ok, when join_collapse_limit = 1 I get this now: explain analyze select recordtext from eventactivity join ( select incidentid from k_r join k_b using (incidentid) where k_r.id = 94 and k_b.id = 107 ) a using (incidentid ); QUERY PLAN ------------------------------------------------------------------------ ----------------------------------------------------------------------- Nested Loop (cost=0.00..156509.08 rows=2948 width=35) (actual time=1.555..340.625 rows=24825 loops=1) -> Nested Loop (cost=0.00..5361.89 rows=6 width=28) (actual time=1.234..142.078 rows=366 loops=1) -> Index Scan using k_b_idx on k_b (cost=0.00..1943.09 rows=681 width=14) (actual time=0.423..56.974 rows=2521 loops=1) Index Cond: (id = 107) -> Index Scan using k_r_idx on k_r (cost=0.00..5.01 rows=1 width=14) (actual time=0.031..0.031 rows=0 loops=2521) Index Cond: ((k_r.id = 94) AND ((k_r.incidentid)::text = ("outer".incidentid)::text)) -> Index Scan using eventactivity1 on eventactivity (cost=0.00..25079.55 rows=8932 width=49) (actual time=0.107..0.481 rows=68 loops=366) Index Cond: ((eventactivity.incidentid)::text = ("outer".incidentid)::text) Total runtime: 347.975 ms MUCH better! Maybe you can help me understand what I did and if I need to make something permanent to get this behavior from now on? > > > > If you have analyzed recently can you do: > SELECT relname, reltuples FROM pg_class WHERE relname='eventactivity'; > > It is a cheaper form than "SELECT count(*) FROM eventactivity" to > get an > approximate estimate of the number of rows. But if it isn't too > expensive, please also give the value from SELECT count(*) FROM > eventactivity. > > Again, that helps us know if your tables are up-to-date. > Sure: select relname, reltuples from pg_class where relname='eventactivity'; relname | reltuples ---------------+------------- eventactivity | 3.16882e+07 select count(*) from eventactivity; count ---------- 31871142 > > >> >> >> >>> I don't know how selective your keys are, but one of these queries >>> should probably structure it better for the planner. It depends >>> a lot on >>> how selective your query is. >>> >> >> >> eventactivity currently has around 36 million rows in it. There >> should >> only be maybe 200-300 incidentids at most that will be matched >> with the >> combination of k_b and k_r. That's why I was thinking I could >> somehow >> get a list of just the incidentids that matched the id = 94 and id = >> 107 in k_b and k_r first. Then, I would only need to grab a few >> hundred >> out of 36 million rows from eventactivity. >> >> > > Well, you can also try: > SELECT count(*) FROM k_b JOIN k_r USING (incidentid) > WHERE k_b.id=?? AND k_r.id=?? > ; > > That will tell you how many rows they have in common. select count(*) from k_b join k_r using (incidentid) where k_b.id=107 and k_r.id=94; count ------- 373 > > Well, if you look at the latest plans, things have gone up from 44M to > 156M, I don't know why it is worse, but it is getting there. I assume this is because r_k and r_b are growing fairly rapidly right now. The time in between queries contained a lot of inserts. I was careful to vacuum analyze before sending statistics, as I did this time. I'm sorry if this has confused the issue. From pgsql-patches-owner@postgresql.org Thu Jul 14 21:22:39 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8AE8D52C84 for ; Thu, 14 Jul 2005 21:22:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40471-02 for ; Fri, 15 Jul 2005 00:22:32 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id D91FF52834 for ; Thu, 14 Jul 2005 21:22:27 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Thu, 14 Jul 2005 20:22:19 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Thu, 14 Jul 2005 20:22:19 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Thu, 14 Jul 2005 20:22:18 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 14 Jul 2005 17:22:18 -0700 Subject: Re: COPY FROM performance improvements From: "Alon Goldshuv" To: pgsql-patches@postgresql.org Message-ID: In-Reply-To: MIME-Version: 1.0 X-OriginalArrivalTime: 15 Jul 2005 00:22:19.0800 (UTC) FILETIME=[43450180:01C588D3] X-WSS-ID: 6EC9DE311J42555262-01-01 Content-Type: multipart/mixed; boundary=B_3204206540_47940875 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.601 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/314 X-Sequence-Number: 16654 > This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. --B_3204206540_47940875 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I revisited my patch and removed the code duplications that were there, and added support for CSV with buffered input, so CSV now runs faster too (although it is not as optimized as the TEXT format parsing). So now TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. Patch attached. Greetings, Alon. --B_3204206540_47940875 Content-Type: application/octet-stream; name=copy_parse_improvements_V15.patch Content-Disposition: attachment; filename=copy_parse_improvements_V15.patch Content-Transfer-Encoding: base64 SW5kZXg6IHNyYy9iYWNrZW5kL2NvbW1hbmRzL2NvcHkuYwo9PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09ClJDUyBm aWxlOiAvcHJvamVjdHMvY3Zzcm9vdC9wZ3NxbC9zcmMvYmFja2VuZC9jb21tYW5kcy9jb3B5 LmMsdgpyZXRyaWV2aW5nIHJldmlzaW9uIDEuMjQ3CmRpZmYgLWMgLXIxLjI0NyBjb3B5LmMK KioqIHNyYy9iYWNrZW5kL2NvbW1hbmRzL2NvcHkuYwkxMCBKdWwgMjAwNSAyMToxMzo1OCAt MDAwMAkxLjI0NwotLS0gc3JjL2JhY2tlbmQvY29tbWFuZHMvY29weS5jCTE1IEp1bCAyMDA1 IDAwOjA3OjQ5IC0wMDAwCioqKioqKioqKioqKioqKgoqKiogOCwxNCAqKioqCiAgICoKICAg KgogICAqIElERU5USUZJQ0FUSU9OCiEgICoJICAkUG9zdGdyZVNRTDogcGdzcWwvc3JjL2Jh Y2tlbmQvY29tbWFuZHMvY29weS5jLHYgMS4yNDcgMjAwNS8wNy8xMCAyMToxMzo1OCB0Z2wg RXhwICQKICAgKgogICAqLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQogICAqLwotLS0gOCwxNCAtLS0t CiAgICoKICAgKgogICAqIElERU5USUZJQ0FUSU9OCiEgICoJICAkUG9zdGdyZVNRTDogcGdz cWwvc3JjL2JhY2tlbmQvY29tbWFuZHMvY29weS5jLHYgMS4yNDYgMjAwNS8wNi8yOCAwNTow ODo1MyB0Z2wgRXhwICQKICAgKgogICAqLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQogICAqLwoqKioq KioqKioqKioqKioKKioqIDUwLDU1ICoqKioKLS0tIDUwLDU2IC0tLS0KICAKICAjZGVmaW5l IElTT0NUQUwoYykgKCgoYykgPj0gJzAnKSAmJiAoKGMpIDw9ICc3JykpCiAgI2RlZmluZSBP Q1RWQUxVRShjKSAoKGMpIC0gJzAnKQorICNkZWZpbmUgQ09QWV9CVUZfU0laRSA2NTUzNgog IAogIC8qCiAgICogUmVwcmVzZW50cyB0aGUgZGlmZmVyZW50IHNvdXJjZS9kZXN0IGNhc2Vz IHdlIG5lZWQgdG8gd29ycnkgYWJvdXQgYXQKKioqKioqKioqKioqKioqCioqKiA2Myw3OCAq KioqCiAgfSBDb3B5RGVzdDsKICAKICAvKgotICAqIFN0YXRlIGluZGljYXRvciBzaG93aW5n IHdoYXQgc3RvcHBlZCBDb3B5UmVhZEF0dHJpYnV0ZSgpCi0gICovCi0gdHlwZWRlZiBlbnVt IENvcHlSZWFkUmVzdWx0Ci0gewotIAlOT1JNQUxfQVRUUiwKLSAJRU5EX09GX0xJTkUsCi0g CVVOVEVSTUlOQVRFRF9GSUVMRAotIH0gQ29weVJlYWRSZXN1bHQ7Ci0gCi0gLyoKICAgKglS ZXByZXNlbnRzIHRoZSBlbmQtb2YtbGluZSB0ZXJtaW5hdG9yIHR5cGUgb2YgdGhlIGlucHV0 CiAgICovCiAgdHlwZWRlZiBlbnVtIEVvbFR5cGUKLS0tIDY0LDY5IC0tLS0KKioqKioqKioq KioqKioqCioqKiA5NywxMDIgKioqKgotLS0gODgsOTYgLS0tLQogIHN0YXRpYyBFb2xUeXBl IGVvbF90eXBlOwkJLyogRU9MIHR5cGUgb2YgaW5wdXQgKi8KICBzdGF0aWMgaW50CWNsaWVu dF9lbmNvZGluZzsJLyogcmVtb3RlIHNpZGUncyBjaGFyYWN0ZXIgZW5jb2RpbmcgKi8KICBz dGF0aWMgaW50CXNlcnZlcl9lbmNvZGluZzsJLyogbG9jYWwgZW5jb2RpbmcgKi8KKyBzdGF0 aWMgY2hhcgllb2xfY2hbMl07CQkJLyogVGhlIGJ5dGUgdmFsdWVzIG9mIHRoZSAxIG9yIDIg ZW9sIGJ5dGVzICovCisgc3RhdGljIGJvb2wgY2xpZW50X2VuY29kaW5nX29ubHk7IC8qIHRy dWUgaWYgY2xpZW50IGVuY29kaW5nIGlzIGEgbm9uCisgICAgICAgICAgICAgICAgICAgICAg ICAgICAgICAgICAgICAqIHN1cHBvcnRlZCBzZXJ2ZXIgZW5jb2RpbmcgKi8KICAKICAvKiB0 aGVzZSBhcmUganVzdCBmb3IgZXJyb3IgbWVzc2FnZXMsIHNlZSBjb3B5X2luX2Vycm9yX2Nh bGxiYWNrICovCiAgc3RhdGljIGJvb2wgY29weV9iaW5hcnk7CQkvKiBpcyBpdCBhIGJpbmFy eSBjb3B5PyAqLwoqKioqKioqKioqKioqKioKKioqIDEwNCwxMjYgKioqKgogIHN0YXRpYyBp bnQJY29weV9saW5lbm87CQkvKiBsaW5lIG51bWJlciBmb3IgZXJyb3IgbWVzc2FnZXMgKi8K ICBzdGF0aWMgY29uc3QgY2hhciAqY29weV9hdHRuYW1lOwkvKiBjdXJyZW50IGF0dCBmb3Ig ZXJyb3IgbWVzc2FnZXMgKi8KICAKICAKICAvKgogICAqIFRoZXNlIHN0YXRpYyB2YXJpYWJs ZXMgYXJlIHVzZWQgdG8gYXZvaWQgaW5jdXJyaW5nIG92ZXJoZWFkIGZvciBlYWNoCiEgICog YXR0cmlidXRlIHByb2Nlc3NlZC4gIGF0dHJpYnV0ZV9idWYgaXMgcmV1c2VkIG9uIGVhY2gg Q29weVJlYWRBdHRyaWJ1dGUKICAgKiBjYWxsIHRvIGhvbGQgdGhlIHN0cmluZyBiZWluZyBy ZWFkIGluLiAgVW5kZXIgbm9ybWFsIHVzZSBpdCB3aWxsIHNvb24KICAgKiBncm93IHRvIGEg c3VpdGFibGUgc2l6ZSwgYW5kIHRoZW4gd2Ugd2lsbCBhdm9pZCBwYWxsb2MvcGZyZWUgb3Zl cmhlYWQKICAgKiBmb3Igc3Vic2VxdWVudCBhdHRyaWJ1dGVzLiAgTm90ZSB0aGF0IENvcHlS ZWFkQXR0cmlidXRlIHJldHVybnMgYSBwb2ludGVyCiEgICogdG8gYXR0cmlidXRlX2J1Zidz IGRhdGEgYnVmZmVyIQogICAqLwohIHN0YXRpYyBTdHJpbmdJbmZvRGF0YSBhdHRyaWJ1dGVf YnVmOwogIAogIC8qCiAgICogU2ltaWxhcmx5LCBsaW5lX2J1ZiBob2xkcyB0aGUgd2hvbGUg aW5wdXQgbGluZSBiZWluZyBwcm9jZXNzZWQgKGl0cwogICAqIGN1cnNvciBmaWVsZCBwb2lu dHMgdG8gdGhlIG5leHQgY2hhcmFjdGVyIHRvIGJlIHJlYWQgYnkgQ29weVJlYWRBdHRyaWJ1 dGUpLgogICAqIFRoZSBpbnB1dCBjeWNsZSBpcyBmaXJzdCB0byByZWFkIHRoZSB3aG9sZSBs aW5lIGludG8gbGluZV9idWYsIGNvbnZlcnQgaXQKICAgKiB0byBzZXJ2ZXIgZW5jb2Rpbmcs IGFuZCB0aGVuIGV4dHJhY3QgaW5kaXZpZHVhbCBhdHRyaWJ1dGUgZmllbGRzIGludG8KISAg KiBhdHRyaWJ1dGVfYnVmLiAgKFdlIHVzZWQgdG8gaGF2ZSBDb3B5UmVhZEF0dHJpYnV0ZSBy ZWFkIHRoZSBpbnB1dCBzb3VyY2UKICAgKiBkaXJlY3RseSwgYnV0IHRoYXQgY2F1c2VkIGEg bG90IG9mIGVuY29kaW5nIGlzc3VlcyBhbmQgdW5uZWNlc3NhcnkgbG9naWMKICAgKiBjb21w bGV4aXR5LikKICAgKi8KLS0tIDk4LDE0MSAtLS0tCiAgc3RhdGljIGludAljb3B5X2xpbmVu bzsJCS8qIGxpbmUgbnVtYmVyIGZvciBlcnJvciBtZXNzYWdlcyAqLwogIHN0YXRpYyBjb25z dCBjaGFyICpjb3B5X2F0dG5hbWU7CS8qIGN1cnJlbnQgYXR0IGZvciBlcnJvciBtZXNzYWdl cyAqLwogIAorIC8qCisgICogU3RhdGljIHZhcmlhYmxlcyBmb3IgYnVmZmVyZWQgaW5wdXQg cGFyc2luZyAKKyAgKi8KKyBzdGF0aWMgY2hhciBpbnB1dF9idWZbQ09QWV9CVUZfU0laRSAr IDFdOyAvKiBleHRyYSBieXRlIGZvciAnXDAnICovCisgc3RhdGljIGludAlidWZmZXJfaW5k ZXg7CQkvKiBpbnB1dCBidWZmZXIgaW5kZXggKi8KKyBzdGF0aWMgYm9vbCBlbmRfbWFya2Vy OworIHN0YXRpYyBjaGFyICpiZWdsb2M7Cisgc3RhdGljIGNoYXIgKmVuZGxvYzsKKyBzdGF0 aWMgYm9vbCBidWZfZG9uZTsJCQkvKiBmaW5pc2hlZCBwcm9jZXNzaW5nIHRoZSBjdXJyZW50 IGJ1ZmZlciAqLworIHN0YXRpYyBib29sIGxpbmVfZG9uZTsJCQkvKiBmaW5pc2hlZCBwcm9j ZXNzaW5nIHRoZSB3aG9sZSBsaW5lIG9yCisgICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgKiBzdG9wcGVkIGluIHRoZSBtaWRkbGUgKi8KKyAvKiB0aGVzZSBhcmUgZm9yIENT ViBmb3JtYXQgKi8KKyBzdGF0aWMgYm9vbAlpbl9xdW90ZTsKKyBzdGF0aWMgYm9vbCBsYXN0 X3dhc19lc2M7CisgCisgLyogdGhlc2UgYXJlIGZvciBURVhUIGZvcm1hdCAqLworIHN0YXRp YyBib29sIGVzY19pbl9wcmV2YnVmOwkJLyogZXNjYXBlIHdhcyBsYXN0IGNoYXJhY3RlciBv ZiB0aGUgZGF0YQorICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICogaW5wdXQg YnVmZmVyICovCisgc3RhdGljIGJvb2wgY3JfaW5fcHJldmJ1ZjsJCS8qIENSIHdhcyBsYXN0 IGNoYXJhY3RlciBvZiB0aGUgZGF0YSBpbnB1dAorICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgICAgICogYnVmZmVyICovCisgCiAgCiAgLyoKICAgKiBUaGVzZSBzdGF0aWMgdmFy aWFibGVzIGFyZSB1c2VkIHRvIGF2b2lkIGluY3VycmluZyBvdmVyaGVhZCBmb3IgZWFjaAoh ICAqIGF0dHJpYnV0ZSBwcm9jZXNzZWQuICBhdHRyX2J1ZiBpcyByZXVzZWQgb24gZWFjaCBD b3B5UmVhZEF0dHJpYnV0ZQogICAqIGNhbGwgdG8gaG9sZCB0aGUgc3RyaW5nIGJlaW5nIHJl YWQgaW4uICBVbmRlciBub3JtYWwgdXNlIGl0IHdpbGwgc29vbgogICAqIGdyb3cgdG8gYSBz dWl0YWJsZSBzaXplLCBhbmQgdGhlbiB3ZSB3aWxsIGF2b2lkIHBhbGxvYy9wZnJlZSBvdmVy aGVhZAogICAqIGZvciBzdWJzZXF1ZW50IGF0dHJpYnV0ZXMuICBOb3RlIHRoYXQgQ29weVJl YWRBdHRyaWJ1dGUgcmV0dXJucyBhIHBvaW50ZXIKISAgKiB0byBhdHRyX2J1ZidzIGRhdGEg YnVmZmVyIQogICAqLwohIHN0YXRpYyBTdHJpbmdJbmZvRGF0YSBhdHRyX2J1ZjsgCiAgCiAg LyoKICAgKiBTaW1pbGFybHksIGxpbmVfYnVmIGhvbGRzIHRoZSB3aG9sZSBpbnB1dCBsaW5l IGJlaW5nIHByb2Nlc3NlZCAoaXRzCiAgICogY3Vyc29yIGZpZWxkIHBvaW50cyB0byB0aGUg bmV4dCBjaGFyYWN0ZXIgdG8gYmUgcmVhZCBieSBDb3B5UmVhZEF0dHJpYnV0ZSkuCiAgICog VGhlIGlucHV0IGN5Y2xlIGlzIGZpcnN0IHRvIHJlYWQgdGhlIHdob2xlIGxpbmUgaW50byBs aW5lX2J1ZiwgY29udmVydCBpdAogICAqIHRvIHNlcnZlciBlbmNvZGluZywgYW5kIHRoZW4g ZXh0cmFjdCBpbmRpdmlkdWFsIGF0dHJpYnV0ZSBmaWVsZHMgaW50bwohICAqIGF0dHJfYnVm LiAgKFdlIHVzZWQgdG8gaGF2ZSBDb3B5UmVhZEF0dHJpYnV0ZSByZWFkIHRoZSBpbnB1dCBz b3VyY2UKICAgKiBkaXJlY3RseSwgYnV0IHRoYXQgY2F1c2VkIGEgbG90IG9mIGVuY29kaW5n IGlzc3VlcyBhbmQgdW5uZWNlc3NhcnkgbG9naWMKICAgKiBjb21wbGV4aXR5LikKICAgKi8K KioqKioqKioqKioqKioqCioqKiAxMzcsMTU1ICoqKioKICBzdGF0aWMgdm9pZCBDb3B5RnJv bShSZWxhdGlvbiByZWwsIExpc3QgKmF0dG51bWxpc3QsIGJvb2wgYmluYXJ5LCBib29sIG9p ZHMsCiAgIGNoYXIgKmRlbGltLCBjaGFyICpudWxsX3ByaW50LCBib29sIGNzdl9tb2RlLCBj aGFyICpxdW90ZSwgY2hhciAqZXNjYXBlLAogIAkJIExpc3QgKmZvcmNlX25vdG51bGxfYXR0 cywgYm9vbCBoZWFkZXJfbGluZSk7CiEgc3RhdGljIGJvb2wgQ29weVJlYWRMaW5lKGNoYXIg KiBxdW90ZSwgY2hhciAqIGVzY2FwZSk7CiEgc3RhdGljIGNoYXIgKkNvcHlSZWFkQXR0cmli dXRlKGNvbnN0IGNoYXIgKmRlbGltLCBjb25zdCBjaGFyICpudWxsX3ByaW50LAohIAkJCQkg IENvcHlSZWFkUmVzdWx0ICpyZXN1bHQsIGJvb2wgKmlzbnVsbCk7CiEgc3RhdGljIGNoYXIg KkNvcHlSZWFkQXR0cmlidXRlQ1NWKGNvbnN0IGNoYXIgKmRlbGltLCBjb25zdCBjaGFyICpu dWxsX3ByaW50LAohIAkJCQkJIGNoYXIgKnF1b3RlLCBjaGFyICplc2NhcGUsCiEgCQkJCQkg Q29weVJlYWRSZXN1bHQgKnJlc3VsdCwgYm9vbCAqaXNudWxsKTsKICBzdGF0aWMgRGF0dW0g Q29weVJlYWRCaW5hcnlBdHRyaWJ1dGUoaW50IGNvbHVtbl9ubywgRm1nckluZm8gKmZsaW5m bywKISAJCQkJCQlPaWQgdHlwaW9wYXJhbSwgaW50MzIgdHlwbW9kLCBib29sICppc251bGwp OwogIHN0YXRpYyB2b2lkIENvcHlBdHRyaWJ1dGVPdXQoY2hhciAqc3RyaW5nLCBjaGFyICpk ZWxpbSk7CiAgc3RhdGljIHZvaWQgQ29weUF0dHJpYnV0ZU91dENTVihjaGFyICpzdHJpbmcs IGNoYXIgKmRlbGltLCBjaGFyICpxdW90ZSwKICAJCQkJCWNoYXIgKmVzY2FwZSwgYm9vbCBm b3JjZV9xdW90ZSk7CiAgc3RhdGljIExpc3QgKkNvcHlHZXRBdHRudW1zKFJlbGF0aW9uIHJl bCwgTGlzdCAqYXR0bmFtZWxpc3QpOwogIHN0YXRpYyB2b2lkIGxpbWl0X3ByaW50b3V0X2xl bmd0aChTdHJpbmdJbmZvIGJ1Zik7CiAgCiAgLyogSW50ZXJuYWwgY29tbXVuaWNhdGlvbnMg ZnVuY3Rpb25zICovCiAgc3RhdGljIHZvaWQgU2VuZENvcHlCZWdpbihib29sIGJpbmFyeSwg aW50IG5hdHRzKTsKLS0tIDE1MiwxNzYgLS0tLQogIHN0YXRpYyB2b2lkIENvcHlGcm9tKFJl bGF0aW9uIHJlbCwgTGlzdCAqYXR0bnVtbGlzdCwgYm9vbCBiaW5hcnksIGJvb2wgb2lkcywK ICAgY2hhciAqZGVsaW0sIGNoYXIgKm51bGxfcHJpbnQsIGJvb2wgY3N2X21vZGUsIGNoYXIg KnF1b3RlLCBjaGFyICplc2NhcGUsCiAgCQkgTGlzdCAqZm9yY2Vfbm90bnVsbF9hdHRzLCBi b29sIGhlYWRlcl9saW5lKTsKISBzdGF0aWMgYm9vbCBDb3B5UmVhZExpbmVUZXh0KHNpemVf dCBieXRlc3JlYWQsIGNoYXIgKmVzY2FwZSk7CiEgc3RhdGljIGJvb2wgQ29weVJlYWRMaW5l Q1NWKHNpemVfdCBieXRlc3JlYWQsIGNoYXIgKnF1b3RlLCBjaGFyICplc2NhcGUpOwohIHN0 YXRpYyB2b2lkIENvcHlSZWFkQXR0cmlidXRlc1RleHQoY29uc3QgY2hhciAqZGVsaW0sIGNv bnN0IGNoYXIgKmVzY2FwZSwgY29uc3QgY2hhciAqbnVsbF9wcmludCwKISAJCQkJCQkJCSAg IGludCBudWxsX3ByaW50X2xlbiwgY2hhciAqbnVsbHMsIExpc3QgKmF0dG51bWxpc3QsIAoh IAkJCQkJCQkJICAgaW50ICphdHRyX29mZnNldHMsIGludCBudW1fcGh5c19hdHRycywgRm9y bV9wZ19hdHRyaWJ1dGUgKmF0dHIpOwohIHN0YXRpYyB2b2lkIENvcHlSZWFkQXR0cmlidXRl c0NTVihjb25zdCBjaGFyICpkZWxpbSwgY29uc3QgY2hhciAqbnVsbF9wcmludCwgY2hhciAq cXVvdGUsCiEgCQkJCQkJCQkgIGNoYXIgKmVzY2FwZSwgaW50IG51bGxfcHJpbnRfbGVuLCBj aGFyICpudWxscywgTGlzdCAqYXR0bnVtbGlzdCwgCiEgCQkJCQkJCQkgIGludCAqYXR0cl9v ZmZzZXRzLCBpbnQgbnVtX3BoeXNfYXR0cnMsIEZvcm1fcGdfYXR0cmlidXRlICphdHRyKTsK ICBzdGF0aWMgRGF0dW0gQ29weVJlYWRCaW5hcnlBdHRyaWJ1dGUoaW50IGNvbHVtbl9ubywg Rm1nckluZm8gKmZsaW5mbywKISAJCQkJCQlPaWQgdHlwaW9wYXJhbSwgYm9vbCAqaXNudWxs KTsKISBzdGF0aWMgY2hhciAqQ29weVJlYWRPaWRBdHRyKGNvbnN0IGNoYXIgKmRlbGltLCBj b25zdCBjaGFyICpudWxsX3ByaW50LCBpbnQgbnVsbF9wcmludF9sZW4sCiEgCQkJCSAgICAg ICAgICAgICBib29sICppc251bGwpOwogIHN0YXRpYyB2b2lkIENvcHlBdHRyaWJ1dGVPdXQo Y2hhciAqc3RyaW5nLCBjaGFyICpkZWxpbSk7CiAgc3RhdGljIHZvaWQgQ29weUF0dHJpYnV0 ZU91dENTVihjaGFyICpzdHJpbmcsIGNoYXIgKmRlbGltLCBjaGFyICpxdW90ZSwKICAJCQkJ CWNoYXIgKmVzY2FwZSwgYm9vbCBmb3JjZV9xdW90ZSk7CiAgc3RhdGljIExpc3QgKkNvcHlH ZXRBdHRudW1zKFJlbGF0aW9uIHJlbCwgTGlzdCAqYXR0bmFtZWxpc3QpOwogIHN0YXRpYyB2 b2lkIGxpbWl0X3ByaW50b3V0X2xlbmd0aChTdHJpbmdJbmZvIGJ1Zik7Cisgc3RhdGljIGJv b2wgRGV0ZWN0TGluZUVuZChzaXplX3QgYnl0ZXNyZWFkLCBjaGFyICpxdW90ZSwgY2hhciAq ZXNjYXBlKTsKKyAKICAKICAvKiBJbnRlcm5hbCBjb21tdW5pY2F0aW9ucyBmdW5jdGlvbnMg Ki8KICBzdGF0aWMgdm9pZCBTZW5kQ29weUJlZ2luKGJvb2wgYmluYXJ5LCBpbnQgbmF0dHMp OwoqKioqKioqKioqKioqKioKKioqIDE1OSwxNzUgKioqKgogIHN0YXRpYyB2b2lkIENvcHlT ZW5kU3RyaW5nKGNvbnN0IGNoYXIgKnN0cik7CiAgc3RhdGljIHZvaWQgQ29weVNlbmRDaGFy KGNoYXIgYyk7CiAgc3RhdGljIHZvaWQgQ29weVNlbmRFbmRPZlJvdyhib29sIGJpbmFyeSk7 CiEgc3RhdGljIHZvaWQgQ29weUdldERhdGEodm9pZCAqZGF0YWJ1ZiwgaW50IGRhdGFzaXpl KTsKISBzdGF0aWMgaW50CUNvcHlHZXRDaGFyKHZvaWQpOwogIAogICNkZWZpbmUgQ29weUdl dEVvZigpICAoZmVfZW9mKQotIHN0YXRpYyBpbnQJQ29weVBlZWtDaGFyKHZvaWQpOwotIHN0 YXRpYyB2b2lkIENvcHlEb25lUGVlayhpbnQgYywgYm9vbCBwaWNrdXApOwogIHN0YXRpYyB2 b2lkIENvcHlTZW5kSW50MzIoaW50MzIgdmFsKTsKICBzdGF0aWMgaW50MzIgQ29weUdldElu dDMyKHZvaWQpOwogIHN0YXRpYyB2b2lkIENvcHlTZW5kSW50MTYoaW50MTYgdmFsKTsKICBz dGF0aWMgaW50MTYgQ29weUdldEludDE2KHZvaWQpOwogIAogIAogIC8qCiAgICogU2VuZCBj b3B5IHN0YXJ0L3N0b3AgbWVzc2FnZXMgZm9yIGZyb250ZW5kIGNvcGllcy4gIFRoZXNlIGhh dmUgY2hhbmdlZAotLS0gMTgwLDE5NyAtLS0tCiAgc3RhdGljIHZvaWQgQ29weVNlbmRTdHJp bmcoY29uc3QgY2hhciAqc3RyKTsKICBzdGF0aWMgdm9pZCBDb3B5U2VuZENoYXIoY2hhciBj KTsKICBzdGF0aWMgdm9pZCBDb3B5U2VuZEVuZE9mUm93KGJvb2wgYmluYXJ5KTsKISBzdGF0 aWMgaW50CUNvcHlHZXREYXRhKHZvaWQgKmRhdGFidWYsIGludCBkYXRhc2l6ZSk7CiAgCiAg I2RlZmluZSBDb3B5R2V0RW9mKCkgIChmZV9lb2YpCiAgc3RhdGljIHZvaWQgQ29weVNlbmRJ bnQzMihpbnQzMiB2YWwpOwogIHN0YXRpYyBpbnQzMiBDb3B5R2V0SW50MzIodm9pZCk7CiAg c3RhdGljIHZvaWQgQ29weVNlbmRJbnQxNihpbnQxNiB2YWwpOwogIHN0YXRpYyBpbnQxNiBD b3B5R2V0SW50MTYodm9pZCk7CiAgCisgLyogYnl0ZSBzY2FuaW5nIHV0aWxzICovCisgc3Rh dGljIGNoYXIgKnNjYW5UZXh0TGluZShjb25zdCBjaGFyICpzLCBjaGFyIGMsIHNpemVfdCBs ZW4pOworIHN0YXRpYyBjaGFyICpzY2FuQ1NWTGluZShjb25zdCBjaGFyICpzLCBjaGFyIGMx LCBjaGFyIGMyLCBjaGFyIGMzLCBzaXplX3QgbGVuKTsKKyBzdGF0aWMgY2hhciAqc2NhblRl eHRBdHRyKGNvbnN0IGNoYXIgKnMsIGNoYXIgYzEsIGNoYXIgYzIsIHNpemVfdCBsZW4pOwog IAogIC8qCiAgICogU2VuZCBjb3B5IHN0YXJ0L3N0b3AgbWVzc2FnZXMgZm9yIGZyb250ZW5k IGNvcGllcy4gIFRoZXNlIGhhdmUgY2hhbmdlZAoqKioqKioqKioqKioqKioKKioqIDM4Miwz OTUgKioqKgogICAqIEl0IHNlZW1zIHVud2lzZSB0byBhbGxvdyB0aGUgQ09QWSBJTiB0byBj b21wbGV0ZSBub3JtYWxseSBpbiB0aGF0IGNhc2UuCiAgICoKICAgKiBOQjogbm8gZGF0YSBj b252ZXJzaW9uIGlzIGFwcGxpZWQgYnkgdGhlc2UgZnVuY3Rpb25zCiAgICovCiEgc3RhdGlj IHZvaWQKICBDb3B5R2V0RGF0YSh2b2lkICpkYXRhYnVmLCBpbnQgZGF0YXNpemUpCiAgewog IAlzd2l0Y2ggKGNvcHlfZGVzdCkKICAJewogIAkJY2FzZSBDT1BZX0ZJTEU6CiEgCQkJZnJl YWQoZGF0YWJ1ZiwgZGF0YXNpemUsIDEsIGNvcHlfZmlsZSk7CiAgCQkJaWYgKGZlb2YoY29w eV9maWxlKSkKICAJCQkJZmVfZW9mID0gdHJ1ZTsKICAJCQlicmVhazsKLS0tIDQwNCw0MjIg LS0tLQogICAqIEl0IHNlZW1zIHVud2lzZSB0byBhbGxvdyB0aGUgQ09QWSBJTiB0byBjb21w bGV0ZSBub3JtYWxseSBpbiB0aGF0IGNhc2UuCiAgICoKICAgKiBOQjogbm8gZGF0YSBjb252 ZXJzaW9uIGlzIGFwcGxpZWQgYnkgdGhlc2UgZnVuY3Rpb25zCisgICoKKyAgKiBSZXR1cm5z OiB0aGUgbnVtYmVyIG9mIGJ5dGVzIHRoYXQgd2VyZSBzdWNjZXNzZnVsbHkgcmVhZAorICAq IGludG8gdGhlIGRhdGEgYnVmZmVyLgogICAqLwohIHN0YXRpYyBpbnQKICBDb3B5R2V0RGF0 YSh2b2lkICpkYXRhYnVmLCBpbnQgZGF0YXNpemUpCiAgeworIAlzaXplX3QJCWJ5dGVzcmVh ZCA9IDA7CisgCiAgCXN3aXRjaCAoY29weV9kZXN0KQogIAl7CiAgCQljYXNlIENPUFlfRklM RToKISAJCQlieXRlc3JlYWQgPSBmcmVhZChkYXRhYnVmLCAxLCBkYXRhc2l6ZSwgY29weV9m aWxlKTsKICAJCQlpZiAoZmVvZihjb3B5X2ZpbGUpKQogIAkJCQlmZV9lb2YgPSB0cnVlOwog IAkJCWJyZWFrOwoqKioqKioqKioqKioqKioKKioqIDQwMSw0MDYgKioqKgotLS0gNDI4LDQz NSAtLS0tCiAgCQkJCQkJKGVycmNvZGUoRVJSQ09ERV9DT05ORUNUSU9OX0ZBSUxVUkUpLAog IAkJCQkJCSBlcnJtc2coInVuZXhwZWN0ZWQgRU9GIG9uIGNsaWVudCBjb25uZWN0aW9uIikp KTsKICAJCQl9CisgCQkJYnl0ZXNyZWFkICs9IGRhdGFzaXplOwkJLyogdXBkYXRlIHRoZSBj b3VudCBvZiBieXRlcyB0aGF0IHdlcmUKKyAJCQkJCQkJCQkJICogcmVhZCBzbyBmYXIgKi8K ICAJCQlicmVhazsKICAJCWNhc2UgQ09QWV9ORVdfRkU6CiAgCQkJd2hpbGUgKGRhdGFzaXpl ID4gMCAmJiAhZmVfZW9mKQoqKioqKioqKioqKioqKioKKioqIDQyOSw0MzUgKioqKgogIAkJ CQkJCWNhc2UgJ2MnOgkJLyogQ29weURvbmUgKi8KICAJCQkJCQkJLyogQ09QWSBJTiBjb3Jy ZWN0bHkgdGVybWluYXRlZCBieSBmcm9udGVuZCAqLwogIAkJCQkJCQlmZV9lb2YgPSB0cnVl OwohIAkJCQkJCQlyZXR1cm47CiAgCQkJCQkJY2FzZSAnZic6CQkvKiBDb3B5RmFpbCAqLwog IAkJCQkJCQllcmVwb3J0KEVSUk9SLAogIAkJCQkJCQkJCShlcnJjb2RlKEVSUkNPREVfUVVF UllfQ0FOQ0VMRUQpLAotLS0gNDU4LDQ2NCAtLS0tCiAgCQkJCQkJY2FzZSAnYyc6CQkvKiBD b3B5RG9uZSAqLwogIAkJCQkJCQkvKiBDT1BZIElOIGNvcnJlY3RseSB0ZXJtaW5hdGVkIGJ5 IGZyb250ZW5kICovCiAgCQkJCQkJCWZlX2VvZiA9IHRydWU7CiEgCQkJCQkJCXJldHVybiBi eXRlc3JlYWQ7CiAgCQkJCQkJY2FzZSAnZic6CQkvKiBDb3B5RmFpbCAqLwogIAkJCQkJCQll cmVwb3J0KEVSUk9SLAogIAkJCQkJCQkJCShlcnJjb2RlKEVSUkNPREVfUVVFUllfQ0FOQ0VM RUQpLAoqKioqKioqKioqKioqKioKKioqIDQ1OSw1OTggKioqKgogIAkJCQkJYXZhaWwgPSBk YXRhc2l6ZTsKICAJCQkJcHFfY29weW1zZ2J5dGVzKGNvcHlfbXNnYnVmLCBkYXRhYnVmLCBh dmFpbCk7CiAgCQkJCWRhdGFidWYgPSAodm9pZCAqKSAoKGNoYXIgKikgZGF0YWJ1ZiArIGF2 YWlsKTsKICAJCQkJZGF0YXNpemUgLT0gYXZhaWw7CiAgCQkJfQogIAkJCWJyZWFrOwogIAl9 Ci0gfQotIAotIHN0YXRpYyBpbnQKLSBDb3B5R2V0Q2hhcih2b2lkKQotIHsKLSAJaW50CQkJ Y2g7Ci0gCi0gCXN3aXRjaCAoY29weV9kZXN0KQotIAl7Ci0gCQljYXNlIENPUFlfRklMRToK LSAJCQljaCA9IGdldGMoY29weV9maWxlKTsKLSAJCQlicmVhazsKLSAJCWNhc2UgQ09QWV9P TERfRkU6Ci0gCQkJY2ggPSBwcV9nZXRieXRlKCk7Ci0gCQkJaWYgKGNoID09IEVPRikKLSAJ CQl7Ci0gCQkJCS8qIE9ubHkgYSBcLiB0ZXJtaW5hdG9yIGlzIGxlZ2FsIEVPRiBpbiBvbGQg cHJvdG9jb2wgKi8KLSAJCQkJZXJlcG9ydChFUlJPUiwKLSAJCQkJCQkoZXJyY29kZShFUlJD T0RFX0NPTk5FQ1RJT05fRkFJTFVSRSksCi0gCQkJCQkJIGVycm1zZygidW5leHBlY3RlZCBF T0Ygb24gY2xpZW50IGNvbm5lY3Rpb24iKSkpOwotIAkJCX0KLSAJCQlicmVhazsKLSAJCWNh c2UgQ09QWV9ORVdfRkU6Ci0gCQkJewotIAkJCQl1bnNpZ25lZCBjaGFyIGNjOwotIAotIAkJ CQlDb3B5R2V0RGF0YSgmY2MsIDEpOwotIAkJCQlpZiAoZmVfZW9mKQotIAkJCQkJY2ggPSBF T0Y7Ci0gCQkJCWVsc2UKLSAJCQkJCWNoID0gY2M7Ci0gCQkJCWJyZWFrOwotIAkJCX0KLSAJ CWRlZmF1bHQ6Ci0gCQkJY2ggPSBFT0Y7Ci0gCQkJYnJlYWs7Ci0gCX0KLSAJaWYgKGNoID09 IEVPRikKLSAJCWZlX2VvZiA9IHRydWU7Ci0gCXJldHVybiBjaDsKLSB9Ci0gCi0gLyoKLSAg KiBDb3B5UGVla0NoYXIgcmVhZHMgYSBieXRlIGluICJwZWVrYWJsZSIgbW9kZS4KLSAgKgot ICAqIGFmdGVyIGVhY2ggY2FsbCB0byBDb3B5UGVla0NoYXIsIGEgY2FsbCB0byBDb3B5RG9u ZVBlZWsgX211c3RfCi0gICogZm9sbG93LCB1bmxlc3MgRU9GIHdhcyByZXR1cm5lZC4KLSAg KgotICAqIENvcHlEb25lUGVlayB3aWxsIGVpdGhlciB0YWtlIHRoZSBwZWVrZWQgY2hhciBv ZmYgdGhlIHN0cmVhbQotICAqIChpZiBwaWNrdXAgaXMgdHJ1ZSkgb3IgbGVhdmUgaXQgb24g dGhlIHN0cmVhbSAoaWYgcGlja3VwIGlzIGZhbHNlKS4KLSAgKi8KLSBzdGF0aWMgaW50Ci0g Q29weVBlZWtDaGFyKHZvaWQpCi0gewotIAlpbnQJCQljaDsKLSAKLSAJc3dpdGNoIChjb3B5 X2Rlc3QpCi0gCXsKLSAJCWNhc2UgQ09QWV9GSUxFOgotIAkJCWNoID0gZ2V0Yyhjb3B5X2Zp bGUpOwotIAkJCWJyZWFrOwotIAkJY2FzZSBDT1BZX09MRF9GRToKLSAJCQljaCA9IHBxX3Bl ZWtieXRlKCk7Ci0gCQkJaWYgKGNoID09IEVPRikKLSAJCQl7Ci0gCQkJCS8qIE9ubHkgYSBc LiB0ZXJtaW5hdG9yIGlzIGxlZ2FsIEVPRiBpbiBvbGQgcHJvdG9jb2wgKi8KLSAJCQkJZXJl cG9ydChFUlJPUiwKLSAJCQkJCQkoZXJyY29kZShFUlJDT0RFX0NPTk5FQ1RJT05fRkFJTFVS RSksCi0gCQkJCQkJIGVycm1zZygidW5leHBlY3RlZCBFT0Ygb24gY2xpZW50IGNvbm5lY3Rp b24iKSkpOwotIAkJCX0KLSAJCQlicmVhazsKLSAJCWNhc2UgQ09QWV9ORVdfRkU6Ci0gCQkJ ewotIAkJCQl1bnNpZ25lZCBjaGFyIGNjOwotIAotIAkJCQlDb3B5R2V0RGF0YSgmY2MsIDEp OwotIAkJCQlpZiAoZmVfZW9mKQotIAkJCQkJY2ggPSBFT0Y7Ci0gCQkJCWVsc2UKLSAJCQkJ CWNoID0gY2M7Ci0gCQkJCWJyZWFrOwotIAkJCX0KLSAJCWRlZmF1bHQ6Ci0gCQkJY2ggPSBF T0Y7Ci0gCQkJYnJlYWs7Ci0gCX0KLSAJaWYgKGNoID09IEVPRikKLSAJCWZlX2VvZiA9IHRy dWU7Ci0gCXJldHVybiBjaDsKLSB9Ci0gCi0gc3RhdGljIHZvaWQKLSBDb3B5RG9uZVBlZWso aW50IGMsIGJvb2wgcGlja3VwKQotIHsKLSAJaWYgKGZlX2VvZikKLSAJCXJldHVybjsJCQkJ CS8qIGNhbid0IHVuZ2V0IGFuIEVPRiAqLwotIAlzd2l0Y2ggKGNvcHlfZGVzdCkKLSAJewot IAkJY2FzZSBDT1BZX0ZJTEU6Ci0gCQkJaWYgKCFwaWNrdXApCi0gCQkJewotIAkJCQkvKiBX ZSBkb24ndCB3YW50IHRvIHBpY2sgaXQgdXAgLSBzbyBwdXQgaXQgYmFjayBpbiB0aGVyZSAq LwotIAkJCQl1bmdldGMoYywgY29weV9maWxlKTsKLSAJCQl9Ci0gCQkJLyogSWYgd2Ugd2Fu dGVkIHRvIHBpY2sgaXQgdXAsIGl0J3MgYWxyZWFkeSBkb25lICovCi0gCQkJYnJlYWs7Ci0g CQljYXNlIENPUFlfT0xEX0ZFOgotIAkJCWlmIChwaWNrdXApCi0gCQkJewotIAkJCQkvKiBX ZSB3YW50IHRvIHBpY2sgaXQgdXAgKi8KLSAJCQkJKHZvaWQpIHBxX2dldGJ5dGUoKTsKLSAJ CQl9CiAgCiEgCQkJLyoKISAJCQkgKiBJZiB3ZSBkaWRuJ3Qgd2FudCB0byBwaWNrIGl0IHVw LCBqdXN0IGxlYXZlIGl0IHdoZXJlIGl0CiEgCQkJICogc2l0cwohIAkJCSAqLwohIAkJCWJy ZWFrOwohIAkJY2FzZSBDT1BZX05FV19GRToKISAJCQlpZiAoIXBpY2t1cCkKISAJCQl7CiEg CQkJCS8qIFdlIGRvbid0IHdhbnQgdG8gcGljayBpdCB1cCAtIHNvIHB1dCBpdCBiYWNrIGlu IHRoZXJlICovCiEgCQkJCWNvcHlfbXNnYnVmLT5jdXJzb3ItLTsKISAJCQl9CiEgCQkJLyog SWYgd2Ugd2FudGVkIHRvIHBpY2sgaXQgdXAsIGl0J3MgYWxyZWFkeSBkb25lICovCiEgCQkJ YnJlYWs7CiEgCX0KICB9CiAgCi0gCiAgLyoKICAgKiBUaGVzZSBmdW5jdGlvbnMgZG8gYXBw bHkgc29tZSBkYXRhIGNvbnZlcnNpb24KICAgKi8KLS0tIDQ4OCw1MDMgLS0tLQogIAkJCQkJ YXZhaWwgPSBkYXRhc2l6ZTsKICAJCQkJcHFfY29weW1zZ2J5dGVzKGNvcHlfbXNnYnVmLCBk YXRhYnVmLCBhdmFpbCk7CiAgCQkJCWRhdGFidWYgPSAodm9pZCAqKSAoKGNoYXIgKikgZGF0 YWJ1ZiArIGF2YWlsKTsKKyAJCQkJYnl0ZXNyZWFkICs9IGF2YWlsOwkJLyogdXBkYXRlIHRo ZSBjb3VudCBvZiBieXRlcyB0aGF0IHdlcmUKKyAJCQkJCQkJCQkJICogcmVhZCBzbyBmYXIg Ki8KICAJCQkJZGF0YXNpemUgLT0gYXZhaWw7CiAgCQkJfQogIAkJCWJyZWFrOwogIAl9CiAg CiEgCXJldHVybiBieXRlc3JlYWQ7CiAgfQogIAogIC8qCiAgICogVGhlc2UgZnVuY3Rpb25z IGRvIGFwcGx5IHNvbWUgZGF0YSBjb252ZXJzaW9uCiAgICovCioqKioqKioqKioqKioqKgoq KiogOTY4LDk4MCAqKioqCiAgCX0KICAKICAJLyogU2V0IHVwIHZhcmlhYmxlcyB0byBhdm9p ZCBwZXItYXR0cmlidXRlIG92ZXJoZWFkLiAqLwohIAlpbml0U3RyaW5nSW5mbygmYXR0cmli dXRlX2J1Zik7CiAgCWluaXRTdHJpbmdJbmZvKCZsaW5lX2J1Zik7CiAgCWxpbmVfYnVmX2Nv bnZlcnRlZCA9IGZhbHNlOwogIAogIAljbGllbnRfZW5jb2RpbmcgPSBwZ19nZXRfY2xpZW50 X2VuY29kaW5nKCk7CiAgCXNlcnZlcl9lbmNvZGluZyA9IEdldERhdGFiYXNlRW5jb2Rpbmco KTsKICAKICAJY29weV9kZXN0ID0gQ09QWV9GSUxFOwkJLyogZGVmYXVsdCAqLwogIAljb3B5 X2ZpbGUgPSBOVUxMOwogIAljb3B5X21zZ2J1ZiA9IE5VTEw7Ci0tLSA4NzMsOTA1IC0tLS0K ICAJfQogIAogIAkvKiBTZXQgdXAgdmFyaWFibGVzIHRvIGF2b2lkIHBlci1hdHRyaWJ1dGUg b3ZlcmhlYWQuICovCiEgCWluaXRTdHJpbmdJbmZvKCZhdHRyX2J1Zik7CiAgCWluaXRTdHJp bmdJbmZvKCZsaW5lX2J1Zik7CiAgCWxpbmVfYnVmX2NvbnZlcnRlZCA9IGZhbHNlOwogIAog IAljbGllbnRfZW5jb2RpbmcgPSBwZ19nZXRfY2xpZW50X2VuY29kaW5nKCk7CiAgCXNlcnZl cl9lbmNvZGluZyA9IEdldERhdGFiYXNlRW5jb2RpbmcoKTsKICAKKyAJLyoKKyAJICogY2hl Y2sgaWYgdGhlIGNsaWVudCBlbmNvZGluZyBpcyBvbmUgb2YgdGhlIDUgZW5jb2RpbmdzCisg CSAqIHRoYXQgYXJlIG5vdCBzdXBwb3J0ZWQgYXMgYSBzZXJ2ZXIgZW5jb2RpbmdzLgorIAkg Ki8KKyAJc3dpdGNoIChjbGllbnRfZW5jb2RpbmcpCisgCXsKKyAJCWNhc2UgUEdfU0pJUzoK KyAJCWNhc2UgUEdfQklHNToKKyAJCWNhc2UgUEdfR0JLOgorIAkJY2FzZSBQR19VSEM6Cisg CQljYXNlIFBHX0dCMTgwMzA6CisgCQkJY2xpZW50X2VuY29kaW5nX29ubHkgPSB0cnVlOwor IAkJCWJyZWFrOworIAkJZGVmYXVsdDoKKyAJCQljbGllbnRfZW5jb2Rpbmdfb25seSA9IGZh bHNlOworIAl9CisgCisgCWlmKCFjc3ZfbW9kZSkKKyAJICAgIGVzY2FwZSA9ICJcXCI7CQkv KiBkZWZhdWx0IGZvciB0ZXh0IGZvcm1hdCAqLworIAogIAljb3B5X2Rlc3QgPSBDT1BZX0ZJ TEU7CQkvKiBkZWZhdWx0ICovCiAgCWNvcHlfZmlsZSA9IE5VTEw7CiAgCWNvcHlfbXNnYnVm ID0gTlVMTDsKKioqKioqKioqKioqKioqCioqKiAxMTA1LDExMTMgKioqKgogIAkJCQkJIGVy cm1zZygiY291bGQgbm90IHdyaXRlIHRvIGZpbGUgXCIlc1wiOiAlbSIsCiAgCQkJCQkJCWZp bGVuYW1lKSkpOwogIAl9CiEgCXBmcmVlKGF0dHJpYnV0ZV9idWYuZGF0YSk7CiAgCXBmcmVl KGxpbmVfYnVmLmRhdGEpOwogIAogIAkvKgogIAkgKiBDbG9zZSB0aGUgcmVsYXRpb24uCUlm IHJlYWRpbmcsIHdlIGNhbiByZWxlYXNlIHRoZSBBY2Nlc3NTaGFyZUxvY2sKICAJICogd2Ug Z290OyBpZiB3cml0aW5nLCB3ZSBzaG91bGQgaG9sZCB0aGUgbG9jayB1bnRpbCBlbmQgb2YK LS0tIDEwMzAsMTAzOSAtLS0tCiAgCQkJCQkgZXJybXNnKCJjb3VsZCBub3Qgd3JpdGUgdG8g ZmlsZSBcIiVzXCI6ICVtIiwKICAJCQkJCQkJZmlsZW5hbWUpKSk7CiAgCX0KISAJcGZyZWUo YXR0cl9idWYuZGF0YSk7CiAgCXBmcmVlKGxpbmVfYnVmLmRhdGEpOwogIAorIAogIAkvKgog IAkgKiBDbG9zZSB0aGUgcmVsYXRpb24uCUlmIHJlYWRpbmcsIHdlIGNhbiByZWxlYXNlIHRo ZSBBY2Nlc3NTaGFyZUxvY2sKICAJICogd2UgZ290OyBpZiB3cml0aW5nLCB3ZSBzaG91bGQg aG9sZCB0aGUgbG9jayB1bnRpbCBlbmQgb2YKKioqKioqKioqKioqKioqCioqKiAxMzQ1LDEz NTEgKioqKgogIAkJCQl7CiAgCQkJCQlieXRlYQkgICAqb3V0cHV0Ynl0ZXM7CiAgCiEgCQkJ CQlvdXRwdXRieXRlcyA9IERhdHVtR2V0Qnl0ZWFQKEZ1bmN0aW9uQ2FsbDEoJm91dF9mdW5j dGlvbnNbYXR0bnVtIC0gMV0sCiAgCQkJCQkJCQkJCQkJCQkJICAgdmFsdWUpKTsKICAJCQkJ CS8qIFdlIGFzc3VtZSB0aGUgcmVzdWx0IHdpbGwgbm90IGhhdmUgYmVlbiB0b2FzdGVkICov CiAgCQkJCQlDb3B5U2VuZEludDMyKFZBUlNJWkUob3V0cHV0Ynl0ZXMpIC0gVkFSSERSU1op OwotLS0gMTI3MSwxMjc4IC0tLS0KICAJCQkJewogIAkJCQkJYnl0ZWEJICAgKm91dHB1dGJ5 dGVzOwogIAohIAkJCQkJb3V0cHV0Ynl0ZXMgPQohIAkJCQkJCURhdHVtR2V0Qnl0ZWFQKEZ1 bmN0aW9uQ2FsbDEoJm91dF9mdW5jdGlvbnNbYXR0bnVtIC0gMV0sCiAgCQkJCQkJCQkJCQkJ CQkJICAgdmFsdWUpKTsKICAJCQkJCS8qIFdlIGFzc3VtZSB0aGUgcmVzdWx0IHdpbGwgbm90 IGhhdmUgYmVlbiB0b2FzdGVkICovCiAgCQkJCQlDb3B5U2VuZEludDMyKFZBUlNJWkUob3V0 cHV0Ynl0ZXMpIC0gVkFSSERSU1opOwoqKioqKioqKioqKioqKioKKioqIDEzOTUsMTQwNCAq KioqCiAgCQlpZiAoY29weV9hdHRuYW1lKQogIAkJewogIAkJCS8qIGVycm9yIGlzIHJlbGV2 YW50IHRvIGEgcGFydGljdWxhciBjb2x1bW4gKi8KISAJCQlsaW1pdF9wcmludG91dF9sZW5n dGgoJmF0dHJpYnV0ZV9idWYpOwogIAkJCWVycmNvbnRleHQoIkNPUFkgJXMsIGxpbmUgJWQs IGNvbHVtbiAlczogXCIlc1wiIiwKICAJCQkJCSAgIGNvcHlfcmVsbmFtZSwgY29weV9saW5l bm8sIGNvcHlfYXR0bmFtZSwKISAJCQkJCSAgIGF0dHJpYnV0ZV9idWYuZGF0YSk7CiAgCQl9 CiAgCQllbHNlCiAgCQl7Ci0tLSAxMzIyLDEzMzEgLS0tLQogIAkJaWYgKGNvcHlfYXR0bmFt ZSkKICAJCXsKICAJCQkvKiBlcnJvciBpcyByZWxldmFudCB0byBhIHBhcnRpY3VsYXIgY29s dW1uICovCiEgCQkJbGltaXRfcHJpbnRvdXRfbGVuZ3RoKCZhdHRyX2J1Zik7CiAgCQkJZXJy Y29udGV4dCgiQ09QWSAlcywgbGluZSAlZCwgY29sdW1uICVzOiBcIiVzXCIiLAogIAkJCQkJ ICAgY29weV9yZWxuYW1lLCBjb3B5X2xpbmVubywgY29weV9hdHRuYW1lLAohIAkJCQkJICAg YXR0cl9idWYuZGF0YSk7CiAgCQl9CiAgCQllbHNlCiAgCQl7CioqKioqKioqKioqKioqKgoq KiogMTQwNiwxNDExICoqKioKLS0tIDEzMzMsMTM0MCAtLS0tCiAgCQkJaWYgKGxpbmVfYnVm X2NvbnZlcnRlZCB8fAogIAkJCQljbGllbnRfZW5jb2RpbmcgPT0gc2VydmVyX2VuY29kaW5n KQogIAkJCXsKKyAJCQkJLyogU3RyaXAgb2ZmIHRoZSBuZXdsaW5lICovCisgCQkJCSoobGlu ZV9idWYuZGF0YSArIGxpbmVfYnVmLmxlbiAtIDEpID0gJ1wwJzsKICAJCQkJbGltaXRfcHJp bnRvdXRfbGVuZ3RoKCZsaW5lX2J1Zik7CiAgCQkJCWVycmNvbnRleHQoIkNPUFkgJXMsIGxp bmUgJWQ6IFwiJXNcIiIsCiAgCQkJCQkJICAgY29weV9yZWxuYW1lLCBjb3B5X2xpbmVubywK KioqKioqKioqKioqKioqCioqKiAxNDg1LDE0OTEgKioqKgogIAlPaWQJCQlpbl9mdW5jX29p ZDsKICAJRGF0dW0JICAgKnZhbHVlczsKICAJY2hhcgkgICAqbnVsbHM7Ci0gCWJvb2wJCWRv bmUgPSBmYWxzZTsKICAJYm9vbAkJaXNudWxsOwogIAlSZXN1bHRSZWxJbmZvICpyZXN1bHRS ZWxJbmZvOwogIAlFU3RhdGUJICAgKmVzdGF0ZSA9IENyZWF0ZUV4ZWN1dG9yU3RhdGUoKTsg LyogZm9yIEV4ZWNDb25zdHJhaW50cygpICovCi0tLSAxNDE0LDE0MTkgLS0tLQoqKioqKioq KioqKioqKioKKioqIDE0OTYsMTUwMSAqKioqCi0tLSAxNDI0LDE0MzQgLS0tLQogIAlFeHBy Q29udGV4dCAqZWNvbnRleHQ7CQkvKiB1c2VkIGZvciBFeGVjRXZhbEV4cHIgZm9yIGRlZmF1 bHQgYXR0cyAqLwogIAlNZW1vcnlDb250ZXh0IG9sZGNvbnRleHQgPSBDdXJyZW50TWVtb3J5 Q29udGV4dDsKICAJRXJyb3JDb250ZXh0Q2FsbGJhY2sgZXJyY29udGV4dDsKKyAgICAgaW50 CQkgICAqYXR0cl9vZmZzZXRzOworIAlpbnQJCQludWxsX3ByaW50X2xlbjsgLyogbGVuZ3Ro IG9mIG51bGwgcHJpbnQgKi8KKyAJYm9vbAkJbm9fbW9yZV9kYXRhOworIAlMaXN0Q2VsbCAg ICpjdXI7CisgCiAgCiAgCXR1cERlc2MgPSBSZWxhdGlvbkdldERlc2NyKHJlbCk7CiAgCWF0 dHIgPSB0dXBEZXNjLT5hdHRyczsKKioqKioqKioqKioqKioqCioqKiAxNjA3LDE2MTMgKioq KgogIAkJfQogIAl9CiAgCiEgCS8qIFByZXBhcmUgdG8gY2F0Y2ggQUZURVIgdHJpZ2dlcnMu ICovCiAgCUFmdGVyVHJpZ2dlckJlZ2luUXVlcnkoKTsKICAKICAJLyoKLS0tIDE1NDAsMTU0 OCAtLS0tCiAgCQl9CiAgCX0KICAKISAJLyoKISAJICogUHJlcGFyZSB0byBjYXRjaCBBRlRF UiB0cmlnZ2Vycy4KISAJICovCiAgCUFmdGVyVHJpZ2dlckJlZ2luUXVlcnkoKTsKICAKICAJ LyoKKioqKioqKioqKioqKioqCioqKiAxNjcxLDE2NzYgKioqKgotLS0gMTYwNiwxNjEyIC0t LS0KICAKICAJdmFsdWVzID0gKERhdHVtICopIHBhbGxvYyhudW1fcGh5c19hdHRycyAqIHNp emVvZihEYXR1bSkpOwogIAludWxscyA9IChjaGFyICopIHBhbGxvYyhudW1fcGh5c19hdHRy cyAqIHNpemVvZihjaGFyKSk7CisgCWF0dHJfb2Zmc2V0cyA9IChpbnQgKikgcGFsbG9jKG51 bV9waHlzX2F0dHJzICogc2l6ZW9mKGludCkpOwogIAogIAkvKiBNYWtlIHJvb20gZm9yIGEg UEFSQU1fRVhFQyB2YWx1ZSBmb3IgZG9tYWluIGNvbnN0cmFpbnQgY2hlY2tzICovCiAgCWlm IChoYXNDb25zdHJhaW50cykKKioqKioqKioqKioqKioqCioqKiAxNjkxLDE3MTIgKioqKgog IAllcnJjb250ZXh0LnByZXZpb3VzID0gZXJyb3JfY29udGV4dF9zdGFjazsKICAJZXJyb3Jf Y29udGV4dF9zdGFjayA9ICZlcnJjb250ZXh0OwogIAohIAkvKiBvbiBpbnB1dCBqdXN0IHRo cm93IHRoZSBoZWFkZXIgbGluZSBhd2F5ICovCiEgCWlmIChoZWFkZXJfbGluZSkKICAJewoh IAkJY29weV9saW5lbm8rKzsKISAJCWRvbmUgPSBDb3B5UmVhZExpbmUocXVvdGUsIGVzY2Fw ZSkgOwogIAl9CiAgCiEgCXdoaWxlICghZG9uZSkKICAJewogIAkJYm9vbAkJc2tpcF90dXBs ZTsKICAJCU9pZAkJCWxvYWRlZF9vaWQgPSBJbnZhbGlkT2lkOwogIAogIAkJQ0hFQ0tfRk9S X0lOVEVSUlVQVFMoKTsKICAKLSAJCWNvcHlfbGluZW5vKys7Ci0gCiAgCQkvKiBSZXNldCB0 aGUgcGVyLXR1cGxlIGV4cHJjb250ZXh0ICovCiAgCQlSZXNldFBlclR1cGxlRXhwckNvbnRl eHQoZXN0YXRlKTsKICAKLS0tIDE2MjcsMTY4MSAtLS0tCiAgCWVycmNvbnRleHQucHJldmlv dXMgPSBlcnJvcl9jb250ZXh0X3N0YWNrOwogIAllcnJvcl9jb250ZXh0X3N0YWNrID0gJmVy cmNvbnRleHQ7CiAgCiEgCS8qCiEgCSAqIGluaXRpYWxpemUgYnVmZmVyZWQgc2NhbiB2YXJp YWJsZXMuCiEgCSAqLwohIAlpZihjc3ZfbW9kZSkKISAJewohIAkgICAgaW5fcXVvdGUgPSBm YWxzZTsKISAJCWxhc3Rfd2FzX2VzYyA9IGZhbHNlOwohIAl9CiEgCiEgCW51bGxfcHJpbnRf bGVuID0gc3RybGVuKG51bGxfcHJpbnQpOwohIAkKISAJLyogU2V0IHVwIGRhdGEgYnVmZmVy IHRvIGhvbGQgYSBjaHVuayBvZiBkYXRhICovCiEgCU1lbVNldChpbnB1dF9idWYsICcgJywg Q09QWV9CVUZfU0laRSAqIHNpemVvZihjaGFyKSk7CiEgCWlucHV0X2J1ZltDT1BZX0JVRl9T SVpFXSA9ICdcMCc7CiEgCiEgCW5vX21vcmVfZGF0YSA9IGZhbHNlOwkJLyogbm8gbW9yZSBp bnB1dCBkYXRhIHRvIHJlYWQgZnJvbSBmaWxlIG9yIEZFICovCiEgCWxpbmVfZG9uZSA9IHRy dWU7CiEgCWJ1Zl9kb25lID0gZmFsc2U7CiEgCiEgCWRvCiAgCXsKISAJCXNpemVfdAkgICBi eXRlc3JlYWQgPSAwOwohIAkJCiEgCQkvKiByZWFkIGEgY2h1bmsgb2YgZGF0YSBpbnRvIHRo ZSBidWZmZXIgKi8KISAJCWlmICghYmluYXJ5KQohIAkJewohIAkJICAgYnl0ZXNyZWFkID0g Q29weUdldERhdGEoaW5wdXRfYnVmLCBDT1BZX0JVRl9TSVpFKTsKISAJCSAgIGJ1Zl9kb25l ID0gZmFsc2U7CiEgCiEgCQkgICAvKiBzZXQgYnVmZmVyIHBvaW50ZXJzIHRvIGJlZ2lubmlu ZyBvZiB0aGUgYnVmZmVyICovCiEgCQkgICBiZWdsb2MgPSBpbnB1dF9idWY7CiEgCQkgICBi dWZmZXJfaW5kZXggPSAwOwogIAl9CiAgCiEgCQkvKgohIAkJICogY29udGludWUgaWYgc29t ZSBieXRlcyB3ZXJlIHJlYWQgb3IgaWYgd2UgZGlkbid0IHJlYWNoIEVPRi4gaWYgd2UKISAJ CSAqIGJvdGggcmVhY2hlZCBFT0YgX2FuZF8gbm8gYnl0ZXMgd2VyZSByZWFkLCBxdWl0IHRo ZSBsb29wIHdlIGFyZQohIAkJICogZG9uZQohIAkJICovCiEgCQlpZiAoYnl0ZXNyZWFkID4g MCB8fCAhZmVfZW9mIHx8IGJpbmFyeSkKISAJCXsKISAKISAJCQl3aGlsZSAoIWJ1Zl9kb25l KQogIAl7CiAgCQlib29sCQlza2lwX3R1cGxlOwogIAkJT2lkCQkJbG9hZGVkX29pZCA9IElu dmFsaWRPaWQ7CiAgCiAgCQlDSEVDS19GT1JfSU5URVJSVVBUUygpOwogIAogIAkJLyogUmVz ZXQgdGhlIHBlci10dXBsZSBleHByY29udGV4dCAqLwogIAkJUmVzZXRQZXJUdXBsZUV4cHJD b250ZXh0KGVzdGF0ZSk7CiAgCioqKioqKioqKioqKioqKgoqKiogMTcxNiwxNzQ2ICoqKioK ICAJCS8qIEluaXRpYWxpemUgYWxsIHZhbHVlcyBmb3Igcm93IHRvIE5VTEwgKi8KICAJCU1l bVNldCh2YWx1ZXMsIDAsIG51bV9waHlzX2F0dHJzICogc2l6ZW9mKERhdHVtKSk7CiAgCQlN ZW1TZXQobnVsbHMsICduJywgbnVtX3BoeXNfYXR0cnMgKiBzaXplb2YoY2hhcikpOwogIAog IAkJaWYgKCFiaW5hcnkpCiAgCQl7Ci0gCQkJQ29weVJlYWRSZXN1bHQgcmVzdWx0ID0gTk9S TUFMX0FUVFI7Ci0gCQkJY2hhcgkgICAqc3RyaW5nOwotIAkJCUxpc3RDZWxsICAgKmN1cjsK ICAKICAJCQkvKiBBY3R1YWxseSByZWFkIHRoZSBsaW5lIGludG8gbWVtb3J5IGhlcmUgKi8K ISAJCQlkb25lID0gY3N2X21vZGUgPyAKISAJCQkJQ29weVJlYWRMaW5lKHF1b3RlLCBlc2Nh cGUpIDogQ29weVJlYWRMaW5lKE5VTEwsIE5VTEwpOwogIAogIAkJCS8qCiEgCQkJICogRU9G IGF0IHN0YXJ0IG9mIGxpbmUgbWVhbnMgd2UncmUgZG9uZS4gIElmIHdlIHNlZSBFT0YgYWZ0 ZXIKISAJCQkgKiBzb21lIGNoYXJhY3RlcnMsIHdlIGFjdCBhcyB0aG91Z2ggaXQgd2FzIG5l d2xpbmUgZm9sbG93ZWQKISAJCQkgKiBieSBFT0YsIGllLCBwcm9jZXNzIHRoZSBsaW5lIGFu ZCB0aGVuIGV4aXQgbG9vcCBvbiBuZXh0CiEgCQkJICogaXRlcmF0aW9uLgogIAkJCSAqLwoh IAkJCWlmIChkb25lICYmIGxpbmVfYnVmLmxlbiA9PSAwKQogIAkJCQlicmVhazsKICAKICAJ CQlpZiAoZmlsZV9oYXNfb2lkcykKICAJCQl7CiAgCQkJCS8qIGNhbid0IGJlIGluIENTViBt b2RlIGhlcmUgKi8KISAJCQkJc3RyaW5nID0gQ29weVJlYWRBdHRyaWJ1dGUoZGVsaW0sIG51 bGxfcHJpbnQsCiEgCQkJCQkJCQkJCSAgICZyZXN1bHQsICZpc251bGwpOwogIAogIAkJCQlp ZiAoaXNudWxsKQogIAkJCQkJZXJlcG9ydChFUlJPUiwKLS0tIDE2ODUsMTcyOSAtLS0tCiAg CQkvKiBJbml0aWFsaXplIGFsbCB2YWx1ZXMgZm9yIHJvdyB0byBOVUxMICovCiAgCQlNZW1T ZXQodmFsdWVzLCAwLCBudW1fcGh5c19hdHRycyAqIHNpemVvZihEYXR1bSkpOwogIAkJTWVt U2V0KG51bGxzLCAnbicsIG51bV9waHlzX2F0dHJzICogc2l6ZW9mKGNoYXIpKTsKKyAJCQkJ LyogcmVzZXQgYXR0cmlidXRlIHBvaW50ZXJzICovCisgCQkJCU1lbVNldChhdHRyX29mZnNl dHMsIDAsIG51bV9waHlzX2F0dHJzICogc2l6ZW9mKGludCkpOwogIAogIAkJaWYgKCFiaW5h cnkpCiAgCQl7CiAgCiAgCQkJLyogQWN0dWFsbHkgcmVhZCB0aGUgbGluZSBpbnRvIG1lbW9y eSBoZXJlICovCiEgCQkJCQlsaW5lX2RvbmUgPSBjc3ZfbW9kZSA/CiEgCQkJCQkJCQlDb3B5 UmVhZExpbmVDU1YoYnl0ZXNyZWFkLCBxdW90ZSwgZXNjYXBlKSA6IAohIAkJCQkJCQkJQ29w eVJlYWRMaW5lVGV4dChieXRlc3JlYWQsIGVzY2FwZSk7CiAgCiAgCQkJLyoKISAJCQkJCSAq IGlmIGZpbmlzaGVkIHByb2Nlc3NpbmcgZGF0YSBsaW5lIC0gaW5jcmVtZW50IGxpbmUgY291 bnQuCiEgCQkJCQkgKiBPdGhlcndpc2UsIGlmIGVvZiBpcyBub3QgeWV0IHJlYWNoZWQsIHdl IHNraXAgYXR0IHBhcnNpbmcgCiEgCQkJCQkgKiBhbmQgcmVhZCBtb3JlIGRhdGEuIEJ1dCBp ZiBlb2YgX3dhc18gcmVhY2hlZCBpdCBtZWFucyAKISAJCQkJCSAqIHRoYXQgdGhlIG9yaWdp bmFsIGxhc3QgZGF0YSBsaW5lIGlzIGRlZmVjdGl2ZSBhbmQgCiEgCQkJCQkgKiB3ZSB3YW50 IHRvIGNhdGNoIHRoYXQgZXJyb3IgbGF0ZXIgb24uCiAgCQkJICovCiEgCQkJCQlpZiAobGlu ZV9kb25lKQohIAkJCQkJCWNvcHlfbGluZW5vKys7CiEgCQkJCQllbHNlIGlmICghZmVfZW9m IHx8IGVuZF9tYXJrZXIgKQogIAkJCQlicmVhazsKICAKKyAJCQkJCWlmIChoZWFkZXJfbGlu ZSkKKyAJCQkJCXsKKyAJCQkJCQlsaW5lX2J1Zi5sZW4gPSAwOwkJLyogd2UgY2FuIHJlc2V0 IGxpbmUgYnVmZmVyIG5vdy4gKi8KKyAJCQkJCQlsaW5lX2J1Zi5kYXRhWzBdID0gJ1wwJzsK KyAJCQkJCQlsaW5lX2J1Zi5jdXJzb3IgPSAwOworIAkJCQkJCWhlYWRlcl9saW5lID0gZmFs c2U7CisgCQkJCQkJY29udGludWU7CisgCQkJCQl9CisgCQkJCQkKICAJCQlpZiAoZmlsZV9o YXNfb2lkcykKICAJCQl7CisgCQkJCQkJY2hhcgkgICAqb2lkX3N0cmluZzsKKyAKICAJCQkJ LyogY2FuJ3QgYmUgaW4gQ1NWIG1vZGUgaGVyZSAqLwohIAkJCQkJCW9pZF9zdHJpbmcgPSBD b3B5UmVhZE9pZEF0dHIoZGVsaW0sIG51bGxfcHJpbnQsIG51bGxfcHJpbnRfbGVuLAohIAkJ CQkJCQkJCQkJCQkgJmlzbnVsbCk7CiAgCiAgCQkJCWlmIChpc251bGwpCiAgCQkJCQllcmVw b3J0KEVSUk9SLAoqKioqKioqKioqKioqKioKKioqIDE3NTAsMTc1NiAqKioqCiAgCQkJCXsK ICAJCQkJCWNvcHlfYXR0bmFtZSA9ICJvaWQiOwogIAkJCQkJbG9hZGVkX29pZCA9IERhdHVt R2V0T2JqZWN0SWQoRGlyZWN0RnVuY3Rpb25DYWxsMShvaWRpbiwKISAJCQkJCQkJCQkJCSAg IENTdHJpbmdHZXREYXR1bShzdHJpbmcpKSk7CiAgCQkJCQlpZiAobG9hZGVkX29pZCA9PSBJ bnZhbGlkT2lkKQogIAkJCQkJCWVyZXBvcnQoRVJST1IsCiAgCQkJCQkJCQkoZXJyY29kZShF UlJDT0RFX0JBRF9DT1BZX0ZJTEVfRk9STUFUKSwKLS0tIDE3MzMsMTczOSAtLS0tCiAgCQkJ CXsKICAJCQkJCWNvcHlfYXR0bmFtZSA9ICJvaWQiOwogIAkJCQkJbG9hZGVkX29pZCA9IERh dHVtR2V0T2JqZWN0SWQoRGlyZWN0RnVuY3Rpb25DYWxsMShvaWRpbiwKISAJCQkJCQkJCQkJ CQkgICBDU3RyaW5nR2V0RGF0dW0ob2lkX3N0cmluZykpKTsKICAJCQkJCWlmIChsb2FkZWRf b2lkID09IEludmFsaWRPaWQpCiAgCQkJCQkJZXJlcG9ydChFUlJPUiwKICAJCQkJCQkJCShl cnJjb2RlKEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JNQVQpLAoqKioqKioqKioqKioqKioK KioqIDE3NTksMTc5MiAqKioqCiAgCQkJCX0KICAJCQl9CiAgCiEgCQkJLyogTG9vcCB0byBy ZWFkIHRoZSB1c2VyIGF0dHJpYnV0ZXMgb24gdGhlIGxpbmUuICovCiAgCQkJZm9yZWFjaChj dXIsIGF0dG51bWxpc3QpCiAgCQkJewogIAkJCQlpbnQJCQlhdHRudW0gPSBsZmlyc3RfaW50 KGN1cik7CiAgCQkJCWludAkJCW0gPSBhdHRudW0gLSAxOwogIAohIAkJCQkvKgohIAkJCQkg KiBJZiBwcmlvciBhdHRyIG9uIHRoaXMgbGluZSB3YXMgZW5kZWQgYnkgbmV3bGluZSwKISAJ CQkJICogY29tcGxhaW4uCiEgCQkJCSAqLwohIAkJCQlpZiAocmVzdWx0ICE9IE5PUk1BTF9B VFRSKQohIAkJCQkJZXJlcG9ydChFUlJPUiwKISAJCQkJCQkJKGVycmNvZGUoRVJSQ09ERV9C QURfQ09QWV9GSUxFX0ZPUk1BVCksCiEgCQkJCQkJCSBlcnJtc2coIm1pc3NpbmcgZGF0YSBm b3IgY29sdW1uIFwiJXNcIiIsCiEgCQkJCQkJCQkJTmFtZVN0cihhdHRyW21dLT5hdHRuYW1l KSkpKTsKICAKISAJCQkJaWYgKGNzdl9tb2RlKQohIAkJCQl7CiEgCQkJCQlzdHJpbmcgPSBD b3B5UmVhZEF0dHJpYnV0ZUNTVihkZWxpbSwgbnVsbF9wcmludCwgcXVvdGUsCiEgCQkJCQkJ CQkJCQkgICBlc2NhcGUsICZyZXN1bHQsICZpc251bGwpOwohIAkJCQkJaWYgKHJlc3VsdCA9 PSBVTlRFUk1JTkFURURfRklFTEQpCiEgCQkJCQkJZXJlcG9ydChFUlJPUiwKISAJCQkJCQkJ CShlcnJjb2RlKEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JNQVQpLAohIAkJCQkJCQkgICBl cnJtc2coInVudGVybWluYXRlZCBDU1YgcXVvdGVkIGZpZWxkIikpKTsKISAJCQkJfQogIAkJ CQllbHNlCiEgCQkJCQlzdHJpbmcgPSBDb3B5UmVhZEF0dHJpYnV0ZShkZWxpbSwgbnVsbF9w cmludCwKISAJCQkJCQkJCQkJCSAgICZyZXN1bHQsICZpc251bGwpOwogIAogIAkJCQlpZiAo Y3N2X21vZGUgJiYgaXNudWxsICYmIGZvcmNlX25vdG51bGxbbV0pCiAgCQkJCXsKLS0tIDE3 NDIsMTc3MCAtLS0tCiAgCQkJCX0KICAJCQl9CiAgCiEgCQkJCQkvKiBwYXJzZSBhbGwgdGhl IGF0dHJpYnV0ZSBpbiB0aGUgZGF0YSBsaW5lICovCiEgCQkJCQlpZihjc3ZfbW9kZSkKISAJ CQkJCQlDb3B5UmVhZEF0dHJpYnV0ZXNDU1YoZGVsaW0sIG51bGxfcHJpbnQsIHF1b3RlLCBl c2NhcGUsIG51bGxfcHJpbnRfbGVuLCAKISAJCQkJCQkJCQkJCW51bGxzLCBhdHRudW1saXN0 LCBhdHRyX29mZnNldHMsIG51bV9waHlzX2F0dHJzLCBhdHRyKTsKISAJCQkJCWVsc2UKISAJ CQkJCQlDb3B5UmVhZEF0dHJpYnV0ZXNUZXh0KGRlbGltLCBlc2NhcGUsIG51bGxfcHJpbnQs IG51bGxfcHJpbnRfbGVuLAohIAkJCQkJCQkJCQkJIG51bGxzLCBhdHRudW1saXN0LCBhdHRy X29mZnNldHMsIG51bV9waHlzX2F0dHJzLCBhdHRyKTsKISAKISAJCQkJCS8qCiEgCQkJCQkg KiBMb29wIHRvIHJlYWQgdGhlIHVzZXIgYXR0cmlidXRlcyBvbiB0aGUgbGluZS4KISAJCQkJ CSAqLwogIAkJCWZvcmVhY2goY3VyLCBhdHRudW1saXN0KQogIAkJCXsKICAJCQkJaW50CQkJ YXR0bnVtID0gbGZpcnN0X2ludChjdXIpOwogIAkJCQlpbnQJCQltID0gYXR0bnVtIC0gMTsK KyAJCQkJCQljaGFyICAgICAgICpzdHJpbmc7CiAgCiEgCQkJCQkJc3RyaW5nID0gYXR0cl9i dWYuZGF0YSArIGF0dHJfb2Zmc2V0c1ttXTsKICAKISAJCQkJCQlpZiAobnVsbHNbbV0gPT0g JyAnKQohIAkJCQkJCQlpc251bGwgPSBmYWxzZTsKICAJCQkJZWxzZQohIAkJCQkJCQlpc251 bGwgPSB0cnVlOwogIAogIAkJCQlpZiAoY3N2X21vZGUgJiYgaXNudWxsICYmIGZvcmNlX25v dG51bGxbbV0pCiAgCQkJCXsKKioqKioqKioqKioqKioqCioqKiAxODA2LDE4MjMgKioqKgog IAkJCQkJY29weV9hdHRuYW1lID0gTlVMTDsKICAJCQkJfQogIAkJCX0KLSAKLSAJCQkvKgot IAkJCSAqIENvbXBsYWluIGlmIHRoZXJlIGFyZSBtb3JlIGZpZWxkcyBvbiB0aGUgaW5wdXQg bGluZS4KLSAJCQkgKgotIAkJCSAqIFNwZWNpYWwgY2FzZTogaWYgd2UncmUgcmVhZGluZyBh IHplcm8tY29sdW1uIHRhYmxlLCB3ZSB3b24ndAotIAkJCSAqIHlldCBoYXZlIGNhbGxlZCBD b3B5UmVhZEF0dHJpYnV0ZSgpIGF0IGFsbDsgc28gbm8gZXJyb3IgaWYKLSAJCQkgKiBsaW5l IGlzIGVtcHR5LgotIAkJCSAqLwotIAkJCWlmIChyZXN1bHQgPT0gTk9STUFMX0FUVFIgJiYg bGluZV9idWYubGVuICE9IDApCi0gCQkJCWVyZXBvcnQoRVJST1IsCi0gCQkJCQkJKGVycmNv ZGUoRVJSQ09ERV9CQURfQ09QWV9GSUxFX0ZPUk1BVCksCi0gCQkJCQkgICBlcnJtc2coImV4 dHJhIGRhdGEgYWZ0ZXIgbGFzdCBleHBlY3RlZCBjb2x1bW4iKSkpOwogIAkJfQogIAkJZWxz ZQogIAkJewotLS0gMTc4NCwxNzg5IC0tLS0KKioqKioqKioqKioqKioqCioqKiAxODI4LDE4 MzQgKioqKgogIAkJCWZsZF9jb3VudCA9IENvcHlHZXRJbnQxNigpOwogIAkJCWlmIChDb3B5 R2V0RW9mKCkgfHwgZmxkX2NvdW50ID09IC0xKQogIAkJCXsKISAJCQkJZG9uZSA9IHRydWU7 CiAgCQkJCWJyZWFrOwogIAkJCX0KICAKLS0tIDE3OTQsMTgwMSAtLS0tCiAgCQkJZmxkX2Nv dW50ID0gQ29weUdldEludDE2KCk7CiAgCQkJaWYgKENvcHlHZXRFb2YoKSB8fCBmbGRfY291 bnQgPT0gLTEpCiAgCQkJewohIAkJCQkJCWJ1Zl9kb25lID0gdHJ1ZTsKISAJCQkJCQlub19t b3JlX2RhdGEgPSB0cnVlOwogIAkJCQlicmVhazsKICAJCQl9CiAgCioqKioqKioqKioqKioq KgoqKiogMTg0NSwxODUxICoqKioKICAJCQkJCURhdHVtR2V0T2JqZWN0SWQoQ29weVJlYWRC aW5hcnlBdHRyaWJ1dGUoMCwKICAJCQkJCQkJCQkJCQkJCQkgJm9pZF9pbl9mdW5jdGlvbiwK ICAJCQkJCQkJCQkJCQkJCQkgb2lkX3R5cGlvcGFyYW0sCi0gCQkJCQkJCQkJCQkJCQkJIC0x LAogIAkJCQkJCQkJCQkJCQkJCSAmaXNudWxsKSk7CiAgCQkJCWlmIChpc251bGwgfHwgbG9h ZGVkX29pZCA9PSBJbnZhbGlkT2lkKQogIAkJCQkJZXJlcG9ydChFUlJPUiwKLS0tIDE4MTIs MTgxNyAtLS0tCioqKioqKioqKioqKioqKgoqKiogMTg2NSwxODcxICoqKioKICAJCQkJdmFs dWVzW21dID0gQ29weVJlYWRCaW5hcnlBdHRyaWJ1dGUoaSwKICAJCQkJCQkJCQkJCQkJJmlu X2Z1bmN0aW9uc1ttXSwKICAJCQkJCQkJCQkJCQkJdHlwaW9wYXJhbXNbbV0sCi0gCQkJCQkJ CQkJCQkJCWF0dHJbbV0tPmF0dHR5cG1vZCwKICAJCQkJCQkJCQkJCQkJJmlzbnVsbCk7CiAg CQkJCW51bGxzW21dID0gaXNudWxsID8gJ24nIDogJyAnOwogIAkJCQljb3B5X2F0dG5hbWUg PSBOVUxMOwotLS0gMTgzMSwxODM2IC0tLS0KKioqKioqKioqKioqKioqCioqKiAxODg1LDE4 OTEgKioqKgogIAkJCQludWxsc1tkZWZtYXBbaV1dID0gJyAnOwogIAkJfQogIAohIAkJLyog TmV4dCBhcHBseSBhbnkgZG9tYWluIGNvbnN0cmFpbnRzICovCiAgCQlpZiAoaGFzQ29uc3Ry YWludHMpCiAgCQl7CiAgCQkJUGFyYW1FeGVjRGF0YSAqcHJtZGF0YSA9ICZlY29udGV4dC0+ ZWN4dF9wYXJhbV9leGVjX3ZhbHNbMF07Ci0tLSAxODUwLDE4NTggLS0tLQogIAkJCQludWxs c1tkZWZtYXBbaV1dID0gJyAnOwogIAkJfQogIAohIAkJCQkvKgohIAkJCQkgKiBOZXh0IGFw cGx5IGFueSBkb21haW4gY29uc3RyYWludHMKISAJCQkJICovCiAgCQlpZiAoaGFzQ29uc3Ry YWludHMpCiAgCQl7CiAgCQkJUGFyYW1FeGVjRGF0YSAqcHJtZGF0YSA9ICZlY29udGV4dC0+ ZWN4dF9wYXJhbV9leGVjX3ZhbHNbMF07CioqKioqKioqKioqKioqKgoqKiogMTkxMiwxOTI0 ICoqKioKICAJCQl9CiAgCQl9CiAgCiEgCQkvKiBBbmQgbm93IHdlIGNhbiBmb3JtIHRoZSBp bnB1dCB0dXBsZS4gKi8KICAJCXR1cGxlID0gaGVhcF9mb3JtdHVwbGUodHVwRGVzYywgdmFs dWVzLCBudWxscyk7CiAgCiAgCQlpZiAob2lkcyAmJiBmaWxlX2hhc19vaWRzKQogIAkJCUhl YXBUdXBsZVNldE9pZCh0dXBsZSwgbG9hZGVkX29pZCk7CiAgCiEgCQkvKiBUcmlnZ2VycyBh bmQgc3R1ZmYgbmVlZCB0byBiZSBpbnZva2VkIGluIHF1ZXJ5IGNvbnRleHQuICovCiAgCQlN ZW1vcnlDb250ZXh0U3dpdGNoVG8ob2xkY29udGV4dCk7CiAgCiAgCQlza2lwX3R1cGxlID0g ZmFsc2U7Ci0tLSAxODc5LDE4OTUgLS0tLQogIAkJCX0KICAJCX0KICAKISAJCQkJLyoKISAJ CQkJICogQW5kIG5vdyB3ZSBjYW4gZm9ybSB0aGUgaW5wdXQgdHVwbGUuCiEgCQkJCSAqLwog IAkJdHVwbGUgPSBoZWFwX2Zvcm10dXBsZSh0dXBEZXNjLCB2YWx1ZXMsIG51bGxzKTsKICAK ICAJCWlmIChvaWRzICYmIGZpbGVfaGFzX29pZHMpCiAgCQkJSGVhcFR1cGxlU2V0T2lkKHR1 cGxlLCBsb2FkZWRfb2lkKTsKICAKISAJCQkJLyoKISAJCQkJICogVHJpZ2dlcnMgYW5kIHN0 dWZmIG5lZWQgdG8gYmUgaW52b2tlZCBpbiBxdWVyeSBjb250ZXh0LgohIAkJCQkgKi8KICAJ CU1lbW9yeUNvbnRleHRTd2l0Y2hUbyhvbGRjb250ZXh0KTsKICAKICAJCXNraXBfdHVwbGUg PSBmYWxzZTsKKioqKioqKioqKioqKioqCioqKiAxOTQ1LDE5NTUgKioqKgogIAkJCS8qIFBs YWNlIHR1cGxlIGluIHR1cGxlIHNsb3QgKi8KICAJCQlFeGVjU3RvcmVUdXBsZSh0dXBsZSwg c2xvdCwgSW52YWxpZEJ1ZmZlciwgZmFsc2UpOwogIAohIAkJCS8qIENoZWNrIHRoZSBjb25z dHJhaW50cyBvZiB0aGUgdHVwbGUgKi8KICAJCQlpZiAocmVsLT5yZF9hdHQtPmNvbnN0cikK ICAJCQkJRXhlY0NvbnN0cmFpbnRzKHJlc3VsdFJlbEluZm8sIHNsb3QsIGVzdGF0ZSk7CiAg CiEgCQkJLyogT0ssIHN0b3JlIHRoZSB0dXBsZSBhbmQgY3JlYXRlIGluZGV4IGVudHJpZXMg Zm9yIGl0ICovCiAgCQkJc2ltcGxlX2hlYXBfaW5zZXJ0KHJlbCwgdHVwbGUpOwogIAogIAkJ CWlmIChyZXN1bHRSZWxJbmZvLT5yaV9OdW1JbmRpY2VzID4gMCkKLS0tIDE5MTYsMTkzMCAt LS0tCiAgCQkJLyogUGxhY2UgdHVwbGUgaW4gdHVwbGUgc2xvdCAqLwogIAkJCUV4ZWNTdG9y ZVR1cGxlKHR1cGxlLCBzbG90LCBJbnZhbGlkQnVmZmVyLCBmYWxzZSk7CiAgCiEgCQkJCQkv KgohIAkJCQkJICogQ2hlY2sgdGhlIGNvbnN0cmFpbnRzIG9mIHRoZSB0dXBsZQohIAkJCQkJ ICovCiAgCQkJaWYgKHJlbC0+cmRfYXR0LT5jb25zdHIpCiAgCQkJCUV4ZWNDb25zdHJhaW50 cyhyZXN1bHRSZWxJbmZvLCBzbG90LCBlc3RhdGUpOwogIAohIAkJCQkJLyoKISAJCQkJCSAq IE9LLCBzdG9yZSB0aGUgdHVwbGUgYW5kIGNyZWF0ZSBpbmRleCBlbnRyaWVzIGZvciBpdAoh IAkJCQkJICovCiAgCQkJc2ltcGxlX2hlYXBfaW5zZXJ0KHJlbCwgdHVwbGUpOwogIAogIAkJ CWlmIChyZXN1bHRSZWxJbmZvLT5yaV9OdW1JbmRpY2VzID4gMCkKKioqKioqKioqKioqKioq CioqKiAxOTU4LDE5NzYgKioqKgogIAkJCS8qIEFGVEVSIFJPVyBJTlNFUlQgVHJpZ2dlcnMg Ki8KICAJCQlFeGVjQVJJbnNlcnRUcmlnZ2Vycyhlc3RhdGUsIHJlc3VsdFJlbEluZm8sIHR1 cGxlKTsKICAJCX0KICAJfQogIAohIAkvKiBEb25lLCBjbGVhbiB1cCAqLwogIAllcnJvcl9j b250ZXh0X3N0YWNrID0gZXJyY29udGV4dC5wcmV2aW91czsKICAKICAJTWVtb3J5Q29udGV4 dFN3aXRjaFRvKG9sZGNvbnRleHQpOwogIAohIAkvKiBFeGVjdXRlIEFGVEVSIFNUQVRFTUVO VCBpbnNlcnRpb24gdHJpZ2dlcnMgKi8KICAJRXhlY0FTSW5zZXJ0VHJpZ2dlcnMoZXN0YXRl LCByZXN1bHRSZWxJbmZvKTsKICAKISAJLyogSGFuZGxlIHF1ZXVlZCBBRlRFUiB0cmlnZ2Vy cyAqLwogIAlBZnRlclRyaWdnZXJFbmRRdWVyeShlc3RhdGUpOwogIAogIAlwZnJlZSh2YWx1 ZXMpOwogIAlwZnJlZShudWxscyk7CiAgCi0tLSAxOTMzLDE5NzAgLS0tLQogIAkJCS8qIEFG VEVSIFJPVyBJTlNFUlQgVHJpZ2dlcnMgKi8KICAJCQlFeGVjQVJJbnNlcnRUcmlnZ2Vycyhl c3RhdGUsIHJlc3VsdFJlbEluZm8sIHR1cGxlKTsKICAJCX0KKyAKKyAJCQkJbGluZV9idWYu bGVuID0gMDsJCS8qIHdlIGNhbiByZXNldCBsaW5lIGJ1ZmZlciBub3cuICovCisgCQkJCWxp bmVfYnVmLmRhdGFbMF0gPSAnXDAnOworIAkJCQlsaW5lX2J1Zi5jdXJzb3IgPSAwOworIAkJ CX0JCQkJCS8qIGVuZCB3aGlsZSghYnVmX2RvbmUpICovCisgCQl9CQkJCQkJLyogZW5kIGlm IChieXRlc3JlYWQgPiAwIHx8ICFmZV9lb2YpICovCisgCQllbHNlCisgCQkJLyogbm8gYnl0 ZXMgcmVhZCwgZW5kIG9mIGRhdGEgKi8KKyAJCXsKKyAJCQlub19tb3JlX2RhdGEgPSB0cnVl OwogIAl9CisgCX0gd2hpbGUgKCFub19tb3JlX2RhdGEpOwogIAohIAkvKgohIAkgKiBEb25l LCBjbGVhbiB1cAohIAkgKi8KICAJZXJyb3JfY29udGV4dF9zdGFjayA9IGVycmNvbnRleHQu cHJldmlvdXM7CiAgCiAgCU1lbW9yeUNvbnRleHRTd2l0Y2hUbyhvbGRjb250ZXh0KTsKICAK ISAJLyogCSAKISAgICAgICogRXhlY3V0ZSBBRlRFUiBTVEFURU1FTlQgaW5zZXJ0aW9uIHRy aWdnZXJzCiEgCSAqLwogIAlFeGVjQVNJbnNlcnRUcmlnZ2Vycyhlc3RhdGUsIHJlc3VsdFJl bEluZm8pOwogIAohIAkvKgohIAkgKiBIYW5kbGUgcXVldWVkIEFGVEVSIHRyaWdnZXJzCiEg CSAqLwogIAlBZnRlclRyaWdnZXJFbmRRdWVyeShlc3RhdGUpOwogIAorIAlwZnJlZShhdHRy X29mZnNldHMpOworIAogIAlwZnJlZSh2YWx1ZXMpOwogIAlwZnJlZShudWxscyk7CiAgCioq KioqKioqKioqKioqKgoqKiogMTk5MCwyMjYyICoqKioKICAKICAKICAvKgohICAqIFJlYWQg dGhlIG5leHQgaW5wdXQgbGluZSBhbmQgc3Rhc2ggaXQgaW4gbGluZV9idWYsIHdpdGggY29u dmVyc2lvbiB0bwohICAqIHNlcnZlciBlbmNvZGluZy4KISAgKgohICAqIFJlc3VsdCBpcyB0 cnVlIGlmIHJlYWQgd2FzIHRlcm1pbmF0ZWQgYnkgRU9GLCBmYWxzZSBpZiB0ZXJtaW5hdGVk CiEgICogYnkgbmV3bGluZS4KICAgKi8KICBzdGF0aWMgYm9vbAohIENvcHlSZWFkTGluZShj aGFyICogcXVvdGUsIGNoYXIgKiBlc2NhcGUpCiAgewohIAlib29sCQlyZXN1bHQ7CiEgCWJv b2wJCWNoYW5nZV9lbmNvZGluZyA9IChjbGllbnRfZW5jb2RpbmcgIT0gc2VydmVyX2VuY29k aW5nKTsKISAJaW50CQkJYzsKISAJaW50CQkJbWJsZW47CiEgCWludAkJCWo7CiEgCXVuc2ln bmVkIGNoYXIgc1syXTsKICAJY2hhcgkgICAqY3Z0OwohIAlib29sICAgICAgICBpbl9xdW90 ZSA9IGZhbHNlLCBsYXN0X3dhc19lc2MgPSBmYWxzZSwgY3N2X21vZGUgPSBmYWxzZTsKISAJ Y2hhciAgICAgICAgcXVvdGVjID0gJ1wwJywgZXNjYXBlYyA9ICdcMCc7CiEgCiEgCWlmIChx dW90ZSkKISAJewohIAkJY3N2X21vZGUgPSB0cnVlOwohIAkJcXVvdGVjID0gcXVvdGVbMF07 CiEgCQllc2NhcGVjID0gZXNjYXBlWzBdOwohIAkJLyogaWdub3JlIHNwZWNpYWwgZXNjYXBl IHByb2Nlc3NpbmcgaWYgaXQncyB0aGUgc2FtZSBhcyBxdW90ZWMgKi8KISAJCWlmIChxdW90 ZWMgPT0gZXNjYXBlYykKISAJCQllc2NhcGVjID0gJ1wwJzsKISAJfQohIAohIAohIAlzWzFd ID0gMDsKISAKISAJLyogcmVzZXQgbGluZV9idWYgdG8gZW1wdHkgKi8KISAJbGluZV9idWYu bGVuID0gMDsKISAJbGluZV9idWYuZGF0YVswXSA9ICdcMCc7CiEgCWxpbmVfYnVmLmN1cnNv ciA9IDA7CiAgCiAgCS8qIG1hcmsgdGhhdCBlbmNvZGluZyBjb252ZXJzaW9uIGhhc24ndCBv Y2N1cnJlZCB5ZXQgKi8KICAJbGluZV9idWZfY29udmVydGVkID0gZmFsc2U7CiAgCiEgCS8q IHNldCBkZWZhdWx0IHN0YXR1cyAqLwohIAlyZXN1bHQgPSBmYWxzZTsKICAKICAJLyoKISAJ ICogSW4gdGhpcyBsb29wIHdlIG9ubHkgY2FyZSBmb3IgZGV0ZWN0aW5nIG5ld2xpbmVzIChc ciBhbmQvb3IgXG4pIGFuZAohIAkgKiB0aGUgZW5kLW9mLWNvcHkgbWFya2VyIChcLikuICAK ISAJICoKISAJICogSW4gVGV4dCBtb2RlLCBmb3IgYmFja3dhcmRzIGNvbXBhdGliaWxpdHkg d2UgYWxsb3cKISAJICogYmFja3NsYXNoZXMgdG8gZXNjYXBlIG5ld2xpbmUgY2hhcmFjdGVy cy4gIEJhY2tzbGFzaGVzIG90aGVyIHRoYW4KISAJICogdGhlIGVuZCBtYXJrZXIgZ2V0IHB1 dCBpbnRvIHRoZSBsaW5lX2J1Ziwgc2luY2UgQ29weVJlYWRBdHRyaWJ1dGUKISAJICogZG9l cyBpdHMgb3duIGVzY2FwZSBwcm9jZXNzaW5nLgkKISAJICoKISAJICogSW4gQ1NWIG1vZGUs IENSIGFuZCBOTCBpbnNpZGUgcSBxdW90ZWQgZmllbGQgYXJlIGp1c3QgcGFydCBvZiB0aGUK ISAJICogZGF0YSB2YWx1ZSBhbmQgYXJlIHB1dCBpbiBsaW5lX2J1Zi4gV2Uga2VlcCBqdXN0 IGVub3VnaCBzdGF0ZQohIAkgKiB0byBrbm93IGlmIHdlIGFyZSBjdXJyZW50bHkgaW4gYSBx dW90ZWQgZmllbGQgb3Igbm90LgohIAkgKgohIAkgKiBUaGVzZSBmb3VyIGNoYXJhY3RlcnMs IGFuZCBvbmx5IHRoZXNlIGZvdXIsIGFyZSBhc3N1bWVkIHRoZSBzYW1lIGluIAohIAkgKiBm cm9udGVuZCBhbmQgYmFja2VuZCBlbmNvZGluZ3MuCiEgCSAqCiEgCSAqIFdlIGRvIG5vdCBh c3N1bWUgdGhhdCBzZWNvbmQgYW5kIGxhdGVyIGJ5dGVzIG9mIGEgZnJvbnRlbmQKISAJICog bXVsdGlieXRlIGNoYXJhY3RlciBjb3VsZG4ndCBsb29rIGxpa2UgQVNDSUkgY2hhcmFjdGVy cy4KICAJICovCiEgCWZvciAoOzspCiAgCXsKISAJCWMgPSBDb3B5R2V0Q2hhcigpOwohIAkJ aWYgKGMgPT0gRU9GKQogIAkJewohIAkJCXJlc3VsdCA9IHRydWU7CiEgCQkJYnJlYWs7CiAg CQl9CiAgCiEgCQlpZiAoY3N2X21vZGUpCiAgCQl7CiAgCQkJLyogIAohIAkJCSAqIERlYWxp bmcgd2l0aCBxdW90ZXMgYW5kIGVzY2FwZXMgaGVyZSBpcyBtaWxkbHkgdHJpY2t5LiBJZiB0 aGUKISAJCQkgKiBxdW90ZSBjaGFyIGlzIGFsc28gdGhlIGVzY2FwZSBjaGFyLCB0aGVyZSdz IG5vIHByb2JsZW0gLSB3ZSAgCiEgCQkJICoganVzdCB1c2UgdGhlIGNoYXIgYXMgYSB0b2dn bGUuIElmIHRoZXkgYXJlIGRpZmZlcmVudCwgd2UgbmVlZAohIAkJCSAqIHRvIGVuc3VyZSB0 aGF0IHdlIG9ubHkgdGFrZSBhY2NvdW50IG9mIGFuIGVzY2FwZSBpbnNpZGUgYSBxdW90ZWQK ISAJCQkgKiBmaWVsZCBhbmQgaW1tZWRpYXRlbHkgcHJlY2VkaW5nIGEgcXVvdGUgY2hhciwg YW5kIG5vdCB0aGUKISAJCQkgKiBzZWNvbmQgaW4gYSBlc2NhcGUtZXNjYXBlIHNlcXVlbmNl LgogIAkJCSAqLyAKICAKISAJCQlpZiAoaW5fcXVvdGUgJiYgYyA9PSBlc2NhcGVjKQohIAkJ CQlsYXN0X3dhc19lc2MgPSAhIGxhc3Rfd2FzX2VzYzsKISAJCQlpZiAoYyA9PSBxdW90ZWMg JiYgISBsYXN0X3dhc19lc2MpCiEgCQkJCWluX3F1b3RlID0gISBpbl9xdW90ZTsKISAJCQlp ZiAoYyAhPSBlc2NhcGVjKQohIAkJCQlsYXN0X3dhc19lc2MgPSBmYWxzZTsKICAKICAJCQkv KgohIAkJCSAqIHVwZGF0aW5nIHRoZSBsaW5lIGNvdW50IGZvciBlbWJlZGRlZCBDUiBhbmQv b3IgTEYgY2hhcnMgaXMgCiEgCQkJICogbmVjZXNzYXJpbHkgYSBsaXR0bGUgZnJhZ2lsZSAt IHRoaXMgdGVzdCBpcyBwcm9iYWJseSBhYm91dCAKISAJCQkgKiB0aGUgYmVzdCB3ZSBjYW4g ZG8uCiAgCQkJICovIAohIAkJCWlmIChpbl9xdW90ZSAmJiBjID09IChlb2xfdHlwZSA9PSBF T0xfQ1IgPyAnXHInIDogJ1xuJykpIAohIAkJCQljb3B5X2xpbmVubysrOyAKISAJCX0KICAK ISAJCWlmICghaW5fcXVvdGUgJiYgYyA9PSAnXHInKQogIAkJewohIAkJCWlmIChlb2xfdHlw ZSA9PSBFT0xfTkwpCiAgCQkJewohIAkJCQlpZiAoISBjc3ZfbW9kZSkKISAJCQkJCWVyZXBv cnQoRVJST1IsCiEgCQkJCQkJCShlcnJjb2RlKEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JN QVQpLAohIAkJCQkJCQkgZXJybXNnKCJsaXRlcmFsIGNhcnJpYWdlIHJldHVybiBmb3VuZCBp biBkYXRhIiksCiEgCQkJCQkJCSBlcnJoaW50KCJVc2UgXCJcXHJcIiB0byByZXByZXNlbnQg Y2FycmlhZ2UgcmV0dXJuLiIpKSk7CiEgCQkJCWVsc2UKISAJCQkJCWVyZXBvcnQoRVJST1Is CiEgCQkJCQkJCShlcnJjb2RlKEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JNQVQpLAohIAkJ CQkJCQkgZXJybXNnKCJ1bnF1b3RlZCBjYXJyaWFnZSByZXR1cm4gZm91bmQgaW4gQ1NWIGRh dGEiKSwKISAJCQkJCQkJIGVycmhpbnQoIlVzZSBxdW90ZWQgQ1NWIGZpZWxkIHRvIHJlcHJl c2VudCBjYXJyaWFnZSByZXR1cm4uIikpKTsKICAJCQl9CiEgCQkJLyogQ2hlY2sgZm9yIFxy XG4gb24gZmlyc3QgbGluZSwgX2FuZF8gaGFuZGxlIFxyXG4uICovCiEgCQkJaWYgKGVvbF90 eXBlID09IEVPTF9VTktOT1dOIHx8IGVvbF90eXBlID09IEVPTF9DUk5MKQogIAkJCXsKISAJ CQkJaW50CQkJYzIgPSBDb3B5UGVla0NoYXIoKTsKICAKISAJCQkJaWYgKGMyID09ICdcbicp CiEgCQkJCXsKISAJCQkJCUNvcHlEb25lUGVlayhjMiwgdHJ1ZSk7CQkvKiBlYXQgbmV3bGlu ZSAqLwohIAkJCQkJZW9sX3R5cGUgPSBFT0xfQ1JOTDsKICAJCQkJfQogIAkJCQllbHNlCiAg CQkJCXsKISAJCQkJCS8qIGZvdW5kIFxyLCBidXQgbm8gXG4gKi8KICAJCQkJCWlmIChlb2xf dHlwZSA9PSBFT0xfQ1JOTCkKICAJCQkJCXsKISAJCQkJCQlpZiAoIWNzdl9tb2RlKQohIAkJ CQkJCQllcmVwb3J0KEVSUk9SLAohIAkJCQkJCQkJCShlcnJjb2RlKEVSUkNPREVfQkFEX0NP UFlfRklMRV9GT1JNQVQpLAohIAkJCQkJCQkJCSBlcnJtc2coImxpdGVyYWwgY2FycmlhZ2Ug cmV0dXJuIGZvdW5kIGluIGRhdGEiKSwKISAJCQkJCQkJCQkgZXJyaGludCgiVXNlIFwiXFxy XCIgdG8gcmVwcmVzZW50IGNhcnJpYWdlIHJldHVybi4iKSkpOwogIAkJCQkJCWVsc2UKISAJ CQkJCQkJZXJlcG9ydChFUlJPUiwKISAJCQkJCQkJCQkoZXJyY29kZShFUlJDT0RFX0JBRF9D T1BZX0ZJTEVfRk9STUFUKSwKISAJCQkJCQkJCQkgZXJybXNnKCJ1bnF1b3RlZCBjYXJyaWFn ZSByZXR1cm4gZm91bmQgaW4gZGF0YSIpLAohIAkJCQkJCQkJCSBlcnJoaW50KCJVc2UgcXVv dGVkIENTViBmaWVsZCB0byByZXByZXNlbnQgY2FycmlhZ2UgcmV0dXJuLiIpKSk7CiEgCiAg CQkJCQl9CiAgCiAgCQkJCQkvKgohIAkJCQkJICogaWYgd2UgZ290IGhlcmUsIGl0IGlzIHRo ZSBmaXJzdCBsaW5lIGFuZCB3ZSBkaWRuJ3QKISAJCQkJCSAqIGdldCBcbiwgc28gcHV0IGl0 IGJhY2sKICAJCQkJCSAqLwohIAkJCQkJQ29weURvbmVQZWVrKGMyLCBmYWxzZSk7CiEgCQkJ CQllb2xfdHlwZSA9IEVPTF9DUjsKICAJCQkJfQogIAkJCX0KLSAJCQlicmVhazsKICAJCX0K ISAJCWlmICghaW5fcXVvdGUgJiYgYyA9PSAnXG4nKQogIAkJewohIAkJCWlmIChlb2xfdHlw ZSA9PSBFT0xfQ1IgfHwgZW9sX3R5cGUgPT0gRU9MX0NSTkwpCiEgCQkJewohIAkJCQlpZiAo IWNzdl9tb2RlKQohIAkJCQkJZXJlcG9ydChFUlJPUiwKISAJCQkJCQkJKGVycmNvZGUoRVJS Q09ERV9CQURfQ09QWV9GSUxFX0ZPUk1BVCksCiEgCQkJCQkJCSBlcnJtc2coImxpdGVyYWwg bmV3bGluZSBmb3VuZCBpbiBkYXRhIiksCiEgCQkJCQkJCSBlcnJoaW50KCJVc2UgXCJcXG5c IiB0byByZXByZXNlbnQgbmV3bGluZS4iKSkpOwohIAkJCQllbHNlCiEgCQkJCQllcmVwb3J0 KEVSUk9SLAohIAkJCQkJCQkoZXJyY29kZShFUlJDT0RFX0JBRF9DT1BZX0ZJTEVfRk9STUFU KSwKISAJCQkJCQkJIGVycm1zZygidW5xdW90ZWQgbmV3bGluZSBmb3VuZCBpbiBkYXRhIiks CiEgCQkJCQkJCSBlcnJoaW50KCJVc2UgcXVvdGVkIENTViBmaWVsZCB0byByZXByZXNlbnQg bmV3bGluZS4iKSkpOwohIAkJCQkJCiAgCQkJfQotIAkJCWVvbF90eXBlID0gRU9MX05MOwot IAkJCWJyZWFrOwotIAkJfQogIAohIAkJaWYgKChsaW5lX2J1Zi5sZW4gPT0gMCB8fCAhY3N2 X21vZGUpICYmIGMgPT0gJ1xcJykKISAJCXsKISAJCQlpbnQgYzI7CiEgCQkJCiEgCQkJaWYg KGNzdl9tb2RlKQohIAkJCQljMiA9IENvcHlQZWVrQ2hhcigpOwohIAkJCWVsc2UKISAJCQkJ YzIgPSBjID0gQ29weUdldENoYXIoKTsKICAKISAJCQlpZiAoYzIgPT0gRU9GKQogIAkJCXsK ISAJCQkJcmVzdWx0ID0gdHJ1ZTsKISAJCQkJaWYgKGNzdl9tb2RlKQohIAkJCQkJQ29weURv bmVQZWVrKGMyLCB0cnVlKTsKICAJCQkJYnJlYWs7CiAgCQkJfQohIAkJCWlmIChjMiA9PSAn LicpCiEgCQkJewohIAkJCQlpZiAoY3N2X21vZGUpCiEgCQkJCQlDb3B5RG9uZVBlZWsoYzIs IHRydWUpOyAvKiBhbGxvdyBrZWVwIGNhbGxpbmcgR2V0Q2hhcigpICovCiAgCiEgCQkJCWlm IChlb2xfdHlwZSA9PSBFT0xfQ1JOTCkKISAJCQkJewohIAkJCQkJYyA9IENvcHlHZXRDaGFy KCk7CiEgCQkJCQlpZiAoYyA9PSAnXG4nKQohIAkJCQkJCWVyZXBvcnQoRVJST1IsCiEgCQkJ CQkJCQkoZXJyY29kZShFUlJDT0RFX0JBRF9DT1BZX0ZJTEVfRk9STUFUKSwKISAJCQkJCQkJ CSBlcnJtc2coImVuZC1vZi1jb3B5IG1hcmtlciBkb2VzIG5vdCBtYXRjaCBwcmV2aW91cyBu ZXdsaW5lIHN0eWxlIikpKTsKISAJCQkJCWlmIChjICE9ICdccicpCiEgCQkJCQkJZXJlcG9y dChFUlJPUiwKISAJCQkJCQkJCShlcnJjb2RlKEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JN QVQpLAohIAkJCQkJCQkJIGVycm1zZygiZW5kLW9mLWNvcHkgbWFya2VyIGNvcnJ1cHQiKSkp OwogIAkJCQl9Ci0gCQkJCWMgPSBDb3B5R2V0Q2hhcigpOwotIAkJCQlpZiAoYyAhPSAnXHIn ICYmIGMgIT0gJ1xuJykKLSAJCQkJCWVyZXBvcnQoRVJST1IsCi0gCQkJCQkJCShlcnJjb2Rl KEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JNQVQpLAotIAkJCQkJCQkgZXJybXNnKCJlbmQt b2YtY29weSBtYXJrZXIgY29ycnVwdCIpKSk7Ci0gCQkJCWlmICgoZW9sX3R5cGUgPT0gRU9M X05MICYmIGMgIT0gJ1xuJykgfHwKLSAJCQkJCShlb2xfdHlwZSA9PSBFT0xfQ1JOTCAmJiBj ICE9ICdcbicpIHx8Ci0gCQkJCQkoZW9sX3R5cGUgPT0gRU9MX0NSICYmIGMgIT0gJ1xyJykp Ci0gCQkJCQllcmVwb3J0KEVSUk9SLAotIAkJCQkJCQkoZXJyY29kZShFUlJDT0RFX0JBRF9D T1BZX0ZJTEVfRk9STUFUKSwKLSAJCQkJCQkJIGVycm1zZygiZW5kLW9mLWNvcHkgbWFya2Vy IGRvZXMgbm90IG1hdGNoIHByZXZpb3VzIG5ld2xpbmUgc3R5bGUiKSkpOwogIAogIAkJCQkv KgohIAkJCQkgKiBJbiBwcm90b2NvbCB2ZXJzaW9uIDMsIHdlIHNob3VsZCBpZ25vcmUgYW55 dGhpbmcgYWZ0ZXIKISAJCQkJICogXC4gdXAgdG8gdGhlIHByb3RvY29sIGVuZCBvZiBjb3B5 IGRhdGEuCShYWFggbWF5YmUKISAJCQkJICogYmV0dGVyIG5vdCB0byB0cmVhdCBcLiBhcyBz cGVjaWFsPykKICAJCQkJICovCiEgCQkJCWlmIChjb3B5X2Rlc3QgPT0gQ09QWV9ORVdfRkUp CiAgCQkJCXsKISAJCQkJCXdoaWxlIChjICE9IEVPRikKISAJCQkJCQljID0gQ29weUdldENo YXIoKTsKICAJCQkJfQohIAkJCQlyZXN1bHQgPSB0cnVlOwkvKiByZXBvcnQgRU9GICovCiAg CQkJCWJyZWFrOwogIAkJCX0KICAJCQkKISAJCQlpZiAoY3N2X21vZGUpCiEgCQkJCUNvcHlE b25lUGVlayhjMiwgZmFsc2UpOyAvKiBub3QgYSBkb3QsIHNvIHB1dCBpdCBiYWNrICovIAog IAkJCWVsc2UKISAJCQkJLyogbm90IEVPRiBtYXJrLCBzbyBlbWl0IFwgYW5kIGZvbGxvd2lu ZyBjaGFyIGxpdGVyYWxseSAqLwohIAkJCQlhcHBlbmRTdHJpbmdJbmZvQ2hhck1hY3JvKCZs aW5lX2J1ZiwgJ1xcJyk7CiAgCQl9CiAgCiEgCQlhcHBlbmRTdHJpbmdJbmZvQ2hhck1hY3Jv KCZsaW5lX2J1ZiwgYyk7CiAgCiAgCQkvKgohIAkJICogV2hlbiBjbGllbnQgZW5jb2Rpbmcg IT0gc2VydmVyLCBtdXN0IGJlIGNhcmVmdWwgdG8gcmVhZCB0aGUKISAJCSAqIGV4dHJhIGJ5 dGVzIG9mIGEgbXVsdGlieXRlIGNoYXJhY3RlciBleGFjdGx5LCBzaW5jZSB0aGUKISAJCSAq IGVuY29kaW5nIG1pZ2h0IG5vdCBlbnN1cmUgdGhleSBkb24ndCBsb29rIGxpa2UgQVNDSUku ICBXaGVuIHRoZQohIAkJICogZW5jb2RpbmdzIGFyZSB0aGUgc2FtZSwgd2UgbmVlZCBub3Qg ZG8gdGhpcywgc2luY2Ugbm8gc2VydmVyCiEgCQkgKiBlbmNvZGluZyB3ZSB1c2UgaGFzIEFT Q0lJLWxpa2UgZm9sbG93aW5nIGJ5dGVzLgogIAkJICovCiEgCQlpZiAoY2hhbmdlX2VuY29k aW5nKQohIAkJewohIAkJCXNbMF0gPSBjOwohIAkJCW1ibGVuID0gcGdfZW5jb2RpbmdfbWJs ZW4oY2xpZW50X2VuY29kaW5nLCBzKTsKISAJCQlmb3IgKGogPSAxOyBqIDwgbWJsZW47IGor KykKISAJCQl7CiEgCQkJCWMgPSBDb3B5R2V0Q2hhcigpOwohIAkJCQlpZiAoYyA9PSBFT0Yp CiAgCQkJCXsKISAJCQkJCXJlc3VsdCA9IHRydWU7CiAgCQkJCQlicmVhazsKICAJCQkJfQot IAkJCQlhcHBlbmRTdHJpbmdJbmZvQ2hhck1hY3JvKCZsaW5lX2J1ZiwgYyk7CiAgCQkJfQoh IAkJCWlmIChyZXN1bHQpCiEgCQkJCWJyZWFrOwkJCS8qIG91dCBvZiBvdXRlciBsb29wICov CiAgCQl9Ci0gCX0JCQkJCQkJLyogZW5kIG9mIG91dGVyIGxvb3AgKi8KICAKISAJLyogRG9u ZSByZWFkaW5nIHRoZSBsaW5lLiAgQ29udmVydCBpdCB0byBzZXJ2ZXIgZW5jb2RpbmcuICov CiEgCWlmIChjaGFuZ2VfZW5jb2RpbmcpCiAgCXsKICAJCWN2dCA9IChjaGFyICopIHBnX2Ns aWVudF90b19zZXJ2ZXIoKHVuc2lnbmVkIGNoYXIgKikgbGluZV9idWYuZGF0YSwKICAJCQkJ CQkJCQkJICAgbGluZV9idWYubGVuKTsKLS0tIDE5ODQsMjQyMiAtLS0tCiAgCiAgCiAgLyoK ISAgKiBGaW5kcyB0aGUgbmV4dCBURVhUIGxpbmUgdGhhdCBpcyBpbiB0aGUgaW5wdXQgYnVm ZmVyIGFuZCBsb2FkcyAKISAgKiBpdCBpbnRvIGxpbmVfYnVmLiBSZXR1cm5zIGFuIGluZGlj YXRpb24gaWYgdGhlIGxpbmUgdGhhdCB3YXMgcmVhZCAKISAgKiBpcyBjb21wbGV0ZSAoaWYg YW4gdW5lc2NhcGVkIGxpbmUtZW5kIHdhcyBlbmNvdW50ZXJlZCkuIElmIHdlIAohICAqIHJl YWNoZWQgdGhlIGVuZCBvZiBidWZmZXIgYmVmb3JlIHRoZSB3aG9sZSBsaW5lIHdhcyB3cml0 dGVuIGludG8gdGhlCiEgICogbGluZSBidWZmZXIgdGhlbiByZXR1cm5zIGZhbHNlLgogICAq LwogIHN0YXRpYyBib29sCiEgQ29weVJlYWRMaW5lVGV4dChzaXplX3QgYnl0ZXNyZWFkLCBj aGFyICplc2NhcGUpCiAgewohIAlpbnQJCQlsaW5lc2l6ZTsKISAJYm9vbAkJdHJhbnNjb2Rl ID0gKGNsaWVudF9lbmNvZGluZyAhPSBzZXJ2ZXJfZW5jb2RpbmcpOwogIAljaGFyCSAgICpj dnQ7CiEgCWNoYXIgICAgICAgIGVzY2FwZWMgPSAnXDAnOwogIAogIAkvKiBtYXJrIHRoYXQg ZW5jb2RpbmcgY29udmVyc2lvbiBoYXNuJ3Qgb2NjdXJyZWQgeWV0ICovCiAgCWxpbmVfYnVm X2NvbnZlcnRlZCA9IGZhbHNlOwogIAohIAkvKgohIAkgKiBzZXQgdGhlIGVzY2FwZSBjaGFy IGZvciB0ZXh0IGZvcm1hdCAoJ1xcJyBieSBkZWZhdWx0KS4KISAJICovCiEgCWVzY2FwZWMg PSBlc2NhcGVbMF07CiAgCiAgCS8qCiEgCSAqIERldGVjdCBlbmQgb2YgbGluZSB0eXBlIGlm IG5vdCBhbHJlYWR5IGRldGVjdGVkLgogIAkgKi8KISAJaWYgKGVvbF90eXBlID09IEVPTF9V TktOT1dOKQogIAl7CiEgCQlib29sCQllb2xfZGV0ZWN0ZWQgPSBEZXRlY3RMaW5lRW5kKGJ5 dGVzcmVhZCwgTlVMTCwgZXNjYXBlKTsKISAKISAJCWlmICghZW9sX2RldGVjdGVkKQogIAkJ ewohIAkJCS8qIGxvYWQgZW50aXJlIGlucHV0IGJ1ZmZlciBpbnRvIGxpbmUgYnVmLCBhbmQg cXVpdCAqLwohIAkJCWFwcGVuZEJpbmFyeVN0cmluZ0luZm8oJmxpbmVfYnVmLCBpbnB1dF9i dWYsIENPUFlfQlVGX1NJWkUpOwohIAkJCWJ1Zl9kb25lID0gdHJ1ZTsKISAJCQlyZXR1cm4g ZmFsc2U7CiEgCQl9CiAgCQl9CiAgCiEgCS8qCiEgCSAqIFNwZWNpYWwgY2FzZTogZW9sIGlz IENSTkwsIGxhc3QgYnl0ZSBvZiBwcmV2aW91cyBidWZmZXIgd2FzIGFuCiEgCSAqIHVuZXNj YXBlZCBDUiBhbmQgMXN0IGJ5dGUgb2YgY3VycmVudCBidWZmZXIgaXMgTkwuIFdlIGNoZWNr IGZvcgohIAkgKiB0aGF0IGhlcmUuCiEgCSAqLwohIAlpZiAoZW9sX3R5cGUgPT0gRU9MX0NS TkwpCiEgCXsKISAJCS8qIGlmIHdlIHN0YXJ0ZWQgc2Nhbm5pbmcgZnJvbSB0aGUgMXN0IGJ5 dGUgb2YgdGhlIGJ1ZmZlciAqLwohIAkJaWYgKGJlZ2xvYyA9PSBpbnB1dF9idWYpCiEgCQl7 CiEgCQkJLyogYW5kIGhhZCBhIENSIGluIGxhc3QgYnl0ZSBvZiBwcmV2IGJ1ZiAqLwohIAkJ CWlmIChjcl9pbl9wcmV2YnVmKQohIAkJCXsKISAJCQkJLyoKISAJCQkJICogaWYgdGhpcyAx c3QgYnl0ZSBpbiBidWZmZXIgaXMgMm5kIGJ5dGUgb2YgbGluZSBlbmQgc2VxdWVuY2UKISAJ CQkJICogKGxpbmVmZWVkKQohIAkJCQkgKi8KISAJCQkJaWYgKCpiZWdsb2MgPT0gZW9sX2No WzFdKQogIAkJewogIAkJCS8qICAKISAJCQkJCSAqIGxvYWQgdGhhdCBvbmUgbGluZWZlZWQg Ynl0ZSBhbmQgaW5kaWNhdGUgd2UgYXJlIGRvbmUKISAJCQkJCSAqIHdpdGggdGhlIGRhdGEg bGluZQogIAkJCSAqLyAKKyAJCQkJCWFwcGVuZEJpbmFyeVN0cmluZ0luZm8oJmxpbmVfYnVm LCBiZWdsb2MsIDEpOworIAkJCQkJYnVmZmVyX2luZGV4Kys7CisgCQkJCQliZWdsb2MrKzsK KyAJCQkJCWVzY19pbl9wcmV2YnVmID0gZmFsc2U7CisgCQkJCQljcl9pbl9wcmV2YnVmID0g ZmFsc2U7CisgCQkJCQlyZXR1cm4gdHJ1ZTsKKyAJCQkJfQorIAkJCX0KICAKISAJCQljcl9p bl9wcmV2YnVmID0gZmFsc2U7CiEgCQl9CiEgCX0KICAKICAJCQkvKgohIAkgKiAod2UgbmVl ZCBhIGxvb3Agc28gdGhhdCBpZiBlb2xfY2ggaXMgZm91bmQsIGJ1dCBwcmV2IGNoIGlzIGJh Y2tzbGFzaCwKISAJICogd2UgY2FuIHNlYXJjaCBmb3IgdGhlIG5leHQgZW9sX2NoKQogIAkJ CSAqLyAKISAJd2hpbGUgKHRydWUpCiEgCXsKISAJCS8qIHJlYWNoZWQgZW5kIG9mIGJ1ZmZl ciAqLwohIAkJaWYgKChlbmRsb2MgPSBzY2FuVGV4dExpbmUoYmVnbG9jLCBlb2xfY2hbMF0s IGJ5dGVzcmVhZCAtIGJ1ZmZlcl9pbmRleCkpID09IE5VTEwpCiEgCQl7CiEgCQkJbGluZXNp emUgPSBDT1BZX0JVRl9TSVpFIC0gKGJlZ2xvYyAtIGlucHV0X2J1Zik7CiEgCQkJYXBwZW5k QmluYXJ5U3RyaW5nSW5mbygmbGluZV9idWYsIGJlZ2xvYywgbGluZXNpemUpOwogIAohIAkJ CWlmIChsaW5lX2J1Zi5sZW4gPiAxKQogIAkJewohIAkJCQljaGFyCSAgICpsYXN0X2NoID0g bGluZV9idWYuZGF0YSArIGxpbmVfYnVmLmxlbiAtIDE7IC8qIGJlZm9yZSB0ZXJtaW5hdGlu ZyBcMCAqLwohIAohIAkJCQlpZiAoKmxhc3RfY2ggPT0gZXNjYXBlYykKICAJCQl7CiEgCQkJ CQllc2NfaW5fcHJldmJ1ZiA9IHRydWU7CiEgCiEgCQkJCQlpZiAobGluZV9idWYubGVuID4g MikKISAJCQkJCXsKISAJCQkJCQlsYXN0X2NoLS07CiEgCQkJCQkJaWYgKCpsYXN0X2NoID09 IGVzY2FwZWMpCiEgCQkJCQkJCWVzY19pbl9wcmV2YnVmID0gZmFsc2U7CiAgCQkJfQohIAkJ CQl9CiEgCQkJCWVsc2UgaWYgKCpsYXN0X2NoID09ICdccicpCiAgCQkJewohIAkJCQkJaWYg KGVvbF90eXBlID09IEVPTF9DUk5MKQohIAkJCQkJCWNyX2luX3ByZXZidWYgPSB0cnVlOwoh IAkJCQl9CiEgCQkJfQogIAohIAkJCWxpbmVfZG9uZSA9IGZhbHNlOwohIAkJCWJ1Zl9kb25l ID0gdHJ1ZTsKISAJCQlicmVhazsKICAJCQkJfQogIAkJCQllbHNlCisgCQkJLyogZm91bmQg dGhlIDFzdCBlb2wgY2ggaW4gaW5wdXRfYnVmLiAqLwogIAkJCQl7CiEgCQkJYm9vbAkJZW9s X2ZvdW5kID0gdHJ1ZTsKISAJCQlib29sCQllb2xfZXNjYXBlZCA9IHRydWU7CiEgCiEgCQkJ LyoKISAJCQkgKiBMb2FkIHRoYXQgcGllY2Ugb2YgZGF0YSAocG90ZW50aWFsbHkgYSBkYXRh IGxpbmUpIGludG8gdGhlIGxpbmUgYnVmZmVyLAohIAkJCSAqIGFuZCB1cGRhdGUgdGhlIHBv aW50ZXJzIGZvciB0aGUgbmV4dCBzY2FuLgohIAkJCSAqLwohIAkJCWxpbmVzaXplID0gZW5k bG9jIC0gYmVnbG9jICsgMTsKISAJCQlhcHBlbmRCaW5hcnlTdHJpbmdJbmZvKCZsaW5lX2J1 ZiwgYmVnbG9jLCBsaW5lc2l6ZSk7CiEgCQkJYnVmZmVyX2luZGV4ICs9IGxpbmVzaXplOwoh IAkJCWJlZ2xvYyA9IGVuZGxvYyArIDE7CiEgCiAgCQkJCQlpZiAoZW9sX3R5cGUgPT0gRU9M X0NSTkwpCiAgCQkJCQl7CiEgCQkJCS8qIGNoZWNrIGlmIHRoZXJlIGlzIGEgJ1xuJyBhZnRl ciB0aGUgJ1xyJyAqLwohIAkJCQlpZiAoKihlbmRsb2MgKyAxKSA9PSAnXG4nKQohIAkJCQl7 CiEgCQkJCQkvKiB0aGlzIGlzIGEgbGluZSBlbmQgKi8KISAJCQkJCWFwcGVuZEJpbmFyeVN0 cmluZ0luZm8oJmxpbmVfYnVmLCBiZWdsb2MsIDEpOwkJLyogbG9hZCB0aGF0ICdcbicgKi8K ISAJCQkJCWJ1ZmZlcl9pbmRleCsrOwohIAkJCQkJYmVnbG9jKys7CiEgCQkJCX0KICAJCQkJ CQllbHNlCiEgCQkJCQkvKiBqdXN0IGEgQ1IsIG5vdCBhIGxpbmUgZW5kICovCiEgCQkJCQll b2xfZm91bmQgPSBmYWxzZTsKICAJCQkJCX0KICAKICAJCQkJCS8qCiEgCQkJICogaW4gc29t ZSBjYXNlcywgdGhpcyBlbmQgb2YgbGluZSBjaGFyIGhhcHBlbnMgdG8gYmUgdGhlCiEgCQkJ ICogbGFzdCBjaGFyYWN0ZXIgaW4gdGhlIGJ1ZmZlci4gd2UgbmVlZCB0byBjYXRjaCB0aGF0 LgogIAkJCQkJICovCiEgCQkJaWYgKGJ1ZmZlcl9pbmRleCA+PSBieXRlc3JlYWQpCiEgCQkJ CWJ1Zl9kb25lID0gdHJ1ZTsKISAKISAJCQkvKgohIAkJCSAqIENoZWNrIGlmIHRoZSAxc3Qg ZW5kIG9mIGxpbmUgY2ggaXMgZXNjYXBlZC4KISAJCQkgKi8KISAJCQlpZiAoZW5kbG9jICE9 IGlucHV0X2J1ZikJLyogY2FuIHdlIGxvb2sgMSBjaGFyIGJhY2s/ICovCiEgCQkJewohIAkJ CQlpZiAoKihlbmRsb2MgLSAxKSAhPSBlc2NhcGVjKQkvKiBwcmV2IGNoYXIgaXMgbm90IGFu IGVzY2FwZSAqLwohIAkJCQkJZW9sX2VzY2FwZWQgPSBmYWxzZTsKISAJCQkJZWxzZQohIAkJ CQkJLyogcHJldiBjaGFyIGlzIGFuIGVzY2FwZSAqLwohIAkJCQl7CiEgCQkJCQlpZiAoZW5k bG9jICE9IChpbnB1dF9idWYgKyAxKSkJCS8qIGNhbiB3ZSBsb29rIGFub3RoZXIKISAJCQkJ CQkJCQkJCQkJCSAqIGNoYXIgYmFjaz8gKi8KISAJCQkJCXsKISAJCQkJCQkvKiBpdCdzIGEg ZG91YmxlIGVzY2FwZSBjaGFyLCBzbyBpdCdzIG5vdCBhbiBlc2NhcGUgKi8KISAJCQkJCQlp ZiAoKihlbmRsb2MgLSAyKSA9PSBlc2NhcGVjKQohIAkJCQkJCQllb2xfZXNjYXBlZCA9IGZh bHNlOwohIAkJCQkJCS8qIGVsc2UgaXQncyBhIHNpbmdsZSBlc2NhcGUgY2hhciwgc28gRU9M IGlzIGFzY2FwZWQgKi8KISAJCQkJCX0KISAJCQkJCWVsc2UKISAJCQkJCXsKISAJCQkJCQkv KiB3ZSBuZWVkIHRvIGNoZWNrIGluIHRoZSBsYXN0IGJ1ZmZlciAqLwohIAkJCQkJCWlmIChl c2NfaW5fcHJldmJ1ZikJCQohIAkJCQkJCQllb2xfZXNjYXBlZCA9IGZhbHNlOwogIAkJCQl9 CiAgCQkJfQogIAkJfQohIAkJCWVsc2UKISAJCQkJLyoKISAJCQkJICogdGhpcyBlb2wgY2gg aXMgZmlyc3QgY2ggaW4gYnVmZmVyLCBjaGVjayBmb3IgZXNjYXBlIGluIHByZXYgYnVmCiEg CQkJCSAqLwogIAkJewohIAkJCQlpZiAoIWVzY19pbl9wcmV2YnVmKQohIAkJCQkJZW9sX2Vz Y2FwZWQgPSBmYWxzZTsKICAJCQl9CiAgCiEgCQkJZXNjX2luX3ByZXZidWYgPSBmYWxzZTsJ CS8qIHJlc2V0IHZhcmlhYmxlICovCiAgCiEgCQkJLyoKISAJCQkgKiBpZiBlb2wgd2FzIGZv dW5kLCBhbmQgaXQgaXNuJ3QgZXNjYXBlZCwgbGluZSBpcyBkb25lCiEgCQkJICovCiEgCQkJ aWYgKChlb2xfZXNjYXBlZCA9PSBmYWxzZSkgJiYgZW9sX2ZvdW5kKQogIAkJCXsKISAJCQkJ bGluZV9kb25lID0gdHJ1ZTsKICAJCQkJYnJlYWs7CiAgCQkJfQohIAkJCQllbHNlCiEgCQkJ CS8qIHN0YXkgaW4gdGhlIGxvb3AgYW5kIHByb2Nlc3Mgc29tZSBtb3JlIGRhdGEuICovCiEg CQkJCWxpbmVfZG9uZSA9IGZhbHNlOwohIAkJCQkJCiEgCQl9CQkJCQkJLyogZW5kIG9mIGZv dW5kIGVvbF9jaCAqLwohIAkJfQogIAohIAkvKgohIAkgKiBEb25lIHJlYWRpbmcgdGhlIGxp bmUuIENvbnZlcnQgaXQgdG8gc2VydmVyIGVuY29kaW5nLgohIAkgKi8KISAJaWYgKHRyYW5z Y29kZSkKISAJewohIAkJY3Z0ID0gKGNoYXIgKikgcGdfY2xpZW50X3RvX3NlcnZlcigodW5z aWduZWQgY2hhciAqKSBsaW5lX2J1Zi5kYXRhLAohIAkJCQkJCQkJCQkgICBsaW5lX2J1Zi5s ZW4pOwohIAkJaWYgKGN2dCAhPSBsaW5lX2J1Zi5kYXRhKQohIAkJewohIAkJCS8qIHRyYW5z ZmVyIGNvbnZlcnRlZCBkYXRhIGJhY2sgdG8gbGluZV9idWYgKi8KISAJCQlsaW5lX2J1Zi5s ZW4gPSAwOwohIAkJCWxpbmVfYnVmLmRhdGFbMF0gPSAnXDAnOwohIAkJCWxpbmVfYnVmLmN1 cnNvciA9IDA7CiEgCQkJYXBwZW5kQmluYXJ5U3RyaW5nSW5mbygmbGluZV9idWYsIGN2dCwg c3RybGVuKGN2dCkpOwohIAkJfQohIAl9CiEgCQkJCiEgCS8qIGluZGljYXRlIHRoYXQgY29u dmVyc2lvbiBoYWQgb2NjdXJlZCAqLwohIAlsaW5lX2J1Zl9jb252ZXJ0ZWQgPSB0cnVlOwoh IAohIAkvKgohIAkgKiBjaGVjayBpZiB0aGlzIGxpbmUgaXMgYW4gZW5kIG1hcmtlciAtLSAi XC4iCiEgCSAqLwohIAllbmRfbWFya2VyID0gZmFsc2U7CiEgCiEgCXN3aXRjaCAoZW9sX3R5 cGUpCiEgCQkJewohIAkJY2FzZSBFT0xfTkw6CiEgCQkJaWYgKCFzdHJjbXAobGluZV9idWYu ZGF0YSwgIlxcLlxuIikpCiEgCQkJCWVuZF9tYXJrZXIgPSB0cnVlOwohIAkJCWJyZWFrOwoh IAkJY2FzZSBFT0xfQ1I6CiEgCQkJaWYgKCFzdHJjbXAobGluZV9idWYuZGF0YSwgIlxcLlxy IikpCiEgCQkJCWVuZF9tYXJrZXIgPSB0cnVlOwohIAkJCWJyZWFrOwohIAkJY2FzZSBFT0xf Q1JOTDoKISAJCQlpZiAoIXN0cmNtcChsaW5lX2J1Zi5kYXRhLCAiXFwuXHJcbiIpKQohIAkJ CQllbmRfbWFya2VyID0gdHJ1ZTsKISAJCQlicmVhazsKISAJCWNhc2UgRU9MX1VOS05PV046 CiEgCQkJCWJyZWFrOwohIAkJCX0KISAKISAJaWYgKGVuZF9tYXJrZXIpCiEgCQkJewohIAkJ ZmVfZW9mID0gdHJ1ZTsKISAJCS8qIHdlIGRvbid0IHdhbnQgdG8gcHJvY2VzcyBhIFwuIGFz IGRhdGEgbGluZSwgd2FudCB0byBxdWl0LiAqLwohIAkJbGluZV9kb25lID0gZmFsc2U7CiEg CQlidWZfZG9uZSA9IHRydWU7CiEgCX0KISAKISAJcmV0dXJuIGxpbmVfZG9uZTsKISB9CiEg CiEgLyoKISAgKiBGaW5kcyB0aGUgbmV4dCBDU1YgbGluZSB0aGF0IGlzIGluIHRoZSBpbnB1 dCBidWZmZXIgYW5kIGxvYWRzIAohICAqIGl0IGludG8gbGluZV9idWYuIFJldHVybnMgYW4g aW5kaWNhdGlvbiBpZiB0aGUgbGluZSB0aGF0IHdhcyByZWFkIAohICAqIGlzIGNvbXBsZXRl IChpZiBhbiB1bmVzY2FwZWQgbGluZS1lbmQgd2FzIGVuY291bnRlcmVkKS4gSWYgd2UgCiEg ICogcmVhY2hlZCB0aGUgZW5kIG9mIGJ1ZmZlciBiZWZvcmUgdGhlIHdob2xlIGxpbmUgd2Fz IHdyaXR0ZW4gaW50byB0aGUKISAgKiBsaW5lIGJ1ZmZlciB0aGVuIHJldHVybnMgZmFsc2Uu CiEgICovCiEgc3RhdGljIGJvb2wKISBDb3B5UmVhZExpbmVDU1Yoc2l6ZV90IGJ5dGVzcmVh ZCwgY2hhciAqcXVvdGUsIGNoYXIgKmVzY2FwZSkKISB7CiEgCWludAkJCWxpbmVzaXplOwoh IAljaGFyCSAgICpjdnQ7CiEgCWNoYXIgICAgICAgIHF1b3RlYyA9ICdcMCcsCiEgICAgICAg ICAgICAgICAgIGVzY2FwZWMgPSAnXDAnOwohIAlib29sCQl0cmFuc2NvZGUgPSAoY2xpZW50 X2VuY29kaW5nICE9IHNlcnZlcl9lbmNvZGluZyk7CiEgCQohIAkvKiBtYXJrIHRoYXQgZW5j b2RpbmcgY29udmVyc2lvbiBoYXNuJ3Qgb2NjdXJyZWQgeWV0ICovCiEgCWxpbmVfYnVmX2Nv bnZlcnRlZCA9IGZhbHNlOwohIAkKISAJZXNjYXBlYyA9IGVzY2FwZVswXTsKISAJcXVvdGVj ID0gcXVvdGVbMF07CiEgCQohICAgICAvKiBpZ25vcmUgc3BlY2lhbCBlc2NhcGUgcHJvY2Vz c2luZyBpZiBpdCdzIHRoZSBzYW1lIGFzIHF1b3RlYyAqLwohIAlpZiAocXVvdGVjID09IGVz Y2FwZWMpCiEgCQllc2NhcGVjID0gJ1wwJzsKISAKISAJLyoKISAJICogRGV0ZWN0IGVuZCBv ZiBsaW5lIHR5cGUgaWYgbm90IGFscmVhZHkgZGV0ZWN0ZWQuCiEgCSAqLwohIAlpZiAoZW9s X3R5cGUgPT0gRU9MX1VOS05PV04pCiEgCXsKISAJCWJvb2wJCWVvbF9kZXRlY3RlZCA9IERl dGVjdExpbmVFbmQoYnl0ZXNyZWFkLCBxdW90ZSwgZXNjYXBlKTsKISAJCQohIAkJaWYgKCFl b2xfZGV0ZWN0ZWQpCiEgCQl7CiEgCQkJLyogbG9hZCBlbnRpcmUgaW5wdXQgYnVmZmVyIGlu dG8gbGluZSBidWYsIGFuZCBxdWl0ICovCiEgCQkJYXBwZW5kQmluYXJ5U3RyaW5nSW5mbygm bGluZV9idWYsIGlucHV0X2J1ZiwgQ09QWV9CVUZfU0laRSk7CiEgCQkJbGluZV9kb25lID0g ZmFsc2U7CiEgCQkJYnVmX2RvbmUgPSB0cnVlOwohIAkJCQohIAkJCXJldHVybiBsaW5lX2Rv bmU7CiEgCQl9CiEgCX0KISAJCiEgCS8qCiEgCSAqIFNwZWNpYWwgY2FzZTogZW9sIGlzIENS TkwsIGxhc3QgYnl0ZSBvZiBwcmV2aW91cyBidWZmZXIgd2FzIGFuCiEgCSAqIHVuZXNjYXBl ZCBDUiBhbmQgMXN0IGJ5dGUgb2YgY3VycmVudCBidWZmZXIgaXMgTkwuIFdlIGNoZWNrIGZv cgohIAkgKiB0aGF0IGhlcmUuCiEgCSAqLwohIAkJCQlpZiAoZW9sX3R5cGUgPT0gRU9MX0NS TkwpCiEgCQkJCXsKISAJCS8qIGlmIHdlIHN0YXJ0ZWQgc2Nhbm5pbmcgZnJvbSB0aGUgMXN0 IGJ5dGUgb2YgdGhlIGJ1ZmZlciAqLwohIAkJaWYgKGJlZ2xvYyA9PSBpbnB1dF9idWYpCiEg CQl7CiEgCQkJLyogYW5kIGhhZCBhIENSIGluIGxhc3QgYnl0ZSBvZiBwcmV2IGJ1ZiAqLwoh IAkJCWlmIChjcl9pbl9wcmV2YnVmKQohIAkJCXsKISAJCQkJLyoKISAJCQkJICogaWYgdGhp cyAxc3QgYnl0ZSBpbiBidWZmZXIgaXMgMm5kIGJ5dGUgb2YgbGluZSBlbmQgc2VxdWVuY2UK ISAJCQkJICogKGxpbmVmZWVkKQohIAkJCQkgKi8KISAJCQkJaWYgKCpiZWdsb2MgPT0gZW9s X2NoWzFdKQohIAkJCQl7CiEgCQkJCQkvKgohIAkJCQkJICogbG9hZCB0aGF0IG9uZSBsaW5l ZmVlZCBieXRlIGFuZCBpbmRpY2F0ZSB3ZSBhcmUgZG9uZQohIAkJCQkJICogd2l0aCB0aGUg ZGF0YSBsaW5lCiEgCQkJCQkgKi8KISAJCQkJCWFwcGVuZEJpbmFyeVN0cmluZ0luZm8oJmxp bmVfYnVmLCBiZWdsb2MsIDEpOwohIAkJCQkJYnVmZmVyX2luZGV4Kys7CiEgCQkJCQliZWds b2MrKzsKISAJCQkJCQohIAkJCQkJbGluZV9kb25lID0gdHJ1ZTsKISAJCQkJCWVzY19pbl9w cmV2YnVmID0gZmFsc2U7CiEgCQkJCQljcl9pbl9wcmV2YnVmID0gZmFsc2U7CiEgCQkJCQkK ISAJCQkJCXJldHVybiBsaW5lX2RvbmU7CiEgCQkJCX0KISAJCQl9CiEgCQkJCiEgCQkJY3Jf aW5fcHJldmJ1ZiA9IGZhbHNlOwohIAkJfQogIAkJCQl9CiAgCiAgCQkJCS8qCiEgCSAqICh3 ZSBuZWVkIGEgbG9vcCBzbyB0aGF0IGlmIGVvbF9jaCBpcyBmb3VuZCwgYnV0IHdlIGFyZSBp biBxdW90ZXMsCiEgCSAqIHdlIGNhbiBzZWFyY2ggZm9yIHRoZSBuZXh0IGVvbF9jaCkKICAJ CQkJICovCiEgCXdoaWxlICh0cnVlKQohIAl7CiEgCQkvKiByZWFjaGVkIGVuZCBvZiBidWZm ZXIgKi8KISAJCWlmICgoZW5kbG9jID0gc2NhbkNTVkxpbmUoYmVnbG9jLCBlb2xfY2hbMF0s IGVzY2FwZWMsIHF1b3RlYywgYnl0ZXNyZWFkIC0gYnVmZmVyX2luZGV4KSkgPT0gTlVMTCkK ICAJCQkJewohIAkJCWxpbmVzaXplID0gQ09QWV9CVUZfU0laRSAtIChiZWdsb2MgLSBpbnB1 dF9idWYpOwohIAkJCWFwcGVuZEJpbmFyeVN0cmluZ0luZm8oJmxpbmVfYnVmLCBiZWdsb2Ms IGxpbmVzaXplKTsKISAJCQkKISAJCQlpZiAobGFzdF93YXNfZXNjKQohIAkJCQllc2NfaW5f cHJldmJ1ZiA9IHRydWU7CiEgCQkJCiEgCQkJaWYgKGxpbmVfYnVmLmxlbiA+IDEpCiEgCQkJ ewohIAkJCQljaGFyCSAgICpsYXN0X2NoID0gbGluZV9idWYuZGF0YSArIGxpbmVfYnVmLmxl biAtIDE7IC8qIGJlZm9yZSB0ZXJtaW5hdGluZyBcMCAqLwohIAkJCQkKISAJCQkJaWYgKCps YXN0X2NoID09ICdccicpCiEgCQkJCXsKISAJCQkJCWlmIChlb2xfdHlwZSA9PSBFT0xfQ1JO TCkKISAJCQkJCQljcl9pbl9wcmV2YnVmID0gdHJ1ZTsKICAJCQkJfQohIAkJCX0KISAJCQkK ISAJCQlsaW5lX2RvbmUgPSBmYWxzZTsKISAJCQlidWZfZG9uZSA9IHRydWU7CiAgCQkJCWJy ZWFrOwogIAkJCX0KKyAJCWVsc2UKKyAJCQkvKiBmb3VuZCAxc3QgZW9sIGNoYXIgaW4gaW5w dXRfYnVmLiAqLworIAkJeworIAkJCWJvb2wJCWVvbF9mb3VuZCA9IHRydWU7CiAgCQkJCiEg CQkJLyoKISAJCQkgKiBMb2FkIHRoYXQgcGllY2Ugb2YgZGF0YSAocG90ZW50aWFsbHkgYSBk YXRhIGxpbmUpIGludG8gdGhlIGxpbmUgYnVmZmVyLAohIAkJCSAqIGFuZCB1cGRhdGUgdGhl IHBvaW50ZXJzIGZvciB0aGUgbmV4dCBzY2FuLgohIAkJCSAqLwohIAkJCWxpbmVzaXplID0g ZW5kbG9jIC0gYmVnbG9jICsgMTsKISAJCQlhcHBlbmRCaW5hcnlTdHJpbmdJbmZvKCZsaW5l X2J1ZiwgYmVnbG9jLCBsaW5lc2l6ZSk7CiEgCQkJYnVmZmVyX2luZGV4ICs9IGxpbmVzaXpl OwohIAkJCWJlZ2xvYyA9IGVuZGxvYyArIDE7CiEgCQkJCiEgCQkJLyogZW5kIG9mIGxpbmUg b25seSBpZiBub3QgaW4gcXVvdGVzICovCiEgCQkJaWYoaW5fcXVvdGUpCiEgCQkJewohIAkJ CQlsaW5lX2RvbmUgPSBmYWxzZTsKISAJCQkJLyogYnVmIGRvbmUsIGJ1dCBzdGlsbCBpbiBx dW90ZSAqLwohIAkJCQlpZiAoYnVmZmVyX2luZGV4ID49IGJ5dGVzcmVhZCkKISAJCQkJewoh IAkJCQkJYnVmX2RvbmUgPSB0cnVlOwohIAkJCQkJYnJlYWs7CiEgCQkJCX0KISAJCQl9CiAg CQkJZWxzZQohIAkJCXsKISAJCQkJLyogaWYgZG9zIGVvbCwgY2hlY2sgZm9yICdcbicgYWZ0 ZXIgdGhlICdccicgKi8KISAJCQkJaWYgKGVvbF90eXBlID09IEVPTF9DUk5MKQohIAkJCQl7 CiEgCQkJCQlpZiAoKihlbmRsb2MgKyAxKSA9PSAnXG4nKQohIAkJCQkJewohIAkJCQkJCS8q IHRoaXMgaXMgYSBsaW5lIGVuZCAqLwohIAkJCQkJCWFwcGVuZEJpbmFyeVN0cmluZ0luZm8o JmxpbmVfYnVmLCBiZWdsb2MsIDEpOwkJLyogbG9hZCB0aGF0ICdcbicgKi8KISAJCQkJCQli dWZmZXJfaW5kZXgrKzsKISAJCQkJCQliZWdsb2MrKzsKISAJCQkJCX0KISAJCQllbHNlCiEg CQkJCQkJLyoganVzdCBhIENSLCBub3QgYSBsaW5lIGVuZCAqLwohIAkJCQkJCWVvbF9mb3Vu ZCA9IGZhbHNlOwogIAkJfQogIAohIAkJCQkvKgohIAkJCQkgKiBpbiBzb21lIGNhc2VzLCB0 aGlzIGVuZCBvZiBsaW5lIGNoYXIgaGFwcGVucyB0byBiZSB0aGUKISAJCQkJICogbGFzdCBj aGFyYWN0ZXIgaW4gdGhlIGJ1ZmZlci4gd2UgbmVlZCB0byBjYXRjaCB0aGF0LgohIAkJCQkg Ki8KISAJCQkJaWYgKGJ1ZmZlcl9pbmRleCA+PSBieXRlc3JlYWQpCiEgCQkJCQlidWZfZG9u ZSA9IHRydWU7CiAgCiAgCQkvKgohIAkJCQkgKiBpZiBlb2wgd2FzIGZvdW5kIGxpbmUgaXMg ZG9uZQogIAkJICovCiEgCQkJCWlmIChlb2xfZm91bmQpCiAgCQkJCXsKISAJCQkJCWxpbmVf ZG9uZSA9IHRydWU7CiAgCQkJCQlicmVhazsKICAJCQkJfQogIAkJCX0KISAJCX0JCQkJCQkv KiBlbmQgb2YgZm91bmQgZW9sX2NoICovCiAgCQl9CiAgCiEgCS8qCiEgCSAqIERvbmUgcmVh ZGluZyB0aGUgbGluZS4gQ29udmVydCBpdCB0byBzZXJ2ZXIgZW5jb2RpbmcuCiEgCSAqLwoh IAlpZiAodHJhbnNjb2RlKQogIAl7CiAgCQljdnQgPSAoY2hhciAqKSBwZ19jbGllbnRfdG9f c2VydmVyKCh1bnNpZ25lZCBjaGFyICopIGxpbmVfYnVmLmRhdGEsCiAgCQkJCQkJCQkJCSAg IGxpbmVfYnVmLmxlbik7CioqKioqKioqKioqKioqKgoqKiogMjI2NSwyMjg1ICoqKioKICAJ CQkvKiB0cmFuc2ZlciBjb252ZXJ0ZWQgZGF0YSBiYWNrIHRvIGxpbmVfYnVmICovCiAgCQkJ bGluZV9idWYubGVuID0gMDsKICAJCQlsaW5lX2J1Zi5kYXRhWzBdID0gJ1wwJzsKICAJCQlh cHBlbmRCaW5hcnlTdHJpbmdJbmZvKCZsaW5lX2J1ZiwgY3Z0LCBzdHJsZW4oY3Z0KSk7CiAg CQl9CiAgCX0KICAKISAJLyogTm93IGl0J3Mgc2FmZSB0byB1c2UgdGhlIGJ1ZmZlciBpbiBl cnJvciBtZXNzYWdlcyAqLwogIAlsaW5lX2J1Zl9jb252ZXJ0ZWQgPSB0cnVlOwogIAohIAly ZXR1cm4gcmVzdWx0OwogIH0KICAKICAvKgogICAqCVJldHVybiBkZWNpbWFsIHZhbHVlIGZv ciBhIGhleGFkZWNpbWFsIGRpZ2l0CiAgICovCiAgc3RhdGljCiEgaW50IEdldERlY2ltYWxG cm9tSGV4KGNoYXIgaGV4KQogIHsKICAJaWYgKGlzZGlnaXQoaGV4KSkKICAJCXJldHVybiBo ZXggLSAnMCc7Ci0tLSAyNDI1LDI0NzggLS0tLQogIAkJCS8qIHRyYW5zZmVyIGNvbnZlcnRl ZCBkYXRhIGJhY2sgdG8gbGluZV9idWYgKi8KICAJCQlsaW5lX2J1Zi5sZW4gPSAwOwogIAkJ CWxpbmVfYnVmLmRhdGFbMF0gPSAnXDAnOworIAkJCWxpbmVfYnVmLmN1cnNvciA9IDA7CiAg CQkJYXBwZW5kQmluYXJ5U3RyaW5nSW5mbygmbGluZV9idWYsIGN2dCwgc3RybGVuKGN2dCkp OwogIAkJfQogIAl9CiAgCiEgCS8qIGluZGljYXRlIHRoYXQgY29udmVyc2lvbiBoYWQgb2Nj dXJlZCAqLwogIAlsaW5lX2J1Zl9jb252ZXJ0ZWQgPSB0cnVlOwogIAohIAkvKgohIAkgKiBj aGVjayBpZiB0aGlzIGxpbmUgaXMgYW4gZW5kIG1hcmtlciAtLSAiXC4iCiEgCSAqLwohIAll bmRfbWFya2VyID0gZmFsc2U7CiEgCiEgCXN3aXRjaCAoZW9sX3R5cGUpCiEgCXsKISAJCWNh c2UgRU9MX05MOgohIAkJCWlmICghc3RyY21wKGxpbmVfYnVmLmRhdGEsICJcXC5cbiIpKQoh IAkJCQllbmRfbWFya2VyID0gdHJ1ZTsKISAJCQlicmVhazsKISAJCWNhc2UgRU9MX0NSOgoh IAkJCWlmICghc3RyY21wKGxpbmVfYnVmLmRhdGEsICJcXC5cciIpKQohIAkJCQllbmRfbWFy a2VyID0gdHJ1ZTsKISAJCQlicmVhazsKISAJCWNhc2UgRU9MX0NSTkw6CiEgCQkJaWYgKCFz dHJjbXAobGluZV9idWYuZGF0YSwgIlxcLlxyXG4iKSkKISAJCQkJZW5kX21hcmtlciA9IHRy dWU7CiEgCQkJYnJlYWs7CiEgCQljYXNlIEVPTF9VTktOT1dOOgohIAkJCWJyZWFrOwohIAl9 CiEgCiEgCWlmIChlbmRfbWFya2VyKQohIAl7CiEgCQlmZV9lb2YgPSB0cnVlOwohIAkJLyog d2UgZG9uJ3Qgd2FudCB0byBwcm9jZXNzIGEgXC4gYXMgZGF0YSBsaW5lLCB3YW50IHRvIHF1 aXQuICovCiEgCQlsaW5lX2RvbmUgPSBmYWxzZTsKISAJCWJ1Zl9kb25lID0gdHJ1ZTsKISAJ fQohIAohIAlyZXR1cm4gbGluZV9kb25lOwogIH0KICAKICAvKgogICAqCVJldHVybiBkZWNp bWFsIHZhbHVlIGZvciBhIGhleGFkZWNpbWFsIGRpZ2l0CiAgICovCiAgc3RhdGljCiEgaW50 CiEgR2V0RGVjaW1hbEZyb21IZXgoY2hhciBoZXgpCiAgewogIAlpZiAoaXNkaWdpdChoZXgp KQogIAkJcmV0dXJuIGhleCAtICcwJzsKKioqKioqKioqKioqKioqCioqKiAyMjg3LDIzNDQg KioqKgogIAkJcmV0dXJuIHRvbG93ZXIoaGV4KSAtICdhJyArIDEwOwogIH0KICAKISAvKi0t LS0tLS0tLS0KISAgKiBSZWFkIHRoZSB2YWx1ZSBvZiBhIHNpbmdsZSBhdHRyaWJ1dGUsIHBl cmZvcm1pbmcgZGUtZXNjYXBpbmcgYXMgbmVlZGVkLgohICAqCiEgICogZGVsaW0gaXMgdGhl IGNvbHVtbiBkZWxpbWl0ZXIgc3RyaW5nIChtdXN0IGJlIGp1c3Qgb25lIGJ5dGUgZm9yIG5v dykuCiEgICogbnVsbF9wcmludCBpcyB0aGUgbnVsbCBtYXJrZXIgc3RyaW5nLiAgTm90ZSB0 aGF0IHRoaXMgaXMgY29tcGFyZWQgdG8KISAgKiB0aGUgcHJlLWRlLWVzY2FwZWQgaW5wdXQg c3RyaW5nLgohICAqCiEgICogKnJlc3VsdCBpcyBzZXQgdG8gaW5kaWNhdGUgd2hhdCB0ZXJt aW5hdGVkIHRoZSByZWFkOgohICAqCQlOT1JNQUxfQVRUUjoJY29sdW1uIGRlbGltaXRlcgoh ICAqCQlFTkRfT0ZfTElORToJZW5kIG9mIGxpbmUKISAgKiBJbiBlaXRoZXIgY2FzZSwgdGhl IHN0cmluZyByZWFkIHVwIHRvIHRoZSB0ZXJtaW5hdG9yIGlzIHJldHVybmVkLgohICAqCiEg ICogKmlzbnVsbCBpcyBzZXQgdHJ1ZSBvciBmYWxzZSBkZXBlbmRpbmcgb24gd2hldGhlciB0 aGUgaW5wdXQgbWF0Y2hlZAohICAqIHRoZSBudWxsIG1hcmtlci4gIE5vdGUgdGhhdCB0aGUg Y2FsbGVyIGNhbm5vdCBjaGVjayB0aGlzIHNpbmNlIHRoZQohICAqIHJldHVybmVkIHN0cmlu ZyB3aWxsIGJlIHRoZSBwb3N0LWRlLWVzY2FwaW5nIGVxdWl2YWxlbnQsIHdoaWNoIG1heQoh ICAqIGxvb2sgdGhlIHNhbWUgYXMgc29tZSB2YWxpZCBkYXRhIHN0cmluZy4KISAgKi0tLS0t LS0tLS0KICAgKi8KISBzdGF0aWMgY2hhciAqCiEgQ29weVJlYWRBdHRyaWJ1dGUoY29uc3Qg Y2hhciAqZGVsaW0sIGNvbnN0IGNoYXIgKm51bGxfcHJpbnQsCiEgCQkJCSAgQ29weVJlYWRS ZXN1bHQgKnJlc3VsdCwgYm9vbCAqaXNudWxsKQogIHsKICAJY2hhcgkJYzsKISAJY2hhcgkJ ZGVsaW1jID0gZGVsaW1bMF07CiEgCWludAkJCXN0YXJ0X2N1cnNvciA9IGxpbmVfYnVmLmN1 cnNvcjsKISAJaW50CQkJZW5kX2N1cnNvcjsKISAJaW50CQkJaW5wdXRfbGVuOwogIAohIAkv KiByZXNldCBhdHRyaWJ1dGVfYnVmIHRvIGVtcHR5ICovCiEgCWF0dHJpYnV0ZV9idWYubGVu ID0gMDsKISAJYXR0cmlidXRlX2J1Zi5kYXRhWzBdID0gJ1wwJzsKICAKISAJLyogc2V0IGRl ZmF1bHQgc3RhdHVzICovCiEgCSpyZXN1bHQgPSBFTkRfT0ZfTElORTsKICAKISAJZm9yICg7 OykKICAJewohIAkJZW5kX2N1cnNvciA9IGxpbmVfYnVmLmN1cnNvcjsKISAJCWlmIChsaW5l X2J1Zi5jdXJzb3IgPj0gbGluZV9idWYubGVuKQohIAkJCWJyZWFrOwohIAkJYyA9IGxpbmVf YnVmLmRhdGFbbGluZV9idWYuY3Vyc29yKytdOwohIAkJaWYgKGMgPT0gZGVsaW1jKQogIAkJ ewohIAkJCSpyZXN1bHQgPSBOT1JNQUxfQVRUUjsKISAJCQlicmVhazsKICAJCX0KISAJCWlm IChjID09ICdcXCcpCiAgCQl7CiEgCQkJaWYgKGxpbmVfYnVmLmN1cnNvciA+PSBsaW5lX2J1 Zi5sZW4pCiEgCQkJCWJyZWFrOwohIAkJCWMgPSBsaW5lX2J1Zi5kYXRhW2xpbmVfYnVmLmN1 cnNvcisrXTsKISAJCQlzd2l0Y2ggKGMpCiAgCQkJewogIAkJCQljYXNlICcwJzoKICAJCQkJ Y2FzZSAnMSc6Ci0tLSAyNDgwLDI3NjkgLS0tLQogIAkJcmV0dXJuIHRvbG93ZXIoaGV4KSAt ICdhJyArIDEwOwogIH0KICAKISAvKgohICAqIERldGVjdGVkIHRoZSBlb2wgdHlwZSBieSBs b29raW5nIGF0IHRoZSBmaXJzdCBkYXRhIHJvdy4KISAgKiBQb3NzaWJsZSBlb2wgdHlwZXMg YXJlIE5MLCBDUiwgb3IgQ1JOTC4gSWYgZW9sIHR5cGUgd2FzCiEgICogZGV0ZWN0ZWQsIGl0 IGlzIHNldCBhbmQgYSBib29sZWFuIHRydWUgaXMgcmV0dXJuZWQgdG8KISAgKiBpbmRpY2F0 ZWQgZGV0ZWN0aW9uIHdhcyBzdWNjZXNzZnVsLiBJZiB0aGUgZmlyc3QgZGF0YSByb3cKISAg KiBpcyBsb25nZXIgdGhhbiB0aGUgaW5wdXQgYnVmZmVyLCB3ZSByZXR1cm4gZmFsc2UgYW5k IHdpbGwKISAgKiB0cnkgYWdhaW4gaW4gdGhlIG5leHQgYnVmZmVyLgogICAqLwohIHN0YXRp YyBib29sCiEgRGV0ZWN0TGluZUVuZChzaXplX3QgYnl0ZXNyZWFkLCBjaGFyICpxdW90ZSwg Y2hhciAqZXNjYXBlKQogIHsKKyAJaW50ICAgICAgICAgaW5kZXggPSAwOwogIAljaGFyCQlj OwohIAljaGFyCQlxdW90ZWMgPSAnXDAnLAohIAkJZXNjYXBlYyA9ICdcMCc7CiEgCWJvb2wg ICAgICAgIGNzdiA9IGZhbHNlOwogIAohIAlpZiAocXVvdGUpIC8qIENTViBmb3JtYXQgKi8K ISAJewohIAkJY3N2ID0gdHJ1ZTsKISAJCXF1b3RlYyA9IHF1b3RlWzBdOwohIAkJZXNjYXBl YyA9IGVzY2FwZVswXTsKISAJCS8qIGlnbm9yZSBzcGVjaWFsIGVzY2FwZSBwcm9jZXNzaW5n IGlmIGl0J3MgdGhlIHNhbWUgYXMgcXVvdGVjICovCiEgCQlpZiAocXVvdGVjID09IGVzY2Fw ZWMpCiEgCQkJZXNjYXBlYyA9ICdcMCc7CiEgCX0KICAKISAJd2hpbGUgKGluZGV4IDwgQ09Q WV9CVUZfU0laRSkKISAJewohIAkJYyA9IGlucHV0X2J1ZltpbmRleF07CiAgCiEgCQlpZihj c3YpCiAgCXsKISAJCQlpZiAoaW5fcXVvdGUgJiYgYyA9PSBlc2NhcGVjKQohIAkJCQlsYXN0 X3dhc19lc2MgPSAhbGFzdF93YXNfZXNjOwohIAkJCWlmIChjID09IHF1b3RlYyAmJiAhbGFz dF93YXNfZXNjKQohIAkJCQlpbl9xdW90ZSA9ICFpbl9xdW90ZTsKISAJCQlpZiAoYyAhPSBl c2NhcGVjKQohIAkJCQlsYXN0X3dhc19lc2MgPSBmYWxzZTsKISAJCX0KISAJCQohIAkJaWYg KGMgPT0gJ1xuJykKICAJCXsKISAJCQlpZighY3N2IHx8IChjc3YgJiYgIWluX3F1b3RlKSkK ISAJCQl7CiEgCQkJCWVvbF90eXBlID0gRU9MX05MOwohIAkJCQllb2xfY2hbMF0gPSAnXG4n OwohIAkJCQllb2xfY2hbMV0gPSAnXDAnOwohIAkJCQkKISAJCQkJaW5fcXVvdGUgPSBmYWxz ZTsKISAJCQkJbGFzdF93YXNfZXNjID0gZmFsc2U7CiEgCQkJCXJldHVybiB0cnVlOwohIAkJ CX0KICAJCX0KISAJCWlmIChjID09ICdccicpCiAgCQl7CiEgCQkJaWYoIWNzdiB8fCAoY3N2 ICYmICFpbl9xdW90ZSkpCiEgICAgICAgICAgICAgewohIAkJCQlpZiAoaW5wdXRfYnVmW2lu ZGV4ICsgMV0gPT0gJ1xuJykgLyogYWx3YXlzIHNhZmUgKi8KISAJCQkJCXsKISAJCQkJCWVv bF90eXBlID0gRU9MX0NSTkw7CiEgCQkJCQllb2xfY2hbMF0gPSAnXHInOwohIAkJCQkJZW9s X2NoWzFdID0gJ1xuJzsKISAJCQkJCX0KISAJCQkJCWVsc2UKISAJCQkJCXsKISAJCQkJCQll b2xfdHlwZSA9IEVPTF9DUjsKISAJCQkJCQllb2xfY2hbMF0gPSAnXHInOwohIAkJCQkJCWVv bF9jaFsxXSA9ICdcMCc7CiEgCQkJCQl9CiEgCQkJCQkKISAJCQkJCWluX3F1b3RlID0gZmFs c2U7CiEgCQkJCQlsYXN0X3dhc19lc2MgPSBmYWxzZTsKISAJCQkJCXJldHVybiB0cnVlOwoh IAkJCX0KISAJCX0KISAKISAJCWluZGV4Kys7CiEgCX0KISAKISByZXR1cm4gZmFsc2U7CiEg fQohIAohIC8qCiEgICogUmVhZCBhbGwgVEVYVCBhdHRyaWJ1dGVzLiBBdHRyaWJ1dGVzIGFy ZSBwYXJzZWQgZnJvbSBsaW5lX2J1ZiBhbmQKISAgKiBpbnNlcnRlZCAoYWxsIGF0IG9uY2Up IHRvIGF0dHJfYnVmLCB3aGlsZSBzYXZpbmcgcG9pbnRlcnMgdG8KISAgKiBlYWNoIGF0dHJp YnV0ZSdzIHN0YXJ0aW5nIHBvc2l0aW9uLgohICAqCiEgICogV2hlbiB0aGlzIHJvdXRpbmUg ZmluaXNoZXMgZXhlY3V0aW9uIGJvdGggdGhlIG51bGxzIGFycmF5IGFuZAohICAqIHRoZSBh dHRyX29mZnNldHMgYXJyYXkgYXJlIHVwZGF0ZWQuIFRoZSBhdHRyX29mZnNldHMgd2lsbCBp bmNsdWRlCiEgICogdGhlIG9mZnNldCBmcm9tIHRoZSBiZWdpbm5pbmcgb2YgdGhlIGF0dHJp YnV0ZSBhcnJheSBvZiB3aGljaAohICAqIGVhY2ggYXR0cmlidXRlIGJlZ2lucy4gSWYgYSBz cGVjaWZpYyBhdHRyaWJ1dGUgaXMgbm90IHVzZWQgZm9yIHRoaXMKISAgKiBDT1BZIGNvbW1h bmQgKG9tbWl0dGVkIGZyb20gdGhlIGNvbHVtbiBsaXN0KSwgYSB2YWx1ZSBvZiAwIHdpbGwg YmUgYXNzaWduZWQuCiEgICogRm9yIGV4YW1wbGU6IGZvciB0YWJsZSBmb28oYSxiLGMsZCxl KSBhbmQgQ09QWSBmb28oYSxiLGUpCiEgICogYXR0cl9vZmZzZXRzIG1heSBsb29rIHNvbWV0 aGluZyBsaWtlIHRoaXMgYWZ0ZXIgdGhpcyByb3V0aW5lCiEgICogcmV0dXJuczogWzAsMjAs MCwwLDU1XS4gVGhhdCBtZWFucyB0aGF0IGNvbHVtbiAiYSIgdmFsdWUgc3RhcnRzCiEgICog YXQgYnl0ZSBvZmZzZXQgMCwgImIiIGluIDIwIGFuZCAiZSIgaW4gNTUsIGluIGF0dHJfYnVm LgohICAqCiEgICogSW4gdGhlIGF0dHJpYnV0ZSBidWZmZXIgKGF0dHJfYnVmKSBlYWNoIGF0 dHJpYnV0ZQohICAqIGlzIHRlcm1pbmF0ZWQgd2l0aCBhICdcMCcsIGFuZCB0aGVyZWZvcmUg YnkgdXNpbmcgdGhlIGF0dHJfb2Zmc2V0cwohICAqIGFycmF5IHdlIGNvdWxkIHBvaW50IHRv IGEgYmVnaW5uaW5nIG9mIGFuIGF0dHJpYnV0ZSBhbmQgaGF2ZSBpdAohICAqIGJlaGF2ZSBh cyBhIEMgc3RyaW5nLCBtdWNoIGxpa2UgcHJldmlvdXNseSBkb25lIGluIENPUFkuCiEgICoK ISAgKiBBbm90aGVyIGFzcGVjdCB0byBpbXByb3ZpbmcgcGVyZm9ybWFuY2UgaXMgcmVkdWNp bmcgdGhlIGZyZXF1ZW5jeQohICAqIG9mIGRhdGEgbG9hZCBpbnRvIGJ1ZmZlcnMuIFRoZSBv cmlnaW5hbCBDT1BZIHJlYWQgYXR0cmlidXRlIGNvZGUKISAgKiBsb2FkZWQgYSBjaGFyYWN0 ZXIgYXQgYSB0aW1lLiBJbiBoZXJlIHdlIHRyeSB0byBsb2FkIGEgY2h1bmsgb2YgZGF0YQoh ICAqIGF0IGEgdGltZS4gVXN1YWxseSBhIGNodW5rIHdpbGwgaW5jbHVkZSBhIGZ1bGwgZGF0 YSByb3cKISAgKiAodW5sZXNzIHdlIGhhdmUgYW4gZXNjYXBlZCBkZWxpbSkuIFRoYXQgZWZm ZWN0aXZlbHkgcmVkdWNlcyB0aGUgbnVtYmVyIG9mCiEgICogbG9hZHMgYnkgYSBmYWN0b3Ig b2YgbnVtYmVyIG9mIGJ5dGVzIHBlciByb3cuIFRoaXMgaW1wcm92ZXMgcGVyZm9ybWFuY2UK ISAgKiBncmVhdGx5LCB1bmZvcnR1bmF0ZWx5IGl0IGFkZCBtb3JlIGNvbXBsZXhpdHkgdG8g dGhlIGNvZGUuCiEgICoKISAgKiBHbG9iYWwgcGFydGljaXBhbnRzIGluIHBhcnNpbmcgbG9n aWM6CiEgICoKISAgKiBsaW5lX2J1Zi5jdXJzb3IgLS0gYW4gb2Zmc2V0IGZyb20gYmVnaW5u aW5nIG9mIHRoZSBsaW5lIGJ1ZmZlcgohICAqIHRoYXQgaW5kaWNhdGVzIHdoZXJlIHdlIGFy ZSBhYm91dCB0byBiZWdpbiB0aGUgbmV4dCBzY2FuLiBOb3RlIHRoYXQKISAgKiBpZiB3ZSBo YXZlIFdJVEggT0lEUyB0aGlzIGN1cnNvciBpcyBhbHJlYWR5IHNoaWZ0ZWQgcGFzdCB0aGUg Zmlyc3QKISAgKiBPSUQgYXR0cmlidXRlLgohICAqCiEgICogYXR0cl9idWYuY3Vyc29yIC0t IGFuIG9mZnNldCBmcm9tIHRoZSBiZWdpbm5pbmcgb2YgdGhlCiEgICogYXR0cmlidXRlIGJ1 ZmZlciB0aGF0IGluZGljYXRlcyB3aGVyZSB0aGUgY3VycmVudCBhdHRyaWJ1dGUgYmVnaW5z LgohICAqLwohIHN0YXRpYyB2b2lkIAohIENvcHlSZWFkQXR0cmlidXRlc1RleHQoY29uc3Qg Y2hhciAqZGVsaW0sIGNvbnN0IGNoYXIgKmVzY2FwZSwgY29uc3QgY2hhciAqbnVsbF9wcmlu dCwKISAJCQkJCSAgIGludCBudWxsX3ByaW50X2xlbiwgY2hhciAqbnVsbHMsIExpc3QgKmF0 dG51bWxpc3QsIAohIAkJCQkJICAgaW50ICphdHRyX29mZnNldHMsIGludCBudW1fcGh5c19h dHRycywgRm9ybV9wZ19hdHRyaWJ1dGUgKmF0dHIpCiEgewohIAljaGFyCQlkZWxpbWMgPSBk ZWxpbVswXTsJCS8qIGRlbGltaXRlciBjaGFyYWN0ZXIgKi8KISAJY2hhciAgICAgICAgZXNj YXBlYyA9IGVzY2FwZVswXTsgICAgLyogZXNjYXBlIGNoYXJhY3RlciAgICAqLwohIAljaGFy CSAgICpzY2FuX3N0YXJ0OwkJLyogcG9pbnRlciB0byBsaW5lIGJ1ZmZlciBmb3Igc2NhbiBz dGFydC4gKi8KISAJY2hhcgkgICAqc2Nhbl9lbmQ7CQkvKiBwb2ludGVyIHRvIGxpbmUgYnVm ZmVyIHdoZXJlIGNoYXIgd2FzIGZvdW5kICovCiEgCWludAkJCWF0dHJfcHJlX2xlbjsJLyog YXR0ciByYXcgbGVuLCBiZWZvcmUgcHJvY2Vzc2luZyBlc2NhcGVzICovCiEgCWludAkJCWF0 dHJfcG9zdF9sZW47CS8qIGN1cnJlbnQgYXR0ciBsZW4gYWZ0ZXIgZXNjYXBpbmcgKi8KISAJ aW50CQkJbTsJCQkJLyogYXR0cmlidXRlIGluZGV4IGJlaW5nIHBhcnNlZCAqLwohIAlpbnQJ CQlieXRlc19yZW1haW5pbmc7LyogbnVtIGJ5dGVzIHJlbWFpbmluZyB0byBiZSBzY2FubmVk IGluIGxpbmUgYnVmICovCiEgCWludAkJCWNodW5rX3N0YXJ0OwkvKiBvZmZzZXQgdG8gYmVn aW5uaW5nIG9mIGxpbmUgY2h1bmsgdG8gbG9hZCAqLwohIAlpbnQJCQljaHVua19sZW47CQkv KiBsZW5ndGggb2YgY2h1bmsgb2YgZGF0YSB0byBsb2FkIHRvIGF0dHIgYnVmICovCiEgCWlu dAkJCW9jdF92YWw7CQkvKiBieXRlIHZhbHVlIGZvciBvY3RhbCBlc2NhcGVzICovCiEgCWlu dCAgICAgICAgIGhleF92YWw7ICAgICAgICAvKiBieXRlIHZhbHVlIGZvciBoZXhhZGVjaW1h bCBlc2NhcGVzICovCiEgCWNoYXIgICAgICAgIGhleGNoYXI7ICAgICAgICAvKiBjaGFyIHRo YXQgYXBwZWFycyBhZnRlciBceCBmb3IgaGV4IHZhbHVlcyAqLwohIAlpbnQJCQlhdHRudW07 CQkJLyogYXR0cmlidXRlIG51bWJlciBiZWluZyBwYXJzZWQgKi8KISAJTGlzdENlbGwgICAq Y3VyOwkJCS8qIGN1cnNvciB0byBhdHRyaWJ1dGUgbGlzdCB1c2VkIGZvciB0aGlzIENPUFkg Ki8KISAJaW50CQkJYXR0cmlidXRlOwohIAkJCiEgCQkvKgohIAkJICogaW5pdCB2YXJpYWJs ZXMgZm9yIGF0dHJpYnV0ZSBzY2FuCiEgCQkgKi8KISAJCWF0dHJfYnVmLmxlbiA9IDA7CiEg CQlhdHRyX2J1Zi5kYXRhWzBdID0gJ1wwJzsKISAJCWF0dHJfYnVmLmN1cnNvciA9IDA7CiEg CQkvKiBjdXJzb3IgaXMgbm93ID4gMCBpZiB3ZSBjb3B5IFdJVEggT0lEUyAqLwohIAkJc2Nh bl9zdGFydCA9IGxpbmVfYnVmLmRhdGEgKyBsaW5lX2J1Zi5jdXJzb3I7CiEgCQljdXIgPSBs aXN0X2hlYWQoYXR0bnVtbGlzdCk7CiEgCQlhdHRudW0gPSBsZmlyc3RfaW50KGN1cik7CiEg CQltID0gYXR0bnVtIC0gMTsKISAJCWNodW5rX3N0YXJ0ID0gbGluZV9idWYuY3Vyc29yOwoh IAkJY2h1bmtfbGVuID0gMDsKISAJCWF0dHJfcHJlX2xlbiA9IDA7CiEgCQlhdHRyX3Bvc3Rf bGVuID0gMDsKISAJCQohIAkJLyoKISAJCSAqIFNjYW4gdGhyb3VnaCB0aGUgbGluZSBidWZm ZXIgdG8gcmVhZCBhbGwgYXR0cmlidXRlcyBkYXRhCiEgCQkgKi8KISAJCXdoaWxlIChsaW5l X2J1Zi5jdXJzb3IgPCBsaW5lX2J1Zi5sZW4pCiEgCQl7CiEgCQkJYnl0ZXNfcmVtYWluaW5n ID0gbGluZV9idWYubGVuIC0gbGluZV9idWYuY3Vyc29yOwohIAkJCQohIAkJCWlmICgoc2Nh bl9lbmQgPSBzY2FuVGV4dEF0dHIoc2Nhbl9zdGFydCwgZGVsaW1jLCBlc2NhcGVjLCBieXRl c19yZW1haW5pbmcpKQohIAkJCQk9PSBOVUxMKQohIAkJCXsKISAJCQkJLyogR09UIFRPIEVO RCBPRiBMSU5FIEJVRkZFUiAqLwohIAkJCQkKISAJCQkJaWYgKGN1ciA9PSBOVUxMKQohIAkJ CQkJZXJlcG9ydChFUlJPUiwKISAJCQkJCQkJKGVycmNvZGUoRVJSQ09ERV9CQURfQ09QWV9G SUxFX0ZPUk1BVCksCiEgCQkJCQkJCSBlcnJtc2coImV4dHJhIGRhdGEgYWZ0ZXIgbGFzdCBl eHBlY3RlZCBjb2x1bW4iKSkpOwohIAkJCQkKISAJCQkJYXR0bnVtID0gbGZpcnN0X2ludChj dXIpOwohIAkJCQltID0gYXR0bnVtIC0gMTsKISAJCQkJCiEgCQkJCS8qIGRvbid0IGNvdW50 IGVvbCBjaGFyKHMpIGluIGF0dHIgYW5kIGNodW5rIGxlbiBjYWxjdWxhdGlvbiAqLwohIAkJ CQlpZiAoZW9sX3R5cGUgPT0gRU9MX0NSTkwpCiEgCQkJCXsKISAJCQkJCWF0dHJfcHJlX2xl biArPSBieXRlc19yZW1haW5pbmcgLSAyOwohIAkJCQkJY2h1bmtfbGVuID0gbGluZV9idWYu bGVuIC0gY2h1bmtfc3RhcnQgLSAyOwohIAkJCQl9CiEgCQkJCWVsc2UKISAJCQkJewohIAkJ CQkJYXR0cl9wcmVfbGVuICs9IGJ5dGVzX3JlbWFpbmluZyAtIDE7CiEgCQkJCQljaHVua19s ZW4gPSBsaW5lX2J1Zi5sZW4gLSBjaHVua19zdGFydCAtIDE7CiEgCQkJCX0KISAJCQkJCiEg CQkJCS8qIGNoZWNrIGlmIHRoaXMgaXMgYSBOVUxMIHZhbHVlIG9yIGRhdGEgdmFsdWUgKGFz c3VtZWQgTlVMTCkgKi8KISAJCQkJaWYgKGF0dHJfcHJlX2xlbiA9PSBudWxsX3ByaW50X2xl bgohIAkJCQkJJiYKISAJCQkJCXN0cm5jbXAobGluZV9idWYuZGF0YSArIGxpbmVfYnVmLmxl biAtIGF0dHJfcHJlX2xlbiAtIDEsIG51bGxfcHJpbnQsIGF0dHJfcHJlX2xlbikKISAJCQkJ CT09IDApCiEgCQkJCQludWxsc1ttXSA9ICduJzsKISAJCQkJZWxzZQohIAkJCQkJbnVsbHNb bV0gPSAnICc7CiEgCQkJCQohIAkJCQlhdHRyX29mZnNldHNbbV0gPSBhdHRyX2J1Zi5jdXJz b3I7CiEgCQkJCQohIAkJCQkKISAJCQkJLyogbG9hZCB0aGUgbGFzdCBjaHVuaywgdGhlIHdo b2xlIGJ1ZmZlciBpbiBtb3N0IGNhc2VzICovCiEgCQkJCWFwcGVuZEJpbmFyeVN0cmluZ0lu Zm8oJmF0dHJfYnVmLCBsaW5lX2J1Zi5kYXRhICsgY2h1bmtfc3RhcnQsIGNodW5rX2xlbik7 CiEgCQkJCQohIAkJCQlsaW5lX2J1Zi5jdXJzb3IgKz0gYXR0cl9wcmVfbGVuICsgMjsJCS8q IHNraXAgZW9sIGNoYXIgYW5kCiEgCQkJCQkqICdcMCcgdG8gZXhpdCBsb29wICovCiEgCQkJ CQohIAkJCQlpZiAobG5leHQoY3VyKSAhPSBOVUxMKQohIAkJCQl7CiEgCQkJCQkvKgohIAkJ CQkJICogRm9yIGFuIGVtcHR5IGRhdGEgbGluZSwgdGhlIHByZXZpb3VzIENPUFkgY29kZSB3 aWxsCiEgCQkJCQkgKiBmYWlsIGl0IGR1cmluZyB0aGUgY29udmVyc2lvbiBzdGFnZS4gV2Ug Y2FuIGZhaWwgaXQgaGVyZQohIAkJCQkJICogYWxyZWFkeSwgYnV0IHRoZW4gd2Ugd2lsbCBm YWlsIHRoZSByZWdyZXNzaW9uIHRlc3RzIGIvYwohIAkJCQkJICogb2YgYSBkaWZmZXJlbnQg ZXJyb3IgbWVzc2FnZS4gdGhhdCdzIHdoeSB3ZSByZXR1cm4gc28gd2UKISAJCQkJCSAqIGNh biBnZXQgdGhlIHNhbWUgZXJyb3IgbWVzc2FnZSB0aGF0IHJlZ3Jlc3MgZXhwZWN0cy4gYWho Li4uCiEgCQkJCQkgKiB0aGlzIGNvbmRpdGlvbmFsIGlzIHVubmVjZXNzYXJ5IGFuZCBzaG91 bGQgYmUgcmVtb3ZlZCBzb29uLgohIAkJCQkJICovCiEgCQkJCQlpZiAobGluZV9idWYubGVu ID4gMSkKISAJCQkJCQllcmVwb3J0KEVSUk9SLAohIAkJCQkJCQkJKGVycmNvZGUoRVJSQ09E RV9CQURfQ09QWV9GSUxFX0ZPUk1BVCksCiEgCQkJCQkJCQkgZXJybXNnKCJtaXNzaW5nIGRh dGEgZm9yIGNvbHVtbiBcIiVzXCIiLAohIAkJCQkJCQkJCQlOYW1lU3RyKGF0dHJbbSArIDFd LT5hdHRuYW1lKSkpKTsKISAJCQkJCWVsc2UKISAJCQkJCQlyZXR1cm47CiEgCQkJCX0KISAJ CQl9CiEgCQkJZWxzZQohIAkJCQkvKiBGT1VORCBBIERFTElNSVRFUiBPUiBFU0NBUEUgKi8K ISAJCQl7CiEgCQkJCWlmIChjdXIgPT0gTlVMTCkKISAJCQkJCWVyZXBvcnQoRVJST1IsCiEg CQkJCQkJCShlcnJjb2RlKEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JNQVQpLAohIAkJCQkJ CQkgZXJybXNnKCJleHRyYSBkYXRhIGFmdGVyIGxhc3QgZXhwZWN0ZWQgY29sdW1uIikpKTsK ISAJCQkJCiEgCQkJCWlmICgqc2Nhbl9lbmQgPT0gZGVsaW1jKQkJLyogZm91bmQgYSBkZWxp bWl0ZXIgKi8KISAJCQkJewohIAkJCQkJYXR0bnVtID0gbGZpcnN0X2ludChjdXIpOwohIAkJ CQkJbSA9IGF0dG51bSAtIDE7CiEgCQkJCQkKISAJCQkJCS8qICh3ZSBkb24ndCBpbmNsdWRl IHRoZSBkZWxpbWl0ZXIgY2ggaW4gbGVuZ3RoKSAqLwohIAkJCQkJYXR0cl9wcmVfbGVuICs9 IHNjYW5fZW5kIC0gc2Nhbl9zdGFydDsKISAJCQkJCS8qICh3ZSBkb24ndCBpbmNsdWRlIHRo ZSBkZWxpbWl0ZXIgY2ggaW4gbGVuZ3RoKSAqLwohIAkJCQkJYXR0cl9wb3N0X2xlbiArPSBz Y2FuX2VuZCAtIHNjYW5fc3RhcnQ7CiEgCQkJCQkKISAJCQkJCS8qIGNoZWNrIGlmIHRoaXMg aXMgYSBudWxsIHByaW50IG9yIGRhdGEgKGFzc3VtZWQgTlVMTCkgKi8KISAJCQkJCWlmIChh dHRyX3ByZV9sZW4gPT0gbnVsbF9wcmludF9sZW4KISAJCQkJCQkmJgohIAkJCQkJCXN0cm5j bXAoc2Nhbl9lbmQgLSBhdHRyX3ByZV9sZW4sIG51bGxfcHJpbnQsIGF0dHJfcHJlX2xlbikK ISAJCQkJCQk9PSAwKQohIAkJCQkJCW51bGxzW21dID0gJ24nOwohIAkJCQkJZWxzZQohIAkJ CQkJCW51bGxzW21dID0gJyAnOwohIAkJCQkJCiEgCQkJCQkvKiBzZXQgdGhlIHBvaW50ZXIg dG8gbmV4dCBhdHRyaWJ1dGUgcG9zaXRpb24gKi8KISAJCQkJCWF0dHJfb2Zmc2V0c1ttXSA9 IGF0dHJfYnVmLmN1cnNvcjsKISAJCQkJCQohIAkJCQkJLyoKISAJCQkJCSAqIHVwZGF0ZSBi dWZmZXIgY3Vyc29ycyB0byBvdXIgY3VycmVudCBsb2NhdGlvbiwgKzEgdG8gc2tpcAohIAkJ CQkJICogdGhlIGRlbGltYwohIAkJCQkJICovCiEgCQkJCQlsaW5lX2J1Zi5jdXJzb3IgPSBz Y2FuX2VuZCAtIGxpbmVfYnVmLmRhdGEgKyAxOwohIAkJCQkJYXR0cl9idWYuY3Vyc29yICs9 IGF0dHJfcG9zdF9sZW4gKyAxOwohIAkJCQkJCiEgCQkJCQkvKiBwcmVwYXJlIHNjYW4gZm9y IG5leHQgYXR0ciAqLwohIAkJCQkJc2Nhbl9zdGFydCA9IGxpbmVfYnVmLmRhdGEgKyBsaW5l X2J1Zi5jdXJzb3I7CiEgCQkJCQljdXIgPSBsbmV4dChjdXIpOwohIAkJCQkJYXR0cl9wcmVf bGVuID0gMDsKISAJCQkJCWF0dHJfcG9zdF9sZW4gPSAwOwohIAkJCQl9CiEgCQkJCWVsc2UK ISAJCQkJCS8qIGZvdW5kIGFuIGVzY2FwZSBjaGFyYWN0ZXIgKi8KISAJCQkJewohIAkJCQkJ Y2hhcgkJbmV4dGMgPSAqKHNjYW5fZW5kICsgMSk7CiEgCQkJCQljaGFyCQluZXdjOwohIAkJ CQkJaW50CQkJc2tpcCA9IDI7CiEgCQkJCQkKISAJCQkJCWNodW5rX2xlbiA9IChzY2FuX2Vu ZCAtIGxpbmVfYnVmLmRhdGEpIC0gY2h1bmtfc3RhcnQgKyAxOwohIAkJCQkJCiEgCQkJCQkv KiBsb2FkIGEgY2h1bmsgb2YgZGF0YSAqLwohIAkJCQkJYXBwZW5kQmluYXJ5U3RyaW5nSW5m bygmYXR0cl9idWYsIGxpbmVfYnVmLmRhdGEgKyBjaHVua19zdGFydCwgY2h1bmtfbGVuKTsK ISAJCQkJCQohIAkJCQkJc3dpdGNoIChuZXh0YykKICAJCQl7CiAgCQkJCWNhc2UgJzAnOgog IAkJCQljYXNlICcxJzoKKioqKioqKioqKioqKioqCioqKiAyMzQ4LDI0NDUgKioqKgogIAkJ CQljYXNlICc1JzoKICAJCQkJY2FzZSAnNic6CiAgCQkJCWNhc2UgJzcnOgohIAkJCQkJLyog aGFuZGxlIFwwMTMgKi8KISAJCQkJCXsKISAJCQkJCQlpbnQJCQl2YWw7CiAgCiEgCQkJCQkJ dmFsID0gT0NUVkFMVUUoYyk7CiEgCQkJCQkJaWYgKGxpbmVfYnVmLmN1cnNvciA8IGxpbmVf YnVmLmxlbikKISAJCQkJCQl7CiEgCQkJCQkJCWMgPSBsaW5lX2J1Zi5kYXRhW2xpbmVfYnVm LmN1cnNvcl07CiEgCQkJCQkJCWlmIChJU09DVEFMKGMpKQohIAkJCQkJCQl7CiEgCQkJCQkJ CQlsaW5lX2J1Zi5jdXJzb3IrKzsKISAJCQkJCQkJCXZhbCA9ICh2YWwgPDwgMykgKyBPQ1RW QUxVRShjKTsKISAJCQkJCQkJCWlmIChsaW5lX2J1Zi5jdXJzb3IgPCBsaW5lX2J1Zi5sZW4p CiAgCQkJCQkJCQl7CiEgCQkJCQkJCQkJYyA9IGxpbmVfYnVmLmRhdGFbbGluZV9idWYuY3Vy c29yXTsKISAJCQkJCQkJCQlpZiAoSVNPQ1RBTChjKSkKICAJCQkJCQkJCQl7CiEgCQkJCQkJ CQkJCWxpbmVfYnVmLmN1cnNvcisrOwohIAkJCQkJCQkJCQl2YWwgPSAodmFsIDw8IDMpICsg T0NUVkFMVUUoYyk7CiAgCQkJCQkJCQkJfQogIAkJCQkJCQkJfQohIAkJCQkJCQl9CiEgCQkJ CQkJfQohIAkJCQkJCWMgPSB2YWwgJiAwMzc3OwohIAkJCQkJfQogIAkJCQkJYnJlYWs7CiAg CQkJCWNhc2UgJ3gnOgogIAkJCQkJLyogSGFuZGxlIFx4M0YgKi8KISAJCQkJCWlmIChsaW5l X2J1Zi5jdXJzb3IgPCBsaW5lX2J1Zi5sZW4pCiEgCQkJCQl7CiEgCQkJCQkJY2hhciBoZXhj aGFyID0gbGluZV9idWYuZGF0YVtsaW5lX2J1Zi5jdXJzb3JdOwogIAogIAkJCQkJCWlmIChp c3hkaWdpdChoZXhjaGFyKSkKICAJCQkJCQl7CiEgCQkJCQkJCWludCB2YWwgPSBHZXREZWNp bWFsRnJvbUhleChoZXhjaGFyKTsKICAKISAJCQkJCQkJbGluZV9idWYuY3Vyc29yKys7CiEg CQkJCQkJCWlmIChsaW5lX2J1Zi5jdXJzb3IgPCBsaW5lX2J1Zi5sZW4pCiEgCQkJCQkJCXsK ISAJCQkJCQkJCWhleGNoYXIgPSBsaW5lX2J1Zi5kYXRhW2xpbmVfYnVmLmN1cnNvcl07CiAg CQkJCQkJCQlpZiAoaXN4ZGlnaXQoaGV4Y2hhcikpCiAgCQkJCQkJCQl7CiEgCQkJCQkJCQkJ bGluZV9idWYuY3Vyc29yKys7CiEgCQkJCQkJCQkJdmFsID0gKHZhbCA8PCA0KSArIEdldERl Y2ltYWxGcm9tSGV4KGhleGNoYXIpOwohIAkJCQkJCQkJfQohIAkJCQkJCQl9CiEgCQkJCQkJ CWMgPSB2YWwgJiAweGZmOwogIAkJCQkJCX0KICAJCQkJCX0KICAJCQkJCWJyZWFrOwogIAkJ CQljYXNlICdiJzoKISAJCQkJCWMgPSAnXGInOwogIAkJCQkJYnJlYWs7CiAgCQkJCWNhc2Ug J2YnOgohIAkJCQkJYyA9ICdcZic7CiAgCQkJCQlicmVhazsKICAJCQkJY2FzZSAnbic6CiEg CQkJCQljID0gJ1xuJzsKICAJCQkJCWJyZWFrOwogIAkJCQljYXNlICdyJzoKISAJCQkJCWMg PSAnXHInOwogIAkJCQkJYnJlYWs7CiAgCQkJCWNhc2UgJ3QnOgohIAkJCQkJYyA9ICdcdCc7 CiAgCQkJCQlicmVhazsKICAJCQkJY2FzZSAndic6CiEgCQkJCQljID0gJ1x2JzsKICAJCQkJ CWJyZWFrOwogIAogIAkJCQkJLyoKISAJCQkJCSAqIGluIGFsbCBvdGhlciBjYXNlcywgdGFr ZSB0aGUgY2hhciBhZnRlciAnXCcKISAJCQkJCSAqIGxpdGVyYWxseQogIAkJCQkJICovCiAg CQkJfQohIAkJfQohIAkJYXBwZW5kU3RyaW5nSW5mb0NoYXJNYWNybygmYXR0cmlidXRlX2J1 ZiwgYyk7CiAgCX0KICAKISAJLyogY2hlY2sgd2hldGhlciByYXcgaW5wdXQgbWF0Y2hlZCBu dWxsIG1hcmtlciAqLwohIAlpbnB1dF9sZW4gPSBlbmRfY3Vyc29yIC0gc3RhcnRfY3Vyc29y OwohIAlpZiAoaW5wdXRfbGVuID09IHN0cmxlbihudWxsX3ByaW50KSAmJgohIAkJc3RybmNt cCgmbGluZV9idWYuZGF0YVtzdGFydF9jdXJzb3JdLCBudWxsX3ByaW50LCBpbnB1dF9sZW4p ID09IDApCiAgCQkqaXNudWxsID0gdHJ1ZTsKISAJZWxzZQohIAkJKmlzbnVsbCA9IGZhbHNl OwogIAohIAlyZXR1cm4gYXR0cmlidXRlX2J1Zi5kYXRhOwohIH0KICAKICAKICAvKgohICAq IFJlYWQgdGhlIHZhbHVlIG9mIGEgc2luZ2xlIGF0dHJpYnV0ZSBpbiBDU1YgbW9kZSwKICAg KiBwZXJmb3JtaW5nIGRlLWVzY2FwaW5nIGFzIG5lZWRlZC4gRXNjYXBpbmcgZG9lcyBub3Qg Zm9sbG93IHRoZSBub3JtYWwKICAgKiBQb3N0Z3JlU1FMIHRleHQgbW9kZSwgYnV0IGluc3Rl YWQgInN0YW5kYXJkIiAoaS5lLiBjb21tb24pIENTViB1c2FnZS4KICAgKgotLS0gMjc3Mywy OTQ0IC0tLS0KICAJCQkJY2FzZSAnNSc6CiAgCQkJCWNhc2UgJzYnOgogIAkJCQljYXNlICc3 JzoKISAJCQkJCQkJb2N0X3ZhbCA9IE9DVFZBTFVFKG5leHRjKTsKISAJCQkJCQkJbmV4dGMg PSAqKHNjYW5fZW5kICsgMik7CiAgCiEgCQkJCQkJCS8qCiEgCQkJCQkJCSAqIChubyBuZWVk IGZvciBvdXQgYmFkIGFjY2VzcyBjaGVjayBzaW5jZSBsaW5lIGlmCiEgCQkJCQkJCQkqIGJ1 ZmZlcmVkKQohIAkJCQkJCQkgKi8KISAJCQkJCQkJaWYgKElTT0NUQUwobmV4dGMpKQogIAkJ CQkJCQkJewohIAkJCQkJCQkJc2tpcCsrOwohIAkJCQkJCQkJb2N0X3ZhbCA9IChvY3RfdmFs IDw8IDMpICsgT0NUVkFMVUUobmV4dGMpOwohIAkJCQkJCQkJbmV4dGMgPSAqKHNjYW5fZW5k ICsgMyk7CiEgCQkJCQkJCQlpZiAoSVNPQ1RBTChuZXh0YykpCiAgCQkJCQkJCQkJewohIAkJ CQkJCQkJCXNraXArKzsKISAJCQkJCQkJCQlvY3RfdmFsID0gKG9jdF92YWwgPDwgMykgKyBP Q1RWQUxVRShuZXh0Yyk7CiAgCQkJCQkJCQkJfQogIAkJCQkJCQkJfQohIAkJCQkJCQkJbmV3 YyA9IG9jdF92YWwgJiAwMzc3OwkvKiB0aGUgZXNjYXBlZCBieXRlIHZhbHVlICovCiAgCQkJ CQlicmVhazsKICAJCQkJY2FzZSAneCc6CiAgCQkJCQkvKiBIYW5kbGUgXHgzRiAqLwohIAkJ CQkJCQloZXhjaGFyID0gKihzY2FuX2VuZCArIDIpOwogIAogIAkJCQkJCWlmIChpc3hkaWdp dChoZXhjaGFyKSkKICAJCQkJCQl7CiEgCQkJCQkJCQlza2lwKys7CiEgCQkJCQkJCQloZXhf dmFsID0gR2V0RGVjaW1hbEZyb21IZXgoaGV4Y2hhcik7CiAgCiEgCQkJCQkJCQloZXhjaGFy ID0gKihzY2FuX2VuZCArIDMpOwogIAkJCQkJCQkJaWYgKGlzeGRpZ2l0KGhleGNoYXIpKQog IAkJCQkJCQkJewohIAkJCQkJCQkJCXNraXArKzsKISAJCQkJCQkJCQloZXhfdmFsID0gKGhl eF92YWwgPDwgNCkgKyBHZXREZWNpbWFsRnJvbUhleChoZXhjaGFyKTsKICAJCQkJCQl9Cisg CQkJCQkJCQluZXdjID0gaGV4X3ZhbCAmIDB4ZmY7CiAgCQkJCQl9CisgCQkJCQkJCQllbHNl IC8qICJceCIgd2l0aCBubyBoZXggdmFsdWUgKi8KKyAJCQkJCQkJCQluZXdjID0gbmV4dGM7 CiAgCQkJCQlicmVhazsKICAJCQkJY2FzZSAnYic6CiEgCQkJCQkJCW5ld2MgPSAnXGInOwog IAkJCQkJYnJlYWs7CiAgCQkJCWNhc2UgJ2YnOgohIAkJCQkJCQluZXdjID0gJ1xmJzsKICAJ CQkJCWJyZWFrOwogIAkJCQljYXNlICduJzoKISAJCQkJCQkJbmV3YyA9ICdcbic7CiAgCQkJ CQlicmVhazsKICAJCQkJY2FzZSAncic6CiEgCQkJCQkJCW5ld2MgPSAnXHInOwogIAkJCQkJ YnJlYWs7CiAgCQkJCWNhc2UgJ3QnOgohIAkJCQkJCQluZXdjID0gJ1x0JzsKICAJCQkJCWJy ZWFrOwogIAkJCQljYXNlICd2JzoKISAJCQkJCQkJbmV3YyA9ICdcdic7CiEgCQkJCQkJCWJy ZWFrOwohIAkJCQkJCWRlZmF1bHQ6CiEgCQkJCQkJCWlmIChuZXh0YyA9PSBkZWxpbWMpCiEg CQkJCQkJCQluZXdjID0gZGVsaW1jOwohIAkJCQkJCQllbHNlIGlmIChuZXh0YyA9PSBlc2Nh cGVjKQohIAkJCQkJCQkJbmV3YyA9IGVzY2FwZWM7CiEgCQkJCQkJCWVsc2UKISAJCQkJCQkJ CS8qIG5vIGVzY2FwZSBzZXF1ZW5jZSwgdGFrZSBuZXh0IGNoYXIgbGl0ZXJhbHkgKi8KISAJ CQkJCQkJCW5ld2MgPSBuZXh0YzsKICAJCQkJCWJyZWFrOworIAkJCQkJfQorIAkJCQkJCisg CQkJCQkvKiB1cGRhdGUgdG8gY3VycmVudCBsZW5ndGgsIGFkZCBlc2NhcGUgYW5kIGVzY2Fw ZWQgY2hhcnMgICovCisgCQkJCQlhdHRyX3ByZV9sZW4gKz0gc2Nhbl9lbmQgLSBzY2FuX3N0 YXJ0ICsgMjsKKyAJCQkJCS8qIHVwZGF0ZSB0byBjdXJyZW50IGxlbmd0aCwgZXNjYXBlZCBj aGFyICovCisgCQkJCQlhdHRyX3Bvc3RfbGVuICs9IHNjYW5fZW5kIC0gc2Nhbl9zdGFydCAr IDE7CiAgCiAgCQkJCQkvKgohIAkJCQkJICogTmVlZCB0byBnZXQgcmlkIG9mIHRoZSBlc2Nh cGUgY2hhcmFjdGVyLiBUaGlzIGlzIGRvbmUgYnkKISAJCQkJCSAqIGxvYWRpbmcgdGhlIGNo dW5rIHVwIHRvIGluY2x1ZGluZyB0aGUgZXNjYXBlIGNoYXJhY3RlcgohIAkJCQkJICogaW50 byB0aGUgYXR0cmlidXRlIGJ1ZmZlci4gVGhlbiBvdmVyd3JpdHRpbmcgdGhlIGJhY2tzbGFz aAohIAkJCQkJICogd2l0aCB0aGUgZXNjYXBlZCBzZXF1ZW5jZSBvciBjaGFyLCBhbmQgY29u dGludWluZyB0byBzY2FuCiEgCQkJCQkgKiBmcm9tICphZnRlciogdGhlIGNoYXIgdGhhbiBp cyBhZnRlciB0aGUgZXNjYXBlIGluIGxpbmUgYnVmLgogIAkJCQkJICovCisgCQkJCQkqKGF0 dHJfYnVmLmRhdGEgKyBhdHRyX2J1Zi5sZW4gLSAxKSA9IG5ld2M7CisgCQkJCQlsaW5lX2J1 Zi5jdXJzb3IgPSBzY2FuX2VuZCAtIGxpbmVfYnVmLmRhdGEgKyBza2lwOworIAkJCQkJc2Nh bl9zdGFydCA9IHNjYW5fZW5kICsgc2tpcDsKKyAJCQkJCWNodW5rX3N0YXJ0ID0gbGluZV9i dWYuY3Vyc29yOworIAkJCQkJY2h1bmtfbGVuID0gMDsKICAJCQl9CiEgCQkJCQohIAkJCX0J CQkJCQkvKiBlbmQgZGVsaW1pdGVyL2JhY2tzbGFzaCAqLwohIAohIAkJfQkJCQkJCQkvKiBl bmQgbGluZSBidWZmZXIgc2Nhbi4gKi8KISAKISAJLyoKISAJICogUmVwbGFjZSBhbGwgZGVs aW1pdGVycyB3aXRoIE5VTEwgZm9yIHN0cmluZyB0ZXJtaW5hdGlvbi4KISAJICogTk9URTog b25seSBkZWxpbWl0ZXJzIChOT1QgbmVjZXNzYXJpbHkgYWxsIGRlbGltYykgYXJlIHJlcGxh Y2VkLgohIAkgKiBFeGFtcGxlIChkZWxpbWMgPSAnfCcpOgohIAkgKiAtIEJlZm9yZTogIGYg IDEJfCAgZiAgXHwgIDIgIHwJZiAgMwohIAkgKiAtIEFmdGVyIDogIGYgIDEgXDAgIGYgICB8 ICAyIFwwCWYgIDMKISAJICovCiEgCWZvciAoYXR0cmlidXRlID0gMTsgYXR0cmlidXRlIDwg bnVtX3BoeXNfYXR0cnM7IGF0dHJpYnV0ZSsrKQohIAl7CiEgCQlpZiAoYXR0cl9vZmZzZXRz W2F0dHJpYnV0ZV0gIT0gMCkKISAJCQkqKGF0dHJfYnVmLmRhdGEgKyBhdHRyX29mZnNldHNb YXR0cmlidXRlXSAtIDEpID0gJ1wwJzsKICAJfQogIAohIH0KISAKISAKISAvKgohICAqIFJl YWQgYSBiaW5hcnkgYXR0cmlidXRlCiEgICovCiEgc3RhdGljIERhdHVtCiEgQ29weVJlYWRC aW5hcnlBdHRyaWJ1dGUoaW50IGNvbHVtbl9ubywgRm1nckluZm8gKmZsaW5mbywgT2lkIHR5 cGlvcGFyYW0sCiEgCQkJCQkJYm9vbCAqaXNudWxsKQohIHsKISAJaW50MzIJCWZsZF9zaXpl OwohIAlEYXR1bQkJcmVzdWx0OwohIAkKISAJZmxkX3NpemUgPSBDb3B5R2V0SW50MzIoKTsK ISAJaWYgKENvcHlHZXRFb2YoKSkKISAJCWVyZXBvcnQoRVJST1IsCiEgCQkJCShlcnJjb2Rl KEVSUkNPREVfQkFEX0NPUFlfRklMRV9GT1JNQVQpLAohIAkJCQkgZXJybXNnKCJ1bmV4cGVj dGVkIEVPRiBpbiBDT1BZIGRhdGEiKSkpOwohIAlpZiAoZmxkX3NpemUgPT0gLTEpCiEgCXsK ICAJCSppc251bGwgPSB0cnVlOwohIAkJcmV0dXJuIChEYXR1bSkgMDsKISAJfQohIAlpZiAo ZmxkX3NpemUgPCAwKQohIAkJZXJlcG9ydChFUlJPUiwKISAJCQkJKGVycmNvZGUoRVJSQ09E RV9CQURfQ09QWV9GSUxFX0ZPUk1BVCksCiEgCQkJCSBlcnJtc2coImludmFsaWQgZmllbGQg c2l6ZSIpKSk7CiAgCiEgCS8qIHJlc2V0IGF0dHJfYnVmIHRvIGVtcHR5LCBhbmQgbG9hZCBy YXcgZGF0YSBpbiBpdCAqLwohIAlhdHRyX2J1Zi5sZW4gPSAwOwohIAlhdHRyX2J1Zi5kYXRh WzBdID0gJ1wwJzsKISAJYXR0cl9idWYuY3Vyc29yID0gMDsKISAJCiEgCWVubGFyZ2VTdHJp bmdJbmZvKCZhdHRyX2J1ZiwgZmxkX3NpemUpOwohIAkKISAJQ29weUdldERhdGEoYXR0cl9i dWYuZGF0YSwgZmxkX3NpemUpOwohIAlpZiAoQ29weUdldEVvZigpKQohIAkJZXJlcG9ydChF UlJPUiwKISAJCQkJKGVycmNvZGUoRVJSQ09ERV9CQURfQ09QWV9GSUxFX0ZPUk1BVCksCiEg CQkJCSBlcnJtc2coInVuZXhwZWN0ZWQgRU9GIGluIENPUFkgZGF0YSIpKSk7CiAgCisgCWF0 dHJfYnVmLmxlbiA9IGZsZF9zaXplOworIAlhdHRyX2J1Zi5kYXRhW2ZsZF9zaXplXSA9ICdc MCc7CisgCQorIAkvKiBDYWxsIHRoZSBjb2x1bW4gdHlwZSdzIGJpbmFyeSBpbnB1dCBjb252 ZXJ0ZXIgKi8KKyAJcmVzdWx0ID0gRnVuY3Rpb25DYWxsMihmbGluZm8sCisgCQkJCQkJICAg UG9pbnRlckdldERhdHVtKCZhdHRyX2J1ZiksCisgCQkJCQkJICAgT2JqZWN0SWRHZXREYXR1 bSh0eXBpb3BhcmFtKSk7CisgCQorIAkvKiBUcm91YmxlIGlmIGl0IGRpZG4ndCBlYXQgdGhl IHdob2xlIGJ1ZmZlciAqLworIAlpZiAoYXR0cl9idWYuY3Vyc29yICE9IGF0dHJfYnVmLmxl bikKKyAJCWVyZXBvcnQoRVJST1IsCisgCQkJCShlcnJjb2RlKEVSUkNPREVfSU5WQUxJRF9C SU5BUllfUkVQUkVTRU5UQVRJT04pLAorIAkJCQkgZXJybXNnKCJpbmNvcnJlY3QgYmluYXJ5 IGRhdGEgZm9ybWF0IikpKTsKKyAJCisgCSppc251bGwgPSBmYWxzZTsKKyAJcmV0dXJuIHJl c3VsdDsKKyB9CiAgCiAgLyoKISAgKiBSZWFkIGFsbCB0aGUgYXR0cmlidXRlcyBvZiB0aGUg ZGF0YSBsaW5lIGluIENTViBtb2RlLAogICAqIHBlcmZvcm1pbmcgZGUtZXNjYXBpbmcgYXMg bmVlZGVkLiBFc2NhcGluZyBkb2VzIG5vdCBmb2xsb3cgdGhlIG5vcm1hbAogICAqIFBvc3Rn cmVTUUwgdGV4dCBtb2RlLCBidXQgaW5zdGVhZCAic3RhbmRhcmQiIChpLmUuIGNvbW1vbikg Q1NWIHVzYWdlLgogICAqCioqKioqKioqKioqKioqKgoqKiogMjQ0OCwyNDcyICoqKioKICAg KgogICAqIG51bGxfcHJpbnQgaXMgdGhlIG51bGwgbWFya2VyIHN0cmluZy4gIE5vdGUgdGhh dCB0aGlzIGlzIGNvbXBhcmVkIHRvCiAgICogdGhlIHByZS1kZS1lc2NhcGVkIGlucHV0IHN0 cmluZyAodGh1cyBpZiBpdCBpcyBxdW90ZWQgaXQgaXMgbm90IGEgTlVMTCkuCi0gICoKLSAg KiAqcmVzdWx0IGlzIHNldCB0byBpbmRpY2F0ZSB3aGF0IHRlcm1pbmF0ZWQgdGhlIHJlYWQ6 Ci0gICoJCU5PUk1BTF9BVFRSOgljb2x1bW4gZGVsaW1pdGVyCi0gICoJCUVORF9PRl9MSU5F OgllbmQgb2YgbGluZQotICAqCQlVTlRFUk1JTkFURURfRklFTEQgbm8gcXVvdGUgZGV0ZWN0 ZWQgYXQgZW5kIG9mIGEgcXVvdGVkIGZpZWxkCi0gICoKLSAgKiBJbiBhbnkgY2FzZSwgdGhl IHN0cmluZyByZWFkIHVwIHRvIHRoZSB0ZXJtaW5hdG9yIChvciBlbmQgb2YgZmlsZSkKLSAg KiBpcyByZXR1cm5lZC4KLSAgKgotICAqICppc251bGwgaXMgc2V0IHRydWUgb3IgZmFsc2Ug ZGVwZW5kaW5nIG9uIHdoZXRoZXIgdGhlIGlucHV0IG1hdGNoZWQKLSAgKiB0aGUgbnVsbCBt YXJrZXIuICBOb3RlIHRoYXQgdGhlIGNhbGxlciBjYW5ub3QgY2hlY2sgdGhpcyBzaW5jZSB0 aGUKLSAgKiByZXR1cm5lZCBzdHJpbmcgd2lsbCBiZSB0aGUgcG9zdC1kZS1lc2NhcGluZyBl cXVpdmFsZW50LCB3aGljaCBtYXkKLSAgKiBsb29rIHRoZSBzYW1lIGFzIHNvbWUgdmFsaWQg ZGF0YSBzdHJpbmcuCiAgICotLS0tLS0tLS0tCiAgICovCiEgCiEgc3RhdGljIGNoYXIgKgoh IENvcHlSZWFkQXR0cmlidXRlQ1NWKGNvbnN0IGNoYXIgKmRlbGltLCBjb25zdCBjaGFyICpu dWxsX3ByaW50LCBjaGFyICpxdW90ZSwKISAJCQkJCSBjaGFyICplc2NhcGUsIENvcHlSZWFk UmVzdWx0ICpyZXN1bHQsIGJvb2wgKmlzbnVsbCkKICB7CiAgCWNoYXIJCWRlbGltYyA9IGRl bGltWzBdOwogIAljaGFyCQlxdW90ZWMgPSBxdW90ZVswXTsKLS0tIDI5NDcsMjk1OCAtLS0t CiAgICoKICAgKiBudWxsX3ByaW50IGlzIHRoZSBudWxsIG1hcmtlciBzdHJpbmcuICBOb3Rl IHRoYXQgdGhpcyBpcyBjb21wYXJlZCB0bwogICAqIHRoZSBwcmUtZGUtZXNjYXBlZCBpbnB1 dCBzdHJpbmcgKHRodXMgaWYgaXQgaXMgcXVvdGVkIGl0IGlzIG5vdCBhIE5VTEwpLgogICAq LS0tLS0tLS0tLQogICAqLwohIHN0YXRpYyB2b2lkCiEgQ29weVJlYWRBdHRyaWJ1dGVzQ1NW KGNvbnN0IGNoYXIgKmRlbGltLCBjb25zdCBjaGFyICpudWxsX3ByaW50LCBjaGFyICpxdW90 ZSwKISAJCQkJCSAgY2hhciAqZXNjYXBlLCBpbnQgbnVsbF9wcmludF9sZW4sIGNoYXIgKm51 bGxzLCBMaXN0ICphdHRudW1saXN0LCAKISAJCQkJCSAgaW50ICphdHRyX29mZnNldHMsIGlu dCBudW1fcGh5c19hdHRycywgRm9ybV9wZ19hdHRyaWJ1dGUgKmF0dHIpCiAgewogIAljaGFy CQlkZWxpbWMgPSBkZWxpbVswXTsKICAJY2hhcgkJcXVvdGVjID0gcXVvdGVbMF07CioqKioq KioqKioqKioqKgoqKiogMjQ3NywyNTAyICoqKioKICAJaW50CQkJaW5wdXRfbGVuOwogIAli b29sCQlpbl9xdW90ZSA9IGZhbHNlOwogIAlib29sCQlzYXdfcXVvdGUgPSBmYWxzZTsKICAK ISAJLyogcmVzZXQgYXR0cmlidXRlX2J1ZiB0byBlbXB0eSAqLwohIAlhdHRyaWJ1dGVfYnVm LmxlbiA9IDA7CiEgCWF0dHJpYnV0ZV9idWYuZGF0YVswXSA9ICdcMCc7CiAgCi0gCS8qIHNl dCBkZWZhdWx0IHN0YXR1cyAqLwotIAkqcmVzdWx0ID0gRU5EX09GX0xJTkU7CiAgCiAgCWZv ciAoOzspCiAgCXsKICAJCWVuZF9jdXJzb3IgPSBsaW5lX2J1Zi5jdXJzb3I7CiAgCQlpZiAo bGluZV9idWYuY3Vyc29yID49IGxpbmVfYnVmLmxlbikKICAJCQlicmVhazsKICAJCWMgPSBs aW5lX2J1Zi5kYXRhW2xpbmVfYnVmLmN1cnNvcisrXTsKICAKICAJCS8qIHVucXVvdGVkIGZp ZWxkIGRlbGltaXRlciAgKi8KICAJCWlmICghaW5fcXVvdGUgJiYgYyA9PSBkZWxpbWMpCiAg CQl7CiEgCQkJKnJlc3VsdCA9IE5PUk1BTF9BVFRSOwohIAkJCWJyZWFrOwogIAkJfQogIAog IAkJLyogc3RhcnQgb2YgcXVvdGVkIGZpZWxkIChvciBwYXJ0IG9mIGZpZWxkKSAqLwotLS0g Mjk2MywzMDQzIC0tLS0KICAJaW50CQkJaW5wdXRfbGVuOwogIAlib29sCQlpbl9xdW90ZSA9 IGZhbHNlOwogIAlib29sCQlzYXdfcXVvdGUgPSBmYWxzZTsKKyAJaW50CQkJYXR0bnVtOwkJ CS8qIGF0dHJpYnV0ZSBudW1iZXIgYmVpbmcgcGFyc2VkICovCisgCWludAkJCW07CQkJCS8q IGF0dHJpYnV0ZSBpbmRleCBiZWluZyBwYXJzZWQgKi8KKyAJTGlzdENlbGwgICAqY3VyOwkJ CS8qIGN1cnNvciB0byBhdHRyaWJ1dGUgbGlzdCB1c2VkIGZvciB0aGlzIENPUFkgKi8KICAK ISAJLyoKISAJICogaW5pdCB2YXJpYWJsZXMgZm9yIGF0dHJpYnV0ZSBzY2FuCiEgCSAqLwoh IAlhdHRyX2J1Zi5sZW4gPSAwOwohIAlhdHRyX2J1Zi5kYXRhWzBdID0gJ1wwJzsKISAJYXR0 cl9idWYuY3Vyc29yID0gMDsKISAJCiEgCWN1ciA9IGxpc3RfaGVhZChhdHRudW1saXN0KTsK ISAJYXR0bnVtID0gbGZpcnN0X2ludChjdXIpOwohIAltID0gYXR0bnVtIC0gMTsKISAJaW5w dXRfbGVuID0gMDsKICAKICAKICAJZm9yICg7OykKICAJewogIAkJZW5kX2N1cnNvciA9IGxp bmVfYnVmLmN1cnNvcjsKKyAJCQorIAkJLyogZmluaXNoZWQgcHJvY2Vzc2luZyBhdHRyaWJ1 dGVzIGluIGxpbmUgKi8KICAJCWlmIChsaW5lX2J1Zi5jdXJzb3IgPj0gbGluZV9idWYubGVu KQorIAkJeworIAkJCS8qIGNoZWNrIHdoZXRoZXIgcmF3IGlucHV0IG1hdGNoZWQgbnVsbCBt YXJrZXIgKi8KKyAJCQlpbnB1dF9sZW4gPSBlbmRfY3Vyc29yIC0gc3RhcnRfY3Vyc29yOwor IAkJCWlmICghc2F3X3F1b3RlICYmIGlucHV0X2xlbiA9PSBudWxsX3ByaW50X2xlbiAmJgor IAkJCQlzdHJuY21wKCZsaW5lX2J1Zi5kYXRhW3N0YXJ0X2N1cnNvcl0sIG51bGxfcHJpbnQs IGlucHV0X2xlbikgPT0gMCkKKyAJCQkJbnVsbHNbbV0gPSAnbic7CisgCQkJZWxzZQorIAkJ CQludWxsc1ttXSA9ICcgJzsKKyAJCQkKKyAJCQlpZiAoaW5fcXVvdGUpCisgCQkJCWVyZXBv cnQoRVJST1IsCisgCQkJCQkJKGVycmNvZGUoRVJSQ09ERV9CQURfQ09QWV9GSUxFX0ZPUk1B VCksCisgCQkJCQkJIGVycm1zZygidW50ZXJtaW5hdGVkIENTViBxdW90ZWQgZmllbGQiKSkp OworIAkJCQorIAkJCWlmIChsbmV4dChjdXIpICE9IE5VTEwpCisgCQkJCWVyZXBvcnQoRVJS T1IsCisgCQkJCQkJKGVycmNvZGUoRVJSQ09ERV9CQURfQ09QWV9GSUxFX0ZPUk1BVCksCisg CQkJCQkJIGVycm1zZygibWlzc2luZyBkYXRhIGZvciBjb2x1bW4gXCIlc1wiIiwKKyAJCQkJ CQkJCU5hbWVTdHIoYXR0clttICsgMV0tPmF0dG5hbWUpKSkpOwogIAkJCWJyZWFrOworIAkJ fQorIAkJCiAgCQljID0gbGluZV9idWYuZGF0YVtsaW5lX2J1Zi5jdXJzb3IrK107CiAgCiAg CQkvKiB1bnF1b3RlZCBmaWVsZCBkZWxpbWl0ZXIgICovCiAgCQlpZiAoIWluX3F1b3RlICYm IGMgPT0gZGVsaW1jKQogIAkJewohIAkJCS8qIGNoZWNrIHdoZXRoZXIgcmF3IGlucHV0IG1h dGNoZWQgbnVsbCBtYXJrZXIgKi8KISAJCQlpbnB1dF9sZW4gPSBlbmRfY3Vyc29yIC0gc3Rh cnRfY3Vyc29yOwohIAkJCWlmICghc2F3X3F1b3RlICYmIGlucHV0X2xlbiA9PSBudWxsX3By aW50X2xlbiAmJgohIAkJCQlzdHJuY21wKCZsaW5lX2J1Zi5kYXRhW3N0YXJ0X2N1cnNvcl0s IG51bGxfcHJpbnQsIGlucHV0X2xlbikgPT0gMCkKISAJCQkJbnVsbHNbbV0gPSAnbic7CiEg CQkJZWxzZQohIAkJCQludWxsc1ttXSA9ICcgJzsKISAgICAgICAgICAgICAKISAJCQkvKiB0 ZXJtaW5hdGUgYXR0ciBzdHJpbmcgd2l0aCAnXDAnICovCiEgCQkJYXBwZW5kU3RyaW5nSW5m b0NoYXJNYWNybygmYXR0cl9idWYsICdcMCcpOwohIAkJCWF0dHJfYnVmLmN1cnNvcisrOwoh IAkJCQohIAkJCS8qIHNldHVwIG5leHQgYXR0cmlidXRlIHNjYW4gKi8KISAJCQljdXIgPSBs bmV4dChjdXIpOwohIAkJCQohIAkJCWlmIChjdXIgPT0gTlVMTCkKISAJCQkJZXJlcG9ydChF UlJPUiwKISAJCQkJCQkoZXJyY29kZShFUlJDT0RFX0JBRF9DT1BZX0ZJTEVfRk9STUFUKSwK ISAJCQkJCQkgZXJybXNnKCJleHRyYSBkYXRhIGFmdGVyIGxhc3QgZXhwZWN0ZWQgY29sdW1u IikpKTsKISAJCQkKISAJCQlhdHRudW0gPSBsZmlyc3RfaW50KGN1cik7CiEgCQkJbSA9IGF0 dG51bSAtIDE7CiEgCQkJYXR0cl9vZmZzZXRzW21dID0gYXR0cl9idWYuY3Vyc29yOwohIAkJ CXN0YXJ0X2N1cnNvciA9IGxpbmVfYnVmLmN1cnNvcjsKISAJCQljb250aW51ZTsKICAJCX0K ICAKICAJCS8qIHN0YXJ0IG9mIHF1b3RlZCBmaWVsZCAob3IgcGFydCBvZiBmaWVsZCkgKi8K KioqKioqKioqKioqKioqCioqKiAyNTIwLDI1MjcgKioqKgogIAogIAkJCQlpZiAobmV4dGMg PT0gZXNjYXBlYyB8fCBuZXh0YyA9PSBxdW90ZWMpCiAgCQkJCXsKISAJCQkJCWFwcGVuZFN0 cmluZ0luZm9DaGFyTWFjcm8oJmF0dHJpYnV0ZV9idWYsIG5leHRjKTsKICAJCQkJCWxpbmVf YnVmLmN1cnNvcisrOwogIAkJCQkJY29udGludWU7CiAgCQkJCX0KICAJCQl9Ci0tLSAzMDYx LDMwNjkgLS0tLQogIAogIAkJCQlpZiAobmV4dGMgPT0gZXNjYXBlYyB8fCBuZXh0YyA9PSBx dW90ZWMpCiAgCQkJCXsKISAJCQkJCWFwcGVuZFN0cmluZ0luZm9DaGFyTWFjcm8oJmF0dHJf YnVmLCBuZXh0Yyk7CiAgCQkJCQlsaW5lX2J1Zi5jdXJzb3IrKzsKKyAJCQkJCWF0dHJfYnVm LmN1cnNvcisrOwogIAkJCQkJY29udGludWU7CiAgCQkJCX0KICAJCQl9CioqKioqKioqKioq KioqKgoqKiogMjUzNywyNjE2ICoqKioKICAJCQlpbl9xdW90ZSA9IGZhbHNlOwogIAkJCWNv bnRpbnVlOwogIAkJfQohIAkJYXBwZW5kU3RyaW5nSW5mb0NoYXJNYWNybygmYXR0cmlidXRl X2J1ZiwgYyk7CiAgCX0KICAKLSAJaWYgKGluX3F1b3RlKQotIAkJKnJlc3VsdCA9IFVOVEVS TUlOQVRFRF9GSUVMRDsKLSAKLSAJLyogY2hlY2sgd2hldGhlciByYXcgaW5wdXQgbWF0Y2hl ZCBudWxsIG1hcmtlciAqLwotIAlpbnB1dF9sZW4gPSBlbmRfY3Vyc29yIC0gc3RhcnRfY3Vy c29yOwotIAlpZiAoIXNhd19xdW90ZSAmJiBpbnB1dF9sZW4gPT0gc3RybGVuKG51bGxfcHJp bnQpICYmCi0gCQlzdHJuY21wKCZsaW5lX2J1Zi5kYXRhW3N0YXJ0X2N1cnNvcl0sIG51bGxf cHJpbnQsIGlucHV0X2xlbikgPT0gMCkKLSAJCSppc251bGwgPSB0cnVlOwotIAllbHNlCi0g CQkqaXNudWxsID0gZmFsc2U7Ci0gCi0gCXJldHVybiBhdHRyaWJ1dGVfYnVmLmRhdGE7CiAg fQogIAogIC8qCiEgICogUmVhZCBhIGJpbmFyeSBhdHRyaWJ1dGUKICAgKi8KISBzdGF0aWMg RGF0dW0KISBDb3B5UmVhZEJpbmFyeUF0dHJpYnV0ZShpbnQgY29sdW1uX25vLCBGbWdySW5m byAqZmxpbmZvLAohIAkJCQkJCU9pZCB0eXBpb3BhcmFtLCBpbnQzMiB0eXBtb2QsCiAgCQkJ CQkJYm9vbCAqaXNudWxsKQogIHsKISAJaW50MzIJCWZsZF9zaXplOwohIAlEYXR1bQkJcmVz dWx0OwohIAohIAlmbGRfc2l6ZSA9IENvcHlHZXRJbnQzMigpOwohIAlpZiAoQ29weUdldEVv ZigpKQohIAkJZXJlcG9ydChFUlJPUiwKISAJCQkJKGVycmNvZGUoRVJSQ09ERV9CQURfQ09Q WV9GSUxFX0ZPUk1BVCksCiEgCQkJCSBlcnJtc2coInVuZXhwZWN0ZWQgRU9GIGluIENPUFkg ZGF0YSIpKSk7CiEgCWlmIChmbGRfc2l6ZSA9PSAtMSkKISAJewohIAkJKmlzbnVsbCA9IHRy dWU7CiEgCQlyZXR1cm4gKERhdHVtKSAwOwogIAl9CiEgCWlmIChmbGRfc2l6ZSA8IDApCiEg CQllcmVwb3J0KEVSUk9SLAohIAkJCQkoZXJyY29kZShFUlJDT0RFX0JBRF9DT1BZX0ZJTEVf Rk9STUFUKSwKISAJCQkJIGVycm1zZygiaW52YWxpZCBmaWVsZCBzaXplIikpKTsKICAKISAJ LyogcmVzZXQgYXR0cmlidXRlX2J1ZiB0byBlbXB0eSwgYW5kIGxvYWQgcmF3IGRhdGEgaW4g aXQgKi8KISAJYXR0cmlidXRlX2J1Zi5sZW4gPSAwOwohIAlhdHRyaWJ1dGVfYnVmLmRhdGFb MF0gPSAnXDAnOwohIAlhdHRyaWJ1dGVfYnVmLmN1cnNvciA9IDA7CiAgCiEgCWVubGFyZ2VT dHJpbmdJbmZvKCZhdHRyaWJ1dGVfYnVmLCBmbGRfc2l6ZSk7CiAgCi0gCUNvcHlHZXREYXRh KGF0dHJpYnV0ZV9idWYuZGF0YSwgZmxkX3NpemUpOwotIAlpZiAoQ29weUdldEVvZigpKQot IAkJZXJlcG9ydChFUlJPUiwKLSAJCQkJKGVycmNvZGUoRVJSQ09ERV9CQURfQ09QWV9GSUxF X0ZPUk1BVCksCi0gCQkJCSBlcnJtc2coInVuZXhwZWN0ZWQgRU9GIGluIENPUFkgZGF0YSIp KSk7CiAgCiEgCWF0dHJpYnV0ZV9idWYubGVuID0gZmxkX3NpemU7CiEgCWF0dHJpYnV0ZV9i dWYuZGF0YVtmbGRfc2l6ZV0gPSAnXDAnOwogIAohIAkvKiBDYWxsIHRoZSBjb2x1bW4gdHlw ZSdzIGJpbmFyeSBpbnB1dCBjb252ZXJ0ZXIgKi8KISAJcmVzdWx0ID0gRnVuY3Rpb25DYWxs MyhmbGluZm8sCiEgCQkJCQkJICAgUG9pbnRlckdldERhdHVtKCZhdHRyaWJ1dGVfYnVmKSwK ISAJCQkJCQkgICBPYmplY3RJZEdldERhdHVtKHR5cGlvcGFyYW0pLAohIAkJCQkJCSAgIElu dDMyR2V0RGF0dW0odHlwbW9kKSk7CiAgCi0gCS8qIFRyb3VibGUgaWYgaXQgZGlkbid0IGVh dCB0aGUgd2hvbGUgYnVmZmVyICovCi0gCWlmIChhdHRyaWJ1dGVfYnVmLmN1cnNvciAhPSBh dHRyaWJ1dGVfYnVmLmxlbikKLSAJCWVyZXBvcnQoRVJST1IsCi0gCQkJCShlcnJjb2RlKEVS UkNPREVfSU5WQUxJRF9CSU5BUllfUkVQUkVTRU5UQVRJT04pLAotIAkJCQkgZXJybXNnKCJp bmNvcnJlY3QgYmluYXJ5IGRhdGEgZm9ybWF0IikpKTsKICAKLSAJKmlzbnVsbCA9IGZhbHNl OwotIAlyZXR1cm4gcmVzdWx0OwotIH0KICAKICAvKgogICAqIFNlbmQgdGV4dCByZXByZXNl bnRhdGlvbiBvZiBvbmUgYXR0cmlidXRlLCB3aXRoIGNvbnZlcnNpb24gYW5kIGVzY2FwaW5n Ci0tLSAzMDc5LDMxNDUgLS0tLQogIAkJCWluX3F1b3RlID0gZmFsc2U7CiAgCQkJY29udGlu dWU7CiAgCQl9CiEgCQlhcHBlbmRTdHJpbmdJbmZvQ2hhck1hY3JvKCZhdHRyX2J1ZiwgYyk7 CiEgCQlhdHRyX2J1Zi5jdXJzb3IrKzsKICAJfQogIAogIH0KICAKICAvKgohICAqIFJlYWQg dGhlIGZpcnN0IGF0dHJpYnV0ZS4gVGhpcyBpcyBtYWlubHkgdXNlZCB0byBtYWludGFpbiBz dXBwb3J0CiEgICogZm9yIGFuIE9JRCBjb2x1bW4uIEFsbCB0aGUgcmVzdCBvZiB0aGUgY29s dW1ucyB3aWxsIGJlIHJlYWQgYXQgb25jZSB3aXRoCiEgICogQ29weVJlYWRBdHRyaWJ1dGVz VGV4dC4KICAgKi8KISBzdGF0aWMgY2hhciAqCiEgQ29weVJlYWRPaWRBdHRyKGNvbnN0IGNo YXIgKmRlbGltLCBjb25zdCBjaGFyICpudWxsX3ByaW50LCBpbnQgbnVsbF9wcmludF9sZW4s CiAgCQkJCQkJYm9vbCAqaXNudWxsKQogIHsKISAJY2hhcgkJZGVsaW1jID0gZGVsaW1bMF07 CiEgCWNoYXIJICAgKnN0YXJ0X2xvYyA9IGxpbmVfYnVmLmRhdGEgKyBsaW5lX2J1Zi5jdXJz b3I7CiEgCWNoYXIJICAgKmVuZF9sb2M7CiEgCWludAkJCWF0dHJfbGVuID0gMDsKISAJaW50 CQkJYnl0ZXNfcmVtYWluaW5nOwohIAohIAkvKiByZXNldCBhdHRyaWJ1dGUgYnVmIHRvIGVt cHR5ICovCiEgCWF0dHJfYnVmLmxlbiA9IDA7CiEgCWF0dHJfYnVmLmRhdGFbMF0gPSAnXDAn OwohIAlhdHRyX2J1Zi5jdXJzb3IgPSAwOwohIAohIAkvKiAjIG9mIGJ5dGVzIHRoYXQgd2Vy ZSBub3QgeWV0IHByb2Nlc3NlZCBpbiB0aGlzIGxpbmUgKi8KISAJYnl0ZXNfcmVtYWluaW5n ID0gbGluZV9idWYubGVuIC0gbGluZV9idWYuY3Vyc29yOwohIAohIAkvKiBnb3QgdG8gZW5k IG9mIGxpbmUgKi8KISAJaWYgKChlbmRfbG9jID0gc2NhblRleHRMaW5lKHN0YXJ0X2xvYywg ZGVsaW1jLCBieXRlc19yZW1haW5pbmcpKSA9PSBOVUxMKQohIAl7CiEgCQlhdHRyX2xlbiA9 IGJ5dGVzX3JlbWFpbmluZyAtIDE7IC8qIGRvbid0IGNvdW50ICdcbicgaW4gbGVuIGNhbGN1 bGF0aW9uICovCiEgCQlhcHBlbmRCaW5hcnlTdHJpbmdJbmZvKCZhdHRyX2J1Ziwgc3RhcnRf bG9jLCBhdHRyX2xlbik7CiEgCQlsaW5lX2J1Zi5jdXJzb3IgKz0gYXR0cl9sZW4gKyAyOwkJ Lyogc2tpcCAnXG4nIGFuZCAnXDAnICovCiAgCX0KISAJZWxzZQohIAkJLyogZm91bmQgYSBk ZWxpbWl0ZXIgKi8KISAJewohIAkJLyoKISAJCSAqICh3ZSBkb24ndCBjYXJlIGlmIGRlbGlt IHdhcyBwcmVjZWRlZCB3aXRoIGEgYmFja3NsYXNoLCBiZWNhdXNlIGl0J3MKISAJCSAqIGFu IGludmFsaWQgT0lEIGFueXdheSkKISAJCSAqLwogIAohIAkJYXR0cl9sZW4gPSBlbmRfbG9j IC0gc3RhcnRfbG9jOyAvKiB3ZSBkb24ndCBpbmNsdWRlIHRoZSBkZWxpbWl0ZXIgY2ggKi8K ICAKISAJCWFwcGVuZEJpbmFyeVN0cmluZ0luZm8oJmF0dHJfYnVmLCBzdGFydF9sb2MsIGF0 dHJfbGVuKTsKISAJCWxpbmVfYnVmLmN1cnNvciArPSBhdHRyX2xlbiArIDE7CiEgCX0KICAK ICAKISAJLyogY2hlY2sgd2hldGhlciByYXcgaW5wdXQgbWF0Y2hlZCBudWxsIG1hcmtlciAq LwohIAlpZiAoYXR0cl9sZW4gPT0gbnVsbF9wcmludF9sZW4gJiYgc3RybmNtcChzdGFydF9s b2MsIG51bGxfcHJpbnQsIGF0dHJfbGVuKSA9PSAwKQohIAkJKmlzbnVsbCA9IHRydWU7CiEg CWVsc2UKISAJCSppc251bGwgPSBmYWxzZTsKICAKISAJcmV0dXJuIGF0dHJfYnVmLmRhdGE7 CiEgfQogIAogIAogIAogIC8qCiAgICogU2VuZCB0ZXh0IHJlcHJlc2VudGF0aW9uIG9mIG9u ZSBhdHRyaWJ1dGUsIHdpdGggY29udmVyc2lvbiBhbmQgZXNjYXBpbmcKKioqKioqKioqKioq KioqCioqKiAyODAwLDI4MDIgKioqKgotLS0gMzMyOSwzNDMyIC0tLS0KICAKICAJcmV0dXJu IGF0dG51bXM7CiAgfQorIAorIC8qCisgICogVGhlIGZvbGxvd2luZyBhcmUgY3VzdG9tIHZl cnNpb25zIG9mIHRoZSBzdHJpbmcgZnVuY3Rpb24gc3RyY2hyKCkuCisgICogQXMgb3Bwb3Nl ZCB0byB0aGUgb3JpZ2luYWwgc3RyY2hyIHdoaWNoIHNlYXJjaGVzIHRocm91Z2gKKyAgKiBh IHN0cmluZyB1bnRpbCB0aGUgdGFyZ2V0IGNoYXJhY3RlciBpcyBmb3VuZCwgb3IgYSBOVUxM IGlzCisgICogZm91bmQsIHRoaXMgdmVyc2lvbiB3aWxsIG5vdCByZXR1cm4gd2hlbiBhIE5V TEwgaXMgZm91bmQuCisgICogSW5zdGVhZCBpdCB3aWxsIHNlYXJjaCB0aHJvdWdoIGEgcHJl LWRlZmluZWQgbGVuZ3RoIG9mCisgICogYnl0ZXMgYW5kIHdpbGwgcmV0dXJuIG9ubHkgaWYg dGhlIHRhcmdldCBjaGFyYWN0ZXIocykgaXMgcmVhY2hlZC4KKyAgKgorICAqIElmIG91ciBj bGllbnQgZW5jb2RpbmcgaXMgbm90IGEgc3VwcG9ydGVkIHNlcnZlciBlbmNvZGluZywgd2UK KyAgKiBrbm93IHRoYXQgaXQgaXMgbm90IHNhZmUgdG8gbG9vayBhdCBlYWNoIGNoYXJhY3Rl ciBhcyB0cmFpbGluZworICAqIGJ5dGUgaW4gYSBtdWx0aWJ5dGUgY2hhcmFjdGVyIG1heSBi ZSBhIDctYml0IEFTQ0lJIGVxdWl2YWxlbnQuCisgICogVGhlcmVmb3JlIHdlIHVzZSBwZ19l bmNvZGluZ19tYmxlbiB0byBza2lwIHRvIHRoZSBlbmQgb2YgdGhlCisgICogY2hhcmFjdGVy LgorICAqCisgICogcmV0dXJuczoKKyAgKgkgcG9pbnRlciB0byBjIC0gaWYgYyBpcyBsb2Nh dGVkIHdpdGhpbiB0aGUgc3RyaW5nLgorICAqCSBOVUxMIC0gaWYgYyB3YXMgbm90IGZvdW5k IGluIHNwZWNpZmllZCBsZW5ndGggb2Ygc2VhcmNoLiBOb3RlOgorICAqCQkJdGhpcyBET0VT TidUIG1lYW4gdGhhdCBhICdcMCcgd2FzIHJlYWNoZWQuCisgICovCisgY2hhciAqCisgc2Nh blRleHRMaW5lKGNvbnN0IGNoYXIgKnMsIGNoYXIgZW9sLCBzaXplX3QgbGVuKQorIHsKKyAJ Y29uc3QgY2hhciAqc3RhcnQ7CisgCisgCWlmIChjbGllbnRfZW5jb2Rpbmdfb25seSAmJiAh bGluZV9idWZfY29udmVydGVkKQorIAl7CisgCQlpbnQJCQltYmxlbiA9IHBnX2VuY29kaW5n X21ibGVuKGNsaWVudF9lbmNvZGluZywgKHVuc2lnbmVkIGNoYXIqKXMpOworIAorIAkJZm9y IChzdGFydCA9IHM7ICpzICE9IGVvbCAmJiBzIDwgc3RhcnQgKyBsZW47IHMgKz0gbWJsZW4p CisgCQkJbWJsZW4gPSBwZ19lbmNvZGluZ19tYmxlbihjbGllbnRfZW5jb2RpbmcsICh1bnNp Z25lZCBjaGFyKilzKTsKKyAJCQorIAkJcmV0dXJuICgoKnMgPT0gZW9sKSA/IChjaGFyICop IHMgOiBOVUxMKTsKKyAJfQorIAllbHNlCisgCQlyZXR1cm4gbWVtY2hyKHMsZW9sLGxlbik7 CisgfQorIAorIAorIGNoYXIgKgorIHNjYW5DU1ZMaW5lKGNvbnN0IGNoYXIgKnMsIGNoYXIg ZW9sLCBjaGFyIGVzY2FwZWMsIGNoYXIgcXVvdGVjLCBzaXplX3QgbGVuKQorIHsKKyAJY29u c3QgY2hhciAqc3RhcnQ7CisgCQorIAlpZiAoY2xpZW50X2VuY29kaW5nX29ubHkgJiYgIWxp bmVfYnVmX2NvbnZlcnRlZCkKKyAJeworIAkJaW50CQkJbWJsZW4gPSBwZ19lbmNvZGluZ19t YmxlbihjbGllbnRfZW5jb2RpbmcsICh1bnNpZ25lZCBjaGFyKilzKTsKKyAJCQorIAkJZm9y IChzdGFydCA9IHM7ICpzICE9IGVvbCAmJiAgcyA8IHN0YXJ0ICsgbGVuOyBzICs9IG1ibGVu KQorIAkJeworIAkJCWlmIChpbl9xdW90ZSAmJiAqcyA9PSBlc2NhcGVjKQorIAkJCQlsYXN0 X3dhc19lc2MgPSAhbGFzdF93YXNfZXNjOworIAkJCWlmICgqcyA9PSBxdW90ZWMgJiYgIWxh c3Rfd2FzX2VzYykKKyAJCQkJaW5fcXVvdGUgPSAhaW5fcXVvdGU7CisgCQkJaWYgKCpzICE9 IGVzY2FwZWMpCisgCQkJCWxhc3Rfd2FzX2VzYyA9IGZhbHNlOworIAkJCQkJCisgCQkJbWJs ZW4gPSBwZ19lbmNvZGluZ19tYmxlbihjbGllbnRfZW5jb2RpbmcsICh1bnNpZ25lZCBjaGFy KilzKTsKKyAgICAgICAgIH0KKyAJfQorIAllbHNlCisgCQkvKiBzYWZlIHRvIHNjcm9sbCBi eXRlIGJ5IGJ5dGUgKi8KKyAJeworIAkJZm9yIChzdGFydCA9IHM7ICpzICE9IGVvbCAgJiYg cyA8IHN0YXJ0ICsgbGVuOyBzKyspCisgCQl7CisgCQkJaWYgKGluX3F1b3RlICYmICpzID09 IGVzY2FwZWMpCisgCQkJCWxhc3Rfd2FzX2VzYyA9ICFsYXN0X3dhc19lc2M7CisgCQkJaWYg KCpzID09IHF1b3RlYyAmJiAhbGFzdF93YXNfZXNjKQorIAkJCQlpbl9xdW90ZSA9ICFpbl9x dW90ZTsKKyAJCQlpZiAoKnMgIT0gZXNjYXBlYykKKyAJCQkJbGFzdF93YXNfZXNjID0gZmFs c2U7CisgCQl9CisgCQkJCisgCX0KKyAJCisgCWlmKCpzICE9IGVzY2FwZWMpIC8qIGZvdW5k IGVvbF9jaCAqLworIAkJbGFzdF93YXNfZXNjID0gZmFsc2U7CisgCQorIAlyZXR1cm4gKCgq cyA9PSBlb2wpID8gKGNoYXIgKikgcyA6IE5VTEwpOworIH0KKyAKKyAvKgorICAqIFNjYW4g Zm9yIDEgb2YgMiBjaGFyYWN0ZXJzIG5lY2Vzc2FyeSBmb3IgYXR0cmlidXRlIHBhcnNpbmcu IE5vIG5lZWQKKyAgKiBmb3IgY2hlY2tpbmcgZm9yIG11bHRpLWJ5dGUgY2hhcmFjdGVycyBz aW5jZSBjb252ZXJzaW9uIGFscmVhZHkgCisgICogaGFwcGVuZWQuCisgICovCisgY2hhciAq Cisgc2NhblRleHRBdHRyKGNvbnN0IGNoYXIgKnMsIGNoYXIgYzEsIGNoYXIgYzIsIHNpemVf dCBsZW4pCisgeworIAljb25zdCBjaGFyICpzdGFydDsKKyAJCisgCWZvciAoc3RhcnQgPSBz OyAqcyAhPSBjMSAmJiAqcyAhPSBjMiAmJiBzIDwgc3RhcnQgKyBsZW47IHMrKykKKyAJCTsK KyAJCisgCXJldHVybiAoKnMgIT0gJ1wwJyA/IChjaGFyICopIHMgOiBOVUxMKTsKKyB9Cisg CisgCisgCisgCisgCg== --B_3204206540_47940875-- From pgsql-performance-owner@postgresql.org Thu Jul 14 21:28:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D751552C86 for ; Thu, 14 Jul 2005 21:28:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40748-03 for ; Fri, 15 Jul 2005 00:28:18 +0000 (GMT) Received: from mr2.surnet.cl (smtp2.surnet.cl [216.155.73.163]) by svr1.postgresql.org (Postfix) with ESMTP id F2C1852C82 for ; Thu, 14 Jul 2005 21:28:17 -0300 (ADT) Received: from smtp2.surnet.cl (216.155.73.169) by mr2.surnet.cl (7.0.031.3) id 4259ADFA01596C03; Thu, 14 Jul 2005 20:28:00 -0400 Received: from smtp1.surnet.cl (smtp1.surnet.cl []) by mr2.surnet.cl ([216.155.73.169]); Fri, 15 Jul 2005 00:27:59 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp1.surnet.cl (7.0.031.3) id 42587C6E010862FF; Thu, 14 Jul 2005 20:27:39 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60003CD464; Thu, 14 Jul 2005 20:28:16 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 0D53DC370EE; Thu, 14 Jul 2005 20:28:24 -0400 (CLT) Date: Thu, 14 Jul 2005 20:28:24 -0400 From: Alvaro Herrera To: Alison Winters Cc: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table Message-ID: <20050715002824.GA23728@alvh.no-ip.org> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> <20050714162114.GA19778@alvh.no-ip.org> <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.384 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/199 X-Sequence-Number: 13440 On Fri, Jul 15, 2005 at 09:42:12AM +1000, Alison Winters wrote: > > > Our application requires a number of processes to select and update rows > > > from a very small (<10 rows) Postgres table on a regular and frequent > > > basis. These processes often run for weeks at a time, but over the > > > space of a few days we find that updates start getting painfully slow. > > > We are running a full vacuum/analyze and reindex on the table every day, > > Full vacuum, eh? I wonder if what you really need is very frequent > > non-full vacuum. Say, once in 15 minutes (exact rate depending on dead > > tuple rate.) > > > Is there a difference between vacuum and vacuum full? Yes. Vacuum full is more aggresive in compacting the table. Though it really works the same in the presence of long-running transactions: tuples just can't be removed. > The most recent output was this: > > INFO: --Relation public.plc_fldio-- > INFO: Pages 1221: Changed 3, reaped 256, Empty 0, New 0; Tup 108137: Vac 4176, Keep/VTL 108133/108133, UnUsed 19, MinLen 84, MaxLen 84; Re-using: Free/Avail. Space 445176/371836; EndEmpty/Avail. Pages 0/256. > CPU 0.04s/0.14u sec elapsed 0.18 sec. > INFO: Index plcpage_idx: Pages 315; Tuples 108137: Deleted 4176. > CPU 0.03s/0.04u sec elapsed 0.14 sec. > INFO: Rel plc_fldio: Pages: 1221 --> 1221; Tuple(s) moved: 0. > CPU 0.03s/0.04u sec elapsed 0.36 sec. > INFO: Analyzing public.plc_fldio Hmm, so it seems your hourly vacuum is enough. I think the bloat theory can be trashed. Unless I'm reading this output wrong; I don't remember the details of this vacuum output. > We'll up it to every 15 minutes, but i don't know if that'll help > because even with the current vacuuming the updates are still getting > slower and slower over the course of several days. What really puzzles > me is why restarting the processes fixes it. I wonder if the problem may be plan caching. I didn't pay full attention to the description of your problem, so I don't remember if it could be an issue, but it's something to consider. > Does PostgreSQL keep some kind of backlog of transactions all for one > database connection? No. There could be a problem if you had very long transactions, but apparently this isn't your problem. > Isn't it normal to have processes that keep a single database > connection open for days at a time? I guess it depends on exactly what you do with it. I know of at least one case where an app keeps a connection open for months, without a problem. (It's been running for four or five years, and monthly "uptime" for that particular daemon is not unheard of.) -- Alvaro Herrera () "Everybody understands Mickey Mouse. Few understand Hermann Hesse. Hardly anybody understands Einstein. And nobody understands Emperor Norton." From pgsql-performance-owner@postgresql.org Thu Jul 14 21:37:32 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 226F152834 for ; Thu, 14 Jul 2005 21:37:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40007-09 for ; Fri, 15 Jul 2005 00:37:22 +0000 (GMT) Received: from mr2.surnet.cl (smtp2.surnet.cl [216.155.73.163]) by svr1.postgresql.org (Postfix) with ESMTP id CE27B52814 for ; Thu, 14 Jul 2005 21:37:21 -0300 (ADT) Received: from smtp2.surnet.cl (216.155.73.169) by mr2.surnet.cl (7.0.031.3) id 4259ADFA01597261; Thu, 14 Jul 2005 20:37:04 -0400 Received: from smtp1.surnet.cl (smtp1.surnet.cl []) by mr2.surnet.cl ([216.155.73.169]); Fri, 15 Jul 2005 00:37:04 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp1.surnet.cl (7.0.031.3) id 42587C6E010866DA; Thu, 14 Jul 2005 20:36:44 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60003CD832; Thu, 14 Jul 2005 20:37:21 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 14989C370EE; Thu, 14 Jul 2005 20:37:30 -0400 (CLT) Date: Thu, 14 Jul 2005 20:37:30 -0400 From: Alvaro Herrera To: Alison Winters Cc: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table Message-ID: <20050715003729.GB23728@alvh.no-ip.org> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> <20050714162114.GA19778@alvh.no-ip.org> <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> <20050715002824.GA23728@alvh.no-ip.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050715002824.GA23728@alvh.no-ip.org> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.383 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/200 X-Sequence-Number: 13441 On Thu, Jul 14, 2005 at 08:28:24PM -0400, Alvaro Herrera wrote: > On Fri, Jul 15, 2005 at 09:42:12AM +1000, Alison Winters wrote: > > > INFO: Pages 1221: Changed 3, reaped 256, Empty 0, New 0; Tup 108137: Vac 4176, Keep/VTL 108133/108133, UnUsed 19, MinLen 84, MaxLen 84; Re-using: Free/Avail. Space 445176/371836; EndEmpty/Avail. Pages 0/256. > > Hmm, so it seems your hourly vacuum is enough. I think the bloat theory > can be trashed. Unless I'm reading this output wrong; I don't remember > the details of this vacuum output. Ok, so I was _very_ wrong :-) Sorry. -- Alvaro Herrera () Essentially, you're proposing Kevlar shoes as a solution for the problem that you want to walk around carrying a loaded gun aimed at your foot. (Tom Lane) From pgsql-performance-owner@postgresql.org Thu Jul 14 21:43:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CF83552D19 for ; Thu, 14 Jul 2005 21:43:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 39841-08 for ; Fri, 15 Jul 2005 00:43:14 +0000 (GMT) Received: from smtp001.bizmail.yahoo.com (smtp001.bizmail.yahoo.com [216.136.172.125]) by svr1.postgresql.org (Postfix) with SMTP id 37D2C52A58 for ; Thu, 14 Jul 2005 21:43:11 -0300 (ADT) Received: (qmail 1265 invoked from network); 15 Jul 2005 00:43:14 -0000 Received: from unknown (HELO ?192.168.3.41?) (david.mitchell@telogis.com@203.98.10.169 with plain) by smtp001.bizmail.yahoo.com with SMTP; 15 Jul 2005 00:43:14 -0000 Message-ID: <42D7069B.5090107@telogis.com> Date: Fri, 15 Jul 2005 12:43:07 +1200 From: David Mitchell User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dennis Cc: Qingqing Zhou , pgsql-performance@postgresql.org Subject: Re: performance problems ... 100 cpu utilization References: <42D567DB.6000100@works4me.com> <42D5DBEB.4010503@works4me.com> <42D5F315.3090301@telogis.com> <42D60C2E.3020004@works4me.com> In-Reply-To: <42D60C2E.3020004@works4me.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/201 X-Sequence-Number: 13442 If your table has got into this state, then vacuum analyze won't fix it. You will have to do a vacuum full to get it back to normal, then regularly vacuum (not full) to keep it in good condition. We vacuum our critical tables every 10 minutes to keep them in good nick. David Dennis wrote: > David Mitchell wrote: > >> What is the load average on this machine? Do you do many updates? If >> you do a lot of updates, perhaps you haven't vacuumed recently. We >> were seeing similar symptoms when we started load testing our stuff >> and it turned out we were vacuuming too infrequently. > > > The load average at the 100% utilization point was about 30! A vacuum > analyze was done before the test was started. I believe there are many > more selects than updates happening at any one time. > > Dennis > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match > > -- David Mitchell Software Engineer Telogis From pgsql-performance-owner@postgresql.org Thu Jul 14 23:26:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 94DBB52ADF for ; Thu, 14 Jul 2005 23:26:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65237-03 for ; Fri, 15 Jul 2005 02:26:07 +0000 (GMT) Received: from ash25e.internode.on.net (ash25e.internode.on.net [203.16.214.182]) by svr1.postgresql.org (Postfix) with ESMTP id 3DC0652A42 for ; Thu, 14 Jul 2005 23:26:06 -0300 (ADT) Received: from pluto.mirrabooka.com (ppp188-133.static.internode.on.net [150.101.188.133]) by ash25e.internode.on.net (8.12.9/8.12.6) with ESMTP id j6F2Q9k2041662 for ; Fri, 15 Jul 2005 11:56:09 +0930 (CST) (envelope-from alison@mirrabooka.com) Received: by pluto.mirrabooka.com (Postfix, from userid 506) id 646292B11E; Fri, 15 Jul 2005 12:26:09 +1000 (EST) Date: Fri, 15 Jul 2005 12:26:09 +1000 To: pgsql-performance@postgresql.org Subject: Re: lots of updates on small table Message-ID: <42D71EC1.nail8BK11QC2Q@pluto.mirrabooka.com> References: <42D5F34E.nail9QE1195IE@pluto.mirrabooka.com> <20050714162114.GA19778@alvh.no-ip.org> <42D6F854.nail3K81VIICH@pluto.mirrabooka.com> <5893.1121385447@sss.pgh.pa.us> In-Reply-To: <5893.1121385447@sss.pgh.pa.us> User-Agent: nail 11.6 9/7/04 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: alison@mirrabooka.com (Alison Winters) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/202 X-Sequence-Number: 13443 Hi all, > No wonder, considering that your "less than 10 rows" table contains > something upwards of 100000 tuples: > > > INFO: --Relation public.plc_fldio-- > > INFO: Pages 1221: Changed 3, reaped 256, Empty 0, New 0; Tup 108137: Vac 4176, Keep/VTL 108133/108133, UnUsed 19, MinLen 84, MaxLen 84; Re-using: Free/Avail. Space 445176/371836; EndEmpty/Avail. Pages 0/256. > > CPU 0.04s/0.14u sec elapsed 0.18 sec. > > What you need to do is find out why VACUUM is unable to reclaim all > those dead row versions. The reason is likely that some process is > sitting on a open transaction for days at a time. > Cheers mate, that was one of our theories but we weren't sure if it'd be worth rebuilding everything to check. We've been compiling without the -t (autocommit) flag to ecpg, and i believe what's happening is sometimes a transaction is begun and then the processes cycle around doing hardware i/o and never commit or only commit way too late. What we're going to try now is remove all the begins and commits from the code and compile with -t to make sure that any updates happen immediately. Hopefully that'll avoid any hanging transactions. We'll also set up a 10-minutely vacuum (not full) as per some other suggestions here. I'll let you know how it goes - we'll probably slot everything in on Monday so we have a week to follow it. Thanks everyone Alison From pgsql-performance-owner@postgresql.org Fri Jul 15 00:50:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D6F4F52A1E for ; Fri, 15 Jul 2005 00:50:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 89667-03 for ; Fri, 15 Jul 2005 03:50:10 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id CA5D752CC2 for ; Fri, 15 Jul 2005 00:50:07 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 407E564424D for ; Thu, 14 Jul 2005 21:49:43 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 13970-09 for ; Thu, 14 Jul 2005 21:49:41 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 05D15644238 for ; Thu, 14 Jul 2005 21:49:41 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <42D70E42.90403@arbash-meinel.com> References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> <42D70E42.90403@arbash-meinel.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: slow joining very large table to smaller ones Date: Thu, 14 Jul 2005 21:49:50 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/203 X-Sequence-Number: 13444 On Jul 14, 2005, at 7:15 PM, John A Meinel wrote: > > > Is the distribution of your rows uneven? Meaning do you have more rows > with a later id than an earlier one? > There are definitely some id's that will have many times more than the others. If I group and count them, the top 10 are fairly dominant in the table. >> > > Hmm.. How to do it permanantly? Well you could always issue "set > join_collapse set 1; select * from ...." > But obviously that isn't what you prefer. :) > > I think there are things you can do to make merge join more expensive > than a nested loop, but I'm not sure what they are. Maybe someone else has some ideas to encourage this behavior for future work? Setting it on a per-connection basis is doable, but would add some burden to us in code. > > What I really don't understand is that the estimates dropped as well. > The actual number of estimate rows drops to 3k instead of > 1M. > The real question is why does the planner think it will be so > expensive? > > >> select count(*) from k_b join k_r using (incidentid) where k_b.id=107 >> and k_r.id=94; >> count >> ------- >> 373 >> >> > > Well, this says that they are indeed much more selective. > Each one has > 1k rows, but together you end up with only 400. > Is this a bad thing? Is this not "selective enough" to make it much faster? Overall, I'm much happier now after seeing the new plan come about, if I can find a way to make that join_collapse behavior permanent, I can certainly live with these numbers. Thanks again for your continued efforts. -Dan From pgsql-performance-owner@postgresql.org Fri Jul 15 01:12:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 79FBE529AA for ; Fri, 15 Jul 2005 01:12:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95070-09 for ; Fri, 15 Jul 2005 04:12:23 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 69A1D529A3 for ; Fri, 15 Jul 2005 01:12:22 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050715041220m9100k83j1e>; Fri, 15 Jul 2005 04:12:20 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id DE84E55FFC; Thu, 14 Jul 2005 23:12:19 -0500 (CDT) Received: from [192.168.1.12] (Jigglypuff.arbash-meinel.com [192.168.1.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id 7016055FF0; Thu, 14 Jul 2005 23:12:15 -0500 (CDT) Message-ID: <42D7379F.3040806@arbash-meinel.com> Date: Thu, 14 Jul 2005 23:12:15 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> <42D70E42.90403@arbash-meinel.com> In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigFA1CDD21736CEC465830FA59" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.047 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/204 X-Sequence-Number: 13445 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigFA1CDD21736CEC465830FA59 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Dan Harris wrote: > > On Jul 14, 2005, at 7:15 PM, John A Meinel wrote: > >> >> >> Is the distribution of your rows uneven? Meaning do you have more rows >> with a later id than an earlier one? >> > > There are definitely some id's that will have many times more than the > others. If I group and count them, the top 10 are fairly dominant in > the table. That usually skews the estimates. Since the estimate is more of an average (unless the statistics are higher). > >>> >> >> Hmm.. How to do it permanantly? Well you could always issue "set >> join_collapse set 1; select * from ...." >> But obviously that isn't what you prefer. :) >> >> I think there are things you can do to make merge join more expensive >> than a nested loop, but I'm not sure what they are. > > > Maybe someone else has some ideas to encourage this behavior for future > work? Setting it on a per-connection basis is doable, but would add > some burden to us in code. My biggest question is why the planner things the Nested Loop would be so expensive. Have you tuned any of the parameters? It seems like something is out of whack. (cpu_tuple_cost, random_page_cost, etc...) > >> >> What I really don't understand is that the estimates dropped as well. >> The actual number of estimate rows drops to 3k instead of > 1M. >> The real question is why does the planner think it will be so expensive? >> >> >>> select count(*) from k_b join k_r using (incidentid) where k_b.id=107 >>> and k_r.id=94; >>> count >>> ------- >>> 373 >>> >>> >> >> Well, this says that they are indeed much more selective. >> Each one has > 1k rows, but together you end up with only 400. >> > > Is this a bad thing? Is this not "selective enough" to make it much > faster? Yes, being more selective is what makes it faster. But the planner doesn't seem to notice it properly. > > Overall, I'm much happier now after seeing the new plan come about, if > I can find a way to make that join_collapse behavior permanent, I can > certainly live with these numbers. > I'm sure there are pieces to tune, but I've reached my limits of parameters to tweak :) > Thanks again for your continued efforts. > > -Dan > John =:-> --------------enigFA1CDD21736CEC465830FA59 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC1zefJdeBCYSNAAMRAvKrAKDP+f07ZRNOAGveCwnKzT7Tk/oXlQCfZbKM esJNh4xOtqMT334XcB+blCc= =/WcL -----END PGP SIGNATURE----- --------------enigFA1CDD21736CEC465830FA59-- From pgsql-performance-owner@postgresql.org Fri Jul 15 04:01:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A22F952D63 for ; Fri, 15 Jul 2005 04:01:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 61410-05 for ; Fri, 15 Jul 2005 07:01:43 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id CA1DD52D60 for ; Fri, 15 Jul 2005 04:01:43 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id 4B8FA35F952 for ; Fri, 15 Jul 2005 07:01:44 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62470-04 for ; Fri, 15 Jul 2005 07:01:37 +0000 (GMT) Received: from [10.0.0.7] (63-229-107-222.phnx.qwest.net [63.229.107.222]) by veggie-fuel.com (Postfix) with ESMTP id 238E535F94E for ; Fri, 15 Jul 2005 07:01:37 +0000 (GMT) Subject: What's a lot of connections? From: Karim Nassar Reply-To: karim.nassar@acm.org To: pgsql-performance@postgresql.org Content-Type: text/plain Date: Fri, 15 Jul 2005 00:00:58 -0700 Message-Id: <1121410859.4715.87.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/211 X-Sequence-Number: 13452 I am working on a system that uses postgresql 7.4.2 (can't change that until 8.1 goes stable). Just figured out that there are about 285,000 connections created over about 11 hours every day. That averages out to about 7.2 connections per second. Is that a lot? I've never seen that many. -- Karim Nassar From pgsql-performance-owner@postgresql.org Fri Jul 15 04:14:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1883852D18 for ; Fri, 15 Jul 2005 04:14:48 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 63856-08 for ; Fri, 15 Jul 2005 07:14:42 +0000 (GMT) Received: from mail28.sea5.speakeasy.net (mail28.sea5.speakeasy.net [69.17.117.30]) by svr1.postgresql.org (Postfix) with ESMTP id 7AA8352D16 for ; Fri, 15 Jul 2005 04:14:41 -0300 (ADT) Received: (qmail 23607 invoked from network); 15 Jul 2005 07:14:42 -0000 Received: from dsl081-060-184.sfo1.dsl.speakeasy.net (HELO noodles) ([64.81.60.184]) (envelope-sender ) by mail28.sea5.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 15 Jul 2005 07:14:42 -0000 Received: from jwb by noodles with local (Exim 3.36 #1 (Debian)) id 1DtKPC-00021N-00; Fri, 15 Jul 2005 00:14:42 -0700 Subject: Re: What's a lot of connections? From: "Jeffrey W. Baker" To: karim.nassar@acm.org Cc: pgsql-performance@postgresql.org In-Reply-To: <1121410859.4715.87.camel@localhost.localdomain> References: <1121410859.4715.87.camel@localhost.localdomain> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Fri, 15 Jul 2005 00:14:42 -0700 Message-Id: <1121411682.7733.0.camel@noodles> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/212 X-Sequence-Number: 13453 On Fri, 2005-07-15 at 00:00 -0700, Karim Nassar wrote: > I am working on a system that uses postgresql 7.4.2 (can't change that > until 8.1 goes stable). Just figured out that there are about 285,000 > connections created over about 11 hours every day. That averages out to > about 7.2 connections per second. > > Is that a lot? I've never seen that many. I see about 8 million connections per full day. Connecting to postgres is cheap. -jwb From pgsql-performance-owner@postgresql.org Fri Jul 15 05:20:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 45B1E52AB2 for ; Fri, 15 Jul 2005 05:19:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 80418-06 for ; Fri, 15 Jul 2005 08:19:20 +0000 (GMT) Received: from moutng.kundenserver.de (moutng.kundenserver.de [212.227.126.177]) by svr1.postgresql.org (Postfix) with ESMTP id BDFE652BC7 for ; Fri, 15 Jul 2005 05:19:16 -0300 (ADT) Received: from p548F1BEE.dip0.t-ipconnect.de [84.143.27.238] (helo=pse.dyndns.org) by mrelayeu.kundenserver.de with ESMTP (Nemesis), id 0ML21M-1DtLPf078E-0005Zr; Fri, 15 Jul 2005 10:19:15 +0200 Received: from pse1 ([192.168.0.3]) by pse.dyndns.org with esmtp (Exim 4.44) id 1DtLPd-0003rO-I9; Fri, 15 Jul 2005 10:19:13 +0200 Message-ID: <42D7717C.7020208@pse-consulting.de> Date: Fri, 15 Jul 2005 08:19:08 +0000 From: Andreas Pflug User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Bruce Momjian Cc: Agha Asif Raza , pgsql-performance@postgresql.org Subject: Re: Profiler for PostgreSQL References: <200507140627.j6E6RWv11806@candle.pha.pa.us> In-Reply-To: <200507140627.j6E6RWv11806@candle.pha.pa.us> X-Enigmail-Version: 0.89.5.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Provags-ID: kundenserver.de abuse@kundenserver.de login:0ce7ee5c3478b8d72edd8e05ccd40b70 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.314 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/213 X-Sequence-Number: 13454 Bruce Momjian wrote: > Agha Asif Raza wrote: > >>Is there any MS-SQL Server like 'Profiler' available for PostgreSQL? A >>profiler is a tool that monitors the database server and outputs a detailed >>trace of all the transactions/queries that are executed on a database during >>a specified period of time. Kindly let me know if any of you knows of such a >>tool for PostgreSQL. >> Agha Asif Raza > > > Sure see log_statement in postgresql.conf. There are a lot of settings > in there to control what is logged. There's nothing really comparable at the moment, but some tasks can be done with log_statement. I'm planning to implement a full-blown profiling like MSSQL's, but don't expect this too soon (I'm thinking about this for a year now. So many plans, so little time). Regards, Andreas From pgsql-performance-owner@postgresql.org Fri Jul 15 06:17:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 043B952A94 for ; Fri, 15 Jul 2005 06:17:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93100-02 for ; Fri, 15 Jul 2005 09:17:13 +0000 (GMT) Received: from my.endian.it (unknown [62.146.87.34]) by svr1.postgresql.org (Postfix) with ESMTP id BD83752A80 for ; Fri, 15 Jul 2005 06:17:11 -0300 (ADT) Received: from dell.1006.org (host232-14.pool80183.interbusiness.it [80.183.14.232]) (authenticated (0 bits)) by my.endian.it (8.11.6/8.11.6) with ESMTP id j6F985M13202; Fri, 15 Jul 2005 11:08:05 +0200 Subject: Re: PostgresSQL vs. Firebird From: Chris Mair To: Relaxin Cc: pgsql-performance@postgresql.org In-Reply-To: References: Content-Type: text/plain Date: Fri, 15 Jul 2005 11:17:11 +0200 Message-Id: <1121419031.2721.9.camel@dell.1006.org> Mime-Version: 1.0 X-Mailer: Evolution 2.2.2 (2.2.2-5) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/214 X-Sequence-Number: 13455 On Thu, 2005-07-14 at 00:19 -0700, Relaxin wrote: > Before I ask, I don't want to start a war. > > Can someone here give me an honest opinion of how PostgresSQL (PG) is better > than Firebird on Windows? A colleague of mine has made some benchmarks using those two: http://www.1006.org/pg/postgresql_firebird_win_linux.pdf He benchmarked inserts done through *his* own Delphi code varying a few parameters. The servers run on Windows in all tests. The clients were on Windows or Linux. The summary is that PG beats FB performance-wise in all tests except when you do many small transactions (autocommit on) with fsync on. Bye, Chris. From pgsql-performance-owner@postgresql.org Fri Jul 15 11:16:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2E3D352B88 for ; Fri, 15 Jul 2005 11:16:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 55579-04 for ; Fri, 15 Jul 2005 14:16:18 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6170752B13 for ; Fri, 15 Jul 2005 11:16:14 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6FEGH3x011127; Fri, 15 Jul 2005 10:16:17 -0400 (EDT) To: "Jeffrey W. Baker" Cc: karim.nassar@acm.org, pgsql-performance@postgresql.org Subject: Re: What's a lot of connections? In-reply-to: <1121411682.7733.0.camel@noodles> References: <1121410859.4715.87.camel@localhost.localdomain> <1121411682.7733.0.camel@noodles> Comments: In-reply-to "Jeffrey W. Baker" message dated "Fri, 15 Jul 2005 00:14:42 -0700" Date: Fri, 15 Jul 2005 10:16:16 -0400 Message-ID: <11126.1121436976@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/215 X-Sequence-Number: 13456 "Jeffrey W. Baker" writes: > On Fri, 2005-07-15 at 00:00 -0700, Karim Nassar wrote: >> I am working on a system that uses postgresql 7.4.2 (can't change that >> until 8.1 goes stable). Just figured out that there are about 285,000 >> connections created over about 11 hours every day. That averages out to >> about 7.2 connections per second. >> >> Is that a lot? I've never seen that many. > I see about 8 million connections per full day. Connecting to postgres > is cheap. It's not *that* cheap. I think you'd get materially better performance if you managed to pool your connections a bit. By the time a backend has started, initialized itself, joined a database, and populated its internal caches with enough catalog entries to get useful work done, you've got a fair number of cycles invested in it. Dropping the backend after only one or two queries is just not going to be efficient. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Jul 15 12:10:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 543E352B75 for ; Fri, 15 Jul 2005 12:10:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68667-05 for ; Fri, 15 Jul 2005 15:09:52 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 34E0E52B22 for ; Fri, 15 Jul 2005 12:09:50 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id E22A1644140 for ; Fri, 15 Jul 2005 09:09:25 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 20037-01 for ; Fri, 15 Jul 2005 09:09:23 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id BD35364411F for ; Fri, 15 Jul 2005 09:09:23 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <42D7379F.3040806@arbash-meinel.com> References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> <42D70E42.90403@arbash-meinel.com> <42D7379F.3040806@arbash-meinel.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: slow joining very large table to smaller ones Date: Fri, 15 Jul 2005 09:09:37 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/216 X-Sequence-Number: 13457 On Jul 14, 2005, at 10:12 PM, John A Meinel wrote: > > My biggest question is why the planner things the Nested Loop would be > so expensive. > Have you tuned any of the parameters? It seems like something is > out of > whack. (cpu_tuple_cost, random_page_cost, etc...) > here's some of my postgresql.conf. Feel free to blast me if I did something idiotic here. shared_buffers = 50000 effective_cache_size = 1348000 random_page_cost = 3 work_mem = 512000 max_fsm_pages = 80000 log_min_duration_statement = 60000 fsync = true ( not sure if I'm daring enough to run without this ) wal_buffers = 1000 checkpoint_segments = 64 checkpoint_timeout = 3000 #---- FOR PG_AUTOVACUUM --# stats_command_string = true stats_row_level = true From pgsql-performance-owner@postgresql.org Fri Jul 15 12:21:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 871FF528DD for ; Fri, 15 Jul 2005 12:21:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 69883-05 for ; Fri, 15 Jul 2005 15:21:44 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id CDBD552849 for ; Fri, 15 Jul 2005 12:21:43 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 6A33E6441FF for ; Fri, 15 Jul 2005 09:21:19 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 20274-09 for ; Fri, 15 Jul 2005 09:21:16 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 7A8E1644202 for ; Fri, 15 Jul 2005 09:21:16 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> <42D70E42.90403@arbash-meinel.com> <42D7379F.3040806@arbash-meinel.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: slow joining very large table to smaller ones Date: Fri, 15 Jul 2005 09:21:31 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/217 X-Sequence-Number: 13458 On Jul 15, 2005, at 9:09 AM, Dan Harris wrote: > > On Jul 14, 2005, at 10:12 PM, John A Meinel wrote: > >> >> My biggest question is why the planner things the Nested Loop >> would be >> so expensive. >> Have you tuned any of the parameters? It seems like something is >> out of >> whack. (cpu_tuple_cost, random_page_cost, etc...) >> >> > > here's some of my postgresql.conf. Feel free to blast me if I did > something idiotic here. > > shared_buffers = 50000 > effective_cache_size = 1348000 > random_page_cost = 3 > work_mem = 512000 > max_fsm_pages = 80000 > log_min_duration_statement = 60000 > fsync = true ( not sure if I'm daring enough to run without this ) > wal_buffers = 1000 > checkpoint_segments = 64 > checkpoint_timeout = 3000 > > > #---- FOR PG_AUTOVACUUM --# > stats_command_string = true > stats_row_level = true > Sorry, I forgot to re-post my hardware specs. HP DL585 4 x 2.2 GHz Opteron 12GB RAM SmartArray RAID controller, 1GB hardware cache, 4x73GB 10k SCSI in RAID 0+1 ext2 filesystem Also, there are 30 databases on the machine, 27 of them are identical schemas. From pgsql-performance-owner@postgresql.org Fri Jul 15 13:03:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E3F6052913 for ; Fri, 15 Jul 2005 13:03:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79139-08 for ; Fri, 15 Jul 2005 16:03:15 +0000 (GMT) Received: from lakermmtao03.cox.net (lakermmtao03.cox.net [68.230.240.36]) by svr1.postgresql.org (Postfix) with ESMTP id A6FC4528DD for ; Fri, 15 Jul 2005 13:03:11 -0300 (ADT) Received: from [127.0.0.1] (really [68.3.22.74]) by lakermmtao03.cox.net (InterMail vM.6.01.04.00 201-2131-118-20041027) with ESMTP id <20050715160310.LTGV23050.lakermmtao03.cox.net@[127.0.0.1]>; Fri, 15 Jul 2005 12:03:10 -0400 Message-ID: <42D7DE40.3070507@works4me.com> Date: Fri, 15 Jul 2005 09:03:12 -0700 From: Dennis User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: David Mitchell Cc: Qingqing Zhou , pgsql-performance@postgresql.org Subject: Re: performance problems ... 100 cpu utilization References: <42D567DB.6000100@works4me.com> <42D5DBEB.4010503@works4me.com> <42D5F315.3090301@telogis.com> <42D60C2E.3020004@works4me.com> <42D7069B.5090107@telogis.com> In-Reply-To: <42D7069B.5090107@telogis.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Antivirus: avast! (VPS 0528-5, 07/15/2005), Outbound message X-Antivirus-Status: Clean X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/218 X-Sequence-Number: 13459 David Mitchell wrote: > If your table has got into this state, then vacuum analyze won't fix > it. You will have to do a vacuum full to get it back to normal, then > regularly vacuum (not full) to keep it in good condition. We vacuum > our critical tables every 10 minutes to keep them in good nick. So should I have vacuum run during the load test? At what level of updates should it run every ten minutes? Dennis From pgsql-performance-owner@postgresql.org Fri Jul 15 13:44:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 499D552ADF for ; Fri, 15 Jul 2005 13:44:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 12499-02 for ; Fri, 15 Jul 2005 16:44:32 +0000 (GMT) Received: from wolff.to (wolff.to [66.93.197.194]) by svr1.postgresql.org (Postfix) with SMTP id B6B50528DD for ; Fri, 15 Jul 2005 13:44:28 -0300 (ADT) Received: (qmail 783 invoked by uid 500); 15 Jul 2005 16:44:58 -0000 Date: Fri, 15 Jul 2005 11:44:58 -0500 From: Bruno Wolff III To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones Message-ID: <20050715164458.GA26178@wolff.to> Mail-Followup-To: Bruno Wolff III , Dan Harris , pgsql-performance@postgresql.org References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.007 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/219 X-Sequence-Number: 13460 On Thu, Jul 14, 2005 at 16:29:58 -0600, Dan Harris wrote: > > Ok, I tried this one. My ssh keeps getting cut off by a router > somewhere between me and the server due to inactivity timeouts, so > all I know is that both the select and explain analyze are taking > over an hour to run. Here's the explain select for that one, since > that's the best I can get. Are you using NAT at home? That's probably where the issue is. If you have control of that box you can probably increase the timeout to a couple of hours. From pgsql-performance-owner@postgresql.org Fri Jul 15 14:39:45 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2595752BE9 for ; Fri, 15 Jul 2005 14:39:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 73174-05 for ; Fri, 15 Jul 2005 17:39:35 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id E8F3252B13 for ; Fri, 15 Jul 2005 14:39:34 -0300 (ADT) Received: (qmail 25483 invoked from network); 15 Jul 2005 19:40:05 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 15 Jul 2005 19:40:05 +0200 To: "Bruno Wolff III" , "Dan Harris" Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <20050715164458.GA26178@wolff.to> Message-ID: From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Date: Fri, 15 Jul 2005 19:39:30 +0200 In-Reply-To: <20050715164458.GA26178@wolff.to> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/220 X-Sequence-Number: 13461 >> Ok, I tried this one. My ssh keeps getting cut off by a router >> somewhere between me and the server due to inactivity timeouts, so >> all I know is that both the select and explain analyze are taking >> over an hour to run. Here's the explain select for that one, since >> that's the best I can get. one word : screen ! one of the most useful little command line utilities... From pgsql-performance-owner@postgresql.org Fri Jul 15 17:39:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3BFDC52971 for ; Fri, 15 Jul 2005 17:39:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 38860-01 for ; Fri, 15 Jul 2005 20:39:47 +0000 (GMT) Received: from betsy.gw.rwsoft.ca (S01060080c8fe76dd.ed.shawcable.net [68.150.80.101]) by svr1.postgresql.org (Postfix) with ESMTP id AEE0752932 for ; Fri, 15 Jul 2005 17:39:45 -0300 (ADT) Received: from deva.gw.rwsoft.ca (unknown [192.168.34.107]) by betsy.gw.rwsoft.ca (Postfix) with ESMTP id D7C66CFBA for ; Fri, 15 Jul 2005 14:39:45 -0600 (MDT) Date: Fri, 15 Jul 2005 14:39:36 -0600 Message-ID: <8764vbby7r.wl%ron@rwsoft.ca> From: Ron Wills To: pgsql-performance@postgresql.org Subject: Really bad diskio User-Agent: Wanderlust/2.12.2 (99 Luftballons) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.4 (i586-pc-linux-gnu) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.187 tagged_above=0 required=5 tests=FORGED_RCVD_HELO, RCVD_IN_SORBS_DUL X-Spam-Level: X-Archive-Number: 200507/221 X-Sequence-Number: 13462 Hello all I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and an 3Ware SATA raid. Currently the database is only 16G with about 2 tables with 500000+ row, one table 200000+ row and a few small tables. The larger tables get updated about every two hours. The problem I having with this server (which is in production) is the disk IO. On the larger tables I'm getting disk IO wait averages of ~70-90%. I've been tweaking the linux kernel as specified in the PostgreSQL documentations and switched to the deadline scheduler. Nothing seems to be fixing this. The queries are as optimized as I can get them. fsync is off in an attempt to help preformance still nothing. Are there any setting I should be look at the could improve on this??? Thanks for and help in advance. Ron From pgsql-performance-owner@postgresql.org Fri Jul 15 17:45:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6112A5281E for ; Fri, 15 Jul 2005 17:45:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 38609-03 for ; Fri, 15 Jul 2005 20:45:06 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 3128852AF8 for ; Fri, 15 Jul 2005 17:45:06 -0300 (ADT) Received: from [192.168.1.52] (fc1smp [66.93.38.87]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6FKj8hr002927; Fri, 15 Jul 2005 13:45:08 -0700 Message-ID: <42D82053.1090309@commandprompt.com> Date: Fri, 15 Jul 2005 13:45:07 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Ron Wills Cc: pgsql-performance@postgresql.org Subject: Re: Really bad diskio References: <8764vbby7r.wl%ron@rwsoft.ca> In-Reply-To: <8764vbby7r.wl%ron@rwsoft.ca> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.014 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/222 X-Sequence-Number: 13463 Ron Wills wrote: > Hello all > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > an 3Ware SATA raid. 2 drives? 4 drives? 8 drives? RAID 1? 0? 10? 5? Currently the database is only 16G with about 2 > tables with 500000+ row, one table 200000+ row and a few small > tables. The larger tables get updated about every two hours. The > problem I having with this server (which is in production) is the disk > IO. On the larger tables I'm getting disk IO wait averages of > ~70-90%. I've been tweaking the linux kernel as specified in the > PostgreSQL documentations and switched to the deadline > scheduler. Nothing seems to be fixing this. The queries are as > optimized as I can get them. fsync is off in an attempt to help > preformance still nothing. Are there any setting I should be look at > the could improve on this??? > > Thanks for and help in advance. > > Ron > > ---------------------------(end of broadcast)--------------------------- > TIP 1: if posting/reading through Usenet, please send an appropriate > subscribe-nomail command to majordomo@postgresql.org so that your > message can get through to the mailing list cleanly -- Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 PostgreSQL Replication, Consulting, Custom Programming, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ From pgsql-performance-owner@postgresql.org Fri Jul 15 18:01:40 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EDD9C52A76 for ; Fri, 15 Jul 2005 18:01:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42725-02 for ; Fri, 15 Jul 2005 21:01:30 +0000 (GMT) Received: from gghcwest.com (adsl-71-128-90-172.dsl.pltn13.pacbell.net [71.128.90.172]) by svr1.postgresql.org (Postfix) with ESMTP id C5F2452971 for ; Fri, 15 Jul 2005 18:01:28 -0300 (ADT) Received: from toonses.gghcwest.com (toonses.gghcwest.com [192.168.168.115]) by gghcwest.com (8.12.10/8.12.9) with ESMTP id j6FL1Qmd013303; Fri, 15 Jul 2005 14:01:27 -0700 Received: from jwb by toonses.gghcwest.com with local (Exim 4.50) id 1DtXHz-0007dA-VK; Fri, 15 Jul 2005 14:00:07 -0700 Subject: Re: Really bad diskio From: "Jeffrey W. Baker" To: Ron Wills Cc: pgsql-performance@postgresql.org In-Reply-To: <8764vbby7r.wl%ron@rwsoft.ca> References: <8764vbby7r.wl%ron@rwsoft.ca> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Fri, 15 Jul 2005 14:00:07 -0700 Message-Id: <1121461207.28449.8.camel@toonses.gghcwest.com> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/223 X-Sequence-Number: 13464 On Fri, 2005-07-15 at 14:39 -0600, Ron Wills wrote: > Hello all > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > an 3Ware SATA raid. Currently the database is only 16G with about 2 > tables with 500000+ row, one table 200000+ row and a few small > tables. The larger tables get updated about every two hours. The > problem I having with this server (which is in production) is the disk > IO. On the larger tables I'm getting disk IO wait averages of > ~70-90%. I've been tweaking the linux kernel as specified in the > PostgreSQL documentations and switched to the deadline > scheduler. Nothing seems to be fixing this. The queries are as > optimized as I can get them. fsync is off in an attempt to help > preformance still nothing. Are there any setting I should be look at > the could improve on this??? Can you please characterize this a bit better? Send the output of vmstat or iostat over several minutes, or similar diagnostic information. Also please describe your hardware more. Regards, Jeff Baker From pgsql-performance-owner@postgresql.org Fri Jul 15 18:04:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8BCF152B5C for ; Fri, 15 Jul 2005 18:04:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 39580-08 for ; Fri, 15 Jul 2005 21:04:46 +0000 (GMT) Received: from betsy.gw.rwsoft.ca (S01060080c8fe76dd.ed.shawcable.net [68.150.80.101]) by svr1.postgresql.org (Postfix) with ESMTP id EABF952971 for ; Fri, 15 Jul 2005 18:04:44 -0300 (ADT) Received: from deva.gw.rwsoft.ca (unknown [192.168.34.107]) by betsy.gw.rwsoft.ca (Postfix) with ESMTP id 2FF0ECDD7; Fri, 15 Jul 2005 15:04:46 -0600 (MDT) Date: Fri, 15 Jul 2005 15:04:35 -0600 Message-ID: <874qavbx24.wl%ron@rwsoft.ca> From: Ron Wills To: "Joshua D. Drake" Cc: Ron Wills , pgsql-performance@postgresql.org Subject: Re: Really bad diskio In-Reply-To: <42D82053.1090309@commandprompt.com> References: <8764vbby7r.wl%ron@rwsoft.ca> <42D82053.1090309@commandprompt.com> User-Agent: Wanderlust/2.12.2 (99 Luftballons) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.4 (i586-pc-linux-gnu) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_IN_SORBS_DUL X-Spam-Level: X-Archive-Number: 200507/224 X-Sequence-Number: 13465 At Fri, 15 Jul 2005 13:45:07 -0700, Joshua D. Drake wrote: > > Ron Wills wrote: > > Hello all > > > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > > an 3Ware SATA raid. > > 2 drives? > 4 drives? > 8 drives? 3 drives raid 5. I don't believe it's the raid. I've tested this by moving the database to the mirrors software raid where the root is found and onto the the SATA raid. Neither relieved the IO problems. I was also was thinking this could be from the transactional subsystem getting overloaded? There are several automated processes that use the DB. Most are just selects, but the data updates and one that updates the smaller tables that are the heavier queries. On their own they seem to work ok, (still high IO, but fairly quick). But if even the simplest select is called during the heavier operation, then everything goes out through the roof. Maybe there's something I'm missing here as well? > RAID 1? 0? 10? 5? > > > Currently the database is only 16G with about 2 > > tables with 500000+ row, one table 200000+ row and a few small > > tables. The larger tables get updated about every two hours. The > > problem I having with this server (which is in production) is the disk > > IO. On the larger tables I'm getting disk IO wait averages of > > ~70-90%. I've been tweaking the linux kernel as specified in the > > PostgreSQL documentations and switched to the deadline > > scheduler. Nothing seems to be fixing this. The queries are as > > optimized as I can get them. fsync is off in an attempt to help > > preformance still nothing. Are there any setting I should be look at > > the could improve on this??? > > > > Thanks for and help in advance. > > > > Ron > > > > ---------------------------(end of broadcast)--------------------------- > > TIP 1: if posting/reading through Usenet, please send an appropriate > > subscribe-nomail command to majordomo@postgresql.org so that your > > message can get through to the mailing list cleanly > > > -- > Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 > PostgreSQL Replication, Consulting, Custom Programming, 24x7 support > Managed Services, Shared and Dedicated Hosting > Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ From pgsql-performance-owner@postgresql.org Fri Jul 15 18:07:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2320B52815 for ; Fri, 15 Jul 2005 18:07:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42241-03 for ; Fri, 15 Jul 2005 21:07:20 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id D7E1852A76 for ; Fri, 15 Jul 2005 18:07:16 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id B95666441A1 for ; Fri, 15 Jul 2005 15:06:54 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 23968-08 for ; Fri, 15 Jul 2005 15:06:52 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id C46F8644181 for ; Fri, 15 Jul 2005 15:06:52 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <8764vbby7r.wl%ron@rwsoft.ca> References: <8764vbby7r.wl%ron@rwsoft.ca> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <977B093F-E2FC-4A7C-89E5-5715D07873C2@drivefaster.net> Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Really bad diskio Date: Fri, 15 Jul 2005 15:07:08 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/225 X-Sequence-Number: 13466 On Jul 15, 2005, at 2:39 PM, Ron Wills wrote: > Hello all > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > an 3Ware SATA raid. Operating System? Which file system are you using? I was having a similar problem just a few days ago and learned that ext3 was the culprit. -Dan From pgsql-performance-owner@postgresql.org Fri Jul 15 18:10:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CA94152CB3 for ; Fri, 15 Jul 2005 18:10:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42809-04 for ; Fri, 15 Jul 2005 21:10:17 +0000 (GMT) Received: from mr1.surnet.cl (smtp1.surnet.cl [216.155.73.162]) by svr1.postgresql.org (Postfix) with ESMTP id 2728652862 for ; Fri, 15 Jul 2005 18:10:16 -0300 (ADT) Received: from smtp1.surnet.cl (216.155.73.168) by mr1.surnet.cl (7.0.031.3) id 42587EDE015BAB24; Fri, 15 Jul 2005 17:09:39 -0400 Received: from smtp2.surnet.cl (smtp2.surnet.cl []) by mr1.surnet.cl ([216.155.73.168]); Fri, 15 Jul 2005 21:09:39 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp2.surnet.cl (7.0.031.3) id 4259ADE300D53D78; Fri, 15 Jul 2005 17:09:58 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60003F14F7; Fri, 15 Jul 2005 17:10:15 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id B39FCC3711D; Fri, 15 Jul 2005 17:10:26 -0400 (CLT) Date: Fri, 15 Jul 2005 17:10:26 -0400 From: Alvaro Herrera To: Ron Wills Cc: "Joshua D. Drake" , pgsql-performance@postgresql.org Subject: Re: Really bad diskio Message-ID: <20050715211026.GA30540@alvh.no-ip.org> References: <8764vbby7r.wl%ron@rwsoft.ca> <42D82053.1090309@commandprompt.com> <874qavbx24.wl%ron@rwsoft.ca> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <874qavbx24.wl%ron@rwsoft.ca> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.382 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/226 X-Sequence-Number: 13467 On Fri, Jul 15, 2005 at 03:04:35PM -0600, Ron Wills wrote: > > > Ron Wills wrote: > > > Hello all > > > > > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > > > an 3Ware SATA raid. > > > 3 drives raid 5. I don't believe it's the raid. I've tested this by > moving the database to the mirrors software raid where the root is > found and onto the the SATA raid. Neither relieved the IO problems. What filesystem is this? -- Alvaro Herrera () Si no sabes adonde vas, es muy probable que acabes en otra parte. From pgsql-performance-owner@postgresql.org Fri Jul 15 18:18:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C589552CB4 for ; Fri, 15 Jul 2005 18:18:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42126-10 for ; Fri, 15 Jul 2005 21:18:55 +0000 (GMT) Received: from gghcwest.com (adsl-71-128-90-172.dsl.pltn13.pacbell.net [71.128.90.172]) by svr1.postgresql.org (Postfix) with ESMTP id 173AE52CAE for ; Fri, 15 Jul 2005 18:18:54 -0300 (ADT) Received: from toonses.gghcwest.com (toonses.gghcwest.com [192.168.168.115]) by gghcwest.com (8.12.10/8.12.9) with ESMTP id j6FLIrmd026100; Fri, 15 Jul 2005 14:18:53 -0700 Received: from jwb by toonses.gghcwest.com with local (Exim 4.50) id 1DtXYt-0007eo-26; Fri, 15 Jul 2005 14:17:35 -0700 Subject: Re: Really bad diskio From: "Jeffrey W. Baker" To: Ron Wills Cc: pgsql-performance@postgresql.org In-Reply-To: <874qavbx24.wl%ron@rwsoft.ca> References: <8764vbby7r.wl%ron@rwsoft.ca> <42D82053.1090309@commandprompt.com> <874qavbx24.wl%ron@rwsoft.ca> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Fri, 15 Jul 2005 14:17:34 -0700 Message-Id: <1121462254.29379.3.camel@toonses.gghcwest.com> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/227 X-Sequence-Number: 13468 On Fri, 2005-07-15 at 15:04 -0600, Ron Wills wrote: > At Fri, 15 Jul 2005 13:45:07 -0700, > Joshua D. Drake wrote: > > > > Ron Wills wrote: > > > Hello all > > > > > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > > > an 3Ware SATA raid. > > > > 2 drives? > > 4 drives? > > 8 drives? > > 3 drives raid 5. I don't believe it's the raid. I've tested this by > moving the database to the mirrors software raid where the root is > found and onto the the SATA raid. Neither relieved the IO problems. Hard or soft RAID? Which controller? Many of the 3Ware controllers (85xx and 95xx) have extremely bad RAID 5 performance. Did you take any pgbench or other benchmark figures before you started using the DB? -jwb From pgsql-performance-owner@postgresql.org Fri Jul 15 18:29:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F17F352918 for ; Fri, 15 Jul 2005 18:29:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42727-09 for ; Fri, 15 Jul 2005 21:29:15 +0000 (GMT) Received: from betsy.gw.rwsoft.ca (S01060080c8fe76dd.ed.shawcable.net [68.150.80.101]) by svr1.postgresql.org (Postfix) with ESMTP id AB94252B88 for ; Fri, 15 Jul 2005 18:29:14 -0300 (ADT) Received: from deva.gw.rwsoft.ca (unknown [192.168.34.107]) by betsy.gw.rwsoft.ca (Postfix) with ESMTP id 1137ACDD7; Fri, 15 Jul 2005 15:29:16 -0600 (MDT) Date: Fri, 15 Jul 2005 15:29:06 -0600 Message-ID: <873bqfbvx9.wl%ron@rwsoft.ca> From: Ron Wills To: "Jeffrey W. Baker" Cc: Ron Wills , pgsql-performance@postgresql.org Subject: Re: Really bad diskio In-Reply-To: <1121461207.28449.8.camel@toonses.gghcwest.com> References: <8764vbby7r.wl%ron@rwsoft.ca> <1121461207.28449.8.camel@toonses.gghcwest.com> User-Agent: Wanderlust/2.12.2 (99 Luftballons) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.4 (i586-pc-linux-gnu) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_IN_SORBS_DUL X-Spam-Level: X-Archive-Number: 200507/228 X-Sequence-Number: 13469 At Fri, 15 Jul 2005 14:00:07 -0700, Jeffrey W. Baker wrote: > > On Fri, 2005-07-15 at 14:39 -0600, Ron Wills wrote: > > Hello all > > > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > > an 3Ware SATA raid. Currently the database is only 16G with about 2 > > tables with 500000+ row, one table 200000+ row and a few small > > tables. The larger tables get updated about every two hours. The > > problem I having with this server (which is in production) is the disk > > IO. On the larger tables I'm getting disk IO wait averages of > > ~70-90%. I've been tweaking the linux kernel as specified in the > > PostgreSQL documentations and switched to the deadline > > scheduler. Nothing seems to be fixing this. The queries are as > > optimized as I can get them. fsync is off in an attempt to help > > preformance still nothing. Are there any setting I should be look at > > the could improve on this??? > > Can you please characterize this a bit better? Send the output of > vmstat or iostat over several minutes, or similar diagnostic > information. > > Also please describe your hardware more. Here's a bit of a dump of the system that should be useful. Processors x2: vendor_id : AuthenticAMD cpu family : 6 model : 8 model name : AMD Athlon(tm) MP 2400+ stepping : 1 cpu MHz : 2000.474 cache size : 256 KB MemTotal: 903804 kB Mandrake 10.0 Linux kernel 2.6.3-19mdk The raid controller, which is using the hardware raid configuration: 3ware 9000 Storage Controller device driver for Linux v2.26.02.001. scsi0 : 3ware 9000 Storage Controller 3w-9xxx: scsi0: Found a 3ware 9000 Storage Controller at 0xe8020000, IRQ: 17. 3w-9xxx: scsi0: Firmware FE9X 2.02.00.011, BIOS BE9X 2.02.01.037, Ports: 4. Vendor: 3ware Model: Logical Disk 00 Rev: 1.00 Type: Direct-Access ANSI SCSI revision: 00 SCSI device sda: 624955392 512-byte hdwr sectors (319977 MB) SCSI device sda: drive cache: write back, no read (daft) This is also on a 3.6 reiser filesystem. Here's the iostat for 10mins every 10secs. I've removed the stats from the idle drives to reduce the size of this email. Linux 2.6.3-19mdksmp (photo_server) 07/15/2005 avg-cpu: %user %nice %sys %iowait %idle 2.85 1.53 2.15 39.52 53.95 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 82.49 4501.73 188.38 1818836580 76110154 avg-cpu: %user %nice %sys %iowait %idle 0.30 0.00 1.00 96.30 2.40 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 87.80 6159.20 340.00 61592 3400 avg-cpu: %user %nice %sys %iowait %idle 2.50 0.00 1.45 94.35 1.70 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 89.60 5402.40 320.80 54024 3208 avg-cpu: %user %nice %sys %iowait %idle 1.00 0.10 1.35 97.55 0.00 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 105.20 5626.40 332.80 56264 3328 avg-cpu: %user %nice %sys %iowait %idle 0.40 0.00 1.00 87.40 11.20 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 92.61 4484.32 515.48 44888 5160 avg-cpu: %user %nice %sys %iowait %idle 0.45 0.00 1.00 92.66 5.89 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 89.10 4596.00 225.60 45960 2256 avg-cpu: %user %nice %sys %iowait %idle 0.30 0.00 0.80 96.30 2.60 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 86.49 3877.48 414.01 38736 4136 avg-cpu: %user %nice %sys %iowait %idle 0.50 0.00 1.00 98.15 0.35 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 97.10 4710.49 405.19 47152 4056 avg-cpu: %user %nice %sys %iowait %idle 0.35 0.00 1.00 98.65 0.00 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 93.30 5324.80 186.40 53248 1864 avg-cpu: %user %nice %sys %iowait %idle 0.40 0.00 1.10 96.70 1.80 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 117.88 5481.72 402.80 54872 4032 avg-cpu: %user %nice %sys %iowait %idle 0.50 0.00 1.05 98.30 0.15 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 124.00 6081.60 403.20 60816 4032 avg-cpu: %user %nice %sys %iowait %idle 8.75 0.00 2.55 84.46 4.25 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 125.20 5609.60 228.80 56096 2288 avg-cpu: %user %nice %sys %iowait %idle 2.25 0.00 1.30 96.00 0.45 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 176.98 6166.17 686.29 61600 6856 avg-cpu: %user %nice %sys %iowait %idle 5.95 0.00 2.25 88.09 3.70 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 154.55 7879.32 295.70 78872 2960 avg-cpu: %user %nice %sys %iowait %idle 10.29 0.00 3.40 81.97 4.35 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 213.19 11422.18 557.84 114336 5584 avg-cpu: %user %nice %sys %iowait %idle 1.90 0.10 3.25 94.75 0.00 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 227.80 12330.40 212.80 123304 2128 avg-cpu: %user %nice %sys %iowait %idle 0.55 0.00 0.85 96.80 1.80 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 96.30 3464.80 568.80 34648 5688 avg-cpu: %user %nice %sys %iowait %idle 0.70 0.00 1.10 97.25 0.95 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 92.60 4989.60 237.60 49896 2376 avg-cpu: %user %nice %sys %iowait %idle 2.75 0.00 2.10 93.55 1.60 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 198.40 10031.63 458.86 100216 4584 avg-cpu: %user %nice %sys %iowait %idle 0.65 0.00 2.40 95.90 1.05 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 250.25 14174.63 231.77 141888 2320 avg-cpu: %user %nice %sys %iowait %idle 0.60 0.00 2.15 97.20 0.05 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 285.50 12127.20 423.20 121272 4232 avg-cpu: %user %nice %sys %iowait %idle 0.60 0.00 2.90 95.65 0.85 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 393.70 14383.20 534.40 143832 5344 avg-cpu: %user %nice %sys %iowait %idle 0.55 0.00 2.15 96.15 1.15 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 252.15 11801.80 246.15 118136 2464 avg-cpu: %user %nice %sys %iowait %idle 0.75 0.00 3.45 95.15 0.65 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 396.00 19980.80 261.60 199808 2616 avg-cpu: %user %nice %sys %iowait %idle 0.70 0.00 2.70 95.70 0.90 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 286.20 14182.40 467.20 141824 4672 avg-cpu: %user %nice %sys %iowait %idle 0.70 0.00 2.70 95.65 0.95 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 344.20 15838.40 473.60 158384 4736 avg-cpu: %user %nice %sys %iowait %idle 0.75 0.00 1.70 97.50 0.05 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 178.72 7495.70 412.39 75032 4128 avg-cpu: %user %nice %sys %iowait %idle 1.05 0.05 1.30 97.05 0.55 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 107.89 4334.87 249.35 43392 2496 avg-cpu: %user %nice %sys %iowait %idle 0.55 0.00 1.30 98.10 0.05 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 107.01 6345.55 321.12 63392 3208 avg-cpu: %user %nice %sys %iowait %idle 0.65 0.00 1.05 97.55 0.75 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 107.79 3908.89 464.34 39128 4648 avg-cpu: %user %nice %sys %iowait %idle 0.50 0.00 1.15 97.75 0.60 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 109.21 4162.56 434.83 41584 4344 avg-cpu: %user %nice %sys %iowait %idle 0.75 0.00 1.15 98.00 0.10 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 104.19 4796.81 211.58 48064 2120 avg-cpu: %user %nice %sys %iowait %idle 0.70 0.00 1.05 97.85 0.40 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 105.50 4690.40 429.60 46904 4296 avg-cpu: %user %nice %sys %iowait %idle 0.75 0.00 1.10 98.15 0.00 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 107.51 4525.33 357.96 45208 3576 avg-cpu: %user %nice %sys %iowait %idle 2.80 0.00 1.65 92.81 2.75 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 123.18 3810.59 512.29 38144 5128 avg-cpu: %user %nice %sys %iowait %idle 0.60 0.00 1.05 97.10 1.25 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 104.60 3780.00 236.00 37800 2360 avg-cpu: %user %nice %sys %iowait %idle 0.70 0.00 1.10 95.96 2.25 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 117.08 3817.78 466.73 38216 4672 avg-cpu: %user %nice %sys %iowait %idle 0.65 0.00 0.90 96.65 1.80 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 117.20 3629.60 477.60 36296 4776 avg-cpu: %user %nice %sys %iowait %idle 0.80 0.00 1.10 97.50 0.60 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 112.79 4258.94 326.07 42632 3264 avg-cpu: %user %nice %sys %iowait %idle 1.05 0.15 1.20 97.50 0.10 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 125.83 2592.99 522.12 25904 5216 avg-cpu: %user %nice %sys %iowait %idle 0.60 0.00 0.55 98.20 0.65 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 104.90 823.98 305.29 8248 3056 avg-cpu: %user %nice %sys %iowait %idle 0.50 0.00 0.65 98.75 0.10 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 109.80 734.40 468.80 7344 4688 avg-cpu: %user %nice %sys %iowait %idle 1.15 0.00 1.05 97.75 0.05 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 107.70 751.20 463.20 7512 4632 avg-cpu: %user %nice %sys %iowait %idle 6.50 0.00 1.85 90.25 1.40 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 98.00 739.14 277.08 7384 2768 avg-cpu: %user %nice %sys %iowait %idle 0.20 0.00 0.40 82.75 16.65 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 83.13 550.90 360.08 5520 3608 avg-cpu: %user %nice %sys %iowait %idle 2.65 0.30 2.15 82.91 11.99 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 100.00 1136.46 503.50 11376 5040 avg-cpu: %user %nice %sys %iowait %idle 1.00 6.25 2.15 89.70 0.90 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 170.17 4106.51 388.39 41024 3880 avg-cpu: %user %nice %sys %iowait %idle 0.75 0.15 1.75 73.70 23.65 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 234.60 5107.20 232.80 51072 2328 avg-cpu: %user %nice %sys %iowait %idle 0.15 0.00 0.65 49.48 49.73 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 175.52 1431.37 122.28 14328 1224 avg-cpu: %user %nice %sys %iowait %idle 0.15 0.00 0.55 50.22 49.08 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 173.50 1464.00 119.20 14640 1192 avg-cpu: %user %nice %sys %iowait %idle 2.00 0.00 0.60 76.18 21.22 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 130.60 1044.80 203.20 10448 2032 avg-cpu: %user %nice %sys %iowait %idle 0.90 0.10 0.75 97.55 0.70 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 92.09 1024.22 197.80 10232 1976 avg-cpu: %user %nice %sys %iowait %idle 0.25 0.00 0.40 73.78 25.57 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 92.81 582.83 506.99 5840 5080 avg-cpu: %user %nice %sys %iowait %idle 0.20 0.00 0.55 98.85 0.40 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 90.80 657.60 383.20 6576 3832 avg-cpu: %user %nice %sys %iowait %idle 16.46 0.00 4.25 77.09 2.20 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 99.60 1174.83 549.85 11760 5504 avg-cpu: %user %nice %sys %iowait %idle 8.05 0.00 2.60 56.92 32.43 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 172.30 2063.20 128.00 20632 1280 avg-cpu: %user %nice %sys %iowait %idle 20.84 0.00 4.75 52.82 21.59 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 174.30 1416.80 484.00 14168 4840 avg-cpu: %user %nice %sys %iowait %idle 1.30 0.00 1.60 56.93 40.17 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 181.02 2858.74 418.78 28616 4192 avg-cpu: %user %nice %sys %iowait %idle 19.17 0.00 4.44 49.78 26.61 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 162.20 1286.40 373.60 12864 3736 avg-cpu: %user %nice %sys %iowait %idle 0.15 0.00 0.60 50.85 48.40 Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 178.08 1436.64 97.70 14352 976 > Regards, > Jeff Baker From pgsql-performance-owner@postgresql.org Fri Jul 15 18:48:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2B27D52918 for ; Fri, 15 Jul 2005 18:48:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 52890-01 for ; Fri, 15 Jul 2005 21:48:23 +0000 (GMT) Received: from betsy.gw.rwsoft.ca (S01060080c8fe76dd.ed.shawcable.net [68.150.80.101]) by svr1.postgresql.org (Postfix) with ESMTP id 9297A52BC7 for ; Fri, 15 Jul 2005 18:48:21 -0300 (ADT) Received: from deva.gw.rwsoft.ca (unknown [192.168.34.107]) by betsy.gw.rwsoft.ca (Postfix) with ESMTP id AEECC6BE015; Fri, 15 Jul 2005 15:48:22 -0600 (MDT) Date: Fri, 15 Jul 2005 15:48:13 -0600 Message-ID: <871x5zbv1e.wl%ron@rwsoft.ca> From: Ron Wills To: "Jeffrey W. Baker" Cc: Ron Wills , pgsql-performance@postgresql.org Subject: Re: Really bad diskio In-Reply-To: <1121462254.29379.3.camel@toonses.gghcwest.com> References: <8764vbby7r.wl%ron@rwsoft.ca> <42D82053.1090309@commandprompt.com> <874qavbx24.wl%ron@rwsoft.ca> <1121462254.29379.3.camel@toonses.gghcwest.com> User-Agent: Wanderlust/2.12.2 (99 Luftballons) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.4 (i586-pc-linux-gnu) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_IN_SORBS_DUL X-Spam-Level: X-Archive-Number: 200507/229 X-Sequence-Number: 13470 At Fri, 15 Jul 2005 14:17:34 -0700, Jeffrey W. Baker wrote: > > On Fri, 2005-07-15 at 15:04 -0600, Ron Wills wrote: > > At Fri, 15 Jul 2005 13:45:07 -0700, > > Joshua D. Drake wrote: > > > > > > Ron Wills wrote: > > > > Hello all > > > > > > > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > > > > an 3Ware SATA raid. > > > > > > 2 drives? > > > 4 drives? > > > 8 drives? > > > > 3 drives raid 5. I don't believe it's the raid. I've tested this by > > moving the database to the mirrors software raid where the root is > > found and onto the the SATA raid. Neither relieved the IO problems. > > Hard or soft RAID? Which controller? Many of the 3Ware controllers > (85xx and 95xx) have extremely bad RAID 5 performance. > > Did you take any pgbench or other benchmark figures before you started > using the DB? No, unfortunatly, I'm more or less just the developer for the automation systems and admin the system to keep everything going. I have very little say in the hardware used and I don't have any physical access to the machine, it's found a province over :P. But, for what the system, this IO seems unreasonable. I run development on a 1.4Ghz Athlon, Gentoo system, with no raid and I can't reproduce this kind of IO :(. > -jwb From pgsql-performance-owner@postgresql.org Fri Jul 15 18:54:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 88C5A52AED for ; Fri, 15 Jul 2005 18:54:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 52137-06 for ; Fri, 15 Jul 2005 21:54:47 +0000 (GMT) Received: from gghcwest.com (adsl-71-128-90-172.dsl.pltn13.pacbell.net [71.128.90.172]) by svr1.postgresql.org (Postfix) with ESMTP id 457BB52918 for ; Fri, 15 Jul 2005 18:54:46 -0300 (ADT) Received: from toonses.gghcwest.com (toonses.gghcwest.com [192.168.168.115]) by gghcwest.com (8.12.10/8.12.9) with ESMTP id j6FLsjmd017625; Fri, 15 Jul 2005 14:54:46 -0700 Received: from jwb by toonses.gghcwest.com with local (Exim 4.50) id 1DtY7a-0007fl-Qq; Fri, 15 Jul 2005 14:53:26 -0700 Subject: Re: Really bad diskio From: "Jeffrey W. Baker" To: Ron Wills Cc: pgsql-performance@postgresql.org In-Reply-To: <873bqfbvx9.wl%ron@rwsoft.ca> References: <8764vbby7r.wl%ron@rwsoft.ca> <1121461207.28449.8.camel@toonses.gghcwest.com> <873bqfbvx9.wl%ron@rwsoft.ca> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Fri, 15 Jul 2005 14:53:26 -0700 Message-Id: <1121464406.29379.12.camel@toonses.gghcwest.com> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/230 X-Sequence-Number: 13471 On Fri, 2005-07-15 at 15:29 -0600, Ron Wills wrote: > Here's a bit of a dump of the system that should be useful. > > Processors x2: > > vendor_id : AuthenticAMD > cpu family : 6 > model : 8 > model name : AMD Athlon(tm) MP 2400+ > stepping : 1 > cpu MHz : 2000.474 > cache size : 256 KB > > MemTotal: 903804 kB > > Mandrake 10.0 Linux kernel 2.6.3-19mdk > > The raid controller, which is using the hardware raid configuration: > > 3ware 9000 Storage Controller device driver for Linux v2.26.02.001. > scsi0 : 3ware 9000 Storage Controller > 3w-9xxx: scsi0: Found a 3ware 9000 Storage Controller at 0xe8020000, IRQ: 17. > 3w-9xxx: scsi0: Firmware FE9X 2.02.00.011, BIOS BE9X 2.02.01.037, Ports: 4. > Vendor: 3ware Model: Logical Disk 00 Rev: 1.00 > Type: Direct-Access ANSI SCSI revision: 00 > SCSI device sda: 624955392 512-byte hdwr sectors (319977 MB) > SCSI device sda: drive cache: write back, no read (daft) > > This is also on a 3.6 reiser filesystem. > > Here's the iostat for 10mins every 10secs. I've removed the stats from > the idle drives to reduce the size of this email. > > Linux 2.6.3-19mdksmp (photo_server) 07/15/2005 > > avg-cpu: %user %nice %sys %iowait %idle > 2.85 1.53 2.15 39.52 53.95 > > Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn > sda 82.49 4501.73 188.38 1818836580 76110154 > > avg-cpu: %user %nice %sys %iowait %idle > 0.30 0.00 1.00 96.30 2.40 > > Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn > sda 87.80 6159.20 340.00 61592 3400 These I/O numbers are not so horrible, really. 100% iowait is not necessarily a symptom of misconfiguration. It just means you are disk limited. With a database 20 times larger than main memory, this is no surprise. If I had to speculate about the best way to improve your performance, I would say: 1a) Get a better RAID controller. The 3ware hardware RAID5 is very bad. 1b) Get more disks. 2) Get a (much) newer kernel. 3) Try XFS or JFS. Reiser3 has never looked good in my pgbench runs By the way, are you experiencing bad application performance, or are you just unhappy with the iostat figures? Regards, jwb From pgsql-performance-owner@postgresql.org Fri Jul 15 19:11:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6B10652BC7 for ; Fri, 15 Jul 2005 19:11:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 53776-10 for ; Fri, 15 Jul 2005 22:11:48 +0000 (GMT) Received: from betsy.gw.rwsoft.ca (S01060080c8fe76dd.ed.shawcable.net [68.150.80.101]) by svr1.postgresql.org (Postfix) with ESMTP id DB42252918 for ; Fri, 15 Jul 2005 19:11:47 -0300 (ADT) Received: from deva.gw.rwsoft.ca (unknown [192.168.34.107]) by betsy.gw.rwsoft.ca (Postfix) with ESMTP id 017BA5FD7; Fri, 15 Jul 2005 16:11:48 -0600 (MDT) Date: Fri, 15 Jul 2005 16:11:39 -0600 Message-ID: <87zmsnafdw.wl%ron@rwsoft.ca> From: Ron Wills To: "Jeffrey W. Baker" Cc: Ron Wills , pgsql-performance@postgresql.org Subject: Re: Really bad diskio In-Reply-To: <1121464406.29379.12.camel@toonses.gghcwest.com> References: <8764vbby7r.wl%ron@rwsoft.ca> <1121461207.28449.8.camel@toonses.gghcwest.com> <873bqfbvx9.wl%ron@rwsoft.ca> <1121464406.29379.12.camel@toonses.gghcwest.com> User-Agent: Wanderlust/2.12.2 (99 Luftballons) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.4 (i586-pc-linux-gnu) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_IN_SORBS_DUL X-Spam-Level: X-Archive-Number: 200507/231 X-Sequence-Number: 13472 At Fri, 15 Jul 2005 14:53:26 -0700, Jeffrey W. Baker wrote: > > On Fri, 2005-07-15 at 15:29 -0600, Ron Wills wrote: > > Here's a bit of a dump of the system that should be useful. > > > > Processors x2: > > > > vendor_id : AuthenticAMD > > cpu family : 6 > > model : 8 > > model name : AMD Athlon(tm) MP 2400+ > > stepping : 1 > > cpu MHz : 2000.474 > > cache size : 256 KB > > > > MemTotal: 903804 kB > > > > Mandrake 10.0 Linux kernel 2.6.3-19mdk > > > > The raid controller, which is using the hardware raid configuration: > > > > 3ware 9000 Storage Controller device driver for Linux v2.26.02.001. > > scsi0 : 3ware 9000 Storage Controller > > 3w-9xxx: scsi0: Found a 3ware 9000 Storage Controller at 0xe8020000, IRQ: 17. > > 3w-9xxx: scsi0: Firmware FE9X 2.02.00.011, BIOS BE9X 2.02.01.037, Ports: 4. > > Vendor: 3ware Model: Logical Disk 00 Rev: 1.00 > > Type: Direct-Access ANSI SCSI revision: 00 > > SCSI device sda: 624955392 512-byte hdwr sectors (319977 MB) > > SCSI device sda: drive cache: write back, no read (daft) > > > > This is also on a 3.6 reiser filesystem. > > > > Here's the iostat for 10mins every 10secs. I've removed the stats from > > the idle drives to reduce the size of this email. > > > > Linux 2.6.3-19mdksmp (photo_server) 07/15/2005 > > > > avg-cpu: %user %nice %sys %iowait %idle > > 2.85 1.53 2.15 39.52 53.95 > > > > Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn > > sda 82.49 4501.73 188.38 1818836580 76110154 > > > > avg-cpu: %user %nice %sys %iowait %idle > > 0.30 0.00 1.00 96.30 2.40 > > > > Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn > > sda 87.80 6159.20 340.00 61592 3400 > > These I/O numbers are not so horrible, really. 100% iowait is not > necessarily a symptom of misconfiguration. It just means you are disk > limited. With a database 20 times larger than main memory, this is no > surprise. > > If I had to speculate about the best way to improve your performance, I > would say: > > 1a) Get a better RAID controller. The 3ware hardware RAID5 is very bad. > 1b) Get more disks. > 2) Get a (much) newer kernel. > 3) Try XFS or JFS. Reiser3 has never looked good in my pgbench runs Not good news :(. I can't change the hardware, hopefully a kernel update and XFS of JFS will make an improvement. I was hoping for software raid (always has worked well), but the client didn't feel conforable with it :P. > By the way, are you experiencing bad application performance, or are you > just unhappy with the iostat figures? It's affecting the whole system. It is sending the load averages through the roof (from 4 to 12) and processes that would take only a few minutes starts going over an hour, until it clears up. Well, I guess I'll have to drum up some more programming magic... and I'm starting to run out of tricks... I love my job some day :$ > Regards, > jwb > From pgsql-performance-owner@postgresql.org Fri Jul 15 21:06:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 19C8F52C17 for ; Fri, 15 Jul 2005 21:06:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79296-04 for ; Sat, 16 Jul 2005 00:06:44 +0000 (GMT) Received: from ionian.i-clic.uihc.uiowa.edu (ionian.i-clic.uihc.uiowa.edu [129.255.211.20]) by svr1.postgresql.org (Postfix) with ESMTP id BF78E52B06 for ; Fri, 15 Jul 2005 21:06:42 -0300 (ADT) Received: by ionian.i-clic.uihc.uiowa.edu (Postfix, from userid 76) id 0E6B561DB0; Fri, 15 Jul 2005 17:59:16 -0500 (CDT) Received: from [192.168.5.232] (69-212-77-18.ded.ameritech.net [69.212.77.18]) by ionian.i-clic.uihc.uiowa.edu (Postfix) with ESMTP id 6A51B61DAF; Fri, 15 Jul 2005 17:59:04 -0500 (CDT) Message-ID: <42D84F86.8070100@arbash-meinel.com> Date: Fri, 15 Jul 2005 19:06:30 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <42D6F177.3000002@arbash-meinel.com> <42D70E42.90403@arbash-meinel.com> <42D7379F.3040806@arbash-meinel.com> In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig003EECB873ACA41B72F2B48F" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/232 X-Sequence-Number: 13473 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig003EECB873ACA41B72F2B48F Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Dan Harris wrote: > > On Jul 14, 2005, at 10:12 PM, John A Meinel wrote: > >> >> My biggest question is why the planner things the Nested Loop would be >> so expensive. >> Have you tuned any of the parameters? It seems like something is out of >> whack. (cpu_tuple_cost, random_page_cost, etc...) >> > > here's some of my postgresql.conf. Feel free to blast me if I did > something idiotic here. > > shared_buffers = 50000 > effective_cache_size = 1348000 > random_page_cost = 3 > work_mem = 512000 Unless you are the only person connecting to this database, your work_mem is very high. And if you haven't modified maintenance_work_mem it is probably very low. work_mem might be causing postgres to think it can fit all of a merge into ram, making it faster, I can't say for sure. > max_fsm_pages = 80000 This seems high, but it depends how many updates/deletes you get in-between vacuums. It may not be too excessive. VACUUM [FULL] VERBOSE replies with how many free pages are left, if you didn't use that already for tuning. Though it should be tuned based on a steady state situation. Not a one time test. > log_min_duration_statement = 60000 > fsync = true ( not sure if I'm daring enough to run without this ) > wal_buffers = 1000 > checkpoint_segments = 64 > checkpoint_timeout = 3000 > These seem fine to me. Can you include the output of EXPLAIN SELECT both with and without SET join_collapselimit? Since your tables have grown, I can't compare the estimated number of rows, and costs very well. EXPLAIN without ANALYZE is fine, since I am wondering what the planner is thinking things cost. John =:-> > > #---- FOR PG_AUTOVACUUM --# > stats_command_string = true > stats_row_level = true > --------------enig003EECB873ACA41B72F2B48F Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC2E+GJdeBCYSNAAMRAmNwAJ4lc3ufVj1sI25w9rG9zKWZzuF2mgCfRJXx WwfkpkIko3+KDaDEKAIcb50= =eXdN -----END PGP SIGNATURE----- --------------enig003EECB873ACA41B72F2B48F-- From pgsql-performance-owner@postgresql.org Sat Jul 16 05:12:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0075752D97 for ; Sat, 16 Jul 2005 05:12:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04759-06 for ; Sat, 16 Jul 2005 08:12:29 +0000 (GMT) Received: from mail22.sea5.speakeasy.net (mail22.sea5.speakeasy.net [69.17.117.24]) by svr1.postgresql.org (Postfix) with ESMTP id 090E152D96 for ; Sat, 16 Jul 2005 05:12:26 -0300 (ADT) Received: (qmail 1490 invoked from network); 16 Jul 2005 08:12:27 -0000 Received: from dsl081-060-184.sfo1.dsl.speakeasy.net (HELO noodles) ([64.81.60.184]) (envelope-sender ) by mail22.sea5.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 16 Jul 2005 08:12:27 -0000 Received: from jwb by noodles with local (Exim 3.36 #1 (Debian)) id 1Dthmd-0001un-00 for ; Sat, 16 Jul 2005 01:12:27 -0700 Subject: more filesystem benchmarks From: "Jeffrey W. Baker" To: pgsql-performance@postgresql.org Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Sat, 16 Jul 2005 01:12:27 -0700 Message-Id: <1121501547.7101.18.camel@noodles> Mime-Version: 1.0 X-Mailer: Evolution 2.3.4 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/233 X-Sequence-Number: 13474 In our last installment, we saw that JFS provides higher pgbench performance than either XFS or ext3. Using a direct-I/O patch stolen from 8.1, JFS achieved 105 tps with 100 clients. To refresh, the machine in question has 5 7200RPM SATA disks, an Areca RAID controller with 128MB cache, and 1GB of main memory. pgbench is being run with a scale factor of 1000 and 100000 total transactions. At the suggestion of Andreas Dilger of clusterfs, I tried modulating the size of the ext3 journal, and the mount options (data=journal, writeback, and ordered). I turns out that you can achieve a substantial improvement (almost 50%) by simply mounting the ext3 volume with data=writeback instead of data=ordered (the default). Changing the journal size did not seem to make a difference, except that 256MB is for some reason pathological (9% slower than the best time). 128MB, the default for a large volume, gave the same performance as 400MB (the max) or 32MB. In the end, the ext3 volume mounted with -o noatime,data=writeback yielded 88 tps with 100 clients. This is about 16% off the performance of JFS with default options. Andreas pointed me to experimental patches to ext3's block allocation code and writeback strategy. I will test these, but I expect the database community, which seems so attached to its data, will be very interested in code that has not yet entered mainstream use. Another frequent suggestion is to put the xlog on a separate device. I tried this, and, for a given number of disks, it appears to be counter-productive. A RAID5 of 5 disks holding both logs and data is about 15% faster than a RAID5 of 3 disks with the data, and a mirror of two disks holding the xlog. Here are the pgbench results for each permutation of ext3: Journal Size | Journal Mode | 1 Client | 10 Clients | 100 Clients ------------------------------------------------------------------ 32 ordered 28 51 57 32 writeback 34 70 88 64 ordered 29 52 61 64 writeback 32 69 87 128 ordered 32 54 62 128 writeback 34 70 88 256 ordered 28 51 60 256 writeback 29 64 79 400 ordered 26 49 59 400 writeback 32 70 87 -jwb From pgsql-performance-owner@postgresql.org Sat Jul 16 08:01:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1EC8852A8A for ; Sat, 16 Jul 2005 08:01:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 33148-06 for ; Sat, 16 Jul 2005 11:01:11 +0000 (GMT) Received: from vms040pub.verizon.net (vms040pub.verizon.net [206.46.252.40]) by svr1.postgresql.org (Postfix) with ESMTP id 3E4E952A81 for ; Sat, 16 Jul 2005 08:01:07 -0300 (ADT) Received: from osgiliath.mathom.us ([70.108.53.154]) by vms040.mailsrvcs.net (Sun Java System Messaging Server 6.2 HotFix 0.04 (built Dec 24 2004)) with ESMTPA id <0IJP0087AVXXZG04@vms040.mailsrvcs.net> for pgsql-performance@postgresql.org; Sat, 16 Jul 2005 06:01:09 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by osgiliath.mathom.us (Postfix) with ESMTP id 1EB796005E6; Sat, 16 Jul 2005 07:01:09 -0400 (EDT) Received: from osgiliath.mathom.us ([127.0.0.1]) by localhost (osgiliath [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 23919-01-8; Sat, 16 Jul 2005 07:01:09 -0400 (EDT) Received: by osgiliath.mathom.us (Postfix, from userid 1000) id F313B600197; Sat, 16 Jul 2005 07:01:08 -0400 (EDT) Date: Sat, 16 Jul 2005 07:01:08 -0400 From: Michael Stone Subject: Re: more filesystem benchmarks In-reply-to: <1121501547.7101.18.camel@noodles> To: "Jeffrey W. Baker" Cc: pgsql-performance@postgresql.org Mail-Followup-To: "Jeffrey W. Baker" , pgsql-performance@postgresql.org Message-id: <20050716110108.GL19080@mathom.us> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-disposition: inline X-Pgp-Fingerprint: 53 FF 38 00 E7 DD 0A 9C 84 52 84 C5 EE DF 7C 88 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mathom.us References: <1121501547.7101.18.camel@noodles> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/234 X-Sequence-Number: 13475 On Sat, Jul 16, 2005 at 01:12:27AM -0700, Jeffrey W. Baker wrote: >Another frequent suggestion is to put the xlog on a separate device. I >tried this, and, for a given number of disks, it appears to be >counter-productive. A RAID5 of 5 disks holding both logs and data is >about 15% faster than a RAID5 of 3 disks with the data, and a mirror of >two disks holding the xlog. Try simply a seperate partition on the same device with a different filesystem (ext2). Mike Stone From pgsql-performance-owner@postgresql.org Sat Jul 16 18:15:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4E5245299A for ; Sat, 16 Jul 2005 18:15:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 43351-09 for ; Sat, 16 Jul 2005 21:15:03 +0000 (GMT) Received: from betsy.gw.rwsoft.ca (S01060080c8fe76dd.ed.shawcable.net [68.150.80.101]) by svr1.postgresql.org (Postfix) with ESMTP id 9AB14529BB for ; Sat, 16 Jul 2005 18:15:00 -0300 (ADT) Received: from deva.gw.rwsoft.ca (unknown [192.168.34.107]) by betsy.gw.rwsoft.ca (Postfix) with ESMTP id 0FA39CDD7 for ; Sat, 16 Jul 2005 15:15:01 -0600 (MDT) Date: Sat, 16 Jul 2005 15:14:55 -0600 Message-ID: <871x5yh2r4.wl%ron@rwsoft.ca> From: Ron Wills To: pgsql-performance@postgresql.org Subject: Re: Really bad diskio In-Reply-To: <8764vbby7r.wl%ron@rwsoft.ca> References: <8764vbby7r.wl%ron@rwsoft.ca> User-Agent: Wanderlust/2.12.2 (99 Luftballons) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.4 (i586-pc-linux-gnu) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_IN_SORBS_DUL X-Spam-Level: X-Archive-Number: 200507/235 X-Sequence-Number: 13476 At Fri, 15 Jul 2005 14:39:36 -0600, Ron Wills wrote: I just wanted to thank everyone for their help. I believe we found a solution that will help with this problem, with the hardware configuration and caching the larger tables into smaller data sets. A valuable lesson learned from this ;) > Hello all > > I'm running a postgres 7.4.5, on a dual 2.4Ghz Athlon, 1Gig RAM and > an 3Ware SATA raid. Currently the database is only 16G with about 2 > tables with 500000+ row, one table 200000+ row and a few small > tables. The larger tables get updated about every two hours. The > problem I having with this server (which is in production) is the disk > IO. On the larger tables I'm getting disk IO wait averages of > ~70-90%. I've been tweaking the linux kernel as specified in the > PostgreSQL documentations and switched to the deadline > scheduler. Nothing seems to be fixing this. The queries are as > optimized as I can get them. fsync is off in an attempt to help > preformance still nothing. Are there any setting I should be look at > the could improve on this??? > > Thanks for and help in advance. > > Ron > > ---------------------------(end of broadcast)--------------------------- > TIP 1: if posting/reading through Usenet, please send an appropriate > subscribe-nomail command to majordomo@postgresql.org so that your > message can get through to the mailing list cleanly From pgsql-performance-owner@postgresql.org Sat Jul 16 18:42:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7E28852986 for ; Sat, 16 Jul 2005 18:42:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 55412-01 for ; Sat, 16 Jul 2005 21:42:24 +0000 (GMT) Received: from e4.ny.us.ibm.com (e4.ny.us.ibm.com [32.97.182.144]) by svr1.postgresql.org (Postfix) with ESMTP id 231E95280D for ; Sat, 16 Jul 2005 18:42:22 -0300 (ADT) Received: from d01relay02.pok.ibm.com (d01relay02.pok.ibm.com [9.56.227.234]) by e4.ny.us.ibm.com (8.12.11/8.12.11) with ESMTP id j6GLgOCc013790 for ; Sat, 16 Jul 2005 17:42:24 -0400 Received: from d01av03.pok.ibm.com (d01av03.pok.ibm.com [9.56.224.217]) by d01relay02.pok.ibm.com (8.12.10/NCO/VERS6.7) with ESMTP id j6GLgOSw259422 for ; Sat, 16 Jul 2005 17:42:24 -0400 Received: from d01av03.pok.ibm.com (loopback [127.0.0.1]) by d01av03.pok.ibm.com (8.12.11/8.13.3) with ESMTP id j6GLgOhd025396 for ; Sat, 16 Jul 2005 17:42:24 -0400 Received: from d01ml255.pok.ibm.com (d01ml255.pok.ibm.com [9.56.227.128]) by d01av03.pok.ibm.com (8.12.11/8.12.11) with ESMTP id j6GLgOPO025393 for ; Sat, 16 Jul 2005 17:42:24 -0400 Subject: Questions about temporary tables and performance To: pgsql-performance@postgresql.org X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 Message-ID: From: Steven Rosenstein Date: Sat, 16 Jul 2005 17:42:21 -0400 X-MIMETrack: Serialize by Router on D01ML255/01/M/IBM(Release 6.53HF103 | November 15, 2004) at 07/16/2005 17:42:23 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.308 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE X-Spam-Level: X-Archive-Number: 200507/236 X-Sequence-Number: 13477 Postgres Version: 7.3.9 and 8.0.1 (different sites use different versions depending on when they first installed Postgres) Migration Plans: All sites on 8.n within the next 6-9 months. Scenario: A temporary table is created via a "SELECT blah INTO TEMPORARY TABLE blah FROM...". The SELECT query is composed of a number of joins on small (thousands of rows) parameter tables. A view is not usable here because the temporary table SELECT query is constructed on the fly in PHP with JOIN parameters and WHERE filters that may change from main query set to main query set. After the table is created, the key main query JOIN parameter (device ID) is indexed. The resulting temporary table is at most 3000-4000 small (128 byte) records. The temporary table is then joined in a series of SELECT queries to other data tables in the database that contain information associated with the records in the temporary table. These secondary tables can have tens of millions of records each. After the queries are executed, the DB connection is closed and the temporary table and index automatically deleted. Are there any performance issues or considerations associated with using a temporary table in this scenario? Is it worth my trying to develop a solution that just incorporates all the logic used to create the temporary table into each of the main queries? How expensive an operation is temporary table creation and joining? Thanks in advance for your advice, --- Steve From pgsql-performance-owner@postgresql.org Sun Jul 17 01:45:05 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0885752808 for ; Sun, 17 Jul 2005 01:45:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92175-10 for ; Sun, 17 Jul 2005 04:44:54 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 087A25298F for ; Sun, 17 Jul 2005 01:44:53 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6H4irIP025142; Sun, 17 Jul 2005 00:44:54 -0400 (EDT) To: Steven Rosenstein Cc: pgsql-performance@postgresql.org Subject: Re: Questions about temporary tables and performance In-reply-to: References: Comments: In-reply-to Steven Rosenstein message dated "Sat, 16 Jul 2005 17:42:21 -0400" Date: Sun, 17 Jul 2005 00:44:53 -0400 Message-ID: <25141.1121575493@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/237 X-Sequence-Number: 13478 Steven Rosenstein writes: > Are there any performance issues or considerations associated with using a > temporary table in this scenario? It's probably worthwhile to ANALYZE the temp table after it's filled, before you start joining to it. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Jul 17 14:08:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BB18852A01 for ; Sun, 17 Jul 2005 14:08:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 30619-04 for ; Sun, 17 Jul 2005 17:08:36 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.198]) by svr1.postgresql.org (Postfix) with ESMTP id F2759529AE for ; Sun, 17 Jul 2005 14:08:34 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i20so896329wra for ; Sun, 17 Jul 2005 10:08:34 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=npB0MedYWEN3MZti1i8wYb+RpUXj7rV2X3itk7CErwulQ3FcXOWCcapGJer/zfxQgV7v9A/CjjR3xGeLoOrziVzDLxU1r+4vN+kdHSlcDceg3b3EJws3cu/fHikb1UrRBm5cUf82o+ibwlnd62p4YE+lUt7TpaXLi54jdpdFqps= Received: by 10.54.32.52 with SMTP id f52mr126241wrf; Sun, 17 Jul 2005 10:08:34 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Sun, 17 Jul 2005 10:08:34 -0700 (PDT) Message-ID: <59d991c4050717100874c08a86@mail.gmail.com> Date: Sun, 17 Jul 2005 13:08:34 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: pgsql-performance@postgresql.org Subject: Impact of checkpoint_segments under continual load conditions Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.128 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/238 X-Sequence-Number: 13479 I have a unique scenerio. My DB is under "continual load", meaning that I am constantly using COPY to insert new data into the DB. There is no "quiet period" for the database, at least not for hours on end.=20 Normally, checkpoint_segments can help absorb some of that, but my experience is that if I crank the number up, it simply delays the impact, and when it occurs, it takes a VERY long time (minutes) to clear. Thoughts? Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Sun Jul 17 15:16:05 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4780652AEC for ; Sun, 17 Jul 2005 15:16:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40779-09 for ; Sun, 17 Jul 2005 18:15:55 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 33DC852A22 for ; Sun, 17 Jul 2005 15:15:54 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6HIFsa7018943; Sun, 17 Jul 2005 14:15:54 -0400 (EDT) To: Christopher Petrilli Cc: pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c4050717100874c08a86@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Sun, 17 Jul 2005 13:08:34 -0400" Date: Sun, 17 Jul 2005 14:15:54 -0400 Message-ID: <18942.1121624154@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/239 X-Sequence-Number: 13480 Christopher Petrilli writes: > I have a unique scenerio. My DB is under "continual load", meaning > that I am constantly using COPY to insert new data into the DB. There > is no "quiet period" for the database, at least not for hours on end. > Normally, checkpoint_segments can help absorb some of that, but my > experience is that if I crank the number up, it simply delays the > impact, and when it occurs, it takes a VERY long time (minutes) to > clear. If you are using 8.0, you can probably alleviate the problem by making the bgwriter more aggressive. I don't have any immediate recommendations for specific settings though. A small checkpoint_segments setting is definitely bad news for performance under heavy load. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 00:34:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F117D52AC6 for ; Mon, 18 Jul 2005 00:34:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65918-01 for ; Mon, 18 Jul 2005 03:34:43 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id B4AF452AC0 for ; Mon, 18 Jul 2005 00:34:41 -0300 (ADT) Received: (qmail 25274 invoked by uid 1010); 17 Jul 2005 23:34:40 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 4.443716 secs); 18 Jul 2005 03:34:40 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 17 Jul 2005 23:34:36 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id A3DFE311CB; Sun, 17 Jul 2005 21:34:34 -0600 (MDT) Date: Sun, 17 Jul 2005 21:34:16 -0600 From: Robert Creager To: PGPerformance , Andy Hewitt Subject: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? Message-ID: <20050717213416.59dcb52e@thunder.logicalchaos.org> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Sun__17_Jul_2005_21_34_16_-0600__=pJJtTXBIjc/Vrz"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/240 X-Sequence-Number: 13481 --Signature_Sun__17_Jul_2005_21_34_16_-0600__=pJJtTXBIjc/Vrz Content-Type: multipart/mixed; boundary="Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ" --Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Sigh... I recently upgraded from 7.4.1 to 8.0.3. The application did not change. = I'm now running both database concurrently (on different ports, same machine) j= ust so I could verify the problem really exists. The application is a custom test application for testing mechanical systems= .=20 The runs in question (4 at a time) each generate 16 queries at a time of wh= ich the results are sent to the mechanical system which processes the request, = which processes them anywhere from 10 to 120 seconds. The system is capable of completing between 4 and 8 jobs at once. So, once the system is running, at most there will be 8 queries per run simultaneously. The entire database fits into RAM (2Gb), as evidenced by no disk activity a= nd relatively small database size. pg_xlog is on different disks from the db. The problem is that on version 8.0.3, once I get 3 or more concurrent runs going, the query times start tanking (>20 seconds). On 7.4.1, the applicat= ions hum along with queries typically below .2 seconds on over 5 concurrent runs= .=20 Needless to say, 7.4.1 behaves as expected... The only change between runs= is the port connecting to. Bot DB's are up at the same time. For 8.03, pg_autovacuum is running. On 7.4.1, I set up a cron job to vacuum analyze every 5 minutes. The system is Mandrake Linux running 2.4.22 kernel with dual Intel Xenon CPU with HT enabled. On an 803 run, the context switching is up around 60k. On 7.4.1, it maxes around 23k and averages < 1k. I've attached four files. 741 has the query and explain analyze. 803 has = the query and explain analyze during loaded and unloaded times. I've also atta= ched the conf files for the two versions running. I've gone through them and do= n't see any explanation for the problem I'm having. I'm guessing this is the CS problem that reared it's head last year? I had= an e-mail exchange in April of last year about this. Any reason this would be worse with 8.0.3? Thanks, Rob --=20 13:33:43 up 3 days, 17:08, 9 users, load average: 0.16, 0.59, 0.40 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ Content-Type: application/octet-stream; name=741 Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename=741 U0VMRUNUIGxvY2F0aW9uX3R5cGUubmFtZSBBUyBsb2NhdGlvbl90eXBlX25hbWUsCmxpYnJhcnks IHJhaWwsIGNvbCwgc2lkZSwgcm93LCBsb2NhdGlvbl9pZCwKaGxpX2xzbSwgaGxpX3BhbmVsLCBo bGlfcm93LCBobGlfY29sCkZST00gbG9jYXRpb24gSk9JTiBsb2NhdGlvbl90eXBlIFVTSU5HKCBs b2NhdGlvbl90eXBlX2lkICkKSk9JTiBjb21wbGV4IFVTSU5HKCBsaWJyYXJ5X2lkICkKTEVGVCBP VVRFUiBKT0lOIGhsaV9sb2NhdGlvbiBVU0lORyggbG9jYXRpb25faWQgKQpMRUZUIE9VVEVSIEpP SU4gYXBwbGljYXRpb24gVVNJTkcoIGFwcGxpY2F0aW9uX2lkICkKV0hFUkUgIGNvbXBsZXguY29t cGxleF9pZCA9ICc0JwpBTkQgbG9jYXRpb25faWQgTk9UIElOCihTRUxFQ1QgbG9jYXRpb25faWQK RlJPTSBsb2NhdGlvbl9sb2NrKQpBTkQgbG9jYXRpb25faWQgTk9UIElOCihTRUxFQ1QgbG9jYXRp b25faWQKRlJPTSBjYXJ0cmlkZ2UpCkFORCBsb2NhdGlvbi5yYW5kID49IHJhbmRvbSgpCkFORCAo bG9jYXRpb25fdHlwZS5uYW1lIH4gJ2NlbGwnIEFORCBhcHBsaWNhdGlvbi5uYW1lIH4gJ2hsaScg IEFORCBobGlfbHNtID0gMSkKCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgUVVFUlkgUExBTiAgICAgICAgICAgICAg ICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgIAotLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIE5lc3RlZCBMb29wICAoY29zdD00NS4wMC4uOTMu OTEgcm93cz0xIHdpZHRoPTcyKSAoYWN0dWFsIHRpbWU9MTU1LjI1NC4uMjg2LjYxMSByb3dzPTQ1 NyBsb29wcz0xKQogICAtPiAgTmVzdGVkIExvb3AgIChjb3N0PTQ1LjAwLi45MC45NyByb3dzPTEg d2lkdGg9NDQpIChhY3R1YWwgdGltZT0xNTUuMTA4Li4yNzguMzIyIHJvd3M9NDg2IGxvb3BzPTEp CiAgICAgICAgIC0+ICBOZXN0ZWQgTG9vcCAgKGNvc3Q9NDUuMDAuLjg4LjA0IHJvd3M9MSB3aWR0 aD00OCkgKGFjdHVhbCB0aW1lPTQxLjYyMi4uMjU3Ljc5NCByb3dzPTE5NjggbG9vcHM9MSkKICAg ICAgICAgICAgICAgLT4gIE5lc3RlZCBMb29wICAoY29zdD0wLjAwLi40MC4xMCByb3dzPTEgd2lk dGg9MjApIChhY3R1YWwgdGltZT0yLjk5OS4uMTI5LjMwNSByb3dzPTYxMjAgbG9vcHM9MSkKICAg ICAgICAgICAgICAgICAgICAgLT4gIFNlcSBTY2FuIG9uIGhsaV9sb2NhdGlvbiAgKGNvc3Q9MC4w MC4uMjIuNTAgcm93cz02IHdpZHRoPTI0KSAoYWN0dWFsIHRpbWU9Mi44MDkuLjM0Ljg2NCByb3dz PTYxMjAgbG9vcHM9MSkKICAgICAgICAgICAgICAgICAgICAgICAgICAgRmlsdGVyOiAoKGhsaV9s c20pOjppbnRlZ2VyID0gMSkKICAgICAgICAgICAgICAgICAgICAgLT4gIEluZGV4IFNjYW4gdXNp bmcgYXBwbGljYXRpb25fcGtleSBvbiBhcHBsaWNhdGlvbiAgKGNvc3Q9MC4wMC4uMi45MiByb3dz PTEgd2lkdGg9NCkgKGFjdHVhbCB0aW1lPTAuMDA4Li4wLjAxMCByb3dzPTEgbG9vcHM9NjEyMCkK ICAgICAgICAgICAgICAgICAgICAgICAgICAgSW5kZXggQ29uZDogKCJvdXRlciIuYXBwbGljYXRp b25faWQgPSBhcHBsaWNhdGlvbi5hcHBsaWNhdGlvbl9pZCkKICAgICAgICAgICAgICAgICAgICAg ICAgICAgRmlsdGVyOiAoKG5hbWUpOjp0ZXh0IH4gJ2hsaSc6OnRleHQpCiAgICAgICAgICAgICAg IC0+ICBJbmRleCBTY2FuIHVzaW5nIGxvY2F0aW9uX3BrZXkgb24gImxvY2F0aW9uIiAgKGNvc3Q9 NDUuMDAuLjQ3LjkzIHJvd3M9MSB3aWR0aD0zMikgKGFjdHVhbCB0aW1lPTAuMDE3Li4wLjAxNyBy b3dzPTAgbG9vcHM9NjEyMCkKICAgICAgICAgICAgICAgICAgICAgSW5kZXggQ29uZDogKCJsb2Nh dGlvbiIubG9jYXRpb25faWQgPSAib3V0ZXIiLmxvY2F0aW9uX2lkKQogICAgICAgICAgICAgICAg ICAgICBGaWx0ZXI6ICgocmFuZCA+PSByYW5kb20oKSkgQU5EIChOT1QgKGhhc2hlZCBzdWJwbGFu KSkgQU5EIChOT1QgKGhhc2hlZCBzdWJwbGFuKSkpCiAgICAgICAgICAgICAgICAgICAgIFN1YlBs YW4KICAgICAgICAgICAgICAgICAgICAgICAtPiAgU2VxIFNjYW4gb24gY2FydHJpZGdlICAoY29z dD0wLjAwLi4yMC4wMCByb3dzPTEwMDAgd2lkdGg9NCkgKGFjdHVhbCB0aW1lPTAuMDE0Li4xOS4z MDIgcm93cz0xMDI0OCBsb29wcz0xKQogICAgICAgICAgICAgICAgICAgICAgIC0+ICBTZXEgU2Nh biBvbiBsb2NhdGlvbl9sb2NrICAoY29zdD0wLjAwLi4yMC4wMCByb3dzPTEwMDAgd2lkdGg9NCkg KGFjdHVhbCB0aW1lPTAuMTM1Li4wLjk3NiByb3dzPTgwIGxvb3BzPTEpCiAgICAgICAgIC0+ICBJ bmRleCBTY2FuIHVzaW5nIGNvbXBsZXhfbGlicmFyeV9pZF9rZXkgb24gY29tcGxleCAgKGNvc3Q9 MC4wMC4uMi45MiByb3dzPTEgd2lkdGg9NCkgKGFjdHVhbCB0aW1lPTAuMDA3Li4wLjAwNyByb3dz PTAgbG9vcHM9MTk2OCkKICAgICAgICAgICAgICAgSW5kZXggQ29uZDogKCJvdXRlciIubGlicmFy eV9pZCA9IGNvbXBsZXgubGlicmFyeV9pZCkKICAgICAgICAgICAgICAgRmlsdGVyOiAoY29tcGxl eF9pZCA9IDQpCiAgIC0+ICBJbmRleCBTY2FuIHVzaW5nIGxvY2F0aW9uX3R5cGVfcGtleSBvbiBs b2NhdGlvbl90eXBlICAoY29zdD0wLjAwLi4yLjkyIHJvd3M9MSB3aWR0aD0zNikgKGFjdHVhbCB0 aW1lPTAuMDA5Li4wLjAxMSByb3dzPTEgbG9vcHM9NDg2KQogICAgICAgICBJbmRleCBDb25kOiAo Im91dGVyIi5sb2NhdGlvbl90eXBlX2lkID0gbG9jYXRpb25fdHlwZS5sb2NhdGlvbl90eXBlX2lk KQogICAgICAgICBGaWx0ZXI6ICgobmFtZSk6OnRleHQgfiAnY2VsbCc6OnRleHQpCiBUb3RhbCBy dW50aW1lOiAyODcuODA1IG1zCigyMiByb3dzKQoK --Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ Content-Type: application/octet-stream; name=803 Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename=803 ClNFVCBUUkFOU0FDVElPTiBJU09MQVRJT04gTEVWRUwgU0VSSUFMSVpBQkxFIFJFQUQgV1JJVEU7 CgpTRUxFQ1QgbG9jYXRpb25fdHlwZS5uYW1lIEFTIGxvY2F0aW9uX3R5cGVfbmFtZSwKbGlicmFy eSwgcmFpbCwgY29sLCBzaWRlLCByb3csIGxvY2F0aW9uX2lkLApobGlfbHNtLCBobGlfcGFuZWws IGhsaV9yb3csIGhsaV9jb2wKRlJPTSBsb2NhdGlvbiBKT0lOIGxvY2F0aW9uX3R5cGUgVVNJTkco IGxvY2F0aW9uX3R5cGVfaWQgKQpKT0lOIGNvbXBsZXggVVNJTkcoIGxpYnJhcnlfaWQgKQpMRUZU IE9VVEVSIEpPSU4gaGxpX2xvY2F0aW9uIFVTSU5HKCBsb2NhdGlvbl9pZCApCkxFRlQgT1VURVIg Sk9JTiBhcHBsaWNhdGlvbiBVU0lORyggYXBwbGljYXRpb25faWQgKQpXSEVSRSAgY29tcGxleC5j b21wbGV4X2lkID0gJzE3NicKQU5EIGxvY2F0aW9uX2lkIE5PVCBJTgooU0VMRUNUIGxvY2F0aW9u X2lkCkZST00gbG9jYXRpb25fbG9jaykKQU5EIGxvY2F0aW9uX2lkIE5PVCBJTgooU0VMRUNUIGxv Y2F0aW9uX2lkCkZST00gY2FydHJpZGdlKQpBTkQgbG9jYXRpb24ucmFuZCA+PSByYW5kb20oKQpB TkQgKGxvY2F0aW9uX3R5cGUubmFtZSB+ICdjZWxsJyBBTkQgYXBwbGljYXRpb24ubmFtZSB+ICdo bGknICBBTkQgaGxpX2xzbSA9IDEpCgogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgUVVFUlkgUExBTiAgICAgICAgICAg ICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAKLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCiBOZXN0ZWQgTG9vcCAgKGNvc3Q9OTgxLjU5Li4y NTkzLjgxIHJvd3M9MTIgd2lkdGg9NDkpIChhY3R1YWwgdGltZT00MDIxMC4zMDcuLjQwMjEwLjMw NyByb3dzPTAgbG9vcHM9MSkKICAgSm9pbiBGaWx0ZXI6ICgiaW5uZXIiLmFwcGxpY2F0aW9uX2lk ID0gIm91dGVyIi5hcHBsaWNhdGlvbl9pZCkKICAgLT4gIFNlcSBTY2FuIG9uIGFwcGxpY2F0aW9u ICAoY29zdD0wLjAwLi4xLjAxIHJvd3M9MSB3aWR0aD00KSAoYWN0dWFsIHRpbWU9MC4wNTUuLjAu MDU5IHJvd3M9MSBsb29wcz0xKQogICAgICAgICBGaWx0ZXI6ICgobmFtZSk6OnRleHQgfiAnaGxp Jzo6dGV4dCkKICAgLT4gIE5lc3RlZCBMb29wICAoY29zdD05ODEuNTkuLjI1OTIuNjQgcm93cz0x MiB3aWR0aD01MykgKGFjdHVhbCB0aW1lPTQwMjEwLjIzNS4uNDAyMTAuMjM1IHJvd3M9MCBsb29w cz0xKQogICAgICAgICAtPiAgSGFzaCBKb2luICAoY29zdD05ODEuNTkuLjI0MDUuNDkgcm93cz01 OSB3aWR0aD0zMykgKGFjdHVhbCB0aW1lPTI4MjU0LjczNy4uNDAxNjEuODUwIHJvd3M9NTg0IGxv b3BzPTEpCiAgICAgICAgICAgICAgIEhhc2ggQ29uZDogKCJvdXRlciIubG9jYXRpb25fdHlwZV9p ZCA9ICJpbm5lciIubG9jYXRpb25fdHlwZV9pZCkKICAgICAgICAgICAgICAgLT4gIEhhc2ggSm9p biAgKGNvc3Q9OTgwLjUwLi4yNDAxLjc1IHJvd3M9NDEzIHdpZHRoPTI4KSAoYWN0dWFsIHRpbWU9 MjgyNTQuMDIyLi40MDE1MS45OTggcm93cz04MDUgbG9vcHM9MSkKICAgICAgICAgICAgICAgICAg ICAgSGFzaCBDb25kOiAoIm91dGVyIi5saWJyYXJ5X2lkID0gImlubmVyIi5saWJyYXJ5X2lkKQog ICAgICAgICAgICAgICAgICAgICAtPiAgU2VxIFNjYW4gb24gImxvY2F0aW9uIiAgKGNvc3Q9OTc5 LjQxLi4yMzgyLjA3IHJvd3M9Mjg5MSB3aWR0aD0zMikgKGFjdHVhbCB0aW1lPTI4MjM1LjMxMC4u NDAxMDUuNjk5IHJvd3M9MTIyODcgbG9vcHM9MSkKICAgICAgICAgICAgICAgICAgICAgICAgICAg RmlsdGVyOiAoKHJhbmQgPj0gcmFuZG9tKCkpIEFORCAoTk9UIChoYXNoZWQgc3VicGxhbikpIEFO RCAoTk9UIChoYXNoZWQgc3VicGxhbikpKQogICAgICAgICAgICAgICAgICAgICAgICAgICBTdWJQ bGFuCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgLT4gIFNlcSBTY2FuIG9uIGNhcnRyaWRn ZSAgKGNvc3Q9MC4wMC4uNzM0LjM2IHJvd3M9ODUzNiB3aWR0aD00KSAoYWN0dWFsIHRpbWU9MC4w NzcuLjIyMTE5Ljg2MCByb3dzPTkyNjAgbG9vcHM9MSkKICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAtPiAgU2VxIFNjYW4gb24gbG9jYXRpb25fbG9jayAgKGNvc3Q9MC4wMC4uMjIzLjU3IHJv d3M9NTcgd2lkdGg9NCkgKGFjdHVhbCB0aW1lPTAuMDI5Li42MDQyLjUwMiByb3dzPTY4IGxvb3Bz PTEpCiAgICAgICAgICAgICAgICAgICAgIC0+ICBIYXNoICAoY29zdD0xLjA5Li4xLjA5IHJvd3M9 MSB3aWR0aD00KSAoYWN0dWFsIHRpbWU9MC43MDEuLjAuNzAxIHJvd3M9MCBsb29wcz0xKQogICAg ICAgICAgICAgICAgICAgICAgICAgICAtPiAgU2VxIFNjYW4gb24gY29tcGxleCAgKGNvc3Q9MC4w MC4uMS4wOSByb3dzPTEgd2lkdGg9NCkgKGFjdHVhbCB0aW1lPTAuNjQxLi4wLjY1NyByb3dzPTEg bG9vcHM9MSkKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgRmlsdGVyOiAoY29tcGxl eF9pZCA9IDE3NikKICAgICAgICAgICAgICAgLT4gIEhhc2ggIChjb3N0PTEuMDkuLjEuMDkgcm93 cz0xIHdpZHRoPTEzKSAoYWN0dWFsIHRpbWU9MC4yMzYuLjAuMjM2IHJvd3M9MCBsb29wcz0xKQog ICAgICAgICAgICAgICAgICAgICAtPiAgU2VxIFNjYW4gb24gbG9jYXRpb25fdHlwZSAgKGNvc3Q9 MC4wMC4uMS4wOSByb3dzPTEgd2lkdGg9MTMpIChhY3R1YWwgdGltZT0wLjE4NS4uMC4xOTcgcm93 cz0xIGxvb3BzPTEpCiAgICAgICAgICAgICAgICAgICAgICAgICAgIEZpbHRlcjogKChuYW1lKTo6 dGV4dCB+ICdjZWxsJzo6dGV4dCkKICAgICAgICAgLT4gIEluZGV4IFNjYW4gdXNpbmcgaGxpX2xv Y2F0aW9uX2xvY2F0aW9uX2lkIG9uIGhsaV9sb2NhdGlvbiAgKGNvc3Q9MC4wMC4uMy4xNiByb3dz PTEgd2lkdGg9MjQpIChhY3R1YWwgdGltZT0wLjA3MS4uMC4wNzEgcm93cz0wIGxvb3BzPTU4NCkK ICAgICAgICAgICAgICAgSW5kZXggQ29uZDogKCJvdXRlciIubG9jYXRpb25faWQgPSBobGlfbG9j YXRpb24ubG9jYXRpb25faWQpCiAgICAgICAgICAgICAgIEZpbHRlcjogKChobGlfbHNtKTo6aW50 ZWdlciA9IDEpCiBUb3RhbCBydW50aW1lOiA0MDIxMS4wNjUgbXMKKDI0IHJvd3MpCgogICAgICAg ICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgUVVFUlkgUExBTiAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgICAgICAgICAgICAKLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCiBO ZXN0ZWQgTG9vcCAgKGNvc3Q9OTgxLjU5Li4yNTkzLjgxIHJvd3M9MTIgd2lkdGg9NDkpIChhY3R1 YWwgdGltZT0yNDYuNTY0Li4yNDYuNTY0IHJvd3M9MCBsb29wcz0xKQogICBKb2luIEZpbHRlcjog KCJpbm5lciIuYXBwbGljYXRpb25faWQgPSAib3V0ZXIiLmFwcGxpY2F0aW9uX2lkKQogICAtPiAg U2VxIFNjYW4gb24gYXBwbGljYXRpb24gIChjb3N0PTAuMDAuLjEuMDEgcm93cz0xIHdpZHRoPTQp IChhY3R1YWwgdGltZT0wLjAzNC4uMC4wMzcgcm93cz0xIGxvb3BzPTEpCiAgICAgICAgIEZpbHRl cjogKChuYW1lKTo6dGV4dCB+ICdobGknOjp0ZXh0KQogICAtPiAgTmVzdGVkIExvb3AgIChjb3N0 PTk4MS41OS4uMjU5Mi42NCByb3dzPTEyIHdpZHRoPTUzKSAoYWN0dWFsIHRpbWU9MjQ2LjUxOC4u MjQ2LjUxOCByb3dzPTAgbG9vcHM9MSkKICAgICAgICAgLT4gIEhhc2ggSm9pbiAgKGNvc3Q9OTgx LjU5Li4yNDA1LjQ5IHJvd3M9NTkgd2lkdGg9MzMpIChhY3R1YWwgdGltZT05OC4yMjAuLjIzOC40 MjUgcm93cz01NzYgbG9vcHM9MSkKICAgICAgICAgICAgICAgSGFzaCBDb25kOiAoIm91dGVyIi5s b2NhdGlvbl90eXBlX2lkID0gImlubmVyIi5sb2NhdGlvbl90eXBlX2lkKQogICAgICAgICAgICAg ICAtPiAgSGFzaCBKb2luICAoY29zdD05ODAuNTAuLjI0MDEuNzUgcm93cz00MTMgd2lkdGg9Mjgp IChhY3R1YWwgdGltZT05Ny44MzAuLjIzNS41NDEgcm93cz03OTggbG9vcHM9MSkKICAgICAgICAg ICAgICAgICAgICAgSGFzaCBDb25kOiAoIm91dGVyIi5saWJyYXJ5X2lkID0gImlubmVyIi5saWJy YXJ5X2lkKQogICAgICAgICAgICAgICAgICAgICAtPiAgU2VxIFNjYW4gb24gImxvY2F0aW9uIiAg KGNvc3Q9OTc5LjQxLi4yMzgyLjA3IHJvd3M9Mjg5MSB3aWR0aD0zMikgKGFjdHVhbCB0aW1lPTk3 LjE2Ni4uMjIwLjk4NCByb3dzPTEyMzk3IGxvb3BzPTEpCiAgICAgICAgICAgICAgICAgICAgICAg ICAgIEZpbHRlcjogKChyYW5kID49IHJhbmRvbSgpKSBBTkQgKE5PVCAoaGFzaGVkIHN1YnBsYW4p KSBBTkQgKE5PVCAoaGFzaGVkIHN1YnBsYW4pKSkKICAgICAgICAgICAgICAgICAgICAgICAgICAg U3ViUGxhbgogICAgICAgICAgICAgICAgICAgICAgICAgICAgIC0+ICBTZXEgU2NhbiBvbiBjYXJ0 cmlkZ2UgIChjb3N0PTAuMDAuLjczNC4zNiByb3dzPTg1MzYgd2lkdGg9NCkgKGFjdHVhbCB0aW1l PTAuMDEyLi41MC44NTQgcm93cz05MjYwIGxvb3BzPTEpCiAgICAgICAgICAgICAgICAgICAgICAg ICAgICAgLT4gIFNlcSBTY2FuIG9uIGxvY2F0aW9uX2xvY2sgIChjb3N0PTAuMDAuLjIyMy41NyBy b3dzPTU3IHdpZHRoPTQpIChhY3R1YWwgdGltZT0wLjAxNy4uMzAuMjk4IHJvd3M9NjUgbG9vcHM9 MSkKICAgICAgICAgICAgICAgICAgICAgLT4gIEhhc2ggIChjb3N0PTEuMDkuLjEuMDkgcm93cz0x IHdpZHRoPTQpIChhY3R1YWwgdGltZT0wLjAzNy4uMC4wMzcgcm93cz0wIGxvb3BzPTEpCiAgICAg ICAgICAgICAgICAgICAgICAgICAgIC0+ICBTZXEgU2NhbiBvbiBjb21wbGV4ICAoY29zdD0wLjAw Li4xLjA5IHJvd3M9MSB3aWR0aD00KSAoYWN0dWFsIHRpbWU9MC4wMTguLjAuMDIyIHJvd3M9MSBs b29wcz0xKQogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICBGaWx0ZXI6IChjb21wbGV4 X2lkID0gMTc2KQogICAgICAgICAgICAgICAtPiAgSGFzaCAgKGNvc3Q9MS4wOS4uMS4wOSByb3dz PTEgd2lkdGg9MTMpIChhY3R1YWwgdGltZT0wLjA2Mi4uMC4wNjIgcm93cz0wIGxvb3BzPTEpCiAg ICAgICAgICAgICAgICAgICAgIC0+ICBTZXEgU2NhbiBvbiBsb2NhdGlvbl90eXBlICAoY29zdD0w LjAwLi4xLjA5IHJvd3M9MSB3aWR0aD0xMykgKGFjdHVhbCB0aW1lPTAuMDQxLi4wLjA0NyByb3dz PTEgbG9vcHM9MSkKICAgICAgICAgICAgICAgICAgICAgICAgICAgRmlsdGVyOiAoKG5hbWUpOjp0 ZXh0IH4gJ2NlbGwnOjp0ZXh0KQogICAgICAgICAtPiAgSW5kZXggU2NhbiB1c2luZyBobGlfbG9j YXRpb25fbG9jYXRpb25faWQgb24gaGxpX2xvY2F0aW9uICAoY29zdD0wLjAwLi4zLjE2IHJvd3M9 MSB3aWR0aD0yNCkgKGFjdHVhbCB0aW1lPTAuMDExLi4wLjAxMSByb3dzPTAgbG9vcHM9NTc2KQog ICAgICAgICAgICAgICBJbmRleCBDb25kOiAoIm91dGVyIi5sb2NhdGlvbl9pZCA9IGhsaV9sb2Nh dGlvbi5sb2NhdGlvbl9pZCkKICAgICAgICAgICAgICAgRmlsdGVyOiAoKGhsaV9sc20pOjppbnRl Z2VyID0gMSkKIFRvdGFsIHJ1bnRpbWU6IDI0Ni45OTUgbXMKCg== --Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ Content-Type: application/octet-stream; name=postgres741.conf Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename=postgres741.conf IyAtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQojIFBvc3RncmVTUUwgY29uZmlndXJhdGlv biBmaWxlCiMgLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIwojIFRoaXMgZmlsZSBjb25z aXN0cyBvZiBsaW5lcyBvZiB0aGUgZm9ybToKIwojICAgbmFtZSA9IHZhbHVlCiMKIyAoVGhlICc9 JyBpcyBvcHRpb25hbC4pIFdoaXRlIHNwYWNlIG1heSBiZSB1c2VkLiBDb21tZW50cyBhcmUgaW50 cm9kdWNlZAojIHdpdGggJyMnIGFueXdoZXJlIG9uIGEgbGluZS4gVGhlIGNvbXBsZXRlIGxpc3Qg b2Ygb3B0aW9uIG5hbWVzIGFuZAojIGFsbG93ZWQgdmFsdWVzIGNhbiBiZSBmb3VuZCBpbiB0aGUg UG9zdGdyZVNRTCBkb2N1bWVudGF0aW9uLiBUaGUKIyBjb21tZW50ZWQtb3V0IHNldHRpbmdzIHNo b3duIGluIHRoaXMgZmlsZSByZXByZXNlbnQgdGhlIGRlZmF1bHQgdmFsdWVzLgojCiMgQW55IG9w dGlvbiBjYW4gYWxzbyBiZSBnaXZlbiBhcyBhIGNvbW1hbmQgbGluZSBzd2l0Y2ggdG8gdGhlCiMg cG9zdG1hc3RlciwgZS5nLiAncG9zdG1hc3RlciAtYyBsb2dfY29ubmVjdGlvbnM9b24nLiBTb21l IG9wdGlvbnMKIyBjYW4gYmUgY2hhbmdlZCBhdCBydW4tdGltZSB3aXRoIHRoZSAnU0VUJyBTUUwg Y29tbWFuZC4KIwojIFRoaXMgZmlsZSBpcyByZWFkIG9uIHBvc3RtYXN0ZXIgc3RhcnR1cCBhbmQg d2hlbiB0aGUgcG9zdG1hc3RlcgojIHJlY2VpdmVzIGEgU0lHSFVQLiBJZiB5b3UgZWRpdCB0aGUg ZmlsZSBvbiBhIHJ1bm5pbmcgc3lzdGVtLCB5b3UgaGF2ZSAKIyB0byBTSUdIVVAgdGhlIHBvc3Rt YXN0ZXIgZm9yIHRoZSBjaGFuZ2VzIHRvIHRha2UgZWZmZWN0LCBvciB1c2UgCiMgInBnX2N0bCBy ZWxvYWQiLgoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBDT05ORUNUSU9OUyBBTkQgQVVUSEVOVElD QVRJT04KIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQoKIyAtIENvbm5lY3Rpb24gU2V0dGluZ3MgLQoKdGNw aXBfc29ja2V0ID0gdHJ1ZQptYXhfY29ubmVjdGlvbnMgPSAxMDAKCSMgbm90ZTogaW5jcmVhc2lu ZyBtYXhfY29ubmVjdGlvbnMgY29zdHMgYWJvdXQgNTAwIGJ5dGVzIG9mIHNoYXJlZAoJIyBtZW1v cnkgcGVyIGNvbm5lY3Rpb24gc2xvdCwgaW4gYWRkaXRpb24gdG8gY29zdHMgZnJvbSBzaGFyZWRf YnVmZmVycwogICAgICAgICMgYW5kIG1heF9sb2Nrc19wZXJfdHJhbnNhY3Rpb24uCiNzdXBlcnVz ZXJfcmVzZXJ2ZWRfY29ubmVjdGlvbnMgPSAyCnBvcnQgPSA1NDMzCiN1bml4X3NvY2tldF9kaXJl Y3RvcnkgPSAnJwojdW5peF9zb2NrZXRfZ3JvdXAgPSAnJwojdW5peF9zb2NrZXRfcGVybWlzc2lv bnMgPSAwNzc3CSMgb2N0YWwKI3ZpcnR1YWxfaG9zdCA9ICcnCQkjIHdoYXQgaW50ZXJmYWNlIHRv IGxpc3RlbiBvbjsgZGVmYXVsdHMgdG8gYW55CiNyZW5kZXp2b3VzX25hbWUgPSAnJwkJIyBkZWZh dWx0cyB0byB0aGUgY29tcHV0ZXIgbmFtZQoKIyAtIFNlY3VyaXR5ICYgQXV0aGVudGljYXRpb24g LQoKI2F1dGhlbnRpY2F0aW9uX3RpbWVvdXQgPSA2MAkjIDEtNjAwLCBpbiBzZWNvbmRzCiNzc2wg PSBmYWxzZQojcGFzc3dvcmRfZW5jcnlwdGlvbiA9IHRydWUKI2tyYl9zZXJ2ZXJfa2V5ZmlsZSA9 ICcnCiNkYl91c2VyX25hbWVzcGFjZSA9IGZhbHNlCgoKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQojIFJF U09VUkNFIFVTQUdFIChleGNlcHQgV0FMKQojLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCgojIC0gTWVtb3J5 IC0KCnNoYXJlZF9idWZmZXJzID0gMzI3NjgJCSMgbWluIDE2LCBhdCBsZWFzdCBtYXhfY29ubmVj dGlvbnMqMiwgOEtCIGVhY2gKc29ydF9tZW0gPSA4MTkyCQkJIyBtaW4gNjQsIHNpemUgaW4gS0IK dmFjdXVtX21lbSA9IDgxOTIJCSMgbWluIDEwMjQsIHNpemUgaW4gS0IKCiMgLSBGcmVlIFNwYWNl IE1hcCAtCgptYXhfZnNtX3BhZ2VzID0gMjAwMDAwCQkjIG1pbiBtYXhfZnNtX3JlbGF0aW9ucyox NiwgNiBieXRlcyBlYWNoCm1heF9mc21fcmVsYXRpb25zID0gMTAwMDAJIyBtaW4gMTAwLCB+NTAg Ynl0ZXMgZWFjaAoKIyAtIEtlcm5lbCBSZXNvdXJjZSBVc2FnZSAtCgojbWF4X2ZpbGVzX3Blcl9w cm9jZXNzID0gMTAwMAkjIG1pbiAyNQojcHJlbG9hZF9saWJyYXJpZXMgPSAnJwoKCiMtLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0KIyBXUklURSBBSEVBRCBMT0cKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQoKIyAtIFNl dHRpbmdzIC0KCmZzeW5jID0gdHJ1ZQkJCSMgdHVybnMgZm9yY2VkIHN5bmNocm9uaXphdGlvbiBv biBvciBvZmYKd2FsX3N5bmNfbWV0aG9kID0gb3Blbl9zeW5jCSMgdGhlIGRlZmF1bHQgdmFyaWVz IGFjcm9zcyBwbGF0Zm9ybXM6CgkJCQkjIGZzeW5jLCBmZGF0YXN5bmMsIG9wZW5fc3luYywgb3Ig b3Blbl9kYXRhc3luYwp3YWxfYnVmZmVycyA9IDMyCQkjIG1pbiA0LCA4S0IgZWFjaAoKIyAtIENo ZWNrcG9pbnRzIC0KCmNoZWNrcG9pbnRfc2VnbWVudHMgPSAzMgkjIGluIGxvZ2ZpbGUgc2VnbWVu dHMsIG1pbiAxLCAxNk1CIGVhY2gKI2NoZWNrcG9pbnRfdGltZW91dCA9IDMwMAkjIHJhbmdlIDMw LTM2MDAsIGluIHNlY29uZHMKI2NoZWNrcG9pbnRfd2FybmluZyA9IDMwCSMgMCBpcyBvZmYsIGlu IHNlY29uZHMKI2NvbW1pdF9kZWxheSA9IDAJCSMgcmFuZ2UgMC0xMDAwMDAsIGluIG1pY3Jvc2Vj b25kcwojY29tbWl0X3NpYmxpbmdzID0gNQkJIyByYW5nZSAxLTEwMDAKCgojLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tCiMgUVVFUlkgVFVOSU5HCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KCiMgLSBQbGFubmVyIE1l dGhvZCBFbmFibGluZyAtCgojZW5hYmxlX2hhc2hhZ2cgPSB0cnVlCiNlbmFibGVfaGFzaGpvaW4g PSB0cnVlCiNlbmFibGVfaW5kZXhzY2FuID0gdHJ1ZQojZW5hYmxlX21lcmdlam9pbiA9IHRydWUK I2VuYWJsZV9uZXN0bG9vcCA9IHRydWUKI2VuYWJsZV9zZXFzY2FuID0gdHJ1ZQojZW5hYmxlX3Nv cnQgPSB0cnVlCiNlbmFibGVfdGlkc2NhbiA9IHRydWUKCiMgLSBQbGFubmVyIENvc3QgQ29uc3Rh bnRzIC0KCmVmZmVjdGl2ZV9jYWNoZV9zaXplID0gMTgwMDAwCSMgdHlwaWNhbGx5IDhLQiBlYWNo CnJhbmRvbV9wYWdlX2Nvc3QgPSAyCQkjIHVuaXRzIGFyZSBvbmUgc2VxdWVudGlhbCBwYWdlIGZl dGNoIGNvc3QKI2NwdV90dXBsZV9jb3N0ID0gMC4wMQkJIyAoc2FtZSkKI2NwdV9pbmRleF90dXBs ZV9jb3N0ID0gMC4wMDEJIyAoc2FtZSkKI2NwdV9vcGVyYXRvcl9jb3N0ID0gMC4wMDI1CSMgKHNh bWUpCgojIC0gR2VuZXRpYyBRdWVyeSBPcHRpbWl6ZXIgLQoKI2dlcW8gPSB0cnVlCiNnZXFvX3Ro cmVzaG9sZCA9IDExCiNnZXFvX2VmZm9ydCA9IDEKI2dlcW9fZ2VuZXJhdGlvbnMgPSAwCiNnZXFv X3Bvb2xfc2l6ZSA9IDAJCSMgZGVmYXVsdCBiYXNlZCBvbiB0YWJsZXMgaW4gc3RhdGVtZW50LAoJ CQkJIyByYW5nZSAxMjgtMTAyNAojZ2Vxb19zZWxlY3Rpb25fYmlhcyA9IDIuMAkjIHJhbmdlIDEu NS0yLjAKCiMgLSBPdGhlciBQbGFubmVyIE9wdGlvbnMgLQoKZGVmYXVsdF9zdGF0aXN0aWNzX3Rh cmdldCA9IDEwMAkjIHJhbmdlIDEtMTAwMAojZnJvbV9jb2xsYXBzZV9saW1pdCA9IDgKI2pvaW5f Y29sbGFwc2VfbGltaXQgPSA4CSMgMSBkaXNhYmxlcyBjb2xsYXBzaW5nIG9mIGV4cGxpY2l0IEpP SU5zCgoKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQojIEVSUk9SIFJFUE9SVElORyBBTkQgTE9HR0lORwoj LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tCgojIC0gU3lzbG9nIC0KCnN5c2xvZyA9IDIJCQkjIHJhbmdlIDAt MjsgMD1zdGRvdXQ7IDE9Ym90aDsgMj1zeXNsb2cKc3lzbG9nX2ZhY2lsaXR5ID0gJ0xPQ0FMNScK c3lzbG9nX2lkZW50ID0gJ3Bvc3RncmVzJwoKIyAtIFdoZW4gdG8gTG9nIC0KCmNsaWVudF9taW5f bWVzc2FnZXMgPSB3YXJuaW5nCSMgVmFsdWVzLCBpbiBvcmRlciBvZiBkZWNyZWFzaW5nIGRldGFp bDoKCQkJCSMgICBkZWJ1ZzUsIGRlYnVnNCwgZGVidWczLCBkZWJ1ZzIsIGRlYnVnMSwKCQkJCSMg ICBsb2csIGluZm8sIG5vdGljZSwgd2FybmluZywgZXJyb3IKCmxvZ19taW5fbWVzc2FnZXMgPSB3 YXJuaW5nCSMgVmFsdWVzLCBpbiBvcmRlciBvZiBkZWNyZWFzaW5nIGRldGFpbDoKCQkJCSMgICBk ZWJ1ZzUsIGRlYnVnNCwgZGVidWczLCBkZWJ1ZzIsIGRlYnVnMSwKCQkJCSMgICBpbmZvLCBub3Rp Y2UsIHdhcm5pbmcsIGVycm9yLCBsb2csIGZhdGFsLAoJCQkJIyAgIHBhbmljCgojbG9nX2Vycm9y X3ZlcmJvc2l0eSA9IGRlZmF1bHQgICAjIHRlcnNlLCBkZWZhdWx0LCBvciB2ZXJib3NlIG1lc3Nh Z2VzCgpsb2dfbWluX2Vycm9yX3N0YXRlbWVudCA9IHdhcm5pbmcgIyBWYWx1ZXMgaW4gb3JkZXIg b2YgaW5jcmVhc2luZyBzZXZlcml0eToKCQkJCSAjICAgZGVidWc1LCBkZWJ1ZzQsIGRlYnVnMywg ZGVidWcyLCBkZWJ1ZzEsCgkJCQkgIyAgIGluZm8sIG5vdGljZSwgd2FybmluZywgZXJyb3IsIHBh bmljKG9mZikKCQkJCSAKI2xvZ19taW5fZHVyYXRpb25fc3RhdGVtZW50ID0gMCAjIExvZyBhbGwg c3RhdGVtZW50cyB3aG9zZQpsb2dfbWluX2R1cmF0aW9uX3N0YXRlbWVudCA9IDEwMDAwICMgTG9n IGFsbCBzdGF0ZW1lbnRzIHdob3NlCgkJCQkgIyBleGVjdXRpb24gdGltZSBleGNlZWRzIHRoZSB2 YWx1ZSwgaW4KCQkJCSAjIG1pbGxpc2Vjb25kcy4gIFplcm8gcHJpbnRzIGFsbCBxdWVyaWVzLgoJ CQkJICMgTWludXMtb25lIGRpc2FibGVzLgoKI3NpbGVudF9tb2RlID0gZmFsc2UJCSAjIERPIE5P VCBVU0Ugd2l0aG91dCBTeXNsb2chCgojIC0gV2hhdCB0byBMb2cgLQoKI2RlYnVnX3ByaW50X3Bh cnNlID0gZmFsc2UKI2RlYnVnX3ByaW50X3Jld3JpdHRlbiA9IGZhbHNlCiNkZWJ1Z19wcmludF9w bGFuID0gZmFsc2UKZGVidWdfcHJldHR5X3ByaW50ID0gdHJ1ZQojbG9nX2Nvbm5lY3Rpb25zID0g dHJ1ZQojbG9nX2R1cmF0aW9uID0gZmFsc2UKbG9nX3BpZCA9IHRydWUKbG9nX3N0YXRlbWVudCA9 IGZhbHNlCmxvZ190aW1lc3RhbXAgPSB0cnVlCmxvZ19ob3N0bmFtZSA9IHRydWUKI2xvZ19zb3Vy Y2VfcG9ydCA9IGZhbHNlCgoKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQojIFJVTlRJTUUgU1RBVElTVElD UwojLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tCgojIC0gU3RhdGlzdGljcyBNb25pdG9yaW5nIC0KCiNsb2df cGFyc2VyX3N0YXRzID0gZmFsc2UKI2xvZ19wbGFubmVyX3N0YXRzID0gZmFsc2UKI2xvZ19leGVj dXRvcl9zdGF0cyA9IGZhbHNlCiNsb2dfc3RhdGVtZW50X3N0YXRzID0gZmFsc2UKCiMgLSBRdWVy eS9JbmRleCBTdGF0aXN0aWNzIENvbGxlY3RvciAtCgpzdGF0c19zdGFydF9jb2xsZWN0b3IgPSB0 cnVlCiNzdGF0c19jb21tYW5kX3N0cmluZyA9IGZhbHNlCiNzdGF0c19ibG9ja19sZXZlbCA9IGZh bHNlCnN0YXRzX3Jvd19sZXZlbCA9IHRydWUKc3RhdHNfcmVzZXRfb25fc2VydmVyX3N0YXJ0ID0g dHJ1ZQoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBDTElFTlQgQ09OTkVDVElPTiBERUZBVUxUUwoj LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tCgojIC0gU3RhdGVtZW50IEJlaGF2aW9yIC0KCiNzZWFyY2hfcGF0 aCA9ICckdXNlcixwdWJsaWMnCSMgc2NoZW1hIG5hbWVzCiNjaGVja19mdW5jdGlvbl9ib2RpZXMg PSB0cnVlCiNkZWZhdWx0X3RyYW5zYWN0aW9uX2lzb2xhdGlvbiA9ICdyZWFkIGNvbW1pdHRlZCcK I2RlZmF1bHRfdHJhbnNhY3Rpb25fcmVhZF9vbmx5ID0gZmFsc2UKI3N0YXRlbWVudF90aW1lb3V0 ID0gMAkJIyAwIGlzIGRpc2FibGVkLCBpbiBtaWxsaXNlY29uZHMKCiMgLSBMb2NhbGUgYW5kIEZv cm1hdHRpbmcgLQoKI2RhdGVzdHlsZSA9ICdpc28sIG1keScKI3RpbWV6b25lID0gdW5rbm93bgkJ IyBhY3R1YWxseSwgZGVmYXVsdHMgdG8gVFogZW52aXJvbm1lbnQgc2V0dGluZwojYXVzdHJhbGlh bl90aW1lem9uZXMgPSBmYWxzZQojZXh0cmFfZmxvYXRfZGlnaXRzID0gMAkJIyBtaW4gLTE1LCBt YXggMgojY2xpZW50X2VuY29kaW5nID0gc3FsX2FzY2lpCSMgYWN0dWFsbHksIGRlZmF1bHRzIHRv IGRhdGFiYXNlIGVuY29kaW5nCgojIFRoZXNlIHNldHRpbmdzIGFyZSBpbml0aWFsaXplZCBieSBp bml0ZGIgLS0gdGhleSBtYXkgYmUgY2hhbmdlZApsY19tZXNzYWdlcyA9ICdDJwkJIyBsb2NhbGUg Zm9yIHN5c3RlbSBlcnJvciBtZXNzYWdlIHN0cmluZ3MKbGNfbW9uZXRhcnkgPSAnQycJCSMgbG9j YWxlIGZvciBtb25ldGFyeSBmb3JtYXR0aW5nCmxjX251bWVyaWMgPSAnQycJCSMgbG9jYWxlIGZv ciBudW1iZXIgZm9ybWF0dGluZwpsY190aW1lID0gJ0MnCQkJIyBsb2NhbGUgZm9yIHRpbWUgZm9y bWF0dGluZwoKIyAtIE90aGVyIERlZmF1bHRzIC0KCiNleHBsYWluX3ByZXR0eV9wcmludCA9IHRy dWUKI2R5bmFtaWNfbGlicmFyeV9wYXRoID0gJyRsaWJkaXInCiNtYXhfZXhwcl9kZXB0aCA9IDEw MDAwCQkjIG1pbiAxMAoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBMT0NLIE1BTkFHRU1FTlQKIy0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLQoKI2RlYWRsb2NrX3RpbWVvdXQgPSAxMDAwCSMgaW4gbWlsbGlzZWNv bmRzCiNtYXhfbG9ja3NfcGVyX3RyYW5zYWN0aW9uID0gNjQJIyBtaW4gMTAsIH4yNjAqbWF4X2Nv bm5lY3Rpb25zIGJ5dGVzIGVhY2gKCgojLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCiMgVkVSU0lPTi9QTEFU Rk9STSBDT01QQVRJQklMSVRZCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KCiMgLSBQcmV2aW91cyBQb3N0 Z3JlcyBWZXJzaW9ucyAtCgphZGRfbWlzc2luZ19mcm9tID0gZmFsc2UKI3JlZ2V4X2ZsYXZvciA9 IGFkdmFuY2VkCSMgYWR2YW5jZWQsIGV4dGVuZGVkLCBvciBiYXNpYwpzcWxfaW5oZXJpdGFuY2Ug PSBmYWxzZQoKIyAtIE90aGVyIFBsYXRmb3JtcyAmIENsaWVudHMgLQoKI3RyYW5zZm9ybV9udWxs X2VxdWFscyA9IGZhbHNlCg== --Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ Content-Type: application/octet-stream; name=postgres803.conf Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename=postgres803.conf IyAtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQojIFBvc3RncmVTUUwgY29uZmlndXJhdGlv biBmaWxlCiMgLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIwojIFRoaXMgZmlsZSBjb25z aXN0cyBvZiBsaW5lcyBvZiB0aGUgZm9ybToKIwojICAgbmFtZSA9IHZhbHVlCiMKIyAoVGhlICc9 JyBpcyBvcHRpb25hbC4pIFdoaXRlIHNwYWNlIG1heSBiZSB1c2VkLiBDb21tZW50cyBhcmUgaW50 cm9kdWNlZAojIHdpdGggJyMnIGFueXdoZXJlIG9uIGEgbGluZS4gVGhlIGNvbXBsZXRlIGxpc3Qg b2Ygb3B0aW9uIG5hbWVzIGFuZAojIGFsbG93ZWQgdmFsdWVzIGNhbiBiZSBmb3VuZCBpbiB0aGUg UG9zdGdyZVNRTCBkb2N1bWVudGF0aW9uLiBUaGUKIyBjb21tZW50ZWQtb3V0IHNldHRpbmdzIHNo b3duIGluIHRoaXMgZmlsZSByZXByZXNlbnQgdGhlIGRlZmF1bHQgdmFsdWVzLgojCiMgUGxlYXNl IG5vdGUgdGhhdCByZS1jb21tZW50aW5nIGEgc2V0dGluZyBpcyBOT1Qgc3VmZmljaWVudCB0byBy ZXZlcnQgaXQKIyB0byB0aGUgZGVmYXVsdCB2YWx1ZSwgdW5sZXNzIHlvdSByZXN0YXJ0IHRoZSBw b3N0bWFzdGVyLgojCiMgQW55IG9wdGlvbiBjYW4gYWxzbyBiZSBnaXZlbiBhcyBhIGNvbW1hbmQg bGluZSBzd2l0Y2ggdG8gdGhlCiMgcG9zdG1hc3RlciwgZS5nLiAncG9zdG1hc3RlciAtYyBsb2df Y29ubmVjdGlvbnM9b24nLiBTb21lIG9wdGlvbnMKIyBjYW4gYmUgY2hhbmdlZCBhdCBydW4tdGlt ZSB3aXRoIHRoZSAnU0VUJyBTUUwgY29tbWFuZC4KIwojIFRoaXMgZmlsZSBpcyByZWFkIG9uIHBv c3RtYXN0ZXIgc3RhcnR1cCBhbmQgd2hlbiB0aGUgcG9zdG1hc3RlcgojIHJlY2VpdmVzIGEgU0lH SFVQLiBJZiB5b3UgZWRpdCB0aGUgZmlsZSBvbiBhIHJ1bm5pbmcgc3lzdGVtLCB5b3UgaGF2ZSAK IyB0byBTSUdIVVAgdGhlIHBvc3RtYXN0ZXIgZm9yIHRoZSBjaGFuZ2VzIHRvIHRha2UgZWZmZWN0 LCBvciB1c2UgCiMgInBnX2N0bCByZWxvYWQiLiBTb21lIHNldHRpbmdzLCBzdWNoIGFzIGxpc3Rl bl9hZGRyZXNzLCByZXF1aXJlCiMgYSBwb3N0bWFzdGVyIHNodXRkb3duIGFuZCByZXN0YXJ0IHRv IHRha2UgZWZmZWN0LgoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBGSUxFIExPQ0FUSU9OUwojLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tCgojIFRoZSBkZWZhdWx0IHZhbHVlcyBvZiB0aGVzZSB2YXJpYWJsZXMg YXJlIGRyaXZlbiBmcm9tIHRoZSAtRCBjb21tYW5kIGxpbmUKIyBzd2l0Y2ggb3IgUEdEQVRBIGVu dmlyb25tZW50IHZhcmlhYmxlLCByZXByZXNlbnRlZCBoZXJlIGFzIENvbmZpZ0Rpci4KIyBkYXRh X2RpcmVjdG9yeSA9ICdDb25maWdEaXInCQkjIHVzZSBkYXRhIGluIGFub3RoZXIgZGlyZWN0b3J5 CiMgaGJhX2ZpbGUgPSAnQ29uZmlnRGlyL3BnX2hiYS5jb25mJwkjIHRoZSBob3N0LWJhc2VkIGF1 dGhlbnRpY2F0aW9uIGZpbGUKIyBpZGVudF9maWxlID0gJ0NvbmZpZ0Rpci9wZ19pZGVudC5jb25m JyAgIyB0aGUgSURFTlQgY29uZmlndXJhdGlvbiBmaWxlCgojIElmIGV4dGVybmFsX3BpZF9maWxl IGlzIG5vdCBleHBsaWNpdGx5IHNldCwgbm8gZXh0cmEgcGlkIGZpbGUgaXMgd3JpdHRlbi4KIyBl eHRlcm5hbF9waWRfZmlsZSA9ICcobm9uZSknCQkjIHdyaXRlIGFuIGV4dHJhIHBpZCBmaWxlCgoK Iy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLQojIENPTk5FQ1RJT05TIEFORCBBVVRIRU5USUNBVElPTgojLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tCgojIC0gQ29ubmVjdGlvbiBTZXR0aW5ncyAtCgpsaXN0ZW5fYWRkcmVz c2VzID0gJyonCQkjIHdoYXQgSVAgaW50ZXJmYWNlKHMpIHRvIGxpc3RlbiBvbjsgCgkJCQkjIGRl ZmF1bHRzIHRvIGxvY2FsaG9zdCwgJyonID0gYW55CiNwb3J0ID0gNTQzMgptYXhfY29ubmVjdGlv bnMgPSAyMDAKCSMgbm90ZTogaW5jcmVhc2luZyBtYXhfY29ubmVjdGlvbnMgY29zdHMgYWJvdXQg NTAwIGJ5dGVzIG9mIHNoYXJlZAoJIyBtZW1vcnkgcGVyIGNvbm5lY3Rpb24gc2xvdCwgaW4gYWRk aXRpb24gdG8gY29zdHMgZnJvbSBzaGFyZWRfYnVmZmVycwoJIyBhbmQgbWF4X2xvY2tzX3Blcl90 cmFuc2FjdGlvbi4KI3N1cGVydXNlcl9yZXNlcnZlZF9jb25uZWN0aW9ucyA9IDIKI3VuaXhfc29j a2V0X2RpcmVjdG9yeSA9ICcnCiN1bml4X3NvY2tldF9ncm91cCA9ICcnCiN1bml4X3NvY2tldF9w ZXJtaXNzaW9ucyA9IDA3NzcJIyBvY3RhbAojcmVuZGV6dm91c19uYW1lID0gJycJCSMgZGVmYXVs dHMgdG8gdGhlIGNvbXB1dGVyIG5hbWUKCiMgLSBTZWN1cml0eSAmIEF1dGhlbnRpY2F0aW9uIC0K CiNhdXRoZW50aWNhdGlvbl90aW1lb3V0ID0gNjAJIyAxLTYwMCwgaW4gc2Vjb25kcwojc3NsID0g ZmFsc2UKI3Bhc3N3b3JkX2VuY3J5cHRpb24gPSB0cnVlCiNrcmJfc2VydmVyX2tleWZpbGUgPSAn JwojZGJfdXNlcl9uYW1lc3BhY2UgPSBmYWxzZQoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBSRVNP VVJDRSBVU0FHRSAoZXhjZXB0IFdBTCkKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQoKIyAtIE1lbW9yeSAt CgpzaGFyZWRfYnVmZmVycyA9IDE2Mzg0CSMgbWluIDE2LCBhdCBsZWFzdCBtYXhfY29ubmVjdGlv bnMqMiwgOEtCIGVhY2gKd29ya19tZW0gPSAzMjc2OAkJIyBtaW4gNjQsIHNpemUgaW4gS0IKbWFp bnRlbmFuY2Vfd29ya19tZW0gPSA2NDAwMAkjIG1pbiAxMDI0LCBzaXplIGluIEtCCiNtYXhfc3Rh Y2tfZGVwdGggPSAyMDQ4CQkjIG1pbiAxMDAsIHNpemUgaW4gS0IKCiMgLSBGcmVlIFNwYWNlIE1h cCAtCgptYXhfZnNtX3BhZ2VzID0gODAwMDAwMAkJIyBtaW4gbWF4X2ZzbV9yZWxhdGlvbnMqMTYs IDYgYnl0ZXMgZWFjaAptYXhfZnNtX3JlbGF0aW9ucyA9IDIwMDAJIyBtaW4gMTAwLCB+NTAgYnl0 ZXMgZWFjaAoKIyAtIEtlcm5lbCBSZXNvdXJjZSBVc2FnZSAtCgojbWF4X2ZpbGVzX3Blcl9wcm9j ZXNzID0gMTAwMAkjIG1pbiAyNQojcHJlbG9hZF9saWJyYXJpZXMgPSAnJwoKIyAtIENvc3QtQmFz ZWQgVmFjdXVtIERlbGF5IC0KCnZhY3V1bV9jb3N0X2RlbGF5ID0gMTAJCSMgMC0xMDAwIG1pbGxp c2Vjb25kcwp2YWN1dW1fY29zdF9wYWdlX2hpdCA9IDEJIyAwLTEwMDAwIGNyZWRpdHMKdmFjdXVt X2Nvc3RfcGFnZV9taXNzID0gMTAJIyAwLTEwMDAwIGNyZWRpdHMKdmFjdXVtX2Nvc3RfcGFnZV9k aXJ0eSA9IDIwCSMgMC0xMDAwMCBjcmVkaXRzCnZhY3V1bV9jb3N0X2xpbWl0ID0gMjAwCSMgMC0x MDAwMCBjcmVkaXRzCgojIC0gQmFja2dyb3VuZCB3cml0ZXIgLQoKYmd3cml0ZXJfZGVsYXkgPSAy MDAJCSMgMTAtMTAwMDAgbWlsbGlzZWNvbmRzIGJldHdlZW4gcm91bmRzCmJnd3JpdGVyX3BlcmNl bnQgPSAyCQkjIDAtMTAwJSBvZiBkaXJ0eSBidWZmZXJzIGluIGVhY2ggcm91bmQKYmd3cml0ZXJf bWF4cGFnZXMgPSAxMDAJIyAwLTEwMDAgYnVmZmVycyBtYXggcGVyIHJvdW5kCgoKIy0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLQojIFdSSVRFIEFIRUFEIExPRwojLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCgojIC0gU2V0 dGluZ3MgLQoKZnN5bmMgPSB0cnVlCQkJIyB0dXJucyBmb3JjZWQgc3luY2hyb25pemF0aW9uIG9u IG9yIG9mZgp3YWxfc3luY19tZXRob2QgPSBvcGVuX3N5bmMJIyB0aGUgZGVmYXVsdCB2YXJpZXMg YWNyb3NzIHBsYXRmb3JtczoKCQkJCSMgZnN5bmMsIGZkYXRhc3luYywgZnN5bmNfd3JpdGV0aHJv dWdoLAoJCQkJIyBvcGVuX3N5bmMsIG9wZW5fZGF0YXN5bmMKd2FsX2J1ZmZlcnMgPSAzMgkJIyBt aW4gNCwgOEtCIGVhY2gKI2NvbW1pdF9kZWxheSA9IDAJCSMgcmFuZ2UgMC0xMDAwMDAsIGluIG1p Y3Jvc2Vjb25kcwojY29tbWl0X3NpYmxpbmdzID0gNQkJIyByYW5nZSAxLTEwMDAKCiMgLSBDaGVj a3BvaW50cyAtCgpjaGVja3BvaW50X3NlZ21lbnRzID0gMzIJIyBpbiBsb2dmaWxlIHNlZ21lbnRz LCBtaW4gMSwgMTZNQiBlYWNoCiNjaGVja3BvaW50X3RpbWVvdXQgPSAzMDAJIyByYW5nZSAzMC0z NjAwLCBpbiBzZWNvbmRzCiNjaGVja3BvaW50X3dhcm5pbmcgPSAzMAkjIDAgaXMgb2ZmLCBpbiBz ZWNvbmRzCgojIC0gQXJjaGl2aW5nIC0KCiNhcmNoaXZlX2NvbW1hbmQgPSAnJwkJIyBjb21tYW5k IHRvIHVzZSB0byBhcmNoaXZlIGEgbG9nZmlsZSBzZWdtZW50CgoKIy0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LQojIFFVRVJZIFRVTklORwojLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCgojIC0gUGxhbm5lciBNZXRob2Qg Q29uZmlndXJhdGlvbiAtCgojZW5hYmxlX2hhc2hhZ2cgPSB0cnVlCiNlbmFibGVfaGFzaGpvaW4g PSB0cnVlCiNlbmFibGVfaW5kZXhzY2FuID0gdHJ1ZQojZW5hYmxlX21lcmdlam9pbiA9IHRydWUK I2VuYWJsZV9uZXN0bG9vcCA9IHRydWUKI2VuYWJsZV9zZXFzY2FuID0gdHJ1ZQojZW5hYmxlX3Nv cnQgPSB0cnVlCiNlbmFibGVfdGlkc2NhbiA9IHRydWUKCiMgLSBQbGFubmVyIENvc3QgQ29uc3Rh bnRzIC0KCmVmZmVjdGl2ZV9jYWNoZV9zaXplID0gMTAwMDAwCSMgdHlwaWNhbGx5IDhLQiBlYWNo CnJhbmRvbV9wYWdlX2Nvc3QgPSAyCQkjIHVuaXRzIGFyZSBvbmUgc2VxdWVudGlhbCBwYWdlIGZl dGNoIGNvc3QKI2NwdV90dXBsZV9jb3N0ID0gMC4wMQkJIyAoc2FtZSkKI2NwdV9pbmRleF90dXBs ZV9jb3N0ID0gMC4wMDEJIyAoc2FtZSkKI2NwdV9vcGVyYXRvcl9jb3N0ID0gMC4wMDI1CSMgKHNh bWUpCgojIC0gR2VuZXRpYyBRdWVyeSBPcHRpbWl6ZXIgLQoKI2dlcW8gPSB0cnVlCiNnZXFvX3Ro cmVzaG9sZCA9IDEyCiNnZXFvX2VmZm9ydCA9IDUJCSMgcmFuZ2UgMS0xMAojZ2Vxb19wb29sX3Np emUgPSAwCQkjIHNlbGVjdHMgZGVmYXVsdCBiYXNlZCBvbiBlZmZvcnQKI2dlcW9fZ2VuZXJhdGlv bnMgPSAwCQkjIHNlbGVjdHMgZGVmYXVsdCBiYXNlZCBvbiBlZmZvcnQKI2dlcW9fc2VsZWN0aW9u X2JpYXMgPSAyLjAJIyByYW5nZSAxLjUtMi4wCgojIC0gT3RoZXIgUGxhbm5lciBPcHRpb25zIC0K CmRlZmF1bHRfc3RhdGlzdGljc190YXJnZXQgPSAxMDAJIyByYW5nZSAxLTEwMDAKI2Zyb21fY29s bGFwc2VfbGltaXQgPSA4CiNqb2luX2NvbGxhcHNlX2xpbWl0ID0gOAkjIDEgZGlzYWJsZXMgY29s bGFwc2luZyBvZiBleHBsaWNpdCBKT0lOcwoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBFUlJPUiBS RVBPUlRJTkcgQU5EIExPR0dJTkcKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLQoKIyAtIFdoZXJlIHRvIExv ZyAtCgpsb2dfZGVzdGluYXRpb24gPSAnc3lzbG9nJwkjIFZhbGlkIHZhbHVlcyBhcmUgY29tYmlu YXRpb25zIG9mIHN0ZGVyciwKICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAjIHN5c2xv ZyBhbmQgZXZlbnRsb2csIGRlcGVuZGluZyBvbgogICAgICAgICAgICAgICAgICAgICAgICAgICAg ICAgICMgcGxhdGZvcm0uCgojIFRoaXMgaXMgcmVsZXZhbnQgd2hlbiBsb2dnaW5nIHRvIHN0ZGVy cjoKI3JlZGlyZWN0X3N0ZGVyciA9IGZhbHNlICAgICMgRW5hYmxlIGNhcHR1cmluZyBvZiBzdGRl cnIgaW50byBsb2cgZmlsZXMuCiMgVGhlc2UgYXJlIG9ubHkgcmVsZXZhbnQgaWYgcmVkaXJlY3Rf c3RkZXJyIGlzIHRydWU6CiNsb2dfZGlyZWN0b3J5ID0gJ3BnX2xvZycgICAjIERpcmVjdG9yeSB3 aGVyZSBsb2cgZmlsZXMgYXJlIHdyaXR0ZW4uCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAj IE1heSBiZSBzcGVjaWZpZWQgYWJzb2x1dGUgb3IgcmVsYXRpdmUgdG8gUEdEQVRBCiNsb2dfZmls ZW5hbWUgPSAncG9zdGdyZXNxbC0lWS0lbS0lZF8lSCVNJVMubG9nJyAjIExvZyBmaWxlIG5hbWUg cGF0dGVybi4KICAgICAgICAgICAgICAgICAgICAgICAgICAgICMgTWF5IGluY2x1ZGUgc3RyZnRp bWUoKSBlc2NhcGVzCiNsb2dfdHJ1bmNhdGVfb25fcm90YXRpb24gPSBmYWxzZSAgIyBJZiB0cnVl LCBhbnkgZXhpc3RpbmcgbG9nIGZpbGUgb2YgdGhlIAogICAgICAgICAgICAgICAgICAgICAgICAg ICAgIyBzYW1lIG5hbWUgYXMgdGhlIG5ldyBsb2cgZmlsZSB3aWxsIGJlIHRydW5jYXRlZAogICAg ICAgICAgICAgICAgICAgICAgICAgICAgIyByYXRoZXIgdGhhbiBhcHBlbmRlZCB0by4gIEJ1dCBz dWNoIHRydW5jYXRpb24KICAgICAgICAgICAgICAgICAgICAgICAgICAgICMgb25seSBvY2N1cnMg b24gdGltZS1kcml2ZW4gcm90YXRpb24sCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAjIG5v dCBvbiByZXN0YXJ0cyBvciBzaXplLWRyaXZlbiByb3RhdGlvbi4KICAgICAgICAgICAgICAgICAg ICAgICAgICAgICMgRGVmYXVsdCBpcyBmYWxzZSwgbWVhbmluZyBhcHBlbmQgdG8gZXhpc3Rpbmcg CiAgICAgICAgICAgICAgICAgICAgICAgICAgICAjIGZpbGVzIGluIGFsbCBjYXNlcy4KI2xvZ19y b3RhdGlvbl9hZ2UgPSAxNDQwICAgICMgQXV0b21hdGljIHJvdGF0aW9uIG9mIGxvZ2ZpbGVzIHdp bGwgaGFwcGVuIGFmdGVyCiAgICAgICAgICAgICAgICAgICAgICAgICAgICAjIHNvIG1hbnkgbWlu dXRlcy4gIDAgdG8gZGlzYWJsZS4KI2xvZ19yb3RhdGlvbl9zaXplID0gMTAyNDAgICMgQXV0b21h dGljIHJvdGF0aW9uIG9mIGxvZ2ZpbGVzIHdpbGwgaGFwcGVuIGFmdGVyCiAgICAgICAgICAgICAg ICAgICAgICAgICAgICAjIHNvIG1hbnkga2lsb2J5dGVzIG9mIGxvZyBvdXRwdXQuICAwIHRvIGRp c2FibGUuCgojIFRoZXNlIGFyZSByZWxldmFudCB3aGVuIGxvZ2dpbmcgdG8gc3lzbG9nOgpzeXNs b2dfZmFjaWxpdHkgPSAnTE9DQUw1Jwojc3lzbG9nX2lkZW50ID0gJ3Bvc3RncmVzJwoKCiMgLSBX aGVuIHRvIExvZyAtCgpjbGllbnRfbWluX21lc3NhZ2VzID0gd2FybmluZwkjIFZhbHVlcywgaW4g b3JkZXIgb2YgZGVjcmVhc2luZyBkZXRhaWw6CgkJCQkjICAgZGVidWc1LCBkZWJ1ZzQsIGRlYnVn MywgZGVidWcyLCBkZWJ1ZzEsCgkJCQkjICAgbG9nLCBub3RpY2UsIHdhcm5pbmcsIGVycm9yCgps b2dfbWluX21lc3NhZ2VzID0gd2FybmluZwkjIFZhbHVlcywgaW4gb3JkZXIgb2YgZGVjcmVhc2lu ZyBkZXRhaWw6CgkJCQkjICAgZGVidWc1LCBkZWJ1ZzQsIGRlYnVnMywgZGVidWcyLCBkZWJ1ZzEs CgkJCQkjICAgaW5mbywgbm90aWNlLCB3YXJuaW5nLCBlcnJvciwgbG9nLCBmYXRhbCwKCQkJCSMg ICBwYW5pYwoKI2xvZ19lcnJvcl92ZXJib3NpdHkgPSBkZWZhdWx0CSMgdGVyc2UsIGRlZmF1bHQs IG9yIHZlcmJvc2UgbWVzc2FnZXMKCmxvZ19taW5fZXJyb3Jfc3RhdGVtZW50ID0gd2FybmluZyAj IFZhbHVlcyBpbiBvcmRlciBvZiBpbmNyZWFzaW5nIHNldmVyaXR5OgoJCQkJICMgICBkZWJ1ZzUs IGRlYnVnNCwgZGVidWczLCBkZWJ1ZzIsIGRlYnVnMSwKCQkJCSAjICAgaW5mbywgbm90aWNlLCB3 YXJuaW5nLCBlcnJvciwgcGFuaWMob2ZmKQoJCQkJIApsb2dfbWluX2R1cmF0aW9uX3N0YXRlbWVu dCA9IDUwMDAgIyAtMSBpcyBkaXNhYmxlZCwgaW4gbWlsbGlzZWNvbmRzLgoKI3NpbGVudF9tb2Rl ID0gZmFsc2UJCSAjIERPIE5PVCBVU0Ugd2l0aG91dCBzeXNsb2cgb3IgcmVkaXJlY3Rfc3RkZXJy CgojIC0gV2hhdCB0byBMb2cgLQoKI2RlYnVnX3ByaW50X3BhcnNlID0gZmFsc2UKI2RlYnVnX3By aW50X3Jld3JpdHRlbiA9IGZhbHNlCiNkZWJ1Z19wcmludF9wbGFuID0gZmFsc2UKZGVidWdfcHJl dHR5X3ByaW50ID0gdHJ1ZQojbG9nX2Nvbm5lY3Rpb25zID0gZmFsc2UKI2xvZ19kaXNjb25uZWN0 aW9ucyA9IGZhbHNlCiNsb2dfZHVyYXRpb24gPSBmYWxzZQojbG9nX2xpbmVfcHJlZml4ID0gJyV1 ICVkICVyJwkJIyBlLmcuICc8JXUlJSVkPiAnIAoJCQkJIyAldT11c2VyIG5hbWUgJWQ9ZGF0YWJh c2UgbmFtZQoJCQkJIyAlcj1yZW1vdGUgaG9zdCBhbmQgcG9ydAoJCQkJIyAlcD1QSUQgJXQ9dGlt ZXN0YW1wICVpPWNvbW1hbmQgdGFnCgkJCQkjICVjPXNlc3Npb24gaWQgJWw9c2Vzc2lvbiBsaW5l IG51bWJlcgoJCQkJIyAlcz1zZXNzaW9uIHN0YXJ0IHRpbWVzdGFtcCAleD10cmFuc2FjdGlvbiBp ZAoJCQkJIyAlcT1zdG9wIGhlcmUgaW4gbm9uLXNlc3Npb24gcHJvY2Vzc2VzCgkJCQkjICUlPScl JwojbG9nX3N0YXRlbWVudCA9ICdub25lJwkJIyBub25lLCBtb2QsIGRkbCwgYWxsCiNsb2dfaG9z dG5hbWUgPSBmYWxzZQoKCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KIyBSVU5USU1FIFNUQVRJU1RJQ1MK Iy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLQoKIyAtIFN0YXRpc3RpY3MgTW9uaXRvcmluZyAtCgojbG9nX3Bh cnNlcl9zdGF0cyA9IGZhbHNlCiNsb2dfcGxhbm5lcl9zdGF0cyA9IGZhbHNlCiNsb2dfZXhlY3V0 b3Jfc3RhdHMgPSBmYWxzZQojbG9nX3N0YXRlbWVudF9zdGF0cyA9IGZhbHNlCgojIC0gUXVlcnkv SW5kZXggU3RhdGlzdGljcyBDb2xsZWN0b3IgLQoKc3RhdHNfc3RhcnRfY29sbGVjdG9yID0gdHJ1 ZQpzdGF0c19jb21tYW5kX3N0cmluZyA9IHRydWUKI3N0YXRzX2Jsb2NrX2xldmVsID0gZmFsc2UK c3RhdHNfcm93X2xldmVsID0gdHJ1ZQpzdGF0c19yZXNldF9vbl9zZXJ2ZXJfc3RhcnQgPSB0cnVl CgoKIy0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLQojIENMSUVOVCBDT05ORUNUSU9OIERFRkFVTFRTCiMtLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0KCiMgLSBTdGF0ZW1lbnQgQmVoYXZpb3IgLQoKI3NlYXJjaF9wYXRoID0g JyR1c2VyLHB1YmxpYycJIyBzY2hlbWEgbmFtZXMKI2RlZmF1bHRfdGFibGVzcGFjZSA9ICcnCSMg YSB0YWJsZXNwYWNlIG5hbWUsIG9yICcnIGZvciBkZWZhdWx0CiNjaGVja19mdW5jdGlvbl9ib2Rp ZXMgPSB0cnVlCiNkZWZhdWx0X3RyYW5zYWN0aW9uX2lzb2xhdGlvbiA9ICdyZWFkIGNvbW1pdHRl ZCcKI2RlZmF1bHRfdHJhbnNhY3Rpb25fcmVhZF9vbmx5ID0gZmFsc2UKI3N0YXRlbWVudF90aW1l b3V0ID0gMAkJIyAwIGlzIGRpc2FibGVkLCBpbiBtaWxsaXNlY29uZHMKCiMgLSBMb2NhbGUgYW5k IEZvcm1hdHRpbmcgLQoKI2RhdGVzdHlsZSA9ICdpc28sIG1keScKI3RpbWV6b25lID0gdW5rbm93 bgkJIyBhY3R1YWxseSwgZGVmYXVsdHMgdG8gVFogZW52aXJvbm1lbnQgc2V0dGluZwojYXVzdHJh bGlhbl90aW1lem9uZXMgPSBmYWxzZQojZXh0cmFfZmxvYXRfZGlnaXRzID0gMAkJIyBtaW4gLTE1 LCBtYXggMgojY2xpZW50X2VuY29kaW5nID0gc3FsX2FzY2lpCSMgYWN0dWFsbHksIGRlZmF1bHRz IHRvIGRhdGFiYXNlIGVuY29kaW5nCgojIFRoZXNlIHNldHRpbmdzIGFyZSBpbml0aWFsaXplZCBi eSBpbml0ZGIgLS0gdGhleSBtaWdodCBiZSBjaGFuZ2VkCmxjX21lc3NhZ2VzID0gJ0MnCQkjIGxv Y2FsZSBmb3Igc3lzdGVtIGVycm9yIG1lc3NhZ2Ugc3RyaW5ncwpsY19tb25ldGFyeSA9ICdDJwkJ IyBsb2NhbGUgZm9yIG1vbmV0YXJ5IGZvcm1hdHRpbmcKbGNfbnVtZXJpYyA9ICdDJwkJIyBsb2Nh bGUgZm9yIG51bWJlciBmb3JtYXR0aW5nCmxjX3RpbWUgPSAnQycJCQkjIGxvY2FsZSBmb3IgdGlt ZSBmb3JtYXR0aW5nCgojIC0gT3RoZXIgRGVmYXVsdHMgLQoKI2V4cGxhaW5fcHJldHR5X3ByaW50 ID0gdHJ1ZQojZHluYW1pY19saWJyYXJ5X3BhdGggPSAnJGxpYmRpcicKCgojLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tCiMgTE9DSyBNQU5BR0VNRU5UCiMtLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0KCiNkZWFkbG9ja190 aW1lb3V0ID0gMTAwMAkjIGluIG1pbGxpc2Vjb25kcwojbWF4X2xvY2tzX3Blcl90cmFuc2FjdGlv biA9IDY0CSMgbWluIDEwLCB+MjAwKm1heF9jb25uZWN0aW9ucyBieXRlcyBlYWNoCgoKIy0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLQojIFZFUlNJT04vUExBVEZPUk0gQ09NUEFUSUJJTElUWQojLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tCgojIC0gUHJldmlvdXMgUG9zdGdyZXMgVmVyc2lvbnMgLQoKYWRkX21pc3Npbmdf ZnJvbSA9IGZhbHNlCiNyZWdleF9mbGF2b3IgPSBhZHZhbmNlZAkjIGFkdmFuY2VkLCBleHRlbmRl ZCwgb3IgYmFzaWMKc3FsX2luaGVyaXRhbmNlID0gZmFsc2UKZGVmYXVsdF93aXRoX29pZHMgPSBm YWxzZQoKIyAtIE90aGVyIFBsYXRmb3JtcyAmIENsaWVudHMgLQoKdHJhbnNmb3JtX251bGxfZXF1 YWxzID0gZmFsc2UK --Multipart_Sun__17_Jul_2005_21_34_16_-0600_=WUE/07VkC53E0WZ-- --Signature_Sun__17_Jul_2005_21_34_16_-0600__=pJJtTXBIjc/Vrz Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbI0oACgkQLQ/DKuwDYzlC8wCdGpe4ptr9dMB1aAED9XKcExYW LB0An3/QkFQVuGmO+RJYcSBSRXQC6hEx =Tro8 -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_21_34_16_-0600__=pJJtTXBIjc/Vrz-- From pgsql-performance-owner@postgresql.org Mon Jul 18 00:49:06 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7129B529C9 for ; Mon, 18 Jul 2005 00:49:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62150-09 for ; Mon, 18 Jul 2005 03:48:59 +0000 (GMT) Received: from outbound.mailhop.org (outbound.mailhop.org [63.208.196.171]) by svr1.postgresql.org (Postfix) with ESMTP id 307EC529AE for ; Mon, 18 Jul 2005 00:48:35 -0300 (ADT) Received: from ool-4353a8a5.dyn.optonline.net ([67.83.168.165] helo=[192.168.1.102]) by outbound.mailhop.org with esmtpa (Exim 4.51) id 1DuMcB-000Pcw-TX; Sun, 17 Jul 2005 23:48:23 -0400 X-Mail-Handler: MailHop Outbound by DynDNS.org X-Originating-IP: 67.83.168.165 X-Report-Abuse-To: abuse@dyndns.org (see http://www.mailhop.org/outbound/abuse.html for abuse reporting information) X-MHO-User: zeut Message-ID: <42DB2684.9020706@zeut.net> Date: Sun, 17 Jul 2005 23:48:20 -0400 From: "Matthew T. O'Connor" User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Robert Creager Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - References: <20050717213416.59dcb52e@thunder.logicalchaos.org> In-Reply-To: <20050717213416.59dcb52e@thunder.logicalchaos.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/241 X-Sequence-Number: 13482 Robert Creager wrote: >For 8.03, pg_autovacuum is running. On 7.4.1, I set up a cron job to vacuum >analyze every 5 minutes. > > Are you sure that pg_autovacuum is doing it's job? Meaning are you sure it's vacuuming as often as needed? Try to run it with -d2 or so and make sure that it is actually doing the vacuuming needed. From pgsql-performance-owner@postgresql.org Mon Jul 18 01:11:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 28A8A52831 for ; Mon, 18 Jul 2005 01:11:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71230-03 for ; Mon, 18 Jul 2005 04:11:02 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 4CD8252AB3 for ; Mon, 18 Jul 2005 01:11:01 -0300 (ADT) Received: (qmail 4967 invoked by uid 1010); 18 Jul 2005 00:11:00 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 7.168994 secs); 18 Jul 2005 04:11:00 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 00:10:53 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 335AA72328; Sun, 17 Jul 2005 22:10:51 -0600 (MDT) Date: Sun, 17 Jul 2005 22:10:50 -0600 From: Robert Creager To: "Matthew T. O'Connor" Cc: PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717221050.7c4ac241@thunder.logicalchaos.org> In-Reply-To: <42DB2684.9020706@zeut.net> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary=Signature_Sun__17_Jul_2005_22_10_50_-0600_f8q37Az1.H73v85v; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/242 X-Sequence-Number: 13483 --Signature_Sun__17_Jul_2005_22_10_50_-0600_f8q37Az1.H73v85v Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable I am, and it is. It's ANALYZING and VACUUM'ing tables every interval (5 mi= nutes - 8.0.3). Right now, for that last 4 hours, I'm not VACUUMing the 7.4.1 database and it's still clicking along at < .2 second queries. Last year (7.4.1), I noticed that it took about a week of heavy activity (for this DB) before I'd really need a vacuum. That's when I put in the 5 min cron. When I first switched over to 8.0.3, I was still running the cron vacuum. = I got into big trouble when I had vacuum's backed up for 6 hours. That's when I started noticing the query problem, and the CS numbers being high. 7.4.1 vacuums every 5 minutes always take < 30 seconds (when I'm watching). Cheers, Rob When grilled further on (Sun, 17 Jul 2005 23:48:20 -0400), "Matthew T. O'Connor" confessed: > Robert Creager wrote: >=20 > >For 8.03, pg_autovacuum is running. On 7.4.1, I set up a cron job to va= cuum > >analyze every 5 minutes. > > =20 > > >=20 > Are you sure that pg_autovacuum is doing it's job? Meaning are you sure= =20 > it's vacuuming as often as needed? Try to run it with -d2 or so and=20 > make sure that it is actually doing the vacuuming needed. --=20 22:04:10 up 4 days, 1:39, 8 users, load average: 0.15, 0.15, 0.12 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_22_10_50_-0600_f8q37Az1.H73v85v Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbK8oACgkQLQ/DKuwDYzkVfwCeI8iUFCNmJVfDnsaZBFeqfeo0 qG4AnjdzMJIaPcEqnwebel07bsMSFgnd =RdSZ -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_22_10_50_-0600_f8q37Az1.H73v85v-- From pgsql-performance-owner@postgresql.org Mon Jul 18 01:11:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3CBDE5286E for ; Mon, 18 Jul 2005 01:11:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 67568-08 for ; Mon, 18 Jul 2005 04:11:05 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id B25F152AB9 for ; Mon, 18 Jul 2005 01:11:01 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6I4Arx9024521; Mon, 18 Jul 2005 00:10:53 -0400 (EDT) To: Robert Creager Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? In-reply-to: <20050717213416.59dcb52e@thunder.logicalchaos.org> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> Comments: In-reply-to Robert Creager message dated "Sun, 17 Jul 2005 21:34:16 -0600" Date: Mon, 18 Jul 2005 00:10:53 -0400 Message-ID: <24520.1121659853@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/243 X-Sequence-Number: 13484 Robert Creager writes: > I'm guessing this is the CS problem that reared it's head last year? The context swap problem was no worse in 8.0 than in prior versions, so that hardly seems like a good explanation. Have you tried reverting to the cron-based vacuuming method you used in 7.4? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 01:18:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AC6DD52A90 for ; Mon, 18 Jul 2005 01:18:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 72432-07 for ; Mon, 18 Jul 2005 04:18:09 +0000 (GMT) Received: from smtp003.bizmail.yahoo.com (smtp003.bizmail.yahoo.com [216.136.130.195]) by svr1.postgresql.org (Postfix) with SMTP id D2AA352831 for ; Mon, 18 Jul 2005 01:18:08 -0300 (ADT) Received: (qmail 25448 invoked from network); 18 Jul 2005 04:18:07 -0000 Received: from unknown (HELO ?192.168.3.41?) (david.mitchell@telogis.com@203.98.10.169 with plain) by smtp003.bizmail.yahoo.com with SMTP; 18 Jul 2005 04:18:07 -0000 Message-ID: <42DB2D72.8090401@telogis.com> Date: Mon, 18 Jul 2005 16:17:54 +1200 From: David Mitchell User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Robert Creager Cc: "Matthew T. O'Connor" , PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> In-Reply-To: <20050717221050.7c4ac241@thunder.logicalchaos.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/244 X-Sequence-Number: 13485 Sounds like either someone is holding a lock on your pg8 db, or maybe you need a vacuum full. No amount of normal vacuuming will fix a table that needs a vacuum full. Although if that were the case I'd expect you to have slow queries regardless of the number of concurrent connections. Maybe you should check who is holding locks. David Robert Creager wrote: > I am, and it is. It's ANALYZING and VACUUM'ing tables every interval (5 minutes > - 8.0.3). Right now, for that last 4 hours, I'm not VACUUMing the 7.4.1 > database and it's still clicking along at < .2 second queries. Last year > (7.4.1), I noticed that it took about a week of heavy activity (for this DB) > before I'd really need a vacuum. That's when I put in the 5 min cron. > > When I first switched over to 8.0.3, I was still running the cron vacuum. I got > into big trouble when I had vacuum's backed up for 6 hours. That's when I > started noticing the query problem, and the CS numbers being high. 7.4.1 > vacuums every 5 minutes always take < 30 seconds (when I'm watching). > > Cheers, > Rob > > When grilled further on (Sun, 17 Jul 2005 23:48:20 -0400), > "Matthew T. O'Connor" confessed: > > >>Robert Creager wrote: >> >> >>>For 8.03, pg_autovacuum is running. On 7.4.1, I set up a cron job to vacuum >>>analyze every 5 minutes. >>> >>> >> >>Are you sure that pg_autovacuum is doing it's job? Meaning are you sure >>it's vacuuming as often as needed? Try to run it with -d2 or so and >>make sure that it is actually doing the vacuuming needed. > > > -- David Mitchell Software Engineer Telogis From pgsql-performance-owner@postgresql.org Mon Jul 18 01:18:45 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6FE5352865 for ; Mon, 18 Jul 2005 01:18:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 75293-01 for ; Mon, 18 Jul 2005 04:18:34 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 2FD1E52A90 for ; Mon, 18 Jul 2005 01:18:33 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6I4IU4s024617; Mon, 18 Jul 2005 00:18:30 -0400 (EDT) To: Robert Creager Cc: "Matthew T. O'Connor" , PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS In-reply-to: <20050717221050.7c4ac241@thunder.logicalchaos.org> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> Comments: In-reply-to Robert Creager message dated "Sun, 17 Jul 2005 22:10:50 -0600" Date: Mon, 18 Jul 2005 00:18:30 -0400 Message-ID: <24616.1121660310@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/245 X-Sequence-Number: 13486 Robert Creager writes: > I am, and it is. It's ANALYZING and VACUUM'ing tables every interval (5 mi= > nutes > - 8.0.3). Right now, for that last 4 hours, I'm not VACUUMing the 7.4.1 > database and it's still clicking along at < .2 second queries. Have you compared physical table sizes? If the autovac daemon did let things get out of hand, you'd need a VACUUM FULL or CLUSTER or TRUNCATE to get the table size back down --- plain VACUUM is unlikely to fix it. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 01:18:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 34B1152A21 for ; Mon, 18 Jul 2005 01:18:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 75195-03 for ; Mon, 18 Jul 2005 04:18:46 +0000 (GMT) Received: from outbound.mailhop.org (outbound.mailhop.org [63.208.196.171]) by svr1.postgresql.org (Postfix) with ESMTP id CC00E52A0D for ; Mon, 18 Jul 2005 01:18:45 -0300 (ADT) Received: from ool-4353a8a5.dyn.optonline.net ([67.83.168.165] helo=[192.168.1.102]) by outbound.mailhop.org with esmtpa (Exim 4.51) id 1DuN5Y-0006qL-8G; Mon, 18 Jul 2005 00:18:44 -0400 X-Mail-Handler: MailHop Outbound by DynDNS.org X-Originating-IP: 67.83.168.165 X-Report-Abuse-To: abuse@dyndns.org (see http://www.mailhop.org/outbound/abuse.html for abuse reporting information) X-MHO-User: zeut Message-ID: <42DB2DA3.6000909@zeut.net> Date: Mon, 18 Jul 2005 00:18:43 -0400 From: "Matthew T. O'Connor" User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Robert Creager Cc: PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> In-Reply-To: <20050717221050.7c4ac241@thunder.logicalchaos.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/246 X-Sequence-Number: 13487 Ok, it doesn't look like an autovacuum problem. The only other thing I can think of is that some query is doing a seq scan rather than an index scan. Have you turned on the query logging to see what queries are taking so long? Matt Robert Creager wrote: >I am, and it is. It's ANALYZING and VACUUM'ing tables every interval (5 minutes >- 8.0.3). Right now, for that last 4 hours, I'm not VACUUMing the 7.4.1 >database and it's still clicking along at < .2 second queries. Last year >(7.4.1), I noticed that it took about a week of heavy activity (for this DB) >before I'd really need a vacuum. That's when I put in the 5 min cron. > >When I first switched over to 8.0.3, I was still running the cron vacuum. I got >into big trouble when I had vacuum's backed up for 6 hours. That's when I >started noticing the query problem, and the CS numbers being high. 7.4.1 >vacuums every 5 minutes always take < 30 seconds (when I'm watching). > >Cheers, >Rob > >When grilled further on (Sun, 17 Jul 2005 23:48:20 -0400), >"Matthew T. O'Connor" confessed: > > > >>Robert Creager wrote: >> >> >> >>>For 8.03, pg_autovacuum is running. On 7.4.1, I set up a cron job to vacuum >>>analyze every 5 minutes. >>> >>> >>> >>> >>Are you sure that pg_autovacuum is doing it's job? Meaning are you sure >>it's vacuuming as often as needed? Try to run it with -d2 or so and >>make sure that it is actually doing the vacuuming needed. >> >> > > > > From pgsql-performance-owner@postgresql.org Mon Jul 18 01:33:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 72ACE52AE5 for ; Mon, 18 Jul 2005 01:33:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74721-08 for ; Mon, 18 Jul 2005 04:33:05 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 3C2EA52AD4 for ; Mon, 18 Jul 2005 01:33:04 -0300 (ADT) Received: (qmail 20635 invoked by uid 1010); 18 Jul 2005 00:33:04 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.892918 secs); 18 Jul 2005 04:33:04 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 00:33:02 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 39C97B31D3; Sun, 17 Jul 2005 22:33:00 -0600 (MDT) Date: Sun, 17 Jul 2005 22:32:59 -0600 From: Robert Creager To: "Matthew T. O'Connor" Cc: PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717223259.4eca3b8c@thunder.logicalchaos.org> In-Reply-To: <42DB2DA3.6000909@zeut.net> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> <42DB2DA3.6000909@zeut.net> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Sun__17_Jul_2005_22_32_59_-0600_=FG.4ymlW8TtsHIL"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/247 X-Sequence-Number: 13488 --Signature_Sun__17_Jul_2005_22_32_59_-0600_=FG.4ymlW8TtsHIL Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 00:18:43 -0400), "Matthew T. O'Connor" confessed: > Have you turned on the query logging to see what queries are=20 > taking so long? >=20 Yeah. In the original message is a typical query. One from 741 and the ot= her on 803. On 803, an explain analyze is done twice. Once during the problem, once when the system is idle. On 741, the query behaves the same no matter what... Cheers, Rob --=20 22:31:18 up 4 days, 2:06, 8 users, load average: 0.25, 0.18, 0.11 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_22_32_59_-0600_=FG.4ymlW8TtsHIL Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbMPsACgkQLQ/DKuwDYzkCpwCgjzwS/ibdEN9d3LAfY3gKRDfD srsAnj3K7fWvN3TADjH9FI2TCzV2ZZp9 =2ZIa -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_22_32_59_-0600_=FG.4ymlW8TtsHIL-- From pgsql-performance-owner@postgresql.org Mon Jul 18 02:02:25 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2979352AEB for ; Mon, 18 Jul 2005 02:02:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86627-04 for ; Mon, 18 Jul 2005 05:02:12 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 4E72052831 for ; Mon, 18 Jul 2005 02:02:11 -0300 (ADT) Received: (qmail 29973 invoked by uid 1010); 18 Jul 2005 01:02:11 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.083139 secs); 18 Jul 2005 05:02:11 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 01:02:09 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 8CC6D311CE; Sun, 17 Jul 2005 23:02:08 -0600 (MDT) Date: Sun, 17 Jul 2005 23:02:08 -0600 From: Robert Creager To: David Mitchell Cc: "Matthew T. O'Connor" , PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717230208.6b6c319d@thunder.logicalchaos.org> In-Reply-To: <42DB2D72.8090401@telogis.com> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> <42DB2D72.8090401@telogis.com> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Sun__17_Jul_2005_23_02_08_-0600_AvgBjWTU=k=9Ss+E"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/248 X-Sequence-Number: 13489 --Signature_Sun__17_Jul_2005_23_02_08_-0600_AvgBjWTU=k=9Ss+E Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 16:17:54 +1200), David Mitchell confessed: > Maybe you should check who is holding locks. Hmmm... The only difference is how the vacuum is run. One by autovacuum, = one by cron (vacuum analyze every 5 minutes). Cheers, Rob --=20 23:01:44 up 4 days, 2:36, 6 users, load average: 0.27, 0.16, 0.10 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_23_02_08_-0600_AvgBjWTU=k=9Ss+E Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbN9AACgkQLQ/DKuwDYznUzACdEf6SIy9mgrMHjZ5sgqNyj8O0 V4MAniyDViWlBoQ1DxWlDiErWOzl365c =3Xah -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_23_02_08_-0600_AvgBjWTU=k=9Ss+E-- From pgsql-performance-owner@postgresql.org Mon Jul 18 02:09:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3FBB452AEB for ; Mon, 18 Jul 2005 02:09:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87060-02 for ; Mon, 18 Jul 2005 05:08:58 +0000 (GMT) Received: from mail22.sea5.speakeasy.net (mail22.sea5.speakeasy.net [69.17.117.24]) by svr1.postgresql.org (Postfix) with ESMTP id A923D52831 for ; Mon, 18 Jul 2005 02:08:56 -0300 (ADT) Received: (qmail 30001 invoked from network); 18 Jul 2005 05:08:56 -0000 Received: from dsl081-060-184.sfo1.dsl.speakeasy.net (HELO noodles) ([64.81.60.184]) (envelope-sender ) by mail22.sea5.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 18 Jul 2005 05:08:56 -0000 Received: from jwb by noodles with local (Exim 3.36 #1 (Debian)) id 1DuNsO-0002sS-00; Sun, 17 Jul 2005 22:09:12 -0700 Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - From: "Jeffrey W. Baker" To: Robert Creager Cc: PGPerformance , Andy Hewitt In-Reply-To: <20050717213416.59dcb52e@thunder.logicalchaos.org> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Sun, 17 Jul 2005 22:09:11 -0700 Message-Id: <1121663351.10108.1.camel@noodles> Mime-Version: 1.0 X-Mailer: Evolution 2.3.5.1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/249 X-Sequence-Number: 13490 On Sun, 2005-07-17 at 21:34 -0600, Robert Creager wrote: > Sigh... > > I recently upgraded from 7.4.1 to 8.0.3. The application did not change. I'm > now running both database concurrently (on different ports, same machine) just > so I could verify the problem really exists. > > The application is a custom test application for testing mechanical systems. > The runs in question (4 at a time) each generate 16 queries at a time of which > the results are sent to the mechanical system which processes the request, which > processes them anywhere from 10 to 120 seconds. The system is capable of > completing between 4 and 8 jobs at once. So, once the system is running, at > most there will be 8 queries per run simultaneously. > > The entire database fits into RAM (2Gb), as evidenced by no disk activity and > relatively small database size. pg_xlog is on different disks from the db. > > The problem is that on version 8.0.3, once I get 3 or more concurrent runs > going, the query times start tanking (>20 seconds). On 7.4.1, the applications > hum along with queries typically below .2 seconds on over 5 concurrent runs. > Needless to say, 7.4.1 behaves as expected... The only change between runs is > the port connecting to. Bot DB's are up at the same time. > > For 8.03, pg_autovacuum is running. On 7.4.1, I set up a cron job to vacuum > analyze every 5 minutes. > > The system is Mandrake Linux running 2.4.22 kernel with dual Intel Xenon CPU > with HT enabled. On an 803 run, the context switching is up around 60k. On > 7.4.1, it maxes around 23k and averages < 1k. Did you build 8.0.3 yourself, or install it from packages? I've seen in the past where pg would build with the wrong kind of mutexes on some machines, and that would send the CS through the roof. If you did build it yourself, check your ./configure logs. If not, try strace. -jwb From pgsql-performance-owner@postgresql.org Mon Jul 18 02:09:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 26A9F52831 for ; Mon, 18 Jul 2005 02:09:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 83890-09 for ; Mon, 18 Jul 2005 05:09:20 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 09D0252AFA for ; Mon, 18 Jul 2005 02:09:19 -0300 (ADT) Received: (qmail 32175 invoked by uid 1010); 18 Jul 2005 01:09:20 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.110409 secs); 18 Jul 2005 05:09:20 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 01:09:18 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 9B4DA311CE; Sun, 17 Jul 2005 23:09:12 -0600 (MDT) Date: Sun, 17 Jul 2005 23:09:12 -0600 From: Robert Creager To: Tom Lane Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717230912.12d51ec9@thunder.logicalchaos.org> In-Reply-To: <24520.1121659853@sss.pgh.pa.us> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary=Signature_Sun__17_Jul_2005_23_09_12_-0600_Rpg7oL+V5OnpZbJ.; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/250 X-Sequence-Number: 13491 --Signature_Sun__17_Jul_2005_23_09_12_-0600_Rpg7oL+V5OnpZbJ. Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 00:10:53 -0400), Tom Lane confessed: > Have you tried reverting > to the cron-based vacuuming method you used in 7.4? >=20 I just stopped autovacuum, ran a manual vacuum analyze on 803 (2064 pages needed, 8000000 FSM setting) and re-started the run (with cron vac enabled)= .=20 The query problem has not showed up yet (1/2 hour). A vacuum on 741 showed= 3434 pages needed, 200000 FSM setting. I'll let it run the night and see if it shows up after a couple of hours. = It has run clean for 1 hour prior. If this runs 'till morning, I'll re-enable= the autovacuum, disable the cron and see if it reproduces itself (the slowdown). Cheers, Rob --=20 22:18:40 up 4 days, 1:53, 8 users, load average: 0.10, 0.20, 0.14 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_23_09_12_-0600_Rpg7oL+V5OnpZbJ. Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbOXgACgkQLQ/DKuwDYznSGwCbBWYsuWu4EBxxx8LU2GWY0t4T T9YAnRhBjfYSAVvq3RkaR7jQWnWpyU5J =VwCW -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_23_09_12_-0600_Rpg7oL+V5OnpZbJ.-- From pgsql-performance-owner@postgresql.org Mon Jul 18 02:27:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9633152B0A for ; Mon, 18 Jul 2005 02:27:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 88157-05 for ; Mon, 18 Jul 2005 05:27:11 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 7A7BE52AED for ; Mon, 18 Jul 2005 02:27:11 -0300 (ADT) Received: (qmail 5148 invoked by uid 1010); 18 Jul 2005 01:27:12 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.583789 secs); 18 Jul 2005 05:27:12 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 01:27:10 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id DDBF772328; Sun, 17 Jul 2005 23:27:08 -0600 (MDT) Date: Sun, 17 Jul 2005 23:27:08 -0600 From: Robert Creager To: "Jeffrey W. Baker" Cc: Robert Creager , PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717232708.382bb715@thunder.logicalchaos.org> In-Reply-To: <1121663351.10108.1.camel@noodles> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <1121663351.10108.1.camel@noodles> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Sun__17_Jul_2005_23_27_08_-0600_aqD+2JK2Qri3H/mX"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/251 X-Sequence-Number: 13492 --Signature_Sun__17_Jul_2005_23_27_08_-0600_aqD+2JK2Qri3H/mX Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Sun, 17 Jul 2005 22:09:11 -0700), "Jeffrey W. Baker" confessed: >=20 > Did you build 8.0.3 yourself, or install it from packages? I've seen in > the past where pg would build with the wrong kind of mutexes on some > machines, and that would send the CS through the roof. If you did build > it yourself, check your ./configure logs. If not, try strace. I always build PG from source. I did check the config.log command line (./configure) and they were similar enough. The system has not changed bet= ween building the two versions (if it ain't broke...). Cheers, Rob --=20 23:25:21 up 4 days, 3:00, 6 users, load average: 0.25, 0.15, 0.11 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_23_27_08_-0600_aqD+2JK2Qri3H/mX Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbPawACgkQLQ/DKuwDYzn38ACgmY+9GhBnWSolj8/ju3/UNvRG pm4AnRw7hD7Agm1/KLLQ4d8awLjdT+lq =11o7 -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_23_27_08_-0600_aqD+2JK2Qri3H/mX-- From pgsql-performance-owner@postgresql.org Mon Jul 18 02:43:37 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1438E52AF8 for ; Mon, 18 Jul 2005 02:43:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94280-02 for ; Mon, 18 Jul 2005 05:43:32 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 113ED52AE1 for ; Mon, 18 Jul 2005 02:43:31 -0300 (ADT) Received: (qmail 9280 invoked by uid 1010); 18 Jul 2005 01:43:32 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.237431 secs); 18 Jul 2005 05:43:32 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 01:43:30 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 64FE4311CE; Sun, 17 Jul 2005 23:43:29 -0600 (MDT) Date: Sun, 17 Jul 2005 23:43:29 -0600 From: Robert Creager To: Tom Lane Cc: "Matthew T. O'Connor" , PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717234329.05ce2e52@thunder.logicalchaos.org> In-Reply-To: <24616.1121660310@sss.pgh.pa.us> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> <24616.1121660310@sss.pgh.pa.us> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary=Signature_Sun__17_Jul_2005_23_43_29_-0600_z2HHaZ2Lg_vObEHB; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/252 X-Sequence-Number: 13493 --Signature_Sun__17_Jul_2005_23_43_29_-0600_z2HHaZ2Lg_vObEHB Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 00:18:30 -0400), Tom Lane confessed: > Robert Creager writes: > > I am, and it is. It's ANALYZING and VACUUM'ing tables every interval (= 5 mi=3D > > nutes > > - 8.0.3). Right now, for that last 4 hours, I'm not VACUUMing the 7.4.1 > > database and it's still clicking along at < .2 second queries. >=20 > Have you compared physical table sizes? If the autovac daemon did let > things get out of hand, you'd need a VACUUM FULL or CLUSTER or TRUNCATE > to get the table size back down --- plain VACUUM is unlikely to fix it. Table sizes, no. Entire DB size is 45Mb for 803 and 29Mb for 741. Cannot = make a direct comparison between the two as I've run against more machines now w= ith 803 than 741, so I'd expect it to be larger. I'm still running relatively clean on 803 with cron vacuum. The CS are jum= ping from 100 to 120k, but it's not steady state like it was before, and queries= are all under 5 seconds (none hitting the logs) and are typically (glancing at = test runs) still under 1 sec, with some hitting ~2 seconds occasionally. I've 6 runs going concurrently. Just saw (vmstat 1) a set of 8 seconds whe= re the CS didn't drop below 90k, but right now its at ~300 for over 30 seconds= ...=20 It's bouncing all over the place, but staying reasonably well behaved overa= ll. Whoop. Spoke too soon. Just hit the wall. CS at ~80k constant, queries o= ver 10 seconds and rising (30+ now)... Looking at ps, the vacuum is currently running. Going back in the logs, the CS and vacuum hit at about the same t= ime. I'm going to go back to 741 with the same load and see what happens by tomo= rrow morning... I'll change the cron vac to hit the 741 db. Cheers, Rob --=20 23:29:24 up 4 days, 3:04, 6 users, load average: 0.02, 0.07, 0.08 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_23_43_29_-0600_z2HHaZ2Lg_vObEHB Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbQYEACgkQLQ/DKuwDYzn8MgCfSWOcbx4kPIaeGUaDpA5t8nF3 newAn3/EzLgp9ysUXGL8S4/CUWKDDCvw =znj8 -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_23_43_29_-0600_z2HHaZ2Lg_vObEHB-- From pgsql-performance-owner@postgresql.org Mon Jul 18 02:56:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BD6A152AA6 for ; Mon, 18 Jul 2005 02:56:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 96888-02 for ; Mon, 18 Jul 2005 05:56:34 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 8F69552A73 for ; Mon, 18 Jul 2005 02:56:33 -0300 (ADT) Received: (qmail 12607 invoked by uid 1010); 18 Jul 2005 01:56:33 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.2516 secs); 18 Jul 2005 05:56:33 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 01:56:31 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 7DBBD72328; Sun, 17 Jul 2005 23:56:30 -0600 (MDT) Date: Sun, 17 Jul 2005 23:56:30 -0600 From: Robert Creager To: Tom Lane Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050717235630.26a7c979@thunder.logicalchaos.org> In-Reply-To: <24520.1121659853@sss.pgh.pa.us> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Sun__17_Jul_2005_23_56_30_-0600_qOYAQc/dgQEn9yoa"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/253 X-Sequence-Number: 13494 --Signature_Sun__17_Jul_2005_23_56_30_-0600_qOYAQc/dgQEn9yoa Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 00:10:53 -0400), Tom Lane confessed: > The context swap problem was no worse in 8.0 than in prior versions, > so that hardly seems like a good explanation. Have you tried reverting > to the cron-based vacuuming method you used in 7.4? >=20 I've "vacuum_cost_delay =3D 10" in the conf file for 803. hit, miss, dirty= and limit are 1, 10, 20 and 200 respectively. Could that be contributing to the problem? I'll know more in an hour or so with 741 running and cron vac and= the same load. Cheers, Rob --=20 23:53:53 up 4 days, 3:28, 6 users, load average: 0.11, 0.13, 0.11 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sun__17_Jul_2005_23_56_30_-0600_qOYAQc/dgQEn9yoa Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbRI4ACgkQLQ/DKuwDYzko9QCeLnwlPcHVNLWZITxl/CXdIZQm pyAAn2pAlqiWxpXlb/B4A+O1mG8HnVOf =TwF/ -----END PGP SIGNATURE----- --Signature_Sun__17_Jul_2005_23_56_30_-0600_qOYAQc/dgQEn9yoa-- From pgsql-performance-owner@postgresql.org Mon Jul 18 03:07:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 404BF52806 for ; Mon, 18 Jul 2005 03:07:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95312-06 for ; Mon, 18 Jul 2005 06:07:18 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 2735E52C6F for ; Mon, 18 Jul 2005 03:07:17 -0300 (ADT) Received: (qmail 15688 invoked by uid 1010); 18 Jul 2005 02:07:17 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 2.472975 secs); 18 Jul 2005 06:07:17 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 02:07:15 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id D1C4B3EA42; Mon, 18 Jul 2005 00:07:13 -0600 (MDT) Date: Mon, 18 Jul 2005 00:07:13 -0600 From: Robert Creager To: Robert Creager Cc: Tom Lane , "Matthew T. O'Connor" , PGPerformance Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050718000713.3366dad9@thunder.logicalchaos.org> In-Reply-To: <20050717234329.05ce2e52@thunder.logicalchaos.org> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <42DB2684.9020706@zeut.net> <20050717221050.7c4ac241@thunder.logicalchaos.org> <24616.1121660310@sss.pgh.pa.us> <20050717234329.05ce2e52@thunder.logicalchaos.org> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Mon__18_Jul_2005_00_07_13_-0600_yzEq77=SL24n3HxN"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/254 X-Sequence-Number: 13495 --Signature_Mon__18_Jul_2005_00_07_13_-0600_yzEq77=SL24n3HxN Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Sun, 17 Jul 2005 23:43:29 -0600), Robert Creager confessed: > I've 6 runs going concurrently. Just saw (vmstat 1) a set of 8 seconds w= here > the CS didn't drop below 90k, but right now its at ~300 for over 30 secon= ds... > It's bouncing all over the place, but staying reasonably well behaved ove= rall. >=20 Against 741 and the same load, CS is steady around 300 with spikes up to 4k= , but it's only been running for about 15 minutes. All queries are < .2 seconds. Cheers, Rob --=20 00:03:33 up 4 days, 3:38, 6 users, load average: 1.67, 0.98, 0.44 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Mon__18_Jul_2005_00_07_13_-0600_yzEq77=SL24n3HxN Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbRxEACgkQLQ/DKuwDYzkjjQCgggOf7Bx5Bg+btMrJQ5Cr2EQB rXYAni1EOPZDVsHyTUz7itBv0htio6jE =VZSz -----END PGP SIGNATURE----- --Signature_Mon__18_Jul_2005_00_07_13_-0600_yzEq77=SL24n3HxN-- From pgsql-performance-owner@postgresql.org Mon Jul 18 04:51:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3064C52C80 for ; Mon, 18 Jul 2005 04:51:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20466-08 for ; Mon, 18 Jul 2005 07:51:41 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.200]) by svr1.postgresql.org (Postfix) with ESMTP id CB33952C7F for ; Mon, 18 Jul 2005 04:51:39 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id i11so818446nzh for ; Mon, 18 Jul 2005 00:51:41 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=Sd+AwLmS/DrRJCR9gMHYldjQyCHPHwtwJihIExzVpY3aIpAPAoHMZ/CnweHLU9Ecek41TcKQ59XXNH8bEDurlZpwGUpWFDobjHrmRKkhmC/HVUQsasoUWwImVMlrOV5oW9ae5f4xih9oYgStlBjEieX4jgPHP3GCQu0LUed3aLg= Received: by 10.36.220.67 with SMTP id s67mr3790563nzg; Mon, 18 Jul 2005 00:51:41 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Mon, 18 Jul 2005 00:51:41 -0700 (PDT) Message-ID: <758d5e7f050718005164cbe9a4@mail.gmail.com> Date: Mon, 18 Jul 2005 09:51:41 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Bruno Wolff III , Dan Harris , pgsql-performance@postgresql.org Subject: Re: slow joining very large table to smaller ones In-Reply-To: <20050715164458.GA26178@wolff.to> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <46AF8F35-B735-4DFA-BA0A-AC82B5DA3CCE@drivefaster.net> <42D687E5.7080202@arbash-meinel.com> <20050715164458.GA26178@wolff.to> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.614 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/255 X-Sequence-Number: 13496 On 7/15/05, Bruno Wolff III wrote: > On Thu, Jul 14, 2005 at 16:29:58 -0600, > Dan Harris wrote: > > > > Ok, I tried this one. My ssh keeps getting cut off by a router > > somewhere between me and the server due to inactivity timeouts, so > > all I know is that both the select and explain analyze are taking > > over an hour to run. Here's the explain select for that one, since > > that's the best I can get. >=20 > Are you using NAT at home? That's probably where the issue is. If you > have control of that box you can probably increase the timeout to a > couple of hours. Some versions of ssh have such a configuration option (in .ssh/config): Host * ServerAliveInterval 600 ...it means that ssh will send a "ping" packet to a sshd every 10 minutes of inactivity. This way NAT will see activity and won't kill the session. I'm using OpenSSH_4.1p1 for this... Oh, and it doesn't have anything to do with TCP keep alive, which is rather for finding dead connections than keeping connections alive. ;) Regards, Dawid From pgsql-performance-owner@postgresql.org Mon Jul 18 10:23:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7A6E852C8E for ; Mon, 18 Jul 2005 10:23:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 01371-04 for ; Mon, 18 Jul 2005 13:23:12 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 0950A52C7F for ; Mon, 18 Jul 2005 10:23:11 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6IDNBGE002467; Mon, 18 Jul 2005 09:23:11 -0400 (EDT) To: Robert Creager Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS In-reply-to: <20050717235630.26a7c979@thunder.logicalchaos.org> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> <20050717235630.26a7c979@thunder.logicalchaos.org> Comments: In-reply-to Robert Creager message dated "Sun, 17 Jul 2005 23:56:30 -0600" Date: Mon, 18 Jul 2005 09:23:11 -0400 Message-ID: <2466.1121692991@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/256 X-Sequence-Number: 13497 Robert Creager writes: > Tom Lane confessed: >> The context swap problem was no worse in 8.0 than in prior versions, >> so that hardly seems like a good explanation. Have you tried reverting >> to the cron-based vacuuming method you used in 7.4? > I've "vacuum_cost_delay = 10" in the conf file for 803. Hmm, did you read this thread? http://archives.postgresql.org/pgsql-performance/2005-07/msg00088.php It's still far from clear what's going on there, but it might be interesting to see if turning off the vacuum delay changes your results with 8.0. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 10:41:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6EA7A52A3C for ; Mon, 18 Jul 2005 10:41:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 03547-07 for ; Mon, 18 Jul 2005 13:41:05 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id DA43D52CA9 for ; Mon, 18 Jul 2005 10:41:03 -0300 (ADT) Received: (qmail 32208 invoked by uid 1010); 18 Jul 2005 09:41:06 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 4.074218 secs); 18 Jul 2005 13:41:06 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 09:41:02 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id B5F0EB32C5; Mon, 18 Jul 2005 07:41:00 -0600 (MDT) Date: Mon, 18 Jul 2005 07:41:00 -0600 From: Robert Creager To: Tom Lane Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050718074100.3159e825@thunder.logicalchaos.org> In-Reply-To: <24520.1121659853@sss.pgh.pa.us> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary=Signature_Mon__18_Jul_2005_07_41_00_-0600_qIUf4mwaPMLUfx2w; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/257 X-Sequence-Number: 13498 --Signature_Mon__18_Jul_2005_07_41_00_-0600_qIUf4mwaPMLUfx2w Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 00:10:53 -0400), Tom Lane confessed: > The context swap problem was no worse in 8.0 than in prior versions, > so that hardly seems like a good explanation. Have you tried reverting > to the cron-based vacuuming method you used in 7.4? >=20 Ran 7 hours on 741 with VACUUM ANALYZE every 5 minutes. The largest CS I s= aw was 40k, with an average of 500 (via script which monitors vmstat output). I've done a VACUUM FULL ANALYZE on 803 and have switched the cron based VAC= UUM ANALYZE to 803 also. The tests are now running again. > Hmm, did you read this thread? > http://archives.postgresql.org/pgsql-performance/2005-07/msg00088.php I just glanced at it. Once I've reproduced (or not) the problem on 803 wit= h the VACUUM FULL, I'll turn off the vacuum delay. Cheers, Rob --=20 07:10:06 up 4 days, 10:45, 6 users, load average: 0.28, 0.40, 0.29 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Mon__18_Jul_2005_07_41_00_-0600_qIUf4mwaPMLUfx2w Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbsWwACgkQLQ/DKuwDYzktbQCfZ39U38nwHZ/2Kme20kghS5fe 9F8AnigtFAQ1h9A/WQn2V53aWivWiMNR =MFbd -----END PGP SIGNATURE----- --Signature_Mon__18_Jul_2005_07_41_00_-0600_qIUf4mwaPMLUfx2w-- From pgsql-performance-owner@postgresql.org Mon Jul 18 10:47:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 50E2452CA9 for ; Mon, 18 Jul 2005 10:47:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07276-04 for ; Mon, 18 Jul 2005 13:47:45 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id 75BEE52C92 for ; Mon, 18 Jul 2005 10:47:44 -0300 (ADT) Received: (qmail 2307 invoked by uid 1010); 18 Jul 2005 09:47:48 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.930243 secs); 18 Jul 2005 13:47:48 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 18 Jul 2005 09:47:46 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id 3FD5EB3281; Mon, 18 Jul 2005 07:47:45 -0600 (MDT) Date: Mon, 18 Jul 2005 07:47:44 -0600 From: Robert Creager To: Tom Lane Cc: Robert Creager , PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050718074744.03afadc3@thunder.logicalchaos.org> In-Reply-To: <2466.1121692991@sss.pgh.pa.us> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> <20050717235630.26a7c979@thunder.logicalchaos.org> <2466.1121692991@sss.pgh.pa.us> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary=Signature_Mon__18_Jul_2005_07_47_44_-0600_XnmEX04qaPa4tVZp; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/258 X-Sequence-Number: 13499 --Signature_Mon__18_Jul_2005_07_47_44_-0600_XnmEX04qaPa4tVZp Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Mon, 18 Jul 2005 09:23:11 -0400), Tom Lane confessed: > It's still far from clear what's going on there, but it might be > interesting to see if turning off the vacuum delay changes your results > with 8.0. >=20 Can that be affected by hupping the server, or do I need a restart? Thanks, Rob --=20 07:46:53 up 4 days, 11:21, 6 users, load average: 0.77, 0.43, 0.27 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Mon__18_Jul_2005_07_47_44_-0600_XnmEX04qaPa4tVZp Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLbswAACgkQLQ/DKuwDYznL2gCfThEJTfYMCMQIovWPcVxUbHSe PQ8An2/1naVwenNBHzQSc2JuyT/HkaK8 =Cs86 -----END PGP SIGNATURE----- --Signature_Mon__18_Jul_2005_07_47_44_-0600_XnmEX04qaPa4tVZp-- From pgsql-performance-owner@postgresql.org Mon Jul 18 11:03:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AA47B52A3C for ; Mon, 18 Jul 2005 11:03:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 32594-01 for ; Mon, 18 Jul 2005 14:03:53 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E660852CAB for ; Mon, 18 Jul 2005 11:03:52 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6IE3sZt002899; Mon, 18 Jul 2005 10:03:54 -0400 (EDT) To: Robert Creager Cc: PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS In-reply-to: <20050718074744.03afadc3@thunder.logicalchaos.org> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> <20050717235630.26a7c979@thunder.logicalchaos.org> <2466.1121692991@sss.pgh.pa.us> <20050718074744.03afadc3@thunder.logicalchaos.org> Comments: In-reply-to Robert Creager message dated "Mon, 18 Jul 2005 07:47:44 -0600" Date: Mon, 18 Jul 2005 10:03:54 -0400 Message-ID: <2898.1121695434@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/259 X-Sequence-Number: 13500 Robert Creager writes: > Tom Lane confessed: >> It's still far from clear what's going on there, but it might be >> interesting to see if turning off the vacuum delay changes your results >> with 8.0. > Can that be affected by hupping the server, or do I need a restart? sighup should be fine. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 11:31:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D064B52CB4 for ; Mon, 18 Jul 2005 11:31:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 43273-05 for ; Mon, 18 Jul 2005 14:31:47 +0000 (GMT) Received: from outbound.mailhop.org (outbound.mailhop.org [63.208.196.171]) by svr1.postgresql.org (Postfix) with ESMTP id 3D7FC52CB0 for ; Mon, 18 Jul 2005 11:31:46 -0300 (ADT) Received: from ool-4350c7ad.dyn.optonline.net ([67.80.199.173] helo=[192.168.0.66]) by outbound.mailhop.org with esmtpsa (TLSv1:AES256-SHA:256) (Exim 4.51) id 1DuWer-000Lcw-3O; Mon, 18 Jul 2005 10:31:49 -0400 X-Mail-Handler: MailHop Outbound by DynDNS.org X-Originating-IP: 67.80.199.173 X-Report-Abuse-To: abuse@dyndns.org (see http://www.mailhop.org/outbound/abuse.html for abuse reporting information) X-MHO-User: zeut Message-ID: <42DBBD4A.6030209@zeut.net> Date: Mon, 18 Jul 2005 10:31:38 -0400 From: "Matthew T. O'Connor" Organization: Terrie O'Connor Realtors User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: Robert Creager , PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> <20050717235630.26a7c979@thunder.logicalchaos.org> <2466.1121692991@sss.pgh.pa.us> In-Reply-To: <2466.1121692991@sss.pgh.pa.us> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.003 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/260 X-Sequence-Number: 13501 Tom Lane wrote: >Robert Creager writes: > > >>I've "vacuum_cost_delay = 10" in the conf file for 803. >> >> > >Hmm, did you read this thread? >http://archives.postgresql.org/pgsql-performance/2005-07/msg00088.php > >It's still far from clear what's going on there, but it might be >interesting to see if turning off the vacuum delay changes your results >with 8.0. > With the contrib autovacuum code if you don't specify vacuum delay settings from the command line, then autovacuum doesn't touch them. Therefore (if you aren't specifying them from the command line), on 803, the vacuum delay settings should be the same for a cron issued vacuum and an autovacuum issued vacuum. So if the vacuum delay settings are the problem, then it should show up either way. Matt From pgsql-performance-owner@postgresql.org Mon Jul 18 11:53:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 779A152A3C for ; Mon, 18 Jul 2005 11:53:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 66483-04 for ; Mon, 18 Jul 2005 14:53:00 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 788F752CB2 for ; Mon, 18 Jul 2005 11:53:00 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6IEr43g003284; Mon, 18 Jul 2005 10:53:04 -0400 (EDT) To: "Matthew T. O'Connor" Cc: Robert Creager , PGPerformance , Andy Hewitt Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - In-reply-to: <42DBBD4A.6030209@zeut.net> References: <20050717213416.59dcb52e@thunder.logicalchaos.org> <24520.1121659853@sss.pgh.pa.us> <20050717235630.26a7c979@thunder.logicalchaos.org> <2466.1121692991@sss.pgh.pa.us> <42DBBD4A.6030209@zeut.net> Comments: In-reply-to "Matthew T. O'Connor" message dated "Mon, 18 Jul 2005 10:31:38 -0400" Date: Mon, 18 Jul 2005 10:53:04 -0400 Message-ID: <3283.1121698384@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/261 X-Sequence-Number: 13502 "Matthew T. O'Connor" writes: > Therefore (if you aren't specifying them from the command line), on 803, > the vacuum delay settings should be the same for a cron issued vacuum > and an autovacuum issued vacuum. So if the vacuum delay settings are > the problem, then it should show up either way. ... as indeed it does according to Robert's recent reports. Still awaiting the definitive test, but I'm starting to think this is another case of the strange behavior Ian Westmacott exhibited. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 14:50:14 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F1C22529B2 for ; Mon, 18 Jul 2005 14:50:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 12107-09 for ; Mon, 18 Jul 2005 17:50:04 +0000 (GMT) Received: from prospect.stortek.com (prospect.stortek.com [129.80.22.147]) by svr1.postgresql.org (Postfix) with ESMTP id ADE1252896 for ; Mon, 18 Jul 2005 14:50:02 -0300 (ADT) Received: from prospect.stortek.com (localhost [127.0.0.1]) by prospect.stortek.com (8.12.10/8.12.0) with ESMTP id j6IHo1Wl013926 for ; Mon, 18 Jul 2005 11:50:02 -0600 (MDT) Received: from unix.stortek.com (burma.stortek.com [129.80.16.110]) by prospect.stortek.com (8.12.10/8.12.0) with ESMTP id j6IHo1bc013921; Mon, 18 Jul 2005 11:50:01 -0600 (MDT) Received: from C118181.stortek.com (c118181.stortek.com [129.80.81.172]) by unix.stortek.com (8.12.10/8.12.0) with ESMTP id j6IHo0vn027391; Mon, 18 Jul 2005 11:50:00 -0600 (MDT) Date: Mon, 18 Jul 2005 11:49:52 -0600 From: Robert Creager To: tgl@sss.pgh.pa.us, matthew@zeut.net, pgsql-performance@postgresql.org Cc: Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? Message-ID: <20050718114952.00004ac5@C118181.stortek.com> Organization: StorageTek X-Mailer: Sylpheed-Claws 1.0.4.1 Win32 (GTK+ 1.3.0; Win32) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/262 X-Sequence-Number: 13503 In regards to http://archives.postgresql.org/pgsql-performance/2005-07/msg00261.php Tom Says: > ... as indeed it does according to Robert's recent reports. Still > awaiting the definitive test, but I'm starting to think this is another > case of the strange behavior Ian Westmacott exhibited. Ok. This morning at around 7:30am I started tests against a freshly VACUUM FULL ANALYZE 803 database with the vacuum delay on and cron running vacuum analyze every 5 minutes. Around 8:15 I was starting to receive hits of a few seconds of high CS hits, higher than the previous 7 hour run on 741. I changed the vacuum delay to 0 and HUP'ed the server (how can I see the value vacuum_cost_delay run time?). By 10:30, I had vacuum jobs backed up since 9:20 and the queries were over 75 seconds. I'm currently running on 741 as I need to get work done today ;-) I'll restart the 803 db, vacuum full analyze again and next opportunity (maybe tonight), start runs again with cron vacuum and a vacuum_cost_delay of 0, unless I should try something else? Cheers, Rob -- Robert Creager Advisory Software Engineer Phone 303.673.2365 Pager 888.912.4458 Fax 303.661.5379 StorageTek From pgsql-performance-owner@postgresql.org Mon Jul 18 14:52:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8236952B4C for ; Mon, 18 Jul 2005 14:52:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 15692-02 for ; Mon, 18 Jul 2005 17:52:55 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6F7A952B24 for ; Mon, 18 Jul 2005 14:52:54 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6IHqrKY015473; Mon, 18 Jul 2005 13:52:53 -0400 (EDT) To: Robert Creager Cc: matthew@zeut.net, pgsql-performance@postgresql.org, Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? In-reply-to: <20050718114952.00004ac5@C118181.stortek.com> References: <20050718114952.00004ac5@C118181.stortek.com> Comments: In-reply-to Robert Creager message dated "Mon, 18 Jul 2005 11:49:52 -0600" Date: Mon, 18 Jul 2005 13:52:53 -0400 Message-ID: <15472.1121709173@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/263 X-Sequence-Number: 13504 Robert Creager writes: > Around 8:15 I was starting to receive hits of a few seconds of high CS hits, > higher than the previous 7 hour run on 741. I changed the vacuum delay to 0 and > HUP'ed the server (how can I see the value vacuum_cost_delay run > time?). Start a fresh psql session and "SHOW vacuum_cost_delay" to verify what the active setting is. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 14:57:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5748D529B2 for ; Mon, 18 Jul 2005 14:57:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 14526-08 for ; Mon, 18 Jul 2005 17:57:37 +0000 (GMT) Received: from gwmta.wicourts.gov (gwmta.wicourts.gov [165.219.244.91]) by svr1.postgresql.org (Postfix) with ESMTP id 1F10E52B11 for ; Mon, 18 Jul 2005 14:57:36 -0300 (ADT) Received: from Courts-MTA by gwmta.wicourts.gov with Novell_GroupWise; Mon, 18 Jul 2005 12:57:35 -0500 Message-Id: X-Mailer: Novell GroupWise Internet Agent 6.5.2 Date: Mon, 18 Jul 2005 12:57:31 -0500 From: "Kevin Grittner" To: , Subject: Re: join and query planner Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.019 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/264 X-Sequence-Number: 13505 Just out of curiosity, does it do any better with the following? SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) >>> "Dario Pudlo" 07/06/05 4:54 PM >>> (first at all, sorry for my english) Hi. - Does "left join" restrict the order in which the planner must join tables? I've read about join, but i'm not sure about left join... - If so: Can I avoid this behavior? I mean, make the planner resolve the query, using statistics (uniqueness, data distribution) rather than join order. My query looks like: SELECT ... FROM a, b, LEFT JOIN c ON (c.key = a.key) LEFT JOIN d on (d.key=a.key) WHERE (a.key = b.key) AND (b.column <= 100) b.column has a lot better selectivity, but planner insist on resolve first c.key = a.key. Of course, I could rewrite something like: SELECT ... FROM (SELECT ... FROM a,b LEFT JOIN d on (d.key=a.key) WHERE (b.column <= 100) ) as aa LEFT JOIN c ON (c.key = aa.key) but this is query is constructed by an application with a "multicolumn" filter. It's dynamic. It means that a user could choose to look for "c.column = 1000". And also, combinations of filters. So, I need the planner to choose the best plan... I've already change statistics, I clustered tables with cluster, ran vacuum analyze, changed work_mem, shared_buffers... Greetings. TIA. ---------------------------(end of broadcast)--------------------------- TIP 2: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Mon Jul 18 15:21:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CB26052BD2 for ; Mon, 18 Jul 2005 15:21:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 23253-01 for ; Mon, 18 Jul 2005 18:21:14 +0000 (GMT) Received: from sherman.stortek.com (sherman.stortek.com [129.80.22.146]) by svr1.postgresql.org (Postfix) with ESMTP id DE9A452C26 for ; Mon, 18 Jul 2005 15:21:12 -0300 (ADT) Received: from sherman.stortek.com (localhost [127.0.0.1]) by sherman.stortek.com (8.12.10/8.12.0) with ESMTP id j6IILDD4000405 for ; Mon, 18 Jul 2005 12:21:13 -0600 (MDT) Received: from unix.stortek.com (burma.stortek.com [129.80.16.110]) by sherman.stortek.com (8.12.10/8.12.0) with ESMTP id j6IILCDC000402; Mon, 18 Jul 2005 12:21:12 -0600 (MDT) Received: from C118181.stortek.com (c118181.stortek.com [129.80.81.172]) by unix.stortek.com (8.12.10/8.12.0) with ESMTP id j6IILBvn028899; Mon, 18 Jul 2005 12:21:11 -0600 (MDT) Date: Mon, 18 Jul 2005 12:21:03 -0600 From: Robert Creager To: Tom Lane Cc: matthew@zeut.net, pgsql-performance@postgresql.org, Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? Message-ID: <20050718122103.0000064a@C118181.stortek.com> In-Reply-To: <15472.1121709173@sss.pgh.pa.us> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> Organization: StorageTek X-Mailer: Sylpheed-Claws 1.0.4.1 Win32 (GTK+ 1.3.0; Win32) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/265 X-Sequence-Number: 13506 On Mon, 18 Jul 2005 13:52:53 -0400 Tom Lane wrote: > Start a fresh psql session and "SHOW vacuum_cost_delay" to verify what > the active setting is. Thanks. It does show 0 for 803 in a session that was up since I thought I had HUPed the server with the new value. This is leading me to believe that 803 doesn't do very well with VACUUM ANALYZE running often, at least in my particular application... I will provide a more definitive statement to that affect, hopefully tonight. Cheers, Rob From pgsql-performance-owner@postgresql.org Mon Jul 18 15:32:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 66A97529A9 for ; Mon, 18 Jul 2005 15:32:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25086-01 for ; Mon, 18 Jul 2005 18:32:17 +0000 (GMT) Received: from yertle.kcilink.com (yertle.kcilink.com [65.205.34.180]) by svr1.postgresql.org (Postfix) with ESMTP id 9C59F52996 for ; Mon, 18 Jul 2005 15:32:15 -0300 (ADT) Received: from [192.168.7.103] (host-103.int.kcilink.com [192.168.7.103]) by yertle.kcilink.com (Postfix) with ESMTP id EDD4DB824 for ; Mon, 18 Jul 2005 14:32:14 -0400 (EDT) Mime-Version: 1.0 (Apple Message framework v733) In-Reply-To: <59d991c4050717100874c08a86@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> Content-Type: multipart/signed; micalg=sha1; boundary=Apple-Mail-2--477069979; protocol="application/pkcs7-signature" Message-Id: From: Vivek Khera Subject: Re: Impact of checkpoint_segments under continual load conditions Date: Mon, 18 Jul 2005 14:32:14 -0400 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.733) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.022 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/266 X-Sequence-Number: 13507 --Apple-Mail-2--477069979 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed On Jul 17, 2005, at 1:08 PM, Christopher Petrilli wrote: > Normally, checkpoint_segments can help absorb some of that, but my > experience is that if I crank the number up, it simply delays the > impact, and when it occurs, it takes a VERY long time (minutes) to > clear. There comes a point where your only recourse is to throw hardware at the problem. I would suspect that getting faster disks and splitting the checkpoint log to its own RAID partition would help you here. Adding more RAM while you're at it always does wonders for me :-) Vivek Khera, Ph.D. +1-301-869-4449 x806 --Apple-Mail-2--477069979 Content-Transfer-Encoding: base64 Content-Type: application/pkcs7-signature; name=smime.p7s Content-Disposition: attachment; filename=smime.p7s MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIGhzCCAz8w ggKooAMCAQICAQ0wDQYJKoZIhvcNAQEFBQAwgdExCzAJBgNVBAYTAlpBMRUwEwYDVQQIEwxXZXN0 ZXJuIENhcGUxEjAQBgNVBAcTCUNhcGUgVG93bjEaMBgGA1UEChMRVGhhd3RlIENvbnN1bHRpbmcx KDAmBgNVBAsTH0NlcnRpZmljYXRpb24gU2VydmljZXMgRGl2aXNpb24xJDAiBgNVBAMTG1RoYXd0 ZSBQZXJzb25hbCBGcmVlbWFpbCBDQTErMCkGCSqGSIb3DQEJARYccGVyc29uYWwtZnJlZW1haWxA dGhhd3RlLmNvbTAeFw0wMzA3MTcwMDAwMDBaFw0xMzA3MTYyMzU5NTlaMGIxCzAJBgNVBAYTAlpB MSUwIwYDVQQKExxUaGF3dGUgQ29uc3VsdGluZyAoUHR5KSBMdGQuMSwwKgYDVQQDEyNUaGF3dGUg UGVyc29uYWwgRnJlZW1haWwgSXNzdWluZyBDQTCBnzANBgkqhkiG9w0BAQEFAAOBjQAwgYkCgYEA xKY8VXNV+065yplaHmjAdQRwnd/p/6Me7L3N9VvyGna9fww6YfK/Uc4B1OVQCjDXAmNaLIkVcI7d yfArhVqqP3FWy688Cwfn8R+RNiQqE88r1fOCdz0Dviv+uxg+B79AgAJk16emu59l0cUqVIUPSAR/ p7bRPGEEQB5kGXJgt/sCAwEAAaOBlDCBkTASBgNVHRMBAf8ECDAGAQH/AgEAMEMGA1UdHwQ8MDow OKA2oDSGMmh0dHA6Ly9jcmwudGhhd3RlLmNvbS9UaGF3dGVQZXJzb25hbEZyZWVtYWlsQ0EuY3Js MAsGA1UdDwQEAwIBBjApBgNVHREEIjAgpB4wHDEaMBgGA1UEAxMRUHJpdmF0ZUxhYmVsMi0xMzgw DQYJKoZIhvcNAQEFBQADgYEASIzRUIPqCy7MDaNmrGcPf6+svsIXoUOWlJ1/TCG4+DYfqi2fNi/A 9BxQIJNwPP2t4WFiw9k6GX6EsZkbAMUaC4J0niVQlGLH2ydxVyWN3amcOY6MIE9lX5Xa9/eH1sYI Tq726jTlEBpbNU1341YheILcIRk13iSx0x1G/11fZU8wggNAMIICqaADAgECAgMOah8wDQYJKoZI hvcNAQEEBQAwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkp IEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFpbCBJc3N1aW5nIENBMB4XDTA1 MDQwNTIwMzEzMloXDTA2MDQwNTIwMzEzMlowgYoxHzAdBgNVBAMTFlRoYXd0ZSBGcmVlbWFpbCBN ZW1iZXIxHjAcBgkqhkiG9w0BCQEWD3ZpdmVrQGtoZXJhLm9yZzEgMB4GCSqGSIb3DQEJARYRa2hl cmFAa2NpbGluay5jb20xJTAjBgkqhkiG9w0BCQEWFnZpdmVrQG1haWxlcm1haWxlci5jb20wggEi MA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQDdo7hywGcY0VvK6WqqXXV77MS/t/4X3WkCaCXo RSl2W58GP4P21hodPn7hlIxUoDOW7x9O+FbqTgE2Ejqr6yA00Mm90tGPFgjFjqPGAqg7xk6IDcv9 uTyMia/FKEHSIynM6zqokXY8JklvdbJOiByE/8VeyEXOANWiflo8o4+GHnhMKpA9982YTXUqeKU6 mMQVaLCBRjTDc7j2XkMC/UNcp2HMyDQdTqYVnhLxbvbLX8CNDBY/7OWFlB9evru46SpGWhe4lhv5 DSgE2RdCKvDytzxRDvP49L8V0TnFjAVeC1C1Pj0/KQsoL/AP4APplROiD4QaUhshQl28pXxJtfbl AgMBAAGjVzBVMEUGA1UdEQQ+MDyBD3ZpdmVrQGtoZXJhLm9yZ4ERa2hlcmFAa2NpbGluay5jb22B FnZpdmVrQG1haWxlcm1haWxlci5jb20wDAYDVR0TAQH/BAIwADANBgkqhkiG9w0BAQQFAAOBgQAr CWop3h28qPwofzLrkoT410J4d7Bqk6FLeVlKZfg/wXlS1MTqYMNcCm4x+JsJbjwsO0fb2elFIuGq 1razoSzPpgi89itydvUT0U0U/u+AkZA5rW4AptTpMZ70YW5u9wzkcvmifqZmcfbaaeGdZfruzUXZ 6qvdXDpNb3ZHeQw6PjGCAucwggLjAgEBMGkwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0 ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFp bCBJc3N1aW5nIENBAgMOah8wCQYFKw4DAhoFAKCCAVMwGAYJKoZIhvcNAQkDMQsGCSqGSIb3DQEH ATAcBgkqhkiG9w0BCQUxDxcNMDUwNzE4MTgzMjE0WjAjBgkqhkiG9w0BCQQxFgQUalSNHAl1c1Lw jPY1i+SwOiUmChIweAYJKwYBBAGCNxAEMWswaTBiMQswCQYDVQQGEwJaQTElMCMGA1UEChMcVGhh d3RlIENvbnN1bHRpbmcgKFB0eSkgTHRkLjEsMCoGA1UEAxMjVGhhd3RlIFBlcnNvbmFsIEZyZWVt YWlsIElzc3VpbmcgQ0ECAw5qHzB6BgsqhkiG9w0BCRACCzFroGkwYjELMAkGA1UEBhMCWkExJTAj BgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJz b25hbCBGcmVlbWFpbCBJc3N1aW5nIENBAgMOah8wDQYJKoZIhvcNAQEBBQAEggEALNRgQiUwFLxt ZHWxekLhjdzdHk1TLMEeKD889yVGFeCMN1dV8jwFg9DLmo0iPDlD/jZo/r/tjXSwEzXFbEjfBNPI 3BHt6wrqO7w6JMT5I41ZSsfWH0jfrO6Vo8zl9rDRvioV3YD6gzjsyH5Yf+047vrSaEKMsmphAPEc SrgDMYHy04L0ehf0dXQYs/Pf0WvqAkmVATocYCkeRtZoWAgbx2n5l6Ybwk1VsaGSQJZJW0JkNQv8 pHqx+UZGfJuISQCeERTy22Tq2TSnZzDJ2xhiJC1espe0UbTDR/+zcJ51ZjO5bJI8wqiD60q9X5K5 WAIMK405IP/H9TCZaloRFVTOeQAAAAAAAA== --Apple-Mail-2--477069979-- From pgsql-performance-owner@postgresql.org Mon Jul 18 15:46:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A852C52BD1 for ; Mon, 18 Jul 2005 15:46:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 23015-07 for ; Mon, 18 Jul 2005 18:46:24 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.203]) by svr1.postgresql.org (Postfix) with ESMTP id 0955152996 for ; Mon, 18 Jul 2005 15:46:23 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i35so1052059wra for ; Mon, 18 Jul 2005 11:46:24 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=fzdyxgwM0EHwlpwkiVXl4cVEn56OlccDYj6X+HmyIRR4ok1+k3IEIf1Z/zrq7qLN5vX1jkLAToOdc8qFKZAgduwmTHpZx6+WMJVvYZ1By4a1WAXT0OwexG6ZWrDqWtyyevp9yEXodPGFFUW8dq7dwCotazZm+P3MnyQhxUyS36E= Received: by 10.54.37.47 with SMTP id k47mr460663wrk; Mon, 18 Jul 2005 11:45:37 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Mon, 18 Jul 2005 11:45:35 -0700 (PDT) Message-ID: <59d991c405071811457ccdd3f1@mail.gmail.com> Date: Mon, 18 Jul 2005 14:45:35 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Vivek Khera Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.126 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/267 X-Sequence-Number: 13508 On 7/18/05, Vivek Khera wrote: >=20 > On Jul 17, 2005, at 1:08 PM, Christopher Petrilli wrote: >=20 > > Normally, checkpoint_segments can help absorb some of that, but my > > experience is that if I crank the number up, it simply delays the > > impact, and when it occurs, it takes a VERY long time (minutes) to > > clear. >=20 > There comes a point where your only recourse is to throw hardware at > the problem. I would suspect that getting faster disks and splitting > the checkpoint log to its own RAID partition would help you here. > Adding more RAM while you're at it always does wonders for me :-) My concern is less with absolute performance, than with the nosedive it goes into. I published some of my earlier findings and comparisons on my blog, but there's a graph here: http://blog.amber.org/diagrams/comparison_mysql_pgsql.png Notice the VERY steep drop off. I'm still trying to get rid of it, but honestly, am not smart enough to know where it's originating. I have no desire to ever use MySQL, but it is a reference point, and since I don't particularly need transactional integrity, a valid comparison. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Mon Jul 18 16:24:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 281C852CDC for ; Mon, 18 Jul 2005 16:24:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 33161-08 for ; Mon, 18 Jul 2005 19:24:24 +0000 (GMT) Received: from netbox.unitech.com.ar (unknown [200.32.92.34]) by svr1.postgresql.org (Postfix) with ESMTP id C37DD52CD7 for ; Mon, 18 Jul 2005 16:24:21 -0300 (ADT) Received: from dariop (dariop.unitech.com.ar [192.168.1.237]) by netbox.unitech.com.ar (8.11.6/8.11.6) with ESMTP id j6IKOhE17938 for ; Mon, 18 Jul 2005 16:24:43 -0400 Received: from 127.0.0.1 (AVG SMTP 7.0.323 [267.9.0]); Mon, 18 Jul 2005 16:24:19 -0300 From: "Dario" To: Subject: Re: join and query planner Date: Mon, 18 Jul 2005 16:24:19 -0300 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) In-Reply-To: X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4952.2800 Importance: Normal X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.104 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/268 X-Sequence-Number: 13509 Hi. > Just out of curiosity, does it do any better with the following? > > SELECT ... Yes, it does. But my query could also be SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) /*new*/ , e WHERE (b.column <= 100) /*new*/ and (e.key = a.key) and (e.field = 'filter') because it's constructed by an application. I needed to know if, somehow, someway, I can "unforce" join order. The only way to solve it so far is changing application. It must build something like SELECT ... FROM b JOIN (a JOIN e ON (e.key = a.key)) ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) and (e.field = 'filter') Supossed that e.field has (should have) better selectivity. But now this problem belongs to programmer's group :-) The query, in fact, has more tables to join. I wonder if lowering geqo threshold could do the work... Thank you. Greetings. Long life, little spam and prosperity! -----Mensaje original----- De: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org]En nombre de Kevin Grittner Enviado el: lunes, 18 de julio de 2005 14:58 Para: pgsql-performance@postgresql.org; dariop@unitech.com.ar Asunto: Re: [PERFORM] join and query planner Just out of curiosity, does it do any better with the following? SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) >>> snipp From pgsql-performance-owner@postgresql.org Mon Jul 18 16:29:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D2FA952CC6 for ; Mon, 18 Jul 2005 16:29:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36518-02 for ; Mon, 18 Jul 2005 19:29:23 +0000 (GMT) Received: from apate.telenet-ops.be (apate.telenet-ops.be [195.130.132.57]) by svr1.postgresql.org (Postfix) with ESMTP id 5C33652CBB for ; Mon, 18 Jul 2005 16:29:21 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by apate.telenet-ops.be (Postfix) with SMTP id 8831A381ED for ; Mon, 18 Jul 2005 21:29:21 +0200 (CEST) Received: from [10.0.1.2] (d5152B1F3.access.telenet.be [81.82.177.243]) by apate.telenet-ops.be (Postfix) with ESMTP id 02ADD38123 for ; Mon, 18 Jul 2005 21:29:21 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v622) To: pgsql-performance@postgresql.org Message-Id: Content-Type: multipart/mixed; boundary=Apple-Mail-57--473644005 From: Yves Vindevogel Subject: Insert performance (OT?) Date: Mon, 18 Jul 2005 21:29:20 +0200 X-Mailer: Apple Mail (2.622) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/269 X-Sequence-Number: 13510 --Apple-Mail-57--473644005 Content-Type: multipart/alternative; boundary=Apple-Mail-58--473644004 --Apple-Mail-58--473644004 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed Hi, Suppose I have a table with 4 fields (f1, f2, f3, f4) I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) I have 3 records A, B, C, D (this will be inserted) A, B, C, E (this will pass u2, but not u1, thus not inserted) A, B, F, D (this will pass u1, but not u2, thus not inserted) Now, for performance ... I have tables like this with 500.000 records where there's a new upload=20= of approx. 20.000 records. It is only now that we say index u2 to be necessary. So, until now, I=20= did something like insert into ... select f1, f2, f2, max(f4) group by=20= f1, f2, f3 That is ok ... and also logically ok because of the data definition I cannot do this with 2 group by's. I tried this on paper and I'm not=20= succeeding. So, I must use a function that will check against u1 and u2, and then=20 insert if it is ok. I know that such a function is way slower that my insert query. So, my question ... How can I keep the same performance, but also with the new index in=20 mind ??? Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements --Apple-Mail-58--473644004 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 Hi, Suppose I have a table with 4 fields (f1, f2, f3, f4) I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) I have 3 records A, B, C, D (this will be inserted) A, B, C, E (this will pass u2, but not u1, thus not inserted) A, B, F, D (this will pass u1, but not u2, thus not inserted) Now, for performance ... I have tables like this with 500.000 records where there's a new upload of approx. 20.000 records. It is only now that we say index u2 to be necessary. So, until now, I did something like insert into ... select f1, f2, f2, max(f4) group by f1, f2, f3 That is ok ... and also logically ok because of the data definition I cannot do this with 2 group by's. I tried this on paper and I'm not succeeding. So, I must use a function that will check against u1 and u2, and then insert if it is ok. I know that such a function is way slower that my insert query. So, my question ... How can I keep the same performance, but also with the new index in mind ??? Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements = --Apple-Mail-58--473644004-- --Apple-Mail-57--473644005 Content-Transfer-Encoding: base64 Content-Type: image/tiff; x-unix-mode=0666; name="Pasted Graphic 2.tiff" Content-Disposition: inline; filename="Pasted Graphic 2.tiff" TU0AKgAAFciAP6BP5/wWDQeEQmFQuGQ2HQ+FP5+v9rOh5P9IMBuP8zK9tP8fJNlP8QoNjP8FndhP 8MnyVjJHSMrqFnv9BL1vP9ett2v98PuJxChUOiUWGQOCUalUujOV4PZ/qBluR/mdVtJ/jVHst/ho 9MF/g08WABHJfv8AWa0WoAHFfWuzgg6sB/hY9ysUohjv8qqVqv9LMZxv9ouV40zEYmHUjFY2HPx+ RNtOl4Rhftt/phiRs6LC/iA5LewnJeWg46W22+02cAHOwa26bC0HPY7TZ7G2G9dv8EHFdP8Om9YP 8rKJov9UtFzv9Dzh/sRvu5/vZ9PzHdeF4zsUqgQRlOR5v9DMBwv8uqe/jZIVwNHu6B0/WAKHvWGr RAA1ra0G1c/vfgANrdraXrTQIAA6JW2TUrQNz/ja0oAjO4YAjSWS0DOWawjys4RkIk4Lj2sAWEUZ B/ieUJqH+PZdnAf5aG06R5nsfLtqW7UaoWZJwukwLyi4VRsn+F5HGYf8QJWBQ7JWAI5tYti2Dm1T bQHAq0De1Q2FwtA1Fof4AjVDIBDQV0vDQWK0DY/oADYWsrNQN81NOuDZtYOM6ycs4Cjq+Q9JWE5D pGJpOpqP5dMwW5sHUf53xlHCExupjqoIbh3Hu55vp6ch3Hof4wlMrAdEWYh/gmP1RtXK0DDi2rXy nVzXjpVs8NM1g1vuNU2gANEugGMBRwu4YADTNsvwyAA3v+0k6LRWLb2dBUpjk2I4NVOx/gYPC6BU QiwCgSsSlWZpxH+ahzsOaBzvCe7qoMpKlUghd1n2f5oHId5/lQaJ0n+RBf3GLpUGwf4iksrg0FVF ICzlXT9QBB0tWO/+FjlKVWtfijTNRBsvDbhrdP3NUAwY38KFfkc0PuNjRWLLz8ytiTUVdVlnjlA2 MWPAWRAAM8LAAMjhhuR5ir4ULjigT5pn+P5co2T6pn+YBup6dp6xoh6kHie95modNOFkah0H+Qhd pyKBNmcf68r2Bo9GGtdpwIAUotM1Q4QMNzUDZhuVZbLsvy7KkqZZCnAb4AFcQY1C3LW1Wb5lOa2t ZvUt78NNjcMN83wEOMDczvuTzXNQ3QEN01WRL2b8XBYEDuugQEFUYgkpIo5FjIJTmhfZkHG8J0nm fB/sifp/EQYLBioUKsBkRKVgWPLX7stA4VnaDa+tZlZel0j+gCMpWS8MRULQNeIDY/4zTIAIyFX0 GFybZnr+rZ8p2a1s6tUNfADXXI1Yb8jLX+MNQk+1Az7zWqwVa/FaJrHpvSLeAQOhdE/l7CSJwrAb BXDXH+PkyA/gUiPGaftASWW6JeDgnGAsCmLNvcYqlz5+FchnZMrqGbeE3JoYhAMAKbEvOGeigtaE CDZwJcgxtiML0uMgQYzllKt2GhtP/EaH6cgAm+fnCtVEVXFOKNU54AC1B/tyLfBx4YJBGwhAAG5m yUoGH/TgxlZ7nGXw3YdEpYTPWOvaSq/JVEX28rEDQsGKDoGPx+QYnE1URg2paZY4dy7EHTx9cc41 upqIfwkjSgJWyW0MgBdK9I1DGAAwnS8G8+4cGYyUWWbKMg/gSiOjQG+FMQ44qqlmgBNUSYYMohKW 2ULFZaR8NdMFucoz/w/ZcAAMzgH+j/AKGV8QAZkhmWNM0AJ/I5zCiJEGYjF1qoGmxI5MqwT7JVlG buLULJhPXlbK+WMt3HwHmIbF8xaA0n6mvPgN7EJhx9Vcax+hsX6wGZuldNDnTSgQDoloGohS6BYE 62gPQqyRhlFE2gIYkCRgdD4WcAUsnxvlYklKgUCn4upi29JkkpG/pbYgtKbr86TLOnbLCF066ZLP kGsI+7NY9rRjZHB+0ew4M5S0AKT4FA8l0BaIUk4XHkj/EYLwzAtVyj/GwRYf49B8D6H+OMdw9R/j DG8vcTgx1xhpFTBoHYilRgaDygQAobXCynVVSSVdRYTLJls5CcIaqX0zpw/Kms76c2DSnTsNNPaA TbpggqOSCg4MkQwcAPiBAzimJqI4Xg3R/itX0P8bY66xD7MiYkgg5h4vAF2NgdY/xNGCH+HQVMIQ WiCNVM2L6Ap4s3ShFixMubA2HnjcUulhab3GuIbGxVjLlxXZoawMKFgsCiaSOUd6nB6XbH+PofKN LtqcH7eMf5AyimQOtegf4/TIEFvYP8dyjTm2eV1KdmNgrlIKr/cOxF+LjmRldTZY88L/U6P/YtyF +b82PDAhkOQsyNkHu8jS76NB4YXH+PvDUG7v3lIFeu8mFbuj6q8PXE2HEaXqIEUEgxnh2T2YhD/B WBZcUunVjO59yMB3PsHc3BONLemxwaP8NwsTMEPvVeMieJKvXmHZk8f48h5GHyYP8e+V3g3tvMQs VY01FLCxjgTHk8b943yBTOVoJEiXJzPgae1zr+2HwYhnIuRzrjczwP8eY8zDmKy7l8NOYcx40zKW nNswizytg+kXHc/qS48x9o7QbNMhZ0yMdvPBG7wmNz/jB6WYs436uFmbHhbFrAFgjhl4Y/g5IvH+ 0EkYFw+NuAHqmoet9bWJwPnDHEfM55E0udjTNW7uZ+y9p6L+oNe3Bxtoa4lACzgQee2kQ5Jwo1RK QPQfK8xejaJ6FG6w/wEh1NZJ+L6ydoH+zfqTUK0dK7AzsY7Yem9jaA0Fu2wVf9eJObqW+TxuwbCS K4JEYVnh2O/vWQM5K+wtCehCCAP5YABybV035xCAJIJqnrTyFqB9H8e1Lu/OumM870MTp3MGn9J8 g5Zsxhk2ZURLfGmrBAEg5m/CISIf4jRekblaCgRoyS0BjWNZVw6uZ70hnsrkND/Fbn3kWWtmz1ID a5NxyLYJ1xujdI3iZTm9dkYy4/P9yFKIcFoDQsbtKW4nT2YbLrpr450V6zTGem7N+YzXVzDZ0PSl ddIb8GQVrlD9sQjy4FKHVy0ZDDeLHCPWuuD/68Y0Vg1LX8c0bQSBiAq9JrVv2pYwZkzq6mqxDi8y XJgBkjZGftMMdUgm5PKyCBnogBI72hYgZlgrD6WWgMzPZkxGdUzfIfIzsdb66PXr/Jxp+XwRGlOL GnuJm2b0l7vo/dJeDL4N7qwWN+qn5CqedNMATuzZmObaad1s4dBMl/6wmG/Pt1+oAAYk2hToiuQc 5FxyjyRoHqH0xYKE+Q8k+UIgH0KAH+HSHqXmGwHSIuEGF0I25eWE8A96/g2akwhIk8Puf25mpi3w /IeG/M8y0O10/WgM3cpUS8l0mS+ocOYgDQS0ASDUTOB6EaJWDkXyXwFyL+GqGkPKHQHMOkxIXmIO HAHAJyyuKgvGIIHgU2H+G0GuHKH+FwF6KwEAtAH+CUEcLOAiNC2QnKcOZSZCmO9ongNep2sA3Y2W nY/KwE9hDazc8w8Sug9mh6cAZ0/UAEC0FKH+AOCOEoJQBeD8H+A2A+DYYGBuD2H+D6DmFCH+F0Fq hDB+PKGcGaaSGsGmJyGOGGg0EuEgP0CqCOEOH+BIA+DQH+AgBSDrD+CAEcH+AICwFNA+pamupfDQ /G1E2asfDksJDesNBCv012x+5WDmQSseeiAaCGEiH+BEAmC0H+A8AcCyH+A4AXGqAqAOCsLqAQCw H+BMAuDIH+CJEWPMCpFKCaCEEASEBFESA2AaC2H+AoAOCqK6AVGqA8AYCvGcAiC9HmBlHZBglBDt F80LF7BNDdBHDg2U0JGI0lGEseWsAgB6EYH+BUAkC4H+BKAuDDGcArH+BCAnH+A8AeC6K7HjG7H4 AoANHtG1G4A1GxGkAfHkBAAjJOBCAqC/FMAuDAH+BOApJ8A0BYDy2aWtGEzI1G2dKRIUwDGDF9BP DooBISnVImB9IsBRJHI2AuDHK3HHI4DFK3K6BGArI8BAAlJ8A8AhJqAlJxJBK3LDLBK9LmBNKCH+ AwBZEZIGcjGM3zKVIRL7BEH8zVKfKojrKlMCQVIk2jKuH/KzH/I5K7MjLFMpMnI5HGBIAvG/I5J9 MvMrK5M/K7LrJ9LxL1DXL5KYp1L/MM80g2wABMzWAGjVDZIc/XDrIiNjKtKxK1MtNBN7MlN9HFJ5 M3J7NDONMnNHLvLzKNKnMS5dNPMAsGpgLKLoAQNswAH8EAGAXGB0EobQAwEEJOblOa1uzJIfNvF9 MXFVMbMfOPODOBOBMxM1K3M7PfPdNFLtNLOZOcfjIOgVPIfeAaD2VGBUEY6ECyFQL+awXWH+HCHa KgW+KoAqDsPu3Mt8rxMO+fOjPTNzMZN3MhPtN/M/PlOJPrPhPvHBPzOXL3KPKhP69Mv4SmYWWs4m QyDGFUGsH+GeHEOkHW204SIED6FmaSBqeZHmQ3Fikw7iPwZWkCSs40Yg0CwTQAxpPUAhPZN5RDS1 PjOHPpRRRFOTP1RZSo5W1skw6MDa6g5omYZW94AYDkN+BOD+NKCoE6SKHy1WA8EOL2Z8cuhGTU/f SlArAu95L2cmiO1xSrQ7PXQ/S/S1RJS9RFUlNBTDRXNPRavw3St25kTXBiQsADBfUFNO/fUC6K+A N+9fTI2SkOfG6YmY9MZWDEFS+0TInFUQSfPI1NQ9MdSzRPUnS5MzRLUdRPUqD02RRaeoseDdSaQn ScQi9G/kPu+e/o+E8UQUoGv/IXMKsOfqZu8O/eV2S89E7QZ7XG8w/edGQYxiTeoSB3FKBQAlRBV9 S3LnWDUjXpN7LCBOAnJ8AyBXKKV0Z7KPKO6i764rXI9wS8DJVq6TAw+hP5MFBJDi0Ofq0i86mQPv XMDRWlBi6QVy6iAYCUEuJIBDESBGA0DNI2AxLjRCAxS4BGAvH5I5J2BLZdZVV9ZtZqDKJIA+DUH+ AiB4ES2a4pDFDE+fVIkg84ZylyTUn7RdYjIZYhQy14fipY9SDTWhSjRg6UAEC+fZGWEfEMBCDbFM AtJ9MzI9LlM8BFZjJ5ZpNBMzLCBJLKH+A+A3ZSAiByEMmcC0FO7Q7eTVSW+ectBY9HQ2oFDTNXak Ns9fIa0hPPVUfqfq5i3OjohIAEC6fYAWB6ERFUBSDxLuAtZSA7JSA/JoH/LPH+BIAtK7bZG4BGAt JCApH+A+AhJPGnJOA0AfH+AiBIDoWuBzc6AICvD8PxUAWJWWj2sfWwsPP9cXWzKc/PGE0jPQvwiE 9kWsAeB+EVI2AaCpdQAbGjGnHkA4AbJPHhI0A2AZI0A6AdI1drG+A6AfG/HhGjfVI1fLfJGpfBGq BGAZHsAvKJA+j7eZencVNSlZGBelINchalStIqbTV7J5I8BDLtdrJPfLfSAZHkA2AXfPfXENJTfl JrK0BFdjMoBOArX7UsYhUxaledgRcZgVBLMTepVU17StSxXlMrZzbgAtLDgrdoAlGiBBK1dhZZLi AxK/OFORRVNNhbhu0PhhafgTW1gXilgbgRhzUbV/OOAwDPFMAxe/ZrPriXXnWJidP3hiuZgPiphk eGA8EUJGWEc3MNhtgdUXSvi5XxS1Z3Zhe/MzbfjNRRUrieVpeejtOg0OseDeNYQexAIEE4GcX2B4 EkRKAMDsNjXSTWcVVVjvi1jzh1WHM/j8AvkAAvkFlHkLjVjdjZF4pPk0nANKLKQIBCESL2D6F6XG XgvWqwHQMqEsGIRaL6RSAgD6bcQAbyNRBkLQDCWNhdNxkPj1V5h3i7N7lLlPlTmtUpjTTHak1tUE DHDyQMDWN+ggQIByEiJOEGOcGIG8Okq7CMXay2KMG8HaU4rcMGCIE8aSBcEWL2Ci4ccggbKXhxlD j3jPPdmxbdLnm3PxNJhZkO1CseseDUNKBIEMJWB8EqhCBqEuGgH+DiFiL+GXQcINAEIhl4McIkXc IICurWWuDulCNiDW85VVi3mplHmuH/j/oZM9odRTohkNNQnjYIc6QIAILmH+BgEOLOtabCUcIPpV qiIOHMKeOQGoJ6C+FaMwAuEJT474DYTfliJQrbgjmrj5OBoXkDobrTJ/LsArX+LQDI72nBlkJQLv C2FBpAEoGOMGGhqsvKH7nlqpqlnpsKKMHgJ+tgGQPKCmFIaSAwD8LOA4EGJWBiCdEFH1G/iPp0Av rXlRrboTdXLDdrI0BSBkD4H/n8JOAkD7o0EwhCECF+PKG6HceBsQRtsPtyOuq4/+22RcFYJGBkBK Ddp4AtI8A2AdHkA/H9uPZZrVp5lNp9OFMzK7iLJ9HhHlgrJ8BSA3FSEWECe+vNsVnkXdt4MVqnvR qiHmHkKgF+F0JqEMD8fECQB4D7uPI8AyAZGjubJ8BAAqClbqAng4AbfZdsKyBTuMDiDOE2H+FmFe JGHMU1vXwrvVwrwxCKagF3pADyDiFEH+CWCAEHJ+A3G4B+BmDmH+DQC7EEFGE4NKHYHUXvwxxqIj t3xtxyIaw0ImGwGus8HgUZx1yGavxxyJyPyRySKIKQFuFibQGOGEYFyVynypymKQCaB/xIDkDOE8 IQFSFEVGE+EsQIIWGEF4L+EwEeP6IOE4EoN2DOC0ErpCDLy6DYDAEyH+FiFU6EHyq6H+DqDUFAH+ G4GwOWIWw0IIEMD68HzsE0H+DgDIE4H+DaDD0cG8G4X2FgFS6EFGE2LAIWG0GsHMH+EoEWTaIOFn uEH+DKCzzkDjzoH+DJ1aH+FuFgbRsGIIEKD2fZyCrEIOGUGMI2E8EqNLz6XmDsDXEgDUC6EwH+DW C9zx1ZzkF4FuOOHUq0D8Dnb8DeDHwd0h0kEeEKQsKQCkCLFKDv2SIQD6DkFJENg2H/ygYFxkPCBu BQDkH+CwCXIsIOCiCJb2CQB0D+H+FEE0LoEYEGTOCgCGEKH/2uMrGfI8GEF7RyIWHwOoH+CEBptU EaEITOFCEyLOF0FoJr16aUDpb9fbH+GAF0RSz2KgCABjdACSB4ED3V3YH+C8CjbCFAEwLP006EG8 G2X35LbnK6B/5iH+GcGSs8FoFehCC0CYEavXAT5WRSDyDdxACKBtvwFwFkJqHEHAUVxeLACSB3HZ 2GNKEyEgP6EsEaNEKQCn3MH+DuDZEgIOEUEAZMDuDbxB7jFL5zbCDQC2Et2cC/zwIP8AH+BoBKDf 1iCxEEC4CfFgEqEYNFCeU4BWA2DSXIGgXGIWu+XmCQB35r2R0EEgEOQyIWEOD88GECDvVmCyCZIs DACpGaE0EkN+DP8GIQEGDyFUH/3LFL0Ue+HkHh18IMHPwoCqCRe54ILoCMBvEKC6Cl8CC18IIWFa FKL2DGCtEEIXE0MH2+H+EKD4fZ10fZ374Z7h7l7p7sIMEF9eRd6d1p1t4H4L/ohCDECsEkIQIAVC MiX+dTUoH++n0/X++4W/4hEHe7Xo/xQGTO/zQXEu/0Cd1S/0Yglg/3g73q/yWPUE/zsa1C/0IeVW /0GeFU/3G4Xa/0MfVY/1or2bQlbRUiiVpJpQ/y+U0jEYggzzOSkRUM/0KfJqfzop3+p1Aw3+53I7 n+UCJWak3m26H+OxYdX+Xiij6leVRY3+YKheaks1Yy3+VCPBEIe5zM5qOxbdICAADgEAAAMAAAAB ADgAAAEBAAMAAAABAEAAAAECAAMAAAAEAAAWdgEDAAMAAAABAAUAAAEGAAMAAAABAAIAAAERAAQA AAABAAAACAEVAAMAAAABAAQAAAEWAAQAAAABAAAAkgEXAAQAAAABAAAVwAEaAAUAAAABAAAWfgEb AAUAAAABAAAWhgEcAAMAAAABAAEAAAEoAAMAAAABAAIAAAFSAAMAAAABAAEAAAAAAAAACAAIAAgA CAAK/IAAACcQAAr8gAAAJxA= --Apple-Mail-57--473644005 Content-Type: multipart/alternative; boundary=Apple-Mail-59--473644003 --Apple-Mail-59--473644003 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-59--473644003 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-59--473644003-- --Apple-Mail-57--473644005-- From pgsql-performance-owner@postgresql.org Mon Jul 18 16:32:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 502DD529C0 for ; Mon, 18 Jul 2005 16:32:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36311-06 for ; Mon, 18 Jul 2005 19:32:39 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 4636452AC0 for ; Mon, 18 Jul 2005 16:32:38 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6IJWaw2016395; Mon, 18 Jul 2005 15:32:36 -0400 (EDT) To: Christopher Petrilli Cc: Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c405071811457ccdd3f1@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Mon, 18 Jul 2005 14:45:35 -0400" Date: Mon, 18 Jul 2005 15:32:36 -0400 Message-ID: <16394.1121715156@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/270 X-Sequence-Number: 13511 Christopher Petrilli writes: > http://blog.amber.org/diagrams/comparison_mysql_pgsql.png > Notice the VERY steep drop off. Hmm. Whatever that is, it's not checkpoint's fault. I would interpret the regular upticks in the Postgres times (every several hundred iterations) as being the effects of checkpoints. You could probably smooth out those peaks some with appropriate hacking on bgwriter parameters, but that's not the issue at hand (is it?). I have no idea at all what's causing the sudden falloff in performance after about 10000 iterations. COPY per se ought to be about a constant-time operation, since APPEND is (or should be) constant-time. What indexes, foreign keys, etc do you have on this table? What else was going on at the time? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 16:36:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8336B52AC3 for ; Mon, 18 Jul 2005 16:36:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36543-06 for ; Mon, 18 Jul 2005 19:36:19 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.205]) by svr1.postgresql.org (Postfix) with ESMTP id 6D9B7529C0 for ; Mon, 18 Jul 2005 16:36:18 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i34so1058753wra for ; Mon, 18 Jul 2005 12:36:20 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=iyjsSuCvUJvaUNJwO+eKhaBdUg3zPl7pUCHYZuln7EatkQvJUaxOMDMxlaQ/NTC/oRermpE4rsfbSYVVMUpykmHGGC0TLeygoEnZbr/LsynpzFrJ9FxZvH16bCrt+//MI9GWBuiTJsHwUDPRjZkm40YTGH19GXcssAkw/YM2JZY= Received: by 10.54.36.58 with SMTP id j58mr477320wrj; Mon, 18 Jul 2005 12:34:57 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Mon, 18 Jul 2005 12:34:57 -0700 (PDT) Message-ID: <59d991c405071812342e290f57@mail.gmail.com> Date: Mon, 18 Jul 2005 15:34:57 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <16394.1121715156@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.123 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/271 X-Sequence-Number: 13512 On 7/18/05, Tom Lane wrote: > Christopher Petrilli writes: > > http://blog.amber.org/diagrams/comparison_mysql_pgsql.png >=20 > > Notice the VERY steep drop off. >=20 > Hmm. Whatever that is, it's not checkpoint's fault. I would interpret > the regular upticks in the Postgres times (every several hundred > iterations) as being the effects of checkpoints. You could probably > smooth out those peaks some with appropriate hacking on bgwriter > parameters, but that's not the issue at hand (is it?). I tried hacking that, turning it up to be more agressive, it got worse. Turned it down, it got worse :-) =20 > I have no idea at all what's causing the sudden falloff in performance > after about 10000 iterations. COPY per se ought to be about a > constant-time operation, since APPEND is (or should be) constant-time. > What indexes, foreign keys, etc do you have on this table? What else > was going on at the time? The table has 15 columns, 5 indexes (character, inet and timestamp). No foreign keys. The only other thing running on the machine was the application actually DOING the benchmarking, written in Python (psycopg), but it was, according to top, using less than 1% of the CPU. It was just talking through a pipe to a psql prompt to do the COPY. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Mon Jul 18 17:32:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CF33152AC2 for ; Mon, 18 Jul 2005 17:32:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 47414-10 for ; Mon, 18 Jul 2005 20:32:24 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 55859529A9 for ; Mon, 18 Jul 2005 17:32:23 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6IKWNkt016795; Mon, 18 Jul 2005 16:32:23 -0400 (EDT) To: Christopher Petrilli Cc: Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c405071812342e290f57@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Mon, 18 Jul 2005 15:34:57 -0400" Date: Mon, 18 Jul 2005 16:32:23 -0400 Message-ID: <16794.1121718743@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/272 X-Sequence-Number: 13513 Christopher Petrilli writes: > On 7/18/05, Tom Lane wrote: >> I have no idea at all what's causing the sudden falloff in performance >> after about 10000 iterations. COPY per se ought to be about a >> constant-time operation, since APPEND is (or should be) constant-time. >> What indexes, foreign keys, etc do you have on this table? What else >> was going on at the time? > The table has 15 columns, 5 indexes (character, inet and timestamp). > No foreign keys. The only other thing running on the machine was the > application actually DOING the benchmarking, written in Python > (psycopg), but it was, according to top, using less than 1% of the > CPU. It was just talking through a pipe to a psql prompt to do the > COPY. Sounds pretty plain-vanilla all right. Are you in a position to try the same benchmark against CVS tip? (The nightly snapshot tarball would be plenty close enough.) I'm just wondering if the old bgwriter behavior of locking down the bufmgr while it examined the ARC/2Q data structures is causing this... regards, tom lane From pgsql-performance-owner@postgresql.org Mon Jul 18 17:47:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D27DB52AFB for ; Mon, 18 Jul 2005 17:47:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 50704-08 for ; Mon, 18 Jul 2005 20:47:40 +0000 (GMT) Received: from gwmta.wicourts.gov (gwmta.wicourts.gov [165.219.244.91]) by svr1.postgresql.org (Postfix) with ESMTP id 8DCB052CDE for ; Mon, 18 Jul 2005 17:47:38 -0300 (ADT) Received: from Courts-MTA by gwmta.wicourts.gov with Novell_GroupWise; Mon, 18 Jul 2005 15:47:41 -0500 Message-Id: X-Mailer: Novell GroupWise Internet Agent 6.5.2 Date: Mon, 18 Jul 2005 15:47:34 -0500 From: "Kevin Grittner" To: , Subject: Re: join and query planner Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.017 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/273 X-Sequence-Number: 13514 You might want to set join_collapse_limit high, and use the JOIN operators rather than the comma-separated lists. We generate the WHERE clause on the fly, based on user input, and this has worked well for us. -Kevin >>> "Dario" 07/18/05 2:24 PM >>> Hi. > Just out of curiosity, does it do any better with the following? > > SELECT ... Yes, it does. But my query could also be SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) /*new*/ , e WHERE (b.column <= 100) /*new*/ and (e.key = a.key) and (e.field = 'filter') because it's constructed by an application. I needed to know if, somehow, someway, I can "unforce" join order. The only way to solve it so far is changing application. It must build something like SELECT ... FROM b JOIN (a JOIN e ON (e.key = a.key)) ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) and (e.field = 'filter') Supossed that e.field has (should have) better selectivity. But now this problem belongs to programmer's group :-) The query, in fact, has more tables to join. I wonder if lowering geqo threshold could do the work... Thank you. Greetings. Long life, little spam and prosperity! -----Mensaje original----- De: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org]En nombre de Kevin Grittner Enviado el: lunes, 18 de julio de 2005 14:58 Para: pgsql-performance@postgresql.org; dariop@unitech.com.ar Asunto: Re: [PERFORM] join and query planner Just out of curiosity, does it do any better with the following? SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) >>> snipp ---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings From pgsql-performance-owner@postgresql.org Mon Jul 18 20:31:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B201A52BB0 for ; Mon, 18 Jul 2005 20:31:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 11061-10 for ; Mon, 18 Jul 2005 23:31:40 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.196]) by svr1.postgresql.org (Postfix) with ESMTP id 4382A52BAE for ; Mon, 18 Jul 2005 20:31:38 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i32so1093156wra for ; Mon, 18 Jul 2005 16:31:42 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=f0WrQ3deGdjBEGTN2q1oYQ2x8FJu7bLD0tmgr0fIdggon0ubd+8xy6j/IFBr2/dOKwQXOHBcx+N2gR8HxsiIf2Pw6J4/XVY1XNfWTAqq8/HLGs1+01Dkh+Yts/F8W++NwLUeJiej2hIJgtc4yQO5fT+K8yQdXC2n7zEoj18xIZY= Received: by 10.54.38.40 with SMTP id l40mr538132wrl; Mon, 18 Jul 2005 16:30:53 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Mon, 18 Jul 2005 16:30:53 -0700 (PDT) Message-ID: <59d991c405071816304af42ef@mail.gmail.com> Date: Mon, 18 Jul 2005 19:30:53 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <16794.1121718743@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.121 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/274 X-Sequence-Number: 13515 On 7/18/05, Tom Lane wrote: > Christopher Petrilli writes: > > On 7/18/05, Tom Lane wrote: > >> I have no idea at all what's causing the sudden falloff in performance > >> after about 10000 iterations. COPY per se ought to be about a > >> constant-time operation, since APPEND is (or should be) constant-time. > >> What indexes, foreign keys, etc do you have on this table? What else > >> was going on at the time? >=20 > > The table has 15 columns, 5 indexes (character, inet and timestamp). > > No foreign keys. The only other thing running on the machine was the > > application actually DOING the benchmarking, written in Python > > (psycopg), but it was, according to top, using less than 1% of the > > CPU. It was just talking through a pipe to a psql prompt to do the > > COPY. >=20 > Sounds pretty plain-vanilla all right. >=20 > Are you in a position to try the same benchmark against CVS tip? > (The nightly snapshot tarball would be plenty close enough.) I'm > just wondering if the old bgwriter behavior of locking down the > bufmgr while it examined the ARC/2Q data structures is causing this... So here's something odd I noticed: 20735 pgsql 16 0 20640 11m 10m R 48.0 1.2 4:09.65 postmaster =20 20734 petrilli 25 0 8640 2108 1368 R 38.1 0.2 4:25.80 psql The 47 and 38.1 are %CPU. Why would psql be burning so much CPU? I've got it attached ,via a pipe to another process that's driving it (until I implement the protocol for COPY later). I wouldn't think it should be uing such a huge percentage of the CPU, no? The Python script that's actually driving it is about 10% o the CPU, which is just because it's generating the incoming data on the fly.=20 Thoughts? I will give the CVS head a spin soon, but I wanted to formalize my benchmarking more first. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 05:35:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F401A529C6 for ; Tue, 19 Jul 2005 05:35:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 66754-06 for ; Tue, 19 Jul 2005 08:35:18 +0000 (GMT) Received: from apate.telenet-ops.be (apate.telenet-ops.be [195.130.132.57]) by svr1.postgresql.org (Postfix) with ESMTP id 295B052983 for ; Tue, 19 Jul 2005 05:35:15 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by apate.telenet-ops.be (Postfix) with SMTP id 4657338175 for ; Tue, 19 Jul 2005 10:35:17 +0200 (CEST) Received: from [10.0.1.2] (d5152B1F3.access.telenet.be [81.82.177.243]) by apate.telenet-ops.be (Postfix) with ESMTP id 6B04B3805D for ; Tue, 19 Jul 2005 10:35:16 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v622) In-Reply-To: References: Content-Type: multipart/mixed; boundary=Apple-Mail-12--426488437 Message-Id: <3508d984c33d87fd74c90913aaa24c4d@implements.be> From: Yves Vindevogel Subject: Re: Insert performance (OT?) Date: Tue, 19 Jul 2005 10:35:15 +0200 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.622) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/275 X-Sequence-Number: 13516 --Apple-Mail-12--426488437 Content-Type: multipart/alternative; boundary=Apple-Mail-13--426488437 --Apple-Mail-13--426488437 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed nobody ? On 18 Jul 2005, at 21:29, Yves Vindevogel wrote: > Hi, > > Suppose I have a table with 4 fields (f1, f2, f3, f4) > I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) > > I have 3 records > A, B, C, D (this will be inserted) > A, B, C, E (this will pass u2, but not u1, thus not inserted) > A, B, F, D (this will pass u1, but not u2, thus not inserted) > > Now, for performance ... > > I have tables like this with 500.000 records where there's a new=20 > upload of approx. 20.000 records. > It is only now that we say index u2 to be necessary. So, until now, I=20= > did something like insert into ... select f1, f2, f2, max(f4) group by=20= > f1, f2, f3 > That is ok ... and also logically ok because of the data definition > > I cannot do this with 2 group by's. I tried this on paper and I'm not=20= > succeeding. > > So, I must use a function that will check against u1 and u2, and then=20= > insert if it is ok. > I know that such a function is way slower that my insert query. > > So, my question ... > How can I keep the same performance, but also with the new index in=20 > mind ??? > > > Met vriendelijke groeten, > Bien =E0 vous, > Kind regards, > > Yves Vindevogel > Implements > > > > Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 > > Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 > > Web: http://www.implements.be > > First they ignore you. Then they laugh at you. Then they fight you. =20= > Then you win. > Mahatma Ghandi. > ---------------------------(end of=20 > broadcast)--------------------------- > TIP 5: don't forget to increase your free space map settings > Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements --Apple-Mail-13--426488437 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 nobody ? On 18 Jul 2005, at 21:29, Yves Vindevogel wrote: Hi, Suppose I have a table with 4 fields (f1, f2, f3, f4) I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) I have 3 records A, B, C, D (this will be inserted) A, B, C, E (this will pass u2, but not u1, thus not inserted) A, B, F, D (this will pass u1, but not u2, thus not inserted) Now, for performance ... I have tables like this with 500.000 records where there's a new upload of approx. 20.000 records. It is only now that we say index u2 to be necessary. So, until now, I did something like insert into ... select f1, f2, f2, max(f4) group by f1, f2, f3 That is ok ... and also logically ok because of the data definition I cannot do this with 2 group by's. I tried this on paper and I'm not succeeding. So, I must use a function that will check against u1 and u2, and then insert if it is ok. I know that such a function is way slower that my insert query. So, my question ... How can I keep the same performance, but also with the new index in mind ??? Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements < Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you.=20 Then you win. Mahatma Ghandi. ---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements = --Apple-Mail-13--426488437-- --Apple-Mail-12--426488437 Content-Transfer-Encoding: base64 Content-Type: image/tiff; x-unix-mode=0666; name="Pasted Graphic 2.tiff" Content-Disposition: inline; filename="Pasted Graphic 2.tiff" TU0AKgAAFciAP6BP5/wWDQeEQmFQuGQ2HQ+FP5+v9rOh5P9IMBuP8zK9tP8fJNlP8QoNjP8FndhP 8MnyVjJHSMrqFnv9BL1vP9ett2v98PuJxChUOiUWGQOCUalUujOV4PZ/qBluR/mdVtJ/jVHst/ho 9MF/g08WABHJfv8AWa0WoAHFfWuzgg6sB/hY9ysUohjv8qqVqv9LMZxv9ouV40zEYmHUjFY2HPx+ RNtOl4Rhftt/phiRs6LC/iA5LewnJeWg46W22+02cAHOwa26bC0HPY7TZ7G2G9dv8EHFdP8Om9YP 8rKJov9UtFzv9Dzh/sRvu5/vZ9PzHdeF4zsUqgQRlOR5v9DMBwv8uqe/jZIVwNHu6B0/WAKHvWGr RAA1ra0G1c/vfgANrdraXrTQIAA6JW2TUrQNz/ja0oAjO4YAjSWS0DOWawjys4RkIk4Lj2sAWEUZ B/ieUJqH+PZdnAf5aG06R5nsfLtqW7UaoWZJwukwLyi4VRsn+F5HGYf8QJWBQ7JWAI5tYti2Dm1T bQHAq0De1Q2FwtA1Fof4AjVDIBDQV0vDQWK0DY/oADYWsrNQN81NOuDZtYOM6ycs4Cjq+Q9JWE5D pGJpOpqP5dMwW5sHUf53xlHCExupjqoIbh3Hu55vp6ch3Hof4wlMrAdEWYh/gmP1RtXK0DDi2rXy nVzXjpVs8NM1g1vuNU2gANEugGMBRwu4YADTNsvwyAA3v+0k6LRWLb2dBUpjk2I4NVOx/gYPC6BU QiwCgSsSlWZpxH+ahzsOaBzvCe7qoMpKlUghd1n2f5oHId5/lQaJ0n+RBf3GLpUGwf4iksrg0FVF ICzlXT9QBB0tWO/+FjlKVWtfijTNRBsvDbhrdP3NUAwY38KFfkc0PuNjRWLLz8ytiTUVdVlnjlA2 MWPAWRAAM8LAAMjhhuR5ir4ULjigT5pn+P5co2T6pn+YBup6dp6xoh6kHie95modNOFkah0H+Qhd pyKBNmcf68r2Bo9GGtdpwIAUotM1Q4QMNzUDZhuVZbLsvy7KkqZZCnAb4AFcQY1C3LW1Wb5lOa2t ZvUt78NNjcMN83wEOMDczvuTzXNQ3QEN01WRL2b8XBYEDuugQEFUYgkpIo5FjIJTmhfZkHG8J0nm fB/sifp/EQYLBioUKsBkRKVgWPLX7stA4VnaDa+tZlZel0j+gCMpWS8MRULQNeIDY/4zTIAIyFX0 GFybZnr+rZ8p2a1s6tUNfADXXI1Yb8jLX+MNQk+1Az7zWqwVa/FaJrHpvSLeAQOhdE/l7CSJwrAb BXDXH+PkyA/gUiPGaftASWW6JeDgnGAsCmLNvcYqlz5+FchnZMrqGbeE3JoYhAMAKbEvOGeigtaE CDZwJcgxtiML0uMgQYzllKt2GhtP/EaH6cgAm+fnCtVEVXFOKNU54AC1B/tyLfBx4YJBGwhAAG5m yUoGH/TgxlZ7nGXw3YdEpYTPWOvaSq/JVEX28rEDQsGKDoGPx+QYnE1URg2paZY4dy7EHTx9cc41 upqIfwkjSgJWyW0MgBdK9I1DGAAwnS8G8+4cGYyUWWbKMg/gSiOjQG+FMQ44qqlmgBNUSYYMohKW 2ULFZaR8NdMFucoz/w/ZcAAMzgH+j/AKGV8QAZkhmWNM0AJ/I5zCiJEGYjF1qoGmxI5MqwT7JVlG buLULJhPXlbK+WMt3HwHmIbF8xaA0n6mvPgN7EJhx9Vcax+hsX6wGZuldNDnTSgQDoloGohS6BYE 62gPQqyRhlFE2gIYkCRgdD4WcAUsnxvlYklKgUCn4upi29JkkpG/pbYgtKbr86TLOnbLCF066ZLP kGsI+7NY9rRjZHB+0ew4M5S0AKT4FA8l0BaIUk4XHkj/EYLwzAtVyj/GwRYf49B8D6H+OMdw9R/j DG8vcTgx1xhpFTBoHYilRgaDygQAobXCynVVSSVdRYTLJls5CcIaqX0zpw/Kms76c2DSnTsNNPaA TbpggqOSCg4MkQwcAPiBAzimJqI4Xg3R/itX0P8bY66xD7MiYkgg5h4vAF2NgdY/xNGCH+HQVMIQ WiCNVM2L6Ap4s3ShFixMubA2HnjcUulhab3GuIbGxVjLlxXZoawMKFgsCiaSOUd6nB6XbH+PofKN LtqcH7eMf5AyimQOtegf4/TIEFvYP8dyjTm2eV1KdmNgrlIKr/cOxF+LjmRldTZY88L/U6P/YtyF +b82PDAhkOQsyNkHu8jS76NB4YXH+PvDUG7v3lIFeu8mFbuj6q8PXE2HEaXqIEUEgxnh2T2YhD/B WBZcUunVjO59yMB3PsHc3BONLemxwaP8NwsTMEPvVeMieJKvXmHZk8f48h5GHyYP8e+V3g3tvMQs VY01FLCxjgTHk8b943yBTOVoJEiXJzPgae1zr+2HwYhnIuRzrjczwP8eY8zDmKy7l8NOYcx40zKW nNswizytg+kXHc/qS48x9o7QbNMhZ0yMdvPBG7wmNz/jB6WYs436uFmbHhbFrAFgjhl4Y/g5IvH+ 0EkYFw+NuAHqmoet9bWJwPnDHEfM55E0udjTNW7uZ+y9p6L+oNe3Bxtoa4lACzgQee2kQ5Jwo1RK QPQfK8xejaJ6FG6w/wEh1NZJ+L6ydoH+zfqTUK0dK7AzsY7Yem9jaA0Fu2wVf9eJObqW+TxuwbCS K4JEYVnh2O/vWQM5K+wtCehCCAP5YABybV035xCAJIJqnrTyFqB9H8e1Lu/OumM870MTp3MGn9J8 g5Zsxhk2ZURLfGmrBAEg5m/CISIf4jRekblaCgRoyS0BjWNZVw6uZ70hnsrkND/Fbn3kWWtmz1ID a5NxyLYJ1xujdI3iZTm9dkYy4/P9yFKIcFoDQsbtKW4nT2YbLrpr450V6zTGem7N+YzXVzDZ0PSl ddIb8GQVrlD9sQjy4FKHVy0ZDDeLHCPWuuD/68Y0Vg1LX8c0bQSBiAq9JrVv2pYwZkzq6mqxDi8y XJgBkjZGftMMdUgm5PKyCBnogBI72hYgZlgrD6WWgMzPZkxGdUzfIfIzsdb66PXr/Jxp+XwRGlOL GnuJm2b0l7vo/dJeDL4N7qwWN+qn5CqedNMATuzZmObaad1s4dBMl/6wmG/Pt1+oAAYk2hToiuQc 5FxyjyRoHqH0xYKE+Q8k+UIgH0KAH+HSHqXmGwHSIuEGF0I25eWE8A96/g2akwhIk8Puf25mpi3w /IeG/M8y0O10/WgM3cpUS8l0mS+ocOYgDQS0ASDUTOB6EaJWDkXyXwFyL+GqGkPKHQHMOkxIXmIO HAHAJyyuKgvGIIHgU2H+G0GuHKH+FwF6KwEAtAH+CUEcLOAiNC2QnKcOZSZCmO9ongNep2sA3Y2W nY/KwE9hDazc8w8Sug9mh6cAZ0/UAEC0FKH+AOCOEoJQBeD8H+A2A+DYYGBuD2H+D6DmFCH+F0Fq hDB+PKGcGaaSGsGmJyGOGGg0EuEgP0CqCOEOH+BIA+DQH+AgBSDrD+CAEcH+AICwFNA+pamupfDQ /G1E2asfDksJDesNBCv012x+5WDmQSseeiAaCGEiH+BEAmC0H+A8AcCyH+A4AXGqAqAOCsLqAQCw H+BMAuDIH+CJEWPMCpFKCaCEEASEBFESA2AaC2H+AoAOCqK6AVGqA8AYCvGcAiC9HmBlHZBglBDt F80LF7BNDdBHDg2U0JGI0lGEseWsAgB6EYH+BUAkC4H+BKAuDDGcArH+BCAnH+A8AeC6K7HjG7H4 AoANHtG1G4A1GxGkAfHkBAAjJOBCAqC/FMAuDAH+BOApJ8A0BYDy2aWtGEzI1G2dKRIUwDGDF9BP DooBISnVImB9IsBRJHI2AuDHK3HHI4DFK3K6BGArI8BAAlJ8A8AhJqAlJxJBK3LDLBK9LmBNKCH+ AwBZEZIGcjGM3zKVIRL7BEH8zVKfKojrKlMCQVIk2jKuH/KzH/I5K7MjLFMpMnI5HGBIAvG/I5J9 MvMrK5M/K7LrJ9LxL1DXL5KYp1L/MM80g2wABMzWAGjVDZIc/XDrIiNjKtKxK1MtNBN7MlN9HFJ5 M3J7NDONMnNHLvLzKNKnMS5dNPMAsGpgLKLoAQNswAH8EAGAXGB0EobQAwEEJOblOa1uzJIfNvF9 MXFVMbMfOPODOBOBMxM1K3M7PfPdNFLtNLOZOcfjIOgVPIfeAaD2VGBUEY6ECyFQL+awXWH+HCHa KgW+KoAqDsPu3Mt8rxMO+fOjPTNzMZN3MhPtN/M/PlOJPrPhPvHBPzOXL3KPKhP69Mv4SmYWWs4m QyDGFUGsH+GeHEOkHW204SIED6FmaSBqeZHmQ3Fikw7iPwZWkCSs40Yg0CwTQAxpPUAhPZN5RDS1 PjOHPpRRRFOTP1RZSo5W1skw6MDa6g5omYZW94AYDkN+BOD+NKCoE6SKHy1WA8EOL2Z8cuhGTU/f SlArAu95L2cmiO1xSrQ7PXQ/S/S1RJS9RFUlNBTDRXNPRavw3St25kTXBiQsADBfUFNO/fUC6K+A N+9fTI2SkOfG6YmY9MZWDEFS+0TInFUQSfPI1NQ9MdSzRPUnS5MzRLUdRPUqD02RRaeoseDdSaQn ScQi9G/kPu+e/o+E8UQUoGv/IXMKsOfqZu8O/eV2S89E7QZ7XG8w/edGQYxiTeoSB3FKBQAlRBV9 S3LnWDUjXpN7LCBOAnJ8AyBXKKV0Z7KPKO6i764rXI9wS8DJVq6TAw+hP5MFBJDi0Ofq0i86mQPv XMDRWlBi6QVy6iAYCUEuJIBDESBGA0DNI2AxLjRCAxS4BGAvH5I5J2BLZdZVV9ZtZqDKJIA+DUH+ AiB4ES2a4pDFDE+fVIkg84ZylyTUn7RdYjIZYhQy14fipY9SDTWhSjRg6UAEC+fZGWEfEMBCDbFM AtJ9MzI9LlM8BFZjJ5ZpNBMzLCBJLKH+A+A3ZSAiByEMmcC0FO7Q7eTVSW+ectBY9HQ2oFDTNXak Ns9fIa0hPPVUfqfq5i3OjohIAEC6fYAWB6ERFUBSDxLuAtZSA7JSA/JoH/LPH+BIAtK7bZG4BGAt JCApH+A+AhJPGnJOA0AfH+AiBIDoWuBzc6AICvD8PxUAWJWWj2sfWwsPP9cXWzKc/PGE0jPQvwiE 9kWsAeB+EVI2AaCpdQAbGjGnHkA4AbJPHhI0A2AZI0A6AdI1drG+A6AfG/HhGjfVI1fLfJGpfBGq BGAZHsAvKJA+j7eZencVNSlZGBelINchalStIqbTV7J5I8BDLtdrJPfLfSAZHkA2AXfPfXENJTfl JrK0BFdjMoBOArX7UsYhUxaledgRcZgVBLMTepVU17StSxXlMrZzbgAtLDgrdoAlGiBBK1dhZZLi AxK/OFORRVNNhbhu0PhhafgTW1gXilgbgRhzUbV/OOAwDPFMAxe/ZrPriXXnWJidP3hiuZgPiphk eGA8EUJGWEc3MNhtgdUXSvi5XxS1Z3Zhe/MzbfjNRRUrieVpeejtOg0OseDeNYQexAIEE4GcX2B4 EkRKAMDsNjXSTWcVVVjvi1jzh1WHM/j8AvkAAvkFlHkLjVjdjZF4pPk0nANKLKQIBCESL2D6F6XG XgvWqwHQMqEsGIRaL6RSAgD6bcQAbyNRBkLQDCWNhdNxkPj1V5h3i7N7lLlPlTmtUpjTTHak1tUE DHDyQMDWN+ggQIByEiJOEGOcGIG8Okq7CMXay2KMG8HaU4rcMGCIE8aSBcEWL2Ci4ccggbKXhxlD j3jPPdmxbdLnm3PxNJhZkO1CseseDUNKBIEMJWB8EqhCBqEuGgH+DiFiL+GXQcINAEIhl4McIkXc IICurWWuDulCNiDW85VVi3mplHmuH/j/oZM9odRTohkNNQnjYIc6QIAILmH+BgEOLOtabCUcIPpV qiIOHMKeOQGoJ6C+FaMwAuEJT474DYTfliJQrbgjmrj5OBoXkDobrTJ/LsArX+LQDI72nBlkJQLv C2FBpAEoGOMGGhqsvKH7nlqpqlnpsKKMHgJ+tgGQPKCmFIaSAwD8LOA4EGJWBiCdEFH1G/iPp0Av rXlRrboTdXLDdrI0BSBkD4H/n8JOAkD7o0EwhCECF+PKG6HceBsQRtsPtyOuq4/+22RcFYJGBkBK Ddp4AtI8A2AdHkA/H9uPZZrVp5lNp9OFMzK7iLJ9HhHlgrJ8BSA3FSEWECe+vNsVnkXdt4MVqnvR qiHmHkKgF+F0JqEMD8fECQB4D7uPI8AyAZGjubJ8BAAqClbqAng4AbfZdsKyBTuMDiDOE2H+FmFe JGHMU1vXwrvVwrwxCKagF3pADyDiFEH+CWCAEHJ+A3G4B+BmDmH+DQC7EEFGE4NKHYHUXvwxxqIj t3xtxyIaw0ImGwGus8HgUZx1yGavxxyJyPyRySKIKQFuFibQGOGEYFyVynypymKQCaB/xIDkDOE8 IQFSFEVGE+EsQIIWGEF4L+EwEeP6IOE4EoN2DOC0ErpCDLy6DYDAEyH+FiFU6EHyq6H+DqDUFAH+ G4GwOWIWw0IIEMD68HzsE0H+DgDIE4H+DaDD0cG8G4X2FgFS6EFGE2LAIWG0GsHMH+EoEWTaIOFn uEH+DKCzzkDjzoH+DJ1aH+FuFgbRsGIIEKD2fZyCrEIOGUGMI2E8EqNLz6XmDsDXEgDUC6EwH+DW C9zx1ZzkF4FuOOHUq0D8Dnb8DeDHwd0h0kEeEKQsKQCkCLFKDv2SIQD6DkFJENg2H/ygYFxkPCBu BQDkH+CwCXIsIOCiCJb2CQB0D+H+FEE0LoEYEGTOCgCGEKH/2uMrGfI8GEF7RyIWHwOoH+CEBptU EaEITOFCEyLOF0FoJr16aUDpb9fbH+GAF0RSz2KgCABjdACSB4ED3V3YH+C8CjbCFAEwLP006EG8 G2X35LbnK6B/5iH+GcGSs8FoFehCC0CYEavXAT5WRSDyDdxACKBtvwFwFkJqHEHAUVxeLACSB3HZ 2GNKEyEgP6EsEaNEKQCn3MH+DuDZEgIOEUEAZMDuDbxB7jFL5zbCDQC2Et2cC/zwIP8AH+BoBKDf 1iCxEEC4CfFgEqEYNFCeU4BWA2DSXIGgXGIWu+XmCQB35r2R0EEgEOQyIWEOD88GECDvVmCyCZIs DACpGaE0EkN+DP8GIQEGDyFUH/3LFL0Ue+HkHh18IMHPwoCqCRe54ILoCMBvEKC6Cl8CC18IIWFa FKL2DGCtEEIXE0MH2+H+EKD4fZ10fZ374Z7h7l7p7sIMEF9eRd6d1p1t4H4L/ohCDECsEkIQIAVC MiX+dTUoH++n0/X++4W/4hEHe7Xo/xQGTO/zQXEu/0Cd1S/0Yglg/3g73q/yWPUE/zsa1C/0IeVW /0GeFU/3G4Xa/0MfVY/1or2bQlbRUiiVpJpQ/y+U0jEYggzzOSkRUM/0KfJqfzop3+p1Aw3+53I7 n+UCJWak3m26H+OxYdX+Xiij6leVRY3+YKheaks1Yy3+VCPBEIe5zM5qOxbdICAADgEAAAMAAAAB ADgAAAEBAAMAAAABAEAAAAECAAMAAAAEAAAWdgEDAAMAAAABAAUAAAEGAAMAAAABAAIAAAERAAQA AAABAAAACAEVAAMAAAABAAQAAAEWAAQAAAABAAAAkgEXAAQAAAABAAAVwAEaAAUAAAABAAAWfgEb AAUAAAABAAAWhgEcAAMAAAABAAEAAAEoAAMAAAABAAIAAAFSAAMAAAABAAEAAAAAAAAACAAIAAgA CAAK/IAAACcQAAr8gAAAJxA= --Apple-Mail-12--426488437 Content-Type: multipart/alternative; boundary=Apple-Mail-14--426488435 --Apple-Mail-14--426488435 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-14--426488435 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-14--426488435-- --Apple-Mail-12--426488437-- From pgsql-performance-owner@postgresql.org Tue Jul 19 06:49:41 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3758E528CB for ; Tue, 19 Jul 2005 06:49:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79689-08 for ; Tue, 19 Jul 2005 09:49:37 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id C22F6528C6 for ; Tue, 19 Jul 2005 06:49:36 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id 1CDE8417853; Tue, 19 Jul 2005 10:49:32 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id 2846A15EE6; Tue, 19 Jul 2005 10:39:11 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 03747-03; Tue, 19 Jul 2005 10:39:07 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 9E5A015EDA; Tue, 19 Jul 2005 10:39:07 +0100 (BST) Message-ID: <42DCCA3B.8070100@archonet.com> Date: Tue, 19 Jul 2005 10:39:07 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Yves Vindevogel Cc: pgsql-performance@postgresql.org Subject: Re: Insert performance (OT?) References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/276 X-Sequence-Number: 13517 Yves Vindevogel wrote: > Hi, > > Suppose I have a table with 4 fields (f1, f2, f3, f4) > I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) > > I have 3 records > A, B, C, D (this will be inserted) > A, B, C, E (this will pass u2, but not u1, thus not inserted) > A, B, F, D (this will pass u1, but not u2, thus not inserted) Are you saying you want to know whether they will be inserted before you try to do so? > Now, for performance ... > > I have tables like this with 500.000 records where there's a new upload > of approx. 20.000 records. > It is only now that we say index u2 to be necessary. So, until now, I > did something like insert into ... select f1, f2, f2, max(f4) group by > f1, f2, f3 > That is ok ... and also logically ok because of the data definition I'm confused here - assuming you meant "select f1,f2,f3", then I don't see how you guarantee the row doesn't alredy exist. > I cannot do this with 2 group by's. I tried this on paper and I'm not > succeeding. I don't see how you can have two group-by's, or what that would mean if you did. > So, I must use a function that will check against u1 and u2, and then > insert if it is ok. > I know that such a function is way slower that my insert query. So - you have a table, called something like "upload" with 20,000 rows and you'd like to know whether it is safe to insert them. Well, it's easy enough to identify which ones are duplicates. SELECT * FROM upload JOIN main_table ON u1=f1 AND u2=f2 AND u3=f3; SELECT * FROM upload JOIN main_table ON u1=f1 AND u2=f2 AND u3=f4; Are you saying that deleting these rows and then inserting takes too long? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Tue Jul 19 07:21:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6082D528C3 for ; Tue, 19 Jul 2005 07:21:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86374-09 for ; Tue, 19 Jul 2005 10:21:09 +0000 (GMT) Received: from apate.telenet-ops.be (apate.telenet-ops.be [195.130.132.57]) by svr1.postgresql.org (Postfix) with ESMTP id 18AFB5288D for ; Tue, 19 Jul 2005 07:21:07 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by apate.telenet-ops.be (Postfix) with SMTP id 3C03C38139 for ; Tue, 19 Jul 2005 12:21:10 +0200 (CEST) Received: from [10.0.1.2] (d5152B1F3.access.telenet.be [81.82.177.243]) by apate.telenet-ops.be (Postfix) with ESMTP id 14019380D5 for ; Tue, 19 Jul 2005 12:21:09 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v622) To: pgsql-performance@postgresql.org Message-Id: <940bb98041b7df475b53f98e35c31ce2@implements.be> Content-Type: multipart/mixed; boundary=Apple-Mail-29--420135631 From: Yves Vindevogel Subject: Fwd: Insert performance (OT?) Date: Tue, 19 Jul 2005 12:21:08 +0200 X-Mailer: Apple Mail (2.622) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/277 X-Sequence-Number: 13518 --Apple-Mail-29--420135631 Content-Type: multipart/alternative; boundary=Apple-Mail-30--420135631 --Apple-Mail-30--420135631 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed BTW: thank you for the idea Begin forwarded message: > From: Yves Vindevogel > Date: Tue 19 Jul 2005 12:20:34 CEST > To: Richard Huxton > Subject: Re: [PERFORM] Insert performance (OT?) > > > On 19 Jul 2005, at 11:39, Richard Huxton wrote: > >> Yves Vindevogel wrote: >>> Hi, >>> Suppose I have a table with 4 fields (f1, f2, f3, f4) >>> I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) >>> I have 3 records >>> A, B, C, D (this will be inserted) >>> A, B, C, E (this will pass u2, but not u1, thus not inserted) >>> A, B, F, D (this will pass u1, but not u2, thus not inserted) >> >> Are you saying you want to know whether they will be inserted before=20= >> you try to do so? >> > No, that is not an issue. Problem is that when I use a big query with=20= > "insert into .. select" and one record is wrong (like above) the=20 > complete insert query is abandonned. > Therefore, I must do it another way. Or I must be able to say, insert=20= > them and dump the rest. > >>> Now, for performance ... >>> I have tables like this with 500.000 records where there's a new=20 >>> upload of approx. 20.000 records. >>> It is only now that we say index u2 to be necessary. So, until now,=20= >>> I did something like insert into ... select f1, f2, f2, max(f4)=20 >>> group by f1, f2, f3 >>> That is ok ... and also logically ok because of the data definition >> >> I'm confused here - assuming you meant "select f1,f2,f3", then I=20 >> don't see how you guarantee the row doesn't alredy exist. >> > No, I meant it with max(f4) because my table has 4 fields. And no, I=20= > can't guarantee that, that is exactly my problem. > But with the unique indexes, I'm certain that it will not get into my=20= > database > >>> I cannot do this with 2 group by's. I tried this on paper and I'm=20= >>> not succeeding. >> >> I don't see how you can have two group-by's, or what that would mean=20= >> if you did. >> > select from ( select from group by) as foo group by > >>> So, I must use a function that will check against u1 and u2, and=20 >>> then insert if it is ok. >>> I know that such a function is way slower that my insert query. >> >> So - you have a table, called something like "upload" with 20,000=20 >> rows and you'd like to know whether it is safe to insert them. Well,=20= >> it's easy enough to identify which ones are duplicates. >> >> SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND = u3=3Df3; >> SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND = u3=3Df4; >> > That is a good idea. I can delete the ones that would fail my first=20= > unique index this way, and then delete the ones that would fail my=20 > second unique index and then upload them. > Hmm, why did I not think of that myself. > >> Are you saying that deleting these rows and then inserting takes too=20= >> long? >> > This goes very fast, but not with a function that checks each record=20= > one by one. > >> -- >> Richard Huxton >> Archonet Ltd >> >> ---------------------------(end of=20 >> broadcast)--------------------------- >> TIP 6: explain analyze is your friend >> >> > Met vriendelijke groeten, > Bien =E0 vous, > Kind regards, > > Yves Vindevogel > Implements > --Apple-Mail-30--420135631 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 BTW: thank you for the idea Begin forwarded message: 0000,0000,0000From: Yves Vindevogel < 0000,0000,0000Date: Tue 19 Jul 2005 12:20:34 CEST 0000,0000,0000To: Richard Huxton < 0000,0000,0000Subject: Re: [PERFORM] Insert performance (OT?) On 19 Jul 2005, at 11:39, Richard Huxton wrote: Yves Vindevogel wrote: Hi, Suppose I have a table with 4 fields (f1, f2, f3, f4) I define 2 unique indexes u1 (f1, f2, f3) and u2 (f1, f2, f4) I have 3 records A, B, C, D (this will be inserted) A, B, C, E (this will pass u2, but not u1, thus not inserted) A, B, F, D (this will pass u1, but not u2, thus not inserted) Are you saying you want to know whether they will be inserted before you try to do so? No, that is not an issue. Problem is that when I use a big query with "insert into .. select" and one record is wrong (like above) the complete insert query is abandonned. Therefore, I must do it another way. Or I must be able to say, insert them and dump the rest. Now, for performance ... I have tables like this with 500.000 records where there's a new upload of approx. 20.000 records. It is only now that we say index u2 to be necessary. So, until now, I did something like insert into ... select f1, f2, f2, max(f4) group by f1, f2, f3 That is ok ... and also logically ok because of the data definition I'm confused here - assuming you meant "select f1,f2,f3", then I don't see how you guarantee the row doesn't alredy exist. No, I meant it with max(f4) because my table has 4 fields.=20 And no, I can't guarantee that, that is exactly my problem. But with the unique indexes, I'm certain that it will not get into my database I cannot do this with 2 group by's. I tried this on paper and I'm not succeeding. I don't see how you can have two group-by's, or what that would mean if you did. select from ( select from group by) as foo group by So, I must use a function that will check against u1 and u2, and then insert if it is ok. I know that such a function is way slower that my insert query. So - you have a table, called something like "upload" with 20,000 rows and you'd like to know whether it is safe to insert them. Well, it's easy enough to identify which ones are duplicates. SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND u3=3Df3; SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND u3=3Df4; That is a good idea. I can delete the ones that would fail my first unique index this way, and then delete the ones that would fail my second unique index and then upload them. Hmm, why did I not think of that myself. Are you saying that deleting these rows and then inserting takes too long? This goes very fast, but not with a function that checks each record one by one. -- Richard Huxton Archonet Ltd ---------------------------(end of broadcast)--------------------------- TIP 6: explain analyze is your friend Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements = --Apple-Mail-30--420135631-- --Apple-Mail-29--420135631 Content-Transfer-Encoding: base64 Content-Type: image/tiff; x-unix-mode=0666; name="Pasted Graphic 2.tiff" Content-Disposition: inline; filename="Pasted Graphic 2.tiff" TU0AKgAAFciAP6BP5/wWDQeEQmFQuGQ2HQ+FP5+v9rOh5P9IMBuP8zK9tP8fJNlP8QoNjP8FndhP 8MnyVjJHSMrqFnv9BL1vP9ett2v98PuJxChUOiUWGQOCUalUujOV4PZ/qBluR/mdVtJ/jVHst/ho 9MF/g08WABHJfv8AWa0WoAHFfWuzgg6sB/hY9ysUohjv8qqVqv9LMZxv9ouV40zEYmHUjFY2HPx+ RNtOl4Rhftt/phiRs6LC/iA5LewnJeWg46W22+02cAHOwa26bC0HPY7TZ7G2G9dv8EHFdP8Om9YP 8rKJov9UtFzv9Dzh/sRvu5/vZ9PzHdeF4zsUqgQRlOR5v9DMBwv8uqe/jZIVwNHu6B0/WAKHvWGr RAA1ra0G1c/vfgANrdraXrTQIAA6JW2TUrQNz/ja0oAjO4YAjSWS0DOWawjys4RkIk4Lj2sAWEUZ B/ieUJqH+PZdnAf5aG06R5nsfLtqW7UaoWZJwukwLyi4VRsn+F5HGYf8QJWBQ7JWAI5tYti2Dm1T bQHAq0De1Q2FwtA1Fof4AjVDIBDQV0vDQWK0DY/oADYWsrNQN81NOuDZtYOM6ycs4Cjq+Q9JWE5D pGJpOpqP5dMwW5sHUf53xlHCExupjqoIbh3Hu55vp6ch3Hof4wlMrAdEWYh/gmP1RtXK0DDi2rXy nVzXjpVs8NM1g1vuNU2gANEugGMBRwu4YADTNsvwyAA3v+0k6LRWLb2dBUpjk2I4NVOx/gYPC6BU QiwCgSsSlWZpxH+ahzsOaBzvCe7qoMpKlUghd1n2f5oHId5/lQaJ0n+RBf3GLpUGwf4iksrg0FVF ICzlXT9QBB0tWO/+FjlKVWtfijTNRBsvDbhrdP3NUAwY38KFfkc0PuNjRWLLz8ytiTUVdVlnjlA2 MWPAWRAAM8LAAMjhhuR5ir4ULjigT5pn+P5co2T6pn+YBup6dp6xoh6kHie95modNOFkah0H+Qhd pyKBNmcf68r2Bo9GGtdpwIAUotM1Q4QMNzUDZhuVZbLsvy7KkqZZCnAb4AFcQY1C3LW1Wb5lOa2t ZvUt78NNjcMN83wEOMDczvuTzXNQ3QEN01WRL2b8XBYEDuugQEFUYgkpIo5FjIJTmhfZkHG8J0nm fB/sifp/EQYLBioUKsBkRKVgWPLX7stA4VnaDa+tZlZel0j+gCMpWS8MRULQNeIDY/4zTIAIyFX0 GFybZnr+rZ8p2a1s6tUNfADXXI1Yb8jLX+MNQk+1Az7zWqwVa/FaJrHpvSLeAQOhdE/l7CSJwrAb BXDXH+PkyA/gUiPGaftASWW6JeDgnGAsCmLNvcYqlz5+FchnZMrqGbeE3JoYhAMAKbEvOGeigtaE CDZwJcgxtiML0uMgQYzllKt2GhtP/EaH6cgAm+fnCtVEVXFOKNU54AC1B/tyLfBx4YJBGwhAAG5m yUoGH/TgxlZ7nGXw3YdEpYTPWOvaSq/JVEX28rEDQsGKDoGPx+QYnE1URg2paZY4dy7EHTx9cc41 upqIfwkjSgJWyW0MgBdK9I1DGAAwnS8G8+4cGYyUWWbKMg/gSiOjQG+FMQ44qqlmgBNUSYYMohKW 2ULFZaR8NdMFucoz/w/ZcAAMzgH+j/AKGV8QAZkhmWNM0AJ/I5zCiJEGYjF1qoGmxI5MqwT7JVlG buLULJhPXlbK+WMt3HwHmIbF8xaA0n6mvPgN7EJhx9Vcax+hsX6wGZuldNDnTSgQDoloGohS6BYE 62gPQqyRhlFE2gIYkCRgdD4WcAUsnxvlYklKgUCn4upi29JkkpG/pbYgtKbr86TLOnbLCF066ZLP kGsI+7NY9rRjZHB+0ew4M5S0AKT4FA8l0BaIUk4XHkj/EYLwzAtVyj/GwRYf49B8D6H+OMdw9R/j DG8vcTgx1xhpFTBoHYilRgaDygQAobXCynVVSSVdRYTLJls5CcIaqX0zpw/Kms76c2DSnTsNNPaA TbpggqOSCg4MkQwcAPiBAzimJqI4Xg3R/itX0P8bY66xD7MiYkgg5h4vAF2NgdY/xNGCH+HQVMIQ WiCNVM2L6Ap4s3ShFixMubA2HnjcUulhab3GuIbGxVjLlxXZoawMKFgsCiaSOUd6nB6XbH+PofKN LtqcH7eMf5AyimQOtegf4/TIEFvYP8dyjTm2eV1KdmNgrlIKr/cOxF+LjmRldTZY88L/U6P/YtyF +b82PDAhkOQsyNkHu8jS76NB4YXH+PvDUG7v3lIFeu8mFbuj6q8PXE2HEaXqIEUEgxnh2T2YhD/B WBZcUunVjO59yMB3PsHc3BONLemxwaP8NwsTMEPvVeMieJKvXmHZk8f48h5GHyYP8e+V3g3tvMQs VY01FLCxjgTHk8b943yBTOVoJEiXJzPgae1zr+2HwYhnIuRzrjczwP8eY8zDmKy7l8NOYcx40zKW nNswizytg+kXHc/qS48x9o7QbNMhZ0yMdvPBG7wmNz/jB6WYs436uFmbHhbFrAFgjhl4Y/g5IvH+ 0EkYFw+NuAHqmoet9bWJwPnDHEfM55E0udjTNW7uZ+y9p6L+oNe3Bxtoa4lACzgQee2kQ5Jwo1RK QPQfK8xejaJ6FG6w/wEh1NZJ+L6ydoH+zfqTUK0dK7AzsY7Yem9jaA0Fu2wVf9eJObqW+TxuwbCS K4JEYVnh2O/vWQM5K+wtCehCCAP5YABybV035xCAJIJqnrTyFqB9H8e1Lu/OumM870MTp3MGn9J8 g5Zsxhk2ZURLfGmrBAEg5m/CISIf4jRekblaCgRoyS0BjWNZVw6uZ70hnsrkND/Fbn3kWWtmz1ID a5NxyLYJ1xujdI3iZTm9dkYy4/P9yFKIcFoDQsbtKW4nT2YbLrpr450V6zTGem7N+YzXVzDZ0PSl ddIb8GQVrlD9sQjy4FKHVy0ZDDeLHCPWuuD/68Y0Vg1LX8c0bQSBiAq9JrVv2pYwZkzq6mqxDi8y XJgBkjZGftMMdUgm5PKyCBnogBI72hYgZlgrD6WWgMzPZkxGdUzfIfIzsdb66PXr/Jxp+XwRGlOL GnuJm2b0l7vo/dJeDL4N7qwWN+qn5CqedNMATuzZmObaad1s4dBMl/6wmG/Pt1+oAAYk2hToiuQc 5FxyjyRoHqH0xYKE+Q8k+UIgH0KAH+HSHqXmGwHSIuEGF0I25eWE8A96/g2akwhIk8Puf25mpi3w /IeG/M8y0O10/WgM3cpUS8l0mS+ocOYgDQS0ASDUTOB6EaJWDkXyXwFyL+GqGkPKHQHMOkxIXmIO HAHAJyyuKgvGIIHgU2H+G0GuHKH+FwF6KwEAtAH+CUEcLOAiNC2QnKcOZSZCmO9ongNep2sA3Y2W nY/KwE9hDazc8w8Sug9mh6cAZ0/UAEC0FKH+AOCOEoJQBeD8H+A2A+DYYGBuD2H+D6DmFCH+F0Fq hDB+PKGcGaaSGsGmJyGOGGg0EuEgP0CqCOEOH+BIA+DQH+AgBSDrD+CAEcH+AICwFNA+pamupfDQ /G1E2asfDksJDesNBCv012x+5WDmQSseeiAaCGEiH+BEAmC0H+A8AcCyH+A4AXGqAqAOCsLqAQCw H+BMAuDIH+CJEWPMCpFKCaCEEASEBFESA2AaC2H+AoAOCqK6AVGqA8AYCvGcAiC9HmBlHZBglBDt F80LF7BNDdBHDg2U0JGI0lGEseWsAgB6EYH+BUAkC4H+BKAuDDGcArH+BCAnH+A8AeC6K7HjG7H4 AoANHtG1G4A1GxGkAfHkBAAjJOBCAqC/FMAuDAH+BOApJ8A0BYDy2aWtGEzI1G2dKRIUwDGDF9BP DooBISnVImB9IsBRJHI2AuDHK3HHI4DFK3K6BGArI8BAAlJ8A8AhJqAlJxJBK3LDLBK9LmBNKCH+ AwBZEZIGcjGM3zKVIRL7BEH8zVKfKojrKlMCQVIk2jKuH/KzH/I5K7MjLFMpMnI5HGBIAvG/I5J9 MvMrK5M/K7LrJ9LxL1DXL5KYp1L/MM80g2wABMzWAGjVDZIc/XDrIiNjKtKxK1MtNBN7MlN9HFJ5 M3J7NDONMnNHLvLzKNKnMS5dNPMAsGpgLKLoAQNswAH8EAGAXGB0EobQAwEEJOblOa1uzJIfNvF9 MXFVMbMfOPODOBOBMxM1K3M7PfPdNFLtNLOZOcfjIOgVPIfeAaD2VGBUEY6ECyFQL+awXWH+HCHa KgW+KoAqDsPu3Mt8rxMO+fOjPTNzMZN3MhPtN/M/PlOJPrPhPvHBPzOXL3KPKhP69Mv4SmYWWs4m QyDGFUGsH+GeHEOkHW204SIED6FmaSBqeZHmQ3Fikw7iPwZWkCSs40Yg0CwTQAxpPUAhPZN5RDS1 PjOHPpRRRFOTP1RZSo5W1skw6MDa6g5omYZW94AYDkN+BOD+NKCoE6SKHy1WA8EOL2Z8cuhGTU/f SlArAu95L2cmiO1xSrQ7PXQ/S/S1RJS9RFUlNBTDRXNPRavw3St25kTXBiQsADBfUFNO/fUC6K+A N+9fTI2SkOfG6YmY9MZWDEFS+0TInFUQSfPI1NQ9MdSzRPUnS5MzRLUdRPUqD02RRaeoseDdSaQn ScQi9G/kPu+e/o+E8UQUoGv/IXMKsOfqZu8O/eV2S89E7QZ7XG8w/edGQYxiTeoSB3FKBQAlRBV9 S3LnWDUjXpN7LCBOAnJ8AyBXKKV0Z7KPKO6i764rXI9wS8DJVq6TAw+hP5MFBJDi0Ofq0i86mQPv XMDRWlBi6QVy6iAYCUEuJIBDESBGA0DNI2AxLjRCAxS4BGAvH5I5J2BLZdZVV9ZtZqDKJIA+DUH+ AiB4ES2a4pDFDE+fVIkg84ZylyTUn7RdYjIZYhQy14fipY9SDTWhSjRg6UAEC+fZGWEfEMBCDbFM AtJ9MzI9LlM8BFZjJ5ZpNBMzLCBJLKH+A+A3ZSAiByEMmcC0FO7Q7eTVSW+ectBY9HQ2oFDTNXak Ns9fIa0hPPVUfqfq5i3OjohIAEC6fYAWB6ERFUBSDxLuAtZSA7JSA/JoH/LPH+BIAtK7bZG4BGAt JCApH+A+AhJPGnJOA0AfH+AiBIDoWuBzc6AICvD8PxUAWJWWj2sfWwsPP9cXWzKc/PGE0jPQvwiE 9kWsAeB+EVI2AaCpdQAbGjGnHkA4AbJPHhI0A2AZI0A6AdI1drG+A6AfG/HhGjfVI1fLfJGpfBGq BGAZHsAvKJA+j7eZencVNSlZGBelINchalStIqbTV7J5I8BDLtdrJPfLfSAZHkA2AXfPfXENJTfl JrK0BFdjMoBOArX7UsYhUxaledgRcZgVBLMTepVU17StSxXlMrZzbgAtLDgrdoAlGiBBK1dhZZLi AxK/OFORRVNNhbhu0PhhafgTW1gXilgbgRhzUbV/OOAwDPFMAxe/ZrPriXXnWJidP3hiuZgPiphk eGA8EUJGWEc3MNhtgdUXSvi5XxS1Z3Zhe/MzbfjNRRUrieVpeejtOg0OseDeNYQexAIEE4GcX2B4 EkRKAMDsNjXSTWcVVVjvi1jzh1WHM/j8AvkAAvkFlHkLjVjdjZF4pPk0nANKLKQIBCESL2D6F6XG XgvWqwHQMqEsGIRaL6RSAgD6bcQAbyNRBkLQDCWNhdNxkPj1V5h3i7N7lLlPlTmtUpjTTHak1tUE DHDyQMDWN+ggQIByEiJOEGOcGIG8Okq7CMXay2KMG8HaU4rcMGCIE8aSBcEWL2Ci4ccggbKXhxlD j3jPPdmxbdLnm3PxNJhZkO1CseseDUNKBIEMJWB8EqhCBqEuGgH+DiFiL+GXQcINAEIhl4McIkXc IICurWWuDulCNiDW85VVi3mplHmuH/j/oZM9odRTohkNNQnjYIc6QIAILmH+BgEOLOtabCUcIPpV qiIOHMKeOQGoJ6C+FaMwAuEJT474DYTfliJQrbgjmrj5OBoXkDobrTJ/LsArX+LQDI72nBlkJQLv C2FBpAEoGOMGGhqsvKH7nlqpqlnpsKKMHgJ+tgGQPKCmFIaSAwD8LOA4EGJWBiCdEFH1G/iPp0Av rXlRrboTdXLDdrI0BSBkD4H/n8JOAkD7o0EwhCECF+PKG6HceBsQRtsPtyOuq4/+22RcFYJGBkBK Ddp4AtI8A2AdHkA/H9uPZZrVp5lNp9OFMzK7iLJ9HhHlgrJ8BSA3FSEWECe+vNsVnkXdt4MVqnvR qiHmHkKgF+F0JqEMD8fECQB4D7uPI8AyAZGjubJ8BAAqClbqAng4AbfZdsKyBTuMDiDOE2H+FmFe JGHMU1vXwrvVwrwxCKagF3pADyDiFEH+CWCAEHJ+A3G4B+BmDmH+DQC7EEFGE4NKHYHUXvwxxqIj t3xtxyIaw0ImGwGus8HgUZx1yGavxxyJyPyRySKIKQFuFibQGOGEYFyVynypymKQCaB/xIDkDOE8 IQFSFEVGE+EsQIIWGEF4L+EwEeP6IOE4EoN2DOC0ErpCDLy6DYDAEyH+FiFU6EHyq6H+DqDUFAH+ G4GwOWIWw0IIEMD68HzsE0H+DgDIE4H+DaDD0cG8G4X2FgFS6EFGE2LAIWG0GsHMH+EoEWTaIOFn uEH+DKCzzkDjzoH+DJ1aH+FuFgbRsGIIEKD2fZyCrEIOGUGMI2E8EqNLz6XmDsDXEgDUC6EwH+DW C9zx1ZzkF4FuOOHUq0D8Dnb8DeDHwd0h0kEeEKQsKQCkCLFKDv2SIQD6DkFJENg2H/ygYFxkPCBu BQDkH+CwCXIsIOCiCJb2CQB0D+H+FEE0LoEYEGTOCgCGEKH/2uMrGfI8GEF7RyIWHwOoH+CEBptU EaEITOFCEyLOF0FoJr16aUDpb9fbH+GAF0RSz2KgCABjdACSB4ED3V3YH+C8CjbCFAEwLP006EG8 G2X35LbnK6B/5iH+GcGSs8FoFehCC0CYEavXAT5WRSDyDdxACKBtvwFwFkJqHEHAUVxeLACSB3HZ 2GNKEyEgP6EsEaNEKQCn3MH+DuDZEgIOEUEAZMDuDbxB7jFL5zbCDQC2Et2cC/zwIP8AH+BoBKDf 1iCxEEC4CfFgEqEYNFCeU4BWA2DSXIGgXGIWu+XmCQB35r2R0EEgEOQyIWEOD88GECDvVmCyCZIs DACpGaE0EkN+DP8GIQEGDyFUH/3LFL0Ue+HkHh18IMHPwoCqCRe54ILoCMBvEKC6Cl8CC18IIWFa FKL2DGCtEEIXE0MH2+H+EKD4fZ10fZ374Z7h7l7p7sIMEF9eRd6d1p1t4H4L/ohCDECsEkIQIAVC MiX+dTUoH++n0/X++4W/4hEHe7Xo/xQGTO/zQXEu/0Cd1S/0Yglg/3g73q/yWPUE/zsa1C/0IeVW /0GeFU/3G4Xa/0MfVY/1or2bQlbRUiiVpJpQ/y+U0jEYggzzOSkRUM/0KfJqfzop3+p1Aw3+53I7 n+UCJWak3m26H+OxYdX+Xiij6leVRY3+YKheaks1Yy3+VCPBEIe5zM5qOxbdICAADgEAAAMAAAAB ADgAAAEBAAMAAAABAEAAAAECAAMAAAAEAAAWdgEDAAMAAAABAAUAAAEGAAMAAAABAAIAAAERAAQA AAABAAAACAEVAAMAAAABAAQAAAEWAAQAAAABAAAAkgEXAAQAAAABAAAVwAEaAAUAAAABAAAWfgEb AAUAAAABAAAWhgEcAAMAAAABAAEAAAEoAAMAAAABAAIAAAFSAAMAAAABAAEAAAAAAAAACAAIAAgA CAAK/IAAACcQAAr8gAAAJxA= --Apple-Mail-29--420135631 Content-Type: multipart/alternative; boundary=Apple-Mail-31--420135629 --Apple-Mail-31--420135629 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > > Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 > > Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 > > Web: http://www.implements.be > > First they ignore you. Then they laugh at you. Then they fight you. =20= > Then you win. > Mahatma Ghandi. > Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements --Apple-Mail-31--420135629 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you.=20 Then you win. Mahatma Ghandi. Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements = --Apple-Mail-31--420135629-- --Apple-Mail-29--420135631 Content-Transfer-Encoding: base64 Content-Type: image/tiff; x-unix-mode=0666; name="Pasted Graphic 2.tiff" Content-Disposition: inline; filename="Pasted Graphic 2.tiff" TU0AKgAAFciAP6BP5/wWDQeEQmFQuGQ2HQ+FP5+v9rOh5P9IMBuP8zK9tP8fJNlP8QoNjP8FndhP 8MnyVjJHSMrqFnv9BL1vP9ett2v98PuJxChUOiUWGQOCUalUujOV4PZ/qBluR/mdVtJ/jVHst/ho 9MF/g08WABHJfv8AWa0WoAHFfWuzgg6sB/hY9ysUohjv8qqVqv9LMZxv9ouV40zEYmHUjFY2HPx+ RNtOl4Rhftt/phiRs6LC/iA5LewnJeWg46W22+02cAHOwa26bC0HPY7TZ7G2G9dv8EHFdP8Om9YP 8rKJov9UtFzv9Dzh/sRvu5/vZ9PzHdeF4zsUqgQRlOR5v9DMBwv8uqe/jZIVwNHu6B0/WAKHvWGr RAA1ra0G1c/vfgANrdraXrTQIAA6JW2TUrQNz/ja0oAjO4YAjSWS0DOWawjys4RkIk4Lj2sAWEUZ B/ieUJqH+PZdnAf5aG06R5nsfLtqW7UaoWZJwukwLyi4VRsn+F5HGYf8QJWBQ7JWAI5tYti2Dm1T bQHAq0De1Q2FwtA1Fof4AjVDIBDQV0vDQWK0DY/oADYWsrNQN81NOuDZtYOM6ycs4Cjq+Q9JWE5D pGJpOpqP5dMwW5sHUf53xlHCExupjqoIbh3Hu55vp6ch3Hof4wlMrAdEWYh/gmP1RtXK0DDi2rXy nVzXjpVs8NM1g1vuNU2gANEugGMBRwu4YADTNsvwyAA3v+0k6LRWLb2dBUpjk2I4NVOx/gYPC6BU QiwCgSsSlWZpxH+ahzsOaBzvCe7qoMpKlUghd1n2f5oHId5/lQaJ0n+RBf3GLpUGwf4iksrg0FVF ICzlXT9QBB0tWO/+FjlKVWtfijTNRBsvDbhrdP3NUAwY38KFfkc0PuNjRWLLz8ytiTUVdVlnjlA2 MWPAWRAAM8LAAMjhhuR5ir4ULjigT5pn+P5co2T6pn+YBup6dp6xoh6kHie95modNOFkah0H+Qhd pyKBNmcf68r2Bo9GGtdpwIAUotM1Q4QMNzUDZhuVZbLsvy7KkqZZCnAb4AFcQY1C3LW1Wb5lOa2t ZvUt78NNjcMN83wEOMDczvuTzXNQ3QEN01WRL2b8XBYEDuugQEFUYgkpIo5FjIJTmhfZkHG8J0nm fB/sifp/EQYLBioUKsBkRKVgWPLX7stA4VnaDa+tZlZel0j+gCMpWS8MRULQNeIDY/4zTIAIyFX0 GFybZnr+rZ8p2a1s6tUNfADXXI1Yb8jLX+MNQk+1Az7zWqwVa/FaJrHpvSLeAQOhdE/l7CSJwrAb BXDXH+PkyA/gUiPGaftASWW6JeDgnGAsCmLNvcYqlz5+FchnZMrqGbeE3JoYhAMAKbEvOGeigtaE CDZwJcgxtiML0uMgQYzllKt2GhtP/EaH6cgAm+fnCtVEVXFOKNU54AC1B/tyLfBx4YJBGwhAAG5m yUoGH/TgxlZ7nGXw3YdEpYTPWOvaSq/JVEX28rEDQsGKDoGPx+QYnE1URg2paZY4dy7EHTx9cc41 upqIfwkjSgJWyW0MgBdK9I1DGAAwnS8G8+4cGYyUWWbKMg/gSiOjQG+FMQ44qqlmgBNUSYYMohKW 2ULFZaR8NdMFucoz/w/ZcAAMzgH+j/AKGV8QAZkhmWNM0AJ/I5zCiJEGYjF1qoGmxI5MqwT7JVlG buLULJhPXlbK+WMt3HwHmIbF8xaA0n6mvPgN7EJhx9Vcax+hsX6wGZuldNDnTSgQDoloGohS6BYE 62gPQqyRhlFE2gIYkCRgdD4WcAUsnxvlYklKgUCn4upi29JkkpG/pbYgtKbr86TLOnbLCF066ZLP kGsI+7NY9rRjZHB+0ew4M5S0AKT4FA8l0BaIUk4XHkj/EYLwzAtVyj/GwRYf49B8D6H+OMdw9R/j DG8vcTgx1xhpFTBoHYilRgaDygQAobXCynVVSSVdRYTLJls5CcIaqX0zpw/Kms76c2DSnTsNNPaA TbpggqOSCg4MkQwcAPiBAzimJqI4Xg3R/itX0P8bY66xD7MiYkgg5h4vAF2NgdY/xNGCH+HQVMIQ WiCNVM2L6Ap4s3ShFixMubA2HnjcUulhab3GuIbGxVjLlxXZoawMKFgsCiaSOUd6nB6XbH+PofKN LtqcH7eMf5AyimQOtegf4/TIEFvYP8dyjTm2eV1KdmNgrlIKr/cOxF+LjmRldTZY88L/U6P/YtyF +b82PDAhkOQsyNkHu8jS76NB4YXH+PvDUG7v3lIFeu8mFbuj6q8PXE2HEaXqIEUEgxnh2T2YhD/B WBZcUunVjO59yMB3PsHc3BONLemxwaP8NwsTMEPvVeMieJKvXmHZk8f48h5GHyYP8e+V3g3tvMQs VY01FLCxjgTHk8b943yBTOVoJEiXJzPgae1zr+2HwYhnIuRzrjczwP8eY8zDmKy7l8NOYcx40zKW nNswizytg+kXHc/qS48x9o7QbNMhZ0yMdvPBG7wmNz/jB6WYs436uFmbHhbFrAFgjhl4Y/g5IvH+ 0EkYFw+NuAHqmoet9bWJwPnDHEfM55E0udjTNW7uZ+y9p6L+oNe3Bxtoa4lACzgQee2kQ5Jwo1RK QPQfK8xejaJ6FG6w/wEh1NZJ+L6ydoH+zfqTUK0dK7AzsY7Yem9jaA0Fu2wVf9eJObqW+TxuwbCS K4JEYVnh2O/vWQM5K+wtCehCCAP5YABybV035xCAJIJqnrTyFqB9H8e1Lu/OumM870MTp3MGn9J8 g5Zsxhk2ZURLfGmrBAEg5m/CISIf4jRekblaCgRoyS0BjWNZVw6uZ70hnsrkND/Fbn3kWWtmz1ID a5NxyLYJ1xujdI3iZTm9dkYy4/P9yFKIcFoDQsbtKW4nT2YbLrpr450V6zTGem7N+YzXVzDZ0PSl ddIb8GQVrlD9sQjy4FKHVy0ZDDeLHCPWuuD/68Y0Vg1LX8c0bQSBiAq9JrVv2pYwZkzq6mqxDi8y XJgBkjZGftMMdUgm5PKyCBnogBI72hYgZlgrD6WWgMzPZkxGdUzfIfIzsdb66PXr/Jxp+XwRGlOL GnuJm2b0l7vo/dJeDL4N7qwWN+qn5CqedNMATuzZmObaad1s4dBMl/6wmG/Pt1+oAAYk2hToiuQc 5FxyjyRoHqH0xYKE+Q8k+UIgH0KAH+HSHqXmGwHSIuEGF0I25eWE8A96/g2akwhIk8Puf25mpi3w /IeG/M8y0O10/WgM3cpUS8l0mS+ocOYgDQS0ASDUTOB6EaJWDkXyXwFyL+GqGkPKHQHMOkxIXmIO HAHAJyyuKgvGIIHgU2H+G0GuHKH+FwF6KwEAtAH+CUEcLOAiNC2QnKcOZSZCmO9ongNep2sA3Y2W nY/KwE9hDazc8w8Sug9mh6cAZ0/UAEC0FKH+AOCOEoJQBeD8H+A2A+DYYGBuD2H+D6DmFCH+F0Fq hDB+PKGcGaaSGsGmJyGOGGg0EuEgP0CqCOEOH+BIA+DQH+AgBSDrD+CAEcH+AICwFNA+pamupfDQ /G1E2asfDksJDesNBCv012x+5WDmQSseeiAaCGEiH+BEAmC0H+A8AcCyH+A4AXGqAqAOCsLqAQCw H+BMAuDIH+CJEWPMCpFKCaCEEASEBFESA2AaC2H+AoAOCqK6AVGqA8AYCvGcAiC9HmBlHZBglBDt F80LF7BNDdBHDg2U0JGI0lGEseWsAgB6EYH+BUAkC4H+BKAuDDGcArH+BCAnH+A8AeC6K7HjG7H4 AoANHtG1G4A1GxGkAfHkBAAjJOBCAqC/FMAuDAH+BOApJ8A0BYDy2aWtGEzI1G2dKRIUwDGDF9BP DooBISnVImB9IsBRJHI2AuDHK3HHI4DFK3K6BGArI8BAAlJ8A8AhJqAlJxJBK3LDLBK9LmBNKCH+ AwBZEZIGcjGM3zKVIRL7BEH8zVKfKojrKlMCQVIk2jKuH/KzH/I5K7MjLFMpMnI5HGBIAvG/I5J9 MvMrK5M/K7LrJ9LxL1DXL5KYp1L/MM80g2wABMzWAGjVDZIc/XDrIiNjKtKxK1MtNBN7MlN9HFJ5 M3J7NDONMnNHLvLzKNKnMS5dNPMAsGpgLKLoAQNswAH8EAGAXGB0EobQAwEEJOblOa1uzJIfNvF9 MXFVMbMfOPODOBOBMxM1K3M7PfPdNFLtNLOZOcfjIOgVPIfeAaD2VGBUEY6ECyFQL+awXWH+HCHa KgW+KoAqDsPu3Mt8rxMO+fOjPTNzMZN3MhPtN/M/PlOJPrPhPvHBPzOXL3KPKhP69Mv4SmYWWs4m QyDGFUGsH+GeHEOkHW204SIED6FmaSBqeZHmQ3Fikw7iPwZWkCSs40Yg0CwTQAxpPUAhPZN5RDS1 PjOHPpRRRFOTP1RZSo5W1skw6MDa6g5omYZW94AYDkN+BOD+NKCoE6SKHy1WA8EOL2Z8cuhGTU/f SlArAu95L2cmiO1xSrQ7PXQ/S/S1RJS9RFUlNBTDRXNPRavw3St25kTXBiQsADBfUFNO/fUC6K+A N+9fTI2SkOfG6YmY9MZWDEFS+0TInFUQSfPI1NQ9MdSzRPUnS5MzRLUdRPUqD02RRaeoseDdSaQn ScQi9G/kPu+e/o+E8UQUoGv/IXMKsOfqZu8O/eV2S89E7QZ7XG8w/edGQYxiTeoSB3FKBQAlRBV9 S3LnWDUjXpN7LCBOAnJ8AyBXKKV0Z7KPKO6i764rXI9wS8DJVq6TAw+hP5MFBJDi0Ofq0i86mQPv XMDRWlBi6QVy6iAYCUEuJIBDESBGA0DNI2AxLjRCAxS4BGAvH5I5J2BLZdZVV9ZtZqDKJIA+DUH+ AiB4ES2a4pDFDE+fVIkg84ZylyTUn7RdYjIZYhQy14fipY9SDTWhSjRg6UAEC+fZGWEfEMBCDbFM AtJ9MzI9LlM8BFZjJ5ZpNBMzLCBJLKH+A+A3ZSAiByEMmcC0FO7Q7eTVSW+ectBY9HQ2oFDTNXak Ns9fIa0hPPVUfqfq5i3OjohIAEC6fYAWB6ERFUBSDxLuAtZSA7JSA/JoH/LPH+BIAtK7bZG4BGAt JCApH+A+AhJPGnJOA0AfH+AiBIDoWuBzc6AICvD8PxUAWJWWj2sfWwsPP9cXWzKc/PGE0jPQvwiE 9kWsAeB+EVI2AaCpdQAbGjGnHkA4AbJPHhI0A2AZI0A6AdI1drG+A6AfG/HhGjfVI1fLfJGpfBGq BGAZHsAvKJA+j7eZencVNSlZGBelINchalStIqbTV7J5I8BDLtdrJPfLfSAZHkA2AXfPfXENJTfl JrK0BFdjMoBOArX7UsYhUxaledgRcZgVBLMTepVU17StSxXlMrZzbgAtLDgrdoAlGiBBK1dhZZLi AxK/OFORRVNNhbhu0PhhafgTW1gXilgbgRhzUbV/OOAwDPFMAxe/ZrPriXXnWJidP3hiuZgPiphk eGA8EUJGWEc3MNhtgdUXSvi5XxS1Z3Zhe/MzbfjNRRUrieVpeejtOg0OseDeNYQexAIEE4GcX2B4 EkRKAMDsNjXSTWcVVVjvi1jzh1WHM/j8AvkAAvkFlHkLjVjdjZF4pPk0nANKLKQIBCESL2D6F6XG XgvWqwHQMqEsGIRaL6RSAgD6bcQAbyNRBkLQDCWNhdNxkPj1V5h3i7N7lLlPlTmtUpjTTHak1tUE DHDyQMDWN+ggQIByEiJOEGOcGIG8Okq7CMXay2KMG8HaU4rcMGCIE8aSBcEWL2Ci4ccggbKXhxlD j3jPPdmxbdLnm3PxNJhZkO1CseseDUNKBIEMJWB8EqhCBqEuGgH+DiFiL+GXQcINAEIhl4McIkXc IICurWWuDulCNiDW85VVi3mplHmuH/j/oZM9odRTohkNNQnjYIc6QIAILmH+BgEOLOtabCUcIPpV qiIOHMKeOQGoJ6C+FaMwAuEJT474DYTfliJQrbgjmrj5OBoXkDobrTJ/LsArX+LQDI72nBlkJQLv C2FBpAEoGOMGGhqsvKH7nlqpqlnpsKKMHgJ+tgGQPKCmFIaSAwD8LOA4EGJWBiCdEFH1G/iPp0Av rXlRrboTdXLDdrI0BSBkD4H/n8JOAkD7o0EwhCECF+PKG6HceBsQRtsPtyOuq4/+22RcFYJGBkBK Ddp4AtI8A2AdHkA/H9uPZZrVp5lNp9OFMzK7iLJ9HhHlgrJ8BSA3FSEWECe+vNsVnkXdt4MVqnvR qiHmHkKgF+F0JqEMD8fECQB4D7uPI8AyAZGjubJ8BAAqClbqAng4AbfZdsKyBTuMDiDOE2H+FmFe JGHMU1vXwrvVwrwxCKagF3pADyDiFEH+CWCAEHJ+A3G4B+BmDmH+DQC7EEFGE4NKHYHUXvwxxqIj t3xtxyIaw0ImGwGus8HgUZx1yGavxxyJyPyRySKIKQFuFibQGOGEYFyVynypymKQCaB/xIDkDOE8 IQFSFEVGE+EsQIIWGEF4L+EwEeP6IOE4EoN2DOC0ErpCDLy6DYDAEyH+FiFU6EHyq6H+DqDUFAH+ G4GwOWIWw0IIEMD68HzsE0H+DgDIE4H+DaDD0cG8G4X2FgFS6EFGE2LAIWG0GsHMH+EoEWTaIOFn uEH+DKCzzkDjzoH+DJ1aH+FuFgbRsGIIEKD2fZyCrEIOGUGMI2E8EqNLz6XmDsDXEgDUC6EwH+DW C9zx1ZzkF4FuOOHUq0D8Dnb8DeDHwd0h0kEeEKQsKQCkCLFKDv2SIQD6DkFJENg2H/ygYFxkPCBu BQDkH+CwCXIsIOCiCJb2CQB0D+H+FEE0LoEYEGTOCgCGEKH/2uMrGfI8GEF7RyIWHwOoH+CEBptU EaEITOFCEyLOF0FoJr16aUDpb9fbH+GAF0RSz2KgCABjdACSB4ED3V3YH+C8CjbCFAEwLP006EG8 G2X35LbnK6B/5iH+GcGSs8FoFehCC0CYEavXAT5WRSDyDdxACKBtvwFwFkJqHEHAUVxeLACSB3HZ 2GNKEyEgP6EsEaNEKQCn3MH+DuDZEgIOEUEAZMDuDbxB7jFL5zbCDQC2Et2cC/zwIP8AH+BoBKDf 1iCxEEC4CfFgEqEYNFCeU4BWA2DSXIGgXGIWu+XmCQB35r2R0EEgEOQyIWEOD88GECDvVmCyCZIs DACpGaE0EkN+DP8GIQEGDyFUH/3LFL0Ue+HkHh18IMHPwoCqCRe54ILoCMBvEKC6Cl8CC18IIWFa FKL2DGCtEEIXE0MH2+H+EKD4fZ10fZ374Z7h7l7p7sIMEF9eRd6d1p1t4H4L/ohCDECsEkIQIAVC MiX+dTUoH++n0/X++4W/4hEHe7Xo/xQGTO/zQXEu/0Cd1S/0Yglg/3g73q/yWPUE/zsa1C/0IeVW /0GeFU/3G4Xa/0MfVY/1or2bQlbRUiiVpJpQ/y+U0jEYggzzOSkRUM/0KfJqfzop3+p1Aw3+53I7 n+UCJWak3m26H+OxYdX+Xiij6leVRY3+YKheaks1Yy3+VCPBEIe5zM5qOxbdICAADgEAAAMAAAAB ADgAAAEBAAMAAAABAEAAAAECAAMAAAAEAAAWdgEDAAMAAAABAAUAAAEGAAMAAAABAAIAAAERAAQA AAABAAAACAEVAAMAAAABAAQAAAEWAAQAAAABAAAAkgEXAAQAAAABAAAVwAEaAAUAAAABAAAWfgEb AAUAAAABAAAWhgEcAAMAAAABAAEAAAEoAAMAAAABAAIAAAFSAAMAAAABAAEAAAAAAAAACAAIAAgA CAAK/IAAACcQAAr8gAAAJxA= --Apple-Mail-29--420135631 Content-Type: multipart/alternative; boundary=Apple-Mail-32--420135628 --Apple-Mail-32--420135628 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-32--420135628 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-32--420135628-- --Apple-Mail-29--420135631-- From pgsql-performance-owner@postgresql.org Tue Jul 19 08:00:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C57DF52914 for ; Tue, 19 Jul 2005 08:00:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95493-02 for ; Tue, 19 Jul 2005 11:00:41 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id 4564E5293E for ; Tue, 19 Jul 2005 08:00:37 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id 517294183D5; Tue, 19 Jul 2005 12:00:20 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id 9600F15EE6; Tue, 19 Jul 2005 11:51:55 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 04342-02; Tue, 19 Jul 2005 11:51:52 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 1CD3815EDA; Tue, 19 Jul 2005 11:51:52 +0100 (BST) Message-ID: <42DCDB47.3060009@archonet.com> Date: Tue, 19 Jul 2005 11:51:51 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Yves Vindevogel Cc: 'Postgresql Performance' Subject: Re: Insert performance (OT?) References: <42DCCA3B.8070100@archonet.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/278 X-Sequence-Number: 13519 Yves Vindevogel wrote: >>> So, I must use a function that will check against u1 and u2, and then >>> insert if it is ok. >>> I know that such a function is way slower that my insert query. >> >> So - you have a table, called something like "upload" with 20,000 rows >> and you'd like to know whether it is safe to insert them. Well, it's >> easy enough to identify which ones are duplicates. >> >> SELECT * FROM upload JOIN main_table ON u1=f1 AND u2=f2 AND u3=f3; >> SELECT * FROM upload JOIN main_table ON u1=f1 AND u2=f2 AND u3=f4; >> > That is a good idea. I can delete the ones that would fail my first > unique index this way, and then delete the ones that would fail my > second unique index and then upload them. > Hmm, why did I not think of that myself. I've spent a lot of time moving data from one system to another, usually having to clean it in the process. At 9pm on a Friday, you decide that on the next job you'll find an efficient way to do it :-) >> Are you saying that deleting these rows and then inserting takes too >> long? >> > This goes very fast, but not with a function that checks each record one > by one. You could get away with one query if you converted them to left-joins: INSERT INTO ... SELECT * FROM upload LEFT JOIN ... WHERE f3 IS NULL UNION SELECT * FROM upload LEFT JOIN ... WHERE f4 IS NULL The UNION will remove duplicates for you, but this might turn out to be slower than two separate queries. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Tue Jul 19 10:38:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5FEC7528C3 for ; Tue, 19 Jul 2005 10:38:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 33527-02 for ; Tue, 19 Jul 2005 13:38:35 +0000 (GMT) Received: from apate.telenet-ops.be (apate.telenet-ops.be [195.130.132.57]) by svr1.postgresql.org (Postfix) with ESMTP id E7B2852818 for ; Tue, 19 Jul 2005 10:38:34 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by apate.telenet-ops.be (Postfix) with SMTP id 7872F380DD for ; Tue, 19 Jul 2005 15:38:37 +0200 (CEST) Received: from [10.0.1.2] (d5152B1F3.access.telenet.be [81.82.177.243]) by apate.telenet-ops.be (Postfix) with ESMTP id DA491380F2 for ; Tue, 19 Jul 2005 15:38:36 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v622) In-Reply-To: <42DCDB47.3060009@archonet.com> References: <42DCCA3B.8070100@archonet.com> <42DCDB47.3060009@archonet.com> Content-Type: multipart/mixed; boundary=Apple-Mail-62--408288062 Message-Id: <646c7b05563f9d76b5492de416f7432a@implements.be> From: Yves Vindevogel Subject: Re: Insert performance (OT?) Date: Tue, 19 Jul 2005 15:38:36 +0200 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.622) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/279 X-Sequence-Number: 13520 --Apple-Mail-62--408288062 Content-Type: multipart/alternative; boundary=Apple-Mail-63--408288062 --Apple-Mail-63--408288062 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1; format=flowed I will use 2 queries. They run within a function fnUpload(), so I'm=20 going to keep it simple. On 19 Jul 2005, at 12:51, Richard Huxton wrote: > Yves Vindevogel wrote: > >>> So, I must use a function that will check against u1 and u2, and=20= > then >>>> insert if it is ok. >>>> I know that such a function is way slower that my insert query. >>> >>> So - you have a table, called something like "upload" with 20,000=20 >>> rows and you'd like to know whether it is safe to insert them. Well,=20= >>> it's easy enough to identify which ones are duplicates. >>> >>> SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND = u3=3Df3; >>> SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND = u3=3Df4; >>> >> That is a good idea. I can delete the ones that would fail my first=20= >> unique index this way, and then delete the ones that would fail my=20 >> second unique index and then upload them. >> Hmm, why did I not think of that myself. > > I've spent a lot of time moving data from one system to another,=20 > usually having to clean it in the process. At 9pm on a Friday, you=20 > decide that on the next job you'll find an efficient way to do it :-) > >>> Are you saying that deleting these rows and then inserting takes too=20= >>> long? >>> >> This goes very fast, but not with a function that checks each record=20= >> one by one. > > You could get away with one query if you converted them to left-joins: > INSERT INTO ... > SELECT * FROM upload LEFT JOIN ... WHERE f3 IS NULL > UNION > SELECT * FROM upload LEFT JOIN ... WHERE f4 IS NULL > > The UNION will remove duplicates for you, but this might turn out to=20= > be slower than two separate queries. > > -- > Richard Huxton > Archonet Ltd > > Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements --Apple-Mail-63--408288062 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 I will use 2 queries. They run within a function fnUpload(), so I'm going to keep it simple. On 19 Jul 2005, at 12:51, Richard Huxton wrote: Yves Vindevogel wrote: >>> So, I must use a function that will check against u1 and u2, and then insert if it is ok. I know that such a function is way slower that my insert query. So - you have a table, called something like "upload" with 20,000 rows and you'd like to know whether it is safe to insert them. Well, it's easy enough to identify which ones are duplicates. SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND u3=3Df3; SELECT * FROM upload JOIN main_table ON u1=3Df1 AND u2=3Df2 AND u3=3Df4; That is a good idea. I can delete the ones that would fail my first unique index this way, and then delete the ones that would fail my second unique index and then upload them. Hmm, why did I not think of that myself. I've spent a lot of time moving data from one system to another, usually having to clean it in the process. At 9pm on a Friday, you decide that on the next job you'll find an efficient way to do it :-) Are you saying that deleting these rows and then inserting takes too long? This goes very fast, but not with a function that checks each record one by one. You could get away with one query if you converted them to left-joins: INSERT INTO ... SELECT * FROM upload LEFT JOIN ... WHERE f3 IS NULL UNION SELECT * FROM upload LEFT JOIN ... WHERE f4 IS NULL The UNION will remove duplicates for you, but this might turn out to be slower than two separate queries. -- Richard Huxton Archonet Ltd Met vriendelijke groeten, Bien =E0 vous, Kind regards, Yves Vindevogel Implements = --Apple-Mail-63--408288062-- --Apple-Mail-62--408288062 Content-Transfer-Encoding: base64 Content-Type: image/tiff; x-unix-mode=0666; name="Pasted Graphic 2.tiff" Content-Disposition: inline; filename="Pasted Graphic 2.tiff" TU0AKgAAFciAP6BP5/wWDQeEQmFQuGQ2HQ+FP5+v9rOh5P9IMBuP8zK9tP8fJNlP8QoNjP8FndhP 8MnyVjJHSMrqFnv9BL1vP9ett2v98PuJxChUOiUWGQOCUalUujOV4PZ/qBluR/mdVtJ/jVHst/ho 9MF/g08WABHJfv8AWa0WoAHFfWuzgg6sB/hY9ysUohjv8qqVqv9LMZxv9ouV40zEYmHUjFY2HPx+ RNtOl4Rhftt/phiRs6LC/iA5LewnJeWg46W22+02cAHOwa26bC0HPY7TZ7G2G9dv8EHFdP8Om9YP 8rKJov9UtFzv9Dzh/sRvu5/vZ9PzHdeF4zsUqgQRlOR5v9DMBwv8uqe/jZIVwNHu6B0/WAKHvWGr RAA1ra0G1c/vfgANrdraXrTQIAA6JW2TUrQNz/ja0oAjO4YAjSWS0DOWawjys4RkIk4Lj2sAWEUZ B/ieUJqH+PZdnAf5aG06R5nsfLtqW7UaoWZJwukwLyi4VRsn+F5HGYf8QJWBQ7JWAI5tYti2Dm1T bQHAq0De1Q2FwtA1Fof4AjVDIBDQV0vDQWK0DY/oADYWsrNQN81NOuDZtYOM6ycs4Cjq+Q9JWE5D pGJpOpqP5dMwW5sHUf53xlHCExupjqoIbh3Hu55vp6ch3Hof4wlMrAdEWYh/gmP1RtXK0DDi2rXy nVzXjpVs8NM1g1vuNU2gANEugGMBRwu4YADTNsvwyAA3v+0k6LRWLb2dBUpjk2I4NVOx/gYPC6BU QiwCgSsSlWZpxH+ahzsOaBzvCe7qoMpKlUghd1n2f5oHId5/lQaJ0n+RBf3GLpUGwf4iksrg0FVF ICzlXT9QBB0tWO/+FjlKVWtfijTNRBsvDbhrdP3NUAwY38KFfkc0PuNjRWLLz8ytiTUVdVlnjlA2 MWPAWRAAM8LAAMjhhuR5ir4ULjigT5pn+P5co2T6pn+YBup6dp6xoh6kHie95modNOFkah0H+Qhd pyKBNmcf68r2Bo9GGtdpwIAUotM1Q4QMNzUDZhuVZbLsvy7KkqZZCnAb4AFcQY1C3LW1Wb5lOa2t ZvUt78NNjcMN83wEOMDczvuTzXNQ3QEN01WRL2b8XBYEDuugQEFUYgkpIo5FjIJTmhfZkHG8J0nm fB/sifp/EQYLBioUKsBkRKVgWPLX7stA4VnaDa+tZlZel0j+gCMpWS8MRULQNeIDY/4zTIAIyFX0 GFybZnr+rZ8p2a1s6tUNfADXXI1Yb8jLX+MNQk+1Az7zWqwVa/FaJrHpvSLeAQOhdE/l7CSJwrAb BXDXH+PkyA/gUiPGaftASWW6JeDgnGAsCmLNvcYqlz5+FchnZMrqGbeE3JoYhAMAKbEvOGeigtaE CDZwJcgxtiML0uMgQYzllKt2GhtP/EaH6cgAm+fnCtVEVXFOKNU54AC1B/tyLfBx4YJBGwhAAG5m yUoGH/TgxlZ7nGXw3YdEpYTPWOvaSq/JVEX28rEDQsGKDoGPx+QYnE1URg2paZY4dy7EHTx9cc41 upqIfwkjSgJWyW0MgBdK9I1DGAAwnS8G8+4cGYyUWWbKMg/gSiOjQG+FMQ44qqlmgBNUSYYMohKW 2ULFZaR8NdMFucoz/w/ZcAAMzgH+j/AKGV8QAZkhmWNM0AJ/I5zCiJEGYjF1qoGmxI5MqwT7JVlG buLULJhPXlbK+WMt3HwHmIbF8xaA0n6mvPgN7EJhx9Vcax+hsX6wGZuldNDnTSgQDoloGohS6BYE 62gPQqyRhlFE2gIYkCRgdD4WcAUsnxvlYklKgUCn4upi29JkkpG/pbYgtKbr86TLOnbLCF066ZLP kGsI+7NY9rRjZHB+0ew4M5S0AKT4FA8l0BaIUk4XHkj/EYLwzAtVyj/GwRYf49B8D6H+OMdw9R/j DG8vcTgx1xhpFTBoHYilRgaDygQAobXCynVVSSVdRYTLJls5CcIaqX0zpw/Kms76c2DSnTsNNPaA TbpggqOSCg4MkQwcAPiBAzimJqI4Xg3R/itX0P8bY66xD7MiYkgg5h4vAF2NgdY/xNGCH+HQVMIQ WiCNVM2L6Ap4s3ShFixMubA2HnjcUulhab3GuIbGxVjLlxXZoawMKFgsCiaSOUd6nB6XbH+PofKN LtqcH7eMf5AyimQOtegf4/TIEFvYP8dyjTm2eV1KdmNgrlIKr/cOxF+LjmRldTZY88L/U6P/YtyF +b82PDAhkOQsyNkHu8jS76NB4YXH+PvDUG7v3lIFeu8mFbuj6q8PXE2HEaXqIEUEgxnh2T2YhD/B WBZcUunVjO59yMB3PsHc3BONLemxwaP8NwsTMEPvVeMieJKvXmHZk8f48h5GHyYP8e+V3g3tvMQs VY01FLCxjgTHk8b943yBTOVoJEiXJzPgae1zr+2HwYhnIuRzrjczwP8eY8zDmKy7l8NOYcx40zKW nNswizytg+kXHc/qS48x9o7QbNMhZ0yMdvPBG7wmNz/jB6WYs436uFmbHhbFrAFgjhl4Y/g5IvH+ 0EkYFw+NuAHqmoet9bWJwPnDHEfM55E0udjTNW7uZ+y9p6L+oNe3Bxtoa4lACzgQee2kQ5Jwo1RK QPQfK8xejaJ6FG6w/wEh1NZJ+L6ydoH+zfqTUK0dK7AzsY7Yem9jaA0Fu2wVf9eJObqW+TxuwbCS K4JEYVnh2O/vWQM5K+wtCehCCAP5YABybV035xCAJIJqnrTyFqB9H8e1Lu/OumM870MTp3MGn9J8 g5Zsxhk2ZURLfGmrBAEg5m/CISIf4jRekblaCgRoyS0BjWNZVw6uZ70hnsrkND/Fbn3kWWtmz1ID a5NxyLYJ1xujdI3iZTm9dkYy4/P9yFKIcFoDQsbtKW4nT2YbLrpr450V6zTGem7N+YzXVzDZ0PSl ddIb8GQVrlD9sQjy4FKHVy0ZDDeLHCPWuuD/68Y0Vg1LX8c0bQSBiAq9JrVv2pYwZkzq6mqxDi8y XJgBkjZGftMMdUgm5PKyCBnogBI72hYgZlgrD6WWgMzPZkxGdUzfIfIzsdb66PXr/Jxp+XwRGlOL GnuJm2b0l7vo/dJeDL4N7qwWN+qn5CqedNMATuzZmObaad1s4dBMl/6wmG/Pt1+oAAYk2hToiuQc 5FxyjyRoHqH0xYKE+Q8k+UIgH0KAH+HSHqXmGwHSIuEGF0I25eWE8A96/g2akwhIk8Puf25mpi3w /IeG/M8y0O10/WgM3cpUS8l0mS+ocOYgDQS0ASDUTOB6EaJWDkXyXwFyL+GqGkPKHQHMOkxIXmIO HAHAJyyuKgvGIIHgU2H+G0GuHKH+FwF6KwEAtAH+CUEcLOAiNC2QnKcOZSZCmO9ongNep2sA3Y2W nY/KwE9hDazc8w8Sug9mh6cAZ0/UAEC0FKH+AOCOEoJQBeD8H+A2A+DYYGBuD2H+D6DmFCH+F0Fq hDB+PKGcGaaSGsGmJyGOGGg0EuEgP0CqCOEOH+BIA+DQH+AgBSDrD+CAEcH+AICwFNA+pamupfDQ /G1E2asfDksJDesNBCv012x+5WDmQSseeiAaCGEiH+BEAmC0H+A8AcCyH+A4AXGqAqAOCsLqAQCw H+BMAuDIH+CJEWPMCpFKCaCEEASEBFESA2AaC2H+AoAOCqK6AVGqA8AYCvGcAiC9HmBlHZBglBDt F80LF7BNDdBHDg2U0JGI0lGEseWsAgB6EYH+BUAkC4H+BKAuDDGcArH+BCAnH+A8AeC6K7HjG7H4 AoANHtG1G4A1GxGkAfHkBAAjJOBCAqC/FMAuDAH+BOApJ8A0BYDy2aWtGEzI1G2dKRIUwDGDF9BP DooBISnVImB9IsBRJHI2AuDHK3HHI4DFK3K6BGArI8BAAlJ8A8AhJqAlJxJBK3LDLBK9LmBNKCH+ AwBZEZIGcjGM3zKVIRL7BEH8zVKfKojrKlMCQVIk2jKuH/KzH/I5K7MjLFMpMnI5HGBIAvG/I5J9 MvMrK5M/K7LrJ9LxL1DXL5KYp1L/MM80g2wABMzWAGjVDZIc/XDrIiNjKtKxK1MtNBN7MlN9HFJ5 M3J7NDONMnNHLvLzKNKnMS5dNPMAsGpgLKLoAQNswAH8EAGAXGB0EobQAwEEJOblOa1uzJIfNvF9 MXFVMbMfOPODOBOBMxM1K3M7PfPdNFLtNLOZOcfjIOgVPIfeAaD2VGBUEY6ECyFQL+awXWH+HCHa KgW+KoAqDsPu3Mt8rxMO+fOjPTNzMZN3MhPtN/M/PlOJPrPhPvHBPzOXL3KPKhP69Mv4SmYWWs4m QyDGFUGsH+GeHEOkHW204SIED6FmaSBqeZHmQ3Fikw7iPwZWkCSs40Yg0CwTQAxpPUAhPZN5RDS1 PjOHPpRRRFOTP1RZSo5W1skw6MDa6g5omYZW94AYDkN+BOD+NKCoE6SKHy1WA8EOL2Z8cuhGTU/f SlArAu95L2cmiO1xSrQ7PXQ/S/S1RJS9RFUlNBTDRXNPRavw3St25kTXBiQsADBfUFNO/fUC6K+A N+9fTI2SkOfG6YmY9MZWDEFS+0TInFUQSfPI1NQ9MdSzRPUnS5MzRLUdRPUqD02RRaeoseDdSaQn ScQi9G/kPu+e/o+E8UQUoGv/IXMKsOfqZu8O/eV2S89E7QZ7XG8w/edGQYxiTeoSB3FKBQAlRBV9 S3LnWDUjXpN7LCBOAnJ8AyBXKKV0Z7KPKO6i764rXI9wS8DJVq6TAw+hP5MFBJDi0Ofq0i86mQPv XMDRWlBi6QVy6iAYCUEuJIBDESBGA0DNI2AxLjRCAxS4BGAvH5I5J2BLZdZVV9ZtZqDKJIA+DUH+ AiB4ES2a4pDFDE+fVIkg84ZylyTUn7RdYjIZYhQy14fipY9SDTWhSjRg6UAEC+fZGWEfEMBCDbFM AtJ9MzI9LlM8BFZjJ5ZpNBMzLCBJLKH+A+A3ZSAiByEMmcC0FO7Q7eTVSW+ectBY9HQ2oFDTNXak Ns9fIa0hPPVUfqfq5i3OjohIAEC6fYAWB6ERFUBSDxLuAtZSA7JSA/JoH/LPH+BIAtK7bZG4BGAt JCApH+A+AhJPGnJOA0AfH+AiBIDoWuBzc6AICvD8PxUAWJWWj2sfWwsPP9cXWzKc/PGE0jPQvwiE 9kWsAeB+EVI2AaCpdQAbGjGnHkA4AbJPHhI0A2AZI0A6AdI1drG+A6AfG/HhGjfVI1fLfJGpfBGq BGAZHsAvKJA+j7eZencVNSlZGBelINchalStIqbTV7J5I8BDLtdrJPfLfSAZHkA2AXfPfXENJTfl JrK0BFdjMoBOArX7UsYhUxaledgRcZgVBLMTepVU17StSxXlMrZzbgAtLDgrdoAlGiBBK1dhZZLi AxK/OFORRVNNhbhu0PhhafgTW1gXilgbgRhzUbV/OOAwDPFMAxe/ZrPriXXnWJidP3hiuZgPiphk eGA8EUJGWEc3MNhtgdUXSvi5XxS1Z3Zhe/MzbfjNRRUrieVpeejtOg0OseDeNYQexAIEE4GcX2B4 EkRKAMDsNjXSTWcVVVjvi1jzh1WHM/j8AvkAAvkFlHkLjVjdjZF4pPk0nANKLKQIBCESL2D6F6XG XgvWqwHQMqEsGIRaL6RSAgD6bcQAbyNRBkLQDCWNhdNxkPj1V5h3i7N7lLlPlTmtUpjTTHak1tUE DHDyQMDWN+ggQIByEiJOEGOcGIG8Okq7CMXay2KMG8HaU4rcMGCIE8aSBcEWL2Ci4ccggbKXhxlD j3jPPdmxbdLnm3PxNJhZkO1CseseDUNKBIEMJWB8EqhCBqEuGgH+DiFiL+GXQcINAEIhl4McIkXc IICurWWuDulCNiDW85VVi3mplHmuH/j/oZM9odRTohkNNQnjYIc6QIAILmH+BgEOLOtabCUcIPpV qiIOHMKeOQGoJ6C+FaMwAuEJT474DYTfliJQrbgjmrj5OBoXkDobrTJ/LsArX+LQDI72nBlkJQLv C2FBpAEoGOMGGhqsvKH7nlqpqlnpsKKMHgJ+tgGQPKCmFIaSAwD8LOA4EGJWBiCdEFH1G/iPp0Av rXlRrboTdXLDdrI0BSBkD4H/n8JOAkD7o0EwhCECF+PKG6HceBsQRtsPtyOuq4/+22RcFYJGBkBK Ddp4AtI8A2AdHkA/H9uPZZrVp5lNp9OFMzK7iLJ9HhHlgrJ8BSA3FSEWECe+vNsVnkXdt4MVqnvR qiHmHkKgF+F0JqEMD8fECQB4D7uPI8AyAZGjubJ8BAAqClbqAng4AbfZdsKyBTuMDiDOE2H+FmFe JGHMU1vXwrvVwrwxCKagF3pADyDiFEH+CWCAEHJ+A3G4B+BmDmH+DQC7EEFGE4NKHYHUXvwxxqIj t3xtxyIaw0ImGwGus8HgUZx1yGavxxyJyPyRySKIKQFuFibQGOGEYFyVynypymKQCaB/xIDkDOE8 IQFSFEVGE+EsQIIWGEF4L+EwEeP6IOE4EoN2DOC0ErpCDLy6DYDAEyH+FiFU6EHyq6H+DqDUFAH+ G4GwOWIWw0IIEMD68HzsE0H+DgDIE4H+DaDD0cG8G4X2FgFS6EFGE2LAIWG0GsHMH+EoEWTaIOFn uEH+DKCzzkDjzoH+DJ1aH+FuFgbRsGIIEKD2fZyCrEIOGUGMI2E8EqNLz6XmDsDXEgDUC6EwH+DW C9zx1ZzkF4FuOOHUq0D8Dnb8DeDHwd0h0kEeEKQsKQCkCLFKDv2SIQD6DkFJENg2H/ygYFxkPCBu BQDkH+CwCXIsIOCiCJb2CQB0D+H+FEE0LoEYEGTOCgCGEKH/2uMrGfI8GEF7RyIWHwOoH+CEBptU EaEITOFCEyLOF0FoJr16aUDpb9fbH+GAF0RSz2KgCABjdACSB4ED3V3YH+C8CjbCFAEwLP006EG8 G2X35LbnK6B/5iH+GcGSs8FoFehCC0CYEavXAT5WRSDyDdxACKBtvwFwFkJqHEHAUVxeLACSB3HZ 2GNKEyEgP6EsEaNEKQCn3MH+DuDZEgIOEUEAZMDuDbxB7jFL5zbCDQC2Et2cC/zwIP8AH+BoBKDf 1iCxEEC4CfFgEqEYNFCeU4BWA2DSXIGgXGIWu+XmCQB35r2R0EEgEOQyIWEOD88GECDvVmCyCZIs DACpGaE0EkN+DP8GIQEGDyFUH/3LFL0Ue+HkHh18IMHPwoCqCRe54ILoCMBvEKC6Cl8CC18IIWFa FKL2DGCtEEIXE0MH2+H+EKD4fZ10fZ374Z7h7l7p7sIMEF9eRd6d1p1t4H4L/ohCDECsEkIQIAVC MiX+dTUoH++n0/X++4W/4hEHe7Xo/xQGTO/zQXEu/0Cd1S/0Yglg/3g73q/yWPUE/zsa1C/0IeVW /0GeFU/3G4Xa/0MfVY/1or2bQlbRUiiVpJpQ/y+U0jEYggzzOSkRUM/0KfJqfzop3+p1Aw3+53I7 n+UCJWak3m26H+OxYdX+Xiij6leVRY3+YKheaks1Yy3+VCPBEIe5zM5qOxbdICAADgEAAAMAAAAB ADgAAAEBAAMAAAABAEAAAAECAAMAAAAEAAAWdgEDAAMAAAABAAUAAAEGAAMAAAABAAIAAAERAAQA AAABAAAACAEVAAMAAAABAAQAAAEWAAQAAAABAAAAkgEXAAQAAAABAAAVwAEaAAUAAAABAAAWfgEb AAUAAAABAAAWhgEcAAMAAAABAAEAAAEoAAMAAAABAAIAAAFSAAMAAAABAAEAAAAAAAAACAAIAAgA CAAK/IAAACcQAAr8gAAAJxA= --Apple-Mail-62--408288062 Content-Type: multipart/alternative; boundary=Apple-Mail-64--408288060 --Apple-Mail-64--408288060 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-64--408288060 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Mail: yves.vindevogel@implements.be - Mobile: +32 (478) 80 82 91 Kempische Steenweg 206 - 3500 Hasselt - Tel-Fax: +32 (11) 43 55 76 Web: http://www.implements.be First they ignore you. Then they laugh at you. Then they fight you. Then you win. Mahatma Ghandi. --Apple-Mail-64--408288060-- --Apple-Mail-62--408288062-- From pgsql-performance-owner@postgresql.org Tue Jul 19 11:49:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AD4855286A for ; Tue, 19 Jul 2005 11:49:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 49865-01 for ; Tue, 19 Jul 2005 14:49:06 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.200]) by svr1.postgresql.org (Postfix) with ESMTP id F15A952995 for ; Tue, 19 Jul 2005 11:49:05 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i35so1210895wra for ; Tue, 19 Jul 2005 07:49:10 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=mhQ0B63ZLQYyzMAVfvQujUqo4yn7S6vRV/eccVvASR+wKrY/MBh3j3eotGoofM1GNySAFXemSreRhyrao6Qh0bpJakI4mo/hOePvdnfIcGL38armR0gnx0oiLZYNIRQWvyd7345NuzWbiIUQLiMyYg11VMErjTgoHK30N6/z9vA= Received: by 10.54.50.62 with SMTP id x62mr752726wrx; Tue, 19 Jul 2005 07:48:42 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 07:48:42 -0700 (PDT) Message-ID: <59d991c405071907482bb0689b@mail.gmail.com> Date: Tue, 19 Jul 2005 10:48:42 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <16794.1121718743@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.164 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/280 X-Sequence-Number: 13521 On 7/18/05, Tom Lane wrote: > > The table has 15 columns, 5 indexes (character, inet and timestamp). > > No foreign keys. The only other thing running on the machine was the > > application actually DOING the benchmarking, written in Python > > (psycopg), but it was, according to top, using less than 1% of the > > CPU. It was just talking through a pipe to a psql prompt to do the > > COPY. >=20 > Sounds pretty plain-vanilla all right. >=20 > Are you in a position to try the same benchmark against CVS tip? > (The nightly snapshot tarball would be plenty close enough.) I'm > just wondering if the old bgwriter behavior of locking down the > bufmgr while it examined the ARC/2Q data structures is causing this... Tom, It looks like the CVS HEAD is definately "better," but not by a huge amount. The only difference is I wasn't run autovacuum in the background (default settings), but I don't think this explains it.=20 Here's a graph of the differences and density of behavior: http://blog.amber.org/diagrams/pgsql_copy_803_cvs.png I can provide the raw data. Each COPY was 500 rows. Note that fsync is turned off here. Maybe it'd be more stable with it turned on? Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 12:04:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 547DE52997 for ; Tue, 19 Jul 2005 12:04:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 48921-10 for ; Tue, 19 Jul 2005 15:04:40 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id BAF5D52989 for ; Tue, 19 Jul 2005 12:04:38 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JF4UHp024015; Tue, 19 Jul 2005 11:04:30 -0400 (EDT) To: Christopher Petrilli Cc: Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c405071907482bb0689b@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Tue, 19 Jul 2005 10:48:42 -0400" Date: Tue, 19 Jul 2005 11:04:30 -0400 Message-ID: <24014.1121785470@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/281 X-Sequence-Number: 13522 Christopher Petrilli writes: > Here's a graph of the differences and density of behavior: > http://blog.amber.org/diagrams/pgsql_copy_803_cvs.png > I can provide the raw data. How about the complete test case? There's something awfully odd going on there, and I'd like to find out what. > Note that fsync is turned off here. Maybe it'd be more stable with it > turned on? Hard to say. I was about to ask if you'd experimented with altering configuration parameters such as shared_buffers or checkpoint_segments to see if you can move the point of onset of slowdown. I'm thinking the behavioral change might be associated with running out of free buffers or some such. (Are you running these tests under a freshly- started postmaster, or one that's been busy for awhile?) regards, tom lane From pgsql-patches-owner@postgresql.org Tue Jul 19 12:23:25 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1654D52A15 for ; Tue, 19 Jul 2005 12:23:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54593-07 for ; Tue, 19 Jul 2005 15:23:22 +0000 (GMT) Received: from trolak.mydnsbox2.com (ns1.mydnsbox2.com [207.44.142.118]) by svr1.postgresql.org (Postfix) with ESMTP id 8052F52A14 for ; Tue, 19 Jul 2005 12:23:21 -0300 (ADT) Received: from [192.168.1.103] (cpe-024-211-165-134.nc.res.rr.com [24.211.165.134]) (authenticated (0 bits)) by trolak.mydnsbox2.com (8.11.6/8.11.6) with ESMTP id j6JEbo528179; Tue, 19 Jul 2005 09:37:50 -0500 Message-ID: <42DD1AE6.6020204@dunslane.net> Date: Tue, 19 Jul 2005 11:23:18 -0400 From: Andrew Dunstan User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Fedora/1.7.8-1.3.1 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Alon Goldshuv Cc: pgsql-patches@postgresql.org Subject: Re: COPY FROM performance improvements References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.04 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/377 X-Sequence-Number: 16717 Alon Goldshuv wrote: >I revisited my patch and removed the code duplications that were there, and >added support for CSV with buffered input, so CSV now runs faster too >(although it is not as optimized as the TEXT format parsing). So now >TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. > >Patch attached. > > > I do not have time to review this 2900 line patch analytically, nor to benchmark it. I have done some functional testing of it on Windows, and tried to break it in text and CSV modes, and with both Unix and Windows type line endings - I have not observed any breakage. This does need lots of eyeballs, though. cheers andrew From pgsql-performance-owner@postgresql.org Tue Jul 19 12:33:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DF32D52A22 for ; Tue, 19 Jul 2005 12:33:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 59115-05 for ; Tue, 19 Jul 2005 15:33:25 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 39BC0529DE for ; Tue, 19 Jul 2005 12:33:23 -0300 (ADT) Received: (qmail 10631 invoked from network); 19 Jul 2005 17:33:57 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 17:33:57 +0200 To: "Christopher Petrilli" , "Tom Lane" Cc: "Vivek Khera" , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> Message-ID: Date: Tue, 19 Jul 2005 17:33:21 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <59d991c405071907482bb0689b@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/282 X-Sequence-Number: 13523 What happens if, say at iteration 6000 (a bit after the mess starts), you pause it for a few minutes and resume. Will it restart with a plateau like at the beginning of the test ? or not ? What if, during this pause, you disconnect and reconnect, or restart the postmaster, or vacuum, or analyze ? > On 7/18/05, Tom Lane wrote: >> > The table has 15 columns, 5 indexes (character, inet and timestamp). >> > No foreign keys. The only other thing running on the machine was the >> > application actually DOING the benchmarking, written in Python >> > (psycopg), but it was, according to top, using less than 1% of the >> > CPU. It was just talking through a pipe to a psql prompt to do the >> > COPY. >> >> Sounds pretty plain-vanilla all right. >> >> Are you in a position to try the same benchmark against CVS tip? >> (The nightly snapshot tarball would be plenty close enough.) I'm >> just wondering if the old bgwriter behavior of locking down the >> bufmgr while it examined the ARC/2Q data structures is causing this... > > Tom, > > It looks like the CVS HEAD is definately "better," but not by a huge > amount. The only difference is I wasn't run autovacuum in the > background (default settings), but I don't think this explains it. > Here's a graph of the differences and density of behavior: > > http://blog.amber.org/diagrams/pgsql_copy_803_cvs.png > > I can provide the raw data. Each COPY was 500 rows. Note that fsync > is turned off here. Maybe it'd be more stable with it turned on? > > Chris From pgsql-performance-owner@postgresql.org Tue Jul 19 12:44:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 723FF5286A for ; Tue, 19 Jul 2005 12:44:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 62414-03 for ; Tue, 19 Jul 2005 15:44:39 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.204]) by svr1.postgresql.org (Postfix) with ESMTP id 1BFB35285B for ; Tue, 19 Jul 2005 12:44:36 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 57so1223174wri for ; Tue, 19 Jul 2005 08:44:36 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=uPps7EP+uLyRHLl77uFz1KEljA8tvCkvzVdw35RgQpCpfXmO3rly/UduK0PemS6d24rtN7ycDz1H3cZc8yItj85cQl+5AzcmfjQ+Cb/dledHobUeaGGWWceRVzm3unWSgBM1OqJYkMApuej/CL//bLl+WlyOMCLL6ah9bgP3fms= Received: by 10.54.24.49 with SMTP id 49mr773923wrx; Tue, 19 Jul 2005 08:44:00 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 08:44:00 -0700 (PDT) Message-ID: <59d991c4050719084460b376f1@mail.gmail.com> Date: Tue, 19 Jul 2005 11:44:00 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: PFC , Vivek Khera , Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.161 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/283 X-Sequence-Number: 13524 On 7/19/05, PFC wrote: >=20 > What happens if, say at iteration 6000 (a bit after the mess star= ts), you > pause it for a few minutes and resume. Will it restart with a plateau lik= e > at the beginning of the test ? or not ? Not sure... my benchmark is designed to represent what the database will do under "typical" circumstances, and unfortunately these are typical for the application. However, I can see about adding some delays, though multiple minutes would be absurd in the application.=20 Perhaps a 5-10 second day? Would that still be interesting? > What if, during this pause, you disconnect and reconnect, or rest= art the > postmaster, or vacuum, or analyze ? Well, I don't have the numbers any more, but restarting the postmaster has no effect, other than the first few hundreds COPYs are worse than anything (3-4x slower), but then it goes back to following the trend line. The data in the chart for v8.0.3 includes running pg_autovacuum (5 minutes). Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 12:57:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 465B752998 for ; Tue, 19 Jul 2005 12:57:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 63494-03 for ; Tue, 19 Jul 2005 15:57:36 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 7E52552995 for ; Tue, 19 Jul 2005 12:57:35 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JFvRCK004542; Tue, 19 Jul 2005 11:57:27 -0400 (EDT) To: Christopher Petrilli Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c4050719084460b376f1@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Tue, 19 Jul 2005 11:44:00 -0400" Date: Tue, 19 Jul 2005 11:57:27 -0400 Message-ID: <4541.1121788647@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/284 X-Sequence-Number: 13525 Christopher Petrilli writes: > On 7/19/05, PFC wrote: >> What happens if, say at iteration 6000 (a bit after the mess starts), you >> pause it for a few minutes and resume. Will it restart with a plateau like >> at the beginning of the test ? or not ? > Not sure... my benchmark is designed to represent what the database > will do under "typical" circumstances, and unfortunately these are > typical for the application. However, I can see about adding some > delays, though multiple minutes would be absurd in the application. > Perhaps a 5-10 second day? Would that still be interesting? I think PFC's question was not directed towards modeling your application, but about helping us understand what is going wrong (so we can fix it). It seemed like a good idea to me. > Well, I don't have the numbers any more, but restarting the postmaster > has no effect, other than the first few hundreds COPYs are worse than > anything (3-4x slower), but then it goes back to following the trend > line. The data in the chart for v8.0.3 includes running pg_autovacuum > (5 minutes). The startup transient probably corresponds to the extra I/O needed to repopulate shared buffers with a useful subset of your indexes. But just to be perfectly clear: you tried this, and after the startup transient it returned to the *original* trend line? In particular, the performance goes into the tank after about 5000 total iterations, and not 5000 iterations after the postmaster restart? I'm suddenly wondering if the performance dropoff corresponds to the point where the indexes have grown large enough to not fit in shared buffers anymore. If I understand correctly, the 5000-iterations mark corresponds to 2.5 million total rows in the table; with 5 indexes you'd have 12.5 million index entries or probably a couple hundred MB total. If the insertion pattern is sufficiently random that the entire index ranges are "hot" then you might not have enough RAM. Again, experimenting with different values of shared_buffers seems like a very worthwhile thing to do. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 13:25:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 71A3F528BB for ; Tue, 19 Jul 2005 13:25:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68400-08 for ; Tue, 19 Jul 2005 16:25:51 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 7ADC15285B for ; Tue, 19 Jul 2005 13:25:49 -0300 (ADT) Received: (qmail 13202 invoked from network); 19 Jul 2005 18:26:24 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 18:26:24 +0200 To: "Tom Lane" , "Christopher Petrilli" Cc: "Vivek Khera" , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> Message-ID: Date: Tue, 19 Jul 2005 18:25:48 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <4541.1121788647@sss.pgh.pa.us> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/285 X-Sequence-Number: 13526 > I think PFC's question was not directed towards modeling your > application, but about helping us understand what is going wrong > (so we can fix it). Exactly, I was wondering if this delay would allow things to get flushed, for instance, which would give information about the problem (if giving it a few minutes of rest resumed normal operation, it would mean that some buffer somewhere is getting filled faster than it can be flushed). So, go ahead with a few minutes even if it's unrealistic, that is not the point, you have to tweak it in various possible manners to understand the causes. And instead of a pause, why not just set the duration of your test to 6000 iterations and run it two times without dropping the test table ? I'm going into wild guesses, but first you should want to know if the problem is because the table is big, or if it's something else. So you run the complete test, stopping a bit after it starts to make a mess, then instead of dumping the table and restarting the test anew, you leave it as it is, do something, then run a new test, but on this table which already has data. 'something' could be one of those : disconnect, reconnect (well you'll have to do that if you run the test twice anyway) just wait restart postgres unmount and remount the volume with the logs/data on it reboot the machine analyze vacuum vacuum analyze cluster vacuum full reindex defrag your files on disk (stopping postgres and copying the database from your disk to anotherone and back will do) or even dump'n'reload the whole database I think useful information can be extracted that way. If one of these fixes your problem it'l give hints. From pgsql-performance-owner@postgresql.org Tue Jul 19 13:27:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2262A52998 for ; Tue, 19 Jul 2005 13:27:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 70648-02 for ; Tue, 19 Jul 2005 16:27:25 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 4208452863 for ; Tue, 19 Jul 2005 13:27:23 -0300 (ADT) Received: (qmail 13247 invoked from network); 19 Jul 2005 18:27:57 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 18:27:57 +0200 To: "Tom Lane" , "Christopher Petrilli" Cc: "Vivek Khera" , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> Message-ID: From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Date: Tue, 19 Jul 2005 18:27:21 +0200 In-Reply-To: <4541.1121788647@sss.pgh.pa.us> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/286 X-Sequence-Number: 13527 > total. If the insertion pattern is sufficiently random that the entire > index ranges are "hot" then you might not have enough RAM. Try doing the test dropping some of your indexes and see if it moves the number of iterations after which it becomes slow. From pgsql-performance-owner@postgresql.org Tue Jul 19 13:31:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DDD6A5286E for ; Tue, 19 Jul 2005 13:31:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 69752-06 for ; Tue, 19 Jul 2005 16:31:12 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.197]) by svr1.postgresql.org (Postfix) with ESMTP id 1EFD252863 for ; Tue, 19 Jul 2005 13:31:11 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i30so1231264wra for ; Tue, 19 Jul 2005 09:31:11 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=CWJYJxlPFuk4Op/Sb2huDRFLcWSy3R/qSOEDpjXN8nd36we9UVcz7B3A2yUFfqAC8tkzkw88Kf9zBB/2uP+FeVMgKA9fqg+u+0e1NtlXLQ223IEI6kvk41umauiW2JZYr9p92tbvTXtwUcCkBlFPm16SuuD1GJiNACpCL8IGtYQ= Received: by 10.54.36.32 with SMTP id j32mr788866wrj; Tue, 19 Jul 2005 09:30:34 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 09:30:34 -0700 (PDT) Message-ID: <59d991c40507190930244ba9bb@mail.gmail.com> Date: Tue, 19 Jul 2005 12:30:34 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <4541.1121788647@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.158 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/287 X-Sequence-Number: 13528 On 7/19/05, Tom Lane wrote: > Christopher Petrilli writes:=20 > > Not sure... my benchmark is designed to represent what the database > > will do under "typical" circumstances, and unfortunately these are > > typical for the application. However, I can see about adding some > > delays, though multiple minutes would be absurd in the application. > > Perhaps a 5-10 second day? Would that still be interesting? >=20 > I think PFC's question was not directed towards modeling your > application, but about helping us understand what is going wrong > (so we can fix it). It seemed like a good idea to me. OK, I can modify the code to do that, and I will post it on the web. > The startup transient probably corresponds to the extra I/O needed to > repopulate shared buffers with a useful subset of your indexes. But > just to be perfectly clear: you tried this, and after the startup > transient it returned to the *original* trend line? In particular, > the performance goes into the tank after about 5000 total iterations, > and not 5000 iterations after the postmaster restart? This is correct, the TOTAL is what matters, not the specific instance count. I did an earlier run with larger batch sizes, and it hit at a similar row count, so it's definately row-count/size related. =20 > I'm suddenly wondering if the performance dropoff corresponds to the > point where the indexes have grown large enough to not fit in shared > buffers anymore. If I understand correctly, the 5000-iterations mark > corresponds to 2.5 million total rows in the table; with 5 indexes > you'd have 12.5 million index entries or probably a couple hundred MB > total. If the insertion pattern is sufficiently random that the entire > index ranges are "hot" then you might not have enough RAM. This is entirely possible, currently: shared_buffers =3D 1000 =20 work_mem =3D 65535 =20 maintenance_work_mem =3D 16384 =20 max_stack_depth =3D 2048 =20 > Again, experimenting with different values of shared_buffers seems like > a very worthwhile thing to do. I miss-understood shared_buffers then, as I thought work_mem was where indexes were kept. If this is where index manipulations happen, then I can up it quite a bit. The machine this is running on has 2GB of RAM. My concern isn't absolute performance, as this is not representative hardware, but instead is the evenness of behavior. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 13:35:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7CB9952833 for ; Tue, 19 Jul 2005 13:35:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 72426-03 for ; Tue, 19 Jul 2005 16:35:21 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.201]) by svr1.postgresql.org (Postfix) with ESMTP id 90594529AE for ; Tue, 19 Jul 2005 13:35:18 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i20so1231315wra for ; Tue, 19 Jul 2005 09:35:19 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=j1fUZ2Fbhmw5a4OFVS9N0HNEW6PVkP1paho8PC8yM+6Zs93ud87u1a5PgcaqKoyNxdhdCJO8+BhA4x2GDuvy1wETM3pKznnXCWS0Ph2sBpHNNpcsGvS7XtvXCmcEW/PuLknMjs6lzREadApRlTv/IrCF3an8rKnaKe62x1QqcKM= Received: by 10.54.68.4 with SMTP id q4mr791998wra; Tue, 19 Jul 2005 09:34:19 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 09:34:19 -0700 (PDT) Message-ID: <59d991c405071909341a10143@mail.gmail.com> Date: Tue, 19 Jul 2005 12:34:19 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: PFC Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: Tom Lane , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.155 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/288 X-Sequence-Number: 13529 On 7/19/05, PFC wrote: >=20 >=20 > > I think PFC's question was not directed towards modeling your > > application, but about helping us understand what is going wrong > > (so we can fix it). >=20 > Exactly, I was wondering if this delay would allow things to get = flushed, > for instance, which would give information about the problem (if giving i= t > a few minutes of rest resumed normal operation, it would mean that some > buffer somewhere is getting filled faster than it can be flushed). >=20 > So, go ahead with a few minutes even if it's unrealistic, that is= not the > point, you have to tweak it in various possible manners to understand the > causes. Totally understand, and appologize if I sounded dismissive. I definately appreciate the insight and input. =20 > And instead of a pause, why not just set the duration of your tes= t to > 6000 iterations and run it two times without dropping the test table ? This I can do. I'll probably set it for 5,000 for the first, and then start the second. In non-benchmark experience, however, this didn't seem to make much difference. > I'm going into wild guesses, but first you should want to know if= the > problem is because the table is big, or if it's something else. So you ru= n > the complete test, stopping a bit after it starts to make a mess, then > instead of dumping the table and restarting the test anew, you leave it a= s > it is, do something, then run a new test, but on this table which already > has data. >=20 > 'something' could be one of those : > disconnect, reconnect (well you'll have to do that if you run the= test > twice anyway) > just wait > restart postgres > unmount and remount the volume with the logs/data on it > reboot the machine > analyze > vacuum > vacuum analyze > cluster > vacuum full > reindex > defrag your files on disk (stopping postgres and copying the data= base > from your disk to anotherone and back will do) > or even dump'n'reload the whole database >=20 > I think useful information can be extracted that way. If one of t= hese > fixes your problem it'l give hints. >=20 This could take a while :-) Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 13:42:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3645952833 for ; Tue, 19 Jul 2005 13:42:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74392-01 for ; Tue, 19 Jul 2005 16:42:20 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id C59615284F for ; Tue, 19 Jul 2005 13:42:18 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JGg8qj007225; Tue, 19 Jul 2005 12:42:08 -0400 (EDT) To: Christopher Petrilli Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c40507190930244ba9bb@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Tue, 19 Jul 2005 12:30:34 -0400" Date: Tue, 19 Jul 2005 12:42:08 -0400 Message-ID: <7224.1121791328@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/289 X-Sequence-Number: 13530 Christopher Petrilli writes: > On 7/19/05, Tom Lane wrote: >> I'm suddenly wondering if the performance dropoff corresponds to the >> point where the indexes have grown large enough to not fit in shared >> buffers anymore. If I understand correctly, the 5000-iterations mark >> corresponds to 2.5 million total rows in the table; with 5 indexes >> you'd have 12.5 million index entries or probably a couple hundred MB >> total. If the insertion pattern is sufficiently random that the entire >> index ranges are "hot" then you might not have enough RAM. > This is entirely possible, currently: > shared_buffers = 1000 Ah-hah --- with that setting, you could be seeing shared-buffer thrashing even if only a fraction of the total index ranges need to be touched. I'd try some runs with shared_buffers at 10000, 50000, 100000. You might also try strace'ing the backend and see if behavior changes noticeably when the performance tanks. FWIW I have seen similar behavior while playing with MySQL's sql-bench test --- the default 1000 shared_buffers is not large enough to hold the "hot" part of the indexes in some of their insertion tests, and so performance tanks --- you can see this happening in strace because the kernel request mix goes from almost all writes to a significant part reads. On a pure data insertion benchmark you'd like to see nothing but writes. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 13:50:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0E5E352A3A for ; Tue, 19 Jul 2005 13:50:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74720-04 for ; Tue, 19 Jul 2005 16:50:26 +0000 (GMT) Received: from sherman.stortek.com (sherman.stortek.com [129.80.22.146]) by svr1.postgresql.org (Postfix) with ESMTP id D189752A1B for ; Tue, 19 Jul 2005 13:50:18 -0300 (ADT) Received: from sherman.stortek.com (localhost [127.0.0.1]) by sherman.stortek.com (8.12.10/8.12.0) with ESMTP id j6JGoHD4003067 for ; Tue, 19 Jul 2005 10:50:17 -0600 (MDT) Received: from unix.stortek.com (burma.stortek.com [129.80.16.110]) by sherman.stortek.com (8.12.10/8.12.0) with ESMTP id j6JGoGDC003064; Tue, 19 Jul 2005 10:50:16 -0600 (MDT) Received: from C118181.stortek.com (c118181.stortek.com [129.80.81.172]) by unix.stortek.com (8.12.10/8.12.0) with ESMTP id j6JGoFvn025709; Tue, 19 Jul 2005 10:50:16 -0600 (MDT) Date: Tue, 19 Jul 2005 10:50:14 -0600 From: Robert Creager To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? Message-ID: <20050719105014.00007635@C118181.stortek.com> In-Reply-To: <15472.1121709173@sss.pgh.pa.us> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> Organization: StorageTek X-Mailer: Sylpheed-Claws 1.0.4.1 Win32 (GTK+ 1.3.0; Win32) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/290 X-Sequence-Number: 13531 On Mon, 18 Jul 2005 13:52:53 -0400 Tom Lane wrote: > > Start a fresh psql session and "SHOW vacuum_cost_delay" to verify what > the active setting is. > Alright. Restarted the 803 database. Cron based vacuum analyze is running every 5 minutes. vacuum_cost_delay is 0. The problem showed up after about 1/2 hour of running. I've got vacuum jobs stacked from the last 35 minutes, with 2 vacuums running at the same time. CS is around 73k. What do I do now? I can bring the db back to normal and not run any cron based vacuum to see if it still happens, but I suspect nothing will happen without the vacuum. I'll leave it in it's current semi-catatonic state as long as possible in case there is something to look at? Cheers, Rob From pgsql-performance-owner@postgresql.org Tue Jul 19 13:54:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A8FF552A51 for ; Tue, 19 Jul 2005 13:54:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 73382-10 for ; Tue, 19 Jul 2005 16:54:23 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6FB8E52A36 for ; Tue, 19 Jul 2005 13:54:22 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JGsMkD007323; Tue, 19 Jul 2005 12:54:22 -0400 (EDT) To: Robert Creager Cc: pgsql-performance@postgresql.org Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? In-reply-to: <20050719105014.00007635@C118181.stortek.com> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> <20050719105014.00007635@C118181.stortek.com> Comments: In-reply-to Robert Creager message dated "Tue, 19 Jul 2005 10:50:14 -0600" Date: Tue, 19 Jul 2005 12:54:22 -0400 Message-ID: <7322.1121792062@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/291 X-Sequence-Number: 13532 Robert Creager writes: > Alright. Restarted the 803 database. Cron based vacuum analyze is > running every 5 minutes. vacuum_cost_delay is 0. The problem showed > up after about 1/2 hour of running. I've got vacuum jobs stacked from > the last 35 minutes, with 2 vacuums running at the same time. CS is > around 73k. Hmm, I hadn't thought about the possible impact of multiple concurrent vacuums. Is the problem caused by that, or has performance already gone into the tank by the time the cron-driven vacuums are taking long enough to overlap? regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 13:55:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AB82552992 for ; Tue, 19 Jul 2005 13:55:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74392-06 for ; Tue, 19 Jul 2005 16:55:37 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.203]) by svr1.postgresql.org (Postfix) with ESMTP id BD17652988 for ; Tue, 19 Jul 2005 13:55:36 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 50so385440wri for ; Tue, 19 Jul 2005 09:55:36 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=KWFv+gVJIe573uxo5XQb/weA2zAVI7YRaQ1bXSx6rD4E0OSMBAMRfZ1pkhJadaLIpJiPy7T0FsGVLx4l1dbC9YgkeZCFc6e5S/rBBKzQu+R/WucZu4bkIxaWBu1S+OzKFA070PwF5R9o4fNu85Dvqqooyhf/aqTJiCaI9yJo86Q= Received: by 10.54.36.59 with SMTP id j59mr799612wrj; Tue, 19 Jul 2005 09:54:35 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 09:54:35 -0700 (PDT) Message-ID: <59d991c4050719095440cd3de5@mail.gmail.com> Date: Tue, 19 Jul 2005 12:54:35 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <7224.1121791328@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.153 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/292 X-Sequence-Number: 13533 As I'm doing this, I'm noticing something *VERY* disturbing to me: postmaster backend: 20.3% CPU psql frontend: 61.2% CPU WTF? The only thing going through the front end is the COPY command, and it's sent to the backend to read from a file? Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 14:05:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9E9DD52A2D for ; Tue, 19 Jul 2005 14:05:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91881-05 for ; Tue, 19 Jul 2005 17:05:31 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 289BA52863 for ; Tue, 19 Jul 2005 14:05:29 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JH5PS6007437; Tue, 19 Jul 2005 13:05:25 -0400 (EDT) To: Christopher Petrilli Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c4050719095440cd3de5@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Tue, 19 Jul 2005 12:54:35 -0400" Date: Tue, 19 Jul 2005 13:05:25 -0400 Message-ID: <7436.1121792725@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/293 X-Sequence-Number: 13534 Christopher Petrilli writes: > As I'm doing this, I'm noticing something *VERY* disturbing to me: > postmaster backend: 20.3% CPU > psql frontend: 61.2% CPU > WTF? The only thing going through the front end is the COPY command, > and it's sent to the backend to read from a file? Are you sure the backend is reading directly from the file, and not through psql? (\copy, or COPY FROM STDIN, would go through psql.) But even so that seems awfully high, considering how little work psql has to do compared to the backend. Has anyone ever profiled psql doing this sort of thing? I know I've spent all my time looking at the backend ... regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 14:14:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 189F0528C3 for ; Tue, 19 Jul 2005 14:14:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 89409-09 for ; Tue, 19 Jul 2005 17:14:03 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.202]) by svr1.postgresql.org (Postfix) with ESMTP id E60D352863 for ; Tue, 19 Jul 2005 14:13:59 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i28so1240542wra for ; Tue, 19 Jul 2005 10:14:00 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=RK6fYl2KXT+VFBt9MqGccMbt3z5NFXUzchXFqXnIx2SwhFbAtAIvIQxg5Fxqag2Xtn2wzWQGNcY99rigGZWdXZFLcUjCYvm+wwgwkOPfl/zrflCe9Hi6cRm1gMoH/Z1wX2bxtSQBEpPvLavzoXOfoMFsy/HlLk6wWZNMJQpN2yg= Received: by 10.54.37.71 with SMTP id k71mr804351wrk; Tue, 19 Jul 2005 10:13:05 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 10:13:05 -0700 (PDT) Message-ID: <59d991c405071910133e28031a@mail.gmail.com> Date: Tue, 19 Jul 2005 13:13:05 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <7436.1121792725@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.15 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/294 X-Sequence-Number: 13535 On 7/19/05, Tom Lane wrote: > Christopher Petrilli writes: > > As I'm doing this, I'm noticing something *VERY* disturbing to me: > > postmaster backend: 20.3% CPU > > psql frontend: 61.2% CPU >=20 > > WTF? The only thing going through the front end is the COPY command, > > and it's sent to the backend to read from a file? >=20 > Are you sure the backend is reading directly from the file, and not > through psql? (\copy, or COPY FROM STDIN, would go through psql.) The exact command is: COPY test (columnlist...) FROM '/tmp/loadfile'; =20 > But even so that seems awfully high, considering how little work psql > has to do compared to the backend. Has anyone ever profiled psql doing > this sort of thing? I know I've spent all my time looking at the > backend ... Linux 2.6, ext3, data=3Dwriteback It's flipped now (stil lrunning), and it's 48% postmaster, 36% psql, but anything more than 1-2% seems absurd. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 14:43:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 59A7A52990 for ; Tue, 19 Jul 2005 14:43:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 23853-03 for ; Tue, 19 Jul 2005 17:43:27 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.204]) by svr1.postgresql.org (Postfix) with ESMTP id C65FF5284F for ; Tue, 19 Jul 2005 14:43:23 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i14so1270743wra for ; Tue, 19 Jul 2005 10:43:24 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type; b=cwrfNIfIFDBf8f+uGSSB+iubc258lk7Tr2ktj+IKTy9PdkvWSegzeOYp4+UKLFqCQ1HWQOOLMQ8AnGefND3l972LEgJGb3yb/kI6iJCLGA+Cye+5JynODMH/B4l6j2IFEM7UauhdHtPP3gZcyWCfnUqS1uiAf17Mkh+4fKQOucE= Received: by 10.54.44.53 with SMTP id r53mr823205wrr; Tue, 19 Jul 2005 10:42:42 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 10:42:41 -0700 (PDT) Message-ID: <1efd553a05071910427b290203@mail.gmail.com> Date: Tue, 19 Jul 2005 13:42:42 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: pgsql-performance@postgresql.org Subject: Looking for tips Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_3181_11694449.1121794962010" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.32 tagged_above=0 required=5 tests=HTML_10_20, HTML_MESSAGE, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/295 X-Sequence-Number: 13536 ------=_Part_3181_11694449.1121794962010 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Hi, I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of ram. Running scripts locally, it takes about 1.5x longer than mysql, and the loa= d=20 on the server is only about 21%. I upped the sort_mem to 8192 (kB), and shared_buffers and=20 effective_cache_size to 65536 (512MB), but neither the timing nor the serve= r=20 load have changed at all. FYI, I'm going to be working on data sets in the= =20 order of GB. I think I've gone about as far as I can with google.. can anybody give me= =20 some advice on how to improve the raw performance before I start looking at= =20 code changes? Thanks in advance. ------=_Part_3181_11694449.1121794962010 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline
Hi,
I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of= ram.
Running scripts locally, it takes about 1.5x longer than mysql, and th= e load on the server is only about 21%.
I upped the sort_mem to 8192 (kB), and shared_buffers and effective_ca= che_size to 65536 (512MB), but neither the timing nor the server load have = changed at all. FYI, I'm going to be working on data sets in the order of G= B.
 
I think I've gone about as far as I can with google.. can anybody give= me some advice on how to improve the raw performance before I start lookin= g at code changes?
 
Thanks in advance.
------=_Part_3181_11694449.1121794962010-- From pgsql-performance-owner@postgresql.org Tue Jul 19 14:50:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9374552932 for ; Tue, 19 Jul 2005 14:50:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31735-03 for ; Tue, 19 Jul 2005 17:50:16 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 122155284F for ; Tue, 19 Jul 2005 14:50:15 -0300 (ADT) Received: from [192.168.1.52] (fc1smp [66.93.38.87]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6JHoDhr028623; Tue, 19 Jul 2005 10:50:13 -0700 Message-ID: <42DD3D55.9020508@commandprompt.com> Date: Tue, 19 Jul 2005 10:50:13 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Oliver Crosby Cc: pgsql-performance@postgresql.org Subject: Re: Looking for tips References: <1efd553a05071910427b290203@mail.gmail.com> In-Reply-To: <1efd553a05071910427b290203@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.014 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/296 X-Sequence-Number: 13537 Oliver Crosby wrote: > Hi, > I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of ram. > Running scripts locally, it takes about 1.5x longer than mysql, and the > load on the server is only about 21%. What queries? What is your structure? Have you tried explain analyze? How many rows in the table? Which OS? How are you testing the speed? What type of RAID? -- Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 PostgreSQL Replication, Consulting, Custom Programming, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ From pgsql-performance-owner@postgresql.org Tue Jul 19 14:58:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 497E752808 for ; Tue, 19 Jul 2005 14:58:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 36412-04 for ; Tue, 19 Jul 2005 17:58:46 +0000 (GMT) Received: from ionian.i-clic.uihc.uiowa.edu (ionian.i-clic.uihc.uiowa.edu [129.255.211.20]) by svr1.postgresql.org (Postfix) with ESMTP id 639D352A1F for ; Tue, 19 Jul 2005 14:58:43 -0300 (ADT) Received: by ionian.i-clic.uihc.uiowa.edu (Postfix, from userid 76) id 1850661DB0; Tue, 19 Jul 2005 11:51:04 -0500 (CDT) Received: from [192.168.1.11] (12-214-18-81.client.mchsi.com [12.214.18.81]) by ionian.i-clic.uihc.uiowa.edu (Postfix) with ESMTP id C3DC061DAF; Tue, 19 Jul 2005 11:51:02 -0500 (CDT) Message-ID: <42DD3F3C.7000502@arbash-meinel.com> Date: Tue, 19 Jul 2005 12:58:20 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Oliver Crosby Cc: pgsql-performance@postgresql.org Subject: Re: Looking for tips References: <1efd553a05071910427b290203@mail.gmail.com> In-Reply-To: <1efd553a05071910427b290203@mail.gmail.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigCB437C5FBC85BBD7FF02B6D1" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.021 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/297 X-Sequence-Number: 13538 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigCB437C5FBC85BBD7FF02B6D1 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Oliver Crosby wrote: > Hi, > I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of ram. > Running scripts locally, it takes about 1.5x longer than mysql, and the > load on the server is only about 21%. > I upped the sort_mem to 8192 (kB), and shared_buffers and > effective_cache_size to 65536 (512MB), but neither the timing nor the > server load have changed at all. FYI, I'm going to be working on data > sets in the order of GB. > > I think I've gone about as far as I can with google.. can anybody give > me some advice on how to improve the raw performance before I start > looking at code changes? > > Thanks in advance. First, try to post in plain-text rather than html, it is easier to read. :) Second, if you can determine what queries are running slow, post the result of EXPLAIN ANALYZE on them, and we can try to help you tune them/postgres to better effect. Just a blanket question like this is hard to answer. Your new shared_buffers are probably *way* too high. They should be at most around 10% of ram. Since this is a dedicated server effective_cache_size should be probably ~75% of ram, or close to 1.2GB. There are quite a few things that you can tweak, so the more information you can give, the more we can help. For instance, if you are loading a lot of data into a table, if possible, you want to use COPY not INSERT. If you have a lot of indexes and are loading a significant portion, it is sometimes faster to drop the indexes, COPY the data in, and then rebuild the indexes. For tables with a lot of inserts/updates, you need to watch out for foreign key constraints. (Generally, you will want an index on both sides of the foreign key. One is required, the other is recommended for faster update/deletes). John =:-> --------------enigCB437C5FBC85BBD7FF02B6D1 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC3T9AJdeBCYSNAAMRAp+FAKDRs5s7EatjgMvPMJRl5VVzbb24XwCdGCVV HqKtYtqNvXBExk7Jv7MYFF8= =/XjI -----END PGP SIGNATURE----- --------------enigCB437C5FBC85BBD7FF02B6D1-- From pgsql-performance-owner@postgresql.org Tue Jul 19 15:10:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BD93F52932 for ; Tue, 19 Jul 2005 15:10:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42519-03 for ; Tue, 19 Jul 2005 18:09:53 +0000 (GMT) Received: from sherman.stortek.com (sherman.stortek.com [129.80.22.146]) by svr1.postgresql.org (Postfix) with ESMTP id 7E5645284F for ; Tue, 19 Jul 2005 15:09:52 -0300 (ADT) Received: from sherman.stortek.com (localhost [127.0.0.1]) by sherman.stortek.com (8.12.10/8.12.0) with ESMTP id j6JI9rD4000814 for ; Tue, 19 Jul 2005 12:09:53 -0600 (MDT) Received: from unix.stortek.com (burma.stortek.com [129.80.16.110]) by sherman.stortek.com (8.12.10/8.12.0) with ESMTP id j6JI9qDC000810; Tue, 19 Jul 2005 12:09:53 -0600 (MDT) Received: from C118181.stortek.com (c118181.stortek.com [129.80.81.172]) by unix.stortek.com (8.12.10/8.12.0) with ESMTP id j6JI9pvn028181; Tue, 19 Jul 2005 12:09:52 -0600 (MDT) Date: Tue, 19 Jul 2005 12:09:51 -0600 From: Robert Creager To: Tom Lane Cc: Robert Creager , pgsql-performance@postgresql.org, Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? Message-ID: <20050719120951.00002520@C118181.stortek.com> In-Reply-To: <7322.1121792062@sss.pgh.pa.us> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> <20050719105014.00007635@C118181.stortek.com> <7322.1121792062@sss.pgh.pa.us> Organization: StorageTek X-Mailer: Sylpheed-Claws 1.0.4.1 Win32 (GTK+ 1.3.0; Win32) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/298 X-Sequence-Number: 13539 On Tue, 19 Jul 2005 12:54:22 -0400 Tom Lane wrote: > Hmm, I hadn't thought about the possible impact of multiple concurrent > vacuums. Is the problem caused by that, or has performance already gone > into the tank by the time the cron-driven vacuums are taking long enough > to overlap? Don't know just yet. When I run the vacuums manually on a healthy system on 741, they take less than 30 seconds. I've stopped the cron vacuum and canceled all the outstanding vacuum processes, but the 803 is still struggling (1/2 hour later). I'll re-start the database, vacuum full analyze and restart the runs without the cron vacuum running. Cheers, Rob From pgsql-performance-owner@postgresql.org Tue Jul 19 15:10:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7EF5B52988 for ; Tue, 19 Jul 2005 15:10:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42519-04 for ; Tue, 19 Jul 2005 18:10:08 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id ED5565298F for ; Tue, 19 Jul 2005 15:10:04 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JI9ueC007844; Tue, 19 Jul 2005 14:09:56 -0400 (EDT) To: Christopher Petrilli Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c405071910133e28031a@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> <59d991c405071910133e28031a@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Tue, 19 Jul 2005 13:13:05 -0400" Date: Tue, 19 Jul 2005 14:09:56 -0400 Message-ID: <7843.1121796596@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/299 X-Sequence-Number: 13540 Christopher Petrilli writes: >> Are you sure the backend is reading directly from the file, and not >> through psql? (\copy, or COPY FROM STDIN, would go through psql.) > The exact command is: > COPY test (columnlist...) FROM '/tmp/loadfile'; I tried to replicate this by putting a ton of COPY commands like that into a file and doing "psql -f file ...". I don't see more than about 0.3% CPU going to psql. So there's something funny about your test conditions. How *exactly* are you invoking psql? regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 15:15:38 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3025E5293E for ; Tue, 19 Jul 2005 15:15:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 48005-07 for ; Tue, 19 Jul 2005 18:15:31 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id 601DD52AAB for ; Tue, 19 Jul 2005 15:15:30 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id A860940C972; Tue, 19 Jul 2005 19:15:16 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id F019B15EE6; Tue, 19 Jul 2005 19:10:24 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 13801-04; Tue, 19 Jul 2005 19:10:20 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 68C6715EE1; Tue, 19 Jul 2005 19:10:20 +0100 (BST) Message-ID: <42DD420C.2010306@archonet.com> Date: Tue, 19 Jul 2005 19:10:20 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Oliver Crosby Cc: pgsql-performance@postgresql.org Subject: Re: Looking for tips References: <1efd553a05071910427b290203@mail.gmail.com> In-Reply-To: <1efd553a05071910427b290203@mail.gmail.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/300 X-Sequence-Number: 13541 Oliver Crosby wrote: > Hi, > I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of ram. > Running scripts locally, it takes about 1.5x longer than mysql, and the load > on the server is only about 21%. What scripts? What do they do? Oh, and 7.4.8 is the latest release - worth upgrading for the fixes. > I upped the sort_mem to 8192 (kB), and shared_buffers and > effective_cache_size to 65536 (512MB), but neither the timing nor the server > load have changed at all. Well, effective_cache_size is the amount of RAM being used by the OS to cache your files, so take a look at top/free and set it based on that (pick a steady load). What sort_mem should be will obviously depend how much sorting you do. Drop shared_buffers down to about 10000 - 20000 (at a guess) You may find the following useful http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php Read the Performance Tuning article, there is an updated one for version 8 at: http://www.powerpostgresql.com/PerfList > FYI, I'm going to be working on data sets in the > order of GB. Fair enough. > I think I've gone about as far as I can with google.. can anybody give me > some advice on how to improve the raw performance before I start looking at > code changes? Identify what the problem is first of all. Some things to consider: - Are there particular queries giving you trouble? - Is your load mostly reads or mostly writes? - Do you have one user or 100? - Are you block-loading data efficiently where necessary? - Have you indexed both sides of your foreign-keys where sensible? - Are your disks being used effectively? - Are your statistics accurate/up to date? Bear in mind that MySQL will probably be quicker for simple queries for one user and always will be. If you have multiple users running a mix of multi-table joins and updates then PG will have a chance to stretch its legs. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Tue Jul 19 15:22:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 16CE45284F for ; Tue, 19 Jul 2005 15:22:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 48325-07 for ; Tue, 19 Jul 2005 18:22:03 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.207]) by svr1.postgresql.org (Postfix) with ESMTP id 502F952988 for ; Tue, 19 Jul 2005 15:22:02 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i16so1276478wra for ; Tue, 19 Jul 2005 11:22:03 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=odu9+kMy++RLL6JzTEn9RNHJvOqav/qQSorz3esHO1HOnm81YSZ6elDsHf+qgHNjbiRBwX+aIgCOnVYMnSlboz2gebK46HGWPyUJQwlU/lly4o/4syzAK4egb2oKAjS83gujv7/Xi/IGTCABU1/Nkuz9D6BKMU2603tztqRDP/Y= Received: by 10.54.109.20 with SMTP id h20mr837828wrc; Tue, 19 Jul 2005 11:21:22 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 11:21:22 -0700 (PDT) Message-ID: <1efd553a050719112126b8d981@mail.gmail.com> Date: Tue, 19 Jul 2005 14:21:22 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: "Joshua D. Drake" Subject: Re: Looking for tips Cc: pgsql-performance@postgresql.org In-Reply-To: <42DD3D55.9020508@commandprompt.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.172 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/301 X-Sequence-Number: 13542 I was hoping to start with tuning postgres to match the hardware, but in any case.. The queries are all simple insert or select statements on single tables. Eg. select x from table where y=3D?; or insert into table (a, b, c) values (?, ?, ?); In the case of selects where it's a large table, there's an index on the column being searched, so in terms of the example above, x is either a pkey column or other related field, and y is a non-pkey column. I'm not sure what you mean by structure. I tried explain analyse on the individual queries, but I'm not sure what can be done to manipulate them when they don't do much. My test environment has about 100k - 300k rows in each table, and for production I'm expecting this to be in the order of 1M+. The OS is Redhat Enterprise 3. I'm using a time command when I call the scripts to get a total running time from start to finish. I don't know what we have for RAID, but I suspect it's just a single 10k or 15k rpm hdd. ---------------------------------------------------------------------------= --------------------------------------------- I'll try your recommendations for shared_buffers and effective_cache_size. Thanks John! We're trying to improve performance on a log processing script to the point where it can be run as close as possible to realtime. A lot of what gets inserted depends on what's already in the db, and it runs item-by-item... so unfortunately I can't take advantage of copy. We tried dropping indices, copying data in, then rebuilding. It works great for a bulk import, but the processing script went a lot slower without them. (Each insert is preceeded by a local cache check and then a db search to see if an ID already exists for an item.) We have no foreign keys at the moment. Would they help? On 7/19/05, Joshua D. Drake wrote: > Oliver Crosby wrote: > > Hi, > > I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of = ram. > > Running scripts locally, it takes about 1.5x longer than mysql, and the > > load on the server is only about 21%. >=20 > What queries? > What is your structure? > Have you tried explain analyze? > How many rows in the table? > Which OS? > How are you testing the speed? > What type of RAID? >=20 >=20 >=20 > -- > Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 > PostgreSQL Replication, Consulting, Custom Programming, 24x7 support > Managed Services, Shared and Dedicated Hosting > Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ > From pgsql-performance-owner@postgresql.org Tue Jul 19 15:23:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 55EE852A38 for ; Tue, 19 Jul 2005 15:23:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 52942-08 for ; Tue, 19 Jul 2005 18:23:27 +0000 (GMT) Received: from cdrsmtp01.yellowbook.com (unknown [64.199.226.83]) by svr1.postgresql.org (Postfix) with ESMTP id 8A62352A90 for ; Tue, 19 Jul 2005 15:23:24 -0300 (ADT) Received: from ybcdrmta01.corp.ybusa.net ([10.5.17.170]) by cdrsmtp01.yellowbook.com with Microsoft SMTPSVC(6.0.3790.0); Tue, 19 Jul 2005 13:26:40 -0500 Received: from YBCDREX01.corp.ybusa.net ([10.5.17.162]) by ybcdrmta01.corp.ybusa.net with Microsoft SMTPSVC(6.0.3790.211); Tue, 19 Jul 2005 13:23:25 -0500 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C58C8E.F3B5F894" Subject: context-switching issue on Xeon Date: Tue, 19 Jul 2005 13:23:25 -0500 Message-ID: <3FB3AAE149F4AD4D8499E15EE8CF41A3460442@YBCDREX01.corp.ybusa.net> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: context-switching issue on Xeon Thread-Index: AcWMjvO1oJAWW8ytTy24jjmTwzCcUg== From: "Sailer, Denis (YBUSA-CDR)" To: X-OriginalArrivalTime: 19 Jul 2005 18:23:25.0323 (UTC) FILETIME=[F3C955B0:01C58C8E] X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.376 tagged_above=0 required=5 tests=AWL, HTML_MESSAGE, HTML_TEXT_AFTER_BODY, HTML_TEXT_AFTER_HTML X-Spam-Level: X-Archive-Number: 200507/302 X-Sequence-Number: 13543 This is a multi-part message in MIME format. ------_=_NextPart_001_01C58C8E.F3B5F894 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable The thread below has the test case that we were able to use to reproduce the issue. =20 http://archives.postgresql.org/pgsql-performance/2004-04/msg00280.php =20 The last messages on this subject are from April of 2005. Has there been any successful ways to significantly reduce the impact this has to multi-processing? I haven't been able to find anything showing a resolution of some kind. =20 We are seeing this on two of our machines: =20 Quad 3.0 GHz XEON with 3GB of memory running PG 7.4.3 with SuSE kernel 2.4 =20 Dual 2.8 GHz XEON with 2GB of memory running PG 8.0.0 with SuSE kernel 2.4 =20 =20 =20 ------_=_NextPart_001_01C58C8E.F3B5F894 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

The thread below has the test case that we were able = to use to reproduce the issue.

 

http://archives.postgresql.org/pgsql-performance/2004-04/msg00280.p= hp

 

The last messages on this subject are from April of 2005.  Has there been any successful ways to significantly reduce = the impact this has to multi-processing?  I haven’t been able to = find anything showing a resolution of some kind.

 

We are seeing this on two of our = machines:

 

Quad 3.0 GHz XEON with 3GB of memory running PG 7.4.3 = with SuSE kernel 2.4

 

Dual 2.8 GHz XEON with 2GB of memory running PG 8.0.0 = with SuSE kernel 2.4

 

 

 

=00 ------_=_NextPart_001_01C58C8E.F3B5F894-- From pgsql-performance-owner@postgresql.org Tue Jul 19 15:38:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CE366529F8 for ; Tue, 19 Jul 2005 15:38:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 57049-09 for ; Tue, 19 Jul 2005 18:38:35 +0000 (GMT) Received: from dns.deep-purple.com (dns.deep-purple.com [209.61.158.121]) by svr1.postgresql.org (Postfix) with ESMTP id C4BC6529E1 for ; Tue, 19 Jul 2005 15:38:31 -0300 (ADT) Received: from [192.168.5.4] (82-133-109-237.dyn.gotadsl.co.uk [82.133.109.237]) (authenticated) by dns.deep-purple.com (8.11.6/8.11.6) with ESMTP id j6JIcMc05842; Tue, 19 Jul 2005 13:38:22 -0500 In-Reply-To: <3FB3AAE149F4AD4D8499E15EE8CF41A3460442@YBCDREX01.corp.ybusa.net> References: <3FB3AAE149F4AD4D8499E15EE8CF41A3460442@YBCDREX01.corp.ybusa.net> Mime-Version: 1.0 (Apple Message framework v733) Content-Type: text/plain; charset=WINDOWS-1252; delsp=yes; format=flowed Message-Id: Cc: Content-Transfer-Encoding: quoted-printable From: David Hodgkinson Subject: Re: context-switching issue on Xeon Date: Tue, 19 Jul 2005 19:38:14 +0100 To: "Sailer, Denis (YBUSA-CDR)" X-Mailer: Apple Mail (2.733) X-RIF3-MailScanner-Information: Please contact the ISP for more information X-RIF3-MailScanner: Found to be clean X-RIF3-MailScanner-SpamCheck: not spam, SpamAssassin (score=-4.842, required 5, autolearn=not spam, AWL 0.06, BAYES_00 -4.90) X-MailScanner-From: daveh@hodgkinson.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/303 X-Sequence-Number: 13544 FWIW, I'm seeing this with a client at the moment. 40-60k CS per second on Dual 3.2GHz. There are plenty of other issues we're dealing with, but this is =20 obviously disconcerting... On 19 Jul 2005, at 19:23, Sailer, Denis (YBUSA-CDR) wrote: > The thread below has the test case that we were able to use to =20 > reproduce the issue. > > > http://archives.postgresql.org/pgsql-performance/2004-04/msg00280.php > > > The last messages on this subject are from April of 2005. Has =20 > there been any successful ways to significantly reduce the impact =20 > this has to multi-processing? I haven=92t been able to find anything =20= > showing a resolution of some kind. > > > We are seeing this on two of our machines: > > > Quad 3.0 GHz XEON with 3GB of memory running PG 7.4.3 with SuSE =20 > kernel 2.4 > > > Dual 2.8 GHz XEON with 2GB of memory running PG 8.0.0 with SuSE =20 > kernel 2.4 > > > > > From pgsql-performance-owner@postgresql.org Tue Jul 19 15:40:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 18C40529F8 for ; Tue, 19 Jul 2005 15:40:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71895-02 for ; Tue, 19 Jul 2005 18:40:47 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id DBBA1529EF for ; Tue, 19 Jul 2005 15:40:45 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JIekuK008217; Tue, 19 Jul 2005 14:40:46 -0400 (EDT) To: "Sailer, Denis (YBUSA-CDR)" Cc: pgsql-performance@postgresql.org Subject: Re: context-switching issue on Xeon In-reply-to: <3FB3AAE149F4AD4D8499E15EE8CF41A3460442@YBCDREX01.corp.ybusa.net> References: <3FB3AAE149F4AD4D8499E15EE8CF41A3460442@YBCDREX01.corp.ybusa.net> Comments: In-reply-to "Sailer, Denis (YBUSA-CDR)" message dated "Tue, 19 Jul 2005 13:23:25 -0500" Date: Tue, 19 Jul 2005 14:40:46 -0400 Message-ID: <8216.1121798446@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/304 X-Sequence-Number: 13545 "Sailer, Denis (YBUSA-CDR)" writes: > http://archives.postgresql.org/pgsql-performance/2004-04/msg00280.php > The last messages on this subject are from April of 2005. Has there > been any successful ways to significantly reduce the impact this has to > multi-processing? CVS tip should largely fix the problem as far as buffer manager contention goes. > I haven't been able to find anything showing a > resolution of some kind. Look at the Feb/March threads concerning buffer manager rewrite, clock sweep, etc ... eg http://archives.postgresql.org/pgsql-patches/2005-03/msg00015.php regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 15:42:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C2A1B529EF for ; Tue, 19 Jul 2005 15:41:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 60510-07 for ; Tue, 19 Jul 2005 18:40:58 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 1E288529C3 for ; Tue, 19 Jul 2005 15:40:56 -0300 (ADT) Received: (qmail 19500 invoked from network); 19 Jul 2005 20:41:32 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 20:41:32 +0200 To: "Oliver Crosby" , "Joshua D. Drake" Cc: pgsql-performance@postgresql.org Subject: Re: Looking for tips References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> Message-ID: From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Date: Tue, 19 Jul 2005 20:40:56 +0200 In-Reply-To: <1efd553a050719112126b8d981@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/305 X-Sequence-Number: 13546 What programming language are these scripts written in ? From pgsql-performance-owner@postgresql.org Tue Jul 19 15:42:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 86415529F4 for ; Tue, 19 Jul 2005 15:42:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76183-02 for ; Tue, 19 Jul 2005 18:42:49 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.198]) by svr1.postgresql.org (Postfix) with ESMTP id 183F6529C3 for ; Tue, 19 Jul 2005 15:42:48 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i16so1281222wra for ; Tue, 19 Jul 2005 11:42:49 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=r5DY2JxmIAD0Btdi9i6RSbxAUB6YePn70bDY2i8ZrcTkgLc7EZJtdmvuitTFmpiTmu+0KbE00lx4RRUZ8khIqFwl+uzaxsCKtLPaZ8L6QQYFt9oAx7Ph0a5qekGTe+2gSOaZYFj/2A3q5axspb6TihI/GytSgpuZWimYY/r0hic= Received: by 10.54.57.79 with SMTP id f79mr846368wra; Tue, 19 Jul 2005 11:41:51 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 11:41:51 -0700 (PDT) Message-ID: <1efd553a05071911413e50ea90@mail.gmail.com> Date: Tue, 19 Jul 2005 14:41:51 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: Richard Huxton Subject: Re: Looking for tips Cc: pgsql-performance@postgresql.org In-Reply-To: <42DD420C.2010306@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a05071910427b290203@mail.gmail.com> <42DD420C.2010306@archonet.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.123 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/306 X-Sequence-Number: 13547 > Identify what the problem is first of all. Some things to consider: > - Are there particular queries giving you trouble? > - Is your load mostly reads or mostly writes? > - Do you have one user or 100? > - Are you block-loading data efficiently where necessary? > - Have you indexed both sides of your foreign-keys where sensible? > - Are your disks being used effectively? > - Are your statistics accurate/up to date? No queries in particular appear to be a problem. I think it's just the overall speed. If any of the configuration settings will help make the simple select queries go faster, that would be ideal. The load is about 50/50 read/write. At the moment it's just one user, but the goal is to have a cluster of servers (probably less than a dozen) updating to a central db. Indices exist for the fields being searched, but we don't have any foreign = keys. I'm not too familiar with effective disk usage or statistics... From pgsql-performance-owner@postgresql.org Tue Jul 19 15:45:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5E7A352A0B for ; Tue, 19 Jul 2005 15:45:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 73990-05 for ; Tue, 19 Jul 2005 18:45:18 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.195]) by svr1.postgresql.org (Postfix) with ESMTP id F0B4052A1F for ; Tue, 19 Jul 2005 15:45:17 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i7so1282687wra for ; Tue, 19 Jul 2005 11:45:18 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=q72Sx/ixBmW4jiCx115bdQJgfXJ2KfJe0SttRfXKuasSgXRM5Ey99wwFYApNcexJnhMpLZbV18ZI/nEGw455fnhmbdbvmGhH6hEfpni5S82YQXpr8x0uiDdfppxcnDFd3JUbShdeGMJRVEEBGZwk5t7L8FXd7X8tAE0+fywiT1s= Received: by 10.54.137.2 with SMTP id k2mr847410wrd; Tue, 19 Jul 2005 11:44:26 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 11:44:26 -0700 (PDT) Message-ID: <1efd553a050719114439b43ec3@mail.gmail.com> Date: Tue, 19 Jul 2005 14:44:26 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: PFC Subject: Re: Looking for tips Cc: "Joshua D. Drake" , pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.109 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/307 X-Sequence-Number: 13548 > What programming language are these scripts written in ? perl. using the DBD:Pg interface instead of command-lining it through psql From pgsql-performance-owner@postgresql.org Tue Jul 19 15:50:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 704D2529E1 for ; Tue, 19 Jul 2005 15:49:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 72073-08 for ; Tue, 19 Jul 2005 18:49:47 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.207]) by svr1.postgresql.org (Postfix) with ESMTP id 35F79529FC for ; Tue, 19 Jul 2005 15:49:46 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 69so405340wra for ; Tue, 19 Jul 2005 11:49:47 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=TVsVqRh4sJhShFyw5ePGO/JPuas/pWnWKAB9pCXSW3Bub5Gc6eHQ5lD/AhxvX2XbQo+YgkfHlOpmEAvAbKuHbCWfjCGz7u7Titwn36it/atbFZnBKagUNaXhDj9KvsJyF0HM+iufNe1nxJTiGUFjJxGEWAo0B8t+rPnUJVrBS+E= Received: by 10.54.37.47 with SMTP id k47mr840228wrk; Tue, 19 Jul 2005 11:48:54 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 11:48:54 -0700 (PDT) Message-ID: <59d991c40507191148e6f9ec5@mail.gmail.com> Date: Tue, 19 Jul 2005 14:48:54 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <7843.1121796596@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> <59d991c405071910133e28031a@mail.gmail.com> <7843.1121796596@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.148 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/308 X-Sequence-Number: 13549 On 7/19/05, Tom Lane wrote: > Christopher Petrilli writes: > >> Are you sure the backend is reading directly from the file, and not > >> through psql? (\copy, or COPY FROM STDIN, would go through psql.) >=20 > > The exact command is: > > COPY test (columnlist...) FROM '/tmp/loadfile'; >=20 > I tried to replicate this by putting a ton of COPY commands like that > into a file and doing "psql -f file ...". I don't see more than about > 0.3% CPU going to psql. So there's something funny about your test > conditions. How *exactly* are you invoking psql? It is a subprocess of a Python process, driven using a pexpect interchange. I send the COPY command, then wait for the '=3D#' to come back. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 15:53:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 90EBD52831 for ; Tue, 19 Jul 2005 15:53:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 82527-02 for ; Tue, 19 Jul 2005 18:53:15 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 5ADAD5281B for ; Tue, 19 Jul 2005 15:53:15 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JIrD4g008376; Tue, 19 Jul 2005 14:53:13 -0400 (EDT) To: Christopher Petrilli Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions In-reply-to: <59d991c40507191148e6f9ec5@mail.gmail.com> References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> <59d991c405071910133e28031a@mail.gmail.com> <7843.1121796596@sss.pgh.pa.us> <59d991c40507191148e6f9ec5@mail.gmail.com> Comments: In-reply-to Christopher Petrilli message dated "Tue, 19 Jul 2005 14:48:54 -0400" Date: Tue, 19 Jul 2005 14:53:13 -0400 Message-ID: <8375.1121799193@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/309 X-Sequence-Number: 13550 Christopher Petrilli writes: > On 7/19/05, Tom Lane wrote: >> How *exactly* are you invoking psql? > It is a subprocess of a Python process, driven using a pexpect > interchange. I send the COPY command, then wait for the '=#' to come > back. Some weird interaction with pexpect maybe? Try adding "-n" (disable readline) to the psql command switches. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 15:58:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5A641529EF for ; Tue, 19 Jul 2005 15:58:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79018-06 for ; Tue, 19 Jul 2005 18:58:27 +0000 (GMT) Received: from gwmta.wicourts.gov (gwmta.wicourts.gov [165.219.244.91]) by svr1.postgresql.org (Postfix) with ESMTP id E48C952869 for ; Tue, 19 Jul 2005 15:58:25 -0300 (ADT) Received: from Courts-MTA by gwmta.wicourts.gov with Novell_GroupWise; Tue, 19 Jul 2005 13:58:24 -0500 Message-Id: X-Mailer: Novell GroupWise Internet Agent 6.5.2 Date: Tue, 19 Jul 2005 13:58:20 -0500 From: "Kevin Grittner" To: , Cc: Subject: Re: Looking for tips Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.016 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/310 X-Sequence-Number: 13551 Hi Oliver, We had low resource utilization and poor throughput on inserts of thousands of rows within a single database transaction. There were a lot of configuration parameters we changed, but the one which helped the most was wal_buffers -- we wound up setting it to 1000. This may be higher than it needs to be, but when we got to something which ran well, we stopped tinkering. The default value clearly caused a bottleneck. You might find this page useful: http://www.varlena.com/varlena/GeneralBits/Tidbits/annotated_conf_e.html -Kevin >>> Oliver Crosby 07/19/05 1:21 PM >>> I was hoping to start with tuning postgres to match the hardware, but in any case.. The queries are all simple insert or select statements on single tables. Eg. select x from table where y=?; or insert into table (a, b, c) values (?, ?, ?); In the case of selects where it's a large table, there's an index on the column being searched, so in terms of the example above, x is either a pkey column or other related field, and y is a non-pkey column. I'm not sure what you mean by structure. I tried explain analyse on the individual queries, but I'm not sure what can be done to manipulate them when they don't do much. My test environment has about 100k - 300k rows in each table, and for production I'm expecting this to be in the order of 1M+. The OS is Redhat Enterprise 3. I'm using a time command when I call the scripts to get a total running time from start to finish. I don't know what we have for RAID, but I suspect it's just a single 10k or 15k rpm hdd. ------------------------------------------------------------------------------------------------------------------------ I'll try your recommendations for shared_buffers and effective_cache_size. Thanks John! We're trying to improve performance on a log processing script to the point where it can be run as close as possible to realtime. A lot of what gets inserted depends on what's already in the db, and it runs item-by-item... so unfortunately I can't take advantage of copy. We tried dropping indices, copying data in, then rebuilding. It works great for a bulk import, but the processing script went a lot slower without them. (Each insert is preceeded by a local cache check and then a db search to see if an ID already exists for an item.) We have no foreign keys at the moment. Would they help? On 7/19/05, Joshua D. Drake wrote: > Oliver Crosby wrote: > > Hi, > > I'm running Postgres 7.4.6 on a dedicated server with about 1.5gigs of ram. > > Running scripts locally, it takes about 1.5x longer than mysql, and the > > load on the server is only about 21%. > > What queries? > What is your structure? > Have you tried explain analyze? > How many rows in the table? > Which OS? > How are you testing the speed? > What type of RAID? > > > > -- > Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 > PostgreSQL Replication, Consulting, Custom Programming, 24x7 support > Managed Services, Shared and Dedicated Hosting > Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ > ---------------------------(end of broadcast)--------------------------- TIP 2: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Tue Jul 19 16:02:32 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5949752A03 for ; Tue, 19 Jul 2005 16:02:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 88427-06 for ; Tue, 19 Jul 2005 19:02:17 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 846D1529FA for ; Tue, 19 Jul 2005 16:02:16 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JJ10ZW008447; Tue, 19 Jul 2005 15:01:00 -0400 (EDT) To: Oliver Crosby Cc: "Joshua D. Drake" , pgsql-performance@postgresql.org Subject: Re: Looking for tips In-reply-to: <1efd553a050719112126b8d981@mail.gmail.com> References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> Comments: In-reply-to Oliver Crosby message dated "Tue, 19 Jul 2005 14:21:22 -0400" Date: Tue, 19 Jul 2005 15:01:00 -0400 Message-ID: <8446.1121799660@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/311 X-Sequence-Number: 13552 Oliver Crosby writes: > The queries are all simple insert or select statements on single tables. > Eg. select x from table where y=?; or insert into table (a, b, c) > values (?, ?, ?); > In the case of selects where it's a large table, there's an index on > the column being searched, so in terms of the example above, x is > either a pkey column or other related field, and y is a non-pkey > column. If you're running only a single query at a time (no multiple clients), then this is pretty much the definition of a MySQL-friendly workload; I'd have to say we are doing really well if we are only 50% slower. Postgres doesn't have any performance advantages until you get into complex queries or a significant amount of concurrency. You could possibly get some improvement if you can re-use prepared plans for the queries; but this will require some fooling with the client code (I'm not sure if DBD::Pg even has support for it at all). regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 16:05:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 04FA7529F5 for ; Tue, 19 Jul 2005 16:05:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87301-06 for ; Tue, 19 Jul 2005 19:05:28 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 34F1F529F8 for ; Tue, 19 Jul 2005 16:05:26 -0300 (ADT) Received: (qmail 20764 invoked from network); 19 Jul 2005 21:06:03 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 21:06:03 +0200 To: "Christopher Petrilli" , "Tom Lane" Cc: "Vivek Khera" , pgsql-performance@postgresql.org Subject: Re: Impact of checkpoint_segments under continual load conditions References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c4050719084460b376f1@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> <59d991c405071910133e28031a@mail.gmail.com> <7843.1121796596@sss.pgh.pa.us> <59d991c40507191148e6f9ec5@mail.gmail.com> Message-ID: From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Date: Tue, 19 Jul 2005 21:05:27 +0200 In-Reply-To: <59d991c40507191148e6f9ec5@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/312 X-Sequence-Number: 13553 > It is a subprocess of a Python process, driven using a pexpect > interchange. I send the COPY command, then wait for the '=#' to come > back. did you try sending the COPY as a normal query through psycopg ? From pgsql-performance-owner@postgresql.org Tue Jul 19 16:12:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D62B5529FA for ; Tue, 19 Jul 2005 16:12:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94696-06 for ; Tue, 19 Jul 2005 19:11:52 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.192]) by svr1.postgresql.org (Postfix) with ESMTP id 6E589529F9 for ; Tue, 19 Jul 2005 16:11:50 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i2so1268136wra for ; Tue, 19 Jul 2005 12:11:52 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=ExHRsy09rEcA4teGs23Jxob88YFIHdpzrWHDDtcQSwZlsCWRWruUQhfIDpZjSS08GUidVM9T7K/Zr3TtN/RttCfmT78kQBne/33tGcwJ7ls5X3EGR1YXTIphAcGg661oEGzp7vVk35aeXvWDcYdwngPx4RRun8n/X3QrPol/Aws= Received: by 10.54.57.79 with SMTP id f79mr857445wra; Tue, 19 Jul 2005 12:11:15 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 12:11:15 -0700 (PDT) Message-ID: <1efd553a05071912111f6a3121@mail.gmail.com> Date: Tue, 19 Jul 2005 15:11:15 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: Tom Lane Subject: Re: Looking for tips Cc: "Joshua D. Drake" , pgsql-performance@postgresql.org In-Reply-To: <8446.1121799660@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> <8446.1121799660@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.083 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/313 X-Sequence-Number: 13554 > If you're running only a single query at a time (no multiple clients), > then this is pretty much the definition of a MySQL-friendly workload; > I'd have to say we are doing really well if we are only 50% slower. > Postgres doesn't have any performance advantages until you get into > complex queries or a significant amount of concurrency. The original port was actually twice as slow. It improved quite a bit after I added transactions and trimmed the schema a bit. > You could possibly get some improvement if you can re-use prepared plans > for the queries; but this will require some fooling with the client code > (I'm not sure if DBD::Pg even has support for it at all). Aye. We have prepared statements. From pgsql-performance-owner@postgresql.org Tue Jul 19 16:12:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B130052A0B for ; Tue, 19 Jul 2005 16:12:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 89678-07 for ; Tue, 19 Jul 2005 19:12:08 +0000 (GMT) Received: from cassarossa.samfundet.no (cassarossa.samfundet.no [129.241.93.19]) by svr1.postgresql.org (Postfix) with ESMTP id 72306529FC for ; Tue, 19 Jul 2005 16:12:07 -0300 (ADT) Received: from trofast.ipv6.sesse.net ([2001:700:300:dc03:20e:cff:fe36:a766] helo=trofast.sesse.net) by cassarossa.samfundet.no with esmtp (Exim 4.50) id 1DuxVb-0007Bb-MB for pgsql-performance@postgresql.org; Tue, 19 Jul 2005 21:12:07 +0200 Received: from sesse by trofast.sesse.net with local (Exim 3.36 #1 (Debian)) id 1DuxVb-0004UN-00 for ; Tue, 19 Jul 2005 21:12:03 +0200 Date: Tue, 19 Jul 2005 21:12:03 +0200 From: "Steinar H. Gunderson" To: pgsql-performance@postgresql.org Subject: Re: Looking for tips Message-ID: <20050719191203.GA16813@uio.no> Mail-Followup-To: pgsql-performance@postgresql.org References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> <8446.1121799660@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <8446.1121799660@sss.pgh.pa.us> X-Operating-System: Linux 2.6.11.8 on a i686 X-Message-Flag: Outlook? --> http://www.mozilla.org/products/thunderbird/ User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.018 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/314 X-Sequence-Number: 13555 On Tue, Jul 19, 2005 at 03:01:00PM -0400, Tom Lane wrote: > You could possibly get some improvement if you can re-use prepared plans > for the queries; but this will require some fooling with the client code > (I'm not sure if DBD::Pg even has support for it at all). Newer versions has, when compiled against the 8.0 client libraries and using an 8.0 server (AFAIK). /* Steinar */ -- Homepage: http://www.sesse.net/ From pgsql-performance-owner@postgresql.org Tue Jul 19 16:14:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 722A852A03 for ; Tue, 19 Jul 2005 16:14:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 01354-05 for ; Tue, 19 Jul 2005 19:14:05 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id F254A52A00 for ; Tue, 19 Jul 2005 16:14:04 -0300 (ADT) Received: (qmail 21110 invoked from network); 19 Jul 2005 21:14:41 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 21:14:41 +0200 Date: Tue, 19 Jul 2005 21:14:04 +0200 To: "Oliver Crosby" , "Richard Huxton" Subject: Re: Looking for tips Cc: pgsql-performance@postgresql.org References: <1efd553a05071910427b290203@mail.gmail.com> <42DD420C.2010306@archonet.com> <1efd553a05071911413e50ea90@mail.gmail.com> From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-ID: In-Reply-To: <1efd553a05071911413e50ea90@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/315 X-Sequence-Number: 13556 I can't say wether MySQL is faster for very small queries (like SELECT'ing one row based on an indexed field). That's why I was asking you about the language... I assume you're using a persistent connection. For simple queries like this, PG 8.x seemed to be a lot faster than PG 7.x. Have you tried 8 ? I was asking you which language, because for such really small queries you have to take into account the library overhead. For instance, in PHP a simple query can be 10 times slower in Postgres than in MySQL and I believe it is because php's MySQL driver has seen a lot of optimization whereas the postgres driver has not. Interestingly, the situation is reversed with Python : its best postgres driver (psycopg 2) is a lot faster than the MySQL adapter, and faster than both php adapters (a lot faster). The same query can get (this is from the back of my head): PHP+Postgres 3-5 ms Python+MySQL 1ms PHP+MySQL 0.5 ms Python+Postgres 0.15 ms And yes, I had queries executing in 150 microseconds or so, this includes time to convert the results to native python objects ! This was on a loop of 10000 times the same query. But psycopg2 is fast. The overhead for parsing a simple query and fetching just a row is really small. This is on my Centrino 1.6G laptop. From pgsql-performance-owner@postgresql.org Tue Jul 19 16:16:38 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C8F7852A38 for ; Tue, 19 Jul 2005 16:16:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94696-10 for ; Tue, 19 Jul 2005 19:16:34 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 49E6352A2F for ; Tue, 19 Jul 2005 16:16:33 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6JJGVhO008614; Tue, 19 Jul 2005 15:16:31 -0400 (EDT) To: Oliver Crosby Cc: "Joshua D. Drake" , pgsql-performance@postgresql.org Subject: Re: Looking for tips In-reply-to: <1efd553a05071912111f6a3121@mail.gmail.com> References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> <8446.1121799660@sss.pgh.pa.us> <1efd553a05071912111f6a3121@mail.gmail.com> Comments: In-reply-to Oliver Crosby message dated "Tue, 19 Jul 2005 15:11:15 -0400" Date: Tue, 19 Jul 2005 15:16:31 -0400 Message-ID: <8613.1121800591@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/316 X-Sequence-Number: 13557 Oliver Crosby writes: >> You could possibly get some improvement if you can re-use prepared plans >> for the queries; but this will require some fooling with the client code >> (I'm not sure if DBD::Pg even has support for it at all). > Aye. We have prepared statements. Ah, but are they really prepared, or is DBD::Pg faking it by inserting parameter values into the query text and then sending the assembled string as a fresh query? It wasn't until about 7.4 that we had adequate backend support to let client libraries support prepared queries properly, and I'm unsure that DBD::Pg has been updated to take advantage of that support. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Jul 19 16:22:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 45D82529F8 for ; Tue, 19 Jul 2005 16:22:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 10783-09 for ; Tue, 19 Jul 2005 19:22:16 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.197]) by svr1.postgresql.org (Postfix) with ESMTP id 845C652A62 for ; Tue, 19 Jul 2005 16:22:14 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i35so1264259wra for ; Tue, 19 Jul 2005 12:22:14 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=Z35hxky/1nuUl1h554gAwOeHoL+fHeQOdYBmJbmpFfJcQ5UnGcEBrj/ZYlHqQKkTo467ESS+fd/XuB9VaroSCLDtfO/MQp+r9mhOI5e4FwwRIzVdU7XGGfJXTHIIyLruBNl2+hMkpf+zpnOCzctf1SGYA/q2FQAdw7EH0Gxum88= Received: by 10.54.107.2 with SMTP id f2mr851530wrc; Tue, 19 Jul 2005 12:22:07 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Tue, 19 Jul 2005 12:22:07 -0700 (PDT) Message-ID: <59d991c4050719122213ac3d81@mail.gmail.com> Date: Tue, 19 Jul 2005 15:22:07 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <8375.1121799193@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <4541.1121788647@sss.pgh.pa.us> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> <59d991c405071910133e28031a@mail.gmail.com> <7843.1121796596@sss.pgh.pa.us> <59d991c40507191148e6f9ec5@mail.gmail.com> <8375.1121799193@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.145 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/317 X-Sequence-Number: 13558 On 7/19/05, Tom Lane wrote: > Christopher Petrilli writes: > > On 7/19/05, Tom Lane wrote: > >> How *exactly* are you invoking psql? >=20 > > It is a subprocess of a Python process, driven using a pexpect > > interchange. I send the COPY command, then wait for the '=3D#' to come > > back. >=20 > Some weird interaction with pexpect maybe? Try adding "-n" (disable > readline) to the psql command switches. Um... WOW! =3D=3D> pgsql_benchmark_803_bigbuffers10000_noreadline.txt <=3D=3D 0 0.0319459438324 0.0263829231262 1 0.0303978919983 0.0263390541077 2 0.0306499004364 0.0273139476776 3 0.0306959152222 0.0270659923553 4 0.0307791233063 0.0278429985046 5 0.0306351184845 0.0278820991516 6 0.0307800769806 0.0335869789124 7 0.0408310890198 0.0370559692383 8 0.0371310710907 0.0344209671021 9 0.0372560024261 0.0334041118622 =3D=3D> pgsql_benchmark_803_bigbuffers10000.txt <=3D=3D 0 0.0352520942688 0.149132013321 1 0.0320160388947 0.146126031876 2 0.0307128429413 0.139330863953 3 0.0306718349457 0.139590978622 4 0.0307030677795 0.140225172043 5 0.0306420326233 0.140012979507 6 0.0307261943817 0.139672994614 7 0.0307750701904 0.140661001205 8 0.0307800769806 0.141661167145 9 0.0306720733643 0.141198158264 =20 First column is iteration, second is "gen time" to generate the load file, and 3rd is "load time". It doesn't stay QUITE that low, but it stays lower... quite a bit.=20 We'll see what happens over time. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Tue Jul 19 16:34:40 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 79A11528BB for ; Tue, 19 Jul 2005 16:34:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 15817-09 for ; Tue, 19 Jul 2005 19:34:27 +0000 (GMT) Received: from prospect.stortek.com (prospect.stortek.com [129.80.22.147]) by svr1.postgresql.org (Postfix) with ESMTP id 020FF52A22 for ; Tue, 19 Jul 2005 16:34:22 -0300 (ADT) Received: from prospect.stortek.com (localhost [127.0.0.1]) by prospect.stortek.com (8.12.10/8.12.0) with ESMTP id j6JJYMWl023978 for ; Tue, 19 Jul 2005 13:34:22 -0600 (MDT) Received: from unix.stortek.com (bullion.stortek.com [129.80.16.43]) by prospect.stortek.com (8.12.10/8.12.0) with ESMTP id j6JJYLbc023975; Tue, 19 Jul 2005 13:34:21 -0600 (MDT) Received: from C118181.stortek.com (c118181.stortek.com [129.80.81.172]) by unix.stortek.com (8.12.10/8.12.0) with ESMTP id j6JJYKZb025007; Tue, 19 Jul 2005 13:34:20 -0600 (MDT) Date: Tue, 19 Jul 2005 13:34:17 -0600 From: Robert Creager To: Tom Lane Cc: Robert Creager , pgsql-performance@postgresql.org, Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS issue? Message-ID: <20050719133417.00001473@C118181.stortek.com> In-Reply-To: <7322.1121792062@sss.pgh.pa.us> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> <20050719105014.00007635@C118181.stortek.com> <7322.1121792062@sss.pgh.pa.us> Organization: StorageTek X-Mailer: Sylpheed-Claws 1.0.4.1 Win32 (GTK+ 1.3.0; Win32) Mime-Version: 1.0 Content-Type: multipart/mixed; boundary=Multipart_Tue__19_Jul_2005_13_34_17_-0600_JtTr6Ylk_Fb3N5yn X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/318 X-Sequence-Number: 13559 --Multipart_Tue__19_Jul_2005_13_34_17_-0600_JtTr6Ylk_Fb3N5yn Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Disposition: inline On Tue, 19 Jul 2005 12:54:22 -0400 Tom Lane wrote: > Robert Creager writes: > > Hmm, I hadn't thought about the possible impact of multiple concurrent > vacuums. Is the problem caused by that, or has performance already gone > into the tank by the time the cron-driven vacuums are taking long enough > to overlap? All statements over 5 seconds are logged. Vacuums are running on the 5 minute mark. Log file shows the first query starts going bad a 9:32:15 (7 seconds), although the second query start before the first . The first vacuum statement logged shows 1148 seconds completing at 9:54:09, so starting at 9:35. Looks like the vacuum is an innocent bystander of the problem. The first problem queries are below. Additionally, I've attached 5 minutes (bzipped) of logs starting at the first event below. Jul 19 09:32:15 annette postgres[17029]: [2-1] LOG: duration: 7146.168 ms statement: Jul 19 09:32:15 annette postgres[17029]: [2-2] ^I SELECT location_id, location_type.name AS type, library, rail Jul 19 09:32:15 annette postgres[17029]: [2-3] ^I FROM location_lock JOIN location USING( location_id ) Jul 19 09:32:15 annette postgres[17029]: [2-4] ^I JOIN location_type USING( location_type_id ) Jul 19 09:32:15 annette postgres[17029]: [2-5] ^I WHERE test_session_id = '5264' Jul 19 09:32:20 annette postgres[17092]: [2-1] LOG: duration: 13389.730 ms statement: Jul 19 09:32:20 annette postgres[17092]: [2-2] ^I SELECT location_type.name AS location_type_name, Jul 19 09:32:20 annette postgres[17092]: [2-3] ^I library, rail, col, side, row, location_id, Jul 19 09:32:20 annette postgres[17092]: [2-4] ^I hli_lsm, hli_panel, hli_row, hli_col Jul 19 09:32:20 annette postgres[17092]: [2-5] ^I FROM location JOIN location_type USING( location_type_id ) Jul 19 09:32:20 annette postgres[17092]: [2-6] ^I JOIN complex USING( library_id ) Jul 19 09:32:20 annette postgres[17092]: [2-7] ^I LEFT OUTER JOIN hli_location USING( location_id ) Jul 19 09:32:20 annette postgres[17092]: [2-8] ^I LEFT OUTER JOIN application USING( application_id ) Jul 19 09:32:20 annette postgres[17092]: [2-9] ^I WHERE complex.complex_id = '13' Jul 19 09:32:20 annette postgres[17092]: [2-10] ^I AND location_id NOT IN Jul 19 09:32:20 annette postgres[17092]: [2-11] ^I (SELECT location_id Jul 19 09:32:20 annette postgres[17092]: [2-12] ^I FROM location_lock) Jul 19 09:32:20 annette postgres[17092]: [2-13] ^I AND location_id NOT IN Jul 19 09:32:20 annette postgres[17092]: [2-14] ^I (SELECT location_id Jul 19 09:32:20 annette postgres[17092]: [2-15] ^I FROM cartridge) Jul 19 09:32:20 annette postgres[17092]: [2-16] ^IAND (location_type.name ~ 'cell' AND application.name ~ 'hli' AND hli_lsm = 1 AND col BETWEEN -2 AND 2) Jul 19 09:32:20 annette postgres[17092]: [2-17] ^I Jul 19 09:32:20 annette postgres[17092]: [2-18] ^I ORDER BY location.usage_count, location.rand LIMIT 1 Jul 19 09:32:20 annette postgres[17092]: [2-19] ^I FOR UPDATE OF location Cheers, Rob --Multipart_Tue__19_Jul_2005_13_34_17_-0600_JtTr6Ylk_Fb3N5yn Content-Type: application/x-bzip2; name=pg.log.bz2 Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename=pg.log.bz2 QlpoOTFBWSZTWWjrIOgAs/Z/gBI2YgBQ73/3P/ffq77v//FgNUAhAAAA74hFHQAABgTcfc3LMAfK Y+A6XWAALx7liSe5uUPs97PvvrqXvPvePln33zt4PZ73ij3idwXB26l3RnB73cTj7AAJK8BGlN2x pI3gAAAAAA77wyjyA+bgtAGkENVQGYaAMoiiqDaIGgyGIyAA0MIwgZGBpmkoQikaepj1RoAADQGJ o0GhoBoaBj1UlQFMmAACYAAAAEwTEZNMQSeqUmpEKepNDQ0AMjQ0epiABiMQDCATVSUnmmIp6mjK GhoAA0ADQ0A0yAAClIgiBCYpMiNPQg8p6RoyaPUw1GjygaMNRBgAwZvxv+L5Hx/R9+YiB9v9q9+g HGYUSqZJINmStCmZEz+CQBPNiW5zeBszY2Iht4SRzALqZCmZYCUolktQbFP4tZisiWS1BsU80GZl gJWRLJag2KVrAFZlMkUGwz3QBKuJYnSGxTzWAKzKZIoNlsxMxLbu6Gy2bNk0xJIbLZiZNMSSGzJW hTyZmpYCebEtzm8DZyelWoiGWoocxmCSNMkkGy2YmTTEkjOOjtygBACDvJ4oaIAQAgBBTucGZvjq GePHPhZxwd+Q8CAEFAA7cEkzte3PZduEvHsaJ+XUknFQeQGybUGiSPLygvtvz8+vNZz5nfnseUlo AHkINkkm0ACmYeXUkeP54fA2+T3QAuZEslqDYp80GfLjt30zEQ3a7vR3ALrsJXEQypagsA7XNi7E zMiZq47YAvvYlu93obM2NiIba3dHcA5XdiIjtz3sxEN2u70dwCyZiIY9SQWAXXJ5NyJ5yYiG5XN4 DgFkzEQx4kgsAv25D+Qzd73aHr+eLwNtsyUxILkkhtU+6p7MVFd2YiG5PKpqlS/ikkzEQxo50e4B zlzvOTRBkY9UKlmFZkgc4SGu2bdnlD10kg27PvI6hMzInuzEQ3Dw8CwC/LnnOTEQ3K5nBzGYWTIB JIa92NiIbeEkLAFXdiJ5zIiG5XN4OYBztyH9gzkiZmRM1IAnvYn1/PD4G22bTPlMSSGzJ7oUzIme zM+yzT0xNoNSlWPH88WZYmfJnvJiICekj0cwDteSzHvYkdfvT0NvJ7rALmZTJFBsMrRM9lmnmRI6 /OHgbdnzZVzEQx8RQ8xmCSNMkkGy2YmTTEkhsyfErURDKlqHMA57c7zkxED1673ejuMJiB2fZPez EQ3K5vBzAOVyd5yYiNfl8PBmAXXss0pdoeP3q6G3Z80Ac5kSOPzh4G3k9UdUzUs092JHX709DbbN pntMSSGzJ7qnsRVTzZiIblc3g8wDtTLAYlEslqDYp5oAKyJZLUGxStM9iO0wE9MS3e70NmSoiOx1 XzZiIblc3g7gHK5O85MRDcrm8HMAuuwlcRDKlqCwDtciO0e9mIhu13ejuAWTMRA8fvOc4OYw5zkx EBdJBsDnNzy2bvexLE7wNmbGxENqJI5gF+3Ifve9uIgevXnh8Hn1Xmqttb/BtWNrRYKixWxUZlmM gqWRgITqCpf1Q8RH8cRyQv/lKn9o9pHqyzGZmZYVH+SwZYxgftRH8kR85XxqsbajVRUbFJr8VyNv 1O3Sbfr92orG0VG18bfGzzx5rXNvFa5qedRt42uauXd20Wp3ba+K+LxT5207tvndt426axrMjMMw ZhvCccaP9Qn5xPVR9iv0J9T3qfohIP6IjcbyyHzCf6FGvqjGVjJllV+OEg1awt/S/QbExURIY0dv NRs1WKZkp0JlGCZLTE5S+SFf+qv78r0xNYuNJoNDRqjVazhR8lGkakP0FfaI+/33/GzWZvea23uA oxQGJEJCMUZAjicbD3PU277npet6118RGrWXWvDGYZ+PesMMuv4hPyiZFGCfBX9MR+WPyyP3kVtE fwif54PAn2iO6VMSXcR2iNx3EdRHQn+ERqKNRR+SEg5oqVr6T5nyvrPoavtPwPuPsd3EbvAv987O 7QvE5PNqdnduLJyZai6J5nZlwLzOzm2q2dXkWTweLUW5wasnRxci5nk5uIsnidHY2d3IupybuYsF zJxGXmLzJ4nR3aFuYZeReYvJxaFsnE2dzyc3EWpNTg5uhdE8zmuRcRzaF5Jk5PF0Lgnc1Nnd4F2T c5ObsWoujd5iwujBuTijicni5i0TsnkXYtiwXgnQsJxPJ4paJ5nJl0LBaJ4F2TiN3YuIsPFsXItR 4zGDQXRMKdCwWC8BeBYTmdnd/BEfvxH8MR9lH9ER/xE4UfwCf2CdidifqE5FcxHER+/Ef2/ca/Pl Bc1OsTAwTqI0lNRGQyI7q19VdVr76tX43RWDI0mSJjEpRlj8tVfaq19ttvmqvy1V/a3DAAAGAwGj aDYNGgMBg1AGA1ivtbW5bRQajRtYtsfn2NJAoCSEgmIYzKzMYzGHKSyDgTuUfqGRHiFbg4E7hXrE cCaEoksXr+r1ejJ7LfpNUXsey7zxfavVvSh1PTtETmp729p6T3gVXgfNLvMRJrHykUsjZgWRel42 LSSerwzsHRb5IMO5VKAq2Dt1e5Rp1tUnmjs5g0jXD7Fvu1WAiA77YiFn7/02ACVXMSvYJMKV62FK 7pC7lR+5KPqO/qJ+6PHlR2o/xHInYTtEbHkfT558fXX0z759s6xvPrnOay33rjLX0QRp+h3WpIYt iK3Iga4l6qzVut0RK2Jd7120br18Te1UdqfAmLkcjXD5gLnmWdGw2qK8HD7bGt57IdSiJxOadGjj 51qOZEv/6wDASuNSX8bAMBYl6in7Pdqsru3E4+N1y6zWs444547aznO/YgvmPzKP6ylX+kf3hNyH 86S6pHH2+u6b++9lve9cbLmZQMwAzEAYkAJIAxIAzADMAMwASwCULmKgZgBmYgDMLbcxKkzMkmJS BhbalbbUkqEoXFbbUrbbmW21K224spElJEpSySq224q8zGBmBMUpbittqWUJQqVttzEq0kwEsGsd olhIsUlMSCKUtzFQMwCULitpiQFuK2JZLMSkmJSRJSURbbiygJIAWIALbcVttSttuK22pW0WAGV5 lCUKstpmLMyeX24vsorc5F4ikUikUikUikUitzkXiKRW5yLxFIrczFMGNtttimEUyMbbFMIpkY2K Zzc25F4ikVuci8RSK3OReIrc2225F4ikUitzkXiKRW5ttmZmZtq3m8z9A4CfUJ+8J+KI7iPeI1Ee AnEGhPA6UdQaE/JCvKjuI6iPt8X1ffEzMb/W7XX9Z57VzS+Mbxyxt5vbxaF587XXrz2rml8Y3jlj bze3i0Lz52uvXntXNL4xvHLFvN7eLQvPna69ePajS+Mbxyxt5vbxaF587XXrz2rml8Y3jljbze3i 0Lz52uvXntXNL4xvHLFvNbuxG8c1vL28WhefOrr1489OVGlzG8cWNrBalB0ximNPHgtrJq1DWJt6 8m81GlzG+OWNne+d83t6tD403lRpcxvjljb9a3Z7G8lh160k7B4pIpZGq1HXjeFNaWtN2uJzm5xG k0217GQbiIy6a452U6muaabVkxttTg4lxS48sjx7sI5wp66zFfarml8YtFuvteLQ8aZ9fbzrwHnn pK+XdJy5Bhmuka6o018eNQUeVGmvHirzUU148agok0fa3c6uWS24ZnyvpZNt0Ie3V5fFoXvztdfT z5X0slt4Mz59VeNL6Y305Yt5Xqybb6hD27re3vvncprXS+ldavJtPmvd8uWhefXa6+nntXNL4xvH LFn1ou+ld1UW+lvrL1aF79drr149qNL4xvHLFvK9WTbfUIe3V5erQvfna6+nnu1ezcxrvrXzyo0v pjeuWNvKjS5jeOWLeb28WhefO119PPauaXxjeOWNvN7eLQvPna69ee1c0vjG8csW83t4tC8+drr1 49qNL4xvHNY2sM3FkSEq0tkD2t3eScrnVYn6fNW2RsunuFa05WW94KzZeaW6FO1lOql6nS+3kg3N m7DzrqjW7dg3Ly6utVixkOHEuoxYth0DMoTR3aO6XeDdKqqZup2C6tbRGc88a28+PGuPGa/PEe6Q fniMZiZlmSNJkiGJlGWMi0YyNJkiYxKUZYlWmh70emDxPM7uItTIsFkrL0DUAG7ET4O0/h9wjazK rdGbtVCiTLDHGOJfMjHxxcmIiFaQ0/aaaUc5/pxRmS1MAzFmZgOd53GT7EPAAZtADNxW9aEWABcB gwbXirmQ8FZ42qLMlWYK51oVvETWiHELz678+nHjvnGtaed7lPFRpgpKuKspE4bAAZABgFmnE+gB gNYCY2jpe+vDgBmLBrzFFuZ2HFF8l5LNBuH3rPQV6YJ4wR5wix1oVvCHtCQb8qMBgMExynt6aWta 0Q2SSCGwS4xBM22222wB3ckA7u7u7uAAAAkOcXO7uAA9qrfJt81wLuycUtxdThDUWrZ7TRNgaAGE MGDE1WVuU88e9m6p4oTEE7doMwZt45T8U0+8jJqz4Azd482AGaHADNx2DADr6aIe/XLZTe97FXO2 hWtdO9f0xAmADD2+8xuRAag4uuUWAYCwWzio768+PHa86wDAdYL344zMA749Ozcdn5zkjsq+sAwF jrANXXffeSkamJ+I646wDAO3QGYTXFbF18Rbj0w2ZDuOMAwHANHguwviBJgRF5UuKvGpRopWlHPh tuJKZ4Qkw46a69PMR5qvbz8c+rfn5dcenDjrd5iMJIPJePHPnv4efHt8uePWI9mELepgjFZO7bO7 3qwCEySTENmLSZM6cAAA7uSB3d3d3cBmZmJrGZmVyV5q6ZGXm5PIvEyYLupu0LJwTBcCwma8b9e+ ta9eUfMXZ6VGdn5vF3TytV5X32JVrwvJvGsTd3AYO7O4cOzfLAMAAD1xzfe9HX92i9GdnLXmFci3 565a9hWu7Ldg6Lo53ZqVpaBrtZd9+XltLBmBYMG+O9tw5+J4QAw6wTBvfQ/nznfYnZdp418d7dYe +MYvHGmfGmvIgQGAQZm7nm+/HZAdx448cO7S492Mz2QK30T3RNe6q9fjrs99adY1lzjMZzpcMdpX ZcfDg6+NPbHt50vJHmHGu+PPu52euFyMfXzyA/YG8ADNWEZMTMUwAA8DBgzcDX56ltm1tr8dWr8v QAkB9++vr1jFaXd3d3db3d3ZS6tbbbeYnN21y2yFkCtvFC2ltrltrhJJJJJJJHkNkHOc25CbaaOc 5znOcANhrWtbi8xdRzPc8TJo2ZJ0L3JnAJdeEJ3eDePdGU9BqcCnDF5mJdmEUZApxAkxjr5ZmAAa zyOTmvcDTy8kE+9D+PD/ER50MGb5YMynmguHBmDgiKcgBmLMw8Gue2vLXSz4SRpSsQ20235Yyzxb YbYM6BcKjAuPLXS1cblt0aa6xVVkhgTHvtR531xkcSZgzAAEAMzsGAaPZ4JeMgn1dPCRXYrKr317 8emtonMXHnw1nnrz67eN+K69udennjq9Os0VU8p8iq6BpbdVq7Vr6/EAIQe97+n9vPVe/Pnu7u7u 63u7uwLq1ttt5im3arLSBQvXVuc665znXTnOupuAADYIAbDbbYADYAElnZmkmpekXNxPSLmnEWjC bJxbFg8zFEllS6ljXJ57z11eC6+FqJOLwCBdLfkBgwD3POQZULX9YMBzuZfPTV3I+Wm/nvWAbh56 JkEFT3CzMIgec195GsAvJSHbfecaOcAZgNGO+ed5vH549jgAb3sxDGzzC5AAG9fypyHm2YAM9jbN daBdeRNaGAAzbDrysHh3joMze+SI5Bh049vlhmAAA+QzMOEAMI1V1VKtKtKtfpfmAEIe+J1eSNLd u7u7u7re7u6wurW223mKW7VbaMKBasdtJBSCkkkkkkkkkZgDGbGDZGAAAa1rUXd5snVh6mLiLBaJ k2LbWwsCWkANLCWhoAYSwDXyoy6m8MGzmZsSrlVs5Evg1RPywDMAwU1rpzSzE7xXoAaqqBZ78R8D sQmDAdIuMt9hEsGDFT2Rzvsc+WudPX5I9sLeRZkNdal6lee3Hnrx45gvOM8YQzADyBo57M5E6BoY bt6HncLtesGYBwJd+01mB7ZkPImSGZgPSjwbPjhJ+8AAyxq7y3dHzZkPUoBgzoOOcuo2dgvDW/Ha 2qv6FWry0+lWlWr8/0gANttt7PW70TTbbpbbba3bW5a22222222222222ud3dW7IbuyG7bbdkkjb bbZd3d27u7Ld3dg22222m2m251yu7u7ggDgAhDnDXd3d3dxAGtr71adr8tSu1flmC0aryaC3MJgs JsnAuwAzgBnADOAwcAMTELI8iLvDSzbURANobRzd+WZmDASsfZgbFUU8KrkSr0E+v4IwzjQ72rfN qGwRL24nRajch43AnVBOr8wccH56NHPJHSCCOxZGB/Lrrir6+jyKu4F7O9egOnNt9y7qPKFbyFUZ o87d2IBYTzzcLxFZpIr42oYJmAxx75mZmY22222356edjbbbdttttrdtUlrbbbbbbbbbbbbbbfi3 d3ZN3dIbpbbskkbbbRbtKWuW2uFAAAAAACSAAgAAtLHKwbkkklttttoAASQAAAAK22222m2m3HG3 l3d3d3cQBrXa/OrK/Ku2t5vvVcnErg1bNybFsWhcRsnjvx4idrc01GC6jy31EVd7NfADMA93ggRC 0aJ54a885SRdZELN98vOZrjmSVz8fHx5XriO3V3TBnFy8czPBOckgMfKgPYO4A3kOnvXh/B3QG7A jl8d9oBgOXexIoBgFB73K8e2eC9uPXz6451143Iyk+IjFWvGZmZjbbbbbezv4vVjbbbe7u7u7u7r e7uwh76AAAAE+e/Pfnp7777eeee+wu7rbbbbbgbo5d1u7d3W5d0AAAAAABgSAAAD3buua2DYADYN 7obt3d3bu7ugAAAAAALFd0s1rWtFeonr7HQ5PJk3bjgWo4FwdE4FuxGp7eMznzx6d89tPOd9a85L usUQctziDxjzfwwDTmaZedEmhZzeZBsQYyQbPx8VLAM3gwLrvnr0/NAYDhsJGKNUJyFpYMA8HcXc 8shduQGAW8FXFkVt0wYDCHrKAag/Z1+1ynp41WAwHNd7alGS881gwHMD6uWsfXc5HFAUzwPoyE4A YD4YBpMlnd3eawCEySTEJgl2Mjum0aAqiDbzataq+x9669SasNHqaK1NXsqzNePbfPnvp0769cRs m5iBciopuW6zdl4svMzsrdxzQ8JeJ52Xg5LzA3ObPtz2EAGbnaEdyzsAMMnHfmCrAYcpnhxUABgI PZpBGbFgMw55eEVyj3pawj0wk7yR6Cu+u9+neukDhhGZC8a1EZuK158+ernvJIh5neySyaSyTdq2 d3+yzAhMkkxDZi0YxMAAGwQA2E200AA2ABZb41sts/xZPbV6JC6iF7ypKrpq6VOyp2qmVuI2bmiY LLBZai0FoWRhk2N1oWplTcwLdkyLUbWhek0bFhs0LLVhxNRai1OKOJq2Lg3MFwLRqcGrVwbTZNmT YuJo3UybriZNGC4nFT1SVzHMmiqm5q5WphtQG4uKXFulBuaNmhaFwTCcJFcmrCnE4N3M4mjdTc0T iwmjiblYLipwbJk3bNE3RcFbiZaFpHAsXFLcm5xanAtzBaNC4RaJu1XEwRhguJuTBcHA0aJoy4i/ BUL+SQyVX7B/wiNSL/kowlMiGQh/tql841Ie4RO+qVXiCeBK6pSt9JWCjxETx0YlK75E1StytKPi jcRtEfrKp7R0P/AxR4kfrK/niPnEfAn/WDIPEjBNKVylK6aIbyFzoLyFCPSI8JLgpV6lKvI2PnEZ LV/FVr+JFoxkaTJExiUoyxda1ciSEUmbFCYUiSIGKIv4+7p3XSEUnLti5cTpd25dMuuuXXSLu4cl y6ndHWLWv49VrVX57W1vfkWjGRpMkTGJSjLGRaMZGkyRLMsxhhmWLMakcED9g9ADcfjophWFKtwg wpVwksG4wfcbiPelT/yUq/GAvuUrCutSuqSb5SvLCakTjSjCqdgHcah7Qhg/SkvrEegnoNKPJSr0 j4iPVE1q2lK8hE2itFdaskrlUPREcDZXyHrEeyjClXqiND5lYVT8kR+qEg5hIP4v2xX6Lup/m8rV bzVy8WjP2bujXSiNNfW43E3NOXjXScgWWTSSz1Za/NjkLhQzGKMUfmHPLMjMY2L/O7RowwU0a8tq +voAAAABAAEBAAAD22r5W33XeeXd1ru7d3bmw2222Qm2TNhsoohttnV3du4A/D6+fPh5tvqSGkma 9+3uq5Gye7b5JCgma/D3VfFtbpt3JJZ2Jbe1p2Zbdp2naglgopsNtttkJtkzYbKKIbbbdXd27gD3 358+HlWvKtferX4Va+Va+1WvbbeL7fhFoxkaTJExiUozXzuvltvpfa23xeW2L70nkToTsTwJ5Ccr qMus870ohhAl2AgMYYCGAcE6Tm6kktSu7qrSSSWpXd1VpJJLUru6q0kklqV3dVaSSS1K7uqtJJJa k7wkkktSqrFVaSWpalVO91VpJalqV3dVaSSS1K4rBWLYeJuU5UpRtXqM2Ddu8CItYNzbs2tK1Kqj YvcRpUlD1lokrUneEkklqTvCKKKRRvAr2ox0UpsTpcmiKqRETl5pSS1La0LNqMdFKbE6XJoiqkRE 5eaUktSu7qrSSSWpbWhZtRjopTYnS5NEVUiIiInLwpJJJXdylas4UqmlaKK1J3hJJJald3VWkkkt Su7qrSSSWpXd1VpJJLUru6qyikkShEJJJIlCISSSRKEQkkiiUIhJJJEoRCSSSJQiEkkkShGXW2TS Sx6rNRSJQiEkkkStVZmzoSVxeoXhFpVtvOBPUyH0Wjoq5KzStKd6eJSSJRKytBs1spyldzGlybIq smrqCdy7URu4U9FXdKVmu97ukpEp3uZMq7vdLqbm9OPmrUEla1JEqHp4lJJFEoRURKSSKJQiEkkk ShEJJFIlCISSSRKGZl1Kwl6rIukZTztJzCwRGXCyzERN1VWhp2tlOUrt50uTRFVW4pJM5mTSxGpi IlTWHSkkkqqEtVnClRMzYmMxKYiJe7JMIq4SJhGJlVl2c3ZqNKuEiYRiZVZdnN2ajSrhImEYmVWX Z3InZ0rISShGJlVl2c3ZqNKuEiYRiZVZdnN2ajSrhImEYmVWXZuyq3UUaLw9UtkmomxOlZBSUIxM qsuzuRMzMxhVwSTCMTKrLs4N2ajSrhImEYmVWXZyIzZqJncrTNalWPlTaW6ZM3UnIOIwjEyqy7O5 E6HDk6C8qje7RTrUIF5QhYitMbcziEi8QhYUtMbczhszhwmNuZxCReKhQqqfHInawpTimtjMJhGJ lVl2ckFYtyEahE3BRMIxMqsuyZCtHNUpHIKShGJlVl2dyJ2dKyEkoRiZVZdnMwq4JShGJlVl2dyJ jCrgkmEYmVWXZzdmo0q4SJhGJlVl2dyJ2dKyEkoRiZVZdncidnSshEh0ktShFB0ktShFB0ktShFB 0ktShFB0ktShFB0hMylqS02HgqaR1JZlZdZmh0rSSpFB0omKobkZk7k4lm26u8OnTKDzKd4qYmRM zEQ8Te5KU0owkOVE7l46VWowkOVEpalC3Q6SWpQig6UzMpalCKDpJalCKDpJalCKDpJaku33Oxu5 wAAQGwTt9vrrbqYvXXOrbbZJbaFVskkkkixVpNNtZ3zEszW371ra/1a1q/RtWpf7eu5xhc4nduLn E5xhc4y5xOcZc4nOJzic4wucTuu5xOcYABc4wu6uLnGXOJ2bXQgraM2mIIKhBTNpmbSYoEM2mZtd tCiGE7ru6uLnE5xhc4wu67nbXRm10ZtMQzaZm0xRUIIZtMzaZm0zNpiozaZm10ZtMQQQQUGbTFEn yWGIxRgifpHMR/UNqJpEdRRhKfzSTQm4V/7E0SxZLIklmtktxRIBRAIABRAJIJRAJRaBIBYLAKJB BBAJQAKBBaJAIAAAABABABRKBIBRQBABALAAIALRepJZL3PDsssn12S286ktgyBV/Ui3USlfwiTo kLbKrUhfZVlEHYIWIVS4SD/AJ3JR/YJgTY/YksiMq2pq1Na2xFBRFGIxERQRGoxoxEFAEGJIxEYi A0RRiMYIgiCIiDEYAEwRatW36f7+rV9vLap/9Q6TAxUwTKUv+kRiapK58KkTFIsSTjhRc8RB9ATi kuiF/co+f64FoTuBft+g5ikGcojmDIo5E0J9YjSQapBiUmAmwTwlPlEd0qfXilT7QTY3CQcbkm5J zCQfFKX3D0lB9pIwE5hIPtEfIpX90H4hNFS8JSOgT21Ki+1JB7RG5SXpFX7R6xKHEoylGDKVJ7jv 4UT0Hkff9xFRqKSY0lLExSVfqwkGJfihip3EcRG4jcVcfelL/FKR4E+gn4QaiONqtv1Pvq1ebWt+ QyYMQQBkCojECZRBQJGYMQRIMQBIJsZtWrSpbbV/8qWrb7RaMZGkyRMYlKMsWtf69WrEqcKJ6gmI j6RGqSyI8FKuBOhMCeKVOaUtUqYSA/dok7ijXUVT6QkHwNQkHXA+Q4KPGRGUF2+kKJ2yJuq7lVLr JOgitkgMj8ISD20pH3oT4QtvPRSd0iZVZkTmqSq7RUvQfAT5AnEK0MUfRUMkn+yFwIXHzlSVWeaJ OSrMTIjIOvSoo9qVPv3FIOJI+oJ13JMiOZJtEc0k3Sp9oSD2CfgPuJwJok4yQOwh6x7QkGUFzJNw kGqqlWtCnxKS8wkHrAuBMIo4CftH3/TCQevQySM3QX4tFH1iMjxmZgAAAAAAAAAAAAqte7Va/Da1 t+NoXFFJ1bpSv/F3JFOFCQaOsg6A --Multipart_Tue__19_Jul_2005_13_34_17_-0600_JtTr6Ylk_Fb3N5yn-- From pgsql-performance-owner@postgresql.org Tue Jul 19 16:36:12 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0C252528A6 for ; Tue, 19 Jul 2005 16:36:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 14737-07 for ; Tue, 19 Jul 2005 19:36:07 +0000 (GMT) Received: from cassarossa.samfundet.no (cassarossa.samfundet.no [129.241.93.19]) by svr1.postgresql.org (Postfix) with ESMTP id B181152A0D for ; Tue, 19 Jul 2005 16:36:02 -0300 (ADT) Received: from trofast.sesse.net ([129.241.93.32]) by cassarossa.samfundet.no with esmtp (Exim 4.50) id 1Duxso-0007WX-0e for pgsql-performance@postgresql.org; Tue, 19 Jul 2005 21:36:03 +0200 Received: from sesse by trofast.sesse.net with local (Exim 3.36 #1 (Debian)) id 1Duxsn-0005dj-00 for ; Tue, 19 Jul 2005 21:36:01 +0200 Date: Tue, 19 Jul 2005 21:36:01 +0200 From: "Steinar H. Gunderson" To: pgsql-performance@postgresql.org Subject: Re: Looking for tips Message-ID: <20050719193601.GA21311@uio.no> Mail-Followup-To: pgsql-performance@postgresql.org References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> <8446.1121799660@sss.pgh.pa.us> <1efd553a05071912111f6a3121@mail.gmail.com> <8613.1121800591@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <8613.1121800591@sss.pgh.pa.us> X-Operating-System: Linux 2.6.11.8 on a i686 X-Message-Flag: Outlook? --> http://www.mozilla.org/products/thunderbird/ User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.017 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/319 X-Sequence-Number: 13560 On Tue, Jul 19, 2005 at 03:16:31PM -0400, Tom Lane wrote: > Ah, but are they really prepared, or is DBD::Pg faking it by inserting > parameter values into the query text and then sending the assembled > string as a fresh query? They are really prepared. /* Steinar */ -- Homepage: http://www.sesse.net/ From pgsql-performance-owner@postgresql.org Tue Jul 19 16:53:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C3F3E528A6 for ; Tue, 19 Jul 2005 16:53:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34282-04 for ; Tue, 19 Jul 2005 19:53:07 +0000 (GMT) Received: from sccrmhc11.comcast.net (sccrmhc11.comcast.net [204.127.202.55]) by svr1.postgresql.org (Postfix) with ESMTP id B485A52A0D for ; Tue, 19 Jul 2005 16:53:05 -0300 (ADT) Received: from jefftrout.com ([24.147.120.205]) by comcast.net (sccrmhc11) with SMTP id <2005071919530501100qg2nae>; Tue, 19 Jul 2005 19:53:05 +0000 Received: (qmail 85244 invoked from network); 19 Jul 2005 19:59:18 -0000 Received: from c-24-147-120-205.hsd1.ma.comcast.net (HELO ?192.168.0.106?) (24.147.120.205) by 192.168.0.109 with SMTP; 19 Jul 2005 19:59:18 -0000 In-Reply-To: <20050719193601.GA21311@uio.no> References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> <8446.1121799660@sss.pgh.pa.us> <1efd553a05071912111f6a3121@mail.gmail.com> <8613.1121800591@sss.pgh.pa.us> <20050719193601.GA21311@uio.no> Mime-Version: 1.0 (Apple Message framework v733) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <654C15AD-4ADF-4792-8082-BE94ECF86F3A@torgo.978.org> Cc: pgsql-performance@postgresql.org Content-Transfer-Encoding: 7bit From: Jeff Trout Subject: Re: Looking for tips Date: Tue, 19 Jul 2005 15:53:04 -0400 To: "Steinar H. Gunderson" X-Mailer: Apple Mail (2.733) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/320 X-Sequence-Number: 13561 On Jul 19, 2005, at 3:36 PM, Steinar H. Gunderson wrote: > On Tue, Jul 19, 2005 at 03:16:31PM -0400, Tom Lane wrote: > >> Ah, but are they really prepared, or is DBD::Pg faking it by >> inserting >> parameter values into the query text and then sending the assembled >> string as a fresh query? >> > > They are really prepared. That depends on what version you are using. Older versions did what Tom mentioned rather than sending PREPARE & EXECUTE. Not sure what version that changed in. -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-patches-owner@postgresql.org Tue Jul 19 16:55:11 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9EE6A52833 for ; Tue, 19 Jul 2005 16:54:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34623-07 for ; Tue, 19 Jul 2005 19:54:36 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id 3BC2052A38 for ; Tue, 19 Jul 2005 16:54:33 -0300 (ADT) Received: from ibm-b.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6JJsVjA003280 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Tue, 19 Jul 2005 12:54:32 -0700 Message-Id: <200507191954.j6JJsVjA003280@smtp.osdl.org> Date: Tue, 19 Jul 2005 12:54:53 -0700 From: Mark Wong To: "Alon Goldshuv" Cc: pgsql-patches@postgresql.org Subject: Re: COPY FROM performance improvements In-Reply-To: References: Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.057 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/381 X-Sequence-Number: 16721 On Thu, 14 Jul 2005 17:22:18 -0700 "Alon Goldshuv" wrote: > I revisited my patch and removed the code duplications that were there, and > added support for CSV with buffered input, so CSV now runs faster too > (although it is not as optimized as the TEXT format parsing). So now > TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. Hi Alon, I'm curious, what kind of system are you testing this on? I'm trying to load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm interested in the results you would expect. Mark From pgsql-performance-owner@postgresql.org Tue Jul 19 17:08:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 73B5B52A0D for ; Tue, 19 Jul 2005 17:08:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 50859-04 for ; Tue, 19 Jul 2005 20:08:44 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.201]) by svr1.postgresql.org (Postfix) with ESMTP id ED0A952A04 for ; Tue, 19 Jul 2005 17:08:43 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i3so1303188wra for ; Tue, 19 Jul 2005 13:08:45 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=PIWUxpRgnErhtfEfneFNyuHwwUInsGpzJUt58vqaIr3TuY0/US0bv11mOoma8GjzBfF7DIZewc5yy7ixildx32uR0G54l3x5MmF0ibGufZLxo/g/hDopyfe1ZLbUeUp78DiDzWVxF9VRFEzZxp8ePAhZNJg9/ufOHD7YNjMtRTM= Received: by 10.54.130.10 with SMTP id c10mr877747wrd; Tue, 19 Jul 2005 13:08:11 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 13:08:11 -0700 (PDT) Message-ID: <1efd553a0507191308219a098b@mail.gmail.com> Date: Tue, 19 Jul 2005 16:08:11 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: Kevin Grittner Subject: Re: Looking for tips Cc: jd@commandprompt.com, pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.073 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/321 X-Sequence-Number: 13562 > We had low resource utilization and poor throughput on inserts of > thousands of rows within a single database transaction. There were a > lot of configuration parameters we changed, but the one which helped the > most was wal_buffers -- we wound up setting it to 1000. This may be > higher than it needs to be, but when we got to something which ran well, > we stopped tinkering. The default value clearly caused a bottleneck. I just tried wal_buffers =3D 1000, sort_mem at 10% and effective_cache_size at 75%. The performance refuses to budge.. I guess that's as good as it'll go? From pgsql-performance-owner@postgresql.org Tue Jul 19 17:19:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 63EB252A1F for ; Tue, 19 Jul 2005 17:19:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51777-06 for ; Tue, 19 Jul 2005 20:19:36 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.196]) by svr1.postgresql.org (Postfix) with ESMTP id 26B60529E6 for ; Tue, 19 Jul 2005 17:19:35 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id x7so1220119nzc for ; Tue, 19 Jul 2005 13:19:34 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=q2Ey2siL1V8GOLhFbg1BCPXEr7q/aHbEcICp04AcHljiDDlal3mqqL4gZy2GWCDqOv1Yj6YI8A4E37BKuz/3UkID0M2djHuelCbN6bb3x1BaOVqJglqZ8hR+uJ81s9Kg0XERJqMLZkoh0wj2SBQssYEtFCQcZv6mOJvWcv2cLys= Received: by 10.36.141.3 with SMTP id o3mr5071734nzd; Tue, 19 Jul 2005 13:19:04 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Tue, 19 Jul 2005 13:19:02 -0700 (PDT) Message-ID: <758d5e7f0507191319636afaa3@mail.gmail.com> Date: Tue, 19 Jul 2005 22:19:03 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Oliver Crosby Subject: Re: Looking for tips Cc: Kevin Grittner , jd@commandprompt.com, pgsql-performance@postgresql.org In-Reply-To: <1efd553a0507191308219a098b@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a0507191308219a098b@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.608 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/322 X-Sequence-Number: 13563 On 7/19/05, Oliver Crosby wrote: > > We had low resource utilization and poor throughput on inserts of > > thousands of rows within a single database transaction. There were a > > lot of configuration parameters we changed, but the one which helped th= e > > most was wal_buffers -- we wound up setting it to 1000. This may be > > higher than it needs to be, but when we got to something which ran well= , > > we stopped tinkering. The default value clearly caused a bottleneck. >=20 > I just tried wal_buffers =3D 1000, sort_mem at 10% and > effective_cache_size at 75%. > The performance refuses to budge.. I guess that's as good as it'll go? If it is possible try: 1) wrapping many inserts into one transaction (BEGIN;INSERT;INSERT;...INSERT;COMMIT;). As PostgreSQL will need to handle less transactions per second (each your insert is a transaction), it may work faster. 2) If you can do 1, you could go further and use a COPY command which is the fastest way to bulk-load a database. Sometimes I insert data info temporary table, and then do: INSERT INTO sometable SELECT * FROM tmp_table; (but I do it when I want to do some select, updates, etc on the data before "commiting" them to main table; dropping temporary table is much cheaper than vacuuming many-a-row table). Regards, Dawid PS: Where can I find benchmarks comparing PHP vs Perl vs Python in terms of speed of executing prepared statements? From pgsql-performance-owner@postgresql.org Tue Jul 19 17:29:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 12BBB52A9C for ; Tue, 19 Jul 2005 17:29:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 56500-03 for ; Tue, 19 Jul 2005 20:29:01 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.198]) by svr1.postgresql.org (Postfix) with ESMTP id B54D852A88 for ; Tue, 19 Jul 2005 17:29:00 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i6so1314855wra for ; Tue, 19 Jul 2005 13:29:02 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=Xjq5mhzWZC3giOSDqH65tB7g4ue5ro9+tdul5/z81YAedYs2Y196iVqVGGKY5b4XgV56YBSN1ew1SmTtv69KijBsF/FgATPkrGOmmkTcDPdYYGLo5v7OXrUEHv053cGvvHbToFt+0vkCUXltcGdYOZg3HKQL6ZB//RC5znbxlrg= Received: by 10.54.61.14 with SMTP id j14mr881742wra; Tue, 19 Jul 2005 13:28:26 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 13:28:26 -0700 (PDT) Message-ID: <1efd553a050719132836c31b78@mail.gmail.com> Date: Tue, 19 Jul 2005 16:28:26 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: Dawid Kuroczko Subject: Re: Looking for tips Cc: Kevin Grittner , jd@commandprompt.com, pgsql-performance@postgresql.org In-Reply-To: <758d5e7f0507191319636afaa3@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.066 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/324 X-Sequence-Number: 13565 > If it is possible try: > 1) wrapping many inserts into one transaction > (BEGIN;INSERT;INSERT;...INSERT;COMMIT;). As PostgreSQL will need to > handle less transactions per second (each your insert is a transaction), = it > may work faster. Aye, that's what I have it doing right now. The transactions do save a HUGE chunk of time. (Cuts it down by about 40%). > 2) If you can do 1, you could go further and use a COPY command which is > the fastest way to bulk-load a database. I don't think I can use COPY in my case because I need to do processing on a per-line basis, and I need to check if the item I want to insert is already there, and if it is, I need to get it's ID so I can use that for further processing. From pgsql-performance-owner@postgresql.org Tue Jul 19 17:28:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B1ADF529C8 for ; Tue, 19 Jul 2005 17:28:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54420-06 for ; Tue, 19 Jul 2005 20:28:46 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 107EB5282F for ; Tue, 19 Jul 2005 17:28:45 -0300 (ADT) Received: (qmail 24565 invoked from network); 19 Jul 2005 22:29:21 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 19 Jul 2005 22:29:21 +0200 To: "Dawid Kuroczko" , "Oliver Crosby" Cc: "Kevin Grittner" , jd@commandprompt.com, pgsql-performance@postgresql.org Subject: Re: Looking for tips References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> Message-ID: Date: Tue, 19 Jul 2005 22:28:44 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <758d5e7f0507191319636afaa3@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/323 X-Sequence-Number: 13564 > PS: Where can I find benchmarks comparing PHP vs Perl vs Python in > terms of speed of executing prepared statements? I'm afraid you'll have to do these yourself ! And, I don't think the Python drivers support real prepared statements (the speed of psycopy is really good though). I don't think PHP either ; they don't even provide a database interface to speak of (ie you have to build the query string by hand including quoting). From pgsql-performance-owner@postgresql.org Tue Jul 19 17:30:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CFDB25282F for ; Tue, 19 Jul 2005 17:30:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54368-09 for ; Tue, 19 Jul 2005 20:30:10 +0000 (GMT) Received: from tektite.k12usa.com (tektite.k12hq.com [12.160.186.24]) by svr1.postgresql.org (Postfix) with SMTP id BAB7852A90 for ; Tue, 19 Jul 2005 17:30:08 -0300 (ADT) Received: (qmail 66476 invoked by uid 1001); 19 Jul 2005 20:30:04 -0000 Date: Tue, 19 Jul 2005 16:30:04 -0400 From: Christopher Weimann To: Oliver Crosby Cc: Richard Huxton , pgsql-performance@postgresql.org Subject: Re: Looking for tips Message-ID: <20050719203004.GA4965@tektite.k12usa.internal> References: <1efd553a05071910427b290203@mail.gmail.com> <42DD420C.2010306@archonet.com> <1efd553a05071911413e50ea90@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1efd553a05071911413e50ea90@mail.gmail.com> User-Agent: Mutt/1.4.2.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/325 X-Sequence-Number: 13566 On 07/19/2005-02:41PM, Oliver Crosby wrote: > > No queries in particular appear to be a problem. That could mean they are ALL a problem. Let see some EXPLAIN ANAYZE results just to rule it out. > At the moment it's just one user, With 1 user PostgreSQL will probobaly never beat MySQL but with hundreds it will. From pgsql-performance-owner@postgresql.org Tue Jul 19 17:45:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 42C3852A88 for ; Tue, 19 Jul 2005 17:45:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 61554-02 for ; Tue, 19 Jul 2005 20:45:04 +0000 (GMT) Received: from smtp-gw-cl-c.dmv.com (smtp-gw-cl-c.dmv.com [216.240.97.41]) by svr1.postgresql.org (Postfix) with ESMTP id 3770352A75 for ; Tue, 19 Jul 2005 17:45:02 -0300 (ADT) Received: from lanshark.dmv.com (lanshark.dmv.com [216.240.97.46]) by smtp-gw-cl-c.dmv.com (8.12.10/8.12.10) with ESMTP id j6JKivYY034048; Tue, 19 Jul 2005 16:44:57 -0400 (EDT) (envelope-from sven@dmv.com) Subject: Re: Looking for tips From: Sven Willenberger To: Oliver Crosby Cc: Dawid Kuroczko , Kevin Grittner , jd@commandprompt.com, pgsql-performance@postgresql.org In-Reply-To: <1efd553a050719132836c31b78@mail.gmail.com> References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> <1efd553a050719132836c31b78@mail.gmail.com> Content-Type: text/plain Date: Tue, 19 Jul 2005 16:46:01 -0400 Message-Id: <1121805961.3674.25.camel@lanshark.dmv.com> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Scanned-By: MIMEDefang 2.39 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.327 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_WHOIS X-Spam-Level: X-Archive-Number: 200507/326 X-Sequence-Number: 13567 On Tue, 2005-07-19 at 16:28 -0400, Oliver Crosby wrote: > > If it is possible try: > > 1) wrapping many inserts into one transaction > > (BEGIN;INSERT;INSERT;...INSERT;COMMIT;). As PostgreSQL will need to > > handle less transactions per second (each your insert is a transaction), it > > may work faster. > > Aye, that's what I have it doing right now. The transactions do save a > HUGE chunk of time. (Cuts it down by about 40%). > > > 2) If you can do 1, you could go further and use a COPY command which is > > the fastest way to bulk-load a database. > > I don't think I can use COPY in my case because I need to do > processing on a per-line basis, and I need to check if the item I want > to insert is already there, and if it is, I need to get it's ID so I > can use that for further processing. > since triggers work with COPY, you could probably write a trigger that looks for this condition and does the ID processsing you need; you could thereby enjoy the enormous speed gain resulting from COPY and maintain your data continuity. Sven From pgsql-performance-owner@postgresql.org Tue Jul 19 18:04:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A160352A1F for ; Tue, 19 Jul 2005 18:04:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 64453-06 for ; Tue, 19 Jul 2005 21:04:43 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.200]) by svr1.postgresql.org (Postfix) with ESMTP id 3CC0D52A1B for ; Tue, 19 Jul 2005 18:04:42 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 37so1319439wra for ; Tue, 19 Jul 2005 14:04:44 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=MymnCGJhN6Bv75ENRG9Q+Mx6d8NLzAV0LS69KhL+S8rd3mCByjdOwVnTkH5sdkdydIwZpF4kvPdxFofU2bVJEIQVXbTvppSORwBdeE7vDOsK3SxnO4niQzbnwA+hvSzMZkjm1KVujB/3HzpA3XgzKM/Y77gNv6IA1z5pvWtUNts= Received: by 10.54.137.17 with SMTP id k17mr892461wrd; Tue, 19 Jul 2005 14:04:04 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 14:04:04 -0700 (PDT) Message-ID: <1efd553a050719140462d41468@mail.gmail.com> Date: Tue, 19 Jul 2005 17:04:04 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: Sven Willenberger Subject: Re: Looking for tips Cc: Dawid Kuroczko , Kevin Grittner , jd@commandprompt.com, pgsql-performance@postgresql.org In-Reply-To: <1121805961.3674.25.camel@lanshark.dmv.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> <1efd553a050719132836c31b78@mail.gmail.com> <1121805961.3674.25.camel@lanshark.dmv.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.061 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/327 X-Sequence-Number: 13568 > since triggers work with COPY, you could probably write a trigger that > looks for this condition and does the ID processsing you need; you could > thereby enjoy the enormous speed gain resulting from COPY and maintain > your data continuity. So... (bear with me here.. trying to make sense of this).. With triggers there's a way I can do the parsing I need to on a log file and react to completed events in non-sequential order (you can ignore that part.. it's just how we piece together different related events) and then have perl/DBD::Pg invoke a copy command (which, from what I can tell, has to operate on a file...) and the copy command can feed the ID I need back to perl so I can work with it... If that doesn't hurt my brain, then I'm at least kinda confused... Anyway. Heading home now. I'll think about this more tonight/tomorrow. From pgsql-patches-owner@postgresql.org Tue Jul 19 18:06:18 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4DF795280C for ; Tue, 19 Jul 2005 18:06:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 59079-09 for ; Tue, 19 Jul 2005 21:06:05 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id 2E9A152A1F for ; Tue, 19 Jul 2005 18:06:03 -0300 (ADT) Received: from 172.16.1.112 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Tue, 19 Jul 2005 17:05:55 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP02.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Tue, 19 Jul 2005 17:05:55 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Tue, 19 Jul 2005 17:05:54 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Tue, 19 Jul 2005 14:05:56 -0700 Subject: Re: COPY FROM performance improvements From: "Alon Goldshuv" To: "Mark Wong" Cc: pgsql-patches@postgresql.org Message-ID: In-Reply-To: <200507191954.j6JJsVjA003280@smtp.osdl.org> MIME-Version: 1.0 X-OriginalArrivalTime: 19 Jul 2005 21:05:55.0052 (UTC) FILETIME=[A713EAC0:01C58CA5] X-WSS-ID: 6EC3B4B91J47477295-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.598 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/384 X-Sequence-Number: 16724 Hi Mark, I improved the data *parsing* capabilities of COPY, and didn't touch the data conversion or data insertion parts of the code. The parsing improvement will vary largely depending on the ratio of parsing -to- converting and inserting. Therefore, the speed increase really depends on the nature of your data: 100GB file with long data rows (lots of parsing) Small number of columns (small number of attr conversions per row) less rows (less tuple insertions) Will show the best performance improvements. However, same file size 100GB with Short data rows (minimal parsing) large number of columns (large number of attr conversions per row) AND/OR more rows (more tuple insertions) Will show improvements but not as significant. In general I'll estimate 40%-95% improvement in load speed for the 1st case and 10%-40% for the 2nd. But that also depends on the hardware, disk speed etc... This is for TEXT format. As for CSV, it may be faster but not as much as I specified here. BINARY will stay the same as before. HTH Alon. On 7/19/05 12:54 PM, "Mark Wong" wrote: > On Thu, 14 Jul 2005 17:22:18 -0700 > "Alon Goldshuv" wrote: > >> I revisited my patch and removed the code duplications that were there, and >> added support for CSV with buffered input, so CSV now runs faster too >> (although it is not as optimized as the TEXT format parsing). So now >> TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. > > Hi Alon, > > I'm curious, what kind of system are you testing this on? I'm trying to > load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm > interested in the results you would expect. > > Mark > From pgsql-performance-owner@postgresql.org Tue Jul 19 18:06:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8AB6552A4B for ; Tue, 19 Jul 2005 18:06:04 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 63749-08 for ; Tue, 19 Jul 2005 21:06:01 +0000 (GMT) Received: from calvin.surfutopia.net (calvin.surfutopia.net [67.120.245.34]) by svr1.postgresql.org (Postfix) with ESMTP id CA55C52A2F for ; Tue, 19 Jul 2005 18:05:59 -0300 (ADT) Received: by calvin.surfutopia.net (Postfix, from userid 1000) id 5839FF23A; Tue, 19 Jul 2005 14:05:57 -0700 (PDT) Date: Tue, 19 Jul 2005 14:05:57 -0700 From: John Mendenhall To: pgsql-performance list Subject: performance decrease after reboot Message-ID: <20050719210557.GA30794@calvin.surfutopia.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/328 X-Sequence-Number: 13569 I tuned a query last week to obtain acceptable performance. Here is my recorded explain analyze results: ----- LOG: duration: 826.505 ms statement: explain analyze SELECT c.id AS contact_id, sr.id AS sales_rep_id, LTRIM(RTRIM(sr.firstname || ' ' || sr.lastname)) AS sales_rep_name, p.id AS partner_id, p.company AS partner_company, coalesce(LTRIM(RTRIM(c.company)), LTRIM(RTRIM(c.firstname || ' ' || c.lastname))) AS contact_company, LTRIM(RTRIM(c.city || ' ' || c.state || ' ' || c.postalcode || ' ' || c.country)) AS contact_location, c.phone AS contact_phone, c.email AS contact_email, co.name AS contact_country, TO_CHAR(c.request_status_last_modified, 'mm/dd/yy hh12:mi pm') AS request_status_last_modified, TO_CHAR(c.request_status_last_modified, 'yyyymmddhh24miss') AS rqst_stat_last_mdfd_sortable, c.token_id FROM sales_reps sr JOIN partners p ON (sr.id = p.sales_rep_id) JOIN contacts c ON (p.id = c.partner_id) JOIN countries co ON (LOWER(c.country) = LOWER(co.code)) JOIN partner_classification pc ON (p.classification_id = pc.id AND pc.classification != 'Sales Rep') WHERE c.lead_deleted IS NULL AND EXISTS ( SELECT lr.id FROM lead_requests lr, lead_request_status lrs WHERE c.id = lr.contact_id AND lr.status_id = lrs.id AND lrs.is_closed = 0 ) ORDER BY contact_company, contact_id QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------- Sort (cost=18266.77..18266.80 rows=11 width=219) (actual time=795.502..795.763 rows=246 loops=1) Sort Key: COALESCE(ltrim(rtrim((c.company)::text)), ltrim(rtrim((((c.firstname)::text || ' '::text) || (c.lastname)::text)))), c.id -> Hash Join (cost=18258.48..18266.58 rows=11 width=219) (actual time=747.551..788.095 rows=246 loops=1) Hash Cond: (lower(("outer".code)::text) = lower(("inner".country)::text)) -> Seq Scan on countries co (cost=0.00..4.42 rows=242 width=19) (actual time=0.040..2.128 rows=242 loops=1) -> Hash (cost=18258.45..18258.45 rows=9 width=206) (actual time=746.653..746.653 rows=0 loops=1) -> Merge Join (cost=18258.12..18258.45 rows=9 width=206) (actual time=729.412..743.691 rows=246 loops=1) Merge Cond: ("outer".sales_rep_id = "inner".id) -> Sort (cost=18255.70..18255.73 rows=9 width=185) (actual time=727.948..728.274 rows=249 loops=1) Sort Key: p.sales_rep_id -> Merge Join (cost=18255.39..18255.56 rows=9 width=185) (actual time=712.747..723.095 rows=249 loops=1) Merge Cond: ("outer".id = "inner".classification_id) -> Sort (cost=1.05..1.05 rows=2 width=10) (actual time=0.192..0.195 rows=2 loops=1) Sort Key: pc.id -> Seq Scan on partner_classification pc (cost=0.00..1.04 rows=2 width=10) (actual time=0.100..0.142 rows=2 loops=1) Filter: ((classification)::text <> 'Sales Rep'::text) -> Sort (cost=18254.35..18254.38 rows=13 width=195) (actual time=712.401..712.675 rows=250 loops=1) Sort Key: p.classification_id -> Merge Join (cost=0.00..18254.11 rows=13 width=195) (actual time=47.844..705.517 rows=448 loops=1) Merge Cond: ("outer".id = "inner".partner_id) -> Index Scan using partners_pkey on partners p (cost=0.00..30.80 rows=395 width=53) (actual time=0.066..5.746 rows=395 loops=1) -> Index Scan using contacts_partner_id_idx on contacts c (cost=0.00..130358.50 rows=93 width=152) (actual time=0.351..662.576 rows=452 loops=1) Filter: ((lead_deleted IS NULL) AND (subplan)) SubPlan -> Nested Loop (cost=0.00..6.76 rows=2 width=10) (actual time=0.094..0.094 rows=0 loops=5573) Join Filter: ("outer".status_id = "inner".id) -> Index Scan using lead_requests_contact_id_idx on lead_requests lr (cost=0.00..4.23 rows=2 width=20) (actual time=0.068..0.069 rows=0 loops=5573) Index Cond: ($0 = contact_id) -> Seq Scan on lead_request_status lrs (cost=0.00..1.16 rows=8 width=10) (actual time=0.030..0.094 rows=4 loops=519) Filter: (is_closed = 0::numeric) -> Sort (cost=2.42..2.52 rows=39 width=31) (actual time=1.334..1.665 rows=267 loops=1) Sort Key: sr.id -> Seq Scan on sales_reps sr (cost=0.00..1.39 rows=39 width=31) (actual time=0.064..0.533 rows=39 loops=1) Total runtime: 798.494 ms (34 rows) ----- I rebooted the database machine later that night. Now, when I run the same query, I get the following results: ----- QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------- Sort (cost=17415.32..17415.35 rows=11 width=219) (actual time=6880.583..6880.738 rows=246 loops=1) Sort Key: COALESCE(ltrim(rtrim((c.company)::text)), ltrim(rtrim((((c.firstname)::text || ' '::text) || (c.lastname)::text)))), c.id -> Merge Join (cost=17414.22..17415.13 rows=11 width=219) (actual time=6828.441..6871.894 rows=246 loops=1) Merge Cond: ("outer".sales_rep_id = "inner".id) -> Sort (cost=17411.80..17411.83 rows=11 width=198) (actual time=6825.227..6825.652 rows=249 loops=1) Sort Key: p.sales_rep_id -> Merge Join (cost=17411.42..17411.61 rows=11 width=198) (actual time=6805.894..6818.717 rows=249 loops=1) Merge Cond: ("outer".id = "inner".classification_id) -> Sort (cost=1.05..1.05 rows=2 width=10) (actual time=0.788..0.792 rows=2 loops=1) Sort Key: pc.id -> Seq Scan on partner_classification pc (cost=0.00..1.04 rows=2 width=10) (actual time=0.094..0.554 rows=2 loops=1) Filter: ((classification)::text <> 'Sales Rep'::text) -> Sort (cost=17410.38..17410.41 rows=15 width=208) (actual time=6804.649..6804.923 rows=250 loops=1) Sort Key: p.classification_id -> Merge Join (cost=4.42..17410.08 rows=15 width=208) (actual time=62.598..6795.704 rows=448 loops=1) Merge Cond: ("outer".partner_id = "inner".id) -> Nested Loop (cost=4.42..130886.19 rows=113 width=165) (actual time=8.807..6712.529 rows=739 loops=1) Join Filter: (lower(("outer".country)::text) = lower(("inner".code)::text)) -> Index Scan using contacts_partner_id_idx on contacts c (cost=0.00..130206.59 rows=93 width=152) (actual time=0.793..4082.343 rows=739 loops=1) Filter: ((lead_deleted IS NULL) AND (subplan)) SubPlan -> Nested Loop (cost=0.00..6.76 rows=2 width=10) (actual time=0.084..0.084 rows=0 loops=37077) Join Filter: ("outer".status_id = "inner".id) -> Index Scan using lead_requests_contact_id_idx on lead_requests lr (cost=0.00..4.23 rows=2 width=20) (actual time=0.066..0.066 rows=0 loops=37077) Index Cond: ($0 = contact_id) -> Seq Scan on lead_request_status lrs (cost=0.00..1.16 rows=8 width=10) (actual time=0.031..0.140 rows=4 loops=1195) Filter: (is_closed = 0::numeric) -> Materialize (cost=4.42..6.84 rows=242 width=19) (actual time=0.003..0.347 rows=242 loops=739) -> Seq Scan on countries co (cost=0.00..4.42 rows=242 width=19) (actual time=0.038..3.162 rows=242 loops=1) -> Index Scan using partners_pkey on partners p (cost=0.00..30.80 rows=395 width=53) (actual time=0.062..15.152 rows=787 loops=1) -> Sort (cost=2.42..2.52 rows=39 width=31) (actual time=1.916..2.723 rows=267 loops=1) Sort Key: sr.id -> Seq Scan on sales_reps sr (cost=0.00..1.39 rows=39 width=31) (actual time=0.065..0.723 rows=39 loops=1) Total runtime: 6886.307 ms (34 rows) ----- There is definitely a difference in the query plans. I am guessing this difference in the performance decrease. However, nothing was changed in the postgresql.conf file. I may have run something in the psql explain analyze session a week ago, but I can't figure out what I changed. So, the bottom line is this: What do I need to do to get back to the better performance? Is it possible to determine what options may have changed from the above query plan differences? And, also, What is the "Materialize" query plan item in the second query plan, the slower plan? If you need any additional configurations, please let me know. Thank you very much in advance for any pointers you can provide. JohnM -- John Mendenhall john@surfutopia.net surf utopia internet services From pgsql-performance-owner@postgresql.org Tue Jul 19 18:14:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BFEAB529AE for ; Tue, 19 Jul 2005 18:14:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 64009-06 for ; Tue, 19 Jul 2005 21:14:05 +0000 (GMT) Received: from smtp-gw-cl-c.dmv.com (smtp-gw-cl-c.dmv.com [216.240.97.41]) by svr1.postgresql.org (Postfix) with ESMTP id 943215280C for ; Tue, 19 Jul 2005 18:14:03 -0300 (ADT) Received: from lanshark.dmv.com (lanshark.dmv.com [216.240.97.46]) by smtp-gw-cl-c.dmv.com (8.12.10/8.12.10) with ESMTP id j6JLE4YY035074; Tue, 19 Jul 2005 17:14:04 -0400 (EDT) (envelope-from sven@dmv.com) Subject: Re: Looking for tips From: Sven Willenberger To: Oliver Crosby Cc: pgsql-performance@postgresql.org In-Reply-To: <1efd553a050719140462d41468@mail.gmail.com> References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> <1efd553a050719132836c31b78@mail.gmail.com> <1121805961.3674.25.camel@lanshark.dmv.com> <1efd553a050719140462d41468@mail.gmail.com> Content-Type: text/plain Date: Tue, 19 Jul 2005 17:15:08 -0400 Message-Id: <1121807708.3673.47.camel@lanshark.dmv.com> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Scanned-By: MIMEDefang 2.39 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.328 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_WHOIS X-Spam-Level: X-Archive-Number: 200507/329 X-Sequence-Number: 13570 On Tue, 2005-07-19 at 17:04 -0400, Oliver Crosby wrote: > > since triggers work with COPY, you could probably write a trigger that > > looks for this condition and does the ID processsing you need; you could > > thereby enjoy the enormous speed gain resulting from COPY and maintain > > your data continuity. > > So... (bear with me here.. trying to make sense of this).. > With triggers there's a way I can do the parsing I need to on a log > file and react to completed events in non-sequential order (you can > ignore that part.. it's just how we piece together different related > events) and then have perl/DBD::Pg invoke a copy command (which, from > what I can tell, has to operate on a file...) and the copy command can > feed the ID I need back to perl so I can work with it... > If that doesn't hurt my brain, then I'm at least kinda confused... > Anyway. Heading home now. I'll think about this more tonight/tomorrow. > Well without knowing the specifics of what you are actually trying to accomplish I cannot say yes or no to your question. I am not sure from where this data is coming that you are inserting into the db. However, if the scenario is this: a) attempt to insert a row b) if row exists already, grab the ID and do other db selects/inserts/deletes based on that ID, then there is no need to feed this information back to the perlscript. Is your perlscript parsing a file and then using the parsed information to insert rows? If so, how is the ID that is returned used? Can you have the trigger use the ID that may be returned to perform whatever it is that your perlscript is trying to accomplish with that ID? It's all kind of vague so my answers may or may not help, but based on the [lack of] specifics you have provided, I fear that is the best suggestion that I can offer at this point. Sven From pgsql-patches-owner@postgresql.org Tue Jul 19 18:37:43 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3E489529A0 for ; Tue, 19 Jul 2005 18:37:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 69413-09 for ; Tue, 19 Jul 2005 21:37:36 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id 4FDAC52989 for ; Tue, 19 Jul 2005 18:37:35 -0300 (ADT) Received: from ibm-b.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6JLbZjA011011 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Tue, 19 Jul 2005 14:37:36 -0700 Message-Id: <200507192137.j6JLbZjA011011@smtp.osdl.org> Date: Tue, 19 Jul 2005 14:37:58 -0700 From: Mark Wong To: "Alon Goldshuv" Cc: pgsql-patches@postgresql.org Subject: Re: COPY FROM performance improvements In-Reply-To: References: <200507191954.j6JJsVjA003280@smtp.osdl.org> Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.057 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/386 X-Sequence-Number: 16726 Hi Alon, Yeah, that helps. I just need to break up my scripts a little to just load the data and not build indexes. Is the following information good enough to give a guess about the data I'm loading, if you don't mind? ;) Here's a link to my script to create tables: http://developer.osdl.org/markw/mt/getfile.py?id=eaf16b7831588729780645b2bb44f7f23437e432&path=scripts/pgsql/create_tables.sh.in File sizes: -rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 customer.tbl -rw-r--r-- 1 markw 50 74G Jul 8 15:03 lineitem.tbl -rw-r--r-- 1 markw 50 2.1K Jul 8 15:03 nation.tbl -rw-r--r-- 1 markw 50 17G Jul 8 15:03 orders.tbl -rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 part.tbl -rw-r--r-- 1 markw 50 12G Jul 8 15:03 partsupp.tbl -rw-r--r-- 1 markw 50 391 Jul 8 15:03 region.tbl -rw-r--r-- 1 markw 50 136M Jul 8 15:03 supplier.tbl Number of rows: # wc -l *.tbl 15000000 customer.tbl 600037902 lineitem.tbl 25 nation.tbl 150000000 orders.tbl 20000000 part.tbl 80000000 partsupp.tbl 5 region.tbl 1000000 supplier.tbl Thanks, Mark On Tue, 19 Jul 2005 14:05:56 -0700 "Alon Goldshuv" wrote: > Hi Mark, > > I improved the data *parsing* capabilities of COPY, and didn't touch the > data conversion or data insertion parts of the code. The parsing improvement > will vary largely depending on the ratio of parsing -to- converting and > inserting. > > Therefore, the speed increase really depends on the nature of your data: > > 100GB file with > long data rows (lots of parsing) > Small number of columns (small number of attr conversions per row) > less rows (less tuple insertions) > > Will show the best performance improvements. > > However, same file size 100GB with > Short data rows (minimal parsing) > large number of columns (large number of attr conversions per row) > AND/OR > more rows (more tuple insertions) > > Will show improvements but not as significant. > In general I'll estimate 40%-95% improvement in load speed for the 1st case > and 10%-40% for the 2nd. But that also depends on the hardware, disk speed > etc... This is for TEXT format. As for CSV, it may be faster but not as much > as I specified here. BINARY will stay the same as before. > > HTH > Alon. > > > > > > > On 7/19/05 12:54 PM, "Mark Wong" wrote: > > > On Thu, 14 Jul 2005 17:22:18 -0700 > > "Alon Goldshuv" wrote: > > > >> I revisited my patch and removed the code duplications that were there, and > >> added support for CSV with buffered input, so CSV now runs faster too > >> (although it is not as optimized as the TEXT format parsing). So now > >> TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. > > > > Hi Alon, > > > > I'm curious, what kind of system are you testing this on? I'm trying to > > load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm > > interested in the results you would expect. > > > > Mark > > > From pgsql-performance-owner@postgresql.org Tue Jul 19 18:41:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 59ADA5298F for ; Tue, 19 Jul 2005 18:41:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 73838-02 for ; Tue, 19 Jul 2005 21:41:23 +0000 (GMT) Received: from netbox.unitech.com.ar (unknown [200.32.92.34]) by svr1.postgresql.org (Postfix) with ESMTP id 3418752989 for ; Tue, 19 Jul 2005 18:41:18 -0300 (ADT) Received: from dariop (dariop.unitech.com.ar [192.168.1.237]) by netbox.unitech.com.ar (8.11.6/8.11.6) with ESMTP id j6JMfZE00423 for ; Tue, 19 Jul 2005 18:41:36 -0400 Received: from 127.0.0.1 (AVG SMTP 7.0.323 [267.9.1]); Tue, 19 Jul 2005 18:41:18 -0300 From: "Dario" To: Subject: Re: join and query planner Date: Tue, 19 Jul 2005 18:41:18 -0300 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4952.2800 Importance: Normal In-Reply-To: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.092 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/330 X-Sequence-Number: 13571 I'll try that. Let you know as soon as I can take a look. Thank you- -----Mensaje original----- De: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org]En nombre de Kevin Grittner Enviado el: lunes, 18 de julio de 2005 17:48 Para: pgsql-performance@postgresql.org; dario_d_s@unitech.com.ar Asunto: Re: [PERFORM] join and query planner You might want to set join_collapse_limit high, and use the JOIN operators rather than the comma-separated lists. We generate the WHERE clause on the fly, based on user input, and this has worked well for us. -Kevin >>> "Dario" 07/18/05 2:24 PM >>> Hi. > Just out of curiosity, does it do any better with the following? > > SELECT ... Yes, it does. But my query could also be SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) /*new*/ , e WHERE (b.column <= 100) /*new*/ and (e.key = a.key) and (e.field = 'filter') because it's constructed by an application. I needed to know if, somehow, someway, I can "unforce" join order. The only way to solve it so far is changing application. It must build something like SELECT ... FROM b JOIN (a JOIN e ON (e.key = a.key)) ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) and (e.field = 'filter') Supossed that e.field has (should have) better selectivity. But now this problem belongs to programmer's group :-) The query, in fact, has more tables to join. I wonder if lowering geqo threshold could do the work... Thank you. Greetings. Long life, little spam and prosperity! -----Mensaje original----- De: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org]En nombre de Kevin Grittner Enviado el: lunes, 18 de julio de 2005 14:58 Para: pgsql-performance@postgresql.org; dariop@unitech.com.ar Asunto: Re: [PERFORM] join and query planner Just out of curiosity, does it do any better with the following? SELECT ... FROM a JOIN b ON (a.key = b.key) LEFT JOIN c ON (c.key = a.key) LEFT JOIN d ON (d.key=a.key) WHERE (b.column <= 100) >>> snipp ---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match From pgsql-patches-owner@postgresql.org Tue Jul 19 19:06:47 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1E9F252989 for ; Tue, 19 Jul 2005 19:06:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 75085-10 for ; Tue, 19 Jul 2005 22:06:35 +0000 (GMT) Received: from mail.Mi8.com (nycgw05.mi8.com [63.240.6.50]) by svr1.postgresql.org (Postfix) with ESMTP id B6527529B6 for ; Tue, 19 Jul 2005 19:06:24 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Tue, 19 Jul 2005 18:06:16 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Tue, 19 Jul 2005 18:06:16 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Tue, 19 Jul 2005 18:06:14 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Tue, 19 Jul 2005 15:06:17 -0700 Subject: Re: COPY FROM performance improvements From: "Alon Goldshuv" To: "Mark Wong" Cc: pgsql-patches@postgresql.org Message-ID: In-Reply-To: <200507192137.j6JLbZjA011011@smtp.osdl.org> MIME-Version: 1.0 X-OriginalArrivalTime: 19 Jul 2005 22:06:16.0572 (UTC) FILETIME=[15AC03C0:01C58CAE] X-WSS-ID: 6EC3A6D21M87483625-02-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.595 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/388 X-Sequence-Number: 16728 Mark, Thanks for the info. Yes, isolating indexes out of the picture is a good idea for this purpose. I can't really give a guess to how fast the load rate should be. I don't know how your system is configured, and all the hardware characteristics (and even if I knew that info I may not be able to guess...). I am pretty confident that the load will be faster than before, I'll risk that ;-) Looking into your TPC-H size and metadata I'll estimate that partsupp,customer and orders will have the most significant increase in load rate. You could start with those. I guess the only way to really know is to try... Load several times with the existing PG-COPY and then load several times with the patched COPY and compare. I'll be curious to hear your results. Thx, Alon. On 7/19/05 2:37 PM, "Mark Wong" wrote: > Hi Alon, > > Yeah, that helps. I just need to break up my scripts a little to just > load the data and not build indexes. > > Is the following information good enough to give a guess about the data > I'm loading, if you don't mind? ;) Here's a link to my script to create > tables: > http://developer.osdl.org/markw/mt/getfile.py?id=eaf16b7831588729780645b2bb44f > 7f23437e432&path=scripts/pgsql/create_tables.sh.in > > File sizes: > -rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 customer.tbl > -rw-r--r-- 1 markw 50 74G Jul 8 15:03 lineitem.tbl > -rw-r--r-- 1 markw 50 2.1K Jul 8 15:03 nation.tbl > -rw-r--r-- 1 markw 50 17G Jul 8 15:03 orders.tbl > -rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 part.tbl > -rw-r--r-- 1 markw 50 12G Jul 8 15:03 partsupp.tbl > -rw-r--r-- 1 markw 50 391 Jul 8 15:03 region.tbl > -rw-r--r-- 1 markw 50 136M Jul 8 15:03 supplier.tbl > > Number of rows: > # wc -l *.tbl > 15000000 customer.tbl > 600037902 lineitem.tbl > 25 nation.tbl > 150000000 orders.tbl > 20000000 part.tbl > 80000000 partsupp.tbl > 5 region.tbl > 1000000 supplier.tbl > > Thanks, > Mark > > On Tue, 19 Jul 2005 14:05:56 -0700 > "Alon Goldshuv" wrote: > >> Hi Mark, >> >> I improved the data *parsing* capabilities of COPY, and didn't touch the >> data conversion or data insertion parts of the code. The parsing improvement >> will vary largely depending on the ratio of parsing -to- converting and >> inserting. >> >> Therefore, the speed increase really depends on the nature of your data: >> >> 100GB file with >> long data rows (lots of parsing) >> Small number of columns (small number of attr conversions per row) >> less rows (less tuple insertions) >> >> Will show the best performance improvements. >> >> However, same file size 100GB with >> Short data rows (minimal parsing) >> large number of columns (large number of attr conversions per row) >> AND/OR >> more rows (more tuple insertions) >> >> Will show improvements but not as significant. >> In general I'll estimate 40%-95% improvement in load speed for the 1st case >> and 10%-40% for the 2nd. But that also depends on the hardware, disk speed >> etc... This is for TEXT format. As for CSV, it may be faster but not as much >> as I specified here. BINARY will stay the same as before. >> >> HTH >> Alon. >> >> >> >> >> >> >> On 7/19/05 12:54 PM, "Mark Wong" wrote: >> >>> On Thu, 14 Jul 2005 17:22:18 -0700 >>> "Alon Goldshuv" wrote: >>> >>>> I revisited my patch and removed the code duplications that were there, and >>>> added support for CSV with buffered input, so CSV now runs faster too >>>> (although it is not as optimized as the TEXT format parsing). So now >>>> TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original >>>> file. >>> >>> Hi Alon, >>> >>> I'm curious, what kind of system are you testing this on? I'm trying to >>> load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm >>> interested in the results you would expect. >>> >>> Mark >>> >> > From pgsql-patches-owner@postgresql.org Tue Jul 19 19:18:16 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2B53052815 for ; Tue, 19 Jul 2005 19:18:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 78923-04 for ; Tue, 19 Jul 2005 22:18:05 +0000 (GMT) Received: from trolak.mydnsbox2.com (ns1.mydnsbox2.com [207.44.142.118]) by svr1.postgresql.org (Postfix) with ESMTP id 2624552869 for ; Tue, 19 Jul 2005 19:18:02 -0300 (ADT) Received: from [192.168.1.103] (cpe-024-211-165-134.nc.res.rr.com [24.211.165.134]) (authenticated (0 bits)) by trolak.mydnsbox2.com (8.11.6/8.11.6) with ESMTP id j6JLWO512530; Tue, 19 Jul 2005 16:32:24 -0500 Message-ID: <42DD7C10.4090008@dunslane.net> Date: Tue, 19 Jul 2005 18:17:52 -0400 From: Andrew Dunstan User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Fedora/1.7.8-1.3.1 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Mark Wong Cc: Alon Goldshuv , pgsql-patches@postgresql.org Subject: Re: COPY FROM performance improvements References: <200507191954.j6JJsVjA003280@smtp.osdl.org> <200507192137.j6JLbZjA011011@smtp.osdl.org> In-Reply-To: <200507192137.j6JLbZjA011011@smtp.osdl.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.04 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/389 X-Sequence-Number: 16729 Mark, You should definitely not be doing this sort of thing, I believe: CREATE TABLE orders ( o_orderkey INTEGER, o_custkey INTEGER, o_orderstatus CHAR(1), o_totalprice REAL, o_orderDATE DATE, o_orderpriority CHAR(15), o_clerk CHAR(15), o_shippriority INTEGER, o_comment VARCHAR(79), PRIMARY KEY (o_orderkey)) Create the table with no constraints, load the data, then set up primary keys and whatever other constraints you want using ALTER TABLE. Last time I did a load like this (albeit 2 orders of magnitude smaller) I saw a 50% speedup from deferring constarint creation. cheers andrew Mark Wong wrote: >Hi Alon, > >Yeah, that helps. I just need to break up my scripts a little to just >load the data and not build indexes. > >Is the following information good enough to give a guess about the data >I'm loading, if you don't mind? ;) Here's a link to my script to create >tables: >http://developer.osdl.org/markw/mt/getfile.py?id=eaf16b7831588729780645b2bb44f7f23437e432&path=scripts/pgsql/create_tables.sh.in > >File sizes: >-rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 customer.tbl >-rw-r--r-- 1 markw 50 74G Jul 8 15:03 lineitem.tbl >-rw-r--r-- 1 markw 50 2.1K Jul 8 15:03 nation.tbl >-rw-r--r-- 1 markw 50 17G Jul 8 15:03 orders.tbl >-rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 part.tbl >-rw-r--r-- 1 markw 50 12G Jul 8 15:03 partsupp.tbl >-rw-r--r-- 1 markw 50 391 Jul 8 15:03 region.tbl >-rw-r--r-- 1 markw 50 136M Jul 8 15:03 supplier.tbl > >Number of rows: ># wc -l *.tbl > 15000000 customer.tbl > 600037902 lineitem.tbl > 25 nation.tbl > 150000000 orders.tbl > 20000000 part.tbl > 80000000 partsupp.tbl > 5 region.tbl > 1000000 supplier.tbl > >Thanks, >Mark > >On Tue, 19 Jul 2005 14:05:56 -0700 >"Alon Goldshuv" wrote: > > > >>Hi Mark, >> >>I improved the data *parsing* capabilities of COPY, and didn't touch the >>data conversion or data insertion parts of the code. The parsing improvement >>will vary largely depending on the ratio of parsing -to- converting and >>inserting. >> >>Therefore, the speed increase really depends on the nature of your data: >> >>100GB file with >>long data rows (lots of parsing) >>Small number of columns (small number of attr conversions per row) >>less rows (less tuple insertions) >> >>Will show the best performance improvements. >> >>However, same file size 100GB with >>Short data rows (minimal parsing) >>large number of columns (large number of attr conversions per row) >>AND/OR >>more rows (more tuple insertions) >> >>Will show improvements but not as significant. >>In general I'll estimate 40%-95% improvement in load speed for the 1st case >>and 10%-40% for the 2nd. But that also depends on the hardware, disk speed >>etc... This is for TEXT format. As for CSV, it may be faster but not as much >>as I specified here. BINARY will stay the same as before. >> >>HTH >>Alon. >> >> >> >> >> >> >>On 7/19/05 12:54 PM, "Mark Wong" wrote: >> >> >> >>>On Thu, 14 Jul 2005 17:22:18 -0700 >>>"Alon Goldshuv" wrote: >>> >>> >>> >>>>I revisited my patch and removed the code duplications that were there, and >>>>added support for CSV with buffered input, so CSV now runs faster too >>>>(although it is not as optimized as the TEXT format parsing). So now >>>>TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. >>>> >>>> >>>Hi Alon, >>> >>>I'm curious, what kind of system are you testing this on? I'm trying to >>>load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm >>>interested in the results you would expect. >>> >>>Mark >>> >>> >>> > >---------------------------(end of broadcast)--------------------------- >TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > > > From pgsql-patches-owner@postgresql.org Tue Jul 19 19:51:23 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 206CC529B6 for ; Tue, 19 Jul 2005 19:51:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 84429-03 for ; Tue, 19 Jul 2005 22:51:12 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id D59D4529FD for ; Tue, 19 Jul 2005 19:51:11 -0300 (ADT) Received: from ibm-b.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6JMpAjA016147 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Tue, 19 Jul 2005 15:51:10 -0700 Message-Id: <200507192251.j6JMpAjA016147@smtp.osdl.org> Date: Tue, 19 Jul 2005 15:51:33 -0700 From: Mark Wong To: Andrew Dunstan Cc: Alon Goldshuv , pgsql-patches@postgresql.org Subject: Re: COPY FROM performance improvements In-Reply-To: <42DD7C10.4090008@dunslane.net> References: <200507191954.j6JJsVjA003280@smtp.osdl.org> <200507192137.j6JLbZjA011011@smtp.osdl.org> <42DD7C10.4090008@dunslane.net> Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.057 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/390 X-Sequence-Number: 16730 Whoopsies, yeah good point about the PRIMARY KEY. I'll fix that. Mark On Tue, 19 Jul 2005 18:17:52 -0400 Andrew Dunstan wrote: > Mark, > > You should definitely not be doing this sort of thing, I believe: > > CREATE TABLE orders ( > o_orderkey INTEGER, > o_custkey INTEGER, > o_orderstatus CHAR(1), > o_totalprice REAL, > o_orderDATE DATE, > o_orderpriority CHAR(15), > o_clerk CHAR(15), > o_shippriority INTEGER, > o_comment VARCHAR(79), > PRIMARY KEY (o_orderkey)) > > Create the table with no constraints, load the data, then set up primary keys and whatever other constraints you want using ALTER TABLE. Last time I did a load like this (albeit 2 orders of magnitude smaller) I saw a 50% speedup from deferring constarint creation. > > > cheers > > andrew > > > > Mark Wong wrote: > > >Hi Alon, > > > >Yeah, that helps. I just need to break up my scripts a little to just > >load the data and not build indexes. > > > >Is the following information good enough to give a guess about the data > >I'm loading, if you don't mind? ;) Here's a link to my script to create > >tables: > >http://developer.osdl.org/markw/mt/getfile.py?id=eaf16b7831588729780645b2bb44f7f23437e432&path=scripts/pgsql/create_tables.sh.in > > > >File sizes: > >-rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 customer.tbl > >-rw-r--r-- 1 markw 50 74G Jul 8 15:03 lineitem.tbl > >-rw-r--r-- 1 markw 50 2.1K Jul 8 15:03 nation.tbl > >-rw-r--r-- 1 markw 50 17G Jul 8 15:03 orders.tbl > >-rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 part.tbl > >-rw-r--r-- 1 markw 50 12G Jul 8 15:03 partsupp.tbl > >-rw-r--r-- 1 markw 50 391 Jul 8 15:03 region.tbl > >-rw-r--r-- 1 markw 50 136M Jul 8 15:03 supplier.tbl > > > >Number of rows: > ># wc -l *.tbl > > 15000000 customer.tbl > > 600037902 lineitem.tbl > > 25 nation.tbl > > 150000000 orders.tbl > > 20000000 part.tbl > > 80000000 partsupp.tbl > > 5 region.tbl > > 1000000 supplier.tbl > > > >Thanks, > >Mark > > > >On Tue, 19 Jul 2005 14:05:56 -0700 > >"Alon Goldshuv" wrote: > > > > > > > >>Hi Mark, > >> > >>I improved the data *parsing* capabilities of COPY, and didn't touch the > >>data conversion or data insertion parts of the code. The parsing improvement > >>will vary largely depending on the ratio of parsing -to- converting and > >>inserting. > >> > >>Therefore, the speed increase really depends on the nature of your data: > >> > >>100GB file with > >>long data rows (lots of parsing) > >>Small number of columns (small number of attr conversions per row) > >>less rows (less tuple insertions) > >> > >>Will show the best performance improvements. > >> > >>However, same file size 100GB with > >>Short data rows (minimal parsing) > >>large number of columns (large number of attr conversions per row) > >>AND/OR > >>more rows (more tuple insertions) > >> > >>Will show improvements but not as significant. > >>In general I'll estimate 40%-95% improvement in load speed for the 1st case > >>and 10%-40% for the 2nd. But that also depends on the hardware, disk speed > >>etc... This is for TEXT format. As for CSV, it may be faster but not as much > >>as I specified here. BINARY will stay the same as before. > >> > >>HTH > >>Alon. > >> > >> > >> > >> > >> > >> > >>On 7/19/05 12:54 PM, "Mark Wong" wrote: > >> > >> > >> > >>>On Thu, 14 Jul 2005 17:22:18 -0700 > >>>"Alon Goldshuv" wrote: > >>> > >>> > >>> > >>>>I revisited my patch and removed the code duplications that were there, and > >>>>added support for CSV with buffered input, so CSV now runs faster too > >>>>(although it is not as optimized as the TEXT format parsing). So now > >>>>TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original file. > >>>> > >>>> > >>>Hi Alon, > >>> > >>>I'm curious, what kind of system are you testing this on? I'm trying to > >>>load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm > >>>interested in the results you would expect. > >>> > >>>Mark > >>> > >>> > >>> > > From pgsql-performance-owner@postgresql.org Tue Jul 19 20:01:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9DA1A52A0A for ; Tue, 19 Jul 2005 20:01:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 90974-01 for ; Tue, 19 Jul 2005 23:01:56 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 9FEAE52A00 for ; Tue, 19 Jul 2005 20:01:54 -0300 (ADT) Received: (qmail 31935 invoked from network); 20 Jul 2005 01:02:32 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 20 Jul 2005 01:02:32 +0200 To: "Oliver Crosby" , "Sven Willenberger" Cc: "Dawid Kuroczko" , "Kevin Grittner" , jd@commandprompt.com, pgsql-performance@postgresql.org Subject: Re: Looking for tips References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> <1efd553a050719132836c31b78@mail.gmail.com> <1121805961.3674.25.camel@lanshark.dmv.com> <1efd553a050719140462d41468@mail.gmail.com> Message-ID: Date: Wed, 20 Jul 2005 01:01:56 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <1efd553a050719140462d41468@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/331 X-Sequence-Number: 13572 You could have a program pre-parse your log and put it in a format understandable by COPY, then load it in a temporary table and write a part of your application simply as a plpgsql function, reading from this table and doing queries (or a plperl function)... > So... (bear with me here.. trying to make sense of this).. > With triggers there's a way I can do the parsing I need to on a log > file and react to completed events in non-sequential order (you can > ignore that part.. it's just how we piece together different related > events) and then have perl/DBD::Pg invoke a copy command (which, from > what I can tell, has to operate on a file...) and the copy command can > feed the ID I need back to perl so I can work with it... > If that doesn't hurt my brain, then I'm at least kinda confused... > Anyway. Heading home now. I'll think about this more tonight/tomorrow. > > ---------------------------(end of broadcast)--------------------------- > TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > From pgsql-patches-owner@postgresql.org Tue Jul 19 21:39:52 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BD8AA52997 for ; Tue, 19 Jul 2005 21:39:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 16869-05 for ; Wed, 20 Jul 2005 00:39:38 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id E9FF4529A1 for ; Tue, 19 Jul 2005 21:39:36 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Tue, 19 Jul 2005 20:39:33 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Tue, 19 Jul 2005 20:39:33 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Tue, 19 Jul 2005 20:39:33 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Tue, 19 Jul 2005 17:39:33 -0700 Subject: Re: COPY FROM performance improvements From: "Luke Lonergan" To: "Mark Wong" , "Andrew Dunstan" Cc: "Alon Goldshuv" , pgsql-patches@postgresql.org Message-ID: In-Reply-To: <200507192251.j6JMpAjA016147@smtp.osdl.org> MIME-Version: 1.0 X-OriginalArrivalTime: 20 Jul 2005 00:39:33.0522 (UTC) FILETIME=[7F7B3720:01C58CC3] X-WSS-ID: 6EC342CF1J47631110-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.574 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/395 X-Sequence-Number: 16735 Good points on all, another element in the performance expectations is the ratio of CPU speed to I/O subsystem speed, as Alon had hinted earlier. This patch substantially (500%) improves the efficiency of parsing in the COPY path, which, on a 3GHz P4 desktop with a commodity disk drive represents 8 of a total of 30 seconds of processing time. So, by reducing the parsing time from 8 seconds to 1.5 seconds, the overall COPY time is reduced from 30 seconds to 23.5 seconds, or a speedup of about 20%. On a dual 2.2GHz Opteron machine with a 6-disk SCSI RAID subsystem capable of 240MB/s sequential read and writes, the ratios change and we see between 35% and 95% increase in COPY performance, with the bottleneck being CPU. The disk is only running at about 90MB/s during this period. I'd expect that as your CPUs slow down relative to your I/O speed, and Itaniums or IT2s are quite slow, you should see an increased effect of the parsing improvements. One good way to validate the effect is to watch the I/O bandwidth using vmstat 1 (on Linux) while the load is progressing. When you watch that with the unpatched source and with the patched source, if they are the same, you should see no benefit from the patch (you are I/O limited). If you check your underlying sequential write speed, you will be bottlenecked at roughly half that in performing COPY because of the write-through the WAL. - Luke On 7/19/05 3:51 PM, "Mark Wong" wrote: > Whoopsies, yeah good point about the PRIMARY KEY. I'll fix that. > > Mark > > On Tue, 19 Jul 2005 18:17:52 -0400 > Andrew Dunstan wrote: > >> Mark, >> >> You should definitely not be doing this sort of thing, I believe: >> >> CREATE TABLE orders ( >> o_orderkey INTEGER, >> o_custkey INTEGER, >> o_orderstatus CHAR(1), >> o_totalprice REAL, >> o_orderDATE DATE, >> o_orderpriority CHAR(15), >> o_clerk CHAR(15), >> o_shippriority INTEGER, >> o_comment VARCHAR(79), >> PRIMARY KEY (o_orderkey)) >> >> Create the table with no constraints, load the data, then set up primary keys >> and whatever other constraints you want using ALTER TABLE. Last time I did a >> load like this (albeit 2 orders of magnitude smaller) I saw a 50% speedup >> from deferring constarint creation. >> >> >> cheers >> >> andrew >> >> >> >> Mark Wong wrote: >> >>> Hi Alon, >>> >>> Yeah, that helps. I just need to break up my scripts a little to just >>> load the data and not build indexes. >>> >>> Is the following information good enough to give a guess about the data >>> I'm loading, if you don't mind? ;) Here's a link to my script to create >>> tables: >>> http://developer.osdl.org/markw/mt/getfile.py?id=eaf16b7831588729780645b2bb4 >>> 4f7f23437e432&path=scripts/pgsql/create_tables.sh.in >>> >>> File sizes: >>> -rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 customer.tbl >>> -rw-r--r-- 1 markw 50 74G Jul 8 15:03 lineitem.tbl >>> -rw-r--r-- 1 markw 50 2.1K Jul 8 15:03 nation.tbl >>> -rw-r--r-- 1 markw 50 17G Jul 8 15:03 orders.tbl >>> -rw-r--r-- 1 markw 50 2.3G Jul 8 15:03 part.tbl >>> -rw-r--r-- 1 markw 50 12G Jul 8 15:03 partsupp.tbl >>> -rw-r--r-- 1 markw 50 391 Jul 8 15:03 region.tbl >>> -rw-r--r-- 1 markw 50 136M Jul 8 15:03 supplier.tbl >>> >>> Number of rows: >>> # wc -l *.tbl >>> 15000000 customer.tbl >>> 600037902 lineitem.tbl >>> 25 nation.tbl >>> 150000000 orders.tbl >>> 20000000 part.tbl >>> 80000000 partsupp.tbl >>> 5 region.tbl >>> 1000000 supplier.tbl >>> >>> Thanks, >>> Mark >>> >>> On Tue, 19 Jul 2005 14:05:56 -0700 >>> "Alon Goldshuv" wrote: >>> >>> >>> >>>> Hi Mark, >>>> >>>> I improved the data *parsing* capabilities of COPY, and didn't touch the >>>> data conversion or data insertion parts of the code. The parsing >>>> improvement >>>> will vary largely depending on the ratio of parsing -to- converting and >>>> inserting. >>>> >>>> Therefore, the speed increase really depends on the nature of your data: >>>> >>>> 100GB file with >>>> long data rows (lots of parsing) >>>> Small number of columns (small number of attr conversions per row) >>>> less rows (less tuple insertions) >>>> >>>> Will show the best performance improvements. >>>> >>>> However, same file size 100GB with >>>> Short data rows (minimal parsing) >>>> large number of columns (large number of attr conversions per row) >>>> AND/OR >>>> more rows (more tuple insertions) >>>> >>>> Will show improvements but not as significant. >>>> In general I'll estimate 40%-95% improvement in load speed for the 1st case >>>> and 10%-40% for the 2nd. But that also depends on the hardware, disk speed >>>> etc... This is for TEXT format. As for CSV, it may be faster but not as >>>> much >>>> as I specified here. BINARY will stay the same as before. >>>> >>>> HTH >>>> Alon. >>>> >>>> >>>> >>>> >>>> >>>> >>>> On 7/19/05 12:54 PM, "Mark Wong" wrote: >>>> >>>> >>>> >>>>> On Thu, 14 Jul 2005 17:22:18 -0700 >>>>> "Alon Goldshuv" wrote: >>>>> >>>>> >>>>> >>>>>> I revisited my patch and removed the code duplications that were there, >>>>>> and >>>>>> added support for CSV with buffered input, so CSV now runs faster too >>>>>> (although it is not as optimized as the TEXT format parsing). So now >>>>>> TEXT,CSV and BINARY are all parsed in CopyFrom(), like in the original >>>>>> file. >>>>>> >>>>>> >>>>> Hi Alon, >>>>> >>>>> I'm curious, what kind of system are you testing this on? I'm trying to >>>>> load 100GB of data in our dbt3 workload on a 4-way itanium2. I'm >>>>> interested in the results you would expect. >>>>> >>>>> Mark >>>>> >>>>> >>>>> >>> > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend > From pgsql-performance-owner@postgresql.org Tue Jul 19 22:51:06 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BDFBB52982 for ; Tue, 19 Jul 2005 22:51:04 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25475-09 for ; Wed, 20 Jul 2005 01:50:55 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.196]) by svr1.postgresql.org (Postfix) with ESMTP id 9342F529E2 for ; Tue, 19 Jul 2005 22:50:54 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i2so1354809wra for ; Tue, 19 Jul 2005 18:50:59 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=PKey9uduzS/3wiTTlxYYsSSe3sBUA6Xw5Gxi0QlM3oT766Slt67D/ol3q5TBpvm7R/9RZ1eWwTFEw4e0XT/OboGaBQ+y6Xiop+eyDZccXjRnLPh6RNRIhz3ze8/tpdp4lZGLL6hj4bttFpmZ9gKwrXwrH07WxfzngdELdzy3ogM= Received: by 10.54.125.16 with SMTP id x16mr966649wrc; Tue, 19 Jul 2005 18:50:18 -0700 (PDT) Received: by 10.54.150.2 with HTTP; Tue, 19 Jul 2005 18:50:18 -0700 (PDT) Message-ID: <1efd553a050719185059de2581@mail.gmail.com> Date: Tue, 19 Jul 2005 21:50:18 -0400 From: Oliver Crosby Reply-To: Oliver Crosby To: PFC Subject: Re: Looking for tips Cc: Sven Willenberger , Dawid Kuroczko , Kevin Grittner , jd@commandprompt.com, pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1efd553a0507191308219a098b@mail.gmail.com> <758d5e7f0507191319636afaa3@mail.gmail.com> <1efd553a050719132836c31b78@mail.gmail.com> <1121805961.3674.25.camel@lanshark.dmv.com> <1efd553a050719140462d41468@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.057 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/332 X-Sequence-Number: 13573 Sorry for the lack of specifics... We have a file generated as a list of events, one per line. Suppose lines 1,2,3,5,7,11,etc were related, then the last one would specify that it's the last event. Gradually this gets assembled by a perl script and when the last event is encountered, it gets inserted into the db. For a given table, let's say it's of the form (a,b,c) where 'a' is a pkey, 'b' is indexed, and 'c' is other related information. The most common 'b' values are cached locally with the perl script to save us having to query the db. So what we end up having is: if 'b' exists in cache, use cached 'a' value and continue else if 'b' exists in the db, use the associated 'a' value and continue else add a new line with 'b', return the new 'a' and continue The local cache was a huge time saver with mysql. I've tried making a plpgsql function that handles everything in one step on the db side, but it didn't show any improvement. Time permitting, I'll try some new approaches with changing the scripts and queries, though right now I was just hoping to tune postgresql.conf to work better with the hardware available. Thanks to everyone for your help. Very much appreciated. From pgsql-performance-owner@postgresql.org Wed Jul 20 01:49:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4682E528BB for ; Wed, 20 Jul 2005 01:49:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 65946-09 for ; Wed, 20 Jul 2005 04:49:13 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id EA781528AD for ; Wed, 20 Jul 2005 01:49:12 -0300 (ADT) Received: (qmail 17371 invoked by uid 1010); 20 Jul 2005 00:49:11 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.20231 secs); 20 Jul 2005 04:49:11 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 20 Jul 2005 00:49:10 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id C141912D00B; Tue, 19 Jul 2005 22:49:08 -0600 (MDT) Date: Tue, 19 Jul 2005 22:49:08 -0600 From: Robert Creager To: Robert Creager Cc: Tom Lane , Robert Creager , pgsql-performance@postgresql.org, Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050719224908.15415222@thunder.logicalchaos.org> In-Reply-To: <20050719120951.00002520@C118181.stortek.com> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> <20050719105014.00007635@C118181.stortek.com> <7322.1121792062@sss.pgh.pa.us> <20050719120951.00002520@C118181.stortek.com> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary=Signature_Tue__19_Jul_2005_22_49_08_-0600_wWEYYWYeiUbpipZS; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/333 X-Sequence-Number: 13574 --Signature_Tue__19_Jul_2005_22_49_08_-0600_wWEYYWYeiUbpipZS Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable When grilled further on (Tue, 19 Jul 2005 12:09:51 -0600), Robert Creager confessed: > On Tue, 19 Jul 2005 12:54:22 -0400 > Tom Lane wrote: >=20 > > Hmm, I hadn't thought about the possible impact of multiple concurrent > > vacuums. Is the problem caused by that, or has performance already gone > > into the tank by the time the cron-driven vacuums are taking long enough > > to overlap? >=20 >=20 > I'll re-start the database, vacuum full analyze and restart the runs with= out the > cron vacuum running. >=20 It took a few hours, but the problem did finally occur with no vacuum runni= ng on 803. CS is averaging 72k. I cannot quantitatively say it took longer to reproduce than with the vacuums running, but it seemed like it did. Can any information be gotten out of this? Should I try CVS HEAD? Thoughts? Thanks, Rob --=20 22:41:36 up 6 days, 2:16, 6 users, load average: 0.15, 0.21, 0.30 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Tue__19_Jul_2005_22_49_08_-0600_wWEYYWYeiUbpipZS Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLd18QACgkQLQ/DKuwDYznK2QCghoAdxDC1d7NdiXZbWbT+H0B8 qzEAoJekKizUGtpYUdlDVo62F6N7KkKs =M0q0 -----END PGP SIGNATURE----- --Signature_Tue__19_Jul_2005_22_49_08_-0600_wWEYYWYeiUbpipZS-- From pgsql-performance-owner@postgresql.org Sat Jul 23 21:36:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C9EC1529A0 for ; Wed, 20 Jul 2005 06:11:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 21737-04 for ; Wed, 20 Jul 2005 09:11:42 +0000 (GMT) Received: from mailgate.intershop.de (mailgate.intershop.de [217.17.202.241]) by svr1.postgresql.org (Postfix) with ESMTP id 12CFB528B9 for ; Wed, 20 Jul 2005 06:11:41 -0300 (ADT) Received: from JENMAIL01.ad.intershop.net ([10.0.87.43]) by mailgate.intershop.de (8.11.6p2-2003-09-18-01/8.11.6) with ESMTP id j6K9BYR17281 for ; Wed, 20 Jul 2005 11:11:34 +0200 (MET DST) X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C58D0A.CB71708E" Subject: Re: Looking for tips Date: Wed, 20 Jul 2005 11:05:17 +0200 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Looking for tips Thread-Index: AcWMz8G6pIPxuVVnSO6CWDN/Q6TuAwAOmQ6e From: "Marc Mamin" To: "Oliver Crosby" Cc: X-Virus-Scanned: by AMaViS-perl11-milter (http://amavis.org/) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.057 tagged_above=0 required=5 tests=HTML_30_40, HTML_MESSAGE X-Spam-Level: X-Archive-Number: 200507/362 X-Sequence-Number: 13603 This is a multi-part message in MIME format. ------_=_NextPart_001_01C58D0A.CB71708E Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable =20 Hi, I have a similar application, but instead of adding new items to the db once at time, I retrieve new IDs from a sequence (actually only every 10'000 times) = and write a csv file from perl. When finished, I load all new record in one run with Copy. =20 hth, =20 Marc Mamin ________________________________ From: pgsql-performance-owner@postgresql.org on behalf of Oliver Crosby Sent: Wed 7/20/2005 3:50 AM To: PFC Cc: Sven Willenberger; Dawid Kuroczko; Kevin Grittner; = jd@commandprompt.com; pgsql-performance@postgresql.org Subject: Re: [PERFORM] Looking for tips Sorry for the lack of specifics... We have a file generated as a list of events, one per line. Suppose lines 1,2,3,5,7,11,etc were related, then the last one would specify that it's the last event. Gradually this gets assembled by a perl script and when the last event is encountered, it gets inserted into the db. For a given table, let's say it's of the form (a,b,c) where 'a' is a pkey, 'b' is indexed, and 'c' is other related information. The most common 'b' values are cached locally with the perl script to save us having to query the db. So what we end up having is: if 'b' exists in cache, use cached 'a' value and continue else if 'b' exists in the db, use the associated 'a' value and continue else add a new line with 'b', return the new 'a' and continue The local cache was a huge time saver with mysql. I've tried making a plpgsql function that handles everything in one step on the db side, but it didn't show any improvement. Time permitting, I'll try some new approaches with changing the scripts and queries, though right now I was just hoping to tune postgresql.conf to work better with the hardware available. Thanks to everyone for your help. Very much appreciated. ---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings ------_=_NextPart_001_01C58D0A.CB71708E Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable =0A= =0A= =0A= =0A= =0A= =0A= Re: [PERFORM] Looking for tips=0A= =0A= =0A=
=0A=
 
=0A=
Hi,
=0A=
I have a similar application,
=0A=
but instead of adding new items to the db once at = time,
=0A=
I retrieve new IDs from a sequence (actually only = every 10'000 =0A= times) and write a csv file from perl.
=0A=
When finished, I load all new record in one run with = Copy.
=0A=
 
=0A=
hth,
=0A=
 
=0A=
Marc Mamin
=0A=
=0A=
=0A=
=0A=
From: =0A= pgsql-performance-owner@postgresql.org on behalf of Oliver =0A= Crosby
Sent: Wed 7/20/2005 3:50 AM
To: = PFC
Cc: =0A= Sven Willenberger; Dawid Kuroczko; Kevin Grittner; jd@commandprompt.com; =0A= pgsql-performance@postgresql.org
Subject: Re: [PERFORM] = Looking for =0A= tips

=0A=
=0A=

Sorry for the lack of specifics...

We have a = file =0A= generated as a list of events, one per line. Suppose
lines = 1,2,3,5,7,11,etc =0A= were related, then the last one would specify
that it's the last = event. =0A= Gradually this gets assembled by a perl
script and when the last = event is =0A= encountered, it gets inserted into
the db. For a given table, let's = say it's =0A= of the form (a,b,c) where
'a' is a pkey, 'b' is indexed, and 'c' is = other =0A= related information.
The most common 'b' values are cached locally = with the =0A= perl script to
save us having to query the db. So what we end up = having =0A= is:

if 'b' exists in cache, use cached 'a' value and = continue
else if =0A= 'b' exists in the db, use the associated 'a' value and continue
else = add a =0A= new line with 'b', return the new 'a' and continue

The local = cache was a =0A= huge time saver with mysql. I've tried making a
plpgsql function that = handles =0A= everything in one step on the db side,
but it didn't show any = improvement. =0A= Time permitting, I'll try some new
approaches with changing the = scripts and =0A= queries, though right now I
was just hoping to tune postgresql.conf = to work =0A= better with the
hardware available.

Thanks to everyone for = your help. =0A= Very much appreciated.

---------------------------(end of =0A= broadcast)---------------------------
TIP 5: don't forget to increase = your =0A= free space map settings

=0A= =0A= =0A= ------_=_NextPart_001_01C58D0A.CB71708E-- From pgsql-performance-owner@postgresql.org Wed Jul 20 12:25:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 04A1552A11 for ; Wed, 20 Jul 2005 12:25:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 99736-01 for ; Wed, 20 Jul 2005 15:25:13 +0000 (GMT) Received: from mail2.aeccom.com (port-212-202-101-158.static.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id 642D052A0C for ; Wed, 20 Jul 2005 12:25:11 -0300 (ADT) Received: from [192.168.2.12] (port-83-236-156-26.static.qsc.de [83.236.156.26]) by mail2.aeccom.com (Postfix) with ESMTP id ED8ED69 for ; Wed, 20 Jul 2005 17:25:09 +0200 (CEST) Message-ID: <42DE6CD5.8070100@aeccom.com> Date: Wed, 20 Jul 2005 17:25:09 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH User-Agent: Mozilla Thunderbird 1.0.6 (X11/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Optimizer seems to be way off, why? Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.065 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/334 X-Sequence-Number: 13575 Hi, I do not under stand the following explain output (pgsql 8.0.3): explain analyze select b.e from b, d where b.r=516081780 and b.c=513652057 and b.e=d.e; QUERY PLAN ---------------------------------------------------------------------------------------------------------------- Nested Loop (cost=0.00..1220.09 rows=1 width=4) (actual time=0.213..2926.845 rows=324503 loops=1) -> Index Scan using b_index on b (cost=0.00..1199.12 rows=1 width=4) (actual time=0.104..17.418 rows=3293 loops=1) Index Cond: (r = 516081780::oid) Filter: (c = 513652057::oid) -> Index Scan using d_e_index on d (cost=0.00..19.22 rows=140 width=4) (actual time=0.009..0.380 rows=99 loops=3293) Index Cond: ("outer".e = d.e) Total runtime: 3638.783 ms (7 rows) Why is the rows estimate for b_index and the nested loop 1? It is actually 3293 and 324503. I did VACUUM ANALYZE before and I also increased the STATISTICS TARGET on b.e to 500. No change. Here is the size of the tables: select count(oid) from b; 3532161 select count(oid) from b where r=516081780 and c=513652057; 3293 select count(oid) from d; 117270 Regards, Dirk From pgsql-performance-owner@postgresql.org Wed Jul 20 12:53:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3B63852A1E for ; Wed, 20 Jul 2005 12:53:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04654-05 for ; Wed, 20 Jul 2005 15:53:33 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.196]) by svr1.postgresql.org (Postfix) with ESMTP id 3C841529F1 for ; Wed, 20 Jul 2005 12:53:32 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i21so1465165wra for ; Wed, 20 Jul 2005 08:53:32 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=K5jEmBr3RI051uvTjdt2tBoxgsiBxxpqw8Q79l/Bbx3IUDg7Jj8E14ab4fk4m9+LdooLzgChlBryzu1x407C/OypdE6xPIpgck5nQ0EdTTFViStzlFuG8rz+m9jWPt+Qsgl7r7hKS1gw9adWX85ZwZZRlHBcyetBp8ROq6zhTyU= Received: by 10.54.13.37 with SMTP id 37mr139716wrm; Wed, 20 Jul 2005 08:52:52 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Wed, 20 Jul 2005 08:52:52 -0700 (PDT) Message-ID: <59d991c405072008523c6914bc@mail.gmail.com> Date: Wed, 20 Jul 2005 11:52:52 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: PFC , Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <59d991c4050719122213ac3d81@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c40507190930244ba9bb@mail.gmail.com> <7224.1121791328@sss.pgh.pa.us> <59d991c4050719095440cd3de5@mail.gmail.com> <7436.1121792725@sss.pgh.pa.us> <59d991c405071910133e28031a@mail.gmail.com> <7843.1121796596@sss.pgh.pa.us> <59d991c40507191148e6f9ec5@mail.gmail.com> <8375.1121799193@sss.pgh.pa.us> <59d991c4050719122213ac3d81@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.143 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/335 X-Sequence-Number: 13576 On 7/19/05, Christopher Petrilli wrote: > On 7/19/05, Tom Lane wrote: > > Christopher Petrilli writes: > > > On 7/19/05, Tom Lane wrote: > > >> How *exactly* are you invoking psql? > > > > > It is a subprocess of a Python process, driven using a pexpect > > > interchange. I send the COPY command, then wait for the '=3D#' to com= e > > > back. > > > > Some weird interaction with pexpect maybe? Try adding "-n" (disable > > readline) to the psql command switches. >=20 > Um... WOW! > It doesn't stay QUITE that low, but it stays lower... quite a bit. > We'll see what happens over time. here's a look at the difference: http://blog.amber.org/diagrams/pgsql_readline_impact.png I'm running additional comparisons AFTER clustering and analyzing the table= s...=20 Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Wed Jul 20 13:17:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D365F529B3 for ; Wed, 20 Jul 2005 13:17:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 11254-02 for ; Wed, 20 Jul 2005 16:17:08 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.193]) by svr1.postgresql.org (Postfix) with ESMTP id 7AB055281E for ; Wed, 20 Jul 2005 13:17:07 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i21so1472455wra for ; Wed, 20 Jul 2005 09:17:07 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=i3FpIplwzZagXyHrL/fYoRMmLi+Q0fMzSVdxMpn8bcrZgWQdZbrQsQSYOLQcF8a90F1V8iZgnSPOXQTgdb1e17VvmtgAUcZ76zIdxuHiMczuP99sRlJ+R6C28F76yjQ1OPldlobZub9FoHauADaHI7jZvwnSIw2wfCrteOIAsIU= Received: by 10.54.68.4 with SMTP id q4mr137747wra; Wed, 20 Jul 2005 09:16:26 -0700 (PDT) Received: by 10.54.120.11 with HTTP; Wed, 20 Jul 2005 09:16:26 -0700 (PDT) Message-ID: <59d991c405072009163b149809@mail.gmail.com> Date: Wed, 20 Jul 2005 12:16:26 -0400 From: Christopher Petrilli Reply-To: Christopher Petrilli To: Tom Lane Subject: Re: Impact of checkpoint_segments under continual load conditions Cc: Vivek Khera , pgsql-performance@postgresql.org In-Reply-To: <59d991c405071907482bb0689b@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <59d991c4050717100874c08a86@mail.gmail.com> <59d991c405071811457ccdd3f1@mail.gmail.com> <16394.1121715156@sss.pgh.pa.us> <59d991c405071812342e290f57@mail.gmail.com> <16794.1121718743@sss.pgh.pa.us> <59d991c405071907482bb0689b@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.141 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/336 X-Sequence-Number: 13577 On 7/19/05, Christopher Petrilli wrote: > It looks like the CVS HEAD is definately "better," but not by a huge > amount. The only difference is I wasn't run autovacuum in the > background (default settings), but I don't think this explains it. > Here's a graph of the differences and density of behavior: >=20 > http://blog.amber.org/diagrams/pgsql_copy_803_cvs.png >=20 > I can provide the raw data. Each COPY was 500 rows. Note that fsync > is turned off here. Maybe it'd be more stable with it turned on? I've updated this with trend-lines. Chris --=20 | Christopher Petrilli | petrilli@gmail.com From pgsql-performance-owner@postgresql.org Wed Jul 20 14:20:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B69CC52833 for ; Wed, 20 Jul 2005 14:20:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 19724-10 for ; Wed, 20 Jul 2005 17:20:14 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id E925A52825 for ; Wed, 20 Jul 2005 14:20:12 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id AF17340DC9F; Wed, 20 Jul 2005 18:19:56 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id E7EAB15EE6; Wed, 20 Jul 2005 18:01:46 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 26694-09; Wed, 20 Jul 2005 18:01:42 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 91A6C15EE1; Wed, 20 Jul 2005 18:01:42 +0100 (BST) Message-ID: <42DE8375.8020205@archonet.com> Date: Wed, 20 Jul 2005 18:01:41 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Cc: pgsql-performance@postgresql.org Subject: Re: Optimizer seems to be way off, why? References: <42DE6CD5.8070100@aeccom.com> In-Reply-To: <42DE6CD5.8070100@aeccom.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/337 X-Sequence-Number: 13578 Dirk Lutzeb=E4ck wrote: > Hi, >=20 > I do not under stand the following explain output (pgsql 8.0.3): >=20 > explain analyze > select b.e from b, d > where b.r=3D516081780 and b.c=3D513652057 and b.e=3Dd.e; >=20 > QUERY PLAN > -----------------------------------------------------------------------= -----------------------------------------=20 >=20 > Nested Loop (cost=3D0.00..1220.09 rows=3D1 width=3D4) (actual=20 > time=3D0.213..2926.845 rows=3D324503 loops=3D1) > -> Index Scan using b_index on b (cost=3D0.00..1199.12 rows=3D1 wid= th=3D4)=20 > (actual time=3D0.104..17.418 rows=3D3293 loops=3D1) > Index Cond: (r =3D 516081780::oid) > Filter: (c =3D 513652057::oid) > -> Index Scan using d_e_index on d (cost=3D0.00..19.22 rows=3D140=20 > width=3D4) (actual time=3D0.009..0.380 rows=3D99 loops=3D3293) > Index Cond: ("outer".e =3D d.e) > Total runtime: 3638.783 ms > (7 rows) >=20 > Why is the rows estimate for b_index and the nested loop 1? It is=20 > actually 3293 and 324503. I'm guessing (and that's all it is) that b.r and b.c have a higher=20 correlation than the planner is expecting. That is, it expects the=20 b.c=3D... to reduce the number of matching rows much more than it is. Try a query just on WHERE b.r=3D516081780 and see if it gets the estimate= =20 right for that. If it's a common query, it might be worth an index on (r,c) -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Wed Jul 20 16:16:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9CDBA52A0E for ; Wed, 20 Jul 2005 16:16:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44506-03 for ; Wed, 20 Jul 2005 19:16:54 +0000 (GMT) Received: from mail2.aeccom.com (port-212-202-101-158.static.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id 12D4B52A04 for ; Wed, 20 Jul 2005 16:16:52 -0300 (ADT) Received: from [192.168.2.132] (openvpn02.core.aeccom.com [192.168.2.132]) by mail2.aeccom.com (Postfix) with ESMTP id 9549369; Wed, 20 Jul 2005 21:16:52 +0200 (CEST) Message-ID: <42DEA308.6030608@aeccom.com> Date: Wed, 20 Jul 2005 21:16:24 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Reply-To: lutzeb@aeccom.com User-Agent: Mozilla Thunderbird 0.9 (X11/20041103) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: pgsql-performance@postgresql.org Subject: Re: Optimizer seems to be way off, why? References: <42DE6CD5.8070100@aeccom.com> <42DE8375.8020205@archonet.com> In-Reply-To: <42DE8375.8020205@archonet.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/338 X-Sequence-Number: 13579 Richard Huxton wrote: > Dirk Lutzeb�ck wrote: > >> Hi, >> >> I do not under stand the following explain output (pgsql 8.0.3): >> >> explain analyze >> select b.e from b, d >> where b.r=516081780 and b.c=513652057 and b.e=d.e; >> >> QUERY PLAN >> ---------------------------------------------------------------------------------------------------------------- >> >> Nested Loop (cost=0.00..1220.09 rows=1 width=4) (actual >> time=0.213..2926.845 rows=324503 loops=1) >> -> Index Scan using b_index on b (cost=0.00..1199.12 rows=1 >> width=4) (actual time=0.104..17.418 rows=3293 loops=1) >> Index Cond: (r = 516081780::oid) >> Filter: (c = 513652057::oid) >> -> Index Scan using d_e_index on d (cost=0.00..19.22 rows=140 >> width=4) (actual time=0.009..0.380 rows=99 loops=3293) >> Index Cond: ("outer".e = d.e) >> Total runtime: 3638.783 ms >> (7 rows) >> >> Why is the rows estimate for b_index and the nested loop 1? It is >> actually 3293 and 324503. > > > I'm guessing (and that's all it is) that b.r and b.c have a higher > correlation than the planner is expecting. That is, it expects the > b.c=... to reduce the number of matching rows much more than it is. > > Try a query just on WHERE b.r=516081780 and see if it gets the estimate > right for that. > > If it's a common query, it might be worth an index on (r,c) > > -- > Richard Huxton > Archonet Ltd > Thanks Richard, dropping the join for b.c now gives better estimates (it also uses a different index now) although not accurate (off by factor 10). This query is embedded in a larger query which now got a 1000 times speed up (!) because I can drop b.c because it is redundant. Though, why can't the planner see this correlation? I think somebody said the planner does not know about multiple column correlations, does it? Regards, Dirk From pgsql-performance-owner@postgresql.org Wed Jul 20 17:28:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3E5C75293B for ; Wed, 20 Jul 2005 17:28:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 57065-04 for ; Wed, 20 Jul 2005 20:28:33 +0000 (GMT) Received: from calvin.surfutopia.net (calvin.surfutopia.net [67.120.245.34]) by svr1.postgresql.org (Postfix) with ESMTP id C9B1252AA5 for ; Wed, 20 Jul 2005 17:28:31 -0300 (ADT) Received: by calvin.surfutopia.net (Postfix, from userid 1000) id 7ADC9F23B; Wed, 20 Jul 2005 13:28:32 -0700 (PDT) Date: Wed, 20 Jul 2005 13:28:32 -0700 From: John Mendenhall To: pgsql-performance list Subject: Re: performance decrease after reboot Message-ID: <20050720202832.GA22109@calvin.surfutopia.net> References: <20050719210557.GA30794@calvin.surfutopia.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050719210557.GA30794@calvin.surfutopia.net> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/339 X-Sequence-Number: 13580 On Tue, 19 Jul 2005, John Mendenhall wrote: > I tuned a query last week to obtain acceptable performance. > Here is my recorded explain analyze results: > > LOG: duration: 826.505 ms statement: explain analyze > [cut for brevity] > > I rebooted the database machine later that night. > Now, when I run the same query, I get the following > results: > > LOG: duration: 6931.701 ms statement: explain analyze > [cut for brevity] I just ran my query again, no changes from yesterday and it is back to normal: LOG: duration: 795.839 ms statement: explain analyze What could have been the problem? The major differences in the query plan are as follows: (1) The one that runs faster uses a Hash Join at the very top of the query plan. It does a Hash Cond on the country and code fields. (2) The one that runs slower uses a Materialize with the subplan, with no Hash items. The Materialize does Seq Scan of the countries table, and above it, a Join Filter is run. (3) The partners_pkey index on the partners table is in a different place in the query. Does anyone know what would cause the query plan to be different like this, for the same server, same query? I run vacuum analyze every night. Is this perhaps the problem? What setting do I need to tweak to make sure the faster plan is always found? Thanks for any pointers in this dilemma. JohnM -- John Mendenhall john@surfutopia.net surf utopia internet services From pgsql-performance-owner@postgresql.org Wed Jul 20 17:58:10 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0B9BD52A44 for ; Wed, 20 Jul 2005 17:58:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 61845-08 for ; Wed, 20 Jul 2005 20:57:59 +0000 (GMT) Received: from ionian.i-clic.uihc.uiowa.edu (ionian.i-clic.uihc.uiowa.edu [129.255.211.20]) by svr1.postgresql.org (Postfix) with ESMTP id 5A00952833 for ; Wed, 20 Jul 2005 17:57:54 -0300 (ADT) Received: by ionian.i-clic.uihc.uiowa.edu (Postfix, from userid 76) id 6B61261DAF; Wed, 20 Jul 2005 14:50:13 -0500 (CDT) Received: from [192.168.1.104] (65-100-17-227.cdrr.qwest.net [65.100.17.227]) by ionian.i-clic.uihc.uiowa.edu (Postfix) with ESMTP id 5EEC461DAE; Wed, 20 Jul 2005 14:50:11 -0500 (CDT) Message-ID: <42DEBACF.2020405@arbash-meinel.com> Date: Wed, 20 Jul 2005 15:57:51 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.6 (Macintosh/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: lutzeb@aeccom.com Cc: Richard Huxton , pgsql-performance@postgresql.org Subject: Re: Optimizer seems to be way off, why? References: <42DE6CD5.8070100@aeccom.com> <42DE8375.8020205@archonet.com> <42DEA308.6030608@aeccom.com> In-Reply-To: <42DEA308.6030608@aeccom.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig94CE97DBB20D147FCE6A8757" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/340 X-Sequence-Number: 13581 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig94CE97DBB20D147FCE6A8757 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Dirk Lutzeb=E4ck wrote: > Richard Huxton wrote: >=20 >> Dirk Lutzeb=E4ck wrote: >> >>> Hi, >>> >>> I do not under stand the following explain output (pgsql 8.0.3): >>> >>> explain analyze >>> select b.e from b, d >>> where b.r=3D516081780 and b.c=3D513652057 and b.e=3Dd.e; >>> >>> QUERY PLAN >>> ---------------------------------------------------------------------= ------------------------------------------- >>> >>> Nested Loop (cost=3D0.00..1220.09 rows=3D1 width=3D4) (actual >>> time=3D0.213..2926.845 rows=3D324503 loops=3D1) >>> -> Index Scan using b_index on b (cost=3D0.00..1199.12 rows=3D1 >>> width=3D4) (actual time=3D0.104..17.418 rows=3D3293 loops=3D1) >>> Index Cond: (r =3D 516081780::oid) >>> Filter: (c =3D 513652057::oid) >>> -> Index Scan using d_e_index on d (cost=3D0.00..19.22 rows=3D140= >>> width=3D4) (actual time=3D0.009..0.380 rows=3D99 loops=3D3293) >>> Index Cond: ("outer".e =3D d.e) >>> Total runtime: 3638.783 ms >>> (7 rows) >>> >>> Why is the rows estimate for b_index and the nested loop 1? It is >>> actually 3293 and 324503. >> >> >> >> I'm guessing (and that's all it is) that b.r and b.c have a higher >> correlation than the planner is expecting. That is, it expects the >> b.c=3D... to reduce the number of matching rows much more than it is. >> >> Try a query just on WHERE b.r=3D516081780 and see if it gets the >> estimate right for that. >> >> If it's a common query, it might be worth an index on (r,c) >> >> --=20 >> Richard Huxton >> Archonet Ltd >> >=20 > Thanks Richard, dropping the join for b.c now gives better estimates (i= t > also uses a different index now) although not accurate (off by factor > 10). This query is embedded in a larger query which now got a 1000 time= s > speed up (!) because I can drop b.c because it is redundant. Well, part of the problem is that the poorly estimated row is not 'b.e' but 'b.r', it expects to only find one row that matches, and instead finds 3293 rows. Now, that *could* be because it mis-estimates the selectivity of b.r & b.= c. It actually estimated the join with d approximately correctly. (It thought that for each row it would find 140, and it averaged 99). >=20 > Though, why can't the planner see this correlation? I think somebody > said the planner does not know about multiple column correlations, does= it? The planner does not maintain cross-column statistics, so you are correct. I believe it assumes distributions are independent. So that if r=3DRRRRR is 10% selective, and c=3DCCCC is 20% selective, the total selectivity of r=3DRRRR AND c=3DCCCC is 2%. I could be wrong on this, but= I think it is approximately correct. Now if you created the index on b(r,c), then it would have a much better idea of how selective that would be. At the very least, it could index on (r,c) rather than indexing on (r) and filtering by (c). Also, if you have very skewed data (where you have 1 value 100k times, and 50 values only 10times each), the planner can overestimate the low values, and underestimate the high one. (It uses random sampling, so it kind of depends where the entries are.) Have you tried increasing the statistics on b.r and or b.c? Do you have an index on b.c or just b.r? To see what the planner thinks, you might try: EXPLAIN ANALYZE select count(*) from b where r=3D516081780; That would tell you how selective the planner thinks the r=3D is. >=20 > Regards, >=20 > Dirk >=20 John =3D:-> --------------enig94CE97DBB20D147FCE6A8757 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC3rrTJdeBCYSNAAMRAihcAJ4ime7NRUFh5qAs5gueJTu6mx5YagCgmYuo +OUMPNe8fv5k22wO96NKP5U= =7G8I -----END PGP SIGNATURE----- --------------enig94CE97DBB20D147FCE6A8757-- From pgsql-performance-owner@postgresql.org Sat Jul 23 21:42:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B98C152B1C for ; Thu, 21 Jul 2005 07:02:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 19776-03 for ; Thu, 21 Jul 2005 10:02:10 +0000 (GMT) Received: from charon.4soft.de (charon.4soft.de [62.159.46.131]) by svr1.postgresql.org (Postfix) with ESMTP id 8974B52AE8 for ; Thu, 21 Jul 2005 07:02:06 -0300 (ADT) Received: from zeus.4soft.de (zeus.4soft.de [172.16.0.1]) by charon.4soft.de (Postfix) with ESMTP id C1672D098 for ; Thu, 21 Jul 2005 12:02:08 +0200 (CEST) Received: from [172.16.1.14] (eos.4soft.de [172.16.1.14]) by zeus.4soft.de (8.12.8/8.12.8) with ESMTP id j6LA25Fh012809 for ; Thu, 21 Jul 2005 12:02:05 +0200 Message-ID: <42DF72AA.3090207@4soft.de> Date: Thu, 21 Jul 2005 12:02:18 +0200 From: Achim Luber User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Size of empty varchar and size of index X-Enigmail-Version: 0.92.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by AMaViS-perl11-milter (http://amavis.org/) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/365 X-Sequence-Number: 13606 Hello, I'm searching for two facts: How much space takes a varchar column if there is no value in it (NULL)? How much space needs a index of an integer column? Hope I post to the right list and hope anybody can help me. Thank you Greetings Achim From pgsql-performance-owner@postgresql.org Thu Jul 21 12:58:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0E19552A81 for ; Thu, 21 Jul 2005 12:58:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 97103-01 for ; Thu, 21 Jul 2005 15:58:54 +0000 (GMT) Received: from rwcrmhc12.comcast.net (rwcrmhc13.comcast.net [204.127.198.39]) by svr1.postgresql.org (Postfix) with ESMTP id 772BD52997 for ; Thu, 21 Jul 2005 12:58:51 -0300 (ADT) Received: from [192.168.1.52] (c-24-60-119-214.hsd1.ma.comcast.net[24.60.119.214]) by comcast.net (rwcrmhc13) with ESMTP id <20050721155850015001v13ce>; Thu, 21 Jul 2005 15:58:51 +0000 Message-ID: <42DFC641.1020404@comcast.net> Date: Thu, 21 Jul 2005 11:58:57 -0400 From: Jeffrey Tenny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7) Gecko/20040616 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: What is best way to stream terabytes of data into postgresql? Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.295 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, DNS_FROM_RFC_POST X-Spam-Level: * X-Archive-Number: 200507/341 X-Sequence-Number: 13582 Preferably via JDBC, but by C/C++ if necessary. Streaming being the operative word. Tips appreciated. From pgsql-performance-owner@postgresql.org Thu Jul 21 14:45:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9E2FC52905 for ; Thu, 21 Jul 2005 14:45:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 35440-02 for ; Thu, 21 Jul 2005 17:45:23 +0000 (GMT) Received: from svr4.postgresql.org (svr4.postgresql.org [66.98.251.159]) by svr1.postgresql.org (Postfix) with ESMTP id 1F89E528F9 for ; Thu, 21 Jul 2005 14:45:19 -0300 (ADT) Received: from stark.xeocode.com (stark.xeocode.com [216.58.44.227]) by svr4.postgresql.org (Postfix) with ESMTP id BE25C5AFAE9 for ; Thu, 21 Jul 2005 16:49:15 +0000 (GMT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1Dve5p-00060p-00; Thu, 21 Jul 2005 12:40:17 -0400 To: John A Meinel Cc: lutzeb@aeccom.com, Richard Huxton , pgsql-performance@postgresql.org Subject: Re: Optimizer seems to be way off, why? References: <42DE6CD5.8070100@aeccom.com> <42DE8375.8020205@archonet.com> <42DEA308.6030608@aeccom.com> <42DEBACF.2020405@arbash-meinel.com> In-Reply-To: <42DEBACF.2020405@arbash-meinel.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 21 Jul 2005 12:40:17 -0400 Message-ID: <87vf34cdu6.fsf@stark.xeocode.com> Lines: 16 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/342 X-Sequence-Number: 13583 John A Meinel writes: > Now if you created the index on b(r,c), then it would have a much better > idea of how selective that would be. At the very least, it could index > on (r,c) rather than indexing on (r) and filtering by (c). There has been some discussion of adding functionality like this but afaik no version of Postgres actually does this yet. Adding the index may still help though. -- greg From pgsql-performance-owner@postgresql.org Thu Jul 21 15:28:40 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0720752B72 for ; Thu, 21 Jul 2005 15:28:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 37303-01 for ; Thu, 21 Jul 2005 18:28:26 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id 1288F52B3B for ; Thu, 21 Jul 2005 15:28:25 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Thu, 21 Jul 2005 14:28:18 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Thu, 21 Jul 2005 14:28:18 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: What is best way to stream terabytes of data into Date: Thu, 21 Jul 2005 14:26:04 -0400 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] What is best way to stream terabytes of data into postgresql? Thread-Index: AcWODTo78umGiTNxTN2A4mg23ktjvQAEfrKQ From: "Frank Wosczyna" To: "Jeffrey Tenny" , pgsql-performance@postgresql.org X-OriginalArrivalTime: 21 Jul 2005 18:28:18.0694 (UTC) FILETIME=[F779B260:01C58E21] X-WSS-ID: 6EC136C81J49368304-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/343 X-Sequence-Number: 13584 =20 > Subject: [PERFORM] What is best way to stream terabytes of=20 > data into postgresql? >=20 > Preferably via JDBC, but by C/C++ if necessary. >=20 > Streaming being the operative word. >=20 > Tips appreciated. >=20 Hi, We granted our Java Loader to the Bizgres Open Source, http://www.bizgres.org/assets/BZ_userguide.htm#50413574_pgfId-110126 You can load from STDIN instead of a file, as long as you prepend the stream with the Loader Control file, for example: for name in customer orders lineitem partsupp supplier part;do;cat TPCH_load_100gb_${name}.ctl /mnt//TPCH-Data/${name}.tbl.* | loader.sh -h localhost -p 10001 -d tpch -t -u mpp; done You can also run the loader from a remote host as well, with the "-h" being the target system with the Postgres database. If you have terabytes of data, you might want to set a batch size (-b switch) to commit occasionally. Feel free to contact me directly if you have questions. Thanks, Frank Frank Wosczyna Systems Engineer Greenplum / Bizgres MPP www.greenplum.com From pgsql-patches-owner@postgresql.org Thu Jul 21 18:55:10 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A2E00529D8 for ; Thu, 21 Jul 2005 18:55:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54040-08 for ; Thu, 21 Jul 2005 21:55:02 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id D9A3B52999 for ; Thu, 21 Jul 2005 18:54:58 -0300 (ADT) Received: from ibm-b.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6LLsdjA002847 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Thu, 21 Jul 2005 14:54:39 -0700 Message-Id: <200507212154.j6LLsdjA002847@smtp.osdl.org> Date: Thu, 21 Jul 2005 14:55:07 -0700 From: Mark Wong To: "Luke Lonergan" Cc: "Andrew Dunstan" , "Alvaro Herrera" , "Bruce Momjian" , "Alon Goldshuv" , pgsql-patches@postgresql.org, maryedie@osdl.org Subject: Re: COPY FROM performance improvements In-Reply-To: References: <42C05F88.7000105@dunslane.net> Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.057 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/424 X-Sequence-Number: 16764 I just ran through a few tests with the v14 patch against 100GB of data from dbt3 and found a 30% improvement; 3.6 hours vs 5.3 hours. Just to give a few details, I only loaded data and started a COPY in parallel for each the data files: http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/ Here's a visual of my disk layout, for those familiar with the database schema: http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/layout-dev4-010-dbt3.html I have 6 arrays of fourteen 15k rpm drives in a split-bus configuration attached to a 4-way itanium2 via 6 compaq smartarray pci-x controllers. Let me know if you have any questions. Mark From pgsql-patches-owner@postgresql.org Thu Jul 21 20:15:06 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A5CAA528DE for ; Thu, 21 Jul 2005 20:15:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 16472-01 for ; Thu, 21 Jul 2005 23:14:57 +0000 (GMT) Received: from mail.Mi8.com (d01gw01.mi8.com [63.240.6.47]) by svr1.postgresql.org (Postfix) with ESMTP id D9BEA52836 for ; Thu, 21 Jul 2005 20:14:55 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D1)); Thu, 21 Jul 2005 19:14:49 -0400 X-Server-Uuid: 241911D6-425B-44B9-A073-E3FE0F8FC774 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Thu, 21 Jul 2005 19:14:48 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Thu, 21 Jul 2005 19:14:48 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 21 Jul 2005 16:14:47 -0700 Subject: Re: COPY FROM performance improvements From: "Luke Lonergan" To: "Mark Wong" Cc: "Andrew Dunstan" , "Alvaro Herrera" , "Bruce Momjian" , "Alon Goldshuv" , pgsql-patches@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <200507212154.j6LLsdjA002847@smtp.osdl.org> MIME-Version: 1.0 X-OriginalArrivalTime: 21 Jul 2005 23:14:48.0751 (UTC) FILETIME=[FD8C3BF0:01C58E49] X-WSS-ID: 6EFEF3E32B44412511-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.574 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/425 X-Sequence-Number: 16765 Cool! At what rate does your disk setup write sequential data, e.g.: time dd if=/dev/zero of=bigfile bs=8k count=500000 (sized for 2x RAM on a system with 2GB) BTW - the Compaq smartarray controllers are pretty broken on Linux from a performance standpoint in our experience. We've had disastrously bad results from the SmartArray 5i and 6 controllers on kernels from 2.4 -> 2.6.10, on the order of 20MB/s. For comparison, the results on our dual opteron with a single LSI SCSI controller with software RAID0 on a 2.6.10 kernel: [llonergan@stinger4 dbfast]$ time dd if=/dev/zero of=bigfile bs=8k count=500000 500000+0 records in 500000+0 records out real 0m24.702s user 0m0.077s sys 0m8.794s Which calculates out to about 161MB/s. - Luke On 7/21/05 2:55 PM, "Mark Wong" wrote: > I just ran through a few tests with the v14 patch against 100GB of data > from dbt3 and found a 30% improvement; 3.6 hours vs 5.3 hours. Just to > give a few details, I only loaded data and started a COPY in parallel > for each the data files: > http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/ > > Here's a visual of my disk layout, for those familiar with the database > schema: > http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/layout-dev4 > -010-dbt3.html > > I have 6 arrays of fourteen 15k rpm drives in a split-bus configuration > attached to a 4-way itanium2 via 6 compaq smartarray pci-x controllers. > > Let me know if you have any questions. > > Mark > From pgsql-patches-owner@postgresql.org Thu Jul 21 21:08:02 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 75076529ED for ; Thu, 21 Jul 2005 21:07:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 46523-04 for ; Fri, 22 Jul 2005 00:07:56 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 9C14552901 for ; Thu, 21 Jul 2005 21:07:56 -0300 (ADT) Received: from [192.168.1.100] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6M07Rhr028248; Thu, 21 Jul 2005 17:07:35 -0700 Message-ID: <42E038E9.1080107@commandprompt.com> Date: Thu, 21 Jul 2005 17:08:09 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Luke Lonergan Cc: Mark Wong , Andrew Dunstan , Alvaro Herrera , Bruce Momjian , Alon Goldshuv , pgsql-patches@postgresql.org, maryedie@osdl.org Subject: Re: COPY FROM performance improvements References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/426 X-Sequence-Number: 16766 Luke Lonergan wrote: > Cool! > > At what rate does your disk setup write sequential data, e.g.: > time dd if=/dev/zero of=bigfile bs=8k count=500000 > > (sized for 2x RAM on a system with 2GB) > > BTW - the Compaq smartarray controllers are pretty broken on Linux from a > performance standpoint in our experience. We've had disastrously bad > results from the SmartArray 5i and 6 controllers on kernels from 2.4 -> > 2.6.10, on the order of 20MB/s. O.k. this strikes me as interesting, now we know that Compaq and Dell are borked for Linux. Is there a name brand server (read Enterprise) that actually does provide reasonable performance? > > For comparison, the results on our dual opteron with a single LSI SCSI > controller with software RAID0 on a 2.6.10 kernel: > > [llonergan@stinger4 dbfast]$ time dd if=/dev/zero of=bigfile bs=8k > count=500000 > 500000+0 records in > 500000+0 records out > > real 0m24.702s > user 0m0.077s > sys 0m8.794s > > Which calculates out to about 161MB/s. > > - Luke > > > On 7/21/05 2:55 PM, "Mark Wong" wrote: > > >>I just ran through a few tests with the v14 patch against 100GB of data >>from dbt3 and found a 30% improvement; 3.6 hours vs 5.3 hours. Just to >>give a few details, I only loaded data and started a COPY in parallel >>for each the data files: >>http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/ >> >>Here's a visual of my disk layout, for those familiar with the database >>schema: >>http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/layout-dev4 >>-010-dbt3.html >> >>I have 6 arrays of fourteen 15k rpm drives in a split-bus configuration >>attached to a 4-way itanium2 via 6 compaq smartarray pci-x controllers. >> >>Let me know if you have any questions. >> >>Mark >> > > > > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Thu Jul 21 21:23:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CFCDB529F5 for ; Thu, 21 Jul 2005 21:23:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 52149-08 for ; Fri, 22 Jul 2005 00:23:18 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 6551B52B91 for ; Thu, 21 Jul 2005 21:23:17 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7639743; Thu, 21 Jul 2005 17:25:34 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Re: What is best way to stream terabytes of data into postgresql? Date: Thu, 21 Jul 2005 17:26:03 -0700 User-Agent: KMail/1.8 Cc: Jeffrey Tenny References: <42DFC641.1020404@comcast.net> In-Reply-To: <42DFC641.1020404@comcast.net> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507211726.03438.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.048 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/344 X-Sequence-Number: 13585 Jeff, > Streaming being the operative word. Not sure how much hacking you want to do, but the TelegraphCQ project is based on PostgreSQL: http://telegraph.cs.berkeley.edu/telegraphcq/v0.2/ -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-patches-owner@postgresql.org Thu Jul 21 23:05:13 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3DE4952BBD for ; Thu, 21 Jul 2005 23:05:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20073-03 for ; Fri, 22 Jul 2005 02:05:01 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.50]) by svr1.postgresql.org (Postfix) with ESMTP id 80AD0529E9 for ; Thu, 21 Jul 2005 23:04:59 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Thu, 21 Jul 2005 22:04:57 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Thu, 21 Jul 2005 22:04:57 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Thu, 21 Jul 2005 22:04:56 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 21 Jul 2005 19:04:55 -0700 Subject: Re: COPY FROM performance improvements From: "Luke Lonergan" To: "Joshua D. Drake" Cc: "Mark Wong" , "Andrew Dunstan" , "Alvaro Herrera" , "Bruce Momjian" , "Alon Goldshuv" , pgsql-patches@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <42E038E9.1080107@commandprompt.com> MIME-Version: 1.0 X-OriginalArrivalTime: 22 Jul 2005 02:04:57.0229 (UTC) FILETIME=[C2465BD0:01C58E61] X-WSS-ID: 6EFE8BC31M89611019-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.574 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/427 X-Sequence-Number: 16767 Joshua, On 7/21/05 5:08 PM, "Joshua D. Drake" wrote: > O.k. this strikes me as interesting, now we know that Compaq and Dell > are borked for Linux. Is there a name brand server (read Enterprise) > that actually does provide reasonable performance? I think late model Dell (post the bad chipset problem, circa 2001-2?) and IBM and Sun servers are fine because they all use simple SCSI adapters from LSI or Adaptec. The HP Smartarray is an aberration, they don't have good driver support for Linux and as a consequence have some pretty bad problems with both performance and stability. On Windows they perform quite well. Also - there are very big issues with some SATA controllers and Linux we've seen, particularly the Silicon Image, Highpoint other non-Intel controllers. Not sure about Nvidia, but the only ones I trust now are 3Ware and the others mentioned in earlier posts. - Luke From pgsql-patches-owner@postgresql.org Thu Jul 21 23:53:33 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 06E6252A70 for ; Thu, 21 Jul 2005 23:53:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25246-07 for ; Fri, 22 Jul 2005 02:53:18 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id E6304528DE for ; Thu, 21 Jul 2005 23:53:17 -0300 (ADT) Received: from [192.168.1.100] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6M2qqhr006958; Thu, 21 Jul 2005 19:52:53 -0700 Message-ID: <42E05FAF.70601@commandprompt.com> Date: Thu, 21 Jul 2005 19:53:35 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Luke Lonergan Cc: Mark Wong , Andrew Dunstan , Alvaro Herrera , Bruce Momjian , Alon Goldshuv , pgsql-patches@postgresql.org, maryedie@osdl.org Subject: Re: COPY FROM performance improvements References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/428 X-Sequence-Number: 16768 > I think late model Dell (post the bad chipset problem, circa 2001-2?) and > IBM and Sun servers are fine because they all use simple SCSI adapters from > LSI or Adaptec. Well I know that isn't true at least not with ANY of the Dells my customers have purchased in the last 18 months. They are still really, really slow. > Also - there are very big issues with some SATA controllers and Linux we've > seen, particularly the Silicon Image, Highpoint other non-Intel controllers. > Not sure about Nvidia, but the only ones I trust now are 3Ware and the > others mentioned in earlier posts. I have great success with Silicon Image as long as I am running them with Linux software RAID. The LSI controllers are also really nice. J > > - Luke > > > -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-patches-owner@postgresql.org Fri Jul 22 01:19:22 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 22D0B52892 for ; Fri, 22 Jul 2005 01:19:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 55075-09 for ; Fri, 22 Jul 2005 04:19:18 +0000 (GMT) Received: from mail.mi8.com (d01gw04.mi8.com [63.240.6.44]) by svr1.postgresql.org (Postfix) with ESMTP id 0902652ABE for ; Fri, 22 Jul 2005 01:19:16 -0300 (ADT) Received: from 172.16.1.118 by mail.mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D4)); Fri, 22 Jul 2005 00:19:07 -0400 X-Server-Uuid: C8FB4D43-1108-484A-A898-3CBCC7906230 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 22 Jul 2005 00:19:06 -0400 Received: from 24.5.173.15 ([24.5.173.15]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Fri, 22 Jul 2005 00:19:05 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 21 Jul 2005 21:19:04 -0700 Subject: Re: COPY FROM performance improvements From: "Luke Lonergan" To: "Joshua D. Drake" Cc: "Mark Wong" , "Andrew Dunstan" , "Alvaro Herrera" , "Bruce Momjian" , "Alon Goldshuv" , pgsql-patches@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <42E05FAF.70601@commandprompt.com> MIME-Version: 1.0 X-OriginalArrivalTime: 22 Jul 2005 04:19:06.0968 (UTC) FILETIME=[804B0D80:01C58E74] X-WSS-ID: 6EFEAC3046K1855287-02-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.6 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/429 X-Sequence-Number: 16769 Joshua, On 7/21/05 7:53 PM, "Joshua D. Drake" wrote: > Well I know that isn't true at least not with ANY of the Dells my > customers have purchased in the last 18 months. They are still really, > really slow. That's too bad, can you cite some model numbers? SCSI? > I have great success with Silicon Image as long as I am running them > with Linux software RAID. The LSI controllers are also really nice. That's good to hear, I gave up on Silicon Image controllers on Linux about 1 year ago, which kernel are you using with success? Silicon Image controllers are the most popular, so it's important to see them supported well, though I'd rather see more SATA headers than 2 off of the built-in chipsets. - Luke From pgsql-patches-owner@postgresql.org Fri Jul 22 03:13:55 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9B31D528DC for ; Fri, 22 Jul 2005 03:13:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87709-01 for ; Fri, 22 Jul 2005 06:13:46 +0000 (GMT) Received: from trolak.mydnsbox2.com (ns1.mydnsbox2.com [207.44.142.118]) by svr1.postgresql.org (Postfix) with ESMTP id D04D752BA1 for ; Fri, 22 Jul 2005 03:13:45 -0300 (ADT) Received: from dunslane.net (localhost.localdomain [127.0.0.1]) (authenticated (0 bits)) by trolak.mydnsbox2.com (8.11.6/8.11.6) with ESMTP id j6M5Res23270; Fri, 22 Jul 2005 00:27:40 -0500 Received: from 24.211.165.134 (SquirrelMail authenticated user andrew@dunslane.net) by www.dunslane.net with HTTP; Fri, 22 Jul 2005 00:27:41 -0500 (CDT) Message-ID: <3277.24.211.165.134.1122010061.squirrel@www.dunslane.net> Date: Fri, 22 Jul 2005 00:27:41 -0500 (CDT) Subject: Re: COPY FROM performance improvements From: "Andrew Dunstan" To: In-Reply-To: References: X-Priority: 3 Importance: Normal X-MSMail-Priority: Normal Cc: , , , , , , , X-Mailer: SquirrelMail (version 1.2.5) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/430 X-Sequence-Number: 16770 this discussion belongs on -performance cheers andrew Luke Lonergan said: > Joshua, > > On 7/21/05 7:53 PM, "Joshua D. Drake" wrote: >> Well I know that isn't true at least not with ANY of the Dells my >> customers have purchased in the last 18 months. They are still really, >> really slow. > > That's too bad, can you cite some model numbers? SCSI? > >> I have great success with Silicon Image as long as I am running them >> with Linux software RAID. The LSI controllers are also really nice. > > That's good to hear, I gave up on Silicon Image controllers on Linux > about 1 year ago, which kernel are you using with success? Silicon > Image > controllers are the most popular, so it's important to see them > supported well, though I'd rather see more SATA headers than 2 off of > the built-in chipsets. > > - Luke From pgsql-performance-owner@postgresql.org Fri Jul 22 10:06:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E28DD5284E for ; Fri, 22 Jul 2005 10:06:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 81088-01 for ; Fri, 22 Jul 2005 13:06:15 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.200]) by svr1.postgresql.org (Postfix) with ESMTP id CCAD952863 for ; Fri, 22 Jul 2005 10:06:14 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id i28so193593nzi for ; Fri, 22 Jul 2005 06:06:15 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=LBh8BtXXDBt6pHS7SX2EytQveROsyFSORjVaaIq2dtCOcGrbQObA0ulaZjN1vThSE2e1MuEP1Ia0hILUCxjTh+5k095uArvfjKV8miTsEV3X3L3YkHLbKUoIWgiQYK9jCtlQ50dwidXFcRIRja5ACaC86pHujF7XxlB9GjYVbK0= Received: by 10.36.60.18 with SMTP id i18mr1407065nza; Fri, 22 Jul 2005 02:10:05 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Fri, 22 Jul 2005 02:10:05 -0700 (PDT) Message-ID: <758d5e7f0507220210bfa4978@mail.gmail.com> Date: Fri, 22 Jul 2005 11:10:05 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Postgresql Performance Subject: Planner doesn't look at LIMIT? Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.612 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/345 X-Sequence-Number: 13586 Hello, I have PostgreSQL 8.0.3 running on a "workstation" with 768 MB of RAM, under FreeBSD. And I have a 47-milion row table: qnex=3D# explain select * from log; QUERY PLAN =20 ----------------------------------------------------------------------- Seq Scan on log (cost=3D0.00..1741852.36 rows=3D47044336 width=3D180) (1 row) ...which is joined with a few smaller ones, like: qnex=3D# explain select * from useragents; QUERY PLAN =20 ------------------------------------------------------------------- Seq Scan on useragents (cost=3D0.00..9475.96 rows=3D364896 width=3D96) (1 row) shared_buffers =3D 5000 random_page_cost =3D 3 work_mem =3D 102400 effective_cache_size =3D 60000 Now, if I do a SELECT: qnex=3D# EXPLAIN SELECT * FROM log NATURAL JOIN useragents LIMIT 1; QUERY PLAN ---------------------------------------------------------------------------= ---------- Limit (cost=3D15912.20..15912.31 rows=3D1 width=3D272) -> Hash Join (cost=3D15912.20..5328368.96 rows=3D47044336 width=3D272) Hash Cond: ("outer".useragent_id =3D "inner".useragent_id) -> Seq Scan on log (cost=3D0.00..1741852.36 rows=3D47044336 widt= h=3D180) -> Hash (cost=3D9475.96..9475.96 rows=3D364896 width=3D96) -> Seq Scan on useragents (cost=3D0.00..9475.96 rows=3D364896 width=3D96) (6 rows) Or: qnex=3D# EXPLAIN SELECT * FROM log NATURAL LEFT JOIN useragents LIMIT 1; QUERY PLAN ---------------------------------------------------------------------------= ---------- Limit (cost=3D15912.20..15912.31 rows=3D1 width=3D272) -> Hash Left Join (cost=3D15912.20..5328368.96 rows=3D47044336 width= =3D272) Hash Cond: ("outer".useragent_id =3D "inner".useragent_id) -> Seq Scan on log (cost=3D0.00..1741852.36 rows=3D47044336 widt= h=3D180) -> Hash (cost=3D9475.96..9475.96 rows=3D364896 width=3D96) -> Seq Scan on useragents (cost=3D0.00..9475.96 rows=3D364896 width=3D96) (6 rows) Time: 2.688 ms ...the query seems to last forever (its hashing 47 million rows!) If I set enable_hashjoin=3Dfalse: qnex=3D# EXPLAIN ANALYZE SELECT * FROM log NATURAL LEFT JOIN useragents LIM= IT 1; QUERY PLAN ---------------------------------------------------------------------------= ---------------------------------------------------------------- Limit (cost=3D0.00..3.07 rows=3D1 width=3D272) (actual time=3D74.214..74.= 216 rows=3D1 loops=3D1) -> Nested Loop Left Join (cost=3D0.00..144295895.01 rows=3D47044336 width=3D272) (actual time=3D74.204..74.204 rows=3D1 loops=3D1) -> Seq Scan on log (cost=3D0.00..1741852.36 rows=3D47044336 width=3D180) (actual time=3D23.270..23.270 rows=3D1 loops=3D1) -> Index Scan using useragents_pkey on useragents=20 (cost=3D0.00..3.02 rows=3D1 width=3D96) (actual time=3D50.867..50.867 rows= =3D1 loops=3D1) Index Cond: ("outer".useragent_id =3D useragents.useragent_i= d) Total runtime: 74.483 ms ...which is way faster. Of course if I did: qnex=3D# EXPLAIN ANALYZE SELECT * FROM log NATURAL LEFT JOIN useragents WHERE logid =3D (SELECT logid FROM log LIMIT 1); QUERY PLAN ---------------------------------------------------------------------------= -------------------------------------------------------- Nested Loop Left Join (cost=3D0.04..6.09 rows=3D1 width=3D272) (actual time=3D61.403..61.419 rows=3D1 loops=3D1) InitPlan -> Limit (cost=3D0.00..0.04 rows=3D1 width=3D4) (actual time=3D0.029..0.032 rows=3D1 loops=3D1) -> Seq Scan on log (cost=3D0.00..1741852.36 rows=3D47044336 width=3D4) (actual time=3D0.023..0.023 rows=3D1 loops=3D1) -> Index Scan using log_pkey on log (cost=3D0.00..3.02 rows=3D1 width=3D180) (actual time=3D61.316..61.319 rows=3D1 loops=3D1) Index Cond: (logid =3D $0) -> Index Scan using useragents_pkey on useragents=20 (cost=3D0.00..3.02 rows=3D1 width=3D96) (actual time=3D0.036..0.042 rows=3D= 1 loops=3D1) Index Cond: ("outer".useragent_id =3D useragents.useragent_id) Total runtime: 61.741 ms (9 rows) ...I tried tweaking cpu_*, work_mem, effective_cache and so on, but without any luck. 47 milion table is huge compared to useragents (I actually need to join the log with 3 similar to useragents tables, and create a view out of it). Also tried using LEFT/RIGHT JOINS insead of (inner) JOINs... Of course the database is freshly vacuum analyzed, and statistics are set at 50... My view of the problem is that planner ignores the "LIMIT" part. It assume= s it _needs_ to return all 47 million rows joined with the useragents table, = so the hashjoin is the only sane approach. But chances are that unless I'll use LIMIT 200000, the nested loop will be much faster. Any ideas how to make it work (other than rewriting the query to use subselects, use explicit id-rows, disabling hashjoin completely)? Or is this a bug? Regards, Dawid From pgsql-performance-owner@postgresql.org Fri Jul 22 10:17:54 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6487F528B7 for ; Fri, 22 Jul 2005 10:17:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 81156-04 for ; Fri, 22 Jul 2005 13:16:56 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 8548A5284E for ; Fri, 22 Jul 2005 10:16:54 -0300 (ADT) Received: (qmail 2673 invoked from network); 22 Jul 2005 15:16:58 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 22 Jul 2005 15:16:58 +0200 To: "Dawid Kuroczko" , "Postgresql Performance" Subject: Re: Planner doesn't look at LIMIT? References: <758d5e7f0507220210bfa4978@mail.gmail.com> Message-ID: Date: Fri, 22 Jul 2005 15:16:57 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <758d5e7f0507220210bfa4978@mail.gmail.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/346 X-Sequence-Number: 13587 Which row do you want ? Do you want 'a row' at random ? I presume you want the N latest rows ? In that case you should use an ORDER BY on an indexed field, the serial primary key will do nicely (ORDER BY id DESC) ; it's indexed so it will use the index and it will fly. > Any ideas how to make it work (other than rewriting the query to use > subselects, use explicit id-rows, disabling hashjoin completely)? > Or is this a bug? > > Regards, > Dawid > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match > From pgsql-performance-owner@postgresql.org Fri Jul 22 11:40:00 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8372A52896 for ; Fri, 22 Jul 2005 11:39:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 01385-06 for ; Fri, 22 Jul 2005 14:39:56 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id A57145288B for ; Fri, 22 Jul 2005 11:39:54 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6MEdvbu029474; Fri, 22 Jul 2005 10:39:57 -0400 (EDT) To: Dawid Kuroczko Cc: Postgresql Performance Subject: Re: Planner doesn't look at LIMIT? In-reply-to: <758d5e7f0507220210bfa4978@mail.gmail.com> References: <758d5e7f0507220210bfa4978@mail.gmail.com> Comments: In-reply-to Dawid Kuroczko message dated "Fri, 22 Jul 2005 11:10:05 +0200" Date: Fri, 22 Jul 2005 10:39:57 -0400 Message-ID: <29473.1122043197@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/347 X-Sequence-Number: 13588 Dawid Kuroczko writes: > qnex=# EXPLAIN SELECT * FROM log NATURAL JOIN useragents LIMIT 1; > Limit (cost=15912.20..15912.31 rows=1 width=272) > -> Hash Join (cost=15912.20..5328368.96 rows=47044336 width=272) > If I set enable_hashjoin=false: > qnex=# EXPLAIN ANALYZE SELECT * FROM log NATURAL LEFT JOIN useragents LIMIT 1; > Limit (cost=0.00..3.07 rows=1 width=272) (actual time=74.214..74.216 > rows=1 loops=1) > -> Nested Loop Left Join (cost=0.00..144295895.01 rows=47044336 > width=272) (actual time=74.204..74.204 rows=1 loops=1) This is quite strange. The nestloop plan definitely should be preferred in the context of the LIMIT, considering that it has far lower estimated cost. And it is preferred in simple tests for me. It seems there must be something specific to your installation that's causing the planner to go wrong. Can you develop a self-contained test case that behaves this way for you? I recall we saw a similar complaint a month or two back, but the complainant never followed up with anything useful for tracking down the problem. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Jul 22 11:48:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9742C52896 for ; Fri, 22 Jul 2005 11:48:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04076-05 for ; Fri, 22 Jul 2005 14:48:10 +0000 (GMT) Received: from tomts40-srv.bellnexxia.net (unknown [209.226.175.97]) by svr1.postgresql.org (Postfix) with ESMTP id 820D552869 for ; Fri, 22 Jul 2005 11:48:08 -0300 (ADT) Received: from [192.168.1.99] ([206.172.223.162]) by tomts40-srv.bellnexxia.net (InterMail vM.5.01.06.10 201-253-122-130-110-20040306) with ESMTP id <20050722144806.CIZI1799.tomts40-srv.bellnexxia.net@[192.168.1.99]> for ; Fri, 22 Jul 2005 10:48:06 -0400 Message-ID: <42E106D2.5080604@alteeve.com> Date: Fri, 22 Jul 2005 10:46:42 -0400 From: Madison Kelly User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Debian/1.7.8-1 X-Accept-Language: en-us, en, ja MIME-Version: 1.0 To: pgsqlperform Subject: Another index question Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/348 X-Sequence-Number: 13589 Hi all, I am trying to do an update on a table but so far I can't seem to come up with a usable index. After my last question/thread the user 'PFC' recommended I store whether a file was to be backed up as either 't'(rue), 'f'(alse) or 'i'(nherit) to speed up changing files and sub directories under a given directory when it was toggled. I've more or less finished implementing this and it is certainly a LOT faster but I am hoping to make it just a little faster still with an Index. Tom Lane pointed out to me that I needed 'text_pattern_ops' on my 'file_parent_dir' column in the index if I wanted to do pattern matching (the C locale wasn't set). Now I have added an additional condition and I think this might be my problem. Here is a sample query I am trying to create my index for: UPDATE file_info_2 SET file_backup='i' WHERE file_backup!='i' AND file_parent_dir='/'; This would be an example of someone changing the backup state of the root of a partition. It could also be: UPDATE file_info_2 SET file_backup='i' WHERE file_backup!='i' AND file_parent_dir='/usr'; If, for example, the user was toggling the backup state of the '/usr' directory. I suspected that because I was using "file_backup!='i'" that maybe I was running into the same problem as before so I tried creating the index: tle-bu=> CREATE INDEX file_info_2_mupdate_idx ON file_info_2 (file_backup bpchar_pattern_ops, file_parent_dir text_pattern_ops); tle-bu=> EXPLAIN ANALYZE UPDATE file_info_2 SET file_backup='i' WHERE file_backup!='i' AND file_parent_dir~'^/'; QUERY PLAN ----------------------------------------------------------------------------------------------------------------- Seq Scan on file_info_2 (cost=0.00..13379.38 rows=1 width=134) (actual time=1623.819..1624.087 rows=4 loops=1) Filter: ((file_backup <> 'i'::bpchar) AND (file_parent_dir ~ '^/'::text)) Total runtime: 1628.053 ms (3 rows) This index wasn't used though, even when I set 'enable_seqscan' to 'OFF'. The column 'file_backup' is 'char(1)' and the column 'file_parent_dir' is 'text'. tle-bu=> \d file_info_2; \di file_info_2_mupdate_idx; Table "public.file_info_2" Column | Type | Modifiers -----------------+--------------+------------------------------ file_group_name | text | file_group_uid | integer | not null file_mod_time | bigint | not null file_name | text | not null file_parent_dir | text | not null file_perm | integer | not null file_size | bigint | not null file_type | character(1) | not null file_user_name | text | file_user_uid | integer | not null file_backup | character(1) | not null default 'i'::bpchar file_display | character(1) | not null default 'i'::bpchar file_restore | character(1) | not null default 'i'::bpchar Indexes: "file_info_2_mupdate_idx" btree (file_backup bpchar_pattern_ops, file_parent_dir text_pattern_ops) "file_info_2_supdate_idx" btree (file_parent_dir, file_name, file_type) List of relations Schema | Name | Type | Owner | Table --------+-------------------------+-------+---------+------------- public | file_info_2_mupdate_idx | index | madison | file_info_2 (1 row) Could it be that there needs to be a certain number of "file_backup!='i'" before the planner will use the index? I have also tried not defining an op_class on both tables (and one at a time) but I can't seem to figure this out. As always, thank you! Madison From pgsql-performance-owner@postgresql.org Fri Jul 22 11:51:37 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1741D528D7 for ; Fri, 22 Jul 2005 11:51:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04076-08 for ; Fri, 22 Jul 2005 14:51:17 +0000 (GMT) Received: from colo.samason.me.uk (unknown [69.55.228.22]) by svr1.postgresql.org (Postfix) with ESMTP id 5FCE25288B for ; Fri, 22 Jul 2005 11:51:17 -0300 (ADT) Received: from colo.samason.me.uk (colo [69.55.228.22]) by colo.samason.me.uk (8.12.11/8.12.11) with ESMTP id j6MEpMQo038585 for ; Fri, 22 Jul 2005 15:51:22 +0100 (BST) (envelope-from sam@colo.samason.me.uk) Received: (from sam@localhost) by colo.samason.me.uk (8.12.11/8.12.11/Submit) id j6MEpM8S038584 for pgsql-performance@postgresql.org; Fri, 22 Jul 2005 15:51:22 +0100 (BST) (envelope-from sam) Date: Fri, 22 Jul 2005 15:51:22 +0100 From: Sam Mason To: Postgresql Performance Subject: Re: Planner doesn't look at LIMIT? Message-ID: <20050722145122.GL62747@colo.samason.me.uk> Mail-Followup-To: Sam Mason , Postgresql Performance References: <758d5e7f0507220210bfa4978@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <758d5e7f0507220210bfa4978@mail.gmail.com> User-Agent: Mutt/1.4.2.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/349 X-Sequence-Number: 13590 Dawid Kuroczko wrote: >work_mem = 102400 >...I tried tweaking cpu_*, work_mem, effective_cache and so on, but without >any luck. I'm hoping you didn't tweak it enough! I posted something similar this a while ago, but haven't since got around to figuring out a useful test case to send to the list. Try reducing your work_mem down to 1000 or so and things should start doing what you expect. Sam From pgsql-patches-owner@postgresql.org Fri Jul 22 12:03:05 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2A8A352803 for ; Fri, 22 Jul 2005 12:03:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 08825-04 for ; Fri, 22 Jul 2005 15:02:57 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 5A2E8528D9 for ; Fri, 22 Jul 2005 12:02:57 -0300 (ADT) Received: from [192.168.1.100] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6MF2Zhr029333; Fri, 22 Jul 2005 08:02:35 -0700 Message-ID: <42E10AC0.1080905@commandprompt.com> Date: Fri, 22 Jul 2005 08:03:28 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Luke Lonergan Cc: Mark Wong , Andrew Dunstan , Alvaro Herrera , Bruce Momjian , Alon Goldshuv , pgsql-patches@postgresql.org, maryedie@osdl.org Subject: Re: COPY FROM performance improvements References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/433 X-Sequence-Number: 16773 Luke Lonergan wrote: > Joshua, > > On 7/21/05 7:53 PM, "Joshua D. Drake" wrote: > >>Well I know that isn't true at least not with ANY of the Dells my >>customers have purchased in the last 18 months. They are still really, >>really slow. > > > That's too bad, can you cite some model numbers? SCSI? Yeah I will get them and post, but yes they are all SCSI. > > >>I have great success with Silicon Image as long as I am running them >>with Linux software RAID. The LSI controllers are also really nice. > > > That's good to hear, I gave up on Silicon Image controllers on Linux about 1 > year ago, which kernel are you using with success? Any of the 2.6 kernels. ALso the laster 2.4 (+22 I believe) support it pretty well as well. Silicon Image > controllers are the most popular, so it's important to see them supported > well, though I'd rather see more SATA headers than 2 off of the built-in > chipsets. > > - Luke > -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Fri Jul 22 12:11:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0C79E528C0 for ; Fri, 22 Jul 2005 12:11:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 06881-10 for ; Fri, 22 Jul 2005 15:11:03 +0000 (GMT) Received: from tomts5-srv.bellnexxia.net (tomts5.bellnexxia.net [209.226.175.25]) by svr1.postgresql.org (Postfix) with ESMTP id CD39F52896 for ; Fri, 22 Jul 2005 12:10:59 -0300 (ADT) Received: from [192.168.1.99] ([206.172.171.182]) by tomts5-srv.bellnexxia.net (InterMail vM.5.01.06.10 201-253-122-130-110-20040306) with ESMTP id <20050722151057.FOLM26128.tomts5-srv.bellnexxia.net@[192.168.1.99]> for ; Fri, 22 Jul 2005 11:10:57 -0400 Message-ID: <42E10C2C.4040202@alteeve.com> Date: Fri, 22 Jul 2005 11:09:32 -0400 From: Madison Kelly User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050513 Debian/1.7.8-1 X-Accept-Language: en-us, en, ja MIME-Version: 1.0 To: pgsqlperform Subject: Solved (was: Re: Another index question) References: <42E106D2.5080604@alteeve.com> In-Reply-To: <42E106D2.5080604@alteeve.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/350 X-Sequence-Number: 13591 Line noise, sorry... After posting I went back to reading the pgsql docs and saw the query: SELECT am.amname AS index_method, opc.opcname AS opclass_name, opr.oprname AS opclass_operator FROM pg_am am, pg_opclass opc, pg_amop amop, pg_operator opr WHERE opc.opcamid = am.oid AND amop.amopclaid = opc.oid AND amop.amopopr = opr.oid ORDER BY index_method, opclass_name, opclass_operator; Which listed all the op_classes. I noticed none of the opclass_operators supported '!=' so I wondered if that was simply an unindexable (is that a word?) operator. So I tried creating the index: tle-bu=> CREATE INDEX file_info_2_mupdate_idx ON file_info_2 (file_backup, file_parent_dir text_pattern_ops); And changing my query to: tle-bu=> EXPLAIN ANALYZE UPDATE file_info_2 SET file_backup='i' WHERE file_backup='t' OR file_backup='f' AND file_parent_dir~'^/'; QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------------------------- Index Scan using file_info_2_mupdate_idx, file_info_2_mupdate_idx on file_info_2 (cost=0.00..10.04 rows=1 width=134) (actual time=0.112..0.718 rows=4 loops=1) Index Cond: ((file_backup = 't'::bpchar) OR ((file_backup = 'f'::bpchar) AND (file_parent_dir ~>=~ '/'::text) AND (file_parent_dir ~<~ '0'::text))) Filter: ((file_backup = 't'::bpchar) OR ((file_backup = 'f'::bpchar) AND (file_parent_dir ~ '^/'::text))) Total runtime: 60.359 ms (4 rows) Bingo! Hopefully someone might find this useful in the archives. :p Madison Madison Kelly wrote: > Hi all, > > I am trying to do an update on a table but so far I can't seem to come > up with a usable index. After my last question/thread the user 'PFC' > recommended I store whether a file was to be backed up as either > 't'(rue), 'f'(alse) or 'i'(nherit) to speed up changing files and sub > directories under a given directory when it was toggled. I've more or > less finished implementing this and it is certainly a LOT faster but I > am hoping to make it just a little faster still with an Index. > > Tom Lane pointed out to me that I needed 'text_pattern_ops' on my > 'file_parent_dir' column in the index if I wanted to do pattern matching > (the C locale wasn't set). Now I have added an additional condition and > I think this might be my problem. Here is a sample query I am trying to > create my index for: > > > UPDATE file_info_2 SET file_backup='i' WHERE file_backup!='i' AND > file_parent_dir='/'; > > This would be an example of someone changing the backup state of the > root of a partition. It could also be: > > > UPDATE file_info_2 SET file_backup='i' WHERE file_backup!='i' AND > file_parent_dir='/usr'; > > If, for example, the user was toggling the backup state of the '/usr' > directory. > > I suspected that because I was using "file_backup!='i'" that maybe I > was running into the same problem as before so I tried creating the index: > > > tle-bu=> CREATE INDEX file_info_2_mupdate_idx ON file_info_2 > (file_backup bpchar_pattern_ops, file_parent_dir text_pattern_ops); > > tle-bu=> EXPLAIN ANALYZE UPDATE file_info_2 SET file_backup='i' WHERE > file_backup!='i' AND file_parent_dir~'^/'; QUERY PLAN > ----------------------------------------------------------------------------------------------------------------- > > Seq Scan on file_info_2 (cost=0.00..13379.38 rows=1 width=134) (actual > time=1623.819..1624.087 rows=4 loops=1) > Filter: ((file_backup <> 'i'::bpchar) AND (file_parent_dir ~ > '^/'::text)) > Total runtime: 1628.053 ms > (3 rows) > > > This index wasn't used though, even when I set 'enable_seqscan' to > 'OFF'. The column 'file_backup' is 'char(1)' and the column > 'file_parent_dir' is 'text'. > > > tle-bu=> \d file_info_2; \di file_info_2_mupdate_idx; Table > "public.file_info_2" > Column | Type | Modifiers > -----------------+--------------+------------------------------ > file_group_name | text | > file_group_uid | integer | not null > file_mod_time | bigint | not null > file_name | text | not null > file_parent_dir | text | not null > file_perm | integer | not null > file_size | bigint | not null > file_type | character(1) | not null > file_user_name | text | > file_user_uid | integer | not null > file_backup | character(1) | not null default 'i'::bpchar > file_display | character(1) | not null default 'i'::bpchar > file_restore | character(1) | not null default 'i'::bpchar > Indexes: > "file_info_2_mupdate_idx" btree (file_backup bpchar_pattern_ops, > file_parent_dir text_pattern_ops) > "file_info_2_supdate_idx" btree (file_parent_dir, file_name, file_type) > > List of relations > Schema | Name | Type | Owner | Table > --------+-------------------------+-------+---------+------------- > public | file_info_2_mupdate_idx | index | madison | file_info_2 > (1 row) > > Could it be that there needs to be a certain number of > "file_backup!='i'" before the planner will use the index? I have also > tried not defining an op_class on both tables (and one at a time) but I > can't seem to figure this out. > > As always, thank you! > > Madison From pgsql-performance-owner@postgresql.org Fri Jul 22 13:09:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9655F5297E for ; Fri, 22 Jul 2005 13:09:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 29777-01 for ; Fri, 22 Jul 2005 16:09:40 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.207]) by svr1.postgresql.org (Postfix) with ESMTP id E8DFF5282F for ; Fri, 22 Jul 2005 13:09:36 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id x3so234625nzd for ; Fri, 22 Jul 2005 09:09:37 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=pWOPRiwqOkO5qXHAKdUUOSISG7INoiC7PyFZzp1QNhVUh8rCDsaR8JHEf8l6PyQQIj5AaYpyU8HYcMtlmXwjFqCltIxLEy485tOHnxPS5YijYa7ZS9xwBt42+yhtSlRKOXQRfVujZ8x/7EFBGEVJAAscPs+ayv+jj2ZOvoQO7nM= Received: by 10.36.3.7 with SMTP id 7mr1584052nzc; Fri, 22 Jul 2005 09:09:37 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Fri, 22 Jul 2005 09:09:37 -0700 (PDT) Message-ID: <758d5e7f05072209094156c984@mail.gmail.com> Date: Fri, 22 Jul 2005 18:09:37 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Tom Lane Subject: Re: Planner doesn't look at LIMIT? Cc: Postgresql Performance In-Reply-To: <29473.1122043197@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <758d5e7f0507220210bfa4978@mail.gmail.com> <29473.1122043197@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.612 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/351 X-Sequence-Number: 13592 On 7/22/05, Tom Lane wrote: > Dawid Kuroczko writes: > > qnex=3D# EXPLAIN SELECT * FROM log NATURAL JOIN useragents LIMIT 1; >=20 > > Limit (cost=3D15912.20..15912.31 rows=3D1 width=3D272) > > -> Hash Join (cost=3D15912.20..5328368.96 rows=3D47044336 width=3D= 272) >=20 > This is quite strange. The nestloop plan definitely should be preferred > in the context of the LIMIT, considering that it has far lower estimated > cost. And it is preferred in simple tests for me. It seems there must > be something specific to your installation that's causing the planner to > go wrong. Can you develop a self-contained test case that behaves this > way for you? Why, certainly. I did test it also on Gentoo Linux PostgreSQL 8.0.1 (yeah, a bit older one), but the behaviour is the same. The test looks like this: -- First lets make a "small" lookup table -- 400000 rows. CREATE TABLE lookup ( lookup_id serial PRIMARY KEY, value integer NOT NULL ); INSERT INTO lookup (value) SELECT * FROM generate_series(1, 400000); VACUUM ANALYZE lookup; -- Then lets make a huge data table... CREATE TABLE huge_data ( huge_data_id serial PRIMARY KEY, lookup_id integer NOT NULL ); INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM lookup; INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 80= 0 000 INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 1 60= 0 000 INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 3 20= 0 000 INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 6 40= 0 000 INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 12 80= 0 000 -- You may want to put ANALYZE and EXPLAIN between each of these -- steps. In my cases, at 12.8 mln rows PostgreSQL seems to go for hashjoi= n -- in each case. YMMV, so you may try to push it up to 1024 mln rows. INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 25 60= 0 000 ANALYZE huge_data; EXPLAIN SELECT * FROM huge_data NATURAL JOIN lookup LIMIT 1; My EXPLAIN FROM Linux (SMP P-III box), with PostgreSQL 8.0.1, during making this test case: qnex=3D# EXPLAIN SELECT * FROM huge_data NATURAL JOIN lookup LIMIT 1; QUERY PLAN ---------------------------------------------------------------------------= --------- Limit (cost=3D0.00..3.21 rows=3D1 width=3D12) -> Nested Loop (cost=3D0.00..19557596.04 rows=3D6094777 width=3D12) -> Seq Scan on huge_data (cost=3D0.00..95372.42 rows=3D6399942 wi= dth=3D8) -> Index Scan using lookup_pkey on lookup (cost=3D0.00..3.02 rows=3D1 width=3D8) Index Cond: ("outer".lookup_id =3D lookup.lookup_id) (5 rows) Time: 4,333 ms qnex=3D# INSERT INTO huge_data (lookup_id) SELECT lookup_id FROM huge_data; -- 12 800 000 INSERT 0 6400000 Time: 501014,692 ms qnex=3D# ANALYZE huge_data; ANALYZE Time: 4243,453 ms qnex=3D# EXPLAIN SELECT * FROM huge_data NATURAL JOIN lookup LIMIT 1; QUERY PLAN ---------------------------------------------------------------------------= ---- Limit (cost=3D11719.00..11719.09 rows=3D1 width=3D12) -> Hash Join (cost=3D11719.00..1212739.73 rows=3D12800185 width=3D12) Hash Cond: ("outer".lookup_id =3D "inner".lookup_id) -> Seq Scan on huge_data (cost=3D0.00..190747.84 rows=3D12800184 = width=3D8) -> Hash (cost=3D5961.00..5961.00 rows=3D400000 width=3D8) -> Seq Scan on lookup (cost=3D0.00..5961.00 rows=3D400000 w= idth=3D8) (6 rows) Regards, Dawid From pgsql-performance-owner@postgresql.org Fri Jul 22 13:20:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7F3CD52840; Fri, 22 Jul 2005 13:20:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31810-06; Fri, 22 Jul 2005 16:20:23 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 92C4152846; Fri, 22 Jul 2005 13:20:21 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6MGKKl2006441; Fri, 22 Jul 2005 12:20:20 -0400 (EDT) To: Dawid Kuroczko Cc: pgsql-performance@postgreSQL.org, pgsql-hackers@postgreSQL.org Subject: Re: Planner doesn't look at LIMIT? In-reply-to: <29473.1122043197@sss.pgh.pa.us> References: <758d5e7f0507220210bfa4978@mail.gmail.com> <29473.1122043197@sss.pgh.pa.us> Comments: In-reply-to Tom Lane message dated "Fri, 22 Jul 2005 10:39:57 -0400" Date: Fri, 22 Jul 2005 12:20:20 -0400 Message-ID: <6440.1122049220@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/352 X-Sequence-Number: 13593 I wrote: > Dawid Kuroczko writes: >> qnex=# EXPLAIN SELECT * FROM log NATURAL JOIN useragents LIMIT 1; >> Limit (cost=15912.20..15912.31 rows=1 width=272) >> -> Hash Join (cost=15912.20..5328368.96 rows=47044336 width=272) >> If I set enable_hashjoin=false: >> qnex=# EXPLAIN ANALYZE SELECT * FROM log NATURAL LEFT JOIN useragents LIMIT 1; >> Limit (cost=0.00..3.07 rows=1 width=272) (actual time=74.214..74.216 >> rows=1 loops=1) >> -> Nested Loop Left Join (cost=0.00..144295895.01 rows=47044336 >> width=272) (actual time=74.204..74.204 rows=1 loops=1) > This is quite strange. The nestloop plan definitely should be preferred > in the context of the LIMIT, considering that it has far lower estimated > cost. And it is preferred in simple tests for me. After a suitable period of contemplating my navel, I figured out what is going on here: the total costs involved are large enough that the still-fairly-high startup cost of the hash is disregarded by compare_fuzzy_path_costs(), and so the nestloop is discarded as not having any significant potential advantage in startup time. I think that this refutes the original scheme of using the same fuzz factor for both startup and total cost comparisons, and therefore propose the attached patch. Comments? regards, tom lane *** src/backend/optimizer/util/pathnode.c.orig Fri Jul 15 13:09:25 2005 --- src/backend/optimizer/util/pathnode.c Fri Jul 22 12:08:25 2005 *************** *** 98,157 **** static int compare_fuzzy_path_costs(Path *path1, Path *path2, CostSelector criterion) { - Cost fuzz; - /* ! * The fuzz factor is set at one percent of the smaller total_cost, ! * but not less than 0.01 cost units (just in case total cost is ! * zero). * * XXX does this percentage need to be user-configurable? */ - fuzz = Min(path1->total_cost, path2->total_cost) * 0.01; - fuzz = Max(fuzz, 0.01); - if (criterion == STARTUP_COST) { ! if (Abs(path1->startup_cost - path2->startup_cost) > fuzz) ! { ! if (path1->startup_cost < path2->startup_cost) ! return -1; ! else ! return +1; ! } /* * If paths have the same startup cost (not at all unlikely), * order them by total cost. */ ! if (Abs(path1->total_cost - path2->total_cost) > fuzz) ! { ! if (path1->total_cost < path2->total_cost) ! return -1; ! else ! return +1; ! } } else { ! if (Abs(path1->total_cost - path2->total_cost) > fuzz) ! { ! if (path1->total_cost < path2->total_cost) ! return -1; ! else ! return +1; ! } /* * If paths have the same total cost, order them by startup cost. */ ! if (Abs(path1->startup_cost - path2->startup_cost) > fuzz) ! { ! if (path1->startup_cost < path2->startup_cost) ! return -1; ! else ! return +1; ! } } return 0; } --- 98,138 ---- static int compare_fuzzy_path_costs(Path *path1, Path *path2, CostSelector criterion) { /* ! * We use a fuzz factor of 1% of the smaller cost. * * XXX does this percentage need to be user-configurable? */ if (criterion == STARTUP_COST) { ! if (path1->startup_cost > path2->startup_cost * 1.01) ! return +1; ! if (path2->startup_cost > path1->startup_cost * 1.01) ! return -1; /* * If paths have the same startup cost (not at all unlikely), * order them by total cost. */ ! if (path1->total_cost > path2->total_cost * 1.01) ! return +1; ! if (path2->total_cost > path1->total_cost * 1.01) ! return -1; } else { ! if (path1->total_cost > path2->total_cost * 1.01) ! return +1; ! if (path2->total_cost > path1->total_cost * 1.01) ! return -1; /* * If paths have the same total cost, order them by startup cost. */ ! if (path1->startup_cost > path2->startup_cost * 1.01) ! return +1; ! if (path2->startup_cost > path1->startup_cost * 1.01) ! return -1; } return 0; } From pgsql-patches-owner@postgresql.org Fri Jul 22 13:49:25 2005 X-Original-To: pgsql-patches-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6F81652997 for ; Fri, 22 Jul 2005 13:49:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40405-08 for ; Fri, 22 Jul 2005 16:49:21 +0000 (GMT) Received: from henry.newn.cam.ac.uk (henry.newn.cam.ac.uk [131.111.204.130]) by svr1.postgresql.org (Postfix) with ESMTP id 8AC305282F for ; Fri, 22 Jul 2005 13:49:20 -0300 (ADT) Received: from quartz.newn.cam.ac.uk ([131.111.204.180]) by henry.newn.cam.ac.uk with esmtp (Exim 4.20) id 1Dw0ht-0003Uz-OX; Fri, 22 Jul 2005 17:49:05 +0100 Received: from prlw1 by quartz.newn.cam.ac.uk with local (Exim 4.20) id 1Dw0ht-00041x-Kl; Fri, 22 Jul 2005 17:49:05 +0100 Date: Fri, 22 Jul 2005 17:49:05 +0100 From: Patrick Welche To: Luke Lonergan Cc: "Joshua D. Drake" , Mark Wong , Andrew Dunstan , Alvaro Herrera , Bruce Momjian , Alon Goldshuv , pgsql-patches@postgresql.org, maryedie@osdl.org Subject: Re: COPY FROM performance improvements Message-ID: <20050722164905.GE18413@quartz.newn.cam.ac.uk> References: <42E05FAF.70601@commandprompt.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/436 X-Sequence-Number: 16776 On Thu, Jul 21, 2005 at 09:19:04PM -0700, Luke Lonergan wrote: > Joshua, > > On 7/21/05 7:53 PM, "Joshua D. Drake" wrote: > > Well I know that isn't true at least not with ANY of the Dells my > > customers have purchased in the last 18 months. They are still really, > > really slow. > > That's too bad, can you cite some model numbers? SCSI? I would be interested too, given http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=30531 Cheers, Patrick From pgsql-performance-owner@postgresql.org Fri Jul 22 14:12:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6791E528AD for ; Fri, 22 Jul 2005 14:12:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 55128-01 for ; Fri, 22 Jul 2005 17:12:02 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id A510852858 for ; Fri, 22 Jul 2005 14:12:02 -0300 (ADT) Received: from [192.168.1.52] (fc1smp [66.93.38.87]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6MHBLhr009212; Fri, 22 Jul 2005 10:11:22 -0700 Message-ID: <42E128BD.2000104@commandprompt.com> Date: Fri, 22 Jul 2005 10:11:25 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Patrick Welche Cc: Luke Lonergan , Mark Wong , Andrew Dunstan , Alvaro Herrera , Bruce Momjian , Alon Goldshuv , pgsql-performance@postgresql.org, maryedie@osdl.org Subject: Re: [PATCHES] COPY FROM performance improvements References: <42E05FAF.70601@commandprompt.com> <20050722164905.GE18413@quartz.newn.cam.ac.uk> In-Reply-To: <20050722164905.GE18413@quartz.newn.cam.ac.uk> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.014 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/353 X-Sequence-Number: 13594 Here is the SCSI output: Web Server SCSI subsystem driver Revision: 1.00 megaraid: v1.18j (Release Date: Mon Jul 7 14:39:55 EDT 2003) megaraid: found 0x1028:0x000f:idx 0:bus 4:slot 3:func 0 scsi0 : Found a MegaRAID controller at 0xf883f000, IRQ: 18 scsi0 : Enabling 64 bit support megaraid: [412W:H406] detected 1 logical drives megaraid: supports extended CDBs. megaraid: channel[1] is raid. megaraid: channel[2] is raid. scsi0 : LSI Logic MegaRAID 412W 254 commands 15 targs 5 chans 7 luns Database Server SCSI subsystem driver Revision: 1.00 megaraid: v1.18j (Release Date: Mon Jul 7 14:39:55 EDT 2003) megaraid: found 0x101e:0x1960:idx 0:bus 5:slot 0:func 0 scsi0 : Found a MegaRAID controller at 0xf883f000, IRQ: 21 scsi0 : Enabling 64 bit support megaraid: [196T:3.33] detected 1 logical drives megaraid: supports extended CDBs. megaraid: channel[1] is raid. megaraid: channel[2] is raid. scsi0 : LSI Logic MegaRAID 196T 254 commands 15 targs 5 chans 7 luns Starting timer : 0 0 blk: queue c5f2d218, I/O limit 4095Mb (mask 0xffffffff) scsi0: scanning virtual channel 0 for logical drives. Vendor: MegaRAID Model: LD 0 RAID5 86G Rev: 196T Type: Direct-Access ANSI SCSI revision: 02 Starting timer : 0 0 The webserver is a 1U and it actually performs better on the IO than the database server even though the database server is running 6 disks versus 3. The database server is a PE (Power Edge) 6600 Database Server IO: [root@master root]# /sbin/hdparm -tT /dev/sda /dev/sda: Timing buffer-cache reads: 1888 MB in 2.00 seconds = 944.00 MB/sec Timing buffered disk reads: 32 MB in 3.06 seconds = 10.46 MB/sec Second Database Server IO: [root@pq-slave root]# /sbin/hdparm -tT /dev/sda /dev/sda: Timing buffer-cache reads: 1816 MB in 2.00 seconds = 908.00 MB/sec Timing buffered disk reads: 26 MB in 3.11 seconds = 8.36 MB/sec [root@pq-slave root]# Which is just horrible. Sincerely, Joshua D. Drake Patrick Welche wrote: > On Thu, Jul 21, 2005 at 09:19:04PM -0700, Luke Lonergan wrote: > >>Joshua, >> >>On 7/21/05 7:53 PM, "Joshua D. Drake" wrote: >> >>>Well I know that isn't true at least not with ANY of the Dells my >>>customers have purchased in the last 18 months. They are still really, >>>really slow. >> >>That's too bad, can you cite some model numbers? SCSI? > > > I would be interested too, given > > http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=30531 > > > Cheers, > > Patrick > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Have you searched our list archives? > > http://archives.postgresql.org -- Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 PostgreSQL Replication, Consulting, Custom Programming, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ From pgsql-performance-owner@postgresql.org Fri Jul 22 14:32:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B7D06529BF; Fri, 22 Jul 2005 14:32:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 59657-03; Fri, 22 Jul 2005 17:32:05 +0000 (GMT) Received: from cmailg4.svr.pol.co.uk (cmailg4.svr.pol.co.uk [195.92.195.174]) by svr1.postgresql.org (Postfix) with ESMTP id 4C5425293D; Fri, 22 Jul 2005 14:32:04 -0300 (ADT) Received: from modem-3356.lynx.dialup.pol.co.uk ([217.135.205.28] helo=192.168.0.102) by cmailg4.svr.pol.co.uk with esmtp (Exim 4.41) id 1Dw1NS-0006CU-Ny; Fri, 22 Jul 2005 18:32:02 +0100 Subject: Re: [HACKERS] Planner doesn't look at LIMIT? From: Simon Riggs To: Tom Lane Cc: Dawid Kuroczko , pgsql-performance@postgreSQL.org, pgsql-hackers@postgreSQL.org In-Reply-To: <6440.1122049220@sss.pgh.pa.us> References: <758d5e7f0507220210bfa4978@mail.gmail.com> <29473.1122043197@sss.pgh.pa.us> <6440.1122049220@sss.pgh.pa.us> Content-Type: text/plain Organization: 2nd Quadrant Date: Fri, 22 Jul 2005 18:31:58 +0100 Message-Id: <1122053518.21502.180.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.052 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/354 X-Sequence-Number: 13595 On Fri, 2005-07-22 at 12:20 -0400, Tom Lane wrote: > I think that this refutes the original scheme of using the same fuzz > factor for both startup and total cost comparisons, and therefore > propose the attached patch. > > Comments? Looks good. I think it explains a few other wierd perf reports also. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Fri Jul 22 15:06:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 783EF529BC; Fri, 22 Jul 2005 15:06:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 66823-02; Fri, 22 Jul 2005 18:06:21 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 0670B52806; Fri, 22 Jul 2005 15:06:20 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6MI6HT8007215; Fri, 22 Jul 2005 14:06:17 -0400 (EDT) To: Simon Riggs Cc: Dawid Kuroczko , pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org Subject: Re: [HACKERS] Planner doesn't look at LIMIT? In-reply-to: <1122053518.21502.180.camel@localhost.localdomain> References: <758d5e7f0507220210bfa4978@mail.gmail.com> <29473.1122043197@sss.pgh.pa.us> <6440.1122049220@sss.pgh.pa.us> <1122053518.21502.180.camel@localhost.localdomain> Comments: In-reply-to Simon Riggs message dated "Fri, 22 Jul 2005 18:31:58 +0100" Date: Fri, 22 Jul 2005 14:06:17 -0400 Message-ID: <7214.1122055577@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/355 X-Sequence-Number: 13596 Simon Riggs writes: > Looks good. I think it explains a few other wierd perf reports also. Could be. I went back to look at Sam Mason's report about three weeks ago, and it definitely seems to explain his issue. The "fuzzy cost comparison" logic is new in 8.0 so it hasn't had all that much testing... regards, tom lane From pgsql-performance-owner@postgresql.org Sat Jul 23 21:41:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DDD1D52910 for ; Fri, 22 Jul 2005 16:29:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 80971-10 for ; Fri, 22 Jul 2005 19:29:16 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.50]) by svr1.postgresql.org (Postfix) with ESMTP id C9F4152917 for ; Fri, 22 Jul 2005 16:29:12 -0300 (ADT) Received: from 172.16.1.112 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Fri, 22 Jul 2005 15:28:46 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP02.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Fri, 22 Jul 2005 15:28:45 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Fri, 22 Jul 2005 15:28:45 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 22 Jul 2005 12:28:43 -0700 Subject: Re: [PATCHES] COPY FROM performance improvements From: "Luke Lonergan" To: "Joshua D. Drake" , "Patrick Welche" Cc: "Mark Wong" , pgsql-performance@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <42E128BD.2000104@commandprompt.com> MIME-Version: 1.0 X-OriginalArrivalTime: 22 Jul 2005 19:28:45.0806 (UTC) FILETIME=[93D0D0E0:01C58EF3] X-WSS-ID: 6EFF97671M810283011-02-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.574 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/364 X-Sequence-Number: 13605 Joshua, On 7/22/05 10:11 AM, "Joshua D. Drake" wrote: > The database server is a PE (Power Edge) 6600 > > Database Server IO: > > [root@master root]# /sbin/hdparm -tT /dev/sda > > /dev/sda: > Timing buffer-cache reads: 1888 MB in 2.00 seconds = 944.00 MB/sec > Timing buffered disk reads: 32 MB in 3.06 seconds = 10.46 MB/sec > > Second Database Server IO: > > [root@pq-slave root]# /sbin/hdparm -tT /dev/sda > > /dev/sda: > Timing buffer-cache reads: 1816 MB in 2.00 seconds = 908.00 MB/sec > Timing buffered disk reads: 26 MB in 3.11 seconds = 8.36 MB/sec > [root@pq-slave root]# Can you post the "time dd if=/dev/zero of=bigfile bs=8k count=500000" results? Also do the reverse (read the file) with "time dd if=bigfile of=/dev/null bs=8k". I think you are observing what we've known for a while, hardware RAID is horribly slow. We've not found a hardware RAID adapter of this class yet that shows reasonable read or write performance. The Adaptec 2400R or the LSI or others have terrible internal I/O compared to raw SCSI with software RAID, and even the CPU usage is higher on these cards while doing slower I/O than linux SW RAID. Notably - we've found that the 3Ware RAID controller does a better job than the low end SCSI RAID at HW RAID support, and also exports JBOD at high speeds. If you export JBOD on the low end SCSI RAID adapters, the performance is also very poor, though generally faster than using HW RAID. - Luke From pgsql-performance-owner@postgresql.org Fri Jul 22 16:47:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E9E00528FD for ; Fri, 22 Jul 2005 16:47:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 85752-08 for ; Fri, 22 Jul 2005 19:46:58 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id 7558A5291C for ; Fri, 22 Jul 2005 16:46:56 -0300 (ADT) Received: from ibm-b.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6MJkljA024095 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Fri, 22 Jul 2005 12:46:47 -0700 Message-Id: <200507221946.j6MJkljA024095@smtp.osdl.org> Date: Fri, 22 Jul 2005 12:47:18 -0700 From: Mark Wong To: "Luke Lonergan" Cc: "Andrew Dunstan" , "Alvaro Herrera" , "Bruce Momjian" , "Alon Goldshuv" , pgsql-performance@postgresql.org, maryedie@osdl.org Subject: Re: [PATCHES] COPY FROM performance improvements In-Reply-To: References: <200507212154.j6LLsdjA002847@smtp.osdl.org> Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/356 X-Sequence-Number: 13597 On a single spindle: $ time dd if=/dev/zero of=bigfile bs=8k count=2000000 2000000+0 records in 2000000+0 records out real 2m8.569s user 0m0.725s sys 0m19.633s None of my drives are partitioned big enough for me to create 2x RAM sized files on a single disk. I have 16MB RAM and only 36GB drives. But here are some number for my 12-disk lvm2 striped volume. $ time dd if=/dev/zero of=bigfile3 bs=8k count=4000000 4000000+0 records in 4000000+0 records out real 1m17.059s user 0m1.479s sys 0m41.293s Mark On Thu, 21 Jul 2005 16:14:47 -0700 "Luke Lonergan" wrote: > Cool! > > At what rate does your disk setup write sequential data, e.g.: > time dd if=/dev/zero of=bigfile bs=8k count=500000 > > (sized for 2x RAM on a system with 2GB) > > BTW - the Compaq smartarray controllers are pretty broken on Linux from a > performance standpoint in our experience. We've had disastrously bad > results from the SmartArray 5i and 6 controllers on kernels from 2.4 -> > 2.6.10, on the order of 20MB/s. > > For comparison, the results on our dual opteron with a single LSI SCSI > controller with software RAID0 on a 2.6.10 kernel: > > [llonergan@stinger4 dbfast]$ time dd if=/dev/zero of=bigfile bs=8k > count=500000 > 500000+0 records in > 500000+0 records out > > real 0m24.702s > user 0m0.077s > sys 0m8.794s > > Which calculates out to about 161MB/s. > > - Luke > > > On 7/21/05 2:55 PM, "Mark Wong" wrote: > > > I just ran through a few tests with the v14 patch against 100GB of data > > from dbt3 and found a 30% improvement; 3.6 hours vs 5.3 hours. Just to > > give a few details, I only loaded data and started a COPY in parallel > > for each the data files: > > http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/ > > > > Here's a visual of my disk layout, for those familiar with the database > > schema: > > http://www.testing.osdl.org/projects/dbt3testing/results/fast_copy/layout-dev4 > > -010-dbt3.html > > > > I have 6 arrays of fourteen 15k rpm drives in a split-bus configuration > > attached to a 4-way itanium2 via 6 compaq smartarray pci-x controllers. > > > > Let me know if you have any questions. > > > > Mark > > > From pgsql-performance-owner@postgresql.org Fri Jul 22 17:49:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2B44252924 for ; Fri, 22 Jul 2005 17:49:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 99695-04 for ; Fri, 22 Jul 2005 20:49:00 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.204]) by svr1.postgresql.org (Postfix) with ESMTP id E5E4C5291E for ; Fri, 22 Jul 2005 17:48:59 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 8so273449nzo for ; Fri, 22 Jul 2005 13:48:58 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=kma9qgwWfBz22gj+btHcjO2Au/c+gom5VXLMdJnlswp6HfIxDX6S2Ynpp4niOeHO+5Thx1YsIm6x9dG/+Zgt46QXmAKOEb68fcKYefT5tZMr4O5DbHThiImzhzIrL73DjSNf5CWZvPJ4YrUA/XkfAB8K0zzIPH36eFURsQbyQz8= Received: by 10.36.5.19 with SMTP id 19mr34086nze; Fri, 22 Jul 2005 13:48:58 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Fri, 22 Jul 2005 13:48:58 -0700 (PDT) Message-ID: <758d5e7f05072213486919c8f4@mail.gmail.com> Date: Fri, 22 Jul 2005 22:48:58 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Tom Lane Subject: Re: Planner doesn't look at LIMIT? Cc: pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org In-Reply-To: <6440.1122049220@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <758d5e7f0507220210bfa4978@mail.gmail.com> <29473.1122043197@sss.pgh.pa.us> <6440.1122049220@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.611 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/357 X-Sequence-Number: 13598 On 7/22/05, Tom Lane wrote: > > This is quite strange. The nestloop plan definitely should be preferre= d > > in the context of the LIMIT, considering that it has far lower estimate= d > > cost. And it is preferred in simple tests for me. >=20 > After a suitable period of contemplating my navel, I figured out > what is going on here: the total costs involved are large enough that > the still-fairly-high startup cost of the hash is disregarded by > compare_fuzzy_path_costs(), and so the nestloop is discarded as not > having any significant potential advantage in startup time. >=20 > I think that this refutes the original scheme of using the same fuzz > factor for both startup and total cost comparisons, and therefore > propose the attached patch. >=20 > Comments? Works great!!! With LIMIT below 4 000 000 rows (its 47-milion row table) it prefers nested loops, then it starts to introduce merge joins. Regards, Dawid From pgsql-performance-owner@postgresql.org Sat Jul 23 21:39:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5B4625283C for ; Sat, 23 Jul 2005 01:09:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95774-02 for ; Sat, 23 Jul 2005 04:09:30 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id AA4F45288F for ; Sat, 23 Jul 2005 01:09:28 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Sat, 23 Jul 2005 00:09:12 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Sat, 23 Jul 2005 00:09:12 -0400 Received: from 24.5.173.15 ([24.5.173.15]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Sat, 23 Jul 2005 00:09:11 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 22 Jul 2005 21:09:10 -0700 Subject: Re: [PATCHES] COPY FROM performance improvements From: "Luke Lonergan" To: "Mark Wong" Cc: "Andrew Dunstan" , "Alvaro Herrera" , "Bruce Momjian" , "Alon Goldshuv" , pgsql-performance@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <200507221946.j6MJkljA024095@smtp.osdl.org> MIME-Version: 1.0 X-OriginalArrivalTime: 23 Jul 2005 04:09:12.0025 (UTC) FILETIME=[4817B490:01C58F3C] X-WSS-ID: 6EFF1D621J410642345-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.599 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/363 X-Sequence-Number: 13604 Mark, On 7/22/05 12:47 PM, "Mark Wong" wrote: > On a single spindle: > > $ time dd if=/dev/zero of=bigfile bs=8k count=2000000 > 2000000+0 records in > 2000000+0 records out > > real 2m8.569s > user 0m0.725s > sys 0m19.633s This is super fast! 124MB/s seems too fast for true write performance on a single spindle. > But here are some number for my 12-disk lvm2 striped volume. So, software striping on how many controllers? > $ time dd if=/dev/zero of=bigfile3 bs=8k count=4000000 > 4000000+0 records in > 4000000+0 records out > > real 1m17.059s > user 0m1.479s > sys 0m41.293s Again - super fast at 416MB/s. How many controllers? When we had our problems with the cciss driver and the smartarray 5i/6 controllers, we found the only way to get any performance out of them was to run them in JBOD mode and software stripe. However, when we did so the CPU usage skyrocketed and the performance of simple SCSI adapters was 50% faster with less CPU consumption. These numbers show 100*(42.8/67) = 64% CPU consumption, I'd expect less with 2 simple U320 SCSI controllers. - Luke From pgsql-performance-owner@postgresql.org Sat Jul 23 07:42:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A38E652995; Sat, 23 Jul 2005 07:42:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 70068-03; Sat, 23 Jul 2005 10:42:36 +0000 (GMT) Received: from colo.samason.me.uk (unknown [69.55.228.22]) by svr1.postgresql.org (Postfix) with ESMTP id C2AEC52993; Sat, 23 Jul 2005 07:42:35 -0300 (ADT) Received: from colo.samason.me.uk (colo [69.55.228.22]) by colo.samason.me.uk (8.12.11/8.12.11) with ESMTP id j6NAgZR8011880; Sat, 23 Jul 2005 11:42:35 +0100 (BST) (envelope-from sam@colo.samason.me.uk) Received: (from sam@localhost) by colo.samason.me.uk (8.12.11/8.12.11/Submit) id j6NAgZEt011877; Sat, 23 Jul 2005 11:42:35 +0100 (BST) (envelope-from sam) Date: Sat, 23 Jul 2005 11:42:35 +0100 From: Sam Mason To: pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org Subject: Re: [HACKERS] Planner doesn't look at LIMIT? Message-ID: <20050723104235.GM62747@colo.samason.me.uk> Mail-Followup-To: Sam Mason , pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org References: <758d5e7f0507220210bfa4978@mail.gmail.com> <29473.1122043197@sss.pgh.pa.us> <6440.1122049220@sss.pgh.pa.us> <1122053518.21502.180.camel@localhost.localdomain> <7214.1122055577@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <7214.1122055577@sss.pgh.pa.us> User-Agent: Mutt/1.4.2.1i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/358 X-Sequence-Number: 13599 Tom Lane wrote: >Could be. I went back to look at Sam Mason's report about three weeks >ago, and it definitely seems to explain his issue. I've just built a patched version as well and it appears to be doing what I think is the right thing now. I.e. actually picking the plan with the lower cost. Thanks! Sam From pgsql-performance-owner@postgresql.org Sat Jul 23 13:50:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1C16852A04 for ; Sat, 23 Jul 2005 13:50:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 41793-04 for ; Sat, 23 Jul 2005 16:50:37 +0000 (GMT) Received: from calvin.surfutopia.net (calvin.surfutopia.net [67.120.245.34]) by svr1.postgresql.org (Postfix) with ESMTP id ACEFC529E7 for ; Sat, 23 Jul 2005 13:50:35 -0300 (ADT) Received: by calvin.surfutopia.net (Postfix, from userid 1000) id A3960F21E; Sat, 23 Jul 2005 09:50:34 -0700 (PDT) Date: Sat, 23 Jul 2005 09:50:34 -0700 From: John Mendenhall To: pgsql-performance list Subject: re: performance decrease after reboot Message-ID: <20050723165034.GA2700@calvin.surfutopia.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/359 X-Sequence-Number: 13600 pgsql performance gurus, I sent the following message earlier this week. I have continued attempting to find something on the net that would explain this strange change of query plans, but nothing seems to apply. Are there any thoughts, such as possibly tweaking the database somehow to see if I can get this to repeat consistently? Please let me know if any of you have any pointers as to the cause of the different query plans. Thank you very much in advance for any pointers you can provide. JohnM On Tue, 19 Jul 2005, John Mendenhall wrote: > I tuned a query last week to obtain acceptable performance. > Here is my recorded explain analyze results: > > LOG: duration: 826.505 ms statement: explain analyze > [cut for brevity] > > I rebooted the database machine later that night. > Now, when I run the same query, I get the following > results: > > LOG: duration: 6931.701 ms statement: explain analyze > [cut for brevity] I just ran my query again, no changes from yesterday and it is back to normal: LOG: duration: 795.839 ms statement: explain analyze What could have been the problem? The major differences in the query plan are as follows: (1) The one that runs faster uses a Hash Join at the very top of the query plan. It does a Hash Cond on the country and code fields. (2) The one that runs slower uses a Materialize with the subplan, with no Hash items. The Materialize does Seq Scan of the countries table, and above it, a Join Filter is run. (3) The partners_pkey index on the partners table is in a different place in the query. Does anyone know what would cause the query plan to be different like this, for the same server, same query? I run vacuum analyze every night. Is this perhaps the problem? What setting do I need to tweak to make sure the faster plan is always found? Thanks for any pointers in this dilemma. JohnM -- John Mendenhall john@surfutopia.net surf utopia internet services From pgsql-performance-owner@postgresql.org Sat Jul 23 14:03:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A412E5290F for ; Sat, 23 Jul 2005 14:03:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44674-01 for ; Sat, 23 Jul 2005 17:03:12 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id 2871B529E6 for ; Sat, 23 Jul 2005 14:03:11 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Sat, 23 Jul 2005 13:02:59 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Sat, 23 Jul 2005 13:02:59 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: re: performance decrease after reboot Date: Sat, 23 Jul 2005 13:02:58 -0400 Message-ID: <3E37B936B592014B978C4415F90D662D5B7198@MI8NYCMAIL06.Mi8.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] re: performance decrease after reboot Thread-Index: AcWPpyrHZrTbCH0yQC2BgTaSghP+IwAATXkZ From: "Luke Lonergan" To: "John Mendenhall" , "pgsql-performance list" X-OriginalArrivalTime: 23 Jul 2005 17:02:59.0397 (UTC) FILETIME=[60F67750:01C58FA8] X-WSS-ID: 6EFCA7C91J411005260-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/360 X-Sequence-Number: 13601 It's likely that data is in filesystem (not database) cache the second = time you run the query. See if the same thing happens when you stop and = restart the postmaster (it likely wont), then do something like this to = flush the filesystem cache (read a big file, can't give you a sample cmd = because my Treo has no equal sign :-) then run the query again. - Luke -----Original Message----- From: John Mendenhall [mailto:john@surfutopia.net] Sent: Sat Jul 23 12:54:18 2005 To: pgsql-performance list Subject: [PERFORM] re: performance decrease after reboot pgsql performance gurus, I sent the following message earlier this week. I have continued attempting to find something on the net that would explain this strange change of query plans, but nothing seems to apply. Are there any thoughts, such as possibly tweaking the database somehow to see if I can get this to repeat consistently? Please let me know if any of you have any pointers as to=20 the cause of the different query plans. Thank you very much in advance for any pointers you can provide. JohnM On Tue, 19 Jul 2005, John Mendenhall wrote: > I tuned a query last week to obtain acceptable performance. > Here is my recorded explain analyze results: > > LOG: duration: 826.505 ms statement: explain analyze > [cut for brevity] >=20 > I rebooted the database machine later that night. > Now, when I run the same query, I get the following > results: >=20 > LOG: duration: 6931.701 ms statement: explain analyze > [cut for brevity] I just ran my query again, no changes from yesterday and it is back to normal: LOG: duration: 795.839 ms statement: explain analyze What could have been the problem? The major differences in the query plan are as follows: (1) The one that runs faster uses a Hash Join at the very top of the query plan. It does a Hash Cond on the country and code fields. (2) The one that runs slower uses a Materialize with the subplan, with no Hash items. The Materialize does Seq Scan of the countries table, and above it, a Join Filter is run. (3) The partners_pkey index on the partners table is in a different place in the query. Does anyone know what would cause the query plan to be different like this, for the same server, same query? I run vacuum analyze every night. Is this perhaps the problem? What setting do I need to tweak to make sure the faster plan is always found? Thanks for any pointers in this dilemma. JohnM --=20 John Mendenhall john@surfutopia.net surf utopia internet services ---------------------------(end of broadcast)--------------------------- TIP 6: explain analyze is your friend From pgsql-performance-owner@postgresql.org Sat Jul 23 17:37:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6C78F5290F for ; Sat, 23 Jul 2005 17:37:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04247-02 for ; Sat, 23 Jul 2005 20:36:55 +0000 (GMT) Received: from rambo.iniquinet.com (rambo.iniquinet.com [69.39.89.10]) by svr1.postgresql.org (Postfix) with ESMTP id E865252B12 for ; Sat, 23 Jul 2005 17:36:53 -0300 (ADT) Received: (qmail 9039 invoked by uid 1010); 23 Jul 2005 16:36:54 -0400 Received: from Robert_Creager@LogicalChaos.org by rambo.iniquinet.com by uid 1002 with qmail-scanner-1.20 (clamscan: 0.70. spamassassin: 3.0.3. Clear:RC:0(63.147.78.131):SA:0(?/?):. Processed in 1.281487 secs); 23 Jul 2005 20:36:54 -0000 Received: from unknown (HELO thunder.logicalchaos.org) (perl?test@logicalchaos.org@63.147.78.131) by rambo.iniquinet.com with AES256-SHA encrypted SMTP; 23 Jul 2005 16:36:53 -0400 Received: from logicalchaos.org (thunder.logicalchaos.org [192.168.0.250]) by thunder.logicalchaos.org (Postfix) with ESMTP id A6774611E8; Sat, 23 Jul 2005 14:36:50 -0600 (MDT) Date: Sat, 23 Jul 2005 14:36:50 -0600 From: Robert Creager Cc: Robert Creager , Tom Lane , pgsql-performance@postgresql.org, Andy Hewitt III Subject: Re: Huge performance problem between 7.4.1 and 8.0.3 - CS Message-ID: <20050723143650.5678061a@thunder.logicalchaos.org> In-Reply-To: <20050719224908.15415222@thunder.logicalchaos.org> References: <20050718114952.00004ac5@C118181.stortek.com> <15472.1121709173@sss.pgh.pa.us> <20050719105014.00007635@C118181.stortek.com> <7322.1121792062@sss.pgh.pa.us> <20050719120951.00002520@C118181.stortek.com> <20050719224908.15415222@thunder.logicalchaos.org> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed-Claws 1.0.3 (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; boundary="Signature_Sat__23_Jul_2005_14_36_50_-0600_4Cp=mBFyqM7DO1VE"; protocol="application/pgp-signature"; micalg=pgp-sha1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.001 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/361 X-Sequence-Number: 13602 --Signature_Sat__23_Jul_2005_14_36_50_-0600_4Cp=mBFyqM7DO1VE Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable I've now backed off to version 7.4.1, which doesn't exhibit the problems th= at 8.0.3 does. I guess I'll wait 'till the next version and see if any progre= ss has occurred. Rob When grilled further on (Tue, 19 Jul 2005 22:49:08 -0600), Robert Creager confessed: > When grilled further on (Tue, 19 Jul 2005 12:09:51 -0600), > Robert Creager confessed: >=20 > > On Tue, 19 Jul 2005 12:54:22 -0400 > > Tom Lane wrote: > >=20 > > > Hmm, I hadn't thought about the possible impact of multiple concurrent > > > vacuums. Is the problem caused by that, or has performance already g= one > > > into the tank by the time the cron-driven vacuums are taking long eno= ugh > > > to overlap? > >=20 > >=20 > > I'll re-start the database, vacuum full analyze and restart the runs wi= thout > the > > cron vacuum running. > >=20 >=20 > It took a few hours, but the problem did finally occur with no vacuum run= ning on > 803. CS is averaging 72k. I cannot quantitatively say it took longer to > reproduce than with the vacuums running, but it seemed like it did. >=20 > Can any information be gotten out of this? Should I try CVS HEAD? >=20 > Thoughts? >=20 > Thanks, > Rob >=20 > --=20 > 22:41:36 up 6 days, 2:16, 6 users, load average: 0.15, 0.21, 0.30 > Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --=20 14:35:32 up 9 days, 18:10, 5 users, load average: 2.17, 2.19, 2.15 Linux 2.6.5-02 #8 SMP Mon Jul 12 21:34:44 MDT 2004 --Signature_Sat__23_Jul_2005_14_36_50_-0600_4Cp=mBFyqM7DO1VE Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkLiqmIACgkQLQ/DKuwDYzksAQCdGNv/wST8EwGMlSgbUY0jWmrB GVgAoI/OWLXYULkLGMvKLzlB1+CeKSNB =Y7SY -----END PGP SIGNATURE----- --Signature_Sat__23_Jul_2005_14_36_50_-0600_4Cp=mBFyqM7DO1VE-- From pgsql-performance-owner@postgresql.org Mon Jul 25 05:26:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DE79852BF0 for ; Mon, 25 Jul 2005 05:26:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20207-01 for ; Mon, 25 Jul 2005 08:26:45 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 22134529B1 for ; Mon, 25 Jul 2005 05:26:43 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id DBED0644170 for ; Mon, 25 Jul 2005 02:26:16 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 12292-01 for ; Mon, 25 Jul 2005 02:26:14 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id C7A71644168 for ; Mon, 25 Jul 2005 02:26:14 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) Content-Transfer-Encoding: 7bit Message-Id: <1F0E51A1-C682-4CE1-95F5-BE50032D7122@drivefaster.net> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: Dan Harris Subject: Coraid/AoE device experience? Date: Mon, 25 Jul 2005 02:27:31 -0600 X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/366 X-Sequence-Number: 13607 Lately, I've been reading a lot about these new Coraid AoE RAID devices ( http://www.coraid.com ). They tout it as being fast and cheap and better than iSCSI due to the lack of TCP/IP over the wire. Is it likely that a 15-drive RAID 10 Linux software RAID would outperform a 4-drive 10k SCSI RAID 0+1 for a heavy-loaded database? If not software RAID, how about their dedicated RAID controller blade? I'm definitely IO bound right now and starving for spindles. Does this make sense or is it too good to be true? Thanks -Dan From pgsql-performance-owner@postgresql.org Mon Jul 25 10:16:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5887852C4C for ; Mon, 25 Jul 2005 10:15:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71774-03 for ; Mon, 25 Jul 2005 13:15:39 +0000 (GMT) Received: from mail.metronet.co.uk (mail.metronet.co.uk [213.162.97.75]) by svr1.postgresql.org (Postfix) with ESMTP id C38D352C45 for ; Mon, 25 Jul 2005 10:15:37 -0300 (ADT) Received: from mainbox.archonet.com (84-51-143-99.archon037.adsl.metronet.co.uk [84.51.143.99]) by smtp.metronet.co.uk (MetroNet Mail) with ESMTP id C4BFE40DC7A; Mon, 25 Jul 2005 14:15:30 +0100 (BST) Received: from localhost (localhost.localdomain [127.0.0.1]) by mainbox.archonet.com (Postfix) with ESMTP id 580C715EDA; Mon, 25 Jul 2005 14:04:04 +0100 (BST) Received: from mainbox.archonet.com ([127.0.0.1]) by localhost (mainbox [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 06609-05; Mon, 25 Jul 2005 14:04:01 +0100 (BST) Received: from [192.168.1.17] (client17.office.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 744E015ED9; Mon, 25 Jul 2005 14:04:01 +0100 (BST) Message-ID: <42E4E341.4050807@archonet.com> Date: Mon, 25 Jul 2005 14:04:01 +0100 From: Richard Huxton User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Shashi Kanth Boddula Cc: pgsql-performance@postgresql.org Subject: Re: Mirroring PostgreSQL database References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/368 X-Sequence-Number: 13609 Shashi Kanth Boddula wrote: > > The customer is using DBmirror tool to mirror the database records of > primary to secondary . The customer is complaining that there is one day > (24 hours) delay between primary and secondray for database > synchronization . They have dedicated line and bandwidth , but still the > problems exists. You don't say what the nature of the problem with dbmirror is. Are they saturating their bandwidth? Are one or both servers unable to keep pace with the updates? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Mon Jul 25 12:20:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E87E952AD2 for ; Mon, 25 Jul 2005 12:20:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07901-05 for ; Mon, 25 Jul 2005 15:20:33 +0000 (GMT) Received: from vscan02.westnet.com.au (vscan02.westnet.com.au [203.10.1.132]) by svr1.postgresql.org (Postfix) with ESMTP id B175152AA0 for ; Mon, 25 Jul 2005 12:20:31 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by localhost (Postfix) with ESMTP id 115E111F2B2; Mon, 25 Jul 2005 23:20:30 +0800 (WST) Received: from vscan02.westnet.com.au ([127.0.0.1]) by localhost (vscan02.westnet.com.au [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 01494-02; Mon, 25 Jul 2005 23:20:30 +0800 (WST) Received: from [202.72.133.22] (dsl-202-72-133-22.wa.westnet.com.au [202.72.133.22]) by vscan02.westnet.com.au (Postfix) with ESMTP id 8B4AD11F58A; Mon, 25 Jul 2005 23:20:29 +0800 (WST) Message-ID: <42E5033F.7090201@familyhealth.com.au> Date: Mon, 25 Jul 2005 23:20:31 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.6 (Windows/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Shashi Kanth Boddula Cc: pgsql-performance@postgresql.org Subject: Re: Mirroring PostgreSQL database References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.247 tagged_above=0 required=5 tests=AWL, INFO_TLD X-Spam-Level: X-Archive-Number: 200507/369 X-Sequence-Number: 13610 Try Slony: www.slony.info Shashi Kanth Boddula wrote: > > Hi, > I have one customer who is using PostgreSQL 7.4.8 on Linux . He has some > problems with database mirroring . The details are follows. > The customer is using Linux on which PostgreSQL 7.4.8 along with Jboss > 3.2.3 is running . He has 2 servers , one is acting as a live server > (primary) and another is acting as a fail-over (secondary) server > . Secondary server is placed in remote location . These servers are > acting as a Attendence server for daily activities . Nearly 50,000 > employees depend on the live server . > > The customer is using DBmirror tool to mirror the database records of > primary to secondary . The customer is complaining that there is one day > (24 hours) delay between primary and secondray for database > synchronization . They have dedicated line and bandwidth , but still the > problems exists. > > I just want to know , for immediate data mirroring , what is the best > way for PostgreSQL . PostgreSQL is offering many mirror tools , but > which one is the best ?. Is there any other way to accomplish the task ? > > Thank you . Waiting for your reply. > > > Thanks & Regards, > Shashi Kanth > Consultant - Linux > RHCE , LPIC-2 > Onward Novell - Bangalore > 9886455567 > > From pgsql-performance-owner@postgresql.org Mon Jul 25 13:03:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7D83C52ABC for ; Mon, 25 Jul 2005 13:03:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 14248-09 for ; Mon, 25 Jul 2005 16:03:41 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id D16B452BA3 for ; Mon, 25 Jul 2005 13:03:38 -0300 (ADT) Received: from [192.168.1.52] (fc1smp [66.93.38.87]) (authenticated bits=0) by hosting.commandprompt.com (8.12.8/8.12.8) with ESMTP id j6PG3Hhr025113; Mon, 25 Jul 2005 09:03:17 -0700 Message-ID: <42E50D59.8020500@commandprompt.com> Date: Mon, 25 Jul 2005 09:03:37 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.6-1.1.fc3 (X11/20050720) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Shashi Kanth Boddula Cc: pgsql-performance@postgresql.org Subject: Re: Mirroring PostgreSQL database References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.254 tagged_above=0 required=5 tests=AWL, INFO_TLD X-Spam-Level: X-Archive-Number: 200507/370 X-Sequence-Number: 13611 > I just want to know , for immediate data mirroring , what is the best > way for PostgreSQL . PostgreSQL is offering many mirror tools , but > which one is the best ?. Is there any other way to accomplish the task ? You want to take a look at Slony-I or Mammoth Replicator. http://www.slony.info/ http://www.commandprompt.com/ > > Thank you . Waiting for your reply. > > > Thanks & Regards, > Shashi Kanth > Consultant - Linux > RHCE , LPIC-2 > Onward Novell - Bangalore > 9886455567 > > -- Your PostgreSQL solutions company - Command Prompt, Inc. 1.800.492.2240 PostgreSQL Replication, Consulting, Custom Programming, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/ From pgsql-performance-owner@postgresql.org Mon Jul 25 09:37:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2D10F52B68 for ; Mon, 25 Jul 2005 09:37:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 63237-03 for ; Mon, 25 Jul 2005 12:36:59 +0000 (GMT) Received: from lucius.provo.novell.com (lucius.provo.novell.com [137.65.81.172]) by svr1.postgresql.org (Postfix) with ESMTP id 9F9E552AEE for ; Mon, 25 Jul 2005 09:36:57 -0300 (ADT) Received: from INET-PRV1-MTA by lucius.provo.novell.com with Novell_GroupWise; Mon, 25 Jul 2005 06:37:00 -0600 Message-Id: X-Mailer: Novell GroupWise Internet Agent 6.5.4 Beta Date: Mon, 25 Jul 2005 12:01:52 -0600 From: "Shashi Kanth Boddula" To: Subject: Mirroring PostgreSQL database Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="=__PartE6C54C00.0__=" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.622 tagged_above=0 required=5 tests=AWL, DATE_IN_FUTURE_03_06, DNS_FROM_RFC_ABUSE, HTML_40_50, HTML_FONT_BIG, HTML_MESSAGE X-Spam-Level: X-Archive-Number: 200507/367 X-Sequence-Number: 13608 This is a MIME message. If you are reading this text, you may want to consider changing to a mail reader or gateway that understands how to properly handle MIME multipart messages. --=__PartE6C54C00.0__= Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Hi, I have one customer who is using PostgreSQL 7.4.8 on Linux . He has some problems with database mirroring . The details are follows. The customer is using Linux on which PostgreSQL 7.4.8 along with Jboss 3.2.3 is running . He has 2 servers , one is acting as a live server (primary) and another is acting as a fail-over (secondary) server . Secondary server is placed in remote location . These servers are acting as a Attendence server for daily activities . Nearly 50,000 employees depend on the live server . The customer is using DBmirror tool to mirror the database records of primary to secondary . The customer is complaining that there is one day (24 hours) delay between primary and secondray for database synchronization . They have dedicated line and bandwidth , but still the problems exists. I just want to know , for immediate data mirroring , what is the best way for PostgreSQL . PostgreSQL is offering many mirror tools , but which one is the best ?. Is there any other way to accomplish the task ? Thank you . Waiting for your reply. Thanks & Regards, Shashi Kanth Consultant - Linux RHCE , LPIC-2 Onward Novell - Bangalore 9886455567 --=__PartE6C54C00.0__= Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable =20
Hi,
I have one customer who is using = PostgreSQL 7.4.8 on Linux . He has some problems with database mirroring . = The details are follows.
The customer is using Linux on which PostgreSQL 7.4.8 along = with Jboss 3.2.3 is running . He has 2 servers , one is acting as a live = server (primary) and another is acting as a fail-over (secondary) = 0;server .  Secondary server is placed in remote location . = These servers are acting as a Attendence server for daily activities . = Nearly 50,000 employees depend on the live server .
 
The customer is using DBmirror tool to mirror the database = records of primary to secondary . The customer is complaining that there = is one day (24 hours) delay between primary and secondray for database = synchronization . They have dedicated line and bandwidth , but still the = problems exists.
 
I just want to know , for immediate data mirroring , what = is the best way for PostgreSQL . PostgreSQL is offering many mirror tools = , but which one is the best ?. Is there any other way to accomplish the = task ?
 
Thank you . Waiting for your reply.
 

Thanks & Regards,
Shashi Kanth
Consultant - Linux
RHCE , = LPIC-2
Onward Novell - Bangalore
9886455567


--=__PartE6C54C00.0__=-- From pgsql-performance-owner@postgresql.org Mon Jul 25 18:31:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9DD0C52CB9 for ; Mon, 25 Jul 2005 18:30:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 97657-10 for ; Mon, 25 Jul 2005 21:30:41 +0000 (GMT) Received: from wsmailap01.firstam.com (outbound-smtp01.firstam.com [208.246.101.197]) by svr1.postgresql.org (Postfix) with ESMTP id 0CF0052C4E for ; Mon, 25 Jul 2005 18:30:40 -0300 (ADT) Received: from 172.21.131.6 by wsmailap01.firstam.com with ESMTP ( Tumbleweed MMS SMTP Relay (MMS v5.6.3)); Mon, 25 Jul 2005 14:30:15 -0700 X-Server-Uuid: 2A3A6F04-3324-429D-9DE4-35238E3EA19C Received: from anammx01.ana.firstamdata.com ([192.168.173.35]) by famarp02.firstam.com (MOS 3.5.6-GR) with SMTP id JHG71200; Mon, 25 Jul 2005 14:30:25 -0700 (PDT) Received: from pisgana01sxch21.ana.firstamdata.com (Not Verified[192.168.173.71]) by mailgateway.firstam.com (post.office MTA v5.0 0924 ) with ESMTP id ; Mon, 25 Jul 2005 14:30:24 -0700 Received: from pisgana01sxch01.ana.firstamdata.com ([192.168.173.70]) by pisgana01sxch21.ana.firstamdata.com with Microsoft SMTPSVC(6.0.3790.211 ); Mon, 25 Jul 2005 14:30:24 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: "Vacuum Full Analyze" taking so long Date: Mon, 25 Jul 2005 14:30:24 -0700 Message-ID: Thread-Topic: "Vacuum Full Analyze" taking so long Thread-Index: AcWRM43r8tyNyV7ORnakp9LzUlpbiwAKySfA From: "Tomeh, Husam" To: pgsql-performance@postgresql.org X-OriginalArrivalTime: 25 Jul 2005 21:30:24.0503 (UTC) FILETIME=[116AD870:01C59160] X-WSS-ID: 6EFB866D0ZK914752-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.029 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/371 X-Sequence-Number: 13612 =20 I have an 8.02 postgresql database with about 180 GB in size, running on 2.6 RedHat kernel with 32 GB of RAM and 2 CPUs. I'm running the vacuum full analyze command, and has been running for at least two consecutive days with no other processes running (it's an offline loading server). I tweaked the maintenanace_mem to its max (2 GB) with work_mem of 8M. I have no issues with my checkpoints. I can still I/O activities against the physical files of the "property" table and its two indexes (primary key and r index). The property files are about 128GB and indexes are about 15 GB. I have run the same maintenance job on a different box (staging) with identical hardware config (except with 64 GB instead of 32) and took less than 12 hours. Any clue or tip is really appreciated.=20 Also read a comment by Tom Lane, that terminating the process should be crash-safe if I had to.=20 Thanks, --=20 =20Husam=20 ---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings ********************************************************************** This message contains confidential information intended only for the=20 use of the addressee(s) named above and may contain information that=20 is legally privileged. If you are not the addressee, or the person=20 responsible for delivering it to the addressee, you are hereby=20 notified that reading, disseminating, distributing or copying this=20 message is strictly prohibited. If you have received this message by=20 mistake, please immediately notify us by replying to the message and=20 delete the original message immediately thereafter. Thank you. FADLD Tag ********************************************************************** From pgsql-performance-owner@postgresql.org Mon Jul 25 20:37:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9958D52EC6 for ; Mon, 25 Jul 2005 19:19:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 10905-10 for ; Mon, 25 Jul 2005 22:19:20 +0000 (GMT) Received: from vms040pub.verizon.net (vms040pub.verizon.net [206.46.252.40]) by svr1.postgresql.org (Postfix) with ESMTP id DDD9A52EB2 for ; Mon, 25 Jul 2005 19:19:16 -0300 (ADT) Received: from osgiliath.mathom.us ([70.108.53.154]) by vms040.mailsrvcs.net (Sun Java System Messaging Server 6.2 HotFix 0.04 (built Dec 24 2004)) with ESMTPA id <0IK700KPAFC6CIB3@vms040.mailsrvcs.net> for pgsql-performance@postgresql.org; Mon, 25 Jul 2005 17:19:18 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by osgiliath.mathom.us (Postfix) with ESMTP id 8312864C998; Mon, 25 Jul 2005 17:30:58 -0400 (EDT) Received: from osgiliath.mathom.us ([127.0.0.1]) by localhost (osgiliath [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 17223-06; Mon, 25 Jul 2005 17:30:58 -0400 (EDT) Received: by osgiliath.mathom.us (Postfix, from userid 1000) id 605C960E567; Mon, 25 Jul 2005 17:30:58 -0400 (EDT) Date: Mon, 25 Jul 2005 17:30:58 -0400 From: Michael Stone Subject: Re: "Vacuum Full Analyze" taking so long In-reply-to: To: "Tomeh, Husam" Cc: pgsql-performance@postgresql.org Mail-Followup-To: "Tomeh, Husam" , pgsql-performance@postgresql.org Message-id: <20050725213058.GQ19080@mathom.us> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-disposition: inline X-Pgp-Fingerprint: 53 FF 38 00 E7 DD 0A 9C 84 52 84 C5 EE DF 7C 88 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mathom.us References: User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.008 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/375 X-Sequence-Number: 13616 I'd say, "don't do that". Unless you've deleted a lot of stuff and are expecting the DB to shrink, a full vacuum shouldn't really be needed. On a DB that big a full vacuum is just going to take a long time. If you really are shrinking, consider structuring things so you can just drop a table instead of vacuuming it (the drop is fairly instantaneous). If you can't do that, consider dropping the indices, vacuuming, and recreating the indices. Mike Stone From pgsql-performance-owner@postgresql.org Mon Jul 25 19:39:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AD47E52B0F for ; Mon, 25 Jul 2005 19:32:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 12486-10 for ; Mon, 25 Jul 2005 22:32:49 +0000 (GMT) Received: from tbmail.tradebot.com (Tradebot-Systems-1096753.cust-rtr.swbell.net [68.90.170.2]) by svr1.postgresql.org (Postfix) with ESMTP id 09D4452A2F for ; Mon, 25 Jul 2005 19:32:48 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C59168.CA5AE400" Subject: COPY insert performance X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Date: Mon, 25 Jul 2005 17:32:50 -0500 Message-ID: <07774C6E31D94A44A2A60E2085944F09071DF0@tbmail.tradebot.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: COPY insert performance Thread-Index: AcWRaMpXtxbNujNETa+mMZ0MXi+QSg== From: "Chris Isaacson" To: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.837 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, HTML_50_60, HTML_MESSAGE, HTML_TEXT_AFTER_BODY, HTML_TEXT_AFTER_HTML X-Spam-Level: * X-Archive-Number: 200507/372 X-Sequence-Number: 13613 This is a multi-part message in MIME format. ------_=_NextPart_001_01C59168.CA5AE400 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable I need COPY via libpqxx to insert millions of rows into two tables. One table has roughly have as many rows and requires half the storage. In production, the largest table will grow by ~30M rows/day. To test the COPY performance I split my transactions into 10,000 rows. I insert roughly 5000 rows into table A for every 10,000 rows into table B. =20 Table A has one unique index: =20 "order_main_pk" UNIQUE, btree (cl_ord_id) =20 Table B has 1 unique index and 2 non-unique indexes: =20 "order_transition_pk" UNIQUE, btree (collating_seq) "order_transition_ak2" btree (orig_cl_ord_id) "order_transition_ak3" btree (exec_id) =20 My testing environment is as follows: -Postgresql 8.0.1 -libpqxx 2.5.0 -Linux 2.6.11.4-21.7-smp x86_64=20 -Dual Opteron 246 -System disk (postgres data resides on this SCSI disk) - Seagate (ST373453LC) - 15K, 73 GB (http://www.seagate.com/cda/products/discsales/marketing/detail/0,1081,5 49,00.html) -2nd logical disk - 10K, 36GB IBM SCSI (IC35L036UCDY10-0) - WAL reside on this disk -NO RAID =20 PostgreSQL Here are the results of copying in 10M rows as fast as possible: (10K/transaction) Total Time: 1129.556 s Rows/sec: 9899.922 Transaction>1.2s 225 Transaction>1.5s 77 Transaction>2.0s 4 Max Transaction 2.325s =20 MySQL I ran a similar test with MySQL 4.1.10a (InnoDB) which produced these results: (I used MySQL's INSERT INTO x VALUES (1,2,3)(4,5,6)(...,...,...) syntax) (10K/transaction) Total Time: 860.000 s Rows/sec: 11627.91 Transaction>1.2s 0 Transaction>1.5s 0 Transaction>2.0s 0 Max Transaction 1.175s =20 Considering the configurations shown below, can anyone offer advice to close the 15% gap and the much worse variability I'm experiencing. Thanks =20 My postgresql.conf has the following non-default values: # ----------------------------- # PostgreSQL configuration file # ----------------------------- listen_addresses =3D '*' # what IP interface(s) to listen on;=20 max_connections =3D 100 #----------------------------------------------------------------------- ---- # RESOURCE USAGE (except WAL) #----------------------------------------------------------------------- ---- shared_buffers =3D 65536 # min 16, at least max_connections*2, 8KB each work_mem =3D 2048 # min 64, size in KB maintenance_work_mem =3D 204800 # min 1024, size in KB max_fsm_pages =3D 2250000 # min max_fsm_relations*16, 6 bytes each bgwriter_delay =3D 200 # 10-10000 milliseconds between rounds bgwriter_percent =3D 10 # 0-100% of dirty buffers in each round bgwriter_maxpages =3D 1000 # 0-1000 buffers max per round #----------------------------------------------------------------------- ---- # WRITE AHEAD LOG #----------------------------------------------------------------------- ---- fsync =3D false # turns forced synchronization on or off wal_buffers =3D 64 # min 4, 8KB each checkpoint_segments =3D 40 # in logfile segments, min 1, 16MB each checkpoint_timeout =3D 600 # range 30-3600, in seconds #----------------------------------------------------------------------- ---- # QUERY TUNING #----------------------------------------------------------------------- ---- effective_cache_size =3D 65536 # typically 8KB each random_page_cost =3D 2 # units are one sequential page fetch cost #----------------------------------------------------------------------- ---- # ERROR REPORTING AND LOGGING #----------------------------------------------------------------------- ---- =20 log_min_duration_statement =3D 250 # -1 is disabled, in milliseconds. log_connections =3D true log_disconnections =3D true log_duration =3D true log_line_prefix =3D '<%r%u%p%t%d%%' # e.g. '<%u%%%d> '=20 # %u=3Duser name %d=3Ddatabase name # %r=3Dremote host and port # %p=3DPID %t=3Dtimestamp %i=3Dcommand tag # %c=3Dsession id %l=3Dsession line number # %s=3Dsession start timestamp %x=3Dtransaction id # %q=3Dstop here in non-session processes # %%=3D'%' log_statement =3D 'none' # none, mod, ddl, all #----------------------------------------------------------------------- ---- # RUNTIME STATISTICS #----------------------------------------------------------------------- ---- # - Query/Index Statistics Collector - stats_start_collector =3D true stats_command_string =3D true stats_block_level =3D true stats_row_level =3D true stats_reset_on_server_start =3D true =20 My MySQL my.ini has the following non default values: innodb_data_home_dir =3D /var/lib/mysql/ innodb_data_file_path =3D ibdata1:10M:autoextend innodb_log_group_home_dir =3D /var/lib/mysql/ innodb_log_arch_dir =3D /var/lib/mysql/ # You can set .._buffer_pool_size up to 50 - 80 % # of RAM but beware of setting memory usage too high innodb_buffer_pool_size =3D 512M innodb_additional_mem_pool_size =3D 64M # Set .._log_file_size to 25 % of buffer pool size innodb_log_file_size =3D 128M innodb_log_buffer_size =3D 64M innodb_flush_log_at_trx_commit =3D 1 innodb_lock_wait_timeout =3D 50 innodb_flush_method =3D O_DSYNC max_allowed_packet =3D 16M =20 =20 =20 ------_=_NextPart_001_01C59168.CA5AE400 Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Message
I = need COPY via=20 libpqxx to insert millions of rows into two tables.  One table has = roughly=20 have as many rows and requires half the storage.  In production, = the=20 largest table will grow by ~30M rows/day.  To test the COPY = performance I=20 split my transactions into 10,000 rows.  I insert roughly 5000 = rows=20 into table A for every 10,000 rows into table = B.
 
Table = A has one=20 unique index:
 
"order_main_pk"=20 UNIQUE, btree (cl_ord_id)
 
Table = B has 1 unique=20 index and 2 non-unique indexes:
 
"order_transition_pk" UNIQUE, btree=20 (collating_seq)
"order_transition_ak2" btree=20 (orig_cl_ord_id)
"order_transition_ak3" btree = (exec_id)
 
My = testing=20 environment is as follows:
-Postgresql=20 8.0.1
-libpqxx=20 2.5.0
-Linux = 2.6.11.4-21.7-smp x86_64=20
-Dual = Opteron=20 246
-System disk=20 (postgres data resides on this SCSI disk) -  Seagate (ST373453LC) - 15K, 73 GB (http://www.seagate.com/cda/products/discsales/marketing= /detail/0,1081,549,00.html)
-2nd logical disk - 10K, = 36GB IBM SCSI=20 (IC35L036UCDY10-0) - WAL reside on this disk
-NO RAID
 
PostgreSQL
Here = are the results=20 of copying in 10M rows as fast as possible:=20 (10K/transaction)
Total=20 Time:            = 1129.556=20 s
Rows/sec:      &= nbsp;     =20 9899.922
Transaction>1.2s   =20 225
Transaction>1.5s    =20 77
Transaction>2.0s    &nb= sp;=20 4
Max=20 Transaction       2.325s=
 
MySQL<= /FONT>
I ran a similar = test with MySQL=20 4.1.10a (InnoDB) which produced these results: (I used MySQL's = INSERT=20 INTO x VALUES (1,2,3)(4,5,6)(...,...,...) syntax)=20 (10K/transaction)
Total=20 Time:         860.000=20 s
Rows/sec:      &= nbsp; 11627.91
Transaction>1.2s    &nb= sp;=20 0
Transaction>1.5s    &nb= sp;=20 0
Transaction>2.0s    &nb= sp;=20 0
Max=20 Transaction       1.175s=
 
Considering the=20 configurations shown below, can anyone offer advice to close the 15% gap = and the=20 much worse variability I'm experiencing. =20 Thanks
 
My=20 postgresql.conf has the following non-default=20 values:
#=20 -----------------------------
# PostgreSQL configuration file
#=20 -----------------------------
listen_addresses =3D '*' # what IP = interface(s) to listen on;
max_connections =3D = 100
#---------------------------------------------= ------------------------------
#=20 RESOURCE USAGE (except=20 WAL)
#----------------------------------------------------------------= -----------
shared_buffers=20 =3D 65536  # min 16, at least max_connections*2, 8KB = each
work_mem =3D=20 2048   # min 64, size in KB
maintenance_work_mem =3D=20 204800 # min 1024, size in KB
max_fsm_pages =3D = 2250000  # min=20 max_fsm_relations*16, 6 bytes each
bgwriter_delay =3D = 200  # 10-10000=20 milliseconds between rounds
bgwriter_percent =3D 10  # = 0-100% of=20 dirty buffers in each round
bgwriter_maxpages =3D 1000 # 0-1000 = buffers=20 max per round
#---------------------------------------------= ------------------------------
#=20 WRITE AHEAD=20 LOG
#-----------------------------------------------------------------= ----------
fsync=20 =3D false   # turns forced synchronization on or = off
wal_buffers=20 =3D 64  # min 4, 8KB each
checkpoint_segments =3D 40 # in logfile = segments,=20 min 1, 16MB each
checkpoint_timeout =3D 600 # range 30-3600, in=20 seconds
#---------------------------------------------= ------------------------------
#=20 QUERY=20 TUNING
#--------------------------------------------------------------= -------------
effective_cache_size=20 =3D 65536 # typically 8KB each
random_page_cost =3D = 2  # units are=20 one sequential page fetch cost
#---------------------------------------------= ------------------------------
#=20 ERROR REPORTING AND=20 LOGGING
#-------------------------------------------------------------= --------------  =20
log_min_duration_statement =3D    250 # -1 is = disabled, in=20 milliseconds.
log_connections =3D=20 true
log_disconnections =3D true
log_duration =3D = true
log_line_prefix =3D=20 '<%r%u%p%t%d%%'  # e.g. '<%u%%%d> '=20
    # %u=3Duser name %d=3Ddatabase=20 name
    # %r=3Dremote host and=20 port
    # %p=3DPID %t=3Dtimestamp %i=3Dcommand=20 tag
    # %c=3Dsession id %l=3Dsession line=20 number
    # %s=3Dsession start timestamp = %x=3Dtransaction=20 id
    # %q=3Dstop here in non-session=20 processes
    # %%=3D'%'
log_statement =3D=20 'none'  # none, mod, ddl, all
#---------------------------------------------= ------------------------------
#=20 RUNTIME=20 STATISTICS
#----------------------------------------------------------= -----------------
#=20 - Query/Index Statistics Collector -
stats_start_collector =3D=20 true
stats_command_string =3D true
stats_block_level =3D=20 true
stats_row_level =3D true
stats_reset_on_server_start =3D=20 true
 
My = MySQL=20 my.ini has the following non default=20 values:
innodb_data_home_dir =3D=20 /var/lib/mysql/
innodb_data_file_path =3D=20 ibdata1:10M:autoextend
innodb_log_group_home_dir =3D=20 /var/lib/mysql/
innodb_log_arch_dir =3D /var/lib/mysql/
# You can = set=20 .._buffer_pool_size up to 50 - 80 %
# of RAM but beware of setting = memory=20 usage too high
innodb_buffer_pool_size =3D=20 512M
innodb_additional_mem_pool_size =3D 64M
# Set = .._log_file_size to 25 %=20 of buffer pool size
innodb_log_file_size =3D = 128M
innodb_log_buffer_size =3D=20 64M
innodb_flush_log_at_trx_commit =3D 1
innodb_lock_wait_timeout = =3D=20 50
innodb_flush_method =3D O_DSYNC
max_allowed_packet =3D = 16M
 
 
 
=00 ------_=_NextPart_001_01C59168.CA5AE400-- From pgsql-performance-owner@postgresql.org Mon Jul 25 21:10:05 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6EE1E52C52 for ; Mon, 25 Jul 2005 19:49:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 17584-01 for ; Mon, 25 Jul 2005 22:48:49 +0000 (GMT) Received: from mail.Mi8.com (d01gw01.mi8.com [63.240.6.47]) by svr1.postgresql.org (Postfix) with ESMTP id DA85652C4F for ; Mon, 25 Jul 2005 19:48:47 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D1)); Mon, 25 Jul 2005 18:48:42 -0400 X-Server-Uuid: 241911D6-425B-44B9-A073-E3FE0F8FC774 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Mon, 25 Jul 2005 18:48:42 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Mon, 25 Jul 2005 18:48:41 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Mon, 25 Jul 2005 15:48:39 -0700 Subject: Re: "Vacuum Full Analyze" taking so long From: "Luke Lonergan" To: "Tomeh, Husam" , pgsql-performance@postgresql.org Message-ID: In-Reply-To: MIME-Version: 1.0 X-OriginalArrivalTime: 25 Jul 2005 22:48:42.0462 (UTC) FILETIME=[019E7FE0:01C5916B] X-WSS-ID: 6EFBB3C02B46200164-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.574 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/376 X-Sequence-Number: 13617 Vacuum full takes an exclusive lock on the tables it runs against, so if you have anything else reading the table while you are trying to run it, the vacuum full will wait, possibly forever until it can get the lock. What does the system load look like while you are running this? What does vmstat 1 show you? Is there load on the system other than the database? Do you really need to run vacuum full instead of vacuum? - Luke On 7/25/05 2:30 PM, "Tomeh, Husam" wrote: > > I have an 8.02 postgresql database with about 180 GB in size, running on > 2.6 RedHat kernel with 32 GB of RAM and 2 CPUs. I'm running the vacuum > full analyze command, and has been running for at least two consecutive > days with no other processes running (it's an offline loading server). I > tweaked the maintenanace_mem to its max (2 GB) with work_mem of 8M. I > have no issues with my checkpoints. I can still I/O activities against > the physical files of the "property" table and its two indexes (primary > key and r index). The property files are about 128GB and indexes are > about 15 GB. I have run the same maintenance job on a different box > (staging) with identical hardware config (except with 64 GB instead of > 32) and took less than 12 hours. Any clue or tip is really > appreciated. > > Also read a comment by Tom Lane, that terminating the process should be > crash-safe if I had to. > > Thanks, > From pgsql-performance-owner@postgresql.org Mon Jul 25 20:32:19 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8EE1252AC4 for ; Mon, 25 Jul 2005 20:08:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18171-10 for ; Mon, 25 Jul 2005 23:08:44 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id EC51452894 for ; Mon, 25 Jul 2005 20:08:43 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050725230844m9200g3m59e>; Mon, 25 Jul 2005 23:08:44 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 4797A5600C; Mon, 25 Jul 2005 18:08:44 -0500 (CDT) Received: from [192.168.1.112] (71-32-73-140.cdrr.qwest.net [71.32.73.140]) by juju.arbash-meinel.com (Postfix) with ESMTP id F166C55FB5; Mon, 25 Jul 2005 18:08:34 -0500 (CDT) Message-ID: <42E570F1.7090500@arbash-meinel.com> Date: Mon, 25 Jul 2005 18:08:33 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.6 (Macintosh/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Chris Isaacson , Postgresql Performance Subject: Re: COPY insert performance References: <07774C6E31D94A44A2A60E2085944F09071DF0@tbmail.tradebot.com> In-Reply-To: <07774C6E31D94A44A2A60E2085944F09071DF0@tbmail.tradebot.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig409DCEA0AEEAA541DD15189C" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/374 X-Sequence-Number: 13615 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig409DCEA0AEEAA541DD15189C Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Chris Isaacson wrote: > I need COPY via libpqxx to insert millions of rows into two tables. One > table has roughly have as many rows and requires half the storage. In > production, the largest table will grow by ~30M rows/day. To test the > COPY performance I split my transactions into 10,000 rows. I insert > roughly 5000 rows into table A for every 10,000 rows into table B. > > Table A has one unique index: > > "order_main_pk" UNIQUE, btree (cl_ord_id) > > Table B has 1 unique index and 2 non-unique indexes: > > "order_transition_pk" UNIQUE, btree (collating_seq) > "order_transition_ak2" btree (orig_cl_ord_id) > "order_transition_ak3" btree (exec_id) Do you have any foreign key references? If you are creating a table for the first time (or loading a large fraction of the data), it is common to drop the indexes and foreign keys first, and then insert/copy, and then drop them again. Is InnoDB the backend with referential integrity, and true transaction support? I believe the default backend does not support either (so it is "cheating" to give you speed, which may be just fine for your needs, especially since you are willing to run fsync=false). I think moving pg_xlog to a dedicated drive (set of drives) could help your performance. As well as increasing checkpoint_segments. I don't know if you gain much by changing the bg_writer settings, if you are streaming everything in at once, you probably want to have it written out right away. My understanding is that bg_writer settings are for the case where you have mixed read and writes going on at the same time, and you want to make sure that the reads have time to execute (ie the writes are not saturating your IO). Also, is any of this tested under load? Having a separate process issue queries while you are loading in data. Traditionally MySQL is faster with a single process inserting/querying for data, but once you have multiple processes hitting it at the same time, it's performance degrades much faster than postgres. You also seem to be giving MySQL 512M of ram to work with, while only giving 2M/200M to postgres. (re)creating indexes uses maintenance_work_mem, but updating indexes could easily use work_mem. You may be RAM starved. John =:-> > > My testing environment is as follows: > -Postgresql 8.0.1 > -libpqxx 2.5.0 > -Linux 2.6.11.4-21.7-smp x86_64 > -Dual Opteron 246 > -System disk (postgres data resides on this SCSI disk) - Seagate > (ST373453LC) - 15K, 73 GB > (http://www.seagate.com/cda/products/discsales/marketing/detail/0,1081,549,00.html) > -2nd logical disk - 10K, 36GB IBM SCSI (IC35L036UCDY10-0) - WAL reside > on this disk > -NO RAID > > *PostgreSQL* > Here are the results of copying in 10M rows as fast as possible: > (10K/transaction) > Total Time: 1129.556 s > Rows/sec: 9899.922 > Transaction>1.2s 225 > Transaction>1.5s 77 > Transaction>2.0s 4 > Max Transaction 2.325s > > **MySQL** > **I ran a similar test with MySQL 4.1.10a (InnoDB) which produced these > results: (I used MySQL's INSERT INTO x VALUES > (1,2,3)(4,5,6)(...,...,...) syntax) (10K/transaction) > Total Time: 860.000 s > Rows/sec: 11627.91 > Transaction>1.2s 0 > Transaction>1.5s 0 > Transaction>2.0s 0 > Max Transaction 1.175s > > Considering the configurations shown below, can anyone offer advice to > close the 15% gap and the much worse variability I'm experiencing. Thanks > > My *postgresql.conf* has the following non-default values: > # ----------------------------- > # PostgreSQL configuration file > # ----------------------------- > listen_addresses = '*' # what IP interface(s) to listen on; > max_connections = 100 > #--------------------------------------------------------------------------- > # RESOURCE USAGE (except WAL) > #--------------------------------------------------------------------------- > shared_buffers = 65536 # min 16, at least max_connections*2, 8KB each > work_mem = 2048 # min 64, size in KB > maintenance_work_mem = 204800 # min 1024, size in KB > max_fsm_pages = 2250000 # min max_fsm_relations*16, 6 bytes each > bgwriter_delay = 200 # 10-10000 milliseconds between rounds > bgwriter_percent = 10 # 0-100% of dirty buffers in each round > bgwriter_maxpages = 1000 # 0-1000 buffers max per round > #--------------------------------------------------------------------------- > # WRITE AHEAD LOG > #--------------------------------------------------------------------------- > fsync = false # turns forced synchronization on or off > wal_buffers = 64 # min 4, 8KB each > checkpoint_segments = 40 # in logfile segments, min 1, 16MB each > checkpoint_timeout = 600 # range 30-3600, in seconds > #--------------------------------------------------------------------------- > # QUERY TUNING > #--------------------------------------------------------------------------- > effective_cache_size = 65536 # typically 8KB each > random_page_cost = 2 # units are one sequential page fetch cost > #--------------------------------------------------------------------------- > # ERROR REPORTING AND LOGGING > #--------------------------------------------------------------------------- > > log_min_duration_statement = 250 # -1 is disabled, in milliseconds. > log_connections = true > log_disconnections = true > log_duration = true > log_line_prefix = '<%r%u%p%t%d%%' # e.g. '<%u%%%d> ' > # %u=user name %d=database name > # %r=remote host and port > # %p=PID %t=timestamp %i=command tag > # %c=session id %l=session line number > # %s=session start timestamp %x=transaction id > # %q=stop here in non-session processes > # %%='%' > log_statement = 'none' # none, mod, ddl, all > #--------------------------------------------------------------------------- > # RUNTIME STATISTICS > #--------------------------------------------------------------------------- > # - Query/Index Statistics Collector - > stats_start_collector = true > stats_command_string = true > stats_block_level = true > stats_row_level = true > stats_reset_on_server_start = true > > My MySQL *my.ini* has the following non default values: > innodb_data_home_dir = /var/lib/mysql/ > innodb_data_file_path = ibdata1:10M:autoextend > innodb_log_group_home_dir = /var/lib/mysql/ > innodb_log_arch_dir = /var/lib/mysql/ > # You can set .._buffer_pool_size up to 50 - 80 % > # of RAM but beware of setting memory usage too high > innodb_buffer_pool_size = 512M > innodb_additional_mem_pool_size = 64M > # Set .._log_file_size to 25 % of buffer pool size > innodb_log_file_size = 128M > innodb_log_buffer_size = 64M > innodb_flush_log_at_trx_commit = 1 > innodb_lock_wait_timeout = 50 > innodb_flush_method = O_DSYNC > max_allowed_packet = 16M > > > --------------enig409DCEA0AEEAA541DD15189C Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5XDxJdeBCYSNAAMRAu8tAJ9DW0ixS8RCamqhXjo9x/AU2MjKOwCfR5Ku g0RlPleYCcbP+4F/kBfNhxg= =/FaG -----END PGP SIGNATURE----- --------------enig409DCEA0AEEAA541DD15189C-- From pgsql-performance-owner@postgresql.org Mon Jul 25 20:20:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B070D52F04 for ; Mon, 25 Jul 2005 20:14:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 22483-02 for ; Mon, 25 Jul 2005 23:14:22 +0000 (GMT) Received: from wsmailap01.firstam.com (outbound-smtp01.firstam.com [208.246.101.197]) by svr1.postgresql.org (Postfix) with ESMTP id 541B252E0C for ; Mon, 25 Jul 2005 20:14:21 -0300 (ADT) Received: from 172.21.131.5 by wsmailap02.firstam.com with ESMTP ( Tumbleweed MMS SMTP Relay (MMS v5.6.3)); Mon, 25 Jul 2005 16:14:07 -0700 X-Server-Uuid: 7F57BFB8-2F8F-44C6-8C75-DD639B93CEA6 Received: from anammx01.ana.firstamdata.com ([192.168.173.35]) by famarp01.firstam.com (MOS 3.5.6-GR) with SMTP id EGA36864; Mon, 25 Jul 2005 16:14:17 -0700 (PDT) Received: from pisgana01sxch21.ana.firstamdata.com (Not Verified[192.168.173.71]) by mailgateway.firstam.com (post.office MTA v5.0 0924 ) with ESMTP id ; Mon, 25 Jul 2005 16:14:17 -0700 Received: from pisgana01sxch01.ana.firstamdata.com ([192.168.173.70]) by pisgana01sxch21.ana.firstamdata.com with Microsoft SMTPSVC(6.0.3790.211 ); Mon, 25 Jul 2005 16:14:17 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: "Vacuum Full Analyze" taking so long Date: Mon, 25 Jul 2005 16:14:16 -0700 Message-ID: Thread-Topic: [PERFORM] "Vacuum Full Analyze" taking so long Thread-Index: AcWRazJ1+TPBVbzoS0KsnTmbFkEs4wAAk6ng From: "Tomeh, Husam" To: "Luke Lonergan" , pgsql-performance@postgresql.org X-OriginalArrivalTime: 25 Jul 2005 23:14:17.0282 (UTC) FILETIME=[94717220:01C5916E] X-WSS-ID: 6EFBADB523K977078-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.027 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/373 X-Sequence-Number: 13614 =20 Nothing was running except the job. The server did not look stressed out looking at top and vmstat. We have seen slower query performance when performing load tests, so I run the re-index on all application indexes and then issue a full vacuum. I ran the same thing on a staging server and it took less than 12 hours. Is there a possibility the DB pages are corrupted. Is there a command to verify that. (In Oracle, there's a dbverify command that checks for corruption on the data files level).=20 The other question I have. What would be the proper approach to rebuild indexes. I re-indexes and then run vacuum/analyze. Should I not use the re-index approach, and instead, drop the indexes, vacuum the tables, and then create the indexes, then run analyze on tables and indexes??=20 Thanks, --=20 =20Husam=20 -----Original Message----- From: Luke Lonergan [mailto:llonergan@greenplum.com]=20 Sent: Monday, July 25, 2005 3:49 PM To: Tomeh, Husam; pgsql-performance@postgresql.org Subject: Re: [PERFORM] "Vacuum Full Analyze" taking so long Vacuum full takes an exclusive lock on the tables it runs against, so if you have anything else reading the table while you are trying to run it, the vacuum full will wait, possibly forever until it can get the lock. What does the system load look like while you are running this? What does vmstat 1 show you? Is there load on the system other than the database? Do you really need to run vacuum full instead of vacuum? - Luke On 7/25/05 2:30 PM, "Tomeh, Husam" wrote: > =20 > I have an 8.02 postgresql database with about 180 GB in size, running=20 > on > 2.6 RedHat kernel with 32 GB of RAM and 2 CPUs. I'm running the vacuum > full analyze command, and has been running for at least two=20 > consecutive days with no other processes running (it's an offline=20 > loading server). I tweaked the maintenanace_mem to its max (2 GB) with > work_mem of 8M. I have no issues with my checkpoints. I can still I/O=20 > activities against the physical files of the "property" table and its=20 > two indexes (primary key and r index). The property files are about=20 > 128GB and indexes are about 15 GB. I have run the same maintenance job > on a different box > (staging) with identical hardware config (except with 64 GB instead of > 32) and took less than 12 hours. Any clue or tip is really > appreciated.=20 >=20 > Also read a comment by Tom Lane, that terminating the process should=20 > be crash-safe if I had to. >=20 > Thanks, >=20 ********************************************************************** This message contains confidential information intended only for the=20 use of the addressee(s) named above and may contain information that=20 is legally privileged. If you are not the addressee, or the person=20 responsible for delivering it to the addressee, you are hereby=20 notified that reading, disseminating, distributing or copying this=20 message is strictly prohibited. If you have received this message by=20 mistake, please immediately notify us by replying to the message and=20 delete the original message immediately thereafter. Thank you. FADLD Tag ********************************************************************** From pgsql-performance-owner@postgresql.org Mon Jul 25 23:21:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7E6EB52C48 for ; Mon, 25 Jul 2005 20:31:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25483-03 for ; Mon, 25 Jul 2005 23:31:13 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id 5D18452FE9 for ; Mon, 25 Jul 2005 20:31:12 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050725233116m9200g50sde>; Mon, 25 Jul 2005 23:31:16 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 614E75600C; Mon, 25 Jul 2005 18:31:15 -0500 (CDT) Received: from [192.168.1.112] (71-32-73-140.cdrr.qwest.net [71.32.73.140]) by juju.arbash-meinel.com (Postfix) with ESMTP id B59EE55FB5; Mon, 25 Jul 2005 18:31:11 -0500 (CDT) Message-ID: <42E5763C.7030804@arbash-meinel.com> Date: Mon, 25 Jul 2005 18:31:08 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.6 (Macintosh/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Tomeh, Husam" Cc: Luke Lonergan , pgsql-performance@postgresql.org Subject: Re: "Vacuum Full Analyze" taking so long References: In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig4E54487029FFB447F5276BE0" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/379 X-Sequence-Number: 13620 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig4E54487029FFB447F5276BE0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Tomeh, Husam wrote: > > Nothing was running except the job. The server did not look stressed out > looking at top and vmstat. We have seen slower query performance when > performing load tests, so I run the re-index on all application indexes > and then issue a full vacuum. I ran the same thing on a staging server > and it took less than 12 hours. Is there a possibility the DB pages are > corrupted. Is there a command to verify that. (In Oracle, there's a > dbverify command that checks for corruption on the data files level). > > The other question I have. What would be the proper approach to rebuild > indexes. I re-indexes and then run vacuum/analyze. Should I not use the > re-index approach, and instead, drop the indexes, vacuum the tables, and > then create the indexes, then run analyze on tables and indexes?? I *think* if you are planning on dropping the indexes anyway, just drop them, VACUUM ANALYZE, and then recreate them, I don't think you have to re-analyze after you have recreated them. John =:-> > > Thanks, > > --------------enig4E54487029FFB447F5276BE0 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5XY9JdeBCYSNAAMRAip9AKClCCUdHAiQHR1kfGxOqBXhgBbv5QCgnxhV DVswER6oKm+snqzCAuuLXgk= =vFll -----END PGP SIGNATURE----- --------------enig4E54487029FFB447F5276BE0-- From pgsql-performance-owner@postgresql.org Mon Jul 25 23:11:12 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A89EA52E8E for ; Mon, 25 Jul 2005 21:31:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 38776-02 for ; Tue, 26 Jul 2005 00:31:08 +0000 (GMT) Received: from mail.Mi8.com (d01gw01.mi8.com [63.240.6.47]) by svr1.postgresql.org (Postfix) with ESMTP id 8543352B57 for ; Mon, 25 Jul 2005 21:31:06 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D1)); Mon, 25 Jul 2005 20:31:03 -0400 X-Server-Uuid: 241911D6-425B-44B9-A073-E3FE0F8FC774 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Mon, 25 Jul 2005 20:31:03 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Mon, 25 Jul 2005 20:31:02 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Mon, 25 Jul 2005 17:31:02 -0700 Subject: Re: COPY insert performance From: "Luke Lonergan" To: "Chris Isaacson" , pgsql-performance@postgresql.org Message-ID: In-Reply-To: <07774C6E31D94A44A2A60E2085944F09071DF0@tbmail.tradebot.com> MIME-Version: 1.0 X-OriginalArrivalTime: 26 Jul 2005 00:31:03.0123 (UTC) FILETIME=[4DBD0230:01C59179] X-WSS-ID: 6EFB5BCD2B46231971-01-01 Content-Type: multipart/alternative; boundary=B_3205157462_11232091 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.718 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, HTML_30_40, HTML_FONT_BIG, HTML_MESSAGE, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/378 X-Sequence-Number: 13619 > This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. --B_3205157462_11232091 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Chris, You can try the Bizgres distribution of postgres (based on version 8.0.3), the COPY support is 30% faster as reported by OSDL (without indexes). This is due to very slow parsing within the COPY command, which is sped up using micro-optimized logic for parsing. There is a patch pending for the development version of Postgres which implements the same code, but you can use Bizgres and get it now instead of waiting for postgres 8.1 to come out. Also, Bizgres is QA tested with the enhanced features. Bizgres is a free / open source distribution of Postgres for Business Intelligence / Data Warehousing. Bizgres currently features postgres 8.0.3 plus these patches: * Bypass WAL when performing =B3CREATE TABLE AS SELECT=B2 * COPY is between 30% and 90% faster on machines with fast I/O * Enhanced support for data partitioning with partition elimination optimization=20 * Bitmap Scan support for multiple index use in queries and better low cardinality column performance * Improved optimization of queries with LIMIT See: http://www.bizgres.org for more. - Luke On 7/25/05 3:32 PM, "Chris Isaacson" wrote: > I need COPY via libpqxx to insert millions of rows into two tables. One = table > has roughly have as many rows and requires half the storage. In producti= on, > the largest table will grow by ~30M rows/day. To test the COPY performan= ce I > split my transactions into 10,000 rows. I insert roughly 5000 rows into = table > A for every 10,000 rows into table B. > =20 > Table A has one unique index: > =20 > "order_main_pk" UNIQUE, btree (cl_ord_id) > =20 > Table B has 1 unique index and 2 non-unique indexes: > =20 > "order_transition_pk" UNIQUE, btree (collating_seq) > "order_transition_ak2" btree (orig_cl_ord_id) > "order_transition_ak3" btree (exec_id) > =20 > My testing environment is as follows: > -Postgresql 8.0.1 > -libpqxx 2.5.0 > -Linux 2.6.11.4-21.7-smp x86_64 > -Dual Opteron 246 > -System disk (postgres data resides on this SCSI disk) - Seagate (ST3734= 53LC) > - 15K, 73 GB=20 > (http://www.seagate.com/cda/products/discsales/marketing/detail/0,1081,54= 9,00. > html) > -2nd logical disk - 10K, 36GB IBM SCSI (IC35L036UCDY10-0) - WAL reside on= this > disk > -NO RAID > =20 > PostgreSQL > Here are the results of copying in 10M rows as fast as possible: > (10K/transaction) > Total Time: 1129.556 s > Rows/sec: 9899.922 > Transaction>1.2s 225 > Transaction>1.5s 77 > Transaction>2.0s 4 > Max Transaction 2.325s > =20 > MySQL > I ran a similar test with MySQL 4.1.10a (InnoDB) which produced these res= ults: > (I used MySQL's INSERT INTO x VALUES (1,2,3)(4,5,6)(...,...,...) syntax) > (10K/transaction) > Total Time: 860.000 s > Rows/sec: 11627.91 > Transaction>1.2s 0 > Transaction>1.5s 0 > Transaction>2.0s 0 > Max Transaction 1.175s > =20 > Considering the configurations shown below, can anyone offer advice to cl= ose > the 15% gap and the much worse variability I'm experiencing. Thanks > =20 > My postgresql.conf has the following non-default values: > # ----------------------------- > # PostgreSQL configuration file > # ----------------------------- > listen_addresses =3D '*' # what IP interface(s) to listen on; > max_connections =3D 100 > #------------------------------------------------------------------------= --- > # RESOURCE USAGE (except WAL) > #------------------------------------------------------------------------= --- > shared_buffers =3D 65536 # min 16, at least max_connections*2, 8KB each > work_mem =3D 2048 # min 64, size in KB > maintenance_work_mem =3D 204800 # min 1024, size in KB > max_fsm_pages =3D 2250000 # min max_fsm_relations*16, 6 bytes each > bgwriter_delay =3D 200 # 10-10000 milliseconds between rounds > bgwriter_percent =3D 10 # 0-100% of dirty buffers in each round > bgwriter_maxpages =3D 1000 # 0-1000 buffers max per round > #------------------------------------------------------------------------= --- > # WRITE AHEAD LOG > #------------------------------------------------------------------------= --- > fsync =3D false # turns forced synchronization on or off > wal_buffers =3D 64 # min 4, 8KB each > checkpoint_segments =3D 40 # in logfile segments, min 1, 16MB each > checkpoint_timeout =3D 600 # range 30-3600, in seconds > #------------------------------------------------------------------------= --- > # QUERY TUNING > #------------------------------------------------------------------------= --- > effective_cache_size =3D 65536 # typically 8KB each > random_page_cost =3D 2 # units are one sequential page fetch cost > #------------------------------------------------------------------------= --- > # ERROR REPORTING AND LOGGING > #------------------------------------------------------------------------= --- > log_min_duration_statement =3D 250 # -1 is disabled, in milliseconds. > log_connections =3D true > log_disconnections =3D true > log_duration =3D true > log_line_prefix =3D '<%r%u%p%t%d%%' # e.g. '<%u%%%d> ' > # %u=3Duser name %d=3Ddatabase name > # %r=3Dremote host and port > # %p=3DPID %t=3Dtimestamp %i=3Dcommand tag > # %c=3Dsession id %l=3Dsession line number > # %s=3Dsession start timestamp %x=3Dtransaction id > # %q=3Dstop here in non-session processes > # %%=3D'%' > log_statement =3D 'none' # none, mod, ddl, all > #------------------------------------------------------------------------= --- > # RUNTIME STATISTICS > #------------------------------------------------------------------------= --- > # - Query/Index Statistics Collector - > stats_start_collector =3D true > stats_command_string =3D true > stats_block_level =3D true > stats_row_level =3D true > stats_reset_on_server_start =3D true > =20 > My MySQL my.ini has the following non default values: > innodb_data_home_dir =3D /var/lib/mysql/ > innodb_data_file_path =3D ibdata1:10M:autoextend > innodb_log_group_home_dir =3D /var/lib/mysql/ > innodb_log_arch_dir =3D /var/lib/mysql/ > # You can set .._buffer_pool_size up to 50 - 80 % > # of RAM but beware of setting memory usage too high > innodb_buffer_pool_size =3D 512M > innodb_additional_mem_pool_size =3D 64M > # Set .._log_file_size to 25 % of buffer pool size > innodb_log_file_size =3D 128M > innodb_log_buffer_size =3D 64M > innodb_flush_log_at_trx_commit =3D 1 > innodb_lock_wait_timeout =3D 50 > innodb_flush_method =3D O_DSYNC > max_allowed_packet =3D 16M > =20 > =20 > =20 >=20 --B_3205157462_11232091 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Re: [PERFORM] COPY insert performance Chris= ,

You can try the Bizgres distribution of postgres (based on version 8.0.3), = the COPY support is 30% faster as reported by OSDL (without indexes).  = This is due to very slow parsing within the COPY command, which is sped up u= sing micro-optimized logic for parsing.  There is a patch pending for t= he development version of Postgres which implements the same code, but you c= an use Bizgres and get it now instead of waiting for postgres 8.1 to come ou= t.  Also, Bizgres is QA tested with the enhanced features.

Bizgres is a free / open source distribution of Postgres for Business Intel= ligence / Data Warehousing.

Bizgres currently features postgres 8.0.3 plus these patches:
  • Bypass WAL when performing “CREATE TABLE AS SELECT= 221;
  • COPY is between 30% and 90% faster on machines with fast I/O
  • Enhanced support for data partitioning with partition eliminati= on optimization
  • Bitmap Scan support for multiple index use in queries and bette= r low cardinality column performance
  • Improved optimization of queries with LIMIT

See: http://www.bizgres.org for more.<= BR>
- Luke


On 7/25/05 3:32 PM, "Chris Isaacson" <cisaacson@tradebotsystem= s.com> wrote:

I need COPY via libpqxx to insert millions of rows into two tables.  O= ne table has roughly have as many rows and requires half the storage.  = In production, the largest table will grow by ~30M rows/day.  To test t= he COPY performance I split my transactions into 10,000 rows.  I insert= roughly 5000 rows into table A for every 10,000 rows into table B.
Table A has one unique index:
"order_main_pk" UNIQUE, btree (cl_ord_i= d)
Table B has 1 unique index and 2 non-unique index= es:
"order_transition_pk" UNIQUE, btree (co= llating_seq)
"order_transition_ak2" btree (orig_cl_ord_id)
"order_transition_ak3" btree (exec_id)
My testing environment is as follows:
-Postgresql 8.0.1
-libpqxx 2.5.0
-Linux 2.6.11.4-21.7-smp x86_64
-Dual Opteron 246
-System disk (postgres data resides on this SCSI disk) - Seag= ate (ST373453LC) - 15K, 73 GB (http://www.seagate.com/cda/p= roducts/discsales/marketing/detail/0,1081,549,00.html)
-2nd logical disk - 10K, 36GB IBM SCSI (IC35L036UCDY10-0) - = WAL reside on this disk
-NO RAID
 
PostgreS= QL
Here are the results of copying in 10M rows as fast as possible: (10K/t= ransaction)
Total Time:           &nb= sp;1129.556 s
Rows/sec:            = ; 9899.922
Transaction>1.2s    225
Transaction>1.5s     77
Transaction>2.0s      4
Max Transaction       2.325s
 
MySQL
I ran a similar test with MySQL 4.1.10a (InnoDB) which produced these r= esults: (I used MySQL's INSERT INTO x VALUES (1,2,3)(4,5,6)(...,...,...) syn= tax) (10K/transaction)
Total Time:         860.000 s
Rows/sec:        11627.91
Transaction>1.2s      0
Transaction>1.5s      0
Transaction>2.0s      0
Max Transaction       1.175s
 
Considering the configurations shown below, can anyone offer advice to clos= e the 15% gap and the much worse variability I'm experiencing.  Thanks<= BR>
My postgresql.conf has the following non-d= efault values:
# -----------------------------
# PostgreSQL configuration file
# -----------------------------
listen_addresses =3D '*' # what IP interface(s) to listen on;
max_connections =3D 100
#--------------------------------------------------------------------------= -
# RESOURCE USAGE (except WAL)
#--------------------------------------------------------------------------= -
shared_buffers =3D 65536  # min 16, at least max_connections*2, 8KB each=
work_mem =3D 2048   # min 64, size in KB
maintenance_work_mem =3D 204800 # min 1024, size in KB
max_fsm_pages =3D 2250000  # min max_fsm_relations*16, 6 bytes each
bgwriter_delay =3D 200  # 10-10000 milliseconds between rounds
bgwriter_percent =3D 10  # 0-100% of dirty buffers in each round
bgwriter_maxpages =3D 1000 # 0-1000 buffers max per round
#--------------------------------------------------------------------------= -
# WRITE AHEAD LOG
#--------------------------------------------------------------------------= -
fsync =3D false   # turns forced synchronization on or off
wal_buffers =3D 64  # min 4, 8KB each
checkpoint_segments =3D 40 # in logfile segments, min 1, 16MB each
checkpoint_timeout =3D 600 # range 30-3600, in seconds
#--------------------------------------------------------------------------= -
# QUERY TUNING
#--------------------------------------------------------------------------= -
effective_cache_size =3D 65536 # typically 8KB each
random_page_cost =3D 2  # units are one sequential page fetch cost
#--------------------------------------------------------------------------= -
# ERROR REPORTING AND LOGGING
#--------------------------------------------------------------------------= -   
log_min_duration_statement =3D    250 # -1 is disabled, in mil= liseconds.
log_connections =3D true
log_disconnections =3D true
log_duration =3D true
log_line_prefix =3D '<%r%u%p%t%d%%'  # e.g. '<%u%%%d> '
    # %u=3Duser name %d=3Ddatabase name
    # %r=3Dremote host and port
    # %p=3DPID %t=3Dtimestamp %i=3Dcommand tag
    # %c=3Dsession id %l=3Dsession line number
    # %s=3Dsession start timestamp %x=3Dtransaction id
    # %q=3Dstop here in non-session processes
    # %%=3D'%'
log_statement =3D 'none'  # none, mod, ddl, all
#--------------------------------------------------------------------------= -
# RUNTIME STATISTICS
#--------------------------------------------------------------------------= -
# - Query/Index Statistics Collector -
stats_start_collector =3D true
stats_command_string =3D true
stats_block_level =3D true
stats_row_level =3D true
stats_reset_on_server_start =3D true
My MySQL my.ini has the following non defa= ult values:
innodb_data_home_dir =3D /var/lib/mysql/
innodb_data_file_path =3D ibdata1:10M:autoextend
innodb_log_group_home_dir =3D /var/lib/mysql/
innodb_log_arch_dir =3D /var/lib/mysql/
# You can set .._buffer_pool_size up to 50 - 80 %
# of RAM but beware of setting memory usage too high
innodb_buffer_pool_size =3D 512M
innodb_additional_mem_pool_size =3D 64M
# Set .._log_file_size to 25 % of buffer pool size
innodb_log_file_size =3D 128M
innodb_log_buffer_size =3D 64M
innodb_flush_log_at_trx_commit =3D 1
innodb_lock_wait_timeout =3D 50
innodb_flush_method =3D O_DSYNC
max_allowed_packet =3D 16M
 


--B_3205157462_11232091-- From pgsql-performance-owner@postgresql.org Mon Jul 25 23:10:13 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BE28352BDC for ; Mon, 25 Jul 2005 21:52:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 41807-09 for ; Tue, 26 Jul 2005 00:51:51 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.50]) by svr1.postgresql.org (Postfix) with ESMTP id 6414852BB6 for ; Mon, 25 Jul 2005 21:51:50 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Mon, 25 Jul 2005 20:51:49 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Mon, 25 Jul 2005 20:51:49 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Mon, 25 Jul 2005 20:51:48 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Mon, 25 Jul 2005 17:51:48 -0700 Subject: Re: "Vacuum Full Analyze" taking so long From: "Luke Lonergan" To: "John A Meinel" , "Tomeh, Husam" Cc: pgsql-performance@postgresql.org Message-ID: In-Reply-To: <42E5763C.7030804@arbash-meinel.com> MIME-Version: 1.0 X-OriginalArrivalTime: 26 Jul 2005 00:51:49.0530 (UTC) FILETIME=[34A79FA0:01C5917C] X-WSS-ID: 6EFB56AF1M812745206-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.576 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/377 X-Sequence-Number: 13618 Husam, On 7/25/05 4:31 PM, "John A Meinel" wrote: > Tomeh, Husam wrote: >> >> Nothing was running except the job. The server did not look stressed out >> looking at top and vmstat. We have seen slower query performance when >> performing load tests, so I run the re-index on all application indexes >> and then issue a full vacuum. I ran the same thing on a staging server >> and it took less than 12 hours. Is there a possibility the DB pages are >> corrupted. Is there a command to verify that. (In Oracle, there's a >> dbverify command that checks for corruption on the data files level). >> >> The other question I have. What would be the proper approach to rebuild >> indexes. I re-indexes and then run vacuum/analyze. Should I not use the >> re-index approach, and instead, drop the indexes, vacuum the tables, and >> then create the indexes, then run analyze on tables and indexes?? > > I *think* if you are planning on dropping the indexes anyway, just drop > them, VACUUM ANALYZE, and then recreate them, I don't think you have to > re-analyze after you have recreated them. I agree - and don't run "VACUUM FULL", it is quite different from "VACUUM". Also - you should only need to vacuum if you've deleted a lot of data. It's job is to reclaim space lost to rows marked deleted. So, in fact, you may not even need to run VACUUM. "VACUUM FULL" is like a disk defragmentation operation within the DBMS, and is only necessary if there is a slowdown in performance from lots and lots of deletes and/or updates and new data isn't finding sequential pages for storage, which is rare. Given the need for locking, it's generally better to dump and restore in that case, but again it's a very rare occasion. I don't know of a command to check for page corruption, but I would think that if you can run VACUUM (not full) you should be OK. - Luke From pgsql-performance-owner@postgresql.org Tue Jul 26 06:58:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id ED96952EAD for ; Tue, 26 Jul 2005 06:58:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 60078-06 for ; Tue, 26 Jul 2005 09:58:24 +0000 (GMT) Received: from mailservice.tudelft.nl (mailservice.tudelft.nl [130.161.131.5]) by svr1.postgresql.org (Postfix) with ESMTP id C8B5F52EF2 for ; Tue, 26 Jul 2005 06:58:19 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by rav.antivirus (Postfix) with ESMTP id CC1F080184 for ; Tue, 26 Jul 2005 11:58:21 +0200 (CEST) Received: from [130.161.3.219] (oli219.office.oli.tudelft.nl [130.161.3.219]) by mx4.tudelft.nl (Postfix) with ESMTP id 0F29380183 for ; Tue, 26 Jul 2005 11:58:21 +0200 (CEST) Message-ID: <42E60939.9010905@oli.tudelft.nl> Date: Tue, 26 Jul 2005 11:58:17 +0200 From: Jochem van Dieten Organization: OnLine Internet User-Agent: Mozilla Thunderbird 1.0.5 (Windows/20050711) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: "Vacuum Full Analyze" taking so long References: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at tudelft.nl X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/380 X-Sequence-Number: 13621 Tomeh, Husam wrote: > The other question I have. What would be the proper approach to rebuild > indexes. I re-indexes and then run vacuum/analyze. Should I not use the > re-index approach, and instead, drop the indexes, vacuum the tables, and > then create the indexes, then run analyze on tables and indexes?? If you just want to rebuild indexes, just drop and recreate. However, you are also running a VACUUM FULL, so I presume you have deleted a significant number of rows and want to recover the space that was in use by them. In that scenario, it is often better to CLUSTER the table to force a rebuild. While VACUUM FULL moves the tuples around inside the existing file(s), CLUSTER simply creates new file(s), moves all the non-deleted tuples there and then swaps the old and the new files. There can be a significant performance increase in doing so (but you obviously need to have some free diskspace). If you CLUSTER your table it will be ordered by the index you specify. There can be a performance increase in doing so, but if you don't want to you can also do a no-op ALTER TABLE and change a column to a datatype that is the same as it already has. This too will force a rewrite of the table but without ordering the tuples. So in short my recommendations: - to rebuild indexes, just drop and recreate the indexes - to rebuild everything because there is space that can bepermanently reclaimed, drop indexes, cluster or alter the table, recreate the indexes and anlyze the table Jochem From pgsql-performance-owner@postgresql.org Tue Jul 26 09:12:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 73AD552E5C for ; Tue, 26 Jul 2005 09:12:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92478-02 for ; Tue, 26 Jul 2005 12:12:20 +0000 (GMT) Received: from linuxworld.com.au (unknown [203.34.46.50]) by svr1.postgresql.org (Postfix) with ESMTP id 0794252CC6 for ; Tue, 26 Jul 2005 09:12:18 -0300 (ADT) Received: from linuxworld.com.au (IDENT:swm@localhost.localdomain [127.0.0.1]) by linuxworld.com.au (8.13.2/8.13.2) with ESMTP id j6QCCMLl015654; Tue, 26 Jul 2005 22:12:22 +1000 Received: from localhost (swm@localhost) by linuxworld.com.au (8.13.2/8.13.2/Submit) with ESMTP id j6QCCL0o015651; Tue, 26 Jul 2005 22:12:22 +1000 X-Authentication-Warning: linuxworld.com.au: swm owned process doing -bs Date: Tue, 26 Jul 2005 22:12:21 +1000 (EST) From: Gavin Sherry X-X-Sender: swm@linuxworld.com.au To: Chris Isaacson Cc: pgsql-performance@postgresql.org Subject: Re: COPY insert performance In-Reply-To: <07774C6E31D94A44A2A60E2085944F09071DF0@tbmail.tradebot.com> Message-ID: References: <07774C6E31D94A44A2A60E2085944F09071DF0@tbmail.tradebot.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/381 X-Sequence-Number: 13622 Hi Chris, Have you considered breaking the data into multiple chunks and COPYing each concurrently? Also, have you ensured that your table isn't storing OIDs? On Mon, 25 Jul 2005, Chris Isaacson wrote: > #----------------------------------------------------------------------- > ---- > # RESOURCE USAGE (except WAL) > #----------------------------------------------------------------------- > ---- > shared_buffers = 65536 # min 16, at least max_connections*2, 8KB each shared_buffers that high has been shown to affect performance. Try 12000. > wal_buffers = 64 # min 4, 8KB each Increasing wal_buffers can also have an effect on performance. Thanks, Gavin From pgsql-performance-owner@postgresql.org Tue Jul 26 09:15:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F04B652E46 for ; Tue, 26 Jul 2005 09:15:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93010-01 for ; Tue, 26 Jul 2005 12:15:12 +0000 (GMT) Received: from tbmail.tradebot.com (Tradebot-Systems-1096753.cust-rtr.swbell.net [68.90.170.2]) by svr1.postgresql.org (Postfix) with ESMTP id 63A7052E56 for ; Tue, 26 Jul 2005 09:15:11 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Subject: Re: COPY insert performance X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Date: Tue, 26 Jul 2005 07:15:15 -0500 Message-ID: <07774C6E31D94A44A2A60E2085944F09071DF1@tbmail.tradebot.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] COPY insert performance Thread-Index: AcWRbc7D9PAWJ1M0SPemeI02RDv+BgAau4hw From: "Chris Isaacson" To: "John A Meinel" , X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.229 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: * X-Archive-Number: 200507/382 X-Sequence-Number: 13623 I do not have any foreign keys and I need the indexes on during the insert/copy b/c in production a few queries heavily dependent on the indexes will be issued. These queries will be infrequent, but must be fast when issued. I am using InnoDB with MySQL which appears to enforce true transaction support. (http://dev.mysql.com/doc/mysql/en/innodb-overview.html) If not, how is InnoDB "cheating"? Sorry for the confusion, but pg_xlog is currently on a dedicated drive (10K SCSI, see below). Would I realize further gains if I had a third drive and put the indexes on that drive? =20 I've played with the checkpoint_segments. I noticed an enormous improvement increasing from the default to 40, but neglible improvement thereafter. Do you have a recommendation for a value? My bg_writer adjustments were a last ditch effort. I found your advice correct and realized no gain. I have not tested under a querying load which is a good next step. I had not thought of the comparative degradation of MySQL vs. PostgreSQL. Thanks for the tip on the RAM usage by indexes. I was under the incorrect assumption that shared_buffers would take care of this. I'll increase work_mem to 512MB and rerun my test. I have 1G of RAM, which is less than we'll be running in production (likely 2G). -----Original Message----- From: John A Meinel [mailto:john@arbash-meinel.com]=20 Sent: Monday, July 25, 2005 6:09 PM To: Chris Isaacson; Postgresql Performance Subject: Re: [PERFORM] COPY insert performance Chris Isaacson wrote: > I need COPY via libpqxx to insert millions of rows into two tables. =20 > One table has roughly have as many rows and requires half the storage. > In production, the largest table will grow by ~30M rows/day. To test=20 > the COPY performance I split my transactions into 10,000 rows. I=20 > insert roughly 5000 rows into table A for every 10,000 rows into table > B. > > Table A has one unique index: > > "order_main_pk" UNIQUE, btree (cl_ord_id) > > Table B has 1 unique index and 2 non-unique indexes: > > "order_transition_pk" UNIQUE, btree (collating_seq)=20 > "order_transition_ak2" btree (orig_cl_ord_id) "order_transition_ak3"=20 > btree (exec_id) Do you have any foreign key references? If you are creating a table for the first time (or loading a large fraction of the data), it is common to drop the indexes and foreign keys first, and then insert/copy, and then drop them again. Is InnoDB the backend with referential integrity, and true transaction support? I believe the default backend does not support either (so it is "cheating" to give you speed, which may be just fine for your needs, especially since you are willing to run fsync=3Dfalse). I think moving pg_xlog to a dedicated drive (set of drives) could help your performance. As well as increasing checkpoint_segments. I don't know if you gain much by changing the bg_writer settings, if you are streaming everything in at once, you probably want to have it written out right away. My understanding is that bg_writer settings are for the case where you have mixed read and writes going on at the same time, and you want to make sure that the reads have time to execute (ie the writes are not saturating your IO). Also, is any of this tested under load? Having a separate process issue queries while you are loading in data. Traditionally MySQL is faster with a single process inserting/querying for data, but once you have multiple processes hitting it at the same time, it's performance degrades much faster than postgres. You also seem to be giving MySQL 512M of ram to work with, while only giving 2M/200M to postgres. (re)creating indexes uses maintenance_work_mem, but updating indexes could easily use work_mem. You may be RAM starved. John =3D:-> > > My testing environment is as follows: > -Postgresql 8.0.1 > -libpqxx 2.5.0 > -Linux 2.6.11.4-21.7-smp x86_64 > -Dual Opteron 246 > -System disk (postgres data resides on this SCSI disk) - Seagate > (ST373453LC) - 15K, 73 GB > (http://www.seagate.com/cda/products/discsales/marketing/detail/0,1081 > ,549,00.html) > -2nd logical disk - 10K, 36GB IBM SCSI (IC35L036UCDY10-0) - WAL reside > on this disk > -NO RAID > > *PostgreSQL* > Here are the results of copying in 10M rows as fast as possible: > (10K/transaction) > Total Time: 1129.556 s > Rows/sec: 9899.922 > Transaction>1.2s 225 > Transaction>1.5s 77 > Transaction>2.0s 4 > Max Transaction 2.325s > > **MySQL** > **I ran a similar test with MySQL 4.1.10a (InnoDB) which produced=20 > these > results: (I used MySQL's INSERT INTO x VALUES > (1,2,3)(4,5,6)(...,...,...) syntax) (10K/transaction) > Total Time: 860.000 s > Rows/sec: 11627.91 > Transaction>1.2s 0 > Transaction>1.5s 0 > Transaction>2.0s 0 > Max Transaction 1.175s > > Considering the configurations shown below, can anyone offer advice to > close the 15% gap and the much worse variability I'm experiencing. =20 > Thanks > > My *postgresql.conf* has the following non-default values: > # ----------------------------- > # PostgreSQL configuration file > # ----------------------------- > listen_addresses =3D '*' # what IP interface(s) to listen on;=20 > max_connections =3D 100 > #--------------------------------------------------------------------- > ------ > # RESOURCE USAGE (except WAL) > #----------------------------------------------------------------------- ---- > shared_buffers =3D 65536 # min 16, at least max_connections*2, 8KB = each > work_mem =3D 2048 # min 64, size in KB > maintenance_work_mem =3D 204800 # min 1024, size in KB > max_fsm_pages =3D 2250000 # min max_fsm_relations*16, 6 bytes each > bgwriter_delay =3D 200 # 10-10000 milliseconds between rounds > bgwriter_percent =3D 10 # 0-100% of dirty buffers in each round > bgwriter_maxpages =3D 1000 # 0-1000 buffers max per round > #----------------------------------------------------------------------- ---- > # WRITE AHEAD LOG > #----------------------------------------------------------------------- ---- > fsync =3D false # turns forced synchronization on or off > wal_buffers =3D 64 # min 4, 8KB each > checkpoint_segments =3D 40 # in logfile segments, min 1, 16MB each > checkpoint_timeout =3D 600 # range 30-3600, in seconds > #----------------------------------------------------------------------- ---- > # QUERY TUNING > #----------------------------------------------------------------------- ---- > effective_cache_size =3D 65536 # typically 8KB each > random_page_cost =3D 2 # units are one sequential page fetch cost > #----------------------------------------------------------------------- ---- > # ERROR REPORTING AND LOGGING > #----------------------------------------------------------------------- ---- > > log_min_duration_statement =3D 250 # -1 is disabled, in = milliseconds. > log_connections =3D true > log_disconnections =3D true > log_duration =3D true > log_line_prefix =3D '<%r%u%p%t%d%%' # e.g. '<%u%%%d> ' > # %u=3Duser name %d=3Ddatabase name > # %r=3Dremote host and port > # %p=3DPID %t=3Dtimestamp %i=3Dcommand tag > # %c=3Dsession id %l=3Dsession line number > # %s=3Dsession start timestamp %x=3Dtransaction id > # %q=3Dstop here in non-session processes > # %%=3D'%' > log_statement =3D 'none' # none, mod, ddl, all > #--------------------------------------------------------------------- > ------ > # RUNTIME STATISTICS > #----------------------------------------------------------------------- ---- > # - Query/Index Statistics Collector - > stats_start_collector =3D true > stats_command_string =3D true > stats_block_level =3D true > stats_row_level =3D true > stats_reset_on_server_start =3D true > > My MySQL *my.ini* has the following non default values:=20 > innodb_data_home_dir =3D /var/lib/mysql/ innodb_data_file_path =3D=20 > ibdata1:10M:autoextend innodb_log_group_home_dir =3D /var/lib/mysql/ > innodb_log_arch_dir =3D /var/lib/mysql/ > # You can set .._buffer_pool_size up to 50 - 80 % > # of RAM but beware of setting memory usage too high > innodb_buffer_pool_size =3D 512M > innodb_additional_mem_pool_size =3D 64M > # Set .._log_file_size to 25 % of buffer pool size > innodb_log_file_size =3D 128M > innodb_log_buffer_size =3D 64M > innodb_flush_log_at_trx_commit =3D 1 > innodb_lock_wait_timeout =3D 50 > innodb_flush_method =3D O_DSYNC > max_allowed_packet =3D 16M > > > From pgsql-performance-owner@postgresql.org Tue Jul 26 09:23:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8B4A752E5F for ; Tue, 26 Jul 2005 09:23:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92399-08 for ; Tue, 26 Jul 2005 12:22:58 +0000 (GMT) Received: from tbmail.tradebot.com (Tradebot-Systems-1096753.cust-rtr.swbell.net [68.90.170.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6F71152E47 for ; Tue, 26 Jul 2005 09:22:58 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Subject: Re: COPY insert performance X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Date: Tue, 26 Jul 2005 07:23:02 -0500 Message-ID: <07774C6E31D94A44A2A60E2085944F09071DF2@tbmail.tradebot.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] COPY insert performance Thread-Index: AcWRbc7D9PAWJ1M0SPemeI02RDv+BgAbrL7g From: "Chris Isaacson" To: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.967 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/383 X-Sequence-Number: 13624 John, (FYI: got a failed to deliver to john@arbash-meinel.com) I do not have any foreign keys and I need the indexes on during the insert/copy b/c in production a few queries heavily dependent on the indexes will be issued. These queries will be infrequent, but must be fast when issued. I am using InnoDB with MySQL which appears to enforce true transaction support. (http://dev.mysql.com/doc/mysql/en/innodb-overview.html) If not, how is InnoDB "cheating"? Sorry for the confusion, but pg_xlog is currently on a dedicated drive (10K SCSI, see below). Would I realize further gains if I had a third drive and put the indexes on that drive? =3D20 I've played with the checkpoint_segments. I noticed an enormous improvement increasing from the default to 40, but neglible improvement thereafter. Do you have a recommendation for a value? My bg_writer adjustments were a last ditch effort. I found your advice correct and realized no gain. I have not tested under a querying load which is a good next step. I had not thought of the comparative degradation of MySQL vs. PostgreSQL. Thanks for the tip on the RAM usage by indexes. I was under the incorrect assumption that shared_buffers would take care of this. I'll increase work_mem to 512MB and rerun my test. I have 1G of RAM, which is less than we'll be running in production (likely 2G). -Chris -----Original Message----- From: John A Meinel [mailto:john@arbash-meinel.com]=20 Sent: Monday, July 25, 2005 6:09 PM To: Chris Isaacson; Postgresql Performance Subject: Re: [PERFORM] COPY insert performance Chris Isaacson wrote: > I need COPY via libpqxx to insert millions of rows into two tables. =20 > One table has roughly have as many rows and requires half the storage. > In production, the largest table will grow by ~30M rows/day. To test=20 > the COPY performance I split my transactions into 10,000 rows. I=20 > insert roughly 5000 rows into table A for every 10,000 rows into table > B. > > Table A has one unique index: > > "order_main_pk" UNIQUE, btree (cl_ord_id) > > Table B has 1 unique index and 2 non-unique indexes: > > "order_transition_pk" UNIQUE, btree (collating_seq)=20 > "order_transition_ak2" btree (orig_cl_ord_id) "order_transition_ak3"=20 > btree (exec_id) Do you have any foreign key references? If you are creating a table for the first time (or loading a large fraction of the data), it is common to drop the indexes and foreign keys first, and then insert/copy, and then drop them again. Is InnoDB the backend with referential integrity, and true transaction support? I believe the default backend does not support either (so it is "cheating" to give you speed, which may be just fine for your needs, especially since you are willing to run fsync=3Dfalse). I think moving pg_xlog to a dedicated drive (set of drives) could help your performance. As well as increasing checkpoint_segments. I don't know if you gain much by changing the bg_writer settings, if you are streaming everything in at once, you probably want to have it written out right away. My understanding is that bg_writer settings are for the case where you have mixed read and writes going on at the same time, and you want to make sure that the reads have time to execute (ie the writes are not saturating your IO). Also, is any of this tested under load? Having a separate process issue queries while you are loading in data. Traditionally MySQL is faster with a single process inserting/querying for data, but once you have multiple processes hitting it at the same time, it's performance degrades much faster than postgres. You also seem to be giving MySQL 512M of ram to work with, while only giving 2M/200M to postgres. (re)creating indexes uses maintenance_work_mem, but updating indexes could easily use work_mem. You may be RAM starved. John =3D:-> > > My testing environment is as follows: > -Postgresql 8.0.1 > -libpqxx 2.5.0 > -Linux 2.6.11.4-21.7-smp x86_64 > -Dual Opteron 246 > -System disk (postgres data resides on this SCSI disk) - Seagate > (ST373453LC) - 15K, 73 GB > (http://www.seagate.com/cda/products/discsales/marketing/detail/0,1081 > ,549,00.html) > -2nd logical disk - 10K, 36GB IBM SCSI (IC35L036UCDY10-0) - WAL reside > on this disk > -NO RAID > > *PostgreSQL* > Here are the results of copying in 10M rows as fast as possible: > (10K/transaction) > Total Time: 1129.556 s > Rows/sec: 9899.922 > Transaction>1.2s 225 > Transaction>1.5s 77 > Transaction>2.0s 4 > Max Transaction 2.325s > > **MySQL** > **I ran a similar test with MySQL 4.1.10a (InnoDB) which produced=20 > these > results: (I used MySQL's INSERT INTO x VALUES > (1,2,3)(4,5,6)(...,...,...) syntax) (10K/transaction) > Total Time: 860.000 s > Rows/sec: 11627.91 > Transaction>1.2s 0 > Transaction>1.5s 0 > Transaction>2.0s 0 > Max Transaction 1.175s > > Considering the configurations shown below, can anyone offer advice to > close the 15% gap and the much worse variability I'm experiencing. =20 > Thanks > > My *postgresql.conf* has the following non-default values: > # ----------------------------- > # PostgreSQL configuration file > # ----------------------------- > listen_addresses =3D '*' # what IP interface(s) to listen on;=20 > max_connections =3D 100 > #--------------------------------------------------------------------- > ------ > # RESOURCE USAGE (except WAL) > #----------------------------------------------------------------------- ---- > shared_buffers =3D 65536 # min 16, at least max_connections*2, 8KB = each > work_mem =3D 2048 # min 64, size in KB > maintenance_work_mem =3D 204800 # min 1024, size in KB > max_fsm_pages =3D 2250000 # min max_fsm_relations*16, 6 bytes each > bgwriter_delay =3D 200 # 10-10000 milliseconds between rounds > bgwriter_percent =3D 10 # 0-100% of dirty buffers in each round > bgwriter_maxpages =3D 1000 # 0-1000 buffers max per round > #----------------------------------------------------------------------- ---- > # WRITE AHEAD LOG > #----------------------------------------------------------------------- ---- > fsync =3D false # turns forced synchronization on or off > wal_buffers =3D 64 # min 4, 8KB each > checkpoint_segments =3D 40 # in logfile segments, min 1, 16MB each > checkpoint_timeout =3D 600 # range 30-3600, in seconds > #----------------------------------------------------------------------- ---- > # QUERY TUNING > #----------------------------------------------------------------------- ---- > effective_cache_size =3D 65536 # typically 8KB each > random_page_cost =3D 2 # units are one sequential page fetch cost > #----------------------------------------------------------------------- ---- > # ERROR REPORTING AND LOGGING > #----------------------------------------------------------------------- ---- > > log_min_duration_statement =3D 250 # -1 is disabled, in = milliseconds. > log_connections =3D true > log_disconnections =3D true > log_duration =3D true > log_line_prefix =3D '<%r%u%p%t%d%%' # e.g. '<%u%%%d> ' > # %u=3Duser name %d=3Ddatabase name > # %r=3Dremote host and port > # %p=3DPID %t=3Dtimestamp %i=3Dcommand tag > # %c=3Dsession id %l=3Dsession line number > # %s=3Dsession start timestamp %x=3Dtransaction id > # %q=3Dstop here in non-session processes > # %%=3D'%' > log_statement =3D 'none' # none, mod, ddl, all > #--------------------------------------------------------------------- > ------ > # RUNTIME STATISTICS > #----------------------------------------------------------------------- ---- > # - Query/Index Statistics Collector - > stats_start_collector =3D true > stats_command_string =3D true > stats_block_level =3D true > stats_row_level =3D true > stats_reset_on_server_start =3D true > > My MySQL *my.ini* has the following non default values:=20 > innodb_data_home_dir =3D /var/lib/mysql/ innodb_data_file_path =3D=20 > ibdata1:10M:autoextend innodb_log_group_home_dir =3D /var/lib/mysql/ > innodb_log_arch_dir =3D /var/lib/mysql/ > # You can set .._buffer_pool_size up to 50 - 80 % > # of RAM but beware of setting memory usage too high > innodb_buffer_pool_size =3D 512M > innodb_additional_mem_pool_size =3D 64M > # Set .._log_file_size to 25 % of buffer pool size > innodb_log_file_size =3D 128M > innodb_log_buffer_size =3D 64M > innodb_flush_log_at_trx_commit =3D 1 > innodb_lock_wait_timeout =3D 50 > innodb_flush_method =3D O_DSYNC > max_allowed_packet =3D 16M > > > From pgsql-performance-owner@postgresql.org Tue Jul 26 09:27:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7CB6852E49 for ; Tue, 26 Jul 2005 09:27:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91839-10 for ; Tue, 26 Jul 2005 12:27:23 +0000 (GMT) Received: from tbmail.tradebot.com (Tradebot-Systems-1096753.cust-rtr.swbell.net [68.90.170.2]) by svr1.postgresql.org (Postfix) with ESMTP id CADBB52E7F for ; Tue, 26 Jul 2005 09:27:22 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Subject: Re: COPY insert performance X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Date: Tue, 26 Jul 2005 07:27:25 -0500 Message-ID: <07774C6E31D94A44A2A60E2085944F09071DF3@tbmail.tradebot.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] COPY insert performance Thread-Index: AcWR20nD+SAAWU0ETn6FROOSF6Su5wAAcBBg From: "Chris Isaacson" To: "Gavin Sherry" , X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.876 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/384 X-Sequence-Number: 13625 I need the chunks for each table COPYed within the same transaction which is why I'm not COPYing concurrently via multiple threads/processes. I will experiment w/o OID's and decreasing the shared_buffers and wal_buffers. Thanks, Chris -----Original Message----- From: Gavin Sherry [mailto:swm@alcove.com.au]=20 Sent: Tuesday, July 26, 2005 7:12 AM To: Chris Isaacson Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] COPY insert performance Hi Chris, Have you considered breaking the data into multiple chunks and COPYing each concurrently? Also, have you ensured that your table isn't storing OIDs? On Mon, 25 Jul 2005, Chris Isaacson wrote: > #--------------------------------------------------------------------- > -- > ---- > # RESOURCE USAGE (except WAL) > #----------------------------------------------------------------------- > ---- > shared_buffers =3D 65536 # min 16, at least max_connections*2, 8KB = each shared_buffers that high has been shown to affect performance. Try 12000. > wal_buffers =3D 64 # min 4, 8KB each Increasing wal_buffers can also have an effect on performance. Thanks, Gavin From pgsql-performance-owner@postgresql.org Tue Jul 26 11:23:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 811BC52ED7 for ; Tue, 26 Jul 2005 11:23:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 19932-09 for ; Tue, 26 Jul 2005 14:23:01 +0000 (GMT) Received: from sccrmhc14.comcast.net (sccrmhc14.comcast.net [204.127.202.59]) by svr1.postgresql.org (Postfix) with ESMTP id 1E2FF52EC6 for ; Tue, 26 Jul 2005 11:23:00 -0300 (ADT) Received: from jefftrout.com ([24.147.120.205]) by comcast.net (sccrmhc14) with SMTP id <2005072614225901400kb558e>; Tue, 26 Jul 2005 14:23:04 +0000 Received: (qmail 77981 invoked from network); 26 Jul 2005 14:29:18 -0000 Received: from c-24-147-120-205.hsd1.ma.comcast.net (HELO ?192.168.0.106?) (24.147.120.205) by 192.168.0.109 with SMTP; 26 Jul 2005 14:29:18 -0000 In-Reply-To: <07774C6E31D94A44A2A60E2085944F09071DF1@tbmail.tradebot.com> References: <07774C6E31D94A44A2A60E2085944F09071DF1@tbmail.tradebot.com> Mime-Version: 1.0 (Apple Message framework v733) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <795D0AC7-3C39-4CE3-A105-921F35BAC8DF@torgo.978.org> Cc: "John A Meinel" , Content-Transfer-Encoding: 7bit From: Jeff Trout Subject: Re: COPY insert performance Date: Tue, 26 Jul 2005 10:22:57 -0400 To: Chris Isaacson X-Mailer: Apple Mail (2.733) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/385 X-Sequence-Number: 13626 On Jul 26, 2005, at 8:15 AM, Chris Isaacson wrote: > > I am using InnoDB with MySQL which appears to enforce true transaction > support. (http://dev.mysql.com/doc/mysql/en/innodb-overview.html) If > not, how is InnoDB "cheating"? > are you sure your tables are innodb? chances are high unless you explcitly stated "type = innodb" when creating that they are myisam. look at "show table status" output to verify. > > I've played with the checkpoint_segments. I noticed an enormous > improvement increasing from the default to 40, but neglible > improvement > thereafter. Do you have a recommendation for a value? there's been a thread on -hackers recently about checkpoint issues.. in a nut shell there isn't much to do. But I'd say give bizgres a try if you're going to be continually loading huge amounts of data. -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Tue Jul 26 12:22:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C05C852BD5 for ; Tue, 26 Jul 2005 12:22:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 39330-07 for ; Tue, 26 Jul 2005 15:22:36 +0000 (GMT) Received: from yertle.kcilink.com (yertle.kcilink.com [65.205.34.180]) by svr1.postgresql.org (Postfix) with ESMTP id 989F252AAB for ; Tue, 26 Jul 2005 12:22:35 -0300 (ADT) Received: from [192.168.7.103] (host-103.int.kcilink.com [192.168.7.103]) by yertle.kcilink.com (Postfix) with ESMTP id D2E1BB80C; Tue, 26 Jul 2005 11:22:33 -0400 (EDT) In-Reply-To: <8446.1121799660@sss.pgh.pa.us> References: <1efd553a05071910427b290203@mail.gmail.com> <42DD3D55.9020508@commandprompt.com> <1efd553a050719112126b8d981@mail.gmail.com> <8446.1121799660@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v733) Content-Type: multipart/signed; micalg=sha1; boundary=Apple-Mail-4-202747623; protocol="application/pkcs7-signature" Message-Id: Cc: pgsql-performance@postgresql.org From: Vivek Khera Subject: Re: Looking for tips Date: Tue, 26 Jul 2005 11:22:31 -0400 To: Tom Lane X-Mailer: Apple Mail (2.733) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.021 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/386 X-Sequence-Number: 13627 --Apple-Mail-4-202747623 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed On Jul 19, 2005, at 3:01 PM, Tom Lane wrote: > You could possibly get some improvement if you can re-use prepared > plans > for the queries; but this will require some fooling with the client > code > (I'm not sure if DBD::Pg even has support for it at all). > DBD::Pg 1.40+ by default uses server-side prepared statements when you do $dbh->prepare() against an 8.x database server. Vivek Khera, Ph.D. +1-301-869-4449 x806 --Apple-Mail-4-202747623 Content-Transfer-Encoding: base64 Content-Type: application/pkcs7-signature; name=smime.p7s Content-Disposition: attachment; filename=smime.p7s MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIGhzCCAz8w ggKooAMCAQICAQ0wDQYJKoZIhvcNAQEFBQAwgdExCzAJBgNVBAYTAlpBMRUwEwYDVQQIEwxXZXN0 ZXJuIENhcGUxEjAQBgNVBAcTCUNhcGUgVG93bjEaMBgGA1UEChMRVGhhd3RlIENvbnN1bHRpbmcx KDAmBgNVBAsTH0NlcnRpZmljYXRpb24gU2VydmljZXMgRGl2aXNpb24xJDAiBgNVBAMTG1RoYXd0 ZSBQZXJzb25hbCBGcmVlbWFpbCBDQTErMCkGCSqGSIb3DQEJARYccGVyc29uYWwtZnJlZW1haWxA dGhhd3RlLmNvbTAeFw0wMzA3MTcwMDAwMDBaFw0xMzA3MTYyMzU5NTlaMGIxCzAJBgNVBAYTAlpB MSUwIwYDVQQKExxUaGF3dGUgQ29uc3VsdGluZyAoUHR5KSBMdGQuMSwwKgYDVQQDEyNUaGF3dGUg UGVyc29uYWwgRnJlZW1haWwgSXNzdWluZyBDQTCBnzANBgkqhkiG9w0BAQEFAAOBjQAwgYkCgYEA xKY8VXNV+065yplaHmjAdQRwnd/p/6Me7L3N9VvyGna9fww6YfK/Uc4B1OVQCjDXAmNaLIkVcI7d yfArhVqqP3FWy688Cwfn8R+RNiQqE88r1fOCdz0Dviv+uxg+B79AgAJk16emu59l0cUqVIUPSAR/ p7bRPGEEQB5kGXJgt/sCAwEAAaOBlDCBkTASBgNVHRMBAf8ECDAGAQH/AgEAMEMGA1UdHwQ8MDow OKA2oDSGMmh0dHA6Ly9jcmwudGhhd3RlLmNvbS9UaGF3dGVQZXJzb25hbEZyZWVtYWlsQ0EuY3Js MAsGA1UdDwQEAwIBBjApBgNVHREEIjAgpB4wHDEaMBgGA1UEAxMRUHJpdmF0ZUxhYmVsMi0xMzgw DQYJKoZIhvcNAQEFBQADgYEASIzRUIPqCy7MDaNmrGcPf6+svsIXoUOWlJ1/TCG4+DYfqi2fNi/A 9BxQIJNwPP2t4WFiw9k6GX6EsZkbAMUaC4J0niVQlGLH2ydxVyWN3amcOY6MIE9lX5Xa9/eH1sYI Tq726jTlEBpbNU1341YheILcIRk13iSx0x1G/11fZU8wggNAMIICqaADAgECAgMOah8wDQYJKoZI hvcNAQEEBQAwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkp IEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFpbCBJc3N1aW5nIENBMB4XDTA1 MDQwNTIwMzEzMloXDTA2MDQwNTIwMzEzMlowgYoxHzAdBgNVBAMTFlRoYXd0ZSBGcmVlbWFpbCBN ZW1iZXIxHjAcBgkqhkiG9w0BCQEWD3ZpdmVrQGtoZXJhLm9yZzEgMB4GCSqGSIb3DQEJARYRa2hl cmFAa2NpbGluay5jb20xJTAjBgkqhkiG9w0BCQEWFnZpdmVrQG1haWxlcm1haWxlci5jb20wggEi MA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQDdo7hywGcY0VvK6WqqXXV77MS/t/4X3WkCaCXo RSl2W58GP4P21hodPn7hlIxUoDOW7x9O+FbqTgE2Ejqr6yA00Mm90tGPFgjFjqPGAqg7xk6IDcv9 uTyMia/FKEHSIynM6zqokXY8JklvdbJOiByE/8VeyEXOANWiflo8o4+GHnhMKpA9982YTXUqeKU6 mMQVaLCBRjTDc7j2XkMC/UNcp2HMyDQdTqYVnhLxbvbLX8CNDBY/7OWFlB9evru46SpGWhe4lhv5 DSgE2RdCKvDytzxRDvP49L8V0TnFjAVeC1C1Pj0/KQsoL/AP4APplROiD4QaUhshQl28pXxJtfbl AgMBAAGjVzBVMEUGA1UdEQQ+MDyBD3ZpdmVrQGtoZXJhLm9yZ4ERa2hlcmFAa2NpbGluay5jb22B FnZpdmVrQG1haWxlcm1haWxlci5jb20wDAYDVR0TAQH/BAIwADANBgkqhkiG9w0BAQQFAAOBgQAr CWop3h28qPwofzLrkoT410J4d7Bqk6FLeVlKZfg/wXlS1MTqYMNcCm4x+JsJbjwsO0fb2elFIuGq 1razoSzPpgi89itydvUT0U0U/u+AkZA5rW4AptTpMZ70YW5u9wzkcvmifqZmcfbaaeGdZfruzUXZ 6qvdXDpNb3ZHeQw6PjGCAucwggLjAgEBMGkwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0 ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFp bCBJc3N1aW5nIENBAgMOah8wCQYFKw4DAhoFAKCCAVMwGAYJKoZIhvcNAQkDMQsGCSqGSIb3DQEH ATAcBgkqhkiG9w0BCQUxDxcNMDUwNzI2MTUyMjMyWjAjBgkqhkiG9w0BCQQxFgQUonRq/1/pmgZu mGszJ53kqbdEpS0weAYJKwYBBAGCNxAEMWswaTBiMQswCQYDVQQGEwJaQTElMCMGA1UEChMcVGhh d3RlIENvbnN1bHRpbmcgKFB0eSkgTHRkLjEsMCoGA1UEAxMjVGhhd3RlIFBlcnNvbmFsIEZyZWVt YWlsIElzc3VpbmcgQ0ECAw5qHzB6BgsqhkiG9w0BCRACCzFroGkwYjELMAkGA1UEBhMCWkExJTAj BgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJz b25hbCBGcmVlbWFpbCBJc3N1aW5nIENBAgMOah8wDQYJKoZIhvcNAQEBBQAEggEAfauwaZUZzZiK nvKzrvVLVho7FCvMl3TSs4u0uBhVK4zP/ZxsIwyS/ecYdD0QFxMw+4ftawmNmu/0jsjhxDzyhcjz DgcEOCVy9FsbCurvaeYDWXj+FbTR9VMzYPk39rHZ3YmXb6ZqXJHVkQ8Tg/N6TU23EZCjc6iSqZXT NRZXanaWS7Rb9awwG0VUow1wDg3uy4z1OP4onaTmOGI5EFhKzD4XcQ8U//Jn0pC3HIE+4+yLVJSB IMiV437bMzGrYsjxBKZUnyvNMr0z3py5VfLbqNdj+c6BK9mArNW59WPNcS7YBuUQQv3UiQbe0kpC DBnJQhRp1Izuf7r4+H97kL8zvgAAAAAAAA== --Apple-Mail-4-202747623-- From pgsql-performance-owner@postgresql.org Tue Jul 26 13:34:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 486B752E65 for ; Tue, 26 Jul 2005 13:34:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76212-05 for ; Tue, 26 Jul 2005 16:34:47 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id CAC3352C76 for ; Tue, 26 Jul 2005 13:34:45 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050726163441m9200g48ije>; Tue, 26 Jul 2005 16:34:41 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 4093C5600C; Tue, 26 Jul 2005 11:34:41 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 29FDB55FB5 for ; Tue, 26 Jul 2005 11:34:36 -0500 (CDT) Message-ID: <42E6661D.3060304@arbash-meinel.com> Date: Tue, 26 Jul 2005 11:34:37 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Postgresql Performance Subject: Cheap RAM disk? X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig6935037441687B1F4F5B8D62" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/387 X-Sequence-Number: 13628 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig6935037441687B1F4F5B8D62 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit I saw a review of a relatively inexpensive RAM disk over at anandtech.com, the Gigabyte i-RAM http://www.anandtech.com/storage/showdoc.aspx?i=2480 Basically, it is a PCI card, which takes standard DDR RAM, and has a SATA port on it, so that to the system, it looks like a normal SATA drive. The card costs about $100-150, and you fill it with your own ram, so for a 4GB (max size) disk, it costs around $500. Looking for solid state storage devices, the cheapest I found was around $5k for 2GB. Gigabyte claims that the battery backup can last up to 16h, which seems decent, if not really long (the $5k solution has a built-in harddrive so that if the power goes out, it uses the battery power to copy the ramdisk onto the harddrive for more permanent storage). Anyway, would something like this be reasonable as a drive for storing pg_xlog? With 4GB you could have as many as 256 checkpoint segments. I'm a little leary as it is definitely a version 1.0 product (it is still using an FPGA as the controller, so they were obviously pushing to get the card into production). But it seems like this might be a decent way to improve insert performance, without setting fsync=false. Probably it should see some serious testing (as in power spikes/pulled plugs, etc). I know the article made some claim that if you actually pull out the card it goes into "high consumption mode" which is somehow greater than if you leave it in the slot with the power off. Which to me seems like a lot of bull, and really means the 16h is only under best-case circumstances. But even 1-2h is sufficient to handle a simple power outage. And if you had a UPS with detection of power failure, you could always sync the ramdisk to a local partition before the power goes out. Though you could do that with a normal in-memory ramdisk (tmpfs) without having to buy the card. Though it does give you up-to an extra 4GB of ram, for machines which have already maxed out their slots. Anyway, I thought I would mention it to the list, to see if anyone else has heard of it, or has any thoughts on the matter. I'm sure there are some people who are using more expensive ram disks, maybe they have some ideas about what this device is missing. (other than costing about 1/10th the price) John =:-> --------------enig6935037441687B1F4F5B8D62 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5mYdJdeBCYSNAAMRAiS/AJ4oTdCEuFAKvdZg9ndBJRbucCqk7ACgszi/ bWcE7rGEDR8YAleWF+GYnSI= =Ef6t -----END PGP SIGNATURE----- --------------enig6935037441687B1F4F5B8D62-- From pgsql-performance-owner@postgresql.org Tue Jul 26 13:49:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C23F252AAB for ; Tue, 26 Jul 2005 13:49:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86783-01 for ; Tue, 26 Jul 2005 16:49:20 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 1640152E8B for ; Tue, 26 Jul 2005 13:49:18 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id 8AEB4644158 for ; Tue, 26 Jul 2005 10:48:51 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28894-10 for ; Tue, 26 Jul 2005 10:48:49 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id 90BDA644121 for ; Tue, 26 Jul 2005 10:48:49 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) Content-Transfer-Encoding: 7bit Message-Id: Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: Dan Harris Subject: faster INSERT with possible pre-existing row? Date: Tue, 26 Jul 2005 10:50:14 -0600 X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/388 X-Sequence-Number: 13629 I am working on a process that will be inserting tens of million rows and need this to be as quick as possible. The catch is that for each row I could potentially insert, I need to look and see if the relationship is already there to prevent multiple entries. Currently I am doing a SELECT before doing the INSERT, but I recognize the speed penalty in doing to operations. I wonder if there is some way I can say "insert this record, only if it doesn't exist already". To see if it exists, I would need to compare 3 fields instead of just enforcing a primary key. Even if this could be a small increase per record, even a few percent faster compounded over the whole load could be a significant reduction. Thanks for any ideas you might have. -Dan From pgsql-performance-owner@postgresql.org Tue Jul 26 14:04:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B706752E6A for ; Tue, 26 Jul 2005 14:04:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 89785-06 for ; Tue, 26 Jul 2005 17:04:05 +0000 (GMT) Received: from floppy.pyrenet.fr (news.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id AA24852C9E for ; Tue, 26 Jul 2005 14:04:04 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id DCB2B30B42; Tue, 26 Jul 2005 19:11:18 +0200 (MET DST) From: Chris Browne X-Newsgroups: pgsql.performance Subject: Re: Cheap RAM disk? Date: Tue, 26 Jul 2005 12:51:14 -0400 Organization: cbbrowne Computing Inc Lines: 64 Message-ID: <60u0ih1pfh.fsf@dba2.int.libertyrms.com> References: <42E6661D.3060304@arbash-meinel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Complaints-To: usenet@news.hub.org User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4.17 (Jumbo Shrimp, linux) Cancel-Lock: sha1:9/1Xq5tnQ92R1zfjK80854jEktA= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.124 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/389 X-Sequence-Number: 13630 john@arbash-meinel.com (John A Meinel) writes: > I saw a review of a relatively inexpensive RAM disk over at > anandtech.com, the Gigabyte i-RAM > http://www.anandtech.com/storage/showdoc.aspx?i=2480 And the review shows that it's not *all* that valuable for many of the cases they looked at. > Basically, it is a PCI card, which takes standard DDR RAM, and has a > SATA port on it, so that to the system, it looks like a normal SATA > drive. > > The card costs about $100-150, and you fill it with your own ram, so > for a 4GB (max size) disk, it costs around $500. Looking for solid > state storage devices, the cheapest I found was around $5k for 2GB. > > Gigabyte claims that the battery backup can last up to 16h, which > seems decent, if not really long (the $5k solution has a built-in > harddrive so that if the power goes out, it uses the battery power to > copy the ramdisk onto the harddrive for more permanent storage). > > Anyway, would something like this be reasonable as a drive for storing > pg_xlog? With 4GB you could have as many as 256 checkpoint segments. > > I'm a little leary as it is definitely a version 1.0 product (it is > still using an FPGA as the controller, so they were obviously pushing > to get the card into production). What disappoints me is that nobody has tried the CF/RAM answer; rather than putting a hard drive on the board, you put on some form of flash device (CompactFlash or such), where if power fails, it pushes data onto the CF. That ought to be cheaper (both in terms of hardware cost and power consumption) than using a hard disk. > But it seems like this might be a decent way to improve insert > performance, without setting fsync=false. That's the case which might prove Ludicrously Quicker than any of the sample cases in the review. > Probably it should see some serious testing (as in power spikes/pulled > plugs, etc). I know the article made some claim that if you actually > pull out the card it goes into "high consumption mode" which is > somehow greater than if you leave it in the slot with the power > off. Which to me seems like a lot of bull, and really means the 16h is > only under best-case circumstances. But even 1-2h is sufficient to > handle a simple power outage. Certainly. > Anyway, I thought I would mention it to the list, to see if anyone > else has heard of it, or has any thoughts on the matter. I'm sure > there are some people who are using more expensive ram disks, maybe > they have some ideas about what this device is missing. (other than > costing about 1/10th the price) Well, if it hits a "2.0" version, it may get interesting... -- (format nil "~S@~S" "cbbrowne" "acm.org") http://www.ntlug.org/~cbbrowne/sap.html Rules of the Evil Overlord #78. "I will not tell my Legions of Terror "And he must be taken alive!" The command will be: ``And try to take him alive if it is reasonably practical.''" From pgsql-performance-owner@postgresql.org Tue Jul 26 14:20:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 71EAC52E88 for ; Tue, 26 Jul 2005 14:20:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92700-06 for ; Tue, 26 Jul 2005 17:20:26 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id B112E52EA2 for ; Tue, 26 Jul 2005 14:20:24 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050726172024m9200g48k4e>; Tue, 26 Jul 2005 17:20:24 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 4176156011; Tue, 26 Jul 2005 11:56:26 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 2307E55FB5; Tue, 26 Jul 2005 11:56:15 -0500 (CDT) Message-ID: <42E66B30.3030108@arbash-meinel.com> Date: Tue, 26 Jul 2005 11:56:16 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: faster INSERT with possible pre-existing row? References: In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig9C21B405ABDA157CADDE32ED" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.045 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/390 X-Sequence-Number: 13631 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig9C21B405ABDA157CADDE32ED Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Dan Harris wrote: > I am working on a process that will be inserting tens of million rows > and need this to be as quick as possible. > > The catch is that for each row I could potentially insert, I need to > look and see if the relationship is already there to prevent multiple > entries. Currently I am doing a SELECT before doing the INSERT, but I > recognize the speed penalty in doing to operations. I wonder if there > is some way I can say "insert this record, only if it doesn't exist > already". To see if it exists, I would need to compare 3 fields > instead of just enforcing a primary key. > > Even if this could be a small increase per record, even a few percent > faster compounded over the whole load could be a significant reduction. > > Thanks for any ideas you might have. > > -Dan > You could insert all of your data into a temporary table, and then do: INSERT INTO final_table SELECT * FROM temp_table WHERE NOT EXISTS (SELECT info FROM final_table WHERE id=id, path=path, y=y); Or you could load it into the temporary table, and then: DELETE FROM temp_table WHERE EXISTS (SELECT FROM final_table WHERE id...); And then do a plain INSERT INTO. I can't say what the specific performance increases would be, but temp_table could certainly be an actual TEMP table (meaning it only exists during the connection), and you could easily do a COPY into that table to load it up quickly, without having to check any constraints. Just a thought, John =:-> --------------enig9C21B405ABDA157CADDE32ED Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5mswJdeBCYSNAAMRAupDAKCrP6dp0q0HGvri2TlUZfeg/zZENwCcDUoS OuTezkceW29aUAbY/2rAe9k= =21Zo -----END PGP SIGNATURE----- --------------enig9C21B405ABDA157CADDE32ED-- From pgsql-performance-owner@postgresql.org Tue Jul 26 14:43:12 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CFFDC52EB4 for ; Tue, 26 Jul 2005 14:43:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95343-10 for ; Tue, 26 Jul 2005 17:43:09 +0000 (GMT) Received: from gghcwest.com (adsl-71-128-90-172.dsl.pltn13.pacbell.net [71.128.90.172]) by svr1.postgresql.org (Postfix) with ESMTP id 1D32652CB6 for ; Tue, 26 Jul 2005 14:43:08 -0300 (ADT) Received: from toonses.gghcwest.com (toonses.gghcwest.com [192.168.168.115]) by gghcwest.com (8.12.10/8.12.9) with ESMTP id j6QHgTmd017672; Tue, 26 Jul 2005 10:42:30 -0700 Received: from jwb by toonses.gghcwest.com with local (Exim 4.50) id 1DxTRc-0003KL-JI; Tue, 26 Jul 2005 10:42:20 -0700 Subject: Re: Cheap RAM disk? From: "Jeffrey W. Baker" To: John A Meinel Cc: Postgresql Performance In-Reply-To: <42E6661D.3060304@arbash-meinel.com> References: <42E6661D.3060304@arbash-meinel.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Tue, 26 Jul 2005 10:42:19 -0700 Message-Id: <1122399740.12728.1.camel@toonses.gghcwest.com> Mime-Version: 1.0 X-Mailer: Evolution 2.3.5.1 X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/391 X-Sequence-Number: 13632 On Tue, 2005-07-26 at 11:34 -0500, John A Meinel wrote: > I saw a review of a relatively inexpensive RAM disk over at > anandtech.com, the Gigabyte i-RAM > http://www.anandtech.com/storage/showdoc.aspx?i=2480 > > Basically, it is a PCI card, which takes standard DDR RAM, and has a > SATA port on it, so that to the system, it looks like a normal SATA drive. > > The card costs about $100-150, and you fill it with your own ram, so for > a 4GB (max size) disk, it costs around $500. Looking for solid state > storage devices, the cheapest I found was around $5k for 2GB. > > Gigabyte claims that the battery backup can last up to 16h, which seems > decent, if not really long (the $5k solution has a built-in harddrive so > that if the power goes out, it uses the battery power to copy the > ramdisk onto the harddrive for more permanent storage). > > Anyway, would something like this be reasonable as a drive for storing > pg_xlog? With 4GB you could have as many as 256 checkpoint segments. I haven't tried this product, but the microbenchmarks seem truly slow. I think you would get a similar benefit by simply sticking a 1GB or 2GB DIMM -- battery-backed, of course -- in your RAID controller. -jwb From pgsql-performance-owner@postgresql.org Tue Jul 26 14:49:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 870FC52CB6 for ; Tue, 26 Jul 2005 14:49:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 97517-09 for ; Tue, 26 Jul 2005 17:49:06 +0000 (GMT) Received: from yertle.kcilink.com (yertle.kcilink.com [65.205.34.180]) by svr1.postgresql.org (Postfix) with ESMTP id 3AF7152E65 for ; Tue, 26 Jul 2005 14:49:05 -0300 (ADT) Received: from [192.168.7.103] (host-103.int.kcilink.com [192.168.7.103]) by yertle.kcilink.com (Postfix) with ESMTP id 1FEF3B80C for ; Tue, 26 Jul 2005 13:49:06 -0400 (EDT) Mime-Version: 1.0 (Apple Message framework v733) In-Reply-To: <42E6661D.3060304@arbash-meinel.com> References: <42E6661D.3060304@arbash-meinel.com> X-Gpgmail-State: !signed Content-Type: multipart/signed; micalg=sha1; boundary=Apple-Mail-7-211540927; protocol="application/pkcs7-signature" Message-Id: <37F30C1F-1EEB-4705-B074-9E4469A1CBC6@khera.org> From: Vivek Khera Subject: Re: Cheap RAM disk? Date: Tue, 26 Jul 2005 13:49:05 -0400 To: Postgresql Performance X-Mailer: Apple Mail (2.733) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.02 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/392 X-Sequence-Number: 13633 --Apple-Mail-7-211540927 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed On Jul 26, 2005, at 12:34 PM, John A Meinel wrote: > Basically, it is a PCI card, which takes standard DDR RAM, and has > a SATA port on it, so that to the system, it looks like a normal > SATA drive. > > The card costs about $100-150, and you fill it with your own ram, > so for a 4GB (max size) disk, it costs around $500. Looking for > solid state storage devices, the cheapest I found was around $5k > for 2GB. > gotta love /. don't ya? This card doesn't accept ECC RAM therefore it is nothing more than a toy. I wouldn't trust it as far as I could throw it. There are other vendors of SSD's out there. Some even have *real* power fail strategies such as dumping to a physical disk. These are not cheap, but you gets what ya pays for... Vivek Khera, Ph.D. +1-301-869-4449 x806 --Apple-Mail-7-211540927 Content-Transfer-Encoding: base64 Content-Type: application/pkcs7-signature; name=smime.p7s Content-Disposition: attachment; filename=smime.p7s MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIIGhzCCAz8w ggKooAMCAQICAQ0wDQYJKoZIhvcNAQEFBQAwgdExCzAJBgNVBAYTAlpBMRUwEwYDVQQIEwxXZXN0 ZXJuIENhcGUxEjAQBgNVBAcTCUNhcGUgVG93bjEaMBgGA1UEChMRVGhhd3RlIENvbnN1bHRpbmcx KDAmBgNVBAsTH0NlcnRpZmljYXRpb24gU2VydmljZXMgRGl2aXNpb24xJDAiBgNVBAMTG1RoYXd0 ZSBQZXJzb25hbCBGcmVlbWFpbCBDQTErMCkGCSqGSIb3DQEJARYccGVyc29uYWwtZnJlZW1haWxA dGhhd3RlLmNvbTAeFw0wMzA3MTcwMDAwMDBaFw0xMzA3MTYyMzU5NTlaMGIxCzAJBgNVBAYTAlpB MSUwIwYDVQQKExxUaGF3dGUgQ29uc3VsdGluZyAoUHR5KSBMdGQuMSwwKgYDVQQDEyNUaGF3dGUg UGVyc29uYWwgRnJlZW1haWwgSXNzdWluZyBDQTCBnzANBgkqhkiG9w0BAQEFAAOBjQAwgYkCgYEA xKY8VXNV+065yplaHmjAdQRwnd/p/6Me7L3N9VvyGna9fww6YfK/Uc4B1OVQCjDXAmNaLIkVcI7d yfArhVqqP3FWy688Cwfn8R+RNiQqE88r1fOCdz0Dviv+uxg+B79AgAJk16emu59l0cUqVIUPSAR/ p7bRPGEEQB5kGXJgt/sCAwEAAaOBlDCBkTASBgNVHRMBAf8ECDAGAQH/AgEAMEMGA1UdHwQ8MDow OKA2oDSGMmh0dHA6Ly9jcmwudGhhd3RlLmNvbS9UaGF3dGVQZXJzb25hbEZyZWVtYWlsQ0EuY3Js MAsGA1UdDwQEAwIBBjApBgNVHREEIjAgpB4wHDEaMBgGA1UEAxMRUHJpdmF0ZUxhYmVsMi0xMzgw DQYJKoZIhvcNAQEFBQADgYEASIzRUIPqCy7MDaNmrGcPf6+svsIXoUOWlJ1/TCG4+DYfqi2fNi/A 9BxQIJNwPP2t4WFiw9k6GX6EsZkbAMUaC4J0niVQlGLH2ydxVyWN3amcOY6MIE9lX5Xa9/eH1sYI Tq726jTlEBpbNU1341YheILcIRk13iSx0x1G/11fZU8wggNAMIICqaADAgECAgMOah8wDQYJKoZI hvcNAQEEBQAwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkp IEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFpbCBJc3N1aW5nIENBMB4XDTA1 MDQwNTIwMzEzMloXDTA2MDQwNTIwMzEzMlowgYoxHzAdBgNVBAMTFlRoYXd0ZSBGcmVlbWFpbCBN ZW1iZXIxHjAcBgkqhkiG9w0BCQEWD3ZpdmVrQGtoZXJhLm9yZzEgMB4GCSqGSIb3DQEJARYRa2hl cmFAa2NpbGluay5jb20xJTAjBgkqhkiG9w0BCQEWFnZpdmVrQG1haWxlcm1haWxlci5jb20wggEi MA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIBAQDdo7hywGcY0VvK6WqqXXV77MS/t/4X3WkCaCXo RSl2W58GP4P21hodPn7hlIxUoDOW7x9O+FbqTgE2Ejqr6yA00Mm90tGPFgjFjqPGAqg7xk6IDcv9 uTyMia/FKEHSIynM6zqokXY8JklvdbJOiByE/8VeyEXOANWiflo8o4+GHnhMKpA9982YTXUqeKU6 mMQVaLCBRjTDc7j2XkMC/UNcp2HMyDQdTqYVnhLxbvbLX8CNDBY/7OWFlB9evru46SpGWhe4lhv5 DSgE2RdCKvDytzxRDvP49L8V0TnFjAVeC1C1Pj0/KQsoL/AP4APplROiD4QaUhshQl28pXxJtfbl AgMBAAGjVzBVMEUGA1UdEQQ+MDyBD3ZpdmVrQGtoZXJhLm9yZ4ERa2hlcmFAa2NpbGluay5jb22B FnZpdmVrQG1haWxlcm1haWxlci5jb20wDAYDVR0TAQH/BAIwADANBgkqhkiG9w0BAQQFAAOBgQAr CWop3h28qPwofzLrkoT410J4d7Bqk6FLeVlKZfg/wXlS1MTqYMNcCm4x+JsJbjwsO0fb2elFIuGq 1razoSzPpgi89itydvUT0U0U/u+AkZA5rW4AptTpMZ70YW5u9wzkcvmifqZmcfbaaeGdZfruzUXZ 6qvdXDpNb3ZHeQw6PjGCAucwggLjAgEBMGkwYjELMAkGA1UEBhMCWkExJTAjBgNVBAoTHFRoYXd0 ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJzb25hbCBGcmVlbWFp bCBJc3N1aW5nIENBAgMOah8wCQYFKw4DAhoFAKCCAVMwGAYJKoZIhvcNAQkDMQsGCSqGSIb3DQEH ATAcBgkqhkiG9w0BCQUxDxcNMDUwNzI2MTc0OTA1WjAjBgkqhkiG9w0BCQQxFgQUkIy/mRJYKQip 7aDFeY+Wc+42BU8weAYJKwYBBAGCNxAEMWswaTBiMQswCQYDVQQGEwJaQTElMCMGA1UEChMcVGhh d3RlIENvbnN1bHRpbmcgKFB0eSkgTHRkLjEsMCoGA1UEAxMjVGhhd3RlIFBlcnNvbmFsIEZyZWVt YWlsIElzc3VpbmcgQ0ECAw5qHzB6BgsqhkiG9w0BCRACCzFroGkwYjELMAkGA1UEBhMCWkExJTAj BgNVBAoTHFRoYXd0ZSBDb25zdWx0aW5nIChQdHkpIEx0ZC4xLDAqBgNVBAMTI1RoYXd0ZSBQZXJz b25hbCBGcmVlbWFpbCBJc3N1aW5nIENBAgMOah8wDQYJKoZIhvcNAQEBBQAEggEAr2H8GOSL5I5D QNozxdVqXxGqqRjGvXLD+yKkXIzgbTNiG8W0Bwiu+T+MOatLLQsDfXAq4QQQUuqT24HKebLQ++Qz 0UV9v9mHEfQMMsUrBrNYBfCt2Kgdr2YU9xGVIairvCv2pTH4m0haFWrokzg2FX/sH45WDMIe7EYS 6Tb09wt8CSQpRodi8ROiACjwRn6imSLyrnMWkeG2K9AG3HFXuN0osv9/CZbrx0W15pI+vWTOTywg 6RWnw7vJAjd5CPSla5hQWoN9s3r6LEpT/iasOwa/k+apw2lPmRQPfNONpcjZHJN6GkvI9+u6InbP y1delMuZ20xdrYYk2v5xDFDxPwAAAAAAAA== --Apple-Mail-7-211540927-- From pgsql-performance-owner@postgresql.org Tue Jul 26 15:17:01 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 14FB852C24 for ; Tue, 26 Jul 2005 15:16:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 05932-05 for ; Tue, 26 Jul 2005 18:16:57 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id E6E9E52C3A for ; Tue, 26 Jul 2005 15:16:56 -0300 (ADT) Received: (qmail 27005 invoked from network); 26 Jul 2005 20:16:59 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 26 Jul 2005 20:16:59 +0200 To: "John A Meinel" , "Postgresql Performance" Subject: Re: Cheap RAM disk? References: <42E6661D.3060304@arbash-meinel.com> Message-ID: Date: Tue, 26 Jul 2005 20:16:59 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <42E6661D.3060304@arbash-meinel.com> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/393 X-Sequence-Number: 13634 > I'm a little leary as it is definitely a version 1.0 product (it is > still using an FPGA as the controller, so they were obviously pushing to > get the card into production). Not necessarily. FPGA's have become a sensible choice now. My RME studio soundcard uses a big FPGA. The performance in the test doesn't look that good, though, but don't forget it was run under windows. For instance they get 77s to copy the Firefox source tree on their Athlon 64/raptor ; my Duron / 7200rpm ide drive does it in 30 seconds, but not with windows of course. However it doesnt' use ECC so... That's a pity, because they could have implemented ECC in "software" inside the chip, and have the benefits of error correction with normal, cheap RAM. Well; wait and see... From pgsql-performance-owner@postgresql.org Tue Jul 26 15:23:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9A9B452AE5 for ; Tue, 26 Jul 2005 15:23:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07828-04 for ; Tue, 26 Jul 2005 18:23:31 +0000 (GMT) Received: from mail.mi8.com (d01gw04.mi8.com [63.240.6.44]) by svr1.postgresql.org (Postfix) with ESMTP id 12E55529E1 for ; Tue, 26 Jul 2005 15:23:30 -0300 (ADT) Received: from 172.16.1.110 by mail.mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D4)); Tue, 26 Jul 2005 14:23:24 -0400 X-Server-Uuid: C8FB4D43-1108-484A-A898-3CBCC7906230 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Tue, 26 Jul 2005 14:23:23 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Tue, 26 Jul 2005 14:23:22 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Tue, 26 Jul 2005 11:23:23 -0700 Subject: Re: Cheap RAM disk? From: "Luke Lonergan" To: "John A Meinel" , "Postgresql Performance" Message-ID: In-Reply-To: <42E6661D.3060304@arbash-meinel.com> MIME-Version: 1.0 X-OriginalArrivalTime: 26 Jul 2005 18:23:23.0642 (UTC) FILETIME=[1BAD15A0:01C5920F] X-WSS-ID: 6EF8A0112VG209975-03-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.578 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/394 X-Sequence-Number: 13635 Yup - interesting and very niche product - it seems like it's only obvious application is for the Postgresql WAL problem :-) The real differentiator is the battery backup part. Otherwise, the filesystem caching is more effective, so put the RAM on the motherboard. - Luke From pgsql-performance-owner@postgresql.org Tue Jul 26 15:28:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 16D2F52C37 for ; Tue, 26 Jul 2005 15:28:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 08000-04 for ; Tue, 26 Jul 2005 18:28:08 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id A7F2B52BDC for ; Tue, 26 Jul 2005 15:28:06 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050726182807m9200g4722e>; Tue, 26 Jul 2005 18:28:07 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 3D8F45600C; Tue, 26 Jul 2005 13:28:07 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id C1BFC55FB5; Tue, 26 Jul 2005 13:28:02 -0500 (CDT) Message-ID: <42E680B3.6020303@arbash-meinel.com> Date: Tue, 26 Jul 2005 13:28:03 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Luke Lonergan Cc: Postgresql Performance Subject: Re: Cheap RAM disk? References: In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig20AADE360217A01B16CEDF6E" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.044 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/395 X-Sequence-Number: 13636 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig20AADE360217A01B16CEDF6E Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Luke Lonergan wrote: > Yup - interesting and very niche product - it seems like it's only obvious > application is for the Postgresql WAL problem :-) Well, you could do it for any journaled system (XFS, JFS, ext3, reiserfs). But yes, it seems specifically designed for a battery backed journal. Though the article reviews it for very different purposes. Though it was a Windows review, and I don't know of any way to make NTFS use a separate device for a journal. (Though I expect it is possible somehow). John =:-> > > The real differentiator is the battery backup part. Otherwise, the > filesystem caching is more effective, so put the RAM on the motherboard. > > - Luke > --------------enig20AADE360217A01B16CEDF6E Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5oC0JdeBCYSNAAMRAlkaAKDNKALSqDb0MRL9daxin3P/bDxhLgCgwJ0S zqcW3/c5LLJCNi6cvF11nIg= =OqSJ -----END PGP SIGNATURE----- --------------enig20AADE360217A01B16CEDF6E-- From pgsql-performance-owner@postgresql.org Tue Jul 26 16:27:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7842152CA1 for ; Tue, 26 Jul 2005 16:27:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 57959-03 for ; Tue, 26 Jul 2005 19:27:16 +0000 (GMT) Received: from vms042pub.verizon.net (vms042pub.verizon.net [206.46.252.42]) by svr1.postgresql.org (Postfix) with ESMTP id 341B152C87 for ; Tue, 26 Jul 2005 16:27:15 -0300 (ADT) Received: from osgiliath.mathom.us ([70.108.53.154]) by vms042.mailsrvcs.net (Sun Java System Messaging Server 6.2 HotFix 0.04 (built Dec 24 2004)) with ESMTPA id <0IK900KOO21CJ9CH@vms042.mailsrvcs.net> for pgsql-performance@postgresql.org; Tue, 26 Jul 2005 14:27:13 -0500 (CDT) Received: from localhost (localhost [127.0.0.1]) by osgiliath.mathom.us (Postfix) with ESMTP id 292246247A1 for ; Tue, 26 Jul 2005 14:33:43 -0400 (EDT) Received: from osgiliath.mathom.us ([127.0.0.1]) by localhost (osgiliath [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 03245-01-16 for ; Tue, 26 Jul 2005 14:33:43 -0400 (EDT) Received: by osgiliath.mathom.us (Postfix, from userid 1000) id 10D43623284; Tue, 26 Jul 2005 14:33:43 -0400 (EDT) Date: Tue, 26 Jul 2005 14:33:43 -0400 From: Michael Stone Subject: Re: Cheap RAM disk? In-reply-to: To: pgsql-performance@postgresql.org Mail-Followup-To: pgsql-performance@postgresql.org Message-id: <20050726183343.GZ19080@mathom.us> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-disposition: inline X-Pgp-Fingerprint: 53 FF 38 00 E7 DD 0A 9C 84 52 84 C5 EE DF 7C 88 X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at mathom.us References: <42E6661D.3060304@arbash-meinel.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.008 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/401 X-Sequence-Number: 13642 On Tue, Jul 26, 2005 at 11:23:23AM -0700, Luke Lonergan wrote: >Yup - interesting and very niche product - it seems like it's only obvious >application is for the Postgresql WAL problem :-) On the contrary--it's not obvious that it is an ideal fit for a WAL. A ram disk like this is optimized for highly random access applications. The WAL is a single sequential writer. If you're in the kind of market that needs a really high performance WAL you'd be much better served by putting a big NVRAM cache in front of a fast disk array than by buying a toy like this. Mike Stone From pgsql-performance-owner@postgresql.org Tue Jul 26 15:47:04 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3F3AC52940 for ; Tue, 26 Jul 2005 15:47:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 13410-09 for ; Tue, 26 Jul 2005 18:46:57 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.50]) by svr1.postgresql.org (Postfix) with ESMTP id CD9D352C90 for ; Tue, 26 Jul 2005 15:46:54 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Tue, 26 Jul 2005 14:46:33 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Tue, 26 Jul 2005 14:46:34 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Tue, 26 Jul 2005 14:46:33 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Tue, 26 Jul 2005 11:46:33 -0700 Subject: Re: faster INSERT with possible pre-existing row? From: "Luke Lonergan" To: "John A Meinel" , "Dan Harris" Cc: pgsql-performance@postgresql.org, "bizgres-general" Message-ID: In-Reply-To: <42E66B30.3030108@arbash-meinel.com> MIME-Version: 1.0 X-OriginalArrivalTime: 26 Jul 2005 18:46:34.0256 (UTC) FILETIME=[588BF100:01C59212] X-WSS-ID: 6EF85A831M813479194-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.578 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/396 X-Sequence-Number: 13637 John, On 7/26/05 9:56 AM, "John A Meinel" wrote: > You could insert all of your data into a temporary table, and then do: > > INSERT INTO final_table SELECT * FROM temp_table WHERE NOT EXISTS > (SELECT info FROM final_table WHERE id=id, path=path, y=y); > > Or you could load it into the temporary table, and then: > DELETE FROM temp_table WHERE EXISTS (SELECT FROM final_table WHERE id...); > > And then do a plain INSERT INTO. > > I can't say what the specific performance increases would be, but > temp_table could certainly be an actual TEMP table (meaning it only > exists during the connection), and you could easily do a COPY into that > table to load it up quickly, without having to check any constraints. Yah - that's a typical approach, and it would be excellent if the COPY bypassed WAL for the temp table load. This is something we discussed in bizgres development a while back. I think we should do this for sure - would nearly double the temp table load rate, and the subsequent temp table delete *should* be fast enough (?) Any performance tests you've done on that delete/subselect operation? - Luke From pgsql-performance-owner@postgresql.org Tue Jul 26 16:02:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A630452C87 for ; Tue, 26 Jul 2005 16:02:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 33563-02 for ; Tue, 26 Jul 2005 19:02:15 +0000 (GMT) Received: from smtp-gw-cl-d.dmv.com (smtp-gw-cl-d.dmv.com [216.240.97.42]) by svr1.postgresql.org (Postfix) with ESMTP id BE67252C88 for ; Tue, 26 Jul 2005 16:02:12 -0300 (ADT) Received: from lanshark.dmv.com (lanshark.dmv.com [216.240.97.46]) by smtp-gw-cl-d.dmv.com (8.12.10/8.12.10) with ESMTP id j6QJ2ACf065855; Tue, 26 Jul 2005 15:02:10 -0400 (EDT) (envelope-from sven@dmv.com) Subject: Re: faster INSERT with possible pre-existing row? From: Sven Willenberger To: Dan Harris Cc: pgsql-performance@postgresql.org In-Reply-To: References: Content-Type: text/plain Date: Tue, 26 Jul 2005 15:03:17 -0400 Message-Id: <1122404597.1605.50.camel@lanshark.dmv.com> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Scanned-By: MIMEDefang 2.48 on 216.240.97.42 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.329 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_WHOIS X-Spam-Level: X-Archive-Number: 200507/397 X-Sequence-Number: 13638 On Tue, 2005-07-26 at 10:50 -0600, Dan Harris wrote: > I am working on a process that will be inserting tens of million rows > and need this to be as quick as possible. > > The catch is that for each row I could potentially insert, I need to > look and see if the relationship is already there to prevent > multiple entries. Currently I am doing a SELECT before doing the > INSERT, but I recognize the speed penalty in doing to operations. I > wonder if there is some way I can say "insert this record, only if it > doesn't exist already". To see if it exists, I would need to compare > 3 fields instead of just enforcing a primary key. > > Even if this could be a small increase per record, even a few percent > faster compounded over the whole load could be a significant reduction. > > Thanks for any ideas you might have. > Perhaps a trigger: CREATE FUNCTION verify_unique() RETURNS TRIGGER AS $func$ BEGIN PERFORM a,b,c FROM table1 WHERE a = NEW.a and b = NEW.b and c = NEW.c; IF FOUND THEN RETURN NULL; END IF; RETURN NEW; END; $func$ LANGUAGE plpgsql STABLE; CREATE TRIGGER verify_unique BEFORE INSERT ON table1 FOR EACH ROW EXECUTE PROCEDURE verify_unique(); Triggers are fired on COPY commands and if table1 is able to be cached and you have an index on table1(a,b,c) the results should be fairly decent. I would be interested in seeing the difference in timing between this approach and the temp table approach. From pgsql-performance-owner@postgresql.org Tue Jul 26 16:10:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C522252C62 for ; Tue, 26 Jul 2005 16:10:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34050-08 for ; Tue, 26 Jul 2005 19:10:11 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.199]) by svr1.postgresql.org (Postfix) with ESMTP id 0DA3252AAB for ; Tue, 26 Jul 2005 16:10:09 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i21so15167wra for ; Tue, 26 Jul 2005 12:10:11 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=lk0S0F4WVvCDkaZpyFc/JWM34Rji2UyoGA5eYne5Qw9lJHbrjAe2QhasvJhLVzAJjt8ZyCeBqmyBuINZh8jUh7TxLM1bxZrgIEpWqcPt6cQept5ndnjPEpWiRilrbIfxyGr/LkzMjENoYL3YcUQmOi+PUsyBpBDW2hP/pTZeO7k= Received: by 10.54.116.17 with SMTP id o17mr36668wrc; Tue, 26 Jul 2005 12:10:11 -0700 (PDT) Received: by 10.54.86.15 with HTTP; Tue, 26 Jul 2005 12:10:11 -0700 (PDT) Message-ID: <33c6269f0507261210400dff27@mail.gmail.com> Date: Tue, 26 Jul 2005 15:10:11 -0400 From: Alex Turner Reply-To: Alex Turner To: PFC Subject: Re: Cheap RAM disk? Cc: John A Meinel , Postgresql Performance In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <42E6661D.3060304@arbash-meinel.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.107 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/398 X-Sequence-Number: 13639 Please see: http://www.newegg.com/Product/Product.asp?Item=3DN82E16820145309 and http://www.newegg.com/Product/Product.asp?Item=3DN82E16820145416 The price of Reg ECC is not significantly higher than regular ram at this point. Plus if you go with super fast 2-2-2-6 then it's actualy more than good ol 2.5 Reg ECC. Alex Turner NetEconomist On 7/26/05, PFC wrote: >=20 > > I'm a little leary as it is definitely a version 1.0 product (it is > > still using an FPGA as the controller, so they were obviously pushing t= o > > get the card into production). >=20 > Not necessarily. FPGA's have become a sensible choice now. My RME= studio > soundcard uses a big FPGA. >=20 > The performance in the test doesn't look that good, though, but d= on't > forget it was run under windows. For instance they get 77s to copy the > Firefox source tree on their Athlon 64/raptor ; my Duron / 7200rpm ide > drive does it in 30 seconds, but not with windows of course. >=20 > However it doesnt' use ECC so... That's a pity, because they coul= d have > implemented ECC in "software" inside the chip, and have the benefits of > error correction with normal, cheap RAM. >=20 > Well; wait and see... >=20 > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend > From pgsql-performance-owner@postgresql.org Tue Jul 26 16:12:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 36B3252CAE for ; Tue, 26 Jul 2005 16:11:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 32814-10 for ; Tue, 26 Jul 2005 19:11:47 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.202]) by svr1.postgresql.org (Postfix) with ESMTP id 9DAFF52C8D for ; Tue, 26 Jul 2005 16:11:41 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id i21so15386wra for ; Tue, 26 Jul 2005 12:11:43 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=i0vJAm2Hw1eaDTX8YCd2ID81QhW+yt3qTgbg5MLVNlaxWHL1ncPToUoAW7co67gZEmasBWwfMqBrxWHXAD+RUn+tYvqxchh80Dxi3nmC2ACSD09tdM23uSoLueMeblAPTscE8p/X/OlJF+Ju5mAcxmMGNoKHmDMgIMbm/mctUzo= Received: by 10.54.106.1 with SMTP id e1mr30604wrc; Tue, 26 Jul 2005 12:11:43 -0700 (PDT) Received: by 10.54.86.15 with HTTP; Tue, 26 Jul 2005 12:11:42 -0700 (PDT) Message-ID: <33c6269f05072612113bd68434@mail.gmail.com> Date: Tue, 26 Jul 2005 15:11:42 -0400 From: Alex Turner Reply-To: Alex Turner To: John A Meinel Subject: Re: Cheap RAM disk? Cc: Postgresql Performance In-Reply-To: <42E6661D.3060304@arbash-meinel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <42E6661D.3060304@arbash-meinel.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.107 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/399 X-Sequence-Number: 13640 Also seems pretty silly to put it on a regular SATA connection, when all that can manage is 150MB/sec. If you made it connection directly to 66/64-bit PCI then it could actualy _use_ the speed of the RAM, not to mention PCI-X. Alex Turner NetEconomist On 7/26/05, John A Meinel wrote: > I saw a review of a relatively inexpensive RAM disk over at > anandtech.com, the Gigabyte i-RAM > http://www.anandtech.com/storage/showdoc.aspx?i=3D2480 >=20 > Basically, it is a PCI card, which takes standard DDR RAM, and has a > SATA port on it, so that to the system, it looks like a normal SATA drive= . >=20 > The card costs about $100-150, and you fill it with your own ram, so for > a 4GB (max size) disk, it costs around $500. Looking for solid state > storage devices, the cheapest I found was around $5k for 2GB. >=20 > Gigabyte claims that the battery backup can last up to 16h, which seems > decent, if not really long (the $5k solution has a built-in harddrive so > that if the power goes out, it uses the battery power to copy the > ramdisk onto the harddrive for more permanent storage). >=20 > Anyway, would something like this be reasonable as a drive for storing > pg_xlog? With 4GB you could have as many as 256 checkpoint segments. >=20 > I'm a little leary as it is definitely a version 1.0 product (it is > still using an FPGA as the controller, so they were obviously pushing to > get the card into production). >=20 > But it seems like this might be a decent way to improve insert > performance, without setting fsync=3Dfalse. >=20 > Probably it should see some serious testing (as in power spikes/pulled > plugs, etc). I know the article made some claim that if you actually > pull out the card it goes into "high consumption mode" which is somehow > greater than if you leave it in the slot with the power off. Which to me > seems like a lot of bull, and really means the 16h is only under > best-case circumstances. But even 1-2h is sufficient to handle a simple > power outage. >=20 > And if you had a UPS with detection of power failure, you could always > sync the ramdisk to a local partition before the power goes out. Though > you could do that with a normal in-memory ramdisk (tmpfs) without having > to buy the card. Though it does give you up-to an extra 4GB of ram, for > machines which have already maxed out their slots. >=20 > Anyway, I thought I would mention it to the list, to see if anyone else > has heard of it, or has any thoughts on the matter. I'm sure there are > some people who are using more expensive ram disks, maybe they have some > ideas about what this device is missing. (other than costing about > 1/10th the price) >=20 > John > =3D:-> >=20 >=20 >=20 > From pgsql-performance-owner@postgresql.org Tue Jul 26 16:15:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0E0CF52C62 for ; Tue, 26 Jul 2005 16:15:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 33869-09 for ; Tue, 26 Jul 2005 19:15:35 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id C8BC752C44 for ; Tue, 26 Jul 2005 16:15:31 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050726191532m9200g488fe>; Tue, 26 Jul 2005 19:15:32 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 98A8E5600C; Tue, 26 Jul 2005 14:15:32 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 7717555FB5; Tue, 26 Jul 2005 14:15:26 -0500 (CDT) Message-ID: <42E68BCF.8010600@arbash-meinel.com> Date: Tue, 26 Jul 2005 14:15:27 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Alex Turner Cc: Postgresql Performance Subject: Re: Cheap RAM disk? References: <42E6661D.3060304@arbash-meinel.com> <33c6269f05072612113bd68434@mail.gmail.com> In-Reply-To: <33c6269f05072612113bd68434@mail.gmail.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig546882E2A9D441DC56323519" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.044 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/400 X-Sequence-Number: 13641 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig546882E2A9D441DC56323519 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Alex Turner wrote: > Also seems pretty silly to put it on a regular SATA connection, when > all that can manage is 150MB/sec. If you made it connection directly > to 66/64-bit PCI then it could actualy _use_ the speed of the RAM, not > to mention PCI-X. > > Alex Turner > NetEconomist > Well, the whole point is to have it look like a normal SATA drive, even to the point that you can boot off of it, without having to load a single driver. Now, you could offer that you could recreate a SATA controller on the card, with a SATA bios, etc. And then you could get the increased speed, and still have bootable functionality. But it is a version 1.0 of a product, and I'm sure they tried to make it as cheap as possible (and within their own capabilities.) John =:-> --------------enig546882E2A9D441DC56323519 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5ovPJdeBCYSNAAMRAuwrAKCuJzSoBX8ZbynBBIh1s64qdMLhiACfUzgq /PRFbMYdJuEPWa2t+Uv7bpg= =hxWU -----END PGP SIGNATURE----- --------------enig546882E2A9D441DC56323519-- From pgsql-performance-owner@postgresql.org Tue Jul 26 16:30:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 1C66352CBB for ; Tue, 26 Jul 2005 16:30:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44598-08 for ; Tue, 26 Jul 2005 19:30:09 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.41]) by svr1.postgresql.org (Postfix) with ESMTP id 9456552C70 for ; Tue, 26 Jul 2005 16:30:04 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Tue, 26 Jul 2005 15:30:00 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Tue, 26 Jul 2005 15:30:00 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Tue, 26 Jul 2005 15:30:00 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Tue, 26 Jul 2005 12:30:00 -0700 Subject: Re: [Bizgres-general] Re: faster INSERT with possible From: "Luke Lonergan" To: "Hannu Krosing" Cc: "John A Meinel" , "Dan Harris" , "bizgres-general" , pgsql-performance@postgresql.org Message-ID: In-Reply-To: <1122404195.31930.4.camel@fuji.krosing.net> MIME-Version: 1.0 X-OriginalArrivalTime: 26 Jul 2005 19:30:00.0612 (UTC) FILETIME=[6A0E4E40:01C59218] X-WSS-ID: 6EF850B221S789033-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.577 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/402 X-Sequence-Number: 13643 Hannu, On 7/26/05 11:56 AM, "Hannu Krosing" wrote: > On T, 2005-07-26 at 11:46 -0700, Luke Lonergan wrote: > >> Yah - that's a typical approach, and it would be excellent if the COPY >> bypassed WAL for the temp table load. > > Don't *all* operations on TEMP tables bypass WAL ? Good question - do they? We had discussed the bypass as an elective option, or an automated one for special conditions (no index on table, empty table) or both. I thought that temp tables was one of those special conditions. Well - now that I test it, it appears you are correct, temp table COPY bypasses WAL - thanks for pointing it out! The following test is on a load of 200MB of table data from an ASCII file with 1 text column of size 145MB. - Luke ===================== TEST =========================== dgtestdb=# create temporary table temp1 (a text); CREATE TABLE dgtestdb=# \timing Timing is on. dgtestdb=# \i copy.ctl COPY Time: 4549.212 ms dgtestdb=# \i copy.ctl COPY Time: 3897.395 ms -- that's two tests, two loads of 200MB each, averaging 4.2 secs dgtestdb=# create table temp2 as select * from temp1; SELECT Time: 5914.803 ms -- a quick comparison to "CREATE TABLE AS SELECT", which bypasses WAL -- on bizgres dgtestdb=# drop table temp1; DROP TABLE Time: 135.782 ms dgtestdb=# drop table temp2; DROP TABLE Time: 3.707 ms dgtestdb=# create table temp1 (a text); CREATE TABLE Time: 1.667 ms dgtestdb=# \i copy.ctl COPY Time: 6034.274 ms dgtestdb=# -- This was a non-temporary table COPY, showing the slower performance of 6 secs. - Luke From pgsql-performance-owner@postgresql.org Tue Jul 26 16:35:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C446452C87 for ; Tue, 26 Jul 2005 16:35:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 61299-07 for ; Tue, 26 Jul 2005 19:35:19 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.192]) by svr1.postgresql.org (Postfix) with ESMTP id B6AAC52C73 for ; Tue, 26 Jul 2005 16:35:15 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 70so22791wra for ; Tue, 26 Jul 2005 12:35:17 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=ssbNqQ1ji0BelG+bDIAq+CRiOY0LmhOT5dkoZvSLcSOgzouX4t4HebnZsjXgcnnrPuHFrYGpvgf7xkrRL0o/qqcopQMjt1RN2xDb2ypu2PsP5nkJULFyKw0bEVKdcsjNNHeCAjqFp4N5AKEu5W3uY1663DnpCMUc5T8J97RX9BQ= Received: by 10.54.24.49 with SMTP id 49mr49888wrx; Tue, 26 Jul 2005 12:35:17 -0700 (PDT) Received: by 10.54.22.15 with HTTP; Tue, 26 Jul 2005 12:35:16 -0700 (PDT) Message-ID: Date: Tue, 26 Jul 2005 14:35:17 -0500 From: Matthew Nuzum Reply-To: newz@bearfruit.org To: Dan Harris Subject: Re: faster INSERT with possible pre-existing row? Cc: pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.144 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/404 X-Sequence-Number: 13645 On 7/26/05, Dan Harris wrote: > I am working on a process that will be inserting tens of million rows > and need this to be as quick as possible. >=20 > The catch is that for each row I could potentially insert, I need to > look and see if the relationship is already there to prevent > multiple entries. Currently I am doing a SELECT before doing the > INSERT, but I recognize the speed penalty in doing to operations. I > wonder if there is some way I can say "insert this record, only if it > doesn't exist already". To see if it exists, I would need to compare > 3 fields instead of just enforcing a primary key. I struggled with this for a while. At first I tried stored procedures and triggers, but it took very long (over 24 hours for my dataset). After several iterations of rewritting it, first into C# then into Python I got the whole process down to under 30 min. My scenario is this: I want to normalize log data. For example, for the IP address in a log entry, I need to look up the unique id of the IP address, or if the IP address is new, insert it and then return the newly created entry. Multiple processes use the data, but only one process, run daily, actually changes it. Because this one process knows that the data is static, it selects the tables into in-memory hash tables (C#) or Dictionaries (Python) and then does the lookups there. It is *super* fast, but it uses a *lot* of ram. ;-) To limit the ram, I wrote a version of the python code that uses gdbm files instead of Dictionaries. This requires a newer version of Python (to allow a gdbm db to work just like a dictionary) but makes life easier in case someone is using my software on a lower end machine. This doubled the time of the lookups from about 15 minutes to 30, bringing the whole process to about 45 minutes. --=20 Matthew Nuzum www.bearfruit.org From pgsql-performance-owner@postgresql.org Tue Jul 26 16:35:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B94E152C78 for ; Tue, 26 Jul 2005 16:35:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51186-08 for ; Tue, 26 Jul 2005 19:35:21 +0000 (GMT) Received: from smtp5.pop.com.br (smtp5.pop.com.br [200.175.8.39]) by svr1.postgresql.org (Postfix) with ESMTP id 6026452C70 for ; Tue, 26 Jul 2005 16:35:18 -0300 (ADT) Received: from smtp5.pop.com.br (localhost [127.0.0.1]) by smtp5.pop.com.br (Postfix) with ESMTP id 8F1BA93928 for ; Tue, 26 Jul 2005 16:35:19 -0300 (BRT) Received: from popmail5.pop.com.br (popmail5.pop.com.br [200.175.8.30]) by smtp5.pop.com.br (Postfix) with SMTP id 808FE9379A for ; Tue, 26 Jul 2005 16:35:19 -0300 (BRT) Received: from 200.101.246.47 (proxying for 192.168.69.6) (POPmail authenticated user beto.neto) by popmail5.pop.com.br with HTTP; Tue, 26 Jul 2005 16:35:19 -0300 (EST) Message-ID: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> Date: Tue, 26 Jul 2005 16:35:19 -0300 (EST) Subject: [IMPORTANT] - My application performance From: "Roberto Germano Vieweg Neto" To: pgsql-performance@postgresql.org User-Agent: POPMail/1.4.2 X-POPMail-client-ip: 200.101.246.47 MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-DCC-POPInternet-Metrics: dcc 1259; Body=1 Fuz1=1 Fuz2=1 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/403 X-Sequence-Number: 13644 My application is using Firebird 1.5.2 I have at my database: - 150 Doamins - 318 tables - 141 Views - 365 Procedures - 407 Triggers - 75 generators - 161 Exceptions - 183 UDFs - 1077 Indexes My question is: Postgre SQL will be more faster than Firebird? How much (in percent)? I need about 20% to 50% more performance at my application. Can I get this migratin to postgresql ? From pgsql-performance-owner@postgresql.org Tue Jul 26 16:51:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 56F8952CAB for ; Tue, 26 Jul 2005 16:51:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 67897-06 for ; Tue, 26 Jul 2005 19:51:21 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 01DD852C78 for ; Tue, 26 Jul 2005 16:51:18 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050726195115m9100613f7e>; Tue, 26 Jul 2005 19:51:19 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 2EB895600C; Tue, 26 Jul 2005 14:51:15 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 23A8655FB5; Tue, 26 Jul 2005 14:51:10 -0500 (CDT) Message-ID: <42E6942F.3070006@arbash-meinel.com> Date: Tue, 26 Jul 2005 14:51:11 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: newz@bearfruit.org Cc: Dan Harris , pgsql-performance@postgresql.org Subject: Re: faster INSERT with possible pre-existing row? References: In-Reply-To: X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigE97DC7894F092B8718A89C85" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.044 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/405 X-Sequence-Number: 13646 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigE97DC7894F092B8718A89C85 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Matthew Nuzum wrote: > On 7/26/05, Dan Harris wrote: > >>I am working on a process that will be inserting tens of million rows >>and need this to be as quick as possible. >> >>The catch is that for each row I could potentially insert, I need to >>look and see if the relationship is already there to prevent >>multiple entries. Currently I am doing a SELECT before doing the >>INSERT, but I recognize the speed penalty in doing to operations. I >>wonder if there is some way I can say "insert this record, only if it >>doesn't exist already". To see if it exists, I would need to compare >>3 fields instead of just enforcing a primary key. > > > I struggled with this for a while. At first I tried stored procedures > and triggers, but it took very long (over 24 hours for my dataset). > After several iterations of rewritting it, first into C# then into > Python I got the whole process down to under 30 min. > > My scenario is this: > I want to normalize log data. For example, for the IP address in a log > entry, I need to look up the unique id of the IP address, or if the IP > address is new, insert it and then return the newly created entry. > Multiple processes use the data, but only one process, run daily, > actually changes it. Because this one process knows that the data is > static, it selects the tables into in-memory hash tables (C#) or > Dictionaries (Python) and then does the lookups there. It is *super* > fast, but it uses a *lot* of ram. ;-) > > To limit the ram, I wrote a version of the python code that uses gdbm > files instead of Dictionaries. This requires a newer version of Python > (to allow a gdbm db to work just like a dictionary) but makes life > easier in case someone is using my software on a lower end machine. > This doubled the time of the lookups from about 15 minutes to 30, > bringing the whole process to about 45 minutes. > Did you ever try the temp table approach? You could: COPY all records into temp_table, with an empty row for ip_id -- Get any entries which already exist UPDATE temp_table SET ip_id = (SELECT ip_id from ipaddress WHERE add=add) WHERE EXISTS (SELECT ip_id FROM ipaddress WHERE add=add); -- Create new entries INSERT INTO ipaddress(add) SELECT add FROM temp_table WHERE ip_id IS NULL; -- Update the rest UPDATE temp_table SET ip_id = (SELECT ip_id from ipaddress WHERE add=add) WHERE ip_id IS NULL AND EXISTS (SELECT ip_id FROM ipaddress WHERE add=add); This would let the database do all of the updating work in bulk on it's side, rather than you pulling all the data out and doing it locally. An alternative would be something like: CREATE TEMP TABLE new_ids (address text, ip_id int); COPY all potentially new addresses into that table. -- Delete all entries which already exist DELETE FROM new_ids WHERE EXISTS (SELECT ip_id FROM ipaddresses WHERE add=new_ids.address); -- Now create the new entries INSERT INTO ipaddresses(add) SELECT address FROM new_ids; -- At this point you are guaranteed to have all addresses existing in -- the database If you then insert your full data into the final table, only leave the ip_id column as null. Then if you have a partial index where ip_id is NULL, you could use the command: UPDATE final_table SET ip_id = (SELECT ip_id FROM ipaddresses WHERE add=final_table.add) WHERE ip_id IS NULL; You could also do this in a temporary table, before bulk inserting into the final table. I don't know what you have tried, but I know that for Dan, he easily has > 36M rows. So I don't think he wants to pull that locally and create a in-memory hash just to insert 100 rows or so. Also, for your situation, if you do keep a local cache, you could certainly save the cache between runs, and use a temp table to determine what new ids you need to add to it. Then you wouldn't have to pull the complete set each time. You just pull new values for entries you haven't added yet. John =:-> --------------enigE97DC7894F092B8718A89C85 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Cygwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC5pQvJdeBCYSNAAMRAgo1AJ4ymsiIWbtLh85+pw1lBF2U23EdOACfUxlX B+NHHBJyDb0k/VGl2/fVkpo= =pXYP -----END PGP SIGNATURE----- --------------enigE97DC7894F092B8718A89C85-- From pgsql-performance-owner@postgresql.org Tue Jul 26 17:02:52 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id AB0E552940 for ; Tue, 26 Jul 2005 17:02:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 70802-03 for ; Tue, 26 Jul 2005 20:02:41 +0000 (GMT) Received: from dfw-gate2.raytheon.com (dfw-gate2.raytheon.com [199.46.199.231]) by svr1.postgresql.org (Postfix) with ESMTP id 8D82C52C73 for ; Tue, 26 Jul 2005 17:02:40 -0300 (ADT) Received: from ds02c00.directory.ray.com (ds02c00.directory.ray.com [147.25.138.118]) by dfw-gate2.raytheon.com (8.12.10/8.12.10) with ESMTP id j6QK2cPC006492 for ; Tue, 26 Jul 2005 15:02:38 -0500 (CDT) Received: from ds02c00 (localhost [127.0.0.1]) by ds02c00.directory.ray.com (Switch-3.1.7/Switch-3.1.0) with ESMTP id j6QK2Y6S013381 for ; Tue, 26 Jul 2005 20:02:34 GMT Received: from ds02c00.directory.ray.com with LMTP by ds02c00 (2.0.6/sieved-2-0-build-559) for ; Tue, 26 Jul 2005 20:02:33 +0000 Received: from notesserver5.ftw.us.ray.com (notesserver5.ftw.us.ray.com [151.168.145.35]) by ds02c00.directory.ray.com (Switch-3.1.7/Switch-3.1.0) with ESMTP id j6QK23Nh013087 sender Richard_D_Levine@raytheon.com for ; Tue, 26 Jul 2005 20:02:03 GMT In-Reply-To: <20050726183343.GZ19080@mathom.us> Subject: Re: Cheap RAM disk? To: pgsql-performance@postgresql.org X-Mailer: Lotus Notes Release 6.5.2 June 01, 2004 Message-ID: From: Richard_D_Levine@raytheon.com Date: Tue, 26 Jul 2005 15:02:01 -0500 X-MIMETrack: Serialize by Router on NotesServer5/HDC(Release 6.5.2|June 01, 2004) at 07/26/2005 03:02:03 PM MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII X-SPAM: 0.00 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.207 tagged_above=0 required=5 tests=AWL, NO_REAL_NAME X-Spam-Level: X-Archive-Number: 200507/406 X-Sequence-Number: 13647 > you'd be much better served by > putting a big NVRAM cache in front of a fast disk array I agree with the point below, but I think price was the issue of the original discussion. That said, it seems that a single high speed spindle would give this a run for its money in both price and performance, and for the same reasons Mike points out. Maybe a SCSI 160 or 320 at 15k, or maybe even something slower. Rick pgsql-performance-owner@postgresql.org wrote on 07/26/2005 01:33:43 PM: > On Tue, Jul 26, 2005 at 11:23:23AM -0700, Luke Lonergan wrote: > >Yup - interesting and very niche product - it seems like it's only obvious > >application is for the Postgresql WAL problem :-) > > On the contrary--it's not obvious that it is an ideal fit for a WAL. A > ram disk like this is optimized for highly random access applications. > The WAL is a single sequential writer. If you're in the kind of market > that needs a really high performance WAL you'd be much better served by > putting a big NVRAM cache in front of a fast disk array than by buying a > toy like this. > > Mike Stone > > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Tue Jul 26 17:05:18 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D124552C73 for ; Tue, 26 Jul 2005 17:05:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 71256-03 for ; Tue, 26 Jul 2005 20:04:59 +0000 (GMT) Received: from flake.decibel.org (flake.decibel.org [67.100.216.10]) by svr1.postgresql.org (Postfix) with ESMTP id A8307529E1 for ; Tue, 26 Jul 2005 17:04:59 -0300 (ADT) Received: by flake.decibel.org (Postfix, from userid 1001) id A8C7915232; Tue, 26 Jul 2005 15:05:00 -0500 (CDT) Date: Tue, 26 Jul 2005 15:05:00 -0500 From: "Jim C. Nasby" To: Roberto Germano Vieweg Neto Cc: pgsql-performance@postgresql.org Subject: Re: [IMPORTANT] - My application performance Message-ID: <20050726200500.GY29346@decibel.org> References: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> X-Operating-System: FreeBSD 4.11-RELEASE-p10 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/407 X-Sequence-Number: 13648 The number of objects in your system has virtually nothing to do with performance (at least on any decent database...) What is your application doing? What's the bottleneck right now? On Tue, Jul 26, 2005 at 04:35:19PM -0300, Roberto Germano Vieweg Neto wrote: > My application is using Firebird 1.5.2 > > I have at my database: > - 150 Doamins > - 318 tables > - 141 Views > - 365 Procedures > - 407 Triggers > - 75 generators > - 161 Exceptions > - 183 UDFs > - 1077 Indexes > > My question is: > > Postgre SQL will be more faster than Firebird? How much (in percent)? > > I need about 20% to 50% more performance at my application. > Can I get this migratin to postgresql ? > > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match > -- Jim C. Nasby, Database Consultant decibel@decibel.org Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Tue Jul 26 18:03:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 57FA652BD5 for ; Tue, 26 Jul 2005 18:03:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 84860-01 for ; Tue, 26 Jul 2005 21:03:18 +0000 (GMT) Received: from dfbsa63.conab.gov.br (dfbsa63.conab.gov.br [200.140.131.205]) by svr1.postgresql.org (Postfix) with ESMTP id 0B22252CB3 for ; Tue, 26 Jul 2005 18:03:16 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by dfbsa63.conab.gov.br (Postfix) with ESMTP id 733C24C0DD for ; Tue, 26 Jul 2005 18:00:53 -0300 (BRT) Received: from dfbsa63.conab.gov.br ([127.0.0.1]) by localhost (dfbsa63 [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 04878-40 for ; Tue, 26 Jul 2005 18:00:52 -0300 (BRT) Received: from gesin557 (unknown [10.1.2.201]) by dfbsa63.conab.gov.br (Postfix) with ESMTP id AB1294C0C7 for ; Tue, 26 Jul 2005 18:00:52 -0300 (BRT) From: "Alvaro Neto" To: Subject: RES: [IMPORTANT] - My application performance Date: Tue, 26 Jul 2005 18:00:52 -0300 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-Mailer: Microsoft Office Outlook, Build 11.0.5510 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 In-Reply-To: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> Thread-Index: AcWSGRQpcSi61uyYSl+z6FlkKRCY0QAC/vrg Message-Id: <20050726210052.AB1294C0C7@dfbsa63.conab.gov.br> X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at conab.gov.br X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.18 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/408 X-Sequence-Number: 13649 What's the your platform? Windows or Linux? What's the data volume (up million records)? -----Mensagem original----- De: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] Em nome de Roberto = Germano Vieweg Neto Enviada em: ter=E7a-feira, 26 de julho de 2005 16:35 Para: pgsql-performance@postgresql.org Assunto: [PERFORM] [IMPORTANT] - My application performance My application is using Firebird 1.5.2 I have at my database: - 150 Doamins - 318 tables - 141 Views - 365 Procedures - 407 Triggers - 75 generators - 161 Exceptions - 183 UDFs - 1077 Indexes My question is: Postgre SQL will be more faster than Firebird? How much (in percent)? I need about 20% to 50% more performance at my application. Can I get this migratin to postgresql ? ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match From pgsql-performance-owner@postgresql.org Tue Jul 26 20:04:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CFFDB52C9E for ; Tue, 26 Jul 2005 20:04:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54863-09 for ; Tue, 26 Jul 2005 23:04:02 +0000 (GMT) Received: from floppy.pyrenet.fr (news.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id 40D3C52C7D for ; Tue, 26 Jul 2005 20:04:01 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id D522030B42; Wed, 27 Jul 2005 01:11:21 +0200 (MET DST) From: Chris Browne X-Newsgroups: pgsql.performance Subject: Re: Cheap RAM disk? Date: Tue, 26 Jul 2005 18:12:10 -0400 Organization: cbbrowne Computing Inc Lines: 22 Message-ID: <60hdeh1akl.fsf@dba2.int.libertyrms.com> References: <42E6661D.3060304@arbash-meinel.com> <1122399740.12728.1.camel@toonses.gghcwest.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Complaints-To: usenet@news.hub.org User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4.17 (Jumbo Shrimp, linux) Cancel-Lock: sha1:upRZs9A+mI8OnclikGjycaSsg0E= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.122 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/410 X-Sequence-Number: 13651 jwbaker@acm.org ("Jeffrey W. Baker") writes: > I haven't tried this product, but the microbenchmarks seem truly > slow. I think you would get a similar benefit by simply sticking a > 1GB or 2GB DIMM -- battery-backed, of course -- in your RAID > controller. Well, the microbenchmarks were pretty pre-sophomoric, essentially trying to express how the device would be useful to a Windows user that *might* play games... I'm sure it's hurt by the fact that it's using a SATA ("version 1") interface rather than something faster. Mind you, I'd like to see the product succeed, because they might come up with a "version 2" of it that is what I'd really like... -- (format nil "~S@~S" "cbbrowne" "acm.org") http://www.ntlug.org/~cbbrowne/sap.html Rules of the Evil Overlord #78. "I will not tell my Legions of Terror "And he must be taken alive!" The command will be: ``And try to take him alive if it is reasonably practical.''" From pgsql-performance-owner@postgresql.org Tue Jul 26 19:29:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4255C52F83 for ; Tue, 26 Jul 2005 19:29:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 32997-10 for ; Tue, 26 Jul 2005 22:29:23 +0000 (GMT) Received: from mir3-fs.mir3.com (mail.mir3.com [216.74.11.46]) by svr1.postgresql.org (Postfix) with ESMTP id 6810352CB0 for ; Tue, 26 Jul 2005 19:29:22 -0300 (ADT) Received: from archimedes ([172.16.2.68]) by mir3-fs.mir3.com with Microsoft SMTPSVC(5.0.2195.6713); Tue, 26 Jul 2005 15:32:17 -0700 Subject: Re: faster INSERT with possible pre-existing row? From: Mark Lewis To: John A Meinel Cc: newz@bearfruit.org, Dan Harris , pgsql-performance@postgresql.org In-Reply-To: <42E6942F.3070006@arbash-meinel.com> References: <42E6942F.3070006@arbash-meinel.com> Content-Type: text/plain Organization: MIR3, Inc. Date: Tue, 26 Jul 2005 15:29:23 -0700 Message-Id: <1122416963.17503.25.camel@archimedes> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-16) Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 26 Jul 2005 22:32:17.0233 (UTC) FILETIME=[E0CA3C10:01C59231] X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/409 X-Sequence-Number: 13650 Easier and faster than doing the custom trigger is to simply define a unique index and let the DB enforce the constraint with an index lookup, something like: create unique index happy_index ON happy_table(col1, col2, col3); That should run faster than the custom trigger, but not as fast as the temp table solution suggested elsewhere because it will need to do an index lookup for each row. With this solution, it is important that your shared_buffers are set high enough that the happy_index can be kept in memory, otherwise performance will drop precipitously. Also, if you are increasing the size of the table by a large percentage, you will want to ANALYZE periodically, as an optimal plan for a small table may be a disaster for a large table, and PostgreSQL won't switch plans unless you run ANALYZE. -- Mark On Tue, 2005-07-26 at 14:51 -0500, John A Meinel wrote: > Matthew Nuzum wrote: > > On 7/26/05, Dan Harris wrote: > > > >>I am working on a process that will be inserting tens of million rows > >>and need this to be as quick as possible. > >> > >>The catch is that for each row I could potentially insert, I need to > >>look and see if the relationship is already there to prevent > >>multiple entries. Currently I am doing a SELECT before doing the > >>INSERT, but I recognize the speed penalty in doing to operations. I > >>wonder if there is some way I can say "insert this record, only if it > >>doesn't exist already". To see if it exists, I would need to compare > >>3 fields instead of just enforcing a primary key. > > > > > > I struggled with this for a while. At first I tried stored procedures > > and triggers, but it took very long (over 24 hours for my dataset). > > After several iterations of rewritting it, first into C# then into > > Python I got the whole process down to under 30 min. > > > > My scenario is this: > > I want to normalize log data. For example, for the IP address in a log > > entry, I need to look up the unique id of the IP address, or if the IP > > address is new, insert it and then return the newly created entry. > > Multiple processes use the data, but only one process, run daily, > > actually changes it. Because this one process knows that the data is > > static, it selects the tables into in-memory hash tables (C#) or > > Dictionaries (Python) and then does the lookups there. It is *super* > > fast, but it uses a *lot* of ram. ;-) > > > > To limit the ram, I wrote a version of the python code that uses gdbm > > files instead of Dictionaries. This requires a newer version of Python > > (to allow a gdbm db to work just like a dictionary) but makes life > > easier in case someone is using my software on a lower end machine. > > This doubled the time of the lookups from about 15 minutes to 30, > > bringing the whole process to about 45 minutes. > > > > Did you ever try the temp table approach? You could: > > COPY all records into temp_table, with an empty row for ip_id > -- Get any entries which already exist > UPDATE temp_table SET ip_id = > (SELECT ip_id from ipaddress WHERE add=add) > WHERE EXISTS (SELECT ip_id FROM ipaddress WHERE add=add); > -- Create new entries > INSERT INTO ipaddress(add) SELECT add FROM temp_table > WHERE ip_id IS NULL; > -- Update the rest > UPDATE temp_table SET ip_id = > (SELECT ip_id from ipaddress WHERE add=add) > WHERE ip_id IS NULL AND > EXISTS (SELECT ip_id FROM ipaddress WHERE add=add); > > This would let the database do all of the updating work in bulk on it's > side, rather than you pulling all the data out and doing it locally. > > An alternative would be something like: > > CREATE TEMP TABLE new_ids (address text, ip_id int); > COPY all potentially new addresses into that table. > -- Delete all entries which already exist > DELETE FROM new_ids WHERE EXISTS > (SELECT ip_id FROM ipaddresses > WHERE add=new_ids.address); > -- Now create the new entries > INSERT INTO ipaddresses(add) SELECT address FROM new_ids; > > -- At this point you are guaranteed to have all addresses existing in > -- the database > > If you then insert your full data into the final table, only leave the > ip_id column as null. Then if you have a partial index where ip_id is > NULL, you could use the command: > > UPDATE final_table SET ip_id = > (SELECT ip_id FROM ipaddresses WHERE add=final_table.add) > WHERE ip_id IS NULL; > > You could also do this in a temporary table, before bulk inserting into > the final table. > > I don't know what you have tried, but I know that for Dan, he easily has > > 36M rows. So I don't think he wants to pull that locally and create a > in-memory hash just to insert 100 rows or so. > > Also, for your situation, if you do keep a local cache, you could > certainly save the cache between runs, and use a temp table to determine > what new ids you need to add to it. Then you wouldn't have to pull the > complete set each time. You just pull new values for entries you haven't > added yet. > > John > =:-> From pgsql-performance-owner@postgresql.org Tue Jul 26 22:52:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7203752C0A for ; Tue, 26 Jul 2005 22:52:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 88723-08 for ; Wed, 27 Jul 2005 01:52:40 +0000 (GMT) Received: from houston.familyhealth.com.au (houston.au.fhnetwork.com [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id E516452BD2 for ; Tue, 26 Jul 2005 22:52:38 -0300 (ADT) Received: from houston.familyhealth.com.au (localhost [127.0.0.1]) by houston.familyhealth.com.au (Postfix) with ESMTP id 1F52124FF7; Wed, 27 Jul 2005 09:52:41 +0800 (WST) Received: from [127.0.0.1] (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (Postfix) with ESMTP id 2D51824FF1; Wed, 27 Jul 2005 09:52:39 +0800 (WST) Message-ID: <42E6E9A3.4000400@familyhealth.com.au> Date: Wed, 27 Jul 2005 09:55:47 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.6 (Windows/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Dan Harris Cc: pgsql-performance@postgresql.org Subject: Re: faster INSERT with possible pre-existing row? References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-familyhealth-MailScanner-Information: Please contact the ISP for more information X-familyhealth-MailScanner: Found to be clean X-familyhealth-MailScanner-From: chriskl@familyhealth.com.au X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.062 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/411 X-Sequence-Number: 13652 Insert into a temp table then use INSERT INTO...SELECT FROM to insert all rows into the proper table that don't have a relationship. Chris Dan Harris wrote: > I am working on a process that will be inserting tens of million rows > and need this to be as quick as possible. > > The catch is that for each row I could potentially insert, I need to > look and see if the relationship is already there to prevent multiple > entries. Currently I am doing a SELECT before doing the INSERT, but I > recognize the speed penalty in doing to operations. I wonder if there > is some way I can say "insert this record, only if it doesn't exist > already". To see if it exists, I would need to compare 3 fields > instead of just enforcing a primary key. > > Even if this could be a small increase per record, even a few percent > faster compounded over the whole load could be a significant reduction. > > Thanks for any ideas you might have. > > -Dan > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend From pgsql-performance-owner@postgresql.org Tue Jul 26 22:59:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E77DE52B29 for ; Tue, 26 Jul 2005 22:59:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87520-10 for ; Wed, 27 Jul 2005 01:59:06 +0000 (GMT) Received: from houston.familyhealth.com.au (houston.au.fhnetwork.com [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 9D5CC52C60 for ; Tue, 26 Jul 2005 22:59:05 -0300 (ADT) Received: from houston.familyhealth.com.au (localhost [127.0.0.1]) by houston.familyhealth.com.au (Postfix) with ESMTP id 7AD0F24FF2; Wed, 27 Jul 2005 09:59:09 +0800 (WST) Received: from [127.0.0.1] (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (Postfix) with ESMTP id 85EAA24FF1; Wed, 27 Jul 2005 09:59:08 +0800 (WST) Message-ID: <42E6EB2D.9040008@familyhealth.com.au> Date: Wed, 27 Jul 2005 10:02:21 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 1.0.6 (Windows/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Roberto Germano Vieweg Neto Cc: pgsql-performance@postgresql.org Subject: Re: [IMPORTANT] - My application performance References: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> In-Reply-To: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-familyhealth-MailScanner-Information: Please contact the ISP for more information X-familyhealth-MailScanner: Found to be clean X-familyhealth-MailScanner-From: chriskl@familyhealth.com.au X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.062 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/412 X-Sequence-Number: 13653 Roberto Germano Vieweg Neto wrote: > My application is using Firebird 1.5.2 > > I have at my database: > - 150 Doamins > - 318 tables > - 141 Views > - 365 Procedures > - 407 Triggers > - 75 generators > - 161 Exceptions > - 183 UDFs > - 1077 Indexes > > My question is: > > Postgre SQL will be more faster than Firebird? How much (in percent)? I think you can probably expect around 10341.426% improvement. ps. Yes, I am joking just in case... From pgsql-performance-owner@postgresql.org Wed Jul 27 04:28:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EA13F52EBB for ; Wed, 27 Jul 2005 04:28:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 64629-07 for ; Wed, 27 Jul 2005 07:28:00 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.201]) by svr1.postgresql.org (Postfix) with ESMTP id DE94152EDD for ; Wed, 27 Jul 2005 04:27:58 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 8so104624nzo for ; Wed, 27 Jul 2005 00:28:00 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=QSCGRSyaqvi0nKmeuQuxyHNuoozhJmEJSxAmlX9/mOJ0j1G3SBRvaDWTQLxqSvFvofKngmLyJqqGg4Gd3WVdDTG6l2zM6puDwLD7CWc1tUVJbWtKTNHBYayHsHoRke7U7iZ1QbWnjyxxqG1FOfm/YONpvVI5XFODcIJ3Y/sVxzo= Received: by 10.37.12.49 with SMTP id p49mr608946nzi; Wed, 27 Jul 2005 00:28:00 -0700 (PDT) Received: by 10.36.22.20 with HTTP; Wed, 27 Jul 2005 00:28:00 -0700 (PDT) Message-ID: <758d5e7f050727002844a1ad3c@mail.gmail.com> Date: Wed, 27 Jul 2005 09:28:00 +0200 From: Dawid Kuroczko Reply-To: Dawid Kuroczko To: Roberto Germano Vieweg Neto Subject: Re: [IMPORTANT] - My application performance Cc: pgsql-performance@postgresql.org In-Reply-To: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <38776.200.101.246.47.1122406519.squirrel@popmail5.pop.com.br> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.617 tagged_above=0 required=5 tests=AWL, FROM_ENDS_IN_NUMS, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/413 X-Sequence-Number: 13654 On 7/26/05, Roberto Germano Vieweg Neto wrote: > My application is using Firebird 1.5.2 >=20 > My question is: >=20 > Postgre SQL will be more faster than Firebird? How much (in percent)? >=20 > I need about 20% to 50% more performance at my application. > Can I get this migratin to postgresql ? The answer is: maybe. There's nothing which stops PostgreSQL from being faster, and likewise there is nothing that stops it from being slower. YMMV. Your route should be: * migrate most speed-demanding part to PostgreSQL * benchmark it, trying to emulate real-world load. * if it is slower than Firebird, post it here, together with EXPLAIN ANALY= ZEs and ask if there's something you can do to speed it up. Regards, Dawid From pgsql-performance-owner@postgresql.org Wed Jul 27 04:56:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B096352EE0 for ; Wed, 27 Jul 2005 04:56:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 72467-08 for ; Wed, 27 Jul 2005 07:56:14 +0000 (GMT) Received: from smtp1.asco.de (smtp1.asco.de [217.13.70.154]) by svr1.postgresql.org (Postfix) with ESMTP id 5102C52EBE for ; Wed, 27 Jul 2005 04:56:12 -0300 (ADT) Received: from [192.168.1.72] (pitr.asco.de [192.168.1.72]) (envelope-sender: ) (authenticated j_schicke CRAM-MD5 bits=0) by smtp1.asco.de (8.13.4/8.13.4/Debian-3) with ESMTP id j6R7uC3U008048 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Wed, 27 Jul 2005 09:56:12 +0200 Date: Wed, 27 Jul 2005 09:56:12 +0200 From: Jens-Wolfhard Schicke Reply-To: Jens-Wolfhard Schicke To: pgsql-performance@postgresql.org Subject: Inherited Table Query Planning (fwd) Message-ID: <66F42056CEA99778F918C7C9@[192.168.1.72]> X-Mailer: Mulberry/3.1.6 (Linux/x86) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; format=flowed Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/414 X-Sequence-Number: 13655 Is there a way to make the query planner consider pulling inner appends=20 outside joins? Example: natural_person inherits from person (obviously) admpostgres3=3D# explain analyze select u.name, p.name from users u, person = p=20 where p.user_id =3D u.id and u.name =3D 's_ohl'; QUERY PLAN ----------------------------------------------------------------------------= ---------------------------------------------------------- Hash Join (cost=3D8.01..3350.14 rows=3D3 width=3D36) (actual=20 time=3D107.391..343.657 rows=3D10 loops=3D1) Hash Cond: ("outer".user_id =3D "inner".id) -> Append (cost=3D0.00..2461.34 rows=3D117434 width=3D20) (actual=20 time=3D0.007..264.910 rows=3D117434 loops=3D1) -> Seq Scan on person p (cost=3D0.00..575.06 rows=3D31606 = width=3D20)=20 (actual time=3D0.005..38.911 rows=3D31606 loops=3D1) -> Seq Scan on natural_person p (cost=3D0.00..1886.28 = rows=3D85828=20 width=3D19) (actual time=3D0.003..104.338 rows=3D85828 loops=3D1) -> Hash (cost=3D8.01..8.01 rows=3D2 width=3D24) (actual = time=3D0.096..0.096=20 rows=3D0 loops=3D1) -> Index Scan using users_name_idx on users u (cost=3D0.00..8.01 = rows=3D2 width=3D24) (actual time=3D0.041..0.081 rows=3D10 loops=3D1) Index Cond: ((name)::text =3D 's_ohl'::text) Total runtime: 343.786 ms (9 rows) admpostgres3=3D# explain analyze select u.name, p.name from users u, only=20 person p where p.user_id =3D u.id and u.name =3D 's_ohl' union all select=20 u.name, p.name from users u, only natural_person p where p.user_id =3D u.id = and u.name =3D 's_ohl'; QUERY PLAN ----------------------------------------------------------------------------= ----------------------------------------------------------------------------= ----- Append (cost=3D0.00..28.19 rows=3D3 width=3D28) (actual = time=3D0.197..0.366=20 rows=3D10 loops=3D1) -> Subquery Scan "*SELECT* 1" (cost=3D0.00..14.12 rows=3D1 width=3D28) = (actual time=3D0.159..0.159 rows=3D0 loops=3D1) -> Nested Loop (cost=3D0.00..14.11 rows=3D1 width=3D28) (actual=20 time=3D0.157..0.157 rows=3D0 loops=3D1) -> Index Scan using users_name_idx on users u=20 (cost=3D0.00..8.01 rows=3D2 width=3D24) (actual time=3D0.039..0.075 = rows=3D10 loops=3D1) Index Cond: ((name)::text =3D 's_ohl'::text) -> Index Scan using person_user_idx on person p=20 (cost=3D0.00..3.03 rows=3D2 width=3D8) (actual time=3D0.006..0.006 rows=3D0 = loops=3D10) Index Cond: (p.user_id =3D "outer".id) -> Subquery Scan "*SELECT* 2" (cost=3D0.00..14.08 rows=3D2 width=3D28) = (actual time=3D0.036..0.193 rows=3D10 loops=3D1) -> Nested Loop (cost=3D0.00..14.06 rows=3D2 width=3D28) (actual=20 time=3D0.033..0.171 rows=3D10 loops=3D1) -> Index Scan using users_name_idx on users u=20 (cost=3D0.00..8.01 rows=3D2 width=3D24) (actual time=3D0.018..0.049 = rows=3D10 loops=3D1) Index Cond: ((name)::text =3D 's_ohl'::text) -> Index Scan using natural_person_user_idx on=20 natural_person p (cost=3D0.00..3.01 rows=3D1 width=3D8) (actual=20 time=3D0.006..0.007 rows=3D1 loops=3D10) Index Cond: (p.user_id =3D "outer".id) Total runtime: 0.475 ms (14 rows) Mit freundlichem Gru=DF Jens Schicke --=20 Jens Schicke j.schicke@asco.de asco GmbH http://www.asco.de Mittelweg 7 Tel 0531/3906-127 38106 Braunschweig Fax 0531/3906-400 From pgsql-performance-owner@postgresql.org Wed Jul 27 11:22:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6D23452C32 for ; Wed, 27 Jul 2005 11:09:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 61958-06 for ; Wed, 27 Jul 2005 14:09:07 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6C2F152C06 for ; Wed, 27 Jul 2005 11:09:07 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6RE98PA025524; Wed, 27 Jul 2005 10:09:08 -0400 (EDT) To: Jens-Wolfhard Schicke Cc: pgsql-performance@postgresql.org Subject: Re: Inherited Table Query Planning (fwd) In-reply-to: <66F42056CEA99778F918C7C9@[192.168.1.72]> References: <66F42056CEA99778F918C7C9@[192.168.1.72]> Comments: In-reply-to Jens-Wolfhard Schicke message dated "Wed, 27 Jul 2005 09:56:12 +0200" Date: Wed, 27 Jul 2005 10:09:08 -0400 Message-ID: <25523.1122473348@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/415 X-Sequence-Number: 13656 Jens-Wolfhard Schicke writes: > Is there a way to make the query planner consider pulling inner appends > outside joins? Not at the moment. regards, tom lane From pgsql-admin-owner@postgresql.org Wed Jul 27 12:12:47 2005 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C312A52C45 for ; Wed, 27 Jul 2005 12:12:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 76070-10 for ; Wed, 27 Jul 2005 15:12:28 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.207]) by svr1.postgresql.org (Postfix) with ESMTP id 8DB7E52BC5 for ; Wed, 27 Jul 2005 12:12:27 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 13so177137nzn for ; Wed, 27 Jul 2005 08:12:27 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=jP5pTM/Kw5CCnHGTd15Jb+rUJ8yLJYNzeSxd27A9iRgfl1zVcMzY06fUXUU/CG4KGAticXT68cF0uSnUlPK1kkDEw4obp96Z7/NITmxUquxrcc+bFBRhL60bdqyAA93iNLm+rSd9caaeVkjryONWv35UwBO+dt3TZE2iEgh1aBg= Received: by 10.36.177.1 with SMTP id z1mr983068nze; Wed, 27 Jul 2005 08:12:27 -0700 (PDT) Received: by 10.36.25.13 with HTTP; Wed, 27 Jul 2005 08:12:27 -0700 (PDT) Message-ID: <1d219a6f05072708122cf6abeb@mail.gmail.com> Date: Wed, 27 Jul 2005 11:12:27 -0400 From: Chris Hoover Reply-To: Chris Hoover To: pgsql-admin@postgresql.org Subject: Help with view performance problem Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=AWL, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/222 X-Sequence-Number: 18278 I am having a problem with a view on one of my db's. This view is trying to sequentially can the 2 tables it is accessing. However, when I explain the view on most of my other db's (all have the same schema's), it is using the indexes. Can anyone please help me understand why postgres is choosing to sequenially scan both tables? Both tables in the view have a primary key defined on inv_nbr, inv_qfr. Vacuum and analyze have been run on the tables in question to try and make sure stats are up to date. Thanks, Chris PG - 7.3.4 RH 2.1 Here is the view definition: SELECT DISTINCT clmcom1.inv_nbr AS inventory_number,=20 clmcom1.inv_qfr AS inventory_qualifier,=20 clmcom1.pat_addr_1 AS patient_address_1,=20 clmcom1.pat_addr_2 AS patient_address_2,=20 clmcom1.pat_city AS patient_city,=20 clmcom1.pat_cntry AS patient_country,=20 clmcom1.pat_dob AS patient_date_of_birth,=20 clmcom1.pat_gender_cd AS patient_gender_code, clmcom1.pat_info_pregnancy_ind AS pregnancy_i= nd, clmcom1.pat_state AS patient_state,=20 clmcom1.pat_suffix AS patient_suffix,=20 clmcom1.pat_zip AS patient_zip_code,=20 clmcom1.payto_addr_1 AS payto_address_1,=20 clmcom1.payto_addr_2 AS payto_address_2,=20 clmcom1.payto_city,=20 clmcom1.payto_cntry AS payto_country,=20 clmcom1.payto_f_name AS payto_first_name,=20 clmcom1.payto_m_name AS payto_middle_name, clmcom1.payto_state,=20 clmcom1.payto_zip AS payto_zip_code,=20 clmcom1.clm_tot_clm_chgs AS total_claim_charg= e,=20 clmcom1.bill_l_name_org AS billing_last_name_or_org, clmcom1.clm_delay_rsn_cd AS claim_delay_reason_code, clmcom1.clm_submit_rsn_cd AS claim_submit_reason_code, clmcom1.payto_l_name_org AS payto_last_name_or_org, clmcom1.payto_prim_id AS payto_primary_id, = =20 clmcom1.bill_prim_id AS billing_prov_primary_= id, clmcom1.clm_tot_ncov_chgs AS total_ncov_charg= e,=20 clmcom2.contract_amt AS contract_amount,=20 clmcom2.svc_fac_or_lab_name,=20 clmcom2.svc_fac_addr_1 AS svc_fac_address_1,= =20 clmcom2.svc_fac_addr_2 AS svc_fac_address_2,= =20 clmcom2.svc_fac_city,=20 clmcom2.svc_fac_zip AS svc_fac_zip_code=20 FROM (clmcom1 LEFT JOIN clmcom2 ON (((clmcom1.inv_nbr =3D clmcom2.inv_nbr) AND =20 (clmcom1.inv_qfr =3D clmcom2.inv_qfr)))) ORDER BY clmcom1.inv_nbr,=20 clmcom1.inv_qfr,=20 clmcom1.pat_addr_1,=20 clmcom1.pat_addr_2,=20 clmcom1.pat_city,=20 clmcom1.pat_cntry, clmcom1.pat_dob,=20 clmcom1.pat_gender_cd,=20 clmcom1.pat_info_pregnancy_ind,=20 clmcom1.pat_state,=20 clmcom1.pat_suffix,=20 clmcom1.pat_zip,=20 clmcom1.payto_addr_1,=20 clmcom1.payto_addr_2,=20 clmcom1.payto_city,=20 clmcom1.payto_cntry, clmcom1.payto_f_name,=20 clmcom1.payto_m_name,=20 clmcom1.payto_state,=20 clmcom1.payto_zip,=20 clmcom1.clm_tot_clm_chgs,=20 clmcom1.bill_l_name_org,=20 clmcom1.clm_delay_rsn_cd,=20 clmcom1.clm_submit_rsn_cd,=20 clmcom1.payto_l_name_org,=20 clmcom1.payto_prim_id,=20 clmcom1.bill_prim_id,=20 clmcom1.clm_tot_ncov_chgs,=20 clmcom2.contract_amt,=20 clmcom2.svc_fac_or_lab_name,=20 clmcom2.svc_fac_addr_1,=20 clmcom2.svc_fac_addr_2,=20 clmcom2.svc_fac_city,=20 clmcom2.svc_fac_zip; Here is the explain analyze from the problem db: prob_db=3D# explain analyze select * from clm_com; =20 QUERY PLAN ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= -------------------------- Subquery Scan clm_com (cost=3D1039824.35..1150697.61 rows=3D126712 width=3D367) (actual time=3D311792.78..405819.03 rows=3D1266114 loops=3D1) -> Unique (cost=3D1039824.35..1150697.61 rows=3D126712 width=3D367) (actual time=3D311792.74..386313.14 rows=3D1266114 loops=3D1) -> Sort (cost=3D1039824.35..1042992.16 rows=3D1267123 width=3D367) (actual time=3D311792.74..338189.48 rows=3D1266114 loops=3D1) Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, clmcom1.pat_zip, clmcom1.payto_addr_1, clmcom1.payto_addr_2, clmcom1.payto_city, clmcom1.payto_cntry, clmcom1.payto_f_name, clmcom1.payto_m_name, clmcom1.payto_state, clmcom1.payto_zip, clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip -> Hash Join (cost=3D132972.78..548171.70 rows=3D1267123 width=3D367) (actual time=3D16999.32..179359.43 rows=3D1266114 loops=3D1) Hash Cond: ("outer".inv_nbr =3D "inner".inv_nbr) Join Filter: ("outer".inv_qfr =3D "inner".inv_qfr) -> Seq Scan on clmcom1 (cost=3D0.00..267017.23 rows=3D1267123 width=3D271) (actual time=3D0.11..84711.83 rows=3D1266114 loops=3D1) -> Hash (cost=3D111200.82..111200.82 rows=3D1269582 width=3D96) (actual time=3D16987.45..16987.45 rows=3D0 loops=3D1) -> Seq Scan on clmcom2=20 (cost=3D0.00..111200.82 rows=3D1269582 width=3D96) (actual time=3D0.07..12164.81 rows=3D1266108 loops=3D1) Total runtime: 407317.47 msec (11 rows) ~ Here is the explain analyze from a good db (on the same postgres cluster); good_db=3D# explain analyze select * from clm_com; =20 =20 =20 =20 =20 QUERY PLAN ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= -------------------------- Subquery Scan clm_com (cost=3D78780.59..89498.29 rows=3D12249 width=3D359= ) (actual time=3D73045.36..79974.37 rows=3D122494 loops=3D1) -> Unique (cost=3D78780.59..89498.29 rows=3D12249 width=3D359) (actual time=3D73045.28..78031.99 rows=3D122494 loops=3D1) -> Sort (cost=3D78780.59..79086.81 rows=3D122488 width=3D359) (actual time=3D73045.28..73362.94 rows=3D122494 loops=3D1) Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, clmcom1.pat_zip, clmcom1.payto_addr_1, clmcom1.payto_addr_2, clmcom1.payto_city, clmcom1.payto_cntry, clmcom1.payto_f_name, clmcom1.payto_m_name, clmcom1.payto_state, clmcom1.payto_zip, clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip -> Merge Join (cost=3D0.00..56945.12 rows=3D122488 width=3D359) (actual time=3D54.76..71635.65 rows=3D122494 loops=3D1) Merge Cond: (("outer".inv_nbr =3D "inner".inv_nbr) AND ("outer".inv_qfr =3D "inner".inv_qfr)) -> Index Scan using clmcom1_pkey on clmcom1=20 (cost=3D0.00..38645.61 rows=3D122488 width=3D267) (actual time=3D25.60..49142.16 rows=3D122494 loops=3D1) -> Index Scan using clmcom2_pkey on clmcom2=20 (cost=3D0.00..16004.08 rows=3D122488 width=3D92) (actual time=3D29.09..19418.94 rows=3D122494 loops=3D1) Total runtime: 80162.26 msec (9 rows) From pgsql-performance-owner@postgresql.org Wed Jul 27 13:20:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 62C6652FBB for ; Wed, 27 Jul 2005 13:20:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 01620-02 for ; Wed, 27 Jul 2005 16:20:01 +0000 (GMT) Received: from wsmailap01.firstam.com (outbound-smtp01.firstam.com [208.246.101.197]) by svr1.postgresql.org (Postfix) with ESMTP id 98C6252FA1 for ; Wed, 27 Jul 2005 13:19:59 -0300 (ADT) Received: from 172.21.131.5 by wsmailap02.firstam.com with ESMTP ( Tumbleweed MMS SMTP Relay (MMS v5.6.3)); Wed, 27 Jul 2005 09:19:33 -0700 X-Server-Uuid: 7F57BFB8-2F8F-44C6-8C75-DD639B93CEA6 Received: from anammx01.ana.firstamdata.com ([192.168.173.35]) by famarp01.firstam.com (MOS 3.5.6-GR) with SMTP id EGN46487; Wed, 27 Jul 2005 09:19:42 -0700 (PDT) Received: from pisgana01sxch21.ana.firstamdata.com (Not Verified[192.168.173.71]) by mailgateway.firstam.com (post.office MTA v5.0 0924 ) with ESMTP id ; Wed, 27 Jul 2005 09:19:42 -0700 Received: from pisgana01sxch01.ana.firstamdata.com ([192.168.173.70]) by pisgana01sxch21.ana.firstamdata.com with Microsoft SMTPSVC(6.0.3790.211 ); Wed, 27 Jul 2005 09:19:42 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: "Vacuum Full Analyze" taking so long Date: Wed, 27 Jul 2005 09:19:42 -0700 Message-ID: Thread-Topic: [PERFORM] "Vacuum Full Analyze" taking so long Thread-Index: AcWRygCzjHVwQ9A7SjSkd+zv60r8ogA/N/LA From: "Tomeh, Husam" To: pgsql-performance@postgresql.org X-OriginalArrivalTime: 27 Jul 2005 16:19:42.0305 (UTC) FILETIME=[FEA0AD10:01C592C6] X-WSS-ID: 6EF96B9B23K1603738-02-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.025 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/416 X-Sequence-Number: 13657 Thank you all for your great input. It sure helped.=20 --=20 =20Husam=20 -----Original Message----- From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of Jochem van Dieten Sent: Tuesday, July 26, 2005 2:58 AM To: pgsql-performance@postgresql.org Subject: Re: [PERFORM] "Vacuum Full Analyze" taking so long Tomeh, Husam wrote: > The other question I have. What would be the proper approach to=20 > rebuild indexes. I re-indexes and then run vacuum/analyze. Should I=20 > not use the re-index approach, and instead, drop the indexes, vacuum=20 > the tables, and then create the indexes, then run analyze on tables and indexes?? If you just want to rebuild indexes, just drop and recreate. However, you are also running a VACUUM FULL, so I presume you have deleted a significant number of rows and want to recover the space that was in use by them. In that scenario, it is often better to CLUSTER the table to force a rebuild. While VACUUM FULL moves the tuples around inside the existing file(s), CLUSTER simply creates new file(s), moves all the non-deleted tuples there and then swaps the old and the new files. There can be a significant performance increase in doing so (but you obviously need to have some free diskspace). If you CLUSTER your table it will be ordered by the index you specify. There can be a performance increase in doing so, but if you don't want to you can also do a no-op ALTER TABLE and change a column to a datatype that is the same as it already has. This too will force a rewrite of the table but without ordering the tuples. So in short my recommendations: - to rebuild indexes, just drop and recreate the indexes - to rebuild everything because there is space that can bepermanently reclaimed, drop indexes, cluster or alter the table, recreate the indexes and anlyze the table Jochem ---------------------------(end of broadcast)--------------------------- TIP 3: Have you checked our extensive FAQ? =20 http://www.postgresql.org/docs/faq ********************************************************************** This message contains confidential information intended only for the=20 use of the addressee(s) named above and may contain information that=20 is legally privileged. If you are not the addressee, or the person=20 responsible for delivering it to the addressee, you are hereby=20 notified that reading, disseminating, distributing or copying this=20 message is strictly prohibited. If you have received this message by=20 mistake, please immediately notify us by replying to the message and=20 delete the original message immediately thereafter. Thank you. FADLD Tag ********************************************************************** From pgsql-performance-owner@postgresql.org Wed Jul 27 13:28:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0CF0E52E52 for ; Wed, 27 Jul 2005 13:28:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 02541-06 for ; Wed, 27 Jul 2005 16:28:19 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id A2A2452C2C for ; Wed, 27 Jul 2005 13:28:18 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [63.195.55.98] (account josh@agliodbs.com HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7660937; Wed, 27 Jul 2005 09:30:29 -0700 From: Josh Berkus Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Re: [Bizgres-general] Re: faster INSERT with possible Date: Wed, 27 Jul 2005 09:29:09 -0700 User-Agent: KMail/1.8 Cc: "Luke Lonergan" , "Hannu Krosing" , "John A Meinel" , "Dan Harris" , "bizgres-general" References: In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507270929.10251.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.012 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/417 X-Sequence-Number: 13658 Luke, > Well - now that I test it, it appears you are correct, temp table COPY > bypasses WAL - thanks for pointing it out! RIght. The problem is bypassing WAL for loading new "scratch" tables which aren't TEMPORARY tables. We need to do this for multi-threaded ETL, since: a) Temp tables can't be shared by several writers, and b) you can't index a temp table. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-admin-owner@postgresql.org Wed Jul 27 13:29:22 2005 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C822E52E87 for ; Wed, 27 Jul 2005 13:29:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04284-02 for ; Wed, 27 Jul 2005 16:29:16 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.203]) by svr1.postgresql.org (Postfix) with ESMTP id 87BE452E52 for ; Wed, 27 Jul 2005 13:29:15 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 8so169217nzo for ; Wed, 27 Jul 2005 09:29:15 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=P8e4w7oXjC0mSb023HnNcF+pzI0BLq9wi4mySnMtyJNKIEOmsgTZMddlHx02WYCSY545ZnRK+iehHOg3Gu/5STzfG2bsy/+S1Kb9OAYegjxvnZ/eRnryZmM1qiIk/naadKGRofBZvpTT+sN5BhlOvp7mpNpck2zHbq2JelVw5Uc= Received: by 10.36.7.20 with SMTP id 20mr1087241nzg; Wed, 27 Jul 2005 09:29:15 -0700 (PDT) Received: by 10.36.25.13 with HTTP; Wed, 27 Jul 2005 09:29:14 -0700 (PDT) Message-ID: <1d219a6f050727092927746db1@mail.gmail.com> Date: Wed, 27 Jul 2005 12:29:14 -0400 From: Chris Hoover Reply-To: Chris Hoover To: pgsql-admin@postgresql.org Subject: Re: Help with view performance problem In-Reply-To: <1d219a6f05072708122cf6abeb@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1d219a6f05072708122cf6abeb@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/226 X-Sequence-Number: 18282 I did some more testing, and ran the explain analyze on the problem.=20 In my session I did a set enable_hashjoin =3D false and then ran the analyze. This caused it to use the indexes as I have been expecting it to do. Now, how can I get it to use the indexes w/o manipulating the environment? What make postgresql want to sequentially scan and use a hash join? thanks, Chris explain analyze with set_hashjoin=3Dfalse; prob_db=3D#explain analyze select * from clm_com; =20 QUERY PLAN ---------------------------------------------------------------------------= ------------------------------------------------------------------- ---------------------------------------------------------------------------= ------------------------------------------------------------------- ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ------------------------------------------ Subquery Scan clm_com (cost=3D1057975.45..1169021.26 rows=3D126910 width=3D366) (actual time=3D142307.99..225997.22 rows=3D1268649 loops=3D1) -> Unique (cost=3D1057975.45..1169021.26 rows=3D126910 width=3D366) (actual time=3D142307.96..206082.30 rows=3D1268649 loops=3D1) -> Sort (cost=3D1057975.45..1061148.19 rows=3D1269095 width=3D366) (actual time=3D142307.95..156019.01 rows=3D1268649 loops=3D1) Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, clmcom1.pat_zip, clmcom1.payto_addr_1, clmcom1.payto_addr_2, clmcom1.payto_city, clmcom1.payto_cntry, clmcom1.payto_f_name, clmcom1.payto_m_name, clmcom1.payto_state, clmcom1.payto_zip, clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip -> Merge Join (cost=3D0.00..565541.46 rows=3D1269095 width=3D366) (actual time=3D464.89..130638.06 rows=3D1268649 loops=3D1) Merge Cond: ("outer".inv_nbr =3D "inner".inv_nbr) Join Filter: ("outer".inv_qfr =3D "inner".inv_qfr) -> Index Scan using clmcom1_inv_nbr_iview_idx on clmcom1 (cost=3D0.00..380534.32 rows=3D1269095 width=3D270) (actual time=3D0.27..82159.37 rows=3D1268649 loops=3D1) -> Index Scan using clmcom2_inv_nbr_iview_idx on clmcom2 (cost=3D0.00..159636.25 rows=3D1271198 width=3D96) (actual time=3D464.56..21774.02 rows=3D1494019 loops=3D1) Total runtime: 227369.39 msec (10 rows) On 7/27/05, Chris Hoover wrote: > I am having a problem with a view on one of my db's. This view is > trying to sequentially can the 2 tables it is accessing. However, > when I explain the view on most of my other db's (all have the same > schema's), it is using the indexes. Can anyone please help me > understand why postgres is choosing to sequenially scan both tables? >=20 > Both tables in the view have a primary key defined on inv_nbr, > inv_qfr. Vacuum and analyze have been run on the tables in question > to try and make sure stats are up to date. >=20 > Thanks, >=20 > Chris > PG - 7.3.4 > RH 2.1 >=20 >=20 > Here is the view definition: > SELECT DISTINCT clmcom1.inv_nbr AS inventory_number, > clmcom1.inv_qfr AS inventory_qualifier, > clmcom1.pat_addr_1 AS patient_address_1, > clmcom1.pat_addr_2 AS patient_address_2, > clmcom1.pat_city AS patient_city, > clmcom1.pat_cntry AS patient_country, > clmcom1.pat_dob AS patient_date_of_birth, > clmcom1.pat_gender_cd AS patient_gender_cod= e, > clmcom1.pat_info_pregnancy_ind AS pregnancy= _ind, > clmcom1.pat_state AS patient_state, > clmcom1.pat_suffix AS patient_suffix, > clmcom1.pat_zip AS patient_zip_code, > clmcom1.payto_addr_1 AS payto_address_1, > clmcom1.payto_addr_2 AS payto_address_2, > clmcom1.payto_city, > clmcom1.payto_cntry AS payto_country, > clmcom1.payto_f_name AS payto_first_name, > clmcom1.payto_m_name AS payto_middle_name, > clmcom1.payto_state, > clmcom1.payto_zip AS payto_zip_code, > clmcom1.clm_tot_clm_chgs AS total_claim_cha= rge, > clmcom1.bill_l_name_org AS > billing_last_name_or_org, > clmcom1.clm_delay_rsn_cd AS > claim_delay_reason_code, > clmcom1.clm_submit_rsn_cd AS > claim_submit_reason_code, > clmcom1.payto_l_name_org AS > payto_last_name_or_org, > clmcom1.payto_prim_id AS payto_primary_id, > clmcom1.bill_prim_id AS billing_prov_primar= y_id, > clmcom1.clm_tot_ncov_chgs AS total_ncov_cha= rge, > clmcom2.contract_amt AS contract_amount, > clmcom2.svc_fac_or_lab_name, > clmcom2.svc_fac_addr_1 AS svc_fac_address_1= , > clmcom2.svc_fac_addr_2 AS svc_fac_address_2= , > clmcom2.svc_fac_city, > clmcom2.svc_fac_zip AS svc_fac_zip_code > FROM (clmcom1 LEFT JOIN clmcom2 ON (((clmcom1.inv_nbr =3D > clmcom2.inv_nbr) AND >=20 > (clmcom1.inv_qfr =3D clmcom2.inv_qfr)))) > ORDER BY clmcom1.inv_nbr, > clmcom1.inv_qfr, > clmcom1.pat_addr_1, > clmcom1.pat_addr_2, > clmcom1.pat_city, > clmcom1.pat_cntry, > clmcom1.pat_dob, > clmcom1.pat_gender_cd, > clmcom1.pat_info_pregnancy_ind, > clmcom1.pat_state, > clmcom1.pat_suffix, > clmcom1.pat_zip, > clmcom1.payto_addr_1, > clmcom1.payto_addr_2, > clmcom1.payto_city, > clmcom1.payto_cntry, > clmcom1.payto_f_name, > clmcom1.payto_m_name, > clmcom1.payto_state, > clmcom1.payto_zip, > clmcom1.clm_tot_clm_chgs, > clmcom1.bill_l_name_org, > clmcom1.clm_delay_rsn_cd, > clmcom1.clm_submit_rsn_cd, > clmcom1.payto_l_name_org, > clmcom1.payto_prim_id, > clmcom1.bill_prim_id, > clmcom1.clm_tot_ncov_chgs, > clmcom2.contract_amt, > clmcom2.svc_fac_or_lab_name, > clmcom2.svc_fac_addr_1, > clmcom2.svc_fac_addr_2, > clmcom2.svc_fac_city, > clmcom2.svc_fac_zip; >=20 >=20 > Here is the explain analyze from the problem db: > prob_db=3D# explain analyze select * from clm_com; >=20 >=20 >=20 > QUERY PLAN >=20 >=20 >=20 > -------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------- > Subquery Scan clm_com (cost=3D1039824.35..1150697.61 rows=3D126712 > width=3D367) (actual time=3D311792.78..405819.03 rows=3D1266114 loops=3D1= ) > -> Unique (cost=3D1039824.35..1150697.61 rows=3D126712 width=3D367) > (actual time=3D311792.74..386313.14 rows=3D1266114 loops=3D1) > -> Sort (cost=3D1039824.35..1042992.16 rows=3D1267123 > width=3D367) (actual time=3D311792.74..338189.48 rows=3D1266114 loops=3D1= ) > Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, > clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, > clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, > clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, > clmcom1.pat_zip, clmcom1.payto_addr_1, > clmcom1.payto_addr_2, clmcom1.payto_city, clmcom1.payto_cntry, > clmcom1.payto_f_name, clmcom1.payto_m_name, clmcom1.payto_state, > clmcom1.payto_zip, clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, > clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, > clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, > clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, > clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, > clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip > -> Hash Join (cost=3D132972.78..548171.70 rows=3D1267123 > width=3D367) (actual time=3D16999.32..179359.43 rows=3D1266114 loops=3D1) > Hash Cond: ("outer".inv_nbr =3D "inner".inv_nbr) > Join Filter: ("outer".inv_qfr =3D "inner".inv_qfr) > -> Seq Scan on clmcom1 (cost=3D0.00..267017.23 > rows=3D1267123 width=3D271) (actual time=3D0.11..84711.83 rows=3D1266114 > loops=3D1) > -> Hash (cost=3D111200.82..111200.82 rows=3D126958= 2 > width=3D96) (actual time=3D16987.45..16987.45 rows=3D0 loops=3D1) > -> Seq Scan on clmcom2 > (cost=3D0.00..111200.82 rows=3D1269582 width=3D96) (actual > time=3D0.07..12164.81 rows=3D1266108 loops=3D1) > Total runtime: 407317.47 msec > (11 rows) > ~ >=20 >=20 >=20 > Here is the explain analyze from a good db (on the same postgres cluster)= ; > good_db=3D# explain analyze select * from clm_com; >=20 >=20 >=20 >=20 >=20 > QUERY PLAN > -------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------- > Subquery Scan clm_com (cost=3D78780.59..89498.29 rows=3D12249 width=3D3= 59) > (actual time=3D73045.36..79974.37 rows=3D122494 loops=3D1) > -> Unique (cost=3D78780.59..89498.29 rows=3D12249 width=3D359) (actu= al > time=3D73045.28..78031.99 rows=3D122494 loops=3D1) > -> Sort (cost=3D78780.59..79086.81 rows=3D122488 width=3D359) > (actual time=3D73045.28..73362.94 rows=3D122494 loops=3D1) > Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, > clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, > clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, > clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, > clmcom1.pat_zip, clmcom1.payto_addr_1, clmcom1.payto_addr_2, > clmcom1.payto_city, clmcom1.payto_cntry, clmcom1.payto_f_name, > clmcom1.payto_m_name, clmcom1.payto_state, clmcom1.payto_zip, > clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, > clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, > clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, > clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, > clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, > clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip > -> Merge Join (cost=3D0.00..56945.12 rows=3D122488 > width=3D359) (actual time=3D54.76..71635.65 rows=3D122494 loops=3D1) > Merge Cond: (("outer".inv_nbr =3D "inner".inv_nbr) > AND ("outer".inv_qfr =3D "inner".inv_qfr)) > -> Index Scan using clmcom1_pkey on clmcom1 > (cost=3D0.00..38645.61 rows=3D122488 width=3D267) (actual > time=3D25.60..49142.16 rows=3D122494 loops=3D1) > -> Index Scan using clmcom2_pkey on clmcom2 > (cost=3D0.00..16004.08 rows=3D122488 width=3D92) (actual > time=3D29.09..19418.94 rows=3D122494 loops=3D1) > Total runtime: 80162.26 msec > (9 rows) > From pgsql-performance-owner@postgresql.org Wed Jul 27 15:28:47 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 54AD5529F9 for ; Wed, 27 Jul 2005 15:28:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28819-04 for ; Wed, 27 Jul 2005 18:28:44 +0000 (GMT) Received: from mail63.csoft.net (leary3.csoft.net [63.111.22.74]) by svr1.postgresql.org (Postfix) with SMTP id 011AA529EF for ; Wed, 27 Jul 2005 15:28:43 -0300 (ADT) Received: (qmail 4263 invoked by uid 1112); 27 Jul 2005 18:28:44 -0000 Date: Wed, 27 Jul 2005 13:28:44 -0500 (EST) From: Kris Jurka X-X-Sender: books@leary.csoft.net To: Josh Berkus Cc: pgsql-performance@postgresql.org, John A Meinel , bizgres-general , Dan Harris Subject: Re: [Bizgres-general] Re: faster INSERT with possible In-Reply-To: <200507270929.10251.josh@agliodbs.com> Message-ID: References: <200507270929.10251.josh@agliodbs.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/418 X-Sequence-Number: 13659 On Wed, 27 Jul 2005, Josh Berkus wrote: > b) you can't index a temp table. > jurka# create temp table t (a int); CREATE jurka# create index myi on t(a); CREATE From pgsql-performance-owner@postgresql.org Wed Jul 27 17:29:42 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B88D852C05; Wed, 27 Jul 2005 17:27:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 69426-06; Wed, 27 Jul 2005 20:27:12 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 8B97E52C11; Wed, 27 Jul 2005 17:27:11 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7662106; Wed, 27 Jul 2005 13:29:26 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: pgsql-hackers@postgresql.org Subject: wal_buffer tests in Date: Wed, 27 Jul 2005 13:30:01 -0700 User-Agent: KMail/1.8 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507271330.02083.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.049 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/419 X-Sequence-Number: 13660 Folks, I ran a wal_buffer test series. It appears that increasing the wal_buffers is indeed very important for OLTP applications, potentially resulting in as much as a 15% average increase in transaction processing. What's interesting is that this is not just true for 8.1, it's true for 8.0.3 as well. More importantly, 8.1 performance is somehow back up to above-8.0 levels. Something was broken in June that's got fixed (this test series is based on July 3 CVS) but I don't know what. Clues? Test results are here: http://pgfoundry.org/docman/view.php/1000041/79/wal_buffer_test.pdf As always, detailed test results are available from OSDL, just use: http://khack.osdl.org/stp/# where # is the test number. -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Wed Jul 27 18:21:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C2A5052C05 for ; Wed, 27 Jul 2005 18:21:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 83233-02 for ; Wed, 27 Jul 2005 21:21:43 +0000 (GMT) Received: from dd01.profihoster.net (dd01.profihoster.net [84.233.130.40]) by svr1.postgresql.org (Postfix) with ESMTP id 861BF52ADA for ; Wed, 27 Jul 2005 18:21:42 -0300 (ADT) Received: from [192.168.54.33] (Iaec4.i.pppool.de [85.73.174.196]) by dd01.profihoster.net (Postfix) with ESMTP id AF0C32383C3 for ; Wed, 27 Jul 2005 23:12:26 +0200 (CEST) Message-ID: <42E7FAE3.7070604@laliluna.de> Date: Wed, 27 Jul 2005 23:21:39 +0200 From: Sebastian Hennebrueder User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: de-DE, de, en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Mirroring PostgreSQL database References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/420 X-Sequence-Number: 13661 Shashi Kanth Boddula schrieb: > Hi, > I have one customer who is using PostgreSQL 7.4.8 on Linux . He has > some problems with database mirroring . The details are follows. > The customer is using Linux on which PostgreSQL 7.4.8 along with Jboss > 3.2.3 is running . He has 2 servers , one is acting as a live server > (primary) and another is acting as a fail-over (secondary) server > . Secondary server is placed in remote location . These servers are > acting as a Attendence server for daily activities . Nearly 50,000 > employees depend on the live server . > > The customer is using DBmirror tool to mirror the database records of > primary to secondary . The customer is complaining that there is one > day (24 hours) delay between primary and secondray for database > synchronization . They have dedicated line and bandwidth , but still > the problems exists. > > I just want to know , for immediate data mirroring , what is the best > way for PostgreSQL . PostgreSQL is offering many mirror tools , but > which one is the best ?. Is there any other way to accomplish the task ? > > Thank you . Waiting for your reply. > > > Thanks & Regards, > Shashi Kanth > Consultant - Linux > RHCE , LPIC-2 > Onward Novell - Bangalore > 9886455567 > > For java based solution you could also have a look at x-jdbc or xjdbc. But before you should find out what the reason for the delay is actually. When the backup server is to slow, it may be not important which mirroring tool you use. -- Best Regards / Viele Gr��e Sebastian Hennebrueder ---- http://www.laliluna.de Tutorials for JSP, JavaServer Faces, Struts, Hibernate and EJB Get support, education and consulting for these technologies - uncomplicated and cheap. From pgsql-performance-owner@postgresql.org Wed Jul 27 18:38:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A9F0252B06 for ; Wed, 27 Jul 2005 18:38:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86196-02 for ; Wed, 27 Jul 2005 21:38:14 +0000 (GMT) Received: from cmailg4.svr.pol.co.uk (cmailg4.svr.pol.co.uk [195.92.195.174]) by svr1.postgresql.org (Postfix) with ESMTP id 3B86D52A0B for ; Wed, 27 Jul 2005 18:38:13 -0300 (ADT) Received: from modem-3758.llama.dialup.pol.co.uk ([217.135.190.174] helo=192.168.0.102) by cmailg4.svr.pol.co.uk with esmtp (Exim 4.41) id 1DxtbO-0004eR-Ni; Wed, 27 Jul 2005 22:38:11 +0100 Subject: Re: [Bizgres-general] Re: faster INSERT with possible From: Simon Riggs To: Josh Berkus Cc: pgsql-performance@postgresql.org, John A Meinel , bizgres-general , Dan Harris In-Reply-To: <200507270929.10251.josh@agliodbs.com> References: <200507270929.10251.josh@agliodbs.com> Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 27 Jul 2005 22:37:47 +0100 Message-Id: <1122500267.3670.184.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.053 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/421 X-Sequence-Number: 13662 On Wed, 2005-07-27 at 09:29 -0700, Josh Berkus wrote: > Luke, > > > Well - now that I test it, it appears you are correct, temp table COPY > > bypasses WAL - thanks for pointing it out! > > RIght. The problem is bypassing WAL for loading new "scratch" tables which > aren't TEMPORARY tables. We need to do this for multi-threaded ETL, since: > a) Temp tables can't be shared by several writers, and > b) you can't index a temp table. The description of "scratch" tables might need some slight clarification. It kindof makes it sound like temp tables. I had in mind the extra tables that an application sometimes needs to operate faster. Denormalisations, pre-joined tables, pre-calculated results, aggregated data. These are not temporary tables, just part of the application - multi-user tables that stay across shutdown/restart. If you have gallons of GB, you will probably by looking to make use of such tables. You can use such tables for the style of ETL known as ELT, but that is not the only use. Best Regards, Simon Riggs From pgsql-hackers-owner@postgresql.org Wed Jul 27 18:41:37 2005 X-Original-To: pgsql-hackers-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 03E1152BD6 for ; Wed, 27 Jul 2005 18:41:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 84623-10 for ; Wed, 27 Jul 2005 21:41:31 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id A4B1E52ADA for ; Wed, 27 Jul 2005 18:41:24 -0300 (ADT) Received: from lemming.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6RLfNjA026772 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Wed, 27 Jul 2005 14:41:23 -0700 Message-Id: <200507272141.j6RLfNjA026772@smtp.osdl.org> Date: Wed, 27 Jul 2005 14:42:01 -0700 From: Mark Wong To: josh@agliodbs.com Cc: pgsql-hackers@postgresql.org Subject: Re: wal_buffer tests in In-Reply-To: <200507271330.02083.josh@agliodbs.com> References: <200507271330.02083.josh@agliodbs.com> Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; x86_64-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/1010 X-Sequence-Number: 71062 On Wed, 27 Jul 2005 13:30:01 -0700 Josh Berkus wrote: > Folks, > > I ran a wal_buffer test series. It appears that increasing the > wal_buffers is indeed very important for OLTP applications, potentially > resulting in as much as a 15% average increase in transaction processing. > What's interesting is that this is not just true for 8.1, it's true for > 8.0.3 as well. > > More importantly, 8.1 performance is somehow back up to above-8.0 levels. > Something was broken in June that's got fixed (this test series is based > on July 3 CVS) but I don't know what. Clues? > > Test results are here: > http://pgfoundry.org/docman/view.php/1000041/79/wal_buffer_test.pdf > > As always, detailed test results are available from OSDL, just use: > http://khack.osdl.org/stp/# > where # is the test number. The increase could actually be higher than 15% as 1800 notpm is about the max throughput you can have with 150 warehouses with the default thinktimes. The rule of thumb is about 12 * warehouses, for the throughput. Mark From pgsql-performance-owner@postgresql.org Wed Jul 27 18:53:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C6BE45295A for ; Wed, 27 Jul 2005 18:53:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86404-10 for ; Wed, 27 Jul 2005 21:52:53 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id D08F75293B for ; Wed, 27 Jul 2005 18:52:51 -0300 (ADT) Received: (qmail 16473 invoked from network); 27 Jul 2005 23:52:57 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 27 Jul 2005 23:52:57 +0200 To: "Simon Riggs" , "Josh Berkus" Cc: pgsql-performance@postgresql.org, "John A Meinel" , bizgres-general , "Dan Harris" Subject: Re: [Bizgres-general] Re: faster INSERT with possible References: <200507270929.10251.josh@agliodbs.com> <1122500267.3670.184.camel@localhost.localdomain> Message-ID: Date: Wed, 27 Jul 2005 23:53:02 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <1122500267.3670.184.camel@localhost.localdomain> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/422 X-Sequence-Number: 13663 > I had in mind the extra tables that an application sometimes needs to > operate faster. Denormalisations, pre-joined tables, pre-calculated > results, aggregated data. These are not temporary tables, just part of > the application - multi-user tables that stay across shutdown/restart. You could also add caching search results for easy pagination without redoing always entirely on each page the Big Slow Search Query that every website has... From pgsql-hackers-owner@postgresql.org Wed Jul 27 19:01:52 2005 X-Original-To: pgsql-hackers-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E734A529A2 for ; Wed, 27 Jul 2005 19:01:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 89107-08 for ; Wed, 27 Jul 2005 22:01:21 +0000 (GMT) Received: from cmailm4.svr.pol.co.uk (cmailm4.svr.pol.co.uk [195.92.193.211]) by svr1.postgresql.org (Postfix) with ESMTP id DED1E529E5 for ; Wed, 27 Jul 2005 19:01:20 -0300 (ADT) Received: from modem-3758.llama.dialup.pol.co.uk ([217.135.190.174] helo=192.168.0.102) by cmailm4.svr.pol.co.uk with esmtp (Exim 4.41) id 1Dxtxo-0007AP-IV; Wed, 27 Jul 2005 23:01:20 +0100 Subject: Re: wal_buffer tests in From: Simon Riggs To: josh@agliodbs.com Cc: pgsql-hackers@postgresql.org In-Reply-To: <200507271330.02083.josh@agliodbs.com> References: <200507271330.02083.josh@agliodbs.com> Content-Type: text/plain Organization: 2nd Quadrant Date: Wed, 27 Jul 2005 23:00:55 +0100 Message-Id: <1122501655.3670.204.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.0.2 (2.0.2-3) Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.053 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/1012 X-Sequence-Number: 71064 On Wed, 2005-07-27 at 13:30 -0700, Josh Berkus wrote: > I ran a wal_buffer test series. It appears that increasing the > wal_buffers is indeed very important for OLTP applications, potentially > resulting in as much as a 15% average increase in transaction processing. > What's interesting is that this is not just true for 8.1, it's true for > 8.0.3 as well. The most important thing about these tests is that for the first time we have eliminated much of the post checkpoint noise-and-delay. Look at the response time charts between http://khack.osdl.org/stp/302959/results/0/rt.html and http://khack.osdl.org/stp/302963/results/0/rt.html This last set of results is a thing of beauty and I must congratulate everybody involved for getting here after much effort. The graphs are smooth, which shows a balanced machine. I'd like to repeat test 302963 with full_page_writes=false, to see if those response time spikes at checkpoint drop down to normal level. I think these results are valid for large DW data loads also. Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Wed Jul 27 19:35:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6C67952B75 for ; Wed, 27 Jul 2005 19:35:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 97480-04 for ; Wed, 27 Jul 2005 22:35:26 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id D9577529A2 for ; Wed, 27 Jul 2005 19:35:24 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6RMmUgr005676 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Wed, 27 Jul 2005 14:48:31 -0800 Message-ID: <42E80C28.2040608@aptalaska.net> Date: Wed, 27 Jul 2005 14:35:20 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Performance problems testing with Spamassassin 3.1.0 Bayes module. X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/423 X-Sequence-Number: 13664 I'm not sure how much this has been discussed on the list, but wasn't able to find anything relevant in the archives. The new Spamassassin is due out pretty soon. They are currently testing 3.1.0pre4. One of the things I hope to get out of this release is bayes word stats moved to a real RDBMS. They have separated the mysql BayesStore module from the PgSQL one so now postgres can use it's own queries. I loaded all of this stuff up on a test server and am finding that the bayes put performance is really not good enough for any real amount of mail load. The performance problems seems to be when the bayes module is inserting/updating. This is now handled by the token_put procedure. After playing with various indexes and what not I simply am unable to make this procedure perform any better. Perhaps someone on the list can spot the bottleneck and reveal why this procedure isn't performing that well or ways to make it better. I put the rest of the schema up at http://www.aptalaska.net/~matt.s/bayes/bayes_pg.sql in case someone needs to see it too. CREATE OR REPLACE FUNCTION put_token(integer, bytea, integer, integer, integer) RETURNS bool AS ' DECLARE inuserid ALIAS for $1; intoken ALIAS for $2; inspam_count ALIAS for $3; inham_count ALIAS for $4; inatime ALIAS for $5; got_token record; updated_atime_p bool; BEGIN updated_atime_p := FALSE; SELECT INTO got_token spam_count, ham_count, atime FROM bayes_token WHERE id = inuserid AND token = intoken; IF NOT FOUND THEN -- we do not insert negative counts, just return true IF (inspam_count < 0 OR inham_count < 0) THEN RETURN TRUE; END IF; INSERT INTO bayes_token (id, token, spam_count, ham_count, atime) VALUES (inuserid, intoken, inspam_count, inham_count, inatime); IF NOT FOUND THEN RAISE EXCEPTION ''unable to insert into bayes_token''; return FALSE; END IF; UPDATE bayes_vars SET token_count = token_count + 1 WHERE id = inuserid; IF NOT FOUND THEN RAISE EXCEPTION ''unable to update token_count in bayes_vars''; return FALSE; END IF; UPDATE bayes_vars SET newest_token_age = inatime WHERE id = inuserid AND newest_token_age < inatime; IF NOT FOUND THEN UPDATE bayes_vars SET oldest_token_age = inatime WHERE id = inuserid AND oldest_token_age > inatime; END IF; return TRUE; ELSE IF (inspam_count != 0) THEN -- no need to update atime if it is < the existing value IF (inatime < got_token.atime) THEN UPDATE bayes_token SET spam_count = spam_count + inspam_count WHERE id = inuserid AND token = intoken AND spam_count + inspam_count >= 0; ELSE UPDATE bayes_token SET spam_count = spam_count + inspam_count, atime = inatime WHERE id = inuserid AND token = intoken AND spam_count + inspam_count >= 0; IF FOUND THEN updated_atime_p := TRUE; END IF; END IF; END IF; IF (inham_count != 0) THEN -- no need to update atime is < the existing value or if it was already updated IF inatime < got_token.atime OR updated_atime_p THEN UPDATE bayes_token SET ham_count = ham_count + inham_count WHERE id = inuserid AND token = intoken AND ham_count + inham_count >= 0; ELSE UPDATE bayes_token SET ham_count = ham_count + inham_count, atime = inatime WHERE id = inuserid AND token = intoken AND ham_count + inham_count >= 0; IF FOUND THEN updated_atime_p := TRUE; END IF; END IF; END IF; IF updated_atime_p THEN UPDATE bayes_vars SET oldest_token_age = inatime WHERE id = inuserid AND oldest_token_age > inatime; END IF; return TRUE; END IF; END; ' LANGUAGE 'plpgsql'; From pgsql-performance-owner@postgresql.org Wed Jul 27 21:09:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C448452B4F for ; Wed, 27 Jul 2005 21:09:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 13471-07 for ; Thu, 28 Jul 2005 00:09:24 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 5DD9E52A1C for ; Wed, 27 Jul 2005 21:09:23 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7662855; Wed, 27 Jul 2005 17:11:39 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 Bayes module. Date: Wed, 27 Jul 2005 17:12:15 -0700 User-Agent: KMail/1.8 Cc: Matthew Schumacher References: <42E80C28.2040608@aptalaska.net> In-Reply-To: <42E80C28.2040608@aptalaska.net> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Message-Id: <200507271712.15299.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.049 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/424 X-Sequence-Number: 13665 Matt, > After playing with various indexes and what not I simply am unable to > make this procedure perform any better. =A0Perhaps someone on the list can > spot the bottleneck and reveal why this procedure isn't performing that > well or ways to make it better. Well, my first thought is that this is a pretty complicated procedure for=20 something you want to peform well. Is all this logic really necessary? = =20 How does it get done for MySQL? =2D-=20 =2D-Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Wed Jul 27 21:41:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C10AE528CF for ; Wed, 27 Jul 2005 21:41:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 74222-08 for ; Thu, 28 Jul 2005 00:41:23 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 9106F528E4 for ; Wed, 27 Jul 2005 21:41:22 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6S0f5ge013605; Wed, 27 Jul 2005 20:41:05 -0400 (EDT) To: Josh Berkus Cc: pgsql-performance@postgresql.org, "Luke Lonergan" , "Hannu Krosing" , "John A Meinel" , "Dan Harris" , "bizgres-general" Subject: Re: [Bizgres-general] Re: faster INSERT with possible In-reply-to: <200507270929.10251.josh@agliodbs.com> References: <200507270929.10251.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Wed, 27 Jul 2005 09:29:09 -0700" Date: Wed, 27 Jul 2005 20:41:05 -0400 Message-ID: <13604.1122511265@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/425 X-Sequence-Number: 13666 Josh Berkus writes: > RIght. The problem is bypassing WAL for loading new "scratch" tables which > aren't TEMPORARY tables. We need to do this for multi-threaded ETL, since: > a) Temp tables can't be shared by several writers, and > b) you can't index a temp table. This may not matter given point (a), but: point (b) is completely wrong. regards, tom lane From pgsql-performance-owner@postgresql.org Wed Jul 27 22:59:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9151A52956 for ; Wed, 27 Jul 2005 22:59:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 93519-01 for ; Thu, 28 Jul 2005 01:59:41 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 7F03F5293D for ; Wed, 27 Jul 2005 22:59:40 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6S2CnoW013649 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Wed, 27 Jul 2005 18:12:49 -0800 Message-ID: <42E83C09.2040508@aptalaska.net> Date: Wed, 27 Jul 2005 17:59:37 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: josh@agliodbs.com Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <200507271712.15299.josh@agliodbs.com> In-Reply-To: <200507271712.15299.josh@agliodbs.com> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/426 X-Sequence-Number: 13667 Josh Berkus wrote: > Matt, > > >>After playing with various indexes and what not I simply am unable to >>make this procedure perform any better. Perhaps someone on the list can >>spot the bottleneck and reveal why this procedure isn't performing that >>well or ways to make it better. > > > Well, my first thought is that this is a pretty complicated procedure for > something you want to peform well. Is all this logic really necessary? > How does it get done for MySQL? > I'm not sure if it's all needed, in mysql they have this simple schema: =============================================== CREATE TABLE bayes_expire ( id int(11) NOT NULL default '0', runtime int(11) NOT NULL default '0', KEY bayes_expire_idx1 (id) ) TYPE=MyISAM; CREATE TABLE bayes_global_vars ( variable varchar(30) NOT NULL default '', value varchar(200) NOT NULL default '', PRIMARY KEY (variable) ) TYPE=MyISAM; INSERT INTO bayes_global_vars VALUES ('VERSION','3'); CREATE TABLE bayes_seen ( id int(11) NOT NULL default '0', msgid varchar(200) binary NOT NULL default '', flag char(1) NOT NULL default '', PRIMARY KEY (id,msgid) ) TYPE=MyISAM; CREATE TABLE bayes_token ( id int(11) NOT NULL default '0', token char(5) NOT NULL default '', spam_count int(11) NOT NULL default '0', ham_count int(11) NOT NULL default '0', atime int(11) NOT NULL default '0', PRIMARY KEY (id, token), INDEX bayes_token_idx1 (token), INDEX bayes_token_idx2 (id, atime) ) TYPE=MyISAM; CREATE TABLE bayes_vars ( id int(11) NOT NULL AUTO_INCREMENT, username varchar(200) NOT NULL default '', spam_count int(11) NOT NULL default '0', ham_count int(11) NOT NULL default '0', token_count int(11) NOT NULL default '0', last_expire int(11) NOT NULL default '0', last_atime_delta int(11) NOT NULL default '0', last_expire_reduce int(11) NOT NULL default '0', oldest_token_age int(11) NOT NULL default '2147483647', newest_token_age int(11) NOT NULL default '0', PRIMARY KEY (id), UNIQUE bayes_vars_idx1 (username) ) TYPE=MyISAM; =============================================== Then they do this to insert the token: INSERT INTO bayes_token ( id, token, spam_count, ham_count, atime ) VALUES ( ?, ?, ?, ?, ? ) ON DUPLICATE KEY UPDATE spam_count = GREATEST(spam_count + ?, 0), ham_count = GREATEST(ham_count + ?, 0), atime = GREATEST(atime, ?) Or update the token: UPDATE bayes_vars SET $token_count_update newest_token_age = GREATEST(newest_token_age, ?), oldest_token_age = LEAST(oldest_token_age, ?) WHERE id = ? I think the reason why the procedure was written for postgres was because of the greatest and least statements performing poorly. Honestly, I'm not real up on writing procs, I was hoping the problem would be obvious to someone. schu From pgsql-performance-owner@postgresql.org Wed Jul 27 23:11:56 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B519E52991 for ; Wed, 27 Jul 2005 23:11:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94603-03 for ; Thu, 28 Jul 2005 02:11:44 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 3119952998 for ; Wed, 27 Jul 2005 23:11:43 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [63.195.55.98] (account josh@agliodbs.com HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7663255; Wed, 27 Jul 2005 19:14:01 -0700 From: Josh Berkus Organization: Aglio Database Solutions To: Matthew Schumacher Subject: Re: Performance problems testing with Spamassassin 3.1.0 Bayes module. Date: Wed, 27 Jul 2005 19:12:39 -0700 User-Agent: KMail/1.8 Cc: pgsql-performance@postgresql.org References: <42E80C28.2040608@aptalaska.net> <200507271712.15299.josh@agliodbs.com> <42E83C09.2040508@aptalaska.net> In-Reply-To: <42E83C09.2040508@aptalaska.net> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507271912.39780.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.012 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/427 X-Sequence-Number: 13668 Matt, > UPDATE bayes_vars SET > $token_count_update > newest_token_age = GREATEST(newest_token_age, ?), > oldest_token_age = LEAST(oldest_token_age, ?) > WHERE id = ? > > > I think the reason why the procedure was written for postgres was > because of the greatest and least statements performing poorly. Well, it might be because we don't have a built-in GREATEST or LEAST prior to 8.1. However, it's pretty darned easy to construct one. > Honestly, I'm not real up on writing procs, I was hoping the problem > would be obvious to someone. Well, there's the general performance tuning stuff of course (postgresql.conf) which if you've not done any of it will pretty dramatically affect your througput rates. And vacuum, analyze, indexes, etc. You should also look at ways to make the SP simpler. For example, you have a cycle that looks like: SELECT IF NOT FOUND INSERT ELSE UPDATE Which could be made shorter as: UPDATE IF NOT FOUND INSERT ... saving you one index scan. Also, I don't quite follow it, but the procedure seems to be doing at least two steps that the MySQL version isn't doing at all. If the PG version is doing more things, of course it's going to take longer. Finally, when you have a proc you're happy with, I suggest having an expert re-write it in C, which should double the procedure performance. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Wed Jul 27 23:17:10 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 4709052952 for ; Wed, 27 Jul 2005 23:17:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 95222-07 for ; Thu, 28 Jul 2005 02:17:05 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 6671752883 for ; Wed, 27 Jul 2005 23:17:05 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6S2UHfr017334 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Wed, 27 Jul 2005 18:30:18 -0800 Message-ID: <42E84021.9040904@aptalaska.net> Date: Wed, 27 Jul 2005 18:17:05 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Josh Berkus Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <200507271712.15299.josh@agliodbs.com> <42E83C09.2040508@aptalaska.net> <200507271912.39780.josh@agliodbs.com> In-Reply-To: <200507271912.39780.josh@agliodbs.com> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: unavailable X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/428 X-Sequence-Number: 13669 Josh Berkus wrote: > Matt, > > Well, it might be because we don't have a built-in GREATEST or LEAST prior to > 8.1. However, it's pretty darned easy to construct one. I was more talking about min() and max() but yea, I think you knew where I was going with it... > > Well, there's the general performance tuning stuff of course (postgresql.conf) > which if you've not done any of it will pretty dramatically affect your > througput rates. And vacuum, analyze, indexes, etc. I have gone though all that. > You should also look at ways to make the SP simpler. For example, you have a > cycle that looks like: > > SELECT > IF NOT FOUND > INSERT > ELSE > UPDATE > > Which could be made shorter as: > > UPDATE > IF NOT FOUND > INSERT > > ... saving you one index scan. > > Also, I don't quite follow it, but the procedure seems to be doing at least > two steps that the MySQL version isn't doing at all. If the PG version is > doing more things, of course it's going to take longer. > > Finally, when you have a proc you're happy with, I suggest having an expert > re-write it in C, which should double the procedure performance. > Sounds like I need to completely understand what the proc is doing and work on a rewrite. I'll look into writing it in C, I need to do some reading about how that works and exactly what it buys you. Thanks for the helpful comments. schu From pgsql-performance-owner@postgresql.org Thu Jul 28 02:16:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 83EDB52933 for ; Thu, 28 Jul 2005 02:16:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42644-03 for ; Thu, 28 Jul 2005 05:15:56 +0000 (GMT) Received: from mail.travelamericas.com (unknown [206.130.134.147]) by svr1.postgresql.org (Postfix) with SMTP id DD32B52903 for ; Thu, 28 Jul 2005 02:15:55 -0300 (ADT) Received: (qmail 28370 invoked from network); 28 Jul 2005 05:15:55 -0000 Received: from unknown (HELO ?10.0.0.253?) (10.0.0.253) by verkiel.travelamericas.com with SMTP; 28 Jul 2005 05:15:55 -0000 Message-ID: <42E86A03.1070803@travelamericas.com> Date: Wed, 27 Jul 2005 22:15:47 -0700 From: Chris Travers User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Left joining against two empty tables makes a query SLOW Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.037 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/429 X-Sequence-Number: 13670 Hi all; I have a customer who currently uses an application which had become slow. After doing some digging, I found the slow query: SELECT c.accno, c.description, c.link, c.category, ac.project_id, p.projectnumber, a.department_id, d.description AS department FROM chart c JOIN acc_trans ac ON (ac.chart_id = c.id) JOIN ar a ON (a.id = ac.trans_id) LEFT JOIN project p ON (ac.project_id = p.id) LEFT JOIN department d ON (d.id = a.department_id) WHERE a.customer_id = 11373 AND a.id IN ( SELECT max(id) FROM ar WHERE customer_id = 11373 ); (reformatted for readability) This is taking 10 seconds to run. Interestingly, both the project and department tables are blank, and if I omit them, the query becomes: SELECT c.accno, c.description, c.link, c.category, ac.project_id FROM chart c JOIN acc_trans ac ON (ac.chart_id = c.id) JOIN ar a ON (a.id = ac.trans_id) WHERE a.customer_id = 11373 AND a.id IN ( SELECT max(id) FROM ar WHERE customer_id = 11373 ); This takes 139ms. 1% of the previous query. The plan for the long query is: QUERY PLAN ---------------------------------------------------------------------------------------------------------------------------------------------------------------- Hash IN Join (cost=87337.25..106344.93 rows=41 width=118) (actual time=7615.843..9850.209 rows=10 loops=1) Hash Cond: ("outer".trans_id = "inner".max) -> Merge Right Join (cost=86620.57..100889.85 rows=947598 width=126) (actual time=7408.830..9200.435 rows=177769 loops=1) Merge Cond: ("outer".id = "inner".department_id) -> Index Scan using department_id_key on department d (cost=0.00..52.66 rows=1060 width=36) (actual time=0.090..0.090 rows=0 loops=1) -> Sort (cost=86620.57..87067.55 rows=178792 width=94) (actual time=7408.709..7925.843 rows=177769 loops=1) Sort Key: a.department_id -> Merge Right Join (cost=45871.18..46952.83 rows=178792 width=94) (actual time=4962.122..6671.319 rows=177769 loops=1) Merge Cond: ("outer".id = "inner".project_id) -> Index Scan using project_id_key on project p (cost=0.00..49.80 rows=800 width=36) (actual time=0.007..0.007 rows=0 loops=1) -> Sort (cost=45871.18..46318.16 rows=178792 width=62) (actual time=4962.084..5475.636 rows=177769 loops=1) Sort Key: ac.project_id -> Hash Join (cost=821.20..13193.43 rows=178792 width=62) (actual time=174.905..4295.685 rows=177769 loops=1) Hash Cond: ("outer".chart_id = "inner".id) -> Hash Join (cost=817.66..10508.02 rows=178791 width=20) (actual time=173.952..2840.824 rows=177769 loops=1) Hash Cond: ("outer".trans_id = "inner".id) -> Seq Scan on acc_trans ac (cost=0.00..3304.38 rows=181538 width=12) (actual time=0.062..537.753 rows=181322 loops=1) -> Hash (cost=659.55..659.55 rows=22844 width=8) (actual time=173.625..173.625 rows=0 loops=1) -> Seq Scan on ar a (cost=0.00..659.55 rows=22844 width=8) (actual time=0.022..101.828 rows=22844 loops=1) Filter: (customer_id = 11373) -> Hash (cost=3.23..3.23 rows=123 width=50) (actual time=0.915..0.915 rows=0 loops=1) -> Seq Scan on chart c (cost=0.00..3.23 rows=123 width=50) (actual time=0.013..0.528 rows=123 loops=1) -> Hash (cost=716.67..716.67 rows=1 width=4) (actual time=129.037..129.037 rows=0 loops=1) -> Subquery Scan "IN_subquery" (cost=716.66..716.67 rows=1 width=4) (actual time=129.017..129.025 rows=1 loops=1) -> Aggregate (cost=716.66..716.66 rows=1 width=4) (actual time=129.008..129.011 rows=1 loops=1) -> Seq Scan on ar (cost=0.00..659.55 rows=22844 width=4) (actual time=0.020..73.266 rows=22844 loops=1) Filter: (customer_id = 11373) Total runtime: 9954.133 ms (28 rows) The shorter query's plan is: QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------------------------------- Hash Join (cost=728.42..732.96 rows=8 width=50) (actual time=130.908..131.593 rows=10 loops=1) Hash Cond: ("outer".id = "inner".chart_id) -> Seq Scan on chart c (cost=0.00..3.23 rows=123 width=50) (actual time=0.006..0.361 rows=123 loops=1) -> Hash (cost=728.40..728.40 rows=8 width=8) (actual time=130.841..130.841 rows=0 loops=1) -> Nested Loop (cost=716.67..728.40 rows=8 width=8) (actual time=130.692..130.805 rows=10 loops=1) -> Nested Loop (cost=716.67..720.89 rows=1 width=8) (actual time=130.626..130.639 rows=1 loops=1) -> HashAggregate (cost=716.67..716.67 rows=1 width=4) (actual time=130.484..130.487 rows=1 loops=1) -> Subquery Scan "IN_subquery" (cost=716.66..716.67 rows=1 width=4) (actual time=130.455..130.464 rows=1 loops=1) -> Aggregate (cost=716.66..716.66 rows=1 width=4) (actual time=130.445..130.448 rows=1 loops=1) -> Seq Scan on ar (cost=0.00..659.55 rows=22844 width=4) (actual time=0.020..74.174 rows=22844 loops=1) Filter: (customer_id = 11373) -> Index Scan using ar_id_key on ar a (cost=0.00..4.20 rows=1 width=4) (actual time=0.122..0.125 rows=1 loops=1) Index Cond: (a.id = "outer".max) Filter: (customer_id = 11373) -> Index Scan using acc_trans_trans_id_key on acc_trans ac (cost=0.00..7.41 rows=8 width=12) (actual time=0.051..0.097 rows=10 loops=1) Index Cond: ("outer".max = ac.trans_id) Total runtime: 131.879 ms (17 rows) I am not sure if I want to remove support for the other two tables yet. However, I wanted to submit this here as a (possibly corner-) case where the plan seems to be far slower than it needs to be. Best Wishes, Chris Travers Metatron Technology Consulting From pgsql-performance-owner@postgresql.org Thu Jul 28 03:19:37 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2FFBC52DEA for ; Thu, 28 Jul 2005 03:19:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51825-10 for ; Thu, 28 Jul 2005 06:19:25 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id C0BA552CB6 for ; Thu, 28 Jul 2005 03:19:24 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6S6JJXp020876; Thu, 28 Jul 2005 02:19:19 -0400 (EDT) To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 Bayes module. In-reply-to: <42E80C28.2040608@aptalaska.net> References: <42E80C28.2040608@aptalaska.net> Comments: In-reply-to Matthew Schumacher message dated "Wed, 27 Jul 2005 14:35:20 -0800" Date: Thu, 28 Jul 2005 02:19:19 -0400 Message-ID: <20875.1122531559@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/430 X-Sequence-Number: 13671 Matthew Schumacher writes: > After playing with various indexes and what not I simply am unable to > make this procedure perform any better. Perhaps someone on the list can > spot the bottleneck and reveal why this procedure isn't performing that > well or ways to make it better. There's not anything obviously wrong with that procedure --- all of the updates are on primary keys, so one would expect reasonably efficient query plans to get chosen. Perhaps it'd be worth the trouble to build the server with profiling enabled and get a gprof trace to see where the time is going. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 28 03:29:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 644D152C7C for ; Thu, 28 Jul 2005 03:29:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 57685-03 for ; Thu, 28 Jul 2005 06:29:25 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id 5FD8C5295A for ; Thu, 28 Jul 2005 03:29:24 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id AE18D35F9DE for ; Thu, 28 Jul 2005 06:29:25 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 56636-05; Thu, 28 Jul 2005 06:29:16 +0000 (GMT) Received: from [10.0.0.7] (63-228-143-68.phnx.qwest.net [63.228.143.68]) by veggie-fuel.com (Postfix) with ESMTP id F348335F9D6; Thu, 28 Jul 2005 06:28:49 +0000 (GMT) Subject: Re: Performance problems testing with Spamassassin 3.1.0 From: Karim Nassar To: Matthew Schumacher Cc: pgsql-performance@postgresql.org In-Reply-To: <42E80C28.2040608@aptalaska.net> References: <42E80C28.2040608@aptalaska.net> Content-Type: text/plain Date: Wed, 27 Jul 2005 23:27:03 -0700 Message-Id: <1122532023.25574.1.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.069 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/431 X-Sequence-Number: 13672 On Wed, 2005-07-27 at 14:35 -0800, Matthew Schumacher wrote: > I put the rest of the schema up at > http://www.aptalaska.net/~matt.s/bayes/bayes_pg.sql in case someone > needs to see it too. Do you have sample data too? -- Karim Nassar Collaborative Computing Lab of NAU Office: (928) 523 5868 -=- Mobile: (928) 699 9221 http://ccl.cens.nau.edu/~kan4 From pgsql-performance-owner@postgresql.org Thu Jul 28 06:22:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3145152801 for ; Thu, 28 Jul 2005 06:22:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 96024-01 for ; Thu, 28 Jul 2005 09:22:39 +0000 (GMT) Received: from purple.bdb.fi (purple.bdb.fi [195.197.212.62]) by svr1.postgresql.org (Postfix) with ESMTP id BAD0D5288B for ; Thu, 28 Jul 2005 06:22:35 -0300 (ADT) Received: by purple.bdb.fi (Postfix, from userid 101) id D691C1C83; Thu, 28 Jul 2005 12:21:12 +0300 (EETDST) Received: from localhost (localhost [127.0.0.1]) by purple.bdb.fi (Postfix) with ESMTP id D359C4A95 for ; Thu, 28 Jul 2005 12:21:12 +0300 (EETDST) Date: Thu, 28 Jul 2005 12:21:12 +0300 (EETDST) From: Kari Lavikka To: pgsql-performance@postgresql.org Subject: Finding bottleneck Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/432 X-Sequence-Number: 13673 Hello, we recently upgraded our dual Xeon Dell to a brand new Sun v40z with 4 opterons, 16GB of memory and MegaRAID with enough disks. OS is Debian Sarge amd64, PostgreSQL is 8.0.3. Size of database is something like 80GB and our website performs about 600 selects and several updates/inserts a second. v40z performs somewhat better than our old Dell but mostly due to increased amount of memory. The problem is.. there seems to by plenty of free CPU available and almost no IO-wait but CPU bound queries seem to linger for some reason. Problem appears very clearly during checkpointing. Queries accumulate and when checkpointing is over, there can be something like 400 queries running but over 50% of cpu is just idling. procs -----------memory------------ ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 3 1 0 494008 159492 14107180 0 0 919 3164 3176 13031 29 12 52 8 5 3 0 477508 159508 14118452 0 0 1071 4479 3474 13298 27 13 47 13 0 0 0 463604 159532 14128832 0 0 922 2903 3352 12627 29 11 52 8 3 1 0 442260 159616 14141668 0 0 1208 3153 3357 13163 28 12 52 9 An example of a lingering query (there's usually several of these or similar): SELECT u.uid, u.nick, u.name, u.showname, i.status, i.stamp, i.image_id, i.info, i.t_width, i.t_height FROM users u INNER JOIN image i ON i.uid = u.uid INNER JOIN user_online uo ON u.uid = uo.uid WHERE u.city_id = 5 AND i.status = 'd' AND u.status = 'a' ORDER BY city_id, upper(u.nick) LIMIT (40 + 1) OFFSET 320 Tables involved contain no more than 4 million rows. Those are constantly accessed and should fit nicely to cache. But database is just slow because of some unknown reason. Any ideas? ----------------->8 Relevant rows from postgresql.conf 8<----------------- shared_buffers = 15000 # min 16, at least max_connections*2, 8KB each work_mem = 1536 # min 64, size in KB maintenance_work_mem = 32768 # min 1024, size in KB max_fsm_pages = 1000000 # min max_fsm_relations*16, 6 bytes each max_fsm_relations = 5000 # min 100, ~50 bytes each vacuum_cost_delay = 15 # 0-1000 milliseconds vacuum_cost_limit = 120 # 0-10000 credits bgwriter_percent = 2 # 0-100% of dirty buffers in each round fsync = true # turns forced synchronization on or off # fsync, fdatasync, open_sync, or open_datasync wal_buffers = 128 # min 4, 8KB each commit_delay = 80000 # range 0-100000, in microseconds commit_siblings = 10 # range 1-1000 checkpoint_segments = 200 # in logfile segments, min 1, 16MB each checkpoint_timeout = 1800 # range 30-3600, in seconds effective_cache_size = 1000000 # typically 8KB each random_page_cost = 1.8 # units are one sequential page fetch cost default_statistics_target = 150 # range 1-1000 stats_start_collector = true stats_command_string = true |\__/| ( oo ) Kari Lavikka - tuner@bdb.fi - (050) 380 3808 __ooO( )Ooo_______ _____ ___ _ _ _ _ _ _ _ "" From pgsql-performance-owner@postgresql.org Thu Jul 28 06:35:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A856F529E2 for ; Thu, 28 Jul 2005 06:35:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 98894-02 for ; Thu, 28 Jul 2005 09:34:55 +0000 (GMT) Received: from linuxworld.com.au (unknown [203.34.46.50]) by svr1.postgresql.org (Postfix) with ESMTP id 67036529C8 for ; Thu, 28 Jul 2005 06:34:53 -0300 (ADT) Received: from linuxworld.com.au (IDENT:swm@localhost.localdomain [127.0.0.1]) by linuxworld.com.au (8.13.2/8.13.2) with ESMTP id j6S9Yoti002779; Thu, 28 Jul 2005 19:34:50 +1000 Received: from localhost (swm@localhost) by linuxworld.com.au (8.13.2/8.13.2/Submit) with ESMTP id j6S9Yorh002776; Thu, 28 Jul 2005 19:34:50 +1000 X-Authentication-Warning: linuxworld.com.au: swm owned process doing -bs Date: Thu, 28 Jul 2005 19:34:50 +1000 (EST) From: Gavin Sherry X-X-Sender: swm@linuxworld.com.au To: Kari Lavikka Cc: pgsql-performance@postgresql.org Subject: Re: Finding bottleneck In-Reply-To: Message-ID: References: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/433 X-Sequence-Number: 13674 Hi, On Thu, 28 Jul 2005, Kari Lavikka wrote: > ----------------->8 Relevant rows from postgresql.conf 8<----------------- > > shared_buffers = 15000 # min 16, at least max_connections*2, 8KB each > work_mem = 1536 # min 64, size in KB As an aside, I'd increase work_mem -- but it doesn't sound like that is your problem. > maintenance_work_mem = 32768 # min 1024, size in KB > > max_fsm_pages = 1000000 # min max_fsm_relations*16, 6 bytes each > max_fsm_relations = 5000 # min 100, ~50 bytes each > > vacuum_cost_delay = 15 # 0-1000 milliseconds > vacuum_cost_limit = 120 # 0-10000 credits > > bgwriter_percent = 2 # 0-100% of dirty buffers in each round > > fsync = true # turns forced synchronization on or off > # fsync, fdatasync, open_sync, or open_datasync > wal_buffers = 128 # min 4, 8KB each Some benchmarking results out today suggest that wal_buffers = 1024 or even 2048 could greatly assist you. > commit_delay = 80000 # range 0-100000, in microseconds > commit_siblings = 10 # range 1-1000 This may explain the fact that you've got backed up queries and idle CPU -- I'm not certain though. What does disabling commit_delay do to your situation? Gavin From pgsql-performance-owner@postgresql.org Thu Jul 28 06:56:09 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 412BA529E0 for ; Thu, 28 Jul 2005 06:56:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 02860-03 for ; Thu, 28 Jul 2005 09:55:56 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.199]) by svr1.postgresql.org (Postfix) with ESMTP id 9A57B52889 for ; Thu, 28 Jul 2005 06:55:54 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id z6so202716nzd for ; Thu, 28 Jul 2005 02:55:57 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:references; b=dkjgoMwUIRku+QtJaAV2+p+gSttSePVx6tx8DBH/SuKSibIANggYvZuulnkq6LyBs/qP399dTPHVtItQfrxKUO4yANADQVz7AUdLQFUacbYksUyRUpSJomQA4Gf2mB7Jq7JLCapnnj/c8oPuV4ZmwcZi+FWa1lcDDDm5T7oRv0k= Received: by 10.37.22.50 with SMTP id z50mr912833nzi; Thu, 28 Jul 2005 02:55:57 -0700 (PDT) Received: by 10.36.119.19 with HTTP; Thu, 28 Jul 2005 02:55:57 -0700 (PDT) Message-ID: Date: Thu, 28 Jul 2005 15:25:57 +0530 From: Gnanavel S Reply-To: Gnanavel S To: Chris Travers Subject: Re: Left joining against two empty tables makes a query SLOW Cc: pgsql-performance@postgresql.org In-Reply-To: <42E86A03.1070803@travelamericas.com> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_2293_14846395.1122544557125" References: <42E86A03.1070803@travelamericas.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.488 tagged_above=0 required=5 tests=AWL, HTML_50_60, HTML_MESSAGE, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/434 X-Sequence-Number: 13675 ------=_Part_2293_14846395.1122544557125 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On 7/28/05, Chris Travers wrote: >=20 > Hi all; >=20 > I have a customer who currently uses an application which had become > slow. After doing some digging, I found the slow query: >=20 > SELECT c.accno, c.description, c.link, c.category, ac.project_id, > p.projectnumber, > a.department_id, d.description AS department > FROM chart c JOIN acc_trans ac ON (ac.chart_id =3D c.id ) > JOIN ar a ON (a.id =3D ac.trans_id) > LEFT JOIN project p ON (ac.project_id =3D p.id ) > LEFT JOIN department d ON (d.id =3D a.department_id) > WHERE a.customer_id =3D 11373 AND a.id IN ( > SELECT max(id) FROM ar WHERE customer_id =3D 11373 > ); >=20 > (reformatted for readability) >=20 > This is taking 10 seconds to run. >=20 > Interestingly, both the project and department tables are blank, and if > I omit them, the query becomes: > SELECT c.accno, c.description, c.link, c.category, ac.project_id > FROM chart c JOIN acc_trans ac ON (ac.chart_id =3D c.id ) > JOIN ar a ON (a.id =3D ac.trans_id) > WHERE a.customer_id =3D 11373 AND a.id IN ( > SELECT max(id) FROM ar WHERE customer_id =3D 11373 > ); >=20 > This takes 139ms. 1% of the previous query. >=20 > The plan for the long query is: >=20 >=20 > QUERY PLAN >=20 > -------------------------------------------------------------------------= ---------------------------------------------------------------------------= ------------ > Hash IN Join (cost=3D87337.25..106344.93 rows=3D41 width=3D118) (actual > time=3D7615.843..9850.209 rows=3D10 loops=3D1) > Hash Cond: ("outer".trans_id =3D "inner".max) > -> Merge Right Join (cost=3D86620.57..100889.85 rows=3D947598 > width=3D126) (actual time=3D7408.830..9200.435 rows=3D177769 loops=3D1) > Merge Cond: ("outer".id =3D "inner".department_id) > -> Index Scan using department_id_key on department d > (cost=3D0.00..52.66 > rows=3D1060 width=3D36) (actual time=3D0.090..0.090 rows=3D0 loops=3D1) vacuum & reindex the department and project table as the planner expects=20 there are 1060 rows but actually returning nothing. -> Sort (cost=3D86620.57..87067.55 rows=3D178792 width=3D94) > (actual time=3D7408.709..7925.843 rows=3D177769 loops=3D1) > Sort Key: a.department_id > -> Merge Right Join (cost=3D45871.18..46952.83 > rows=3D178792 width=3D94) (actual time=3D4962.122..6671.319 rows=3D177769= loops=3D1) > Merge Cond: ("outer".id =3D "inner".project_id) > -> Index Scan using project_id_key on project p > (cost=3D0.00..49.80 rows=3D800 width=3D36) (actual time=3D0.007..0.007 ro= ws=3D0 > loops=3D1) > -> Sort (cost=3D45871.18..46318.16 rows=3D178792 > width=3D62) (actual time=3D4962.084..5475.636 rows=3D177769 loops=3D1) > Sort Key: ac.project_id > -> Hash Join (cost=3D821.20..13193.43 > rows=3D178792 width=3D62) (actual time=3D174.905..4295.685 rows=3D177769 = loops=3D1) > Hash Cond: ("outer".chart_id =3D "inner".id) > -> Hash Join (cost=3D817.66..10508.02 > rows=3D178791 > width=3D20) (actual time=3D173.952..2840.824 rows=3D177769 loops=3D1) > Hash Cond: ("outer".trans_id =3D > "inner".id) > -> Seq Scan on acc_trans ac > (cost=3D0.00..3304.38 rows=3D181538 width=3D12) (actual time=3D0.062..537= .753 > rows=3D181322 loops=3D1) > -> Hash (cost=3D659.55..659.55 > rows=3D22844 width=3D8) (actual time=3D173.625..173.625 rows=3D0 loops=3D= 1) > -> Seq Scan on ar a > (cost=3D0.00..659.55 rows=3D22844 width=3D8) (actual time=3D0.022..101.82= 8 > rows=3D22844 loops=3D1) > Filter: (customer_id > =3D 11373) > -> Hash (cost=3D3.23..3.23 rows=3D123 > width=3D50) (actual time=3D0.915..0.915 rows=3D0 loops=3D1) > -> Seq Scan on chart c > (cost=3D0.00..3.23 rows=3D123 width=3D50) (actual time=3D0.013..0.528 row= s=3D123 > loops=3D1) > -> Hash (cost=3D716.67..716.67 rows=3D1 width=3D4) (actual > time=3D129.037..129.037 rows=3D0 loops=3D1) > -> Subquery Scan "IN_subquery" (cost=3D716.66..716.67 rows=3D1 > width=3D4) (actual time=3D129.017..129.025 rows=3D1 loops=3D1) > -> Aggregate (cost=3D716.66..716.66 rows=3D1 width=3D4) > (actual time=3D129.008..129.011 rows=3D1 loops=3D1) > -> Seq Scan on ar (cost=3D0.00..659.55 rows=3D22844 > width=3D4) (actual time=3D0.020..73.266 rows=3D22844 loops=3D1) > Filter: (customer_id =3D 11373) > Total runtime: 9954.133 ms > (28 rows) >=20 > The shorter query's plan is: >=20 >=20 > QUERY PLAN >=20 >=20 > -------------------------------------------------------------------------= ---------------------------------------------------------------------------= ----- > Hash Join (cost=3D728.42..732.96 rows=3D8 width=3D50) (actual > time=3D130.908..131.593 rows=3D10 loops=3D1) > Hash Cond: ("outer".id =3D "inner".chart_id) > -> Seq Scan on chart c (cost=3D0.00..3.23 rows=3D123 width=3D50) (actual > time=3D0.006..0.361 rows=3D123 loops=3D1) > -> Hash (cost=3D728.40..728.40 rows=3D8 width=3D8) (actual > time=3D130.841..130.841 rows=3D0 loops=3D1) > -> Nested Loop (cost=3D716.67..728.40 rows=3D8 width=3D8) (actual > time=3D130.692..130.805 rows=3D10 loops=3D1) > -> Nested Loop (cost=3D716.67..720.89 rows=3D1 width=3D8) > (actual time=3D130.626..130.639 rows=3D1 loops=3D1) > -> HashAggregate (cost=3D716.67..716.67 rows=3D1 > width=3D4) (actual time=3D130.484..130.487 rows=3D1 loops=3D1) > -> Subquery Scan "IN_subquery" > (cost=3D716.66..716.67 rows=3D1 width=3D4) (actual time=3D130.455..130.46= 4 > rows=3D1 loops=3D1) > -> Aggregate (cost=3D716.66..716.66 > rows=3D1 width=3D4) (actual time=3D130.445..130.448 rows=3D1 loops=3D1) > -> Seq Scan on ar > (cost=3D0.00..659.55 rows=3D22844 width=3D4) (actual time=3D0.020..74.174 > rows=3D22844 loops=3D1) > Filter: (customer_id =3D 11373) > -> Index Scan using ar_id_key on ar a > (cost=3D0.00..4.20 rows=3D1 width=3D4) (actual time=3D0.122..0.125 rows= =3D1 loops=3D1) > Index Cond: (a.id =3D "outer".max) > Filter: (customer_id =3D 11373) > -> Index Scan using acc_trans_trans_id_key on acc_trans > ac (cost=3D0.00..7.41 rows=3D8 width=3D12) (actual time=3D0.051..0.097 ro= ws=3D10 > loops=3D1) > Index Cond: ("outer".max =3D ac.trans_id) > Total runtime: 131.879 ms > (17 rows) >=20 > I am not sure if I want to remove support for the other two tables > yet. However, I wanted to submit this here as a (possibly corner-) > case where the plan seems to be far slower than it needs to be. >=20 > Best Wishes, > Chris Travers > Metatron Technology Consulting >=20 > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend >=20 --=20 with regards, S.Gnanavel Satyam Computer Services Ltd. ------=_Part_2293_14846395.1122544557125 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline

On 7/28/05, Chris Travers <c= hris@travelamericas.com> wrote:
Hi all;

I have a customer who currently uses an application which ha= d become
slow.  After doing some digging, I found the slow que= ry:

SELECT c.accno, c.description, c.link, c.category, ac.project_id= ,
p.projectnumber ,
        a.department_id, d.des= cription AS department
FROM chart c JOIN acc_trans ac ON (ac.chart_id = =3D c.id)
     &nbs= p;  JOIN ar a ON (a.id =3D ac.trans_i= d)
        LEFT JOIN project p O= N ( ac.project_id =3D p.id)
   &n= bsp;    LEFT JOIN department d ON (d.id =3D a.department_id)
WHERE a.customer_id =3D 11373 AND a.id IN (
      &n= bsp; SELECT max(id) FROM ar WHERE customer_id =3D 11373
);

(reformatted for readability)

This is taking 10 second= s to run.

Interestingly, both the project and department tables are = blank, and if
I omit them, the query becomes:
SELECT c.accno, c.descr= iption , c.link, c.category, ac.project_id
FROM chart c JOIN acc_trans ac ON (a= c.chart_id =3D c.id)
    = ;    JOIN ar a ON (a.id =3D= ac.trans_id)
WHERE a.customer_id =3D 11373 AND=20 a.id IN (
     &nbs= p;  SELECT max(id) FROM ar WHERE customer_id =3D 11373
);
<= br>This takes 139ms.  1% of the previous query.

The plan f= or the long query is:


QUERY PLAN
----------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------
Hash IN Join  (cost=3D87337.25..106344.93 rows=3D41 width=3D= 118) (actual
time=3D7615.843..9850.209 rows=3D10 loops=3D1)
 &nb= sp; Hash Cond: ("outer".trans_id =3D "inner".max)
&n= bsp;  ->  Merge Right Join  (cost=3D 86620.57..100889.85 rows=3D947598
width=3D126) (actual time=3D7408.830..= 9200.435 rows=3D177769 loops=3D1)
      &n= bsp;  Merge Cond: ("outer".id =3D "inner".departme= nt_id)
         ->  = ;Index Scan using department_id_key on department d
(cost=3D0.00..52.66
rows=3D1060 width=3D36) (actual time=3D0.090..0.= 090 rows=3D0 loops=3D1)

vacuum & reindex the department and project table as the planner expect= s there are 1060 rows but actually returning nothing.

 &= nbsp;       ->  Sort  = (cost=3D86620.57..87067.55 rows=3D178792 width=3D94)
(actual time=3D 7408.709..7925.843 rows=3D177769 loops=3D1)
    &nbs= p;          Sort Key: a.depart= ment_id
          &nbs= p;    ->  Merge Right Join  (cost=3D45871.18..46952.83
= rows=3D178792 width=3D94) (actual time=3D4962.122..6671.319 rows=3D177769 l= oops=3D1)
          &n= bsp;          Merge Cond: ("outer".id =3D "inner".project_id)
&nbs= p;            &= nbsp;       ->  Index Scan using project_id_key on project p
(cost=3D0.= 00..49.80 rows=3D800 width=3D36) (actual time=3D0.007..0.007 rows=3D0
lo= ops=3D1)
          &nb= sp;          ->  Sort  (cost=3D45871.18..46318.16 rows=3D178792width=3D62) (actual time=3D4962.084..5475.636 rows=3D177769 loops=3D1)            &n= bsp;            = ;  Sort Key: ac.project_id
        =             &nb= sp;      ->  Hash Join  (cost=3D821.20..13193.43
rows=3D17= 8792 width=3D62) (actual time=3D174.905..4295.685 rows=3D177769 loops=3D1)<= br>            =             &nb= sp;        Hash Cond: ("outer".chart_id =3D "inner".id)
 &= nbsp;           &nbs= p;            &= nbsp;      ->  Hash Join  (cost=3D817.66..10508.02
rows=3D17= 8791
width=3D20) (actual time=3D173.952..2840.824 rows=3D177769 loops=3D= 1)
           &nb= sp;            =             &nb= sp;  Hash Cond: ("outer".trans_id =3D
"inner".id)
&nbs= p;            &= nbsp;           &nbs= p;             ->  Seq Scan on acc_trans ac
(cost=3D0.00..3304.38 rows=3D1= 81538 width=3D12) (actual time=3D0.062..537.753
rows=3D181322 loops=3D1)=
            = ;            &n= bsp;            = ;  ->  Hash  (cost=3D659.55..659.55
rows=3D22844 wid= th=3D8) (actual time=3D173.625..173.625 rows=3D0 loops=3D1)
  =             &nb= sp;            =             &nb= sp;     ->  Seq Scan on ar a
(cost=3D0.00..659.55 rows=3D22844 widt= h=3D8) (actual time=3D0.022..101.828
rows=3D22844 loops=3D1)
 &n= bsp;            = ;            &n= bsp;            = ;            Filter: (customer_id
=3D 11373)
      &= nbsp;           &nbs= p;            &= nbsp; ->  Hash  (cost=3D3.23..3.23 rows=3D123
width=3D5= 0) (actual time=3D0.915..0.915 rows=3D0 loops=3D1)
   &nb= sp;            =             &nb= sp;          ->  Seq Scan on chart c
(cost=3D0.00..3.23 rows=3D123 width= =3D50) (actual time=3D0.013..0.528 rows=3D123
loops=3D1)
  = ->  Hash  (cost=3D716.67..716.67 rows=3D1 width=3D4= ) (actual
time=3D129.037..129.037 rows=3D0 loops=3D1)
         ->  Subquery Scan "IN_subquery"  (cost=3D716.66..716.67 rows=3D1
width= =3D4) (actual time=3D129.017..129.025 rows=3D1 loops=3D1)
  &n= bsp;            ->  Aggregate  (cost=3D716.66..716.66 rows=3D1 width=3D4)
(actual time=3D129.008..129.011 rows=3D1 loops=3D1)
 =             &nb= sp;       ->  Seq Scan on ar  (cost=3D0.00..659.55 rows=3D2284= 4
width=3D4) (actual time=3D0.020..73.266 rows=3D22844 loops=3D1)
&nb= sp;            =             &nb= sp; Filter: (customer_id =3D 11373)
Total runtime: 9954.133 ms
(28 rows)=

The shorter query's plan is:


QUERY PLAN

---------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------
Hash Join  (cost=3D728.42..732.96 rows=3D8 width=3D50) (actu= al
time=3D130.908..131.593 rows=3D10 loops=3D1)
   Hash Con= d: ("outer".id =3D "inner".chart_id)
   -&= gt;  Seq Scan on chart c  (cost=3D0.00..3.23 rows=3D123 width=3D50) (actual
time=3D0.006..0.361 rows=3D123 loops=3D1= )
   ->  Hash  (cost=3D728.40..728.40 r= ows=3D8 width=3D8) (actual
time=3D130.841..130.841 rows=3D0 loops=3D1)         ->  Nested Loop  (cost=3D716.67..728.40 rows=3D8 width=3D8) (actual
time=3D130.692..130.805 rows=3D10 loops=3D1)
 = ;            &n= bsp; ->  Nested Loop  (cost=3D716.67..720.89 rows=3D1 width=3D8)
(actual time=3D130.626..130.639 rows=3D1 loops=3D1)
 =             &nb= sp;       ->  HashAggregate  (cost=3D716.67..716.67 rows=3D1width=3D4) (actual time=3D130.484..130.487 rows=3D1 loops=3D1)
 &= nbsp;           &nbs= p;             ->  Subquery Scan "IN_subquery"
(cost=3D716.66..7= 16.67 rows=3D1 width=3D4) (actual time=3D130.455..130.464
rows=3D1 loops= =3D1)
           =             &nb= sp;         ->  Aggregate  (cost=3D716.66..716.66
rows=3D1 wi= dth=3D4) (actual time=3D130.445..130.448 rows=3D1 loops=3D1)
  = ;            &n= bsp;            = ;            ->  Seq Scan on ar
(cost=3D0.00..659.55 rows=3D22844 width= =3D4) (actual time=3D0.020..74.174
rows=3D22844 loops=3D1)
 &nbs= p;            &= nbsp;           &nbs= p;            &= nbsp;     Filter: (customer_id =3D 11373)
      &nbs= p;            &= nbsp; ->  Index Scan using ar_id_key on ar a
(cost=3D0.00..4.20 r= ows=3D1 width=3D4) (actual time=3D0.122..0.125 rows=3D1 loops=3D1)
 = ;            &n= bsp;            = ; Index Cond: (a.id =3D "outer".max)            &n= bsp;            = ;  Filter: (customer_id =3D 11373)
      &nbs= p;        ->  Index Scan using acc_trans_trans_id_key on acc_trans
ac=   (cost=3D0.00..7.41 rows=3D8 width=3D12) (actual time=3D0.051..0= .097 rows=3D10
loops=3D1)
       &= nbsp;           &nbs= p; Index Cond: ("outer".max =3D ac.trans_id)
Total runtime: 131.= 879 ms
(17 rows)

I am not sure if I want to remove support for th= e other two tables
yet.   However, I wanted to submit this her= e as a (possibly corner-)
case where the plan seems to be far slower than it needs to be.

= Best Wishes,
Chris Travers
Metatron Technology Consulting

----= -----------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend



--
with regards,
S.Gnanavel
Satyam Computer Services= Ltd. ------=_Part_2293_14846395.1122544557125-- From pgsql-hackers-owner@postgresql.org Thu Jul 28 08:49:57 2005 X-Original-To: pgsql-hackers-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7524B529F9 for ; Thu, 28 Jul 2005 08:49:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 27592-07 for ; Thu, 28 Jul 2005 11:49:45 +0000 (GMT) Received: from stark.xeocode.com (stark.xeocode.com [216.58.44.227]) by svr1.postgresql.org (Postfix) with ESMTP id DA193529EA for ; Thu, 28 Jul 2005 08:49:44 -0300 (ADT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1Dy6tF-0007uU-00; Thu, 28 Jul 2005 07:49:29 -0400 To: josh@agliodbs.com Cc: pgsql-hackers@postgresql.org Subject: Re: wal_buffer tests in References: <200507271330.02083.josh@agliodbs.com> In-Reply-To: <200507271330.02083.josh@agliodbs.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 28 Jul 2005 07:49:29 -0400 Message-ID: <87oe8n881i.fsf@stark.xeocode.com> Lines: 16 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.003 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/1045 X-Sequence-Number: 71097 Josh Berkus writes: > Folks, > > I ran a wal_buffer test series. It appears that increasing the > wal_buffers is indeed very important for OLTP applications, potentially > resulting in as much as a 15% average increase in transaction processing. > What's interesting is that this is not just true for 8.1, it's true for > 8.0.3 as well. You have wal_buffer set to 2048? That's pretty radical compared to the default of just 5. Your tests shows you had to go to this large a value to see the maximum effect? -- greg From pgsql-performance-owner@postgresql.org Thu Jul 28 08:52:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A726752A24 for ; Thu, 28 Jul 2005 08:52:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 25784-09 for ; Thu, 28 Jul 2005 11:52:01 +0000 (GMT) Received: from wproxy.gmail.com (wproxy.gmail.com [64.233.184.192]) by svr1.postgresql.org (Postfix) with ESMTP id 2CFEA52A21 for ; Thu, 28 Jul 2005 08:52:00 -0300 (ADT) Received: by wproxy.gmail.com with SMTP id 70so405905wra for ; Thu, 28 Jul 2005 04:52:04 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=tKrcFYMs/fZ5rJKx9Who9FkYMnDk4ApO/1Yy4EL/CUykFHKPSiiCtnt9fjvbzTPck9Dzhu2JW8LIAYisLneHl9jHxLB2ww/Uwnu/Sk7ipJKiIrkJVOs62/mIO7vKJKgC3YxrQ/65auPTSVU+O4M3CHIL0bcT52gR2YTFgoukmSg= Received: by 10.54.120.4 with SMTP id s4mr767597wrc; Thu, 28 Jul 2005 04:52:04 -0700 (PDT) Received: by 10.54.144.11 with HTTP; Thu, 28 Jul 2005 04:52:03 -0700 (PDT) Message-ID: Date: Thu, 28 Jul 2005 13:52:03 +0200 From: Claus Guttesen Reply-To: Claus Guttesen To: Kari Lavikka Subject: Re: Finding bottleneck Cc: pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/435 X-Sequence-Number: 13676 > effective_cache_size =3D 1000000 # typically 8KB each I have this setting on postgresql 7.4.8 on FreeBSD with 4 GB RAM: effective_cache_size =3D 27462 So eventhough your machine runs Debian and you have four times as much RAM as mine your effective_cache_size is 36 times larger. You could try lowering this setting. regards Claus From pgsql-performance-owner@postgresql.org Thu Jul 28 09:20:41 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CA65D529A4 for ; Thu, 28 Jul 2005 09:20:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 37175-06 for ; Thu, 28 Jul 2005 12:20:32 +0000 (GMT) Received: from Herge.rcsinc.local (mail.rcsonline.com [205.217.85.91]) by svr1.postgresql.org (Postfix) with ESMTP id 4FE325298B for ; Thu, 28 Jul 2005 09:20:31 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Subject: Re: Performance problems testing with Spamassassin 3.1.0 Bayes module. Date: Thu, 28 Jul 2005 08:20:35 -0400 Message-ID: <6EE64EF3AB31D5448D0007DD34EEB3417DCF21@Herge.rcsinc.local> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Performance problems testing with Spamassassin 3.1.0 Bayes module. Thread-Index: AcWS/JSuYQG7b/vqQHCbJnr1jWca5gAcaJCQ From: "Merlin Moncure" To: "Matthew Schumacher" Cc: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/436 X-Sequence-Number: 13677 > I'm not sure how much this has been discussed on the list, but wasn't > able to find anything relevant in the archives. >=20 > The new Spamassassin is due out pretty soon. They are currently testing > 3.1.0pre4. One of the things I hope to get out of this release is bayes > word stats moved to a real RDBMS. They have separated the mysql > BayesStore module from the PgSQL one so now postgres can use it's own > queries. >=20 > I loaded all of this stuff up on a test server and am finding that the > bayes put performance is really not good enough for any real amount of > mail load. >=20 > The performance problems seems to be when the bayes module is > inserting/updating. This is now handled by the token_put procedure. 1. you need high performance client side timing (sub 1 millisecond). on win32 use QueryPerformanceCounter 2. one by one, convert queries inside your routine into dynamic versions. That is, use execute 'query string' 3. Identify the problem. Something somewhere is not using the index. Because of the way the planner works you have to do this sometimes. Merlin From pgsql-performance-owner@postgresql.org Thu Jul 28 10:02:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 28CDE52932 for ; Thu, 28 Jul 2005 10:02:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 44180-05 for ; Thu, 28 Jul 2005 13:02:18 +0000 (GMT) Received: from Herge.rcsinc.local (mail.rcsonline.com [205.217.85.91]) by svr1.postgresql.org (Postfix) with ESMTP id 339D65291A for ; Thu, 28 Jul 2005 10:02:15 -0300 (ADT) Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Subject: Re: Finding bottleneck Date: Thu, 28 Jul 2005 09:02:18 -0400 Message-ID: <6EE64EF3AB31D5448D0007DD34EEB3417DCF28@Herge.rcsinc.local> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Finding bottleneck Thread-Index: AcWTV1FLZLMsPgagRTmZQRBQ3VnXWAAHFC9A From: "Merlin Moncure" To: "Kari Lavikka" Cc: X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.056 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/437 X-Sequence-Number: 13678 Kari Lavikka wrote: > shared_buffers =3D 15000 =20 you can play around with this one but in my experience it doesn't make much difference anymore (it used to). > work_mem =3D 1536 # min 64, size in KB this seems low. are you sure you are not getting sorts swapped to disk? > fsync =3D true # turns forced synchronization on or off does turning this to off make a difference? This would help narrow down where the problem is. > commit_delay =3D 80000 # range 0-100000, in microseconds hm! how did you arrive at this number? try setting to zero and comparing. > stats_start_collector =3D true > stats_command_string =3D true with a high query load you may want to consider turning this off. On win32, I've had some problem with stat's collector under high load conditions. Not un unix, but it's something to look at. Just turn off stats for a while and see if it helps. good luck! your hardware should be more than adequate. Merlin From pgsql-performance-owner@postgresql.org Thu Jul 28 11:38:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 31281529C3 for ; Thu, 28 Jul 2005 11:38:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68310-08 for ; Thu, 28 Jul 2005 14:38:28 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.196]) by svr1.postgresql.org (Postfix) with ESMTP id 1B4F6529A2 for ; Thu, 28 Jul 2005 11:38:25 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 8so325879nzo for ; Thu, 28 Jul 2005 07:38:28 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=Mw+vxWEIT6/opwXb2A2+XZJ7+eKPGM+cT3pnmdh2HxIn+ib1e6XA6HN7v3+x7HD4OBkRtU4GCPXapo6g8pHdTIcpeMXiFxsrbXdWhDkCvMcysvQzI4pJp9skHk8iZOmWL+q6+89PGEQshooTsEYyW8aKPHlz4IioTOIBr0Z+Y08= Received: by 10.36.71.12 with SMTP id t12mr1976998nza; Thu, 28 Jul 2005 07:38:07 -0700 (PDT) Received: by 10.36.25.13 with HTTP; Thu, 28 Jul 2005 07:38:06 -0700 (PDT) Message-ID: <1d219a6f0507280738791f0ea3@mail.gmail.com> Date: Thu, 28 Jul 2005 10:38:06 -0400 From: Chris Hoover Reply-To: Chris Hoover To: pgsql-performance@postgresql.org Subject: Fwd: Help with view performance problem In-Reply-To: <1d219a6f050727092927746db1@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1d219a6f05072708122cf6abeb@mail.gmail.com> <1d219a6f050727092927746db1@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/438 X-Sequence-Number: 13679 Does anyone have any suggestions on this? I did not get any response from the admin list. Thanks, Chris ---------- Forwarded message ---------- From: Chris Hoover Date: Jul 27, 2005 12:29 PM Subject: Re: Help with view performance problem To: pgsql-admin@postgresql.org I did some more testing, and ran the explain analyze on the problem. In my session I did a set enable_hashjoin =3D false and then ran the analyze. This caused it to use the indexes as I have been expecting it to do. Now, how can I get it to use the indexes w/o manipulating the environment? What make postgresql want to sequentially scan and use a hash join? thanks, Chris explain analyze with set_hashjoin=3Dfalse; prob_db=3D#explain analyze select * from clm_com; QUERY PLAN ---------------------------------------------------------------------------= ------------------------------------------------------------------- ---------------------------------------------------------------------------= ------------------------------------------------------------------- ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ------------------------------------------ Subquery Scan clm_com (cost=3D1057975.45..1169021.26 rows=3D126910 width=3D366) (actual time=3D142307.99..225997.22 rows=3D1268649 loops=3D1) -> Unique (cost=3D1057975.45..1169021.26 rows=3D126910 width=3D366) (actual time=3D142307.96..206082.30 rows=3D1268649 loops=3D1) -> Sort (cost=3D1057975.45..1061148.19 rows=3D1269095 width=3D366) (actual time=3D142307.95..156019.01 rows=3D1268649 loops=3D1) Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, clmcom1.pat_zip, clmcom1.payto_addr_1, clmcom1.payto_addr_2, clmcom1.payto_city, clmcom1.payto_cntry, clmcom1.payto_f_name, clmcom1.payto_m_name, clmcom1.payto_state, clmcom1.payto_zip, clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip -> Merge Join (cost=3D0.00..565541.46 rows=3D1269095 width=3D366) (actual time=3D464.89..130638.06 rows=3D1268649 loops=3D1) Merge Cond: ("outer".inv_nbr =3D "inner".inv_nbr) Join Filter: ("outer".inv_qfr =3D "inner".inv_qfr) -> Index Scan using clmcom1_inv_nbr_iview_idx on clmcom1 (cost=3D0.00..380534.32 rows=3D1269095 width=3D270) (actual time=3D0.27..82159.37 rows=3D1268649 loops=3D1) -> Index Scan using clmcom2_inv_nbr_iview_idx on clmcom2 (cost=3D0.00..159636.25 rows=3D1271198 width=3D96) (actual time=3D464.56..21774.02 rows=3D1494019 loops=3D1) Total runtime: 227369.39 msec (10 rows) On 7/27/05, Chris Hoover wrote: > I am having a problem with a view on one of my db's. This view is > trying to sequentially can the 2 tables it is accessing. However, > when I explain the view on most of my other db's (all have the same > schema's), it is using the indexes. Can anyone please help me > understand why postgres is choosing to sequenially scan both tables? > > Both tables in the view have a primary key defined on inv_nbr, > inv_qfr. Vacuum and analyze have been run on the tables in question > to try and make sure stats are up to date. > > Thanks, > > Chris > PG - 7.3.4 > RH 2.1 > > > Here is the view definition: > SELECT DISTINCT clmcom1.inv_nbr AS inventory_number, > clmcom1.inv_qfr AS inventory_qualifier, > clmcom1.pat_addr_1 AS patient_address_1, > clmcom1.pat_addr_2 AS patient_address_2, > clmcom1.pat_city AS patient_city, > clmcom1.pat_cntry AS patient_country, > clmcom1.pat_dob AS patient_date_of_birth, > clmcom1.pat_gender_cd AS patient_gender_cod= e, > clmcom1.pat_info_pregnancy_ind AS pregnancy= _ind, > clmcom1.pat_state AS patient_state, > clmcom1.pat_suffix AS patient_suffix, > clmcom1.pat_zip AS patient_zip_code, > clmcom1.payto_addr_1 AS payto_address_1, > clmcom1.payto_addr_2 AS payto_address_2, > clmcom1.payto_city, > clmcom1.payto_cntry AS payto_country, > clmcom1.payto_f_name AS payto_first_name, > clmcom1.payto_m_name AS payto_middle_name, > clmcom1.payto_state, > clmcom1.payto_zip AS payto_zip_code, > clmcom1.clm_tot_clm_chgs AS total_claim_cha= rge, > clmcom1.bill_l_name_org AS > billing_last_name_or_org, > clmcom1.clm_delay_rsn_cd AS > claim_delay_reason_code, > clmcom1.clm_submit_rsn_cd AS > claim_submit_reason_code, > clmcom1.payto_l_name_org AS > payto_last_name_or_org, > clmcom1.payto_prim_id AS payto_primary_id, > clmcom1.bill_prim_id AS billing_prov_primar= y_id, > clmcom1.clm_tot_ncov_chgs AS total_ncov_cha= rge, > clmcom2.contract_amt AS contract_amount, > clmcom2.svc_fac_or_lab_name, > clmcom2.svc_fac_addr_1 AS svc_fac_address_1= , > clmcom2.svc_fac_addr_2 AS svc_fac_address_2= , > clmcom2.svc_fac_city, > clmcom2.svc_fac_zip AS svc_fac_zip_code > FROM (clmcom1 LEFT JOIN clmcom2 ON (((clmcom1.inv_nbr =3D > clmcom2.inv_nbr) AND > > (clmcom1.inv_qfr =3D clmcom2.inv_qfr)))) > ORDER BY clmcom1.inv_nbr, > clmcom1.inv_qfr, > clmcom1.pat_addr_1, > clmcom1.pat_addr_2, > clmcom1.pat_city, > clmcom1.pat_cntry, > clmcom1.pat_dob, > clmcom1.pat_gender_cd, > clmcom1.pat_info_pregnancy_ind, > clmcom1.pat_state, > clmcom1.pat_suffix, > clmcom1.pat_zip, > clmcom1.payto_addr_1, > clmcom1.payto_addr_2, > clmcom1.payto_city, > clmcom1.payto_cntry, > clmcom1.payto_f_name, > clmcom1.payto_m_name, > clmcom1.payto_state, > clmcom1.payto_zip, > clmcom1.clm_tot_clm_chgs, > clmcom1.bill_l_name_org, > clmcom1.clm_delay_rsn_cd, > clmcom1.clm_submit_rsn_cd, > clmcom1.payto_l_name_org, > clmcom1.payto_prim_id, > clmcom1.bill_prim_id, > clmcom1.clm_tot_ncov_chgs, > clmcom2.contract_amt, > clmcom2.svc_fac_or_lab_name, > clmcom2.svc_fac_addr_1, > clmcom2.svc_fac_addr_2, > clmcom2.svc_fac_city, > clmcom2.svc_fac_zip; > > > Here is the explain analyze from the problem db: > prob_db=3D# explain analyze select * from clm_com; > > > > QUERY PLAN > > > > -------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------- > Subquery Scan clm_com (cost=3D1039824.35..1150697.61 rows=3D126712 > width=3D367) (actual time=3D311792.78..405819.03 rows=3D1266114 loops=3D1= ) > -> Unique (cost=3D1039824.35..1150697.61 rows=3D126712 width=3D367) > (actual time=3D311792.74..386313.14 rows=3D1266114 loops=3D1) > -> Sort (cost=3D1039824.35..1042992.16 rows=3D1267123 > width=3D367) (actual time=3D311792.74..338189.48 rows=3D1266114 loops=3D1= ) > Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, > clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, > clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, > clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, > clmcom1.pat_zip, clmcom1.payto_addr_1, > clmcom1.payto_addr_2, clmcom1.payto_city, clmcom1.payto_cntry, > clmcom1.payto_f_name, clmcom1.payto_m_name, clmcom1.payto_state, > clmcom1.payto_zip, clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, > clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, > clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, > clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, > clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, > clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip > -> Hash Join (cost=3D132972.78..548171.70 rows=3D1267123 > width=3D367) (actual time=3D16999.32..179359.43 rows=3D1266114 loops=3D1) > Hash Cond: ("outer".inv_nbr =3D "inner".inv_nbr) > Join Filter: ("outer".inv_qfr =3D "inner".inv_qfr) > -> Seq Scan on clmcom1 (cost=3D0.00..267017.23 > rows=3D1267123 width=3D271) (actual time=3D0.11..84711.83 rows=3D1266114 > loops=3D1) > -> Hash (cost=3D111200.82..111200.82 rows=3D126958= 2 > width=3D96) (actual time=3D16987.45..16987.45 rows=3D0 loops=3D1) > -> Seq Scan on clmcom2 > (cost=3D0.00..111200.82 rows=3D1269582 width=3D96) (actual > time=3D0.07..12164.81 rows=3D1266108 loops=3D1) > Total runtime: 407317.47 msec > (11 rows) > ~ > > > > Here is the explain analyze from a good db (on the same postgres cluster)= ; > good_db=3D# explain analyze select * from clm_com; > > > > > > QUERY PLAN > -------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------------------------------------------------------= ---------------------------- > Subquery Scan clm_com (cost=3D78780.59..89498.29 rows=3D12249 width=3D3= 59) > (actual time=3D73045.36..79974.37 rows=3D122494 loops=3D1) > -> Unique (cost=3D78780.59..89498.29 rows=3D12249 width=3D359) (actu= al > time=3D73045.28..78031.99 rows=3D122494 loops=3D1) > -> Sort (cost=3D78780.59..79086.81 rows=3D122488 width=3D359) > (actual time=3D73045.28..73362.94 rows=3D122494 loops=3D1) > Sort Key: clmcom1.inv_nbr, clmcom1.inv_qfr, > clmcom1.pat_addr_1, clmcom1.pat_addr_2, clmcom1.pat_city, > clmcom1.pat_cntry, clmcom1.pat_dob, clmcom1.pat_gender_cd, > clmcom1.pat_info_pregnancy_ind, clmcom1.pat_state, clmcom1.pat_suffix, > clmcom1.pat_zip, clmcom1.payto_addr_1, clmcom1.payto_addr_2, > clmcom1.payto_city, clmcom1.payto_cntry, clmcom1.payto_f_name, > clmcom1.payto_m_name, clmcom1.payto_state, clmcom1.payto_zip, > clmcom1.clm_tot_clm_chgs, clmcom1.bill_l_name_org, > clmcom1.clm_delay_rsn_cd, clmcom1.clm_submit_rsn_cd, > clmcom1.payto_l_name_org, clmcom1.payto_prim_id, clmcom1.bill_prim_id, > clmcom1.clm_tot_ncov_chgs, clmcom2.contract_amt, > clmcom2.svc_fac_or_lab_name, clmcom2.svc_fac_addr_1, > clmcom2.svc_fac_addr_2, clmcom2.svc_fac_city, clmcom2.svc_fac_zip > -> Merge Join (cost=3D0.00..56945.12 rows=3D122488 > width=3D359) (actual time=3D54.76..71635.65 rows=3D122494 loops=3D1) > Merge Cond: (("outer".inv_nbr =3D "inner".inv_nbr) > AND ("outer".inv_qfr =3D "inner".inv_qfr)) > -> Index Scan using clmcom1_pkey on clmcom1 > (cost=3D0.00..38645.61 rows=3D122488 width=3D267) (actual > time=3D25.60..49142.16 rows=3D122494 loops=3D1) > -> Index Scan using clmcom2_pkey on clmcom2 > (cost=3D0.00..16004.08 rows=3D122488 width=3D92) (actual > time=3D29.09..19418.94 rows=3D122494 loops=3D1) > Total runtime: 80162.26 msec > (9 rows) > From pgsql-performance-owner@postgresql.org Thu Jul 28 13:13:51 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 378BE529A0 for ; Thu, 28 Jul 2005 13:13:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91158-10 for ; Thu, 28 Jul 2005 16:13:26 +0000 (GMT) Received: from crestone.coronasolutions.com (crestone.coronasolutions.com [66.45.104.24]) by svr1.postgresql.org (Postfix) with ESMTP id 0E17752889 for ; Thu, 28 Jul 2005 13:13:24 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by crestone.coronasolutions.com (Postfix) with ESMTP id E2B0864419C for ; Thu, 28 Jul 2005 10:12:56 -0600 (MDT) Received: from crestone.coronasolutions.com ([127.0.0.1]) by localhost (crestone.coronasolutions.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 20114-07 for ; Thu, 28 Jul 2005 10:12:55 -0600 (MDT) Received: from [192.168.1.12] (c-24-9-24-35.hsd1.co.comcast.net [24.9.24.35]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by crestone.coronasolutions.com (Postfix) with ESMTP id DF3D4644178 for ; Thu, 28 Jul 2005 10:12:54 -0600 (MDT) Mime-Version: 1.0 (Apple Message framework v730) In-Reply-To: <1d219a6f0507280738791f0ea3@mail.gmail.com> References: <1d219a6f05072708122cf6abeb@mail.gmail.com> <1d219a6f050727092927746db1@mail.gmail.com> <1d219a6f0507280738791f0ea3@mail.gmail.com> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Dan Harris Subject: Re: Fwd: Help with view performance problem Date: Thu, 28 Jul 2005 10:14:32 -0600 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.730) X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at drivefaster.net X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/439 X-Sequence-Number: 13680 On Jul 28, 2005, at 8:38 AM, Chris Hoover wrote: > > > I did some more testing, and ran the explain analyze on the problem. > In my session I did a set enable_hashjoin = false and then ran the > analyze. This caused it to use the indexes as I have been expecting > it to do. > > Now, how can I get it to use the indexes w/o manipulating the > environment? What make postgresql want to sequentially scan and use a > hash join? > > thanks, > > Chris > > explain analyze with set_hashjoin=false; > prob_db=#explain analyze select * from clm_com; > > I had something similar to this happen recently. The planner was choosing a merge join and seq scan because my 'random_page_cost' was set too high. I had it at 3 , and ended up settling at 1.8 to get it to correctly use my indices. Once that change was in place, the planner did the 'right' thing for me. Not sure if this will help you, but it sounds similar. -Dan From pgsql-performance-owner@postgresql.org Thu Jul 28 14:02:46 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EBD1952B49 for ; Thu, 28 Jul 2005 14:02:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04843-04 for ; Thu, 28 Jul 2005 17:02:39 +0000 (GMT) Received: from mail.travelamericas.com (unknown [206.130.134.147]) by svr1.postgresql.org (Postfix) with SMTP id 9BDDC52B5D for ; Thu, 28 Jul 2005 14:02:37 -0300 (ADT) Received: (qmail 6500 invoked from network); 28 Jul 2005 16:55:55 -0000 Received: from unknown (HELO ?10.0.0.253?) (10.0.0.253) by verkiel.travelamericas.com with SMTP; 28 Jul 2005 16:55:55 -0000 Message-ID: <42E90E19.3030702@travelamericas.com> Date: Thu, 28 Jul 2005 09:55:53 -0700 From: Chris Travers User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Gnanavel S Cc: pgsql-performance@postgresql.org Subject: Re: Left joining against two empty tables makes a query References: <42E86A03.1070803@travelamericas.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.037 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/441 X-Sequence-Number: 13682 Gnanavel S wrote: > > > vacuum & reindex the department and project table as the planner > expects there are 1060 rows but actually returning nothing. I guess I should have mentioned that I have been vacuuming and reindexing at least once a week, and I did so just before running this test. Normally I do: vacuum analyze; reindex database ....; Secondly, the project table has *never* had anything in it. So where are these numbers coming from? Best Wishes, Chris Travers Metatron Technology Consulting From pgsql-performance-owner@postgresql.org Thu Jul 28 14:00:40 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 886BE529CC for ; Thu, 28 Jul 2005 14:00:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04559-03 for ; Thu, 28 Jul 2005 17:00:29 +0000 (GMT) Received: from e5.ny.us.ibm.com (e5.ny.us.ibm.com [32.97.182.145]) by svr1.postgresql.org (Postfix) with ESMTP id E754C52AF0 for ; Thu, 28 Jul 2005 14:00:21 -0300 (ADT) Received: from d01relay02.pok.ibm.com (d01relay02.pok.ibm.com [9.56.227.234]) by e5.ny.us.ibm.com (8.12.11/8.12.11) with ESMTP id j6SH0KlA025210 for ; Thu, 28 Jul 2005 13:00:20 -0400 Received: from d01av02.pok.ibm.com (d01av02.pok.ibm.com [9.56.224.216]) by d01relay02.pok.ibm.com (8.12.10/NCO/VERS6.7) with ESMTP id j6SH0KEk262098 for ; Thu, 28 Jul 2005 13:00:20 -0400 Received: from d01av02.pok.ibm.com (loopback [127.0.0.1]) by d01av02.pok.ibm.com (8.12.11/8.13.3) with ESMTP id j6SH0JcL016934 for ; Thu, 28 Jul 2005 13:00:20 -0400 Received: from d01ml255.pok.ibm.com (d01ml255.pok.ibm.com [9.56.227.128]) by d01av02.pok.ibm.com (8.12.11/8.12.11) with ESMTP id j6SH0IZf016498 for ; Thu, 28 Jul 2005 13:00:19 -0400 Subject: Unable to explain DB error To: pgsql-performance@postgresql.org X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 Message-ID: From: Steven Rosenstein Date: Thu, 28 Jul 2005 13:00:05 -0400 X-MIMETrack: Serialize by Router on D01ML255/01/M/IBM(Release 6.53HF103 | November 15, 2004) at 07/28/2005 13:00:18 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.305 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE X-Spam-Level: X-Archive-Number: 200507/440 X-Sequence-Number: 13681 Postgres V7.3.9-2. While executing a query in psql, the following error was generated: vsa=# select * from vsa.dtbl_logged_event_20050318 where id=2689472; PANIC: open of /vsa/db/pg_clog/0FC0 failed: No such file or directory server closed the connection unexpectedly This probably means the server terminated abnormally before or while processing the request. The connection to the server was lost. Attempting reset: Failed. !# I checked in the /vsa/db/pg_clog directory, and the files have monotonically increasing filenames starting with 0000. The most recent names are: -rw------- 1 postgres postgres 262144 Jul 25 21:39 04CA -rw------- 1 postgres postgres 262144 Jul 26 01:10 04CB -rw------- 1 postgres postgres 262144 Jul 26 05:39 04CC -rw------- 1 postgres postgres 262144 Jul 28 00:01 04CD -rw------- 1 postgres postgres 237568 Jul 28 11:31 04CE Any idea why Postgres would be looking for a clog file name 0FC0 when the most recent filename is 04CE? Any help and suggestions for recovery are appreciated. --- Steve ___________________________________________________________________________________ Steven Rosenstein IT Architect/Developer | IBM Virtual Server Administration Voice/FAX: 845-689-2064 | Cell: 646-345-6978 | Tieline: 930-6001 Text Messaging: 6463456978 @ mobile.mycingular.com Email: srosenst @ us.ibm.com "Learn from the mistakes of others because you can't live long enough to make them all yourself." -- Eleanor Roosevelt From pgsql-performance-owner@postgresql.org Thu Jul 28 14:35:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 97F68529FD for ; Thu, 28 Jul 2005 14:35:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 10065-08 for ; Thu, 28 Jul 2005 17:35:22 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.200]) by svr1.postgresql.org (Postfix) with ESMTP id ECA5D52B84 for ; Thu, 28 Jul 2005 14:35:16 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id 8so356729nzo for ; Thu, 28 Jul 2005 10:35:17 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=F6GEESUg4xHYNlUCSmV9SK0dGMvszWV0gDoWOlCsFrsMAa5Zjsjr2nz8fzu0Yd/YX2bkjoZ55iKepmOVlRKjLTAL6NX4tppMEUmpeMZdI1FPOIOVVxh00iDHc1VAJarUaTCe1z5+19mTzgeOtcJsrA8QY1fC8TFNJdpLCenIVHA= Received: by 10.36.133.15 with SMTP id g15mr2073505nzd; Thu, 28 Jul 2005 10:34:55 -0700 (PDT) Received: by 10.36.25.13 with HTTP; Thu, 28 Jul 2005 10:34:54 -0700 (PDT) Message-ID: <1d219a6f05072810341d67c525@mail.gmail.com> Date: Thu, 28 Jul 2005 13:34:54 -0400 From: Chris Hoover Reply-To: Chris Hoover To: Dan Harris Subject: Re: Fwd: Help with view performance problem Cc: pgsql-performance@postgresql.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline References: <1d219a6f05072708122cf6abeb@mail.gmail.com> <1d219a6f050727092927746db1@mail.gmail.com> <1d219a6f0507280738791f0ea3@mail.gmail.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/442 X-Sequence-Number: 13683 I'm alreading running at 1.5. It looks like if I drop the random_page_cost t0 1.39, it starts using the indexes. Are there any unseen issues with dropping the random_page_cost this low? Thanks, Chris On 7/28/05, Dan Harris wrote: >=20 > On Jul 28, 2005, at 8:38 AM, Chris Hoover wrote: > > > > > > I did some more testing, and ran the explain analyze on the problem. > > In my session I did a set enable_hashjoin =3D false and then ran the > > analyze. This caused it to use the indexes as I have been expecting > > it to do. > > > > Now, how can I get it to use the indexes w/o manipulating the > > environment? What make postgresql want to sequentially scan and use a > > hash join? > > > > thanks, > > > > Chris > > > > explain analyze with set_hashjoin=3Dfalse; > > prob_db=3D#explain analyze select * from clm_com; > > > > >=20 > I had something similar to this happen recently. The planner was > choosing a merge join and seq scan because my 'random_page_cost' was > set too high. I had it at 3 , and ended up settling at 1.8 to get it > to correctly use my indices. Once that change was in place, the > planner did the 'right' thing for me. >=20 > Not sure if this will help you, but it sounds similar. >=20 > -Dan >=20 > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster > From pgsql-performance-owner@postgresql.org Thu Jul 28 15:14:03 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DC3D052B84 for ; Thu, 28 Jul 2005 15:14:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18749-03 for ; Thu, 28 Jul 2005 18:13:53 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 251F652AAB for ; Thu, 28 Jul 2005 15:13:52 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6SIDpfF025700; Thu, 28 Jul 2005 14:13:51 -0400 (EDT) To: Chris Travers Cc: Gnanavel S , pgsql-performance@postgresql.org Subject: Re: Left joining against two empty tables makes a query In-reply-to: <42E90E19.3030702@travelamericas.com> References: <42E86A03.1070803@travelamericas.com> <42E90E19.3030702@travelamericas.com> Comments: In-reply-to Chris Travers message dated "Thu, 28 Jul 2005 09:55:53 -0700" Date: Thu, 28 Jul 2005 14:13:51 -0400 Message-ID: <25699.1122574431@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/443 X-Sequence-Number: 13684 Chris Travers writes: > Secondly, the project table has *never* had anything in it. So where > are these numbers coming from? The planner is designed to assume a certain minimum size (10 pages) when it sees that a table is of zero physical length. The reason for this is that there are lots of scenarios where a plan created just after a table is first created will continue to be used while the table is filled, and if we optimized on the assumption of zero size we would produce plans that seriously suck once the table gets big. Assuming a few thousand rows keeps us out of the worst problems of this type. (If we had an infrastructure for regenerating cached plans then we could fix this more directly, by replanning whenever the table size changes "too much". We don't yet but I hope there will be something by 8.2.) You might try going ahead and actually putting a row or two into projects; vacuuming that will change the state to where the planner will believe the small size. (If you aren't ever planning to have anything in projects, why have the table at all?) regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 28 15:19:38 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DAAB65285A for ; Thu, 28 Jul 2005 15:19:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20061-05 for ; Thu, 28 Jul 2005 18:19:29 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id AEA0052A01 for ; Thu, 28 Jul 2005 15:19:28 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6SIJS06025775; Thu, 28 Jul 2005 14:19:29 -0400 (EDT) To: Steven Rosenstein Cc: pgsql-performance@postgresql.org Subject: Re: Unable to explain DB error In-reply-to: References: Comments: In-reply-to Steven Rosenstein message dated "Thu, 28 Jul 2005 13:00:05 -0400" Date: Thu, 28 Jul 2005 14:19:28 -0400 Message-ID: <25774.1122574768@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/444 X-Sequence-Number: 13685 Steven Rosenstein writes: > Any idea why Postgres would be looking for a clog file name 0FC0 when the most recent filename is 04CE? Corrupt data --- specifically a bad transaction number in a tuple header. (In practice, this is the first field looked at in which we can readily detect an error, so you tend to see this symptom for any serious data corruption situation. The actual fault may well be something like a corrupt page header causing the code to follow "tuple pointers" that point to garbage.) See the PG list archives for past discussions of dealing with corrupt data. pgsql-performance is pretty off-topic for this. BTW, PG 7.4 and up handle this sort of thing much more gracefully ... they can't resurrect corrupt data of course, but they tend not to panic. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Jul 28 15:27:49 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E002B52A35 for ; Thu, 28 Jul 2005 15:27:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20591-06 for ; Thu, 28 Jul 2005 18:27:45 +0000 (GMT) Received: from mail.Mi8.com (mail.mi8.com [63.240.6.50]) by svr1.postgresql.org (Postfix) with ESMTP id 104B352A15 for ; Thu, 28 Jul 2005 15:27:44 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Thu, 28 Jul 2005 14:27:38 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Thu, 28 Jul 2005 14:27:37 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Thu, 28 Jul 2005 14:27:37 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 28 Jul 2005 11:27:36 -0700 Subject: Re: Finding bottleneck From: "Luke Lonergan" To: "Kari Lavikka" , pgsql-performance@postgresql.org Message-ID: In-Reply-To: MIME-Version: 1.0 X-OriginalArrivalTime: 28 Jul 2005 18:27:37.0969 (UTC) FILETIME=[0817CE10:01C593A2] X-WSS-ID: 6EF7FC101M815461377-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.577 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/445 X-Sequence-Number: 13686 On 7/28/05 2:21 AM, "Kari Lavikka" wrote: There's a new profiling tool called oprofile: http://oprofile.sourceforge.net/download/ that can be run without instrumenting the binaries beforehand. To actually find out what the code is doing during these stalls, oprofile can show you in which routines the CPU is spending time when you start/stop the profiling. As an alternative to the "guess->change parameters->repeat" approach, this is the most direct way to find the exact nature of the problem. - Luke From pgsql-hackers-owner@postgresql.org Thu Jul 28 18:54:29 2005 X-Original-To: pgsql-hackers-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6643D5372B for ; Thu, 28 Jul 2005 18:54:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07598-02 for ; Thu, 28 Jul 2005 21:54:27 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 668A7537AD for ; Thu, 28 Jul 2005 18:54:19 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7666796; Thu, 28 Jul 2005 14:56:34 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Greg Stark Subject: Re: wal_buffer tests in Date: Thu, 28 Jul 2005 14:57:10 -0700 User-Agent: KMail/1.8 Cc: pgsql-hackers@postgresql.org References: <200507271330.02083.josh@agliodbs.com> <87oe8n881i.fsf@stark.xeocode.com> In-Reply-To: <87oe8n881i.fsf@stark.xeocode.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507281457.10637.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.049 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/1061 X-Sequence-Number: 71113 Greg, > You have wal_buffer set to 2048? That's pretty radical compared to the > default of just 5. Your tests shows you had to go to this large a value > to see the maximum effect? No, take a look at the graph. It looks like we got the maximum effect from a wal_buffers somewhere between 64 and 256. On the DBT2 runs, any variation less than 5% is just noise. -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Jul 28 20:06:36 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7084852AA3 for ; Thu, 28 Jul 2005 20:06:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 42277-04 for ; Thu, 28 Jul 2005 23:06:29 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id 1714B52AFA for ; Thu, 28 Jul 2005 20:06:11 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id 8222D35F9D9 for ; Thu, 28 Jul 2005 23:06:13 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 37468-10 for ; Thu, 28 Jul 2005 23:06:04 +0000 (GMT) Received: from [10.0.0.7] (63-228-143-68.phnx.qwest.net [63.228.143.68]) by veggie-fuel.com (Postfix) with ESMTP id CD81035F9D6 for ; Thu, 28 Jul 2005 23:06:04 +0000 (GMT) Subject: Two queries are better than one? From: Karim Nassar Reply-To: karim.nassar@acm.org To: pgsql-performance@postgresql.org Content-Type: text/plain Date: Thu, 28 Jul 2005 16:04:25 -0700 Message-Id: <1122591866.11860.11.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.069 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/446 X-Sequence-Number: 13687 I ran into a situation today maintaining someone else's code where the sum time running 2 queries seems to be faster than 1. The original code was split into two queries. I thought about joining them, but considering the intelligence of my predecessor, I wanted to test it. The question is, which technique is really faster? Is there some hidden setup cost I don't see with explain analyze? Postgres 7.4.7, Redhat AES 3 Each query individually: test=> explain analyze test-> select * from order WHERE ord_batch='343B' AND ord_id='12-645'; QUERY PLAN ---------------------------------------------------------------------------------------------------------------------------- Index Scan using order_pkey on order (cost=0.00..6.02 rows=1 width=486) (actual time=0.063..0.066 rows=1 loops=1) Index Cond: ((ord_batch = '343B'::bpchar) AND (ord_id = '12-645'::bpchar)) Total runtime: 0.172 ms (3 rows) test=> explain analyze test-> select cli_name from client where cli_code='1837'; QUERY PLAN --------------------------------------------------------------------------------------------------------------------- Index Scan using client_pkey on client (cost=0.00..5.98 rows=2 width=39) (actual time=0.043..0.047 rows=1 loops=1) Index Cond: (cli_code = '1837'::bpchar) Total runtime: 0.112 ms (3 rows) Joined: test=> explain analyze test-> SELECT cli_name,order.* test-> FROM order test-> JOIN client ON (ord_client = cli_code) test-> WHERE ord_batch='343B' AND ord_id='12-645'; QUERY PLAN ---------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=0.00..12.00 rows=2 width=525) (actual time=0.120..0.128 rows=1 loops=1) -> Index Scan using order_pkey on order (cost=0.00..6.02 rows=1 width=486) (actual time=0.064..0.066 rows=1 loops=1) Index Cond: ((ord_batch = '343B'::bpchar) AND (ord_id = '12-645'::bpchar)) -> Index Scan using client_pkey on client (cost=0.00..5.98 rows=1 width=51) (actual time=0.023..0.026 rows=1 loops=1) Index Cond: ("outer".ord_client = client.cli_code) Total runtime: 0.328 ms (6 rows) -- Karim Nassar From pgsql-performance-owner@postgresql.org Thu Jul 28 20:43:39 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2F00C529B6 for ; Thu, 28 Jul 2005 20:43:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 47294-08 for ; Thu, 28 Jul 2005 23:43:35 +0000 (GMT) Received: from smtp.osdl.org (smtp.osdl.org [65.172.181.4]) by svr1.postgresql.org (Postfix) with ESMTP id 4B15252971 for ; Thu, 28 Jul 2005 20:43:32 -0300 (ADT) Received: from lemming.pdx.osdl.net (fw.osdl.org [65.172.181.6]) (authenticated bits=0) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j6SNhEjA013300 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Thu, 28 Jul 2005 16:43:15 -0700 Message-Id: <200507282343.j6SNhEjA013300@smtp.osdl.org> Date: Thu, 28 Jul 2005 16:43:52 -0700 From: Mark Wong To: "Luke Lonergan" Cc: "Joshua D. Drake" , "Patrick Welche" , pgsql-performance@postgresql.org, maryedie@osdl.org Subject: Re: [PATCHES] COPY FROM performance improvements In-Reply-To: References: <42E128BD.2000104@commandprompt.com> Organization: OSDL X-Mailer: Sylpheed-Claws 1.0.4a (GTK+ 1.2.10; x86_64-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-MIMEDefang-Filter: osdl$Revision: 1.113 $ X-Scanned-By: MIMEDefang 2.36 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.055 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/447 X-Sequence-Number: 13688 On Fri, 22 Jul 2005 12:28:43 -0700 "Luke Lonergan" wrote: > Joshua, > > On 7/22/05 10:11 AM, "Joshua D. Drake" wrote: > > The database server is a PE (Power Edge) 6600 > > > > Database Server IO: > > > > [root@master root]# /sbin/hdparm -tT /dev/sda > > > > /dev/sda: > > Timing buffer-cache reads: 1888 MB in 2.00 seconds = 944.00 MB/sec > > Timing buffered disk reads: 32 MB in 3.06 seconds = 10.46 MB/sec > > > > Second Database Server IO: > > > > [root@pq-slave root]# /sbin/hdparm -tT /dev/sda > > > > /dev/sda: > > Timing buffer-cache reads: 1816 MB in 2.00 seconds = 908.00 MB/sec > > Timing buffered disk reads: 26 MB in 3.11 seconds = 8.36 MB/sec > > [root@pq-slave root]# > > Can you post the "time dd if=/dev/zero of=bigfile bs=8k count=500000" > results? Also do the reverse (read the file) with "time dd if=bigfile > of=/dev/null bs=8k". > > I think you are observing what we've known for a while, hardware RAID is > horribly slow. We've not found a hardware RAID adapter of this class yet > that shows reasonable read or write performance. The Adaptec 2400R or the > LSI or others have terrible internal I/O compared to raw SCSI with software > RAID, and even the CPU usage is higher on these cards while doing slower I/O > than linux SW RAID. > > Notably - we've found that the 3Ware RAID controller does a better job than > the low end SCSI RAID at HW RAID support, and also exports JBOD at high > speeds. If you export JBOD on the low end SCSI RAID adapters, the > performance is also very poor, though generally faster than using HW RAID. Are there any recommendations for Qlogic controllers on Linux, scsi or fiber channel? I might be able to my hands on some. I have pci-x slots for AMD, Itanium, or POWER5 if the architecture makes a difference. Mark From pgsql-performance-owner@postgresql.org Thu Jul 28 20:57:59 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id EF8B652A48 for ; Thu, 28 Jul 2005 20:57:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 53326-06 for ; Thu, 28 Jul 2005 23:57:54 +0000 (GMT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 6FDA452A38 for ; Thu, 28 Jul 2005 20:57:52 -0300 (ADT) Received: from [192.168.1.101] (clbb-248.saw.net [64.146.135.248]) (authenticated bits=0) by hosting.commandprompt.com (8.13.4/8.13.4) with ESMTP id j6SNv7M6023411; Thu, 28 Jul 2005 16:57:08 -0700 Message-ID: <42E970E4.4010902@commandprompt.com> Date: Thu, 28 Jul 2005 16:57:24 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Mark Wong Cc: Luke Lonergan , Patrick Welche , pgsql-performance@postgresql.org, maryedie@osdl.org Subject: Re: [PATCHES] COPY FROM performance improvements References: <42E128BD.2000104@commandprompt.com> <200507282343.j6SNhEjA013300@smtp.osdl.org> In-Reply-To: <200507282343.j6SNhEjA013300@smtp.osdl.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.024 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/448 X-Sequence-Number: 13689 >> >>>[root@pq-slave root]# /sbin/hdparm -tT /dev/sda >>> >>>/dev/sda: >>> Timing buffer-cache reads: 1816 MB in 2.00 seconds = 908.00 MB/sec >>> Timing buffered disk reads: 26 MB in 3.11 seconds = 8.36 MB/sec >>>[root@pq-slave root]# >> >>Can you post the "time dd if=/dev/zero of=bigfile bs=8k count=500000" >>results? Also do the reverse (read the file) with "time dd if=bigfile >>of=/dev/null bs=8k". I didn't see this come across before... here ya go: time dd if=/dev/zero of=bigfile bs=8k count=500000 500000+0 records in 500000+0 records out real 1m52.738s user 0m0.310s sys 0m36.590s time dd if=bigfile of=/dev/null bs=8k time dd if=bigfile of=/dev/null bs=8k 500000+0 records in 500000+0 records out real 4m38.742s user 0m0.320s sys 0m27.870s FYI on your hardware raid comment... I easily get 50 megs a second on my 3ware controllers and faster on my LSI SATA controllers. Sincerely, Joshua D. Drake > -- Your PostgreSQL solutions provider, Command Prompt, Inc. 24x7 support - 1.800.492.2240, programming, and consulting Home of PostgreSQL Replicator, plPHP, plPerlNG and pgPHPToolkit http://www.commandprompt.com / http://www.postgresql.org From pgsql-performance-owner@postgresql.org Thu Jul 28 21:13:33 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3D1955298B for ; Thu, 28 Jul 2005 21:13:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 53994-07 for ; Fri, 29 Jul 2005 00:13:27 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 487EE529B6 for ; Thu, 28 Jul 2005 21:13:22 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6T0Qgja015145 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Thu, 28 Jul 2005 16:26:43 -0800 Message-ID: <42E9749F.6000709@aptalaska.net> Date: Thu, 28 Jul 2005 16:13:19 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> In-Reply-To: <1122532023.25574.1.camel@localhost.localdomain> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/449 X-Sequence-Number: 13690 Karim Nassar wrote: > On Wed, 2005-07-27 at 14:35 -0800, Matthew Schumacher wrote: > > >>I put the rest of the schema up at >>http://www.aptalaska.net/~matt.s/bayes/bayes_pg.sql in case someone >>needs to see it too. > > > Do you have sample data too? > Ok, I finally got some test data together so that others can test without installing SA. The schema and test dataset is over at http://www.aptalaska.net/~matt.s/bayes/bayesBenchmark.tar.gz I have a pretty fast machine with a tuned postgres and it takes it about 2 minutes 30 seconds to load the test data. Since the test data is the bayes information on 616 spam messages than comes out to be about 250ms per message. While that is doable, it does add quite a bit of overhead to the email system. Perhaps this is as fast as I can expect it to go, if that's the case I may have to look at mysql, but I really don't want to do that. I will be working on some other benchmarks, and reading though exactly how bayes works, but at least there is some data to play with. schu From pgsql-performance-owner@postgresql.org Thu Jul 28 22:55:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6490A52923 for ; Thu, 28 Jul 2005 22:55:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 79420-10 for ; Fri, 29 Jul 2005 01:55:03 +0000 (GMT) Received: from tigger.fuhr.org (tigger.fuhr.org [63.214.45.158]) by svr1.postgresql.org (Postfix) with ESMTP id 353C45280D for ; Thu, 28 Jul 2005 22:55:00 -0300 (ADT) Received: from winnie.fuhr.org (winnie.fuhr.org [10.1.0.1]) by tigger.fuhr.org (8.13.3/8.13.3) with ESMTP id j6T1rNIj028754 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Thu, 28 Jul 2005 19:53:25 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: from winnie.fuhr.org (localhost [127.0.0.1]) by winnie.fuhr.org (8.13.3/8.13.3) with ESMTP id j6T1rM3X025132; Thu, 28 Jul 2005 19:53:22 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: (from mfuhr@localhost) by winnie.fuhr.org (8.13.3/8.13.3/Submit) id j6T1rMcT025131; Thu, 28 Jul 2005 19:53:22 -0600 (MDT) (envelope-from mfuhr) Date: Thu, 28 Jul 2005 19:53:22 -0600 From: Michael Fuhr To: Karim Nassar Cc: pgsql-performance@postgresql.org Subject: Re: Two queries are better than one? Message-ID: <20050729015322.GA25003@winnie.fuhr.org> References: <1122591866.11860.11.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1122591866.11860.11.camel@localhost.localdomain> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/450 X-Sequence-Number: 13691 On Thu, Jul 28, 2005 at 04:04:25PM -0700, Karim Nassar wrote: > I ran into a situation today maintaining someone else's code where the > sum time running 2 queries seems to be faster than 1. The original code > was split into two queries. I thought about joining them, but > considering the intelligence of my predecessor, I wanted to test it. > > The question is, which technique is really faster? Is there some hidden > setup cost I don't see with explain analyze? To see which technique will be faster in your application, time the application code. The queries you show are taking fractions of a millisecond; the communications overhead of executing two queries might make that technique significantly slower than just the server execution time that EXPLAIN ANALYZE shows. -- Michael Fuhr http://www.fuhr.org/~mfuhr/ From pgsql-performance-owner@postgresql.org Thu Jul 28 23:01:11 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 079B4528F5 for ; Thu, 28 Jul 2005 23:01:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 80494-05 for ; Fri, 29 Jul 2005 02:01:05 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 4D17B5299F for ; Thu, 28 Jul 2005 23:01:00 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050729020103m9100g6kese>; Fri, 29 Jul 2005 02:01:03 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 85DF256011; Thu, 28 Jul 2005 21:01:03 -0500 (CDT) Received: from [192.168.1.12] (Jigglypuff.arbash-meinel.com [192.168.1.12]) by juju.arbash-meinel.com (Postfix) with ESMTP id 7F12456006; Thu, 28 Jul 2005 21:00:59 -0500 (CDT) Message-ID: <42E98DDD.5000505@arbash-meinel.com> Date: Thu, 28 Jul 2005 21:01:01 -0500 From: John A Meinel User-Agent: Mozilla Thunderbird 1.0.6 (Macintosh/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: karim.nassar@acm.org, Postgresql Performance Subject: Re: Two queries are better than one? References: <1122591866.11860.11.camel@localhost.localdomain> In-Reply-To: <1122591866.11860.11.camel@localhost.localdomain> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigAB99DF19056F44E02782CD13" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.044 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/451 X-Sequence-Number: 13692 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigAB99DF19056F44E02782CD13 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Karim Nassar wrote: > I ran into a situation today maintaining someone else's code where the > sum time running 2 queries seems to be faster than 1. The original code > was split into two queries. I thought about joining them, but > considering the intelligence of my predecessor, I wanted to test it. > > The question is, which technique is really faster? Is there some hidden > setup cost I don't see with explain analyze? Yes, the time it takes your user code to parse the result, and create the new query. :) It does seem like you are taking an extra 0.1ms for the combined query, but that means you don't have another round trip to the database. So that would mean one less context switch, and you don't need to know what the cli_code is before you can get the cli_name. I would guess the overhead is the time for postgres to parse out the text, place another index query, and then combine the rows. It seems like this shouldn't take 0.1ms, but then again, that isn't very long. Also, did you run it *lots* of times to make sure that this isn't just noise? John =:-> > > Postgres 7.4.7, Redhat AES 3 > > Each query individually: > > test=> explain analyze > test-> select * from order WHERE ord_batch='343B' AND ord_id='12-645'; > QUERY PLAN > ---------------------------------------------------------------------------------------------------------------------------- > Index Scan using order_pkey on order (cost=0.00..6.02 rows=1 width=486) (actual time=0.063..0.066 rows=1 loops=1) > Index Cond: ((ord_batch = '343B'::bpchar) AND (ord_id = '12-645'::bpchar)) > Total runtime: 0.172 ms > (3 rows) > > > test=> explain analyze > test-> select cli_name from client where cli_code='1837'; > QUERY PLAN > --------------------------------------------------------------------------------------------------------------------- > Index Scan using client_pkey on client (cost=0.00..5.98 rows=2 width=39) (actual time=0.043..0.047 rows=1 loops=1) > Index Cond: (cli_code = '1837'::bpchar) > Total runtime: 0.112 ms > (3 rows) > > Joined: > > test=> explain analyze > test-> SELECT cli_name,order.* > test-> FROM order > test-> JOIN client ON (ord_client = cli_code) > test-> WHERE ord_batch='343B' AND ord_id='12-645'; > QUERY PLAN > ---------------------------------------------------------------------------------------------------------------------------------- > Nested Loop (cost=0.00..12.00 rows=2 width=525) (actual time=0.120..0.128 rows=1 loops=1) > -> Index Scan using order_pkey on order (cost=0.00..6.02 rows=1 width=486) (actual time=0.064..0.066 rows=1 loops=1) > Index Cond: ((ord_batch = '343B'::bpchar) AND (ord_id = '12-645'::bpchar)) > -> Index Scan using client_pkey on client (cost=0.00..5.98 rows=1 width=51) (actual time=0.023..0.026 rows=1 loops=1) > Index Cond: ("outer".ord_client = client.cli_code) > Total runtime: 0.328 ms > (6 rows) > > --------------enigAB99DF19056F44E02782CD13 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (Darwin) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC6Y3eJdeBCYSNAAMRAvTBAKDC2CENfXNdYiCbQ5RjQ/M4tlnIOQCfVFaK CSN0dsCFMpgB4e/eXMTqFmA= =Nvyw -----END PGP SIGNATURE----- --------------enigAB99DF19056F44E02782CD13-- From pgsql-performance-owner@postgresql.org Thu Jul 28 23:04:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5BC89528CC for ; Thu, 28 Jul 2005 23:04:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 83168-08 for ; Fri, 29 Jul 2005 02:04:10 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id 1628E529B3 for ; Thu, 28 Jul 2005 23:04:08 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id 2A81435F9DE for ; Fri, 29 Jul 2005 02:04:12 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 81961-08; Fri, 29 Jul 2005 02:04:02 +0000 (GMT) Received: from [10.0.0.7] (63-228-143-68.phnx.qwest.net [63.228.143.68]) by veggie-fuel.com (Postfix) with ESMTP id DCB4A35F9D6; Fri, 29 Jul 2005 02:04:03 +0000 (GMT) Subject: Re: Two queries are better than one? From: Karim Nassar Reply-To: karim.nassar@acm.org To: John A Meinel Cc: Postgresql Performance In-Reply-To: <42E98DDD.5000505@arbash-meinel.com> References: <1122591866.11860.11.camel@localhost.localdomain> <42E98DDD.5000505@arbash-meinel.com> Content-Type: text/plain Date: Thu, 28 Jul 2005 19:02:13 -0700 Message-Id: <1122602533.11862.35.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.034 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/452 X-Sequence-Number: 13693 On Thu, 2005-07-28 at 21:01 -0500, John A Meinel wrote: > > Also, did you run it *lots* of times to make sure that this isn't just > noise? If a dozen is lots, yes. :-) It was very consistent as I repeatedly ran it. -- Karim Nassar From pgsql-performance-owner@postgresql.org Fri Jul 29 00:57:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E5ACB52840 for ; Fri, 29 Jul 2005 00:57:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 14607-05 for ; Fri, 29 Jul 2005 03:57:31 +0000 (GMT) Received: from linuxworld.com.au (unknown [203.34.46.50]) by svr1.postgresql.org (Postfix) with ESMTP id 0788852816 for ; Fri, 29 Jul 2005 00:57:29 -0300 (ADT) Received: from linuxworld.com.au (IDENT:swm@localhost.localdomain [127.0.0.1]) by linuxworld.com.au (8.13.2/8.13.2) with ESMTP id j6T3vG6D010635; Fri, 29 Jul 2005 13:57:17 +1000 Received: from localhost (swm@localhost) by linuxworld.com.au (8.13.2/8.13.2/Submit) with ESMTP id j6T3vFZ9010632; Fri, 29 Jul 2005 13:57:16 +1000 X-Authentication-Warning: linuxworld.com.au: swm owned process doing -bs Date: Fri, 29 Jul 2005 13:57:15 +1000 (EST) From: Gavin Sherry X-X-Sender: swm@linuxworld.com.au To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 In-Reply-To: <42E9749F.6000709@aptalaska.net> Message-ID: References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/453 X-Sequence-Number: 13694 On Thu, 28 Jul 2005, Matthew Schumacher wrote: > Karim Nassar wrote: > > On Wed, 2005-07-27 at 14:35 -0800, Matthew Schumacher wrote: > > > > > >>I put the rest of the schema up at > >>http://www.aptalaska.net/~matt.s/bayes/bayes_pg.sql in case someone > >>needs to see it too. > > > > > > Do you have sample data too? > > > > Ok, I finally got some test data together so that others can test > without installing SA. > > The schema and test dataset is over at > http://www.aptalaska.net/~matt.s/bayes/bayesBenchmark.tar.gz > > I have a pretty fast machine with a tuned postgres and it takes it about > 2 minutes 30 seconds to load the test data. Since the test data is the > bayes information on 616 spam messages than comes out to be about 250ms > per message. While that is doable, it does add quite a bit of overhead > to the email system. I had a look at your data -- thanks. I have a question though: put_token() is invoked 120596 times in your benchmark... for 616 messages. That's nearly 200 queries (not even counting the 1-8 (??) inside the function itself) per message. Something doesn't seem right there.... Gavin From pgsql-performance-owner@postgresql.org Fri Jul 29 01:03:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 82B1D5281B for ; Fri, 29 Jul 2005 01:03:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 14318-06 for ; Fri, 29 Jul 2005 04:03:30 +0000 (GMT) Received: from mail.Mi8.com (unknown [63.240.6.46]) by svr1.postgresql.org (Postfix) with ESMTP id 31FD952B3E for ; Fri, 29 Jul 2005 01:03:26 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Fri, 29 Jul 2005 00:03:21 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 29 Jul 2005 00:03:21 -0400 Received: from 24.5.173.15 ([24.5.173.15]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Fri, 29 Jul 2005 00:03:20 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 28 Jul 2005 21:03:20 -0700 Subject: Re: [PATCHES] COPY FROM performance improvements From: "Luke Lonergan" To: "Joshua D. Drake" , "Mark Wong" Cc: "Patrick Welche" , pgsql-performance@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <42E970E4.4010902@commandprompt.com> MIME-Version: 1.0 X-OriginalArrivalTime: 29 Jul 2005 04:03:21.0606 (UTC) FILETIME=[75B48260:01C593F2] X-WSS-ID: 6EF775031M815823967-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.569 tagged_above=0 required=5 tests=AWL, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/454 X-Sequence-Number: 13695 >>> Can you post the "time dd if=/dev/zero of=bigfile bs=8k count=500000" >>> results? Also do the reverse (read the file) with "time dd if=bigfile >>> of=/dev/null bs=8k". > > I didn't see this come across before... here ya go: > > time dd if=/dev/zero of=bigfile bs=8k count=500000 > > 500000+0 records in > 500000+0 records out > > real 1m52.738s > user 0m0.310s > sys 0m36.590s So, that's 35MB/s, or 1/2 of a single disk drive. > time dd if=bigfile of=/dev/null bs=8k > > time dd if=bigfile of=/dev/null bs=8k > 500000+0 records in > 500000+0 records out > > real 4m38.742s > user 0m0.320s > sys 0m27.870s And that's 14MB/s, or < 1/4 of a single disk drive. > FYI on your hardware raid comment... I easily get 50 megs a second on my > 3ware controllers and faster on my LSI SATA controllers. Then you are almost getting one disk worth of bandwidth. By comparison, we get this using Linux software RAID on Xeon or Opteron: $ time dd if=/dev/zero of=bigfile bs=8k count=500000 500000+0 records in 500000+0 records out real 0m26.927s user 0m0.074s sys 0m8.769s $ time dd if=bigfile of=/dev/null bs=8k 500000+0 records in 500000+0 records out real 0m28.190s user 0m0.039s sys 0m8.349s with less CPU usage than HW SCSI RAID controllers. - Luke From pgsql-performance-owner@postgresql.org Fri Jul 29 01:06:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 41D5652862 for ; Fri, 29 Jul 2005 01:06:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 12253-10 for ; Fri, 29 Jul 2005 04:06:19 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.200]) by svr1.postgresql.org (Postfix) with ESMTP id B532952816 for ; Fri, 29 Jul 2005 01:06:17 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id z6so339988nzd for ; Thu, 28 Jul 2005 21:06:17 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:references; b=TUEV79VD8e9JTcb1p4513QPlPedvJehLd0PCsSd6RIs0lKNS64sb+pEE1rkvE/rkOVHY1BVxMxf+jxC9E7edJt1XhVF2N1AkU2a2Esk5t6wZNQedyzQzU73lRqXDGFLp8aHPmGVi88D4Mc3N1JxjDev2XjeiR8WUaeJrnoZoNxY= Received: by 10.36.36.10 with SMTP id j10mr2380637nzj; Thu, 28 Jul 2005 21:05:40 -0700 (PDT) Received: by 10.36.119.19 with HTTP; Thu, 28 Jul 2005 21:05:40 -0700 (PDT) Message-ID: Date: Fri, 29 Jul 2005 09:35:40 +0530 From: Gnanavel S Reply-To: Gnanavel S To: Chris Travers Subject: Re: Left joining against two empty tables makes a query SLOW Cc: pgsql-performance@postgresql.org In-Reply-To: <42E90E19.3030702@travelamericas.com> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_8056_14123843.1122609940529" References: <42E86A03.1070803@travelamericas.com> <42E90E19.3030702@travelamericas.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.435 tagged_above=0 required=5 tests=AWL, HTML_50_60, HTML_MESSAGE, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/455 X-Sequence-Number: 13696 ------=_Part_8056_14123843.1122609940529 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On 7/28/05, Chris Travers wrote: >=20 > Gnanavel S wrote: >=20 > > > > > > vacuum & reindex the department and project table as the planner > > expects there are 1060 rows but actually returning nothing. >=20 > I guess I should have mentioned that I have been vacuuming and > reindexing at least once a week, and I did so just before running this=20 > test. > Normally I do: > vacuum analyze; > reindex database ....; reindex the tables separately. Secondly, the project table has *never* had anything in it. So where > are these numbers coming from? pg_statistics=20 Best Wishes, > Chris Travers > Metatron Technology Consulting >=20 --=20 with regards, S.Gnanavel Satyam Computer Services Ltd. ------=_Part_8056_14123843.1122609940529 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline

On 7/28/05, Chris Travers <c= hris@travelamericas.com> wrote:
Gnanavel S wrote:

>
>
> vacuum & reindex the depa= rtment and project table as the planner
> expects there are 1060 rows= but actually returning nothing.

I guess I should have mentioned tha= t I have been vacuuming and
reindexing at least once a week, and I did so just before running this = test.
Normally I do:
vacuum analyze;
reindex database ....;

reindex the tables separately.

Secondl= y, the project table has *never* had anything in it.  So whereare these numbers coming from?

pg_statistics

Best Wi= shes,
Chris Travers
Metatron Technology Consulting



--
with regards,
S.Gnanavel
Satyam Comp= uter Services Ltd. ------=_Part_8056_14123843.1122609940529-- From pgsql-performance-owner@postgresql.org Fri Jul 29 01:07:31 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6210852939 for ; Fri, 29 Jul 2005 01:07:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 17986-01 for ; Fri, 29 Jul 2005 04:07:23 +0000 (GMT) Received: from mail.Mi8.com (d01gw01.mi8.com [63.240.6.47]) by svr1.postgresql.org (Postfix) with ESMTP id BBF6952932 for ; Fri, 29 Jul 2005 01:07:21 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D1)); Fri, 29 Jul 2005 00:06:59 -0400 X-Server-Uuid: 241911D6-425B-44B9-A073-E3FE0F8FC774 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Fri, 29 Jul 2005 00:07:13 -0400 Received: from 24.5.173.15 ([24.5.173.15]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.106]) with Microsoft Exchange Server HTTP-DAV ; Fri, 29 Jul 2005 00:07:13 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Thu, 28 Jul 2005 21:07:12 -0700 Subject: Re: [PATCHES] COPY FROM performance improvements From: "Luke Lonergan" To: "Mark Wong" Cc: "Joshua D. Drake" , "Patrick Welche" , pgsql-performance@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <200507282343.j6SNhEjA013300@smtp.osdl.org> MIME-Version: 1.0 X-OriginalArrivalTime: 29 Jul 2005 04:07:13.0620 (UTC) FILETIME=[FFFF0940:01C593F2] X-WSS-ID: 6EF774E92B48094506-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.593 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/456 X-Sequence-Number: 13697 Mark, On 7/28/05 4:43 PM, "Mark Wong" wrote: > Are there any recommendations for Qlogic controllers on Linux, scsi or > fiber channel? I might be able to my hands on some. I have pci-x slots > for AMD, Itanium, or POWER5 if the architecture makes a difference. I don't have a recommendation for a particular one, it's been too long (1998) since I've used one with Linux. However, I'd like to see a comparison between Emulex and Qlogic and a winner chosen. We've had some apparent driver issues with a client running Emulex on Linux, even using many different versions of the kernel. - Luke From pgsql-performance-owner@postgresql.org Fri Jul 29 01:12:30 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E617652825 for ; Fri, 29 Jul 2005 01:12:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 17890-01 for ; Fri, 29 Jul 2005 04:12:28 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.202]) by svr1.postgresql.org (Postfix) with ESMTP id A3D0C52816 for ; Fri, 29 Jul 2005 01:12:26 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id z6so340693nzd for ; Thu, 28 Jul 2005 21:12:26 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:references; b=msI83EWGX0bxbMwMVT/loWZGAPMC1Dy8MfnswBjdR61ko5EDGBND1Jfst2aSiEtGaq1xeqzGhu9o6gqjpodhw8NzScR80oHQutMqkn4SrJHT47polFp98hZYgh2SkPC3om9WDiVSvaBy0qAR4Qrf1PTjdrhOoiwOgDFccvw9rsY= Received: by 10.36.252.6 with SMTP id z6mr1565962nzh; Thu, 28 Jul 2005 21:11:31 -0700 (PDT) Received: by 10.36.119.19 with HTTP; Thu, 28 Jul 2005 21:11:31 -0700 (PDT) Message-ID: Date: Fri, 29 Jul 2005 09:41:31 +0530 From: Gnanavel S Reply-To: Gnanavel S To: karim.nassar@acm.org Subject: Re: Two queries are better than one? Cc: pgsql-performance@postgresql.org In-Reply-To: <1122591866.11860.11.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_8088_21102920.1122610291112" References: <1122591866.11860.11.camel@localhost.localdomain> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.414 tagged_above=0 required=5 tests=AWL, HTML_40_50, HTML_MESSAGE, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/457 X-Sequence-Number: 13698 ------=_Part_8088_21102920.1122610291112 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On 7/29/05, Karim Nassar wrote: >=20 > I ran into a situation today maintaining someone else's code where the > sum time running 2 queries seems to be faster than 1. The original code > was split into two queries. I thought about joining them, but > considering the intelligence of my predecessor, I wanted to test it. >=20 > The question is, which technique is really faster? Is there some hidden > setup cost I don't see with explain analyze? >=20 > Postgres 7.4.7, Redhat AES 3 >=20 > Each query individually: >=20 > test=3D> explain analyze > test-> select * from order WHERE ord_batch=3D'343B' AND ord_id=3D'12-645'= ; > QUERY PLAN >=20 > -------------------------------------------------------------------------= --------------------------------------------------- > Index Scan using order_pkey on order (cost=3D0.00..6.02 rows=3D1 width=3D= 486)=20 > (actual time=3D0.063..0.066 rows=3D1 loops=3D1) > Index Cond: ((ord_batch =3D '343B'::bpchar) AND (ord_id =3D '12-645'::bpc= har)) > Total runtime: 0.172 ms > (3 rows) >=20 >=20 > test=3D> explain analyze > test-> select cli_name from client where cli_code=3D'1837'; > QUERY PLAN >=20 > -------------------------------------------------------------------------= -------------------------------------------- > Index Scan using client_pkey on client (cost=3D0.00..5.98 rows=3D2 width= =3D39)=20 > (actual time=3D0.043..0.047 rows=3D1 loops=3D1) > Index Cond: (cli_code =3D '1837'::bpchar) > Total runtime: 0.112 ms > (3 rows) >=20 > Joined: >=20 > test=3D> explain analyze > test-> SELECT cli_name,order.* > test-> FROM order > test-> JOIN client ON (ord_client =3D cli_code) > test-> WHERE ord_batch=3D'343B' AND ord_id=3D'12-645'; where is the cli_code condition in the above query? QUERY PLAN >=20 > -------------------------------------------------------------------------= --------------------------------------------------------- > Nested Loop (cost=3D0.00..12.00 rows=3D2 width=3D525) (actual time=3D0.12= 0..0.128rows=3D1 loops=3D1) > -> Index Scan using order_pkey on order (cost=3D0.00..6.02 rows=3D1 width= =3D486)=20 > (actual time=3D0.064..0.066 rows=3D1 loops=3D1) > Index Cond: ((ord_batch =3D '343B'::bpchar) AND (ord_id =3D '12-645'::bpc= har)) > -> Index Scan using client_pkey on client (cost=3D0.00..5.98 rows=3D1=20 > width=3D51) (actual time=3D0.023..0.026 rows=3D1 loops=3D1) > Index Cond: ("outer".ord_client =3D client.cli_code) > Total runtime: 0.328 ms > (6 rows) >=20 >=20 > -- > Karim Nassar >=20 >=20 > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend >=20 --=20 with regards, S.Gnanavel Satyam Computer Services Ltd. ------=_Part_8088_21102920.1122610291112 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline

On 7/29/05, Karim Nassar <karim.= nassar@acm.org> wrote:
I ran into a situation today maintaining someone else's code where the
s= um time running 2 queries seems to be faster than 1. The original code
w= as split into two queries. I thought about joining them, but
considering= the intelligence of my predecessor, I wanted to test it.

The question is, which technique is really faster? Is there some hi= dden
setup cost I don't see with explain analyze?

Postgres 7.4.7,= Redhat AES 3

Each query individually:

test=3D> explain an= alyze
test-> select * from order  WHERE ord_batch=3D'343B' AND o= rd_id=3D'12-645';
         =             &nb= sp;            =             &nb= sp;          QUERY PLAN
-------------------------------------------------------------= ---------------------------------------------------------------
Index S= can using order_pkey on order  (cost=3D0.00..6.02 rows=3D1 width= =3D486) (actual time=3D 0.063..0.066 rows=3D1 loops=3D1)
   Index Cond: ((ord_batch = =3D '343B'::bpchar) AND (ord_id =3D '12-645'::bpchar))
Total runtime: 0= .172 ms
(3 rows)


test=3D> explain analyze
test->&nbs= p;    select cli_name from client where cli_code=3D'1837';
            = ;            &n= bsp;            = ;            &n= bsp;   QUERY PLAN
-------------------------------------------------------------= --------------------------------------------------------
Index Scan usi= ng client_pkey on client  (cost=3D0.00..5.98 rows=3D2 width=3D39)= (actual time=3D 0.043..0.047 rows=3D1 loops=3D1)
   Index Cond: (cli_code =3D = '1837'::bpchar)
Total runtime: 0.112 ms
(3 rows)

Joined:
<= br>test=3D> explain analyze
test->    SELECT c= li_name,order.*
test->        = ;       FROM order
test->          &n= bsp;    JOIN client ON (ord_client =3D cli_code)
test->   &nbs= p;          WHERE ord_batch=3D'343B' AND ord_id=3D'12-645';

where is the cli_code condition in the above query?
 

&= nbsp;           &nbs= p;            &= nbsp;           &nbs= p;            &= nbsp;         QUERY PLAN
-------------------------------------------------------------------= ---------------------------------------------------------------
Nested = Loop  (cost=3D0.00..12.00 rows=3D2 width=3D525) (actual time=3D0.= 120..0.128 rows=3D1 loops=3D1)
   ->  Index Scan using order_pkey on order  (cost=3D0.00..6.02 rows=3D1 width=3D486) (actual time=3D0.064..0.066 rows=3D1 loops=3D1)
     &n= bsp;   Index Cond: ((ord_batch =3D '343B'::bpchar) AND (ord_id = =3D '12-645'::bpchar))
   ->  Index Scan using client_pkey on client  (cost=3D0.00..5.98 rows=3D1 width=3D51) (actual time=3D0.023..0.026 rows=3D1 loops=3D1)
     &n= bsp;   Index Cond: ("outer".ord_client =3D client.cli_c= ode)
Total runtime: 0.328 ms
(6 rows)


--
Karim Nassar = <karim.nassar@acm.org >


---------------------------(end of broadcast)----------= -----------------
TIP 6: explain analyze is your friend
=



--
with regards,
S.Gnanavel
Satya= m Computer Services Ltd. ------=_Part_8088_21102920.1122610291112-- From pgsql-performance-owner@postgresql.org Fri Jul 29 01:23:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BC668529B7 for ; Fri, 29 Jul 2005 01:23:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20792-01 for ; Fri, 29 Jul 2005 04:23:24 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id 75B66529B4 for ; Fri, 29 Jul 2005 01:23:23 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id E2A9C35F9DE for ; Fri, 29 Jul 2005 04:23:23 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18957-03; Fri, 29 Jul 2005 04:23:21 +0000 (GMT) Received: from [10.0.0.7] (63-228-143-68.phnx.qwest.net [63.228.143.68]) by veggie-fuel.com (Postfix) with ESMTP id 46CC735F9D6; Fri, 29 Jul 2005 04:23:21 +0000 (GMT) Subject: Re: Two queries are better than one? From: Karim Nassar Reply-To: karim.nassar@acm.org To: Gnanavel S Cc: pgsql-performance@postgresql.org In-Reply-To: References: <1122591866.11860.11.camel@localhost.localdomain> Content-Type: text/plain Date: Thu, 28 Jul 2005 21:21:38 -0700 Message-Id: <1122610899.11862.63.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.029 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/458 X-Sequence-Number: 13699 On Fri, 2005-07-29 at 09:41 +0530, Gnanavel S wrote: > > Joined: > > test=> explain analyze > test-> SELECT cli_name,order.* > test-> FROM order > test-> JOIN client ON (ord_client = cli_code) > test-> WHERE ord_batch='343B' AND > ord_id='12-645'; > > where is the cli_code condition in the above query? I don't understand the question. ord_client is the client code, and cli_code is the client code, for their respective tables. batch/id is unique, so there is only one record from order, and only one client to associate. Clearer? -- Karim Nassar From pgsql-performance-owner@postgresql.org Fri Jul 29 01:25:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 640CB529B7 for ; Fri, 29 Jul 2005 01:25:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18999-06 for ; Fri, 29 Jul 2005 04:25:05 +0000 (GMT) Received: from mailbox.samurai.com (mailbox.samurai.com [205.207.28.82]) by svr1.postgresql.org (Postfix) with ESMTP id 31704529B4 for ; Fri, 29 Jul 2005 01:25:02 -0300 (ADT) Received: from localhost (mailbox.samurai.com [205.207.28.82]) by mailbox.samurai.com (Postfix) with ESMTP id 9A7602394CF; Fri, 29 Jul 2005 00:25:02 -0400 (EDT) Received: from mailbox.samurai.com ([205.207.28.82]) by localhost (mailbox.samurai.com [205.207.28.82]) (amavisd-new, port 10024) with LMTP id 83867-01-4; Fri, 29 Jul 2005 00:25:01 -0400 (EDT) Received: from [61.88.101.19] (unknown [61.88.101.19]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by mailbox.samurai.com (Postfix) with ESMTP id EC129239519; Fri, 29 Jul 2005 00:24:59 -0400 (EDT) Message-ID: <42E9AFF5.50209@samurai.com> Date: Fri, 29 Jul 2005 14:26:29 +1000 From: Neil Conway User-Agent: Debian Thunderbird 1.0.2 (X11/20050602) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Gnanavel S Cc: Chris Travers , pgsql-performance@postgresql.org Subject: Re: Left joining against two empty tables makes a query References: <42E86A03.1070803@travelamericas.com> <42E90E19.3030702@travelamericas.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at mailbox.samurai.com X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.016 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/459 X-Sequence-Number: 13700 Gnanavel S wrote: > reindex the tables separately. Reindexing should not affect this problem, anyway. -Neil From pgsql-performance-owner@postgresql.org Fri Jul 29 01:39:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E824552B4F for ; Fri, 29 Jul 2005 01:39:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 18912-10 for ; Fri, 29 Jul 2005 04:39:26 +0000 (GMT) Received: from zproxy.gmail.com (zproxy.gmail.com [64.233.162.194]) by svr1.postgresql.org (Postfix) with ESMTP id 53DEE52B4A for ; Fri, 29 Jul 2005 01:39:23 -0300 (ADT) Received: by zproxy.gmail.com with SMTP id z6so343758nzd for ; Thu, 28 Jul 2005 21:39:23 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:references; b=C1JPfgFGU/zmYZ0Hbmgw0IhfIeR52GFGPBNzamg/bvYTDNzb5Ly7PKIr270oqYGHF8YOVguYkn6HCnE3JFUgEJUdzaPpqrqOHN2lxUjlz1X/Lql5u3Ba/GBELbtq8muWZ2WZebmp6ia7HXFQy9lHLYw4+6so4/87CSqBOhUvyok= Received: by 10.37.22.3 with SMTP id z3mr2408679nzi; Thu, 28 Jul 2005 21:38:49 -0700 (PDT) Received: by 10.36.119.19 with HTTP; Thu, 28 Jul 2005 21:38:49 -0700 (PDT) Message-ID: Date: Fri, 29 Jul 2005 10:08:49 +0530 From: Gnanavel S Reply-To: Gnanavel S To: karim.nassar@acm.org Subject: Re: Two queries are better than one? Cc: pgsql-performance@postgresql.org In-Reply-To: <1122610899.11862.63.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_8206_18523250.1122611929124" References: <1122591866.11860.11.camel@localhost.localdomain> <1122610899.11862.63.camel@localhost.localdomain> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.403 tagged_above=0 required=5 tests=AWL, HTML_50_60, HTML_MESSAGE, RCVD_BY_IP X-Spam-Level: X-Archive-Number: 200507/460 X-Sequence-Number: 13701 ------=_Part_8206_18523250.1122611929124 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On 7/29/05, Karim Nassar wrote: >=20 > On Fri, 2005-07-29 at 09:41 +0530, Gnanavel S wrote: >=20 > > > > Joined: > > > > test=3D> explain analyze > > test-> SELECT cli_name,order.* > > test-> FROM order > > test-> JOIN client ON (ord_client =3D cli_code) > > test-> WHERE ord_batch=3D'343B' AND > > ord_id=3D'12-645'; > > > > where is the cli_code condition in the above query? >=20 > I don't understand the question. ord_client is the client code, and > cli_code is the client code, for their respective tables. batch/id is > unique, so there is only one record from order, and only one client to > associate. >=20 > Clearer? ok. Reason might be comparing with a literal value (previous case) is cheaper= =20 than comparing with column(as it has to be evaluated). But with the previou= s=20 case getting and assigning the cli_code in the application and executing in= =20 db will be time consuming as it includes IPC cost. -- > Karim Nassar >=20 >=20 --=20 with regards, S.Gnanavel Satyam Computer Services Ltd. ------=_Part_8206_18523250.1122611929124 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline

On 7/29/05, Karim Nassar <karim.= nassar@acm.org> wrote:
On Fri, 2005-07-29 at 09:41 +0530, Gnanavel S wrote:

>
>&nb= sp;        Joined:
>
> =         test=3D> explain analyze
&= gt;         test->  &n= bsp; SELECT cli_name,order.*
>     &nbs= p;   test->           =     FROM order
>         test->           =     JOIN client ON (ord_client =3D cli_code)
>    &nb= sp;    test->           =    WHERE ord_batch=3D'343B' AND
>       &nb= sp; ord_id=3D'12-645';
>
> where is the cli_code condition in t= he above query?

I don't understand the question. ord_client is the c= lient code, and
cli_code is the client code, for their respective tables= . batch/id is
unique, so there is only one record from order, and only one client to<= br>associate.

Clearer?

ok.

 Reason might be comparing with a literal value (previous case) is cheaper than comparing with column(as it has to be evaluated).  But with the previous case getting and assigning the cli_code in the application and executing in db will be time consuming as it includes IPC cost.

--
Karim= Nassar <karim.nassar@acm.org >




--
with regard= s,
S.Gnanavel
Satyam Computer Services Ltd. ------=_Part_8206_18523250.1122611929124-- From pgsql-performance-owner@postgresql.org Fri Jul 29 02:10:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DF14B5283C for ; Fri, 29 Jul 2005 02:10:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28131-05 for ; Fri, 29 Jul 2005 05:10:24 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id DF9EA52831 for ; Fri, 29 Jul 2005 02:10:20 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6T5NYQO005229 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Thu, 28 Jul 2005 21:23:35 -0800 Message-ID: <42E9BA2F.1090900@aptalaska.net> Date: Thu, 28 Jul 2005 21:10:07 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Gavin Sherry Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> In-Reply-To: X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/461 X-Sequence-Number: 13702 Gavin Sherry wrote: > > I had a look at your data -- thanks. > > I have a question though: put_token() is invoked 120596 times in your > benchmark... for 616 messages. That's nearly 200 queries (not even > counting the 1-8 (??) inside the function itself) per message. Something > doesn't seem right there.... > > Gavin I am pretty sure that's right because it is doing word statistics on email messages. I need to spend some time studying the code, I just haven't found time yet. Would it be safe to say that there isn't any glaring performance penalties other than the sheer volume of queries? Thanks, schu From pgsql-performance-owner@postgresql.org Fri Jul 29 02:23:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 3932E52968 for ; Fri, 29 Jul 2005 02:23:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 31920-10 for ; Fri, 29 Jul 2005 05:23:17 +0000 (GMT) Received: from mail.travelamericas.com (unknown [206.130.134.147]) by svr1.postgresql.org (Postfix) with SMTP id 6C191528D2 for ; Fri, 29 Jul 2005 02:23:16 -0300 (ADT) Received: (qmail 12915 invoked from network); 29 Jul 2005 05:23:16 -0000 Received: from unknown (HELO ?10.0.0.253?) (10.0.0.253) by verkiel.travelamericas.com with SMTP; 29 Jul 2005 05:23:16 -0000 Message-ID: <42E9BD43.6060203@travelamericas.com> Date: Thu, 28 Jul 2005 22:23:15 -0700 From: Chris Travers User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Gnanavel S Cc: Chris Travers , pgsql-performance@postgresql.org Subject: Re: Left joining against two empty tables makes a query References: <42E86A03.1070803@travelamericas.com> <42E90E19.3030702@travelamericas.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.037 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/462 X-Sequence-Number: 13703 > > Secondly, the project table has *never* had anything in it. So where > are these numbers coming from? > > > pg_statistics I very much doubt that. I was unable to locate any rows in pg_statistic where the pg_class.oid for either table matched any row's starelid. Tom's argument that this is behavior by design makes sense. I assumed that something like that had to be going on, otherwise there would be nowhere for the numbers to come from. I.e. if there never were any rows in the table, then if pg_statistic is showing 1060 rows, we have bigger problems than a bad query plan. I hope however that eventually tables which are truly empty can be treated intelligently sometime in the future in Left Joins. Otherwise this limits the usefulness of out of the box solutions which may have functionality that we don't use. Such solutions can then kill the database performance quite easily. Chris Travers Metatron Technology Consulting From pgsql-performance-owner@postgresql.org Fri Jul 29 02:50:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id BE56552825 for ; Fri, 29 Jul 2005 02:50:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40532-02 for ; Fri, 29 Jul 2005 05:50:15 +0000 (GMT) Received: from mail1.catalyst.net.nz (godel.catalyst.net.nz [202.78.240.40]) by svr1.postgresql.org (Postfix) with ESMTP id 79D035282E for ; Fri, 29 Jul 2005 02:50:13 -0300 (ADT) Received: from leibniz.catalyst.net.nz ([202.78.240.7] helo=lamb.mcmillan.net.nz) by mail1.catalyst.net.nz with esmtpsa (TLS-1.0:DHE_RSA_AES_256_CBC_SHA:32) (Exim 4.50) id 1DyNl6-0004VQ-Ho; Fri, 29 Jul 2005 17:50:12 +1200 Received: from lamb.mcmillan.net.nz (lamb.mcmillan.net.nz [127.0.0.1]) by lamb.mcmillan.net.nz (Postfix) with ESMTP id 2200D5B8006; Fri, 29 Jul 2005 17:50:12 +1200 (NZST) Subject: Re: Performance problems testing with Spamassassin 3.1.0 From: Andrew McMillan To: Matthew Schumacher Cc: pgsql-performance@postgresql.org In-Reply-To: <42E9749F.6000709@aptalaska.net> References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="=-91YWh4BVVTMDV4QRW/zg" Date: Fri, 29 Jul 2005 17:50:11 +1200 Message-Id: <1122616211.5691.340.camel@lamb.mcmillan.net.nz> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/463 X-Sequence-Number: 13704 --=-91YWh4BVVTMDV4QRW/zg Content-Type: text/plain Content-Transfer-Encoding: quoted-printable On Thu, 2005-07-28 at 16:13 -0800, Matthew Schumacher wrote: >=20 > Ok, I finally got some test data together so that others can test > without installing SA. >=20 > The schema and test dataset is over at > http://www.aptalaska.net/~matt.s/bayes/bayesBenchmark.tar.gz >=20 > I have a pretty fast machine with a tuned postgres and it takes it about > 2 minutes 30 seconds to load the test data. Since the test data is the > bayes information on 616 spam messages than comes out to be about 250ms > per message. While that is doable, it does add quite a bit of overhead > to the email system. On my laptop this takes: real 1m33.758s user 0m4.285s sys 0m1.181s One interesting effect is the data in bayes_vars has a huge number of updates and needs vacuum _frequently_. After the run a vacuum full compacts it down from 461 pages to 1 page. Regards, Andrew. ------------------------------------------------------------------------- Andrew @ Catalyst .Net .NZ Ltd, PO Box 11-053, Manners St, Wellington WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St DDI: +64(4)803-2201 MOB: +64(272)DEBIAN OFFICE: +64(4)499-2267 I don't do it for the money. -- Donald Trump, Art of the Deal ------------------------------------------------------------------------- --=-91YWh4BVVTMDV4QRW/zg Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQBC6cOTjJA0f48GgBIRAoDGAJ46aWRNT5DpkFeA0vPBOfn6CPcesACeLj4T v51QgQ+CZI5hJVrtZsm39IY= =VAB8 -----END PGP SIGNATURE----- --=-91YWh4BVVTMDV4QRW/zg-- From pgsql-performance-owner@postgresql.org Fri Jul 29 02:58:43 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E3B425282E for ; Fri, 29 Jul 2005 02:58:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 40064-07 for ; Fri, 29 Jul 2005 05:58:32 +0000 (GMT) Received: from linuxworld.com.au (unknown [203.34.46.50]) by svr1.postgresql.org (Postfix) with ESMTP id 249BE52825 for ; Fri, 29 Jul 2005 02:58:30 -0300 (ADT) Received: from linuxworld.com.au (IDENT:swm@localhost.localdomain [127.0.0.1]) by linuxworld.com.au (8.13.2/8.13.2) with ESMTP id j6T5wOss011092; Fri, 29 Jul 2005 15:58:24 +1000 Received: from localhost (swm@localhost) by linuxworld.com.au (8.13.2/8.13.2/Submit) with ESMTP id j6T5wOfv011089; Fri, 29 Jul 2005 15:58:24 +1000 X-Authentication-Warning: linuxworld.com.au: swm owned process doing -bs Date: Fri, 29 Jul 2005 15:58:24 +1000 (EST) From: Gavin Sherry X-X-Sender: swm@linuxworld.com.au To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 In-Reply-To: <42E9BA2F.1090900@aptalaska.net> Message-ID: References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <42E9BA2F.1090900@aptalaska.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.002 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/464 X-Sequence-Number: 13705 zOn Thu, 28 Jul 2005, Matthew Schumacher wrote: > Gavin Sherry wrote: > > > > > I had a look at your data -- thanks. > > > > I have a question though: put_token() is invoked 120596 times in your > > benchmark... for 616 messages. That's nearly 200 queries (not even > > counting the 1-8 (??) inside the function itself) per message. Something > > doesn't seem right there.... > > > > Gavin > > I am pretty sure that's right because it is doing word statistics on > email messages. > > I need to spend some time studying the code, I just haven't found time yet. > > Would it be safe to say that there isn't any glaring performance > penalties other than the sheer volume of queries? Well, everything relating to one message should be issued in a transaction block. Secondly, the initial select may be unnecessary -- I haven't looked at the logic that closely. There is, potentially, some parser overhead. In C, you could get around this with PQprepare() et al. It would also be interesting to look at the cost of a C function. Gavin From pgsql-performance-owner@postgresql.org Fri Jul 29 03:49:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A04AE52ABE for ; Fri, 29 Jul 2005 03:49:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 49971-09 for ; Fri, 29 Jul 2005 06:48:51 +0000 (GMT) Received: from zigo.dhs.org (ua-83-227-204-174.cust.bredbandsbolaget.se [83.227.204.174]) by svr1.postgresql.org (Postfix) with ESMTP id 0041152815 for ; Fri, 29 Jul 2005 03:48:50 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 6CDC78467; Fri, 29 Jul 2005 08:48:48 +0200 (CEST) Date: Fri, 29 Jul 2005 08:48:48 +0200 (CEST) From: Dennis Bjorklund To: Matthew Schumacher Cc: josh@agliodbs.com, Subject: Re: Performance problems testing with Spamassassin 3.1.0 In-Reply-To: <42E83C09.2040508@aptalaska.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.422 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/465 X-Sequence-Number: 13706 On Wed, 27 Jul 2005, Matthew Schumacher wrote: > Then they do this to insert the token: > > INSERT INTO bayes_token ( > id, > token, > spam_count, > ham_count, > atime > ) VALUES ( > ?, > ?, > ?, > ?, > ? > ) ON DUPLICATE KEY > UPDATE > spam_count = GREATEST(spam_count + ?, 0), > ham_count = GREATEST(ham_count + ?, 0), > atime = GREATEST(atime, ?) > > Or update the token: > > UPDATE bayes_vars SET > $token_count_update > newest_token_age = GREATEST(newest_token_age, ?), > oldest_token_age = LEAST(oldest_token_age, ?) > WHERE id = ? > > > I think the reason why the procedure was written for postgres was > because of the greatest and least statements performing poorly. How can they perform poorly when they are dead simple? Here are 2 functions that work for the above cases of greatest: CREATE FUNCTION greatest_int (integer, integer) RETURNS INTEGER IMMUTABLE STRICT AS 'SELECT CASE WHEN $1 < $2 THEN $2 ELSE $1 END;' LANGUAGE SQL; CREATE FUNCTION least_int (integer, integer) RETURNS INTEGER IMMUTABLE STRICT AS 'SELECT CASE WHEN $1 < $2 THEN $1 ELSE $2 END;' LANGUAGE SQL; and these should be inlined by pg and very fast to execute. I wrote a function that should do what the insert above does. The update I've not looked at (I don't know what $token_count_update is) but the update looks simple enough to just implement the same way in pg as in mysql. For the insert or replace case you can probably use this function: CREATE FUNCTION insert_or_update_token (xid INTEGER, xtoken BYTEA, xspam_count INTEGER, xham_count INTEGER, xatime INTEGER) RETURNS VOID AS $$ BEGIN LOOP UPDATE bayes_token SET spam_count = greatest_int (spam_count + xspam_count, 0), ham_count = greatest_int (ham_count + xham_count, 0), atime = greatest_int (atime, xatime) WHERE id = xid AND token = xtoken; IF found THEN RETURN; END IF; BEGIN INSERT INTO bayes_token VALUES (xid, xtoken, xspam_count, xham_count, xatime); RETURN; EXCEPTION WHEN unique_violation THEN -- do nothing END; END LOOP; END; $$ LANGUAGE plpgsql; It's not really tested so I can't tell if it's faster then what you have. What it does do is mimic the way you insert values in mysql. It only work on pg 8.0 and later however since the exception handling was added in 8.0. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Fri Jul 29 04:01:27 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9464B52AC7 for ; Fri, 29 Jul 2005 04:01:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 52624-05 for ; Fri, 29 Jul 2005 07:01:17 +0000 (GMT) Received: from mail.Mi8.com (unknown [63.240.6.42]) by svr1.postgresql.org (Postfix) with ESMTP id 5CDFF52ABE for ; Fri, 29 Jul 2005 04:01:15 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Fri, 29 Jul 2005 03:01:07 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Fri, 29 Jul 2005 03:01:07 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Subject: Re: Performance problems testing with Spamassassin Date: Fri, 29 Jul 2005 03:01:07 -0400 Message-ID: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Performance problems testing with Spamassassin 3.1.0 Thread-Index: AcWUAmWmvDU/LT66SYiCywCdDqFp2wABvk4+ From: "Luke Lonergan" To: "Andrew McMillan" , "Matthew Schumacher" Cc: pgsql-performance@postgresql.org X-OriginalArrivalTime: 29 Jul 2005 07:01:07.0562 (UTC) FILETIME=[4B1C3CA0:01C5940B] X-WSS-ID: 6EF70BB921S3163737-01-01 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/466 X-Sequence-Number: 13707 work_mem =3D 131072 # min 64, size in KB shared_buffers =3D 16000 # min 16, at least max_connections*2, = 8KB each checkpoint_segments =3D 128 # in logfile segments, min 1, 16MB = each effective_cache_size =3D 750000 # typically 8KB each fsync=3Dfalse # turns forced synchronization on or = off =20 ------------------------------------------ On Bizgres (0_7_2) running on a 2GHz Opteron: ------------------------------------------ [llonergan@stinger4 bayesBenchmark]$ ./test.sh real 0m38.348s user 0m1.422s sys 0m1.870s =20 ------------------------------------------ On a 2.4GHz AMD64: ------------------------------------------ [llonergan@kite15 bayesBenchmark]$ ./test.sh real 0m35.497s user 0m2.250s sys 0m0.470s =20 Now we turn fsync=3Dtrue: =20 ------------------------------------------ On a 2.4GHz AMD64: ------------------------------------------ [llonergan@kite15 bayesBenchmark]$ ./test.sh real 2m7.368s user 0m2.560s sys 0m0.750s =20 I guess we see the real culprit here. Anyone surprised it's the WAL? =20 - Luke ________________________________ From: pgsql-performance-owner@postgresql.org on behalf of Andrew = McMillan Sent: Thu 7/28/2005 10:50 PM To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] Performance problems testing with Spamassassin = 3.1.0 On Thu, 2005-07-28 at 16:13 -0800, Matthew Schumacher wrote: > > Ok, I finally got some test data together so that others can test > without installing SA. > > The schema and test dataset is over at > http://www.aptalaska.net/~matt.s/bayes/bayesBenchmark.tar.gz > > I have a pretty fast machine with a tuned postgres and it takes it = about > 2 minutes 30 seconds to load the test data. Since the test data is = the > bayes information on 616 spam messages than comes out to be about = 250ms > per message. While that is doable, it does add quite a bit of = overhead > to the email system. On my laptop this takes: real 1m33.758s user 0m4.285s sys 0m1.181s One interesting effect is the data in bayes_vars has a huge number of updates and needs vacuum _frequently_. After the run a vacuum full compacts it down from 461 pages to 1 page. Regards, Andrew. -------------------------------------------------------------------------= Andrew @ Catalyst .Net .NZ Ltd, PO Box 11-053, Manners St, Wellington WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St DDI: +64(4)803-2201 MOB: +64(272)DEBIAN OFFICE: +64(4)499-2267 I don't do it for the money. -- Donald Trump, Art of the Deal -------------------------------------------------------------------------= From pgsql-performance-owner@postgresql.org Fri Jul 29 09:37:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E5031529B4 for ; Fri, 29 Jul 2005 09:37:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 28595-04 for ; Fri, 29 Jul 2005 12:37:14 +0000 (GMT) Received: from candle.pha.pa.us (candle.pha.pa.us [64.139.89.126]) by svr1.postgresql.org (Postfix) with ESMTP id AAEC9529A2 for ; Fri, 29 Jul 2005 09:37:09 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id j6TCb0A10838; Fri, 29 Jul 2005 08:37:00 -0400 (EDT) From: Bruce Momjian Message-Id: <200507291237.j6TCb0A10838@candle.pha.pa.us> Subject: Re: [PATCHES] COPY FROM performance improvements In-Reply-To: To: Luke Lonergan Date: Fri, 29 Jul 2005 08:37:00 -0400 (EDT) Cc: Mark Wong , "Joshua D. Drake" , Patrick Welche , pgsql-performance@postgresql.org, maryedie@osdl.org X-Mailer: ELM [version 2.4ME+ PL121 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/467 X-Sequence-Number: 13708 Luke Lonergan wrote: > Mark, > > On 7/28/05 4:43 PM, "Mark Wong" wrote: > > > Are there any recommendations for Qlogic controllers on Linux, scsi or > > fiber channel? I might be able to my hands on some. I have pci-x slots > > for AMD, Itanium, or POWER5 if the architecture makes a difference. > > I don't have a recommendation for a particular one, it's been too long > (1998) since I've used one with Linux. However, I'd like to see a > comparison between Emulex and Qlogic and a winner chosen. We've had some > apparent driver issues with a client running Emulex on Linux, even using > many different versions of the kernel. Where is the most recent version of the COPY patch? -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073 From pgsql-bugs-owner@postgresql.org Fri Jul 29 10:11:28 2005 X-Original-To: pgsql-bugs-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0EEDB529BD for ; Fri, 29 Jul 2005 09:56:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 32818-03 for ; Fri, 29 Jul 2005 12:56:49 +0000 (GMT) Received: from svr2.postgresql.org (svr2.postgresql.org [65.19.161.25]) by svr1.postgresql.org (Postfix) with ESMTP id 176DC52998 for ; Fri, 29 Jul 2005 09:56:49 -0300 (ADT) Received: by svr2.postgresql.org (Postfix, from userid 80) id 1166CF0B14; Fri, 29 Jul 2005 13:52:45 +0100 (BST) To: pgsql-bugs@postgresql.org Subject: BUG #1797: Problem using Limit in a function, seqscan From: "Magno Leite" Message-Id: <20050729125245.1166CF0B14@svr2.postgresql.org> Date: Fri, 29 Jul 2005 13:52:45 +0100 (BST) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/248 X-Sequence-Number: 12411 The following bug has been logged online: Bug reference: 1797 Logged by: Magno Leite Email address: magnomilk@yahoo.com.br PostgreSQL version: 8.0 Operating system: Windows XP Professional Edition Description: Problem using Limit in a function, seqscan Details: I looked for about this problem in BUG REPORT but I can't find. This is my problem, when I try to use limit in a function, the Postgre doesn't use my index, then it use sequencial scan. What is the problem ? From pgsql-performance-owner@postgresql.org Fri Jul 29 10:23:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A5B5B529C0 for ; Fri, 29 Jul 2005 10:23:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 38219-08 for ; Fri, 29 Jul 2005 13:23:05 +0000 (GMT) Received: from mr2.surnet.cl (smtp2.surnet.cl [216.155.73.163]) by svr1.postgresql.org (Postfix) with ESMTP id 15D3952939 for ; Fri, 29 Jul 2005 10:23:04 -0300 (ADT) Received: from smtp2.surnet.cl (216.155.73.169) by mr2.surnet.cl (7.0.031.3) id 4259ADFA0191741E; Fri, 29 Jul 2005 09:21:47 -0400 Received: from smtp1.surnet.cl (smtp1.surnet.cl []) by mr2.surnet.cl ([216.155.73.169]); Fri, 29 Jul 2005 13:21:47 +0000 Received: from cluster.surnet.cl (216.155.73.164) by smtp1.surnet.cl (7.0.031.3) id 42587C6E012F051D; Fri, 29 Jul 2005 09:21:38 -0400 Received: from alvh.no-ip.org (200.85.218.66) by cluster.surnet.cl (7.0.043) (authenticated as alvherre@surnet.cl) id 42B3EF60005D5F25; Fri, 29 Jul 2005 09:22:17 -0400 Received: by alvh.no-ip.org (Postfix, from userid 1000) id 7426DC2D450; Fri, 29 Jul 2005 09:23:19 -0400 (CLT) Date: Fri, 29 Jul 2005 09:23:19 -0400 From: Alvaro Herrera To: Luke Lonergan Cc: Andrew McMillan , Matthew Schumacher , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin Message-ID: <20050729132319.GA13680@alvh.no-ip.org> Mail-Followup-To: Luke Lonergan , Andrew McMillan , Matthew Schumacher , pgsql-performance@postgresql.org References: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.427 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, FORGED_RCVD_HELO, RCVD_IN_NJABL_PROXY X-Spam-Level: * X-Archive-Number: 200507/468 X-Sequence-Number: 13709 On Fri, Jul 29, 2005 at 03:01:07AM -0400, Luke Lonergan wrote: > I guess we see the real culprit here. Anyone surprised it's the WAL? So what? Are you planning to suggest people to turn fsync=false? I just had a person lose 3 days of data on some tables because of that, even when checkpoints were 5 minutes apart. With fsync off, there's no work _at all_ going on, not just the WAL -- heap/index file fsync at checkpoint is also skipped. This is no good. -- Alvaro Herrera () "In a specialized industrial society, it would be a disaster to have kids running around loose." (Paul Graham) From pgsql-bugs-owner@postgresql.org Fri Jul 29 11:00:44 2005 X-Original-To: pgsql-bugs-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9EB53529F6 for ; Fri, 29 Jul 2005 11:00:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 48716-05 for ; Fri, 29 Jul 2005 14:00:40 +0000 (GMT) Received: from tigger.fuhr.org (tigger.fuhr.org [63.214.45.158]) by svr1.postgresql.org (Postfix) with ESMTP id D3935529F3 for ; Fri, 29 Jul 2005 11:00:38 -0300 (ADT) Received: from winnie.fuhr.org (winnie.fuhr.org [10.1.0.1]) by tigger.fuhr.org (8.13.3/8.13.3) with ESMTP id j6TE0dhM029609 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Fri, 29 Jul 2005 08:00:42 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: from winnie.fuhr.org (localhost [127.0.0.1]) by winnie.fuhr.org (8.13.3/8.13.3) with ESMTP id j6TE0c4k083725; Fri, 29 Jul 2005 08:00:38 -0600 (MDT) (envelope-from mfuhr@winnie.fuhr.org) Received: (from mfuhr@localhost) by winnie.fuhr.org (8.13.3/8.13.3/Submit) id j6TE0cNO083724; Fri, 29 Jul 2005 08:00:38 -0600 (MDT) (envelope-from mfuhr) Date: Fri, 29 Jul 2005 08:00:38 -0600 From: Michael Fuhr To: Magno Leite Cc: pgsql-bugs@postgresql.org Subject: Re: BUG #1797: Problem using Limit in a function, seqscan Message-ID: <20050729140038.GA83653@winnie.fuhr.org> References: <20050729125245.1166CF0B14@svr2.postgresql.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050729125245.1166CF0B14@svr2.postgresql.org> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/249 X-Sequence-Number: 12412 On Fri, Jul 29, 2005 at 01:52:45PM +0100, Magno Leite wrote: > I looked for about this problem in BUG REPORT but I can't find. This is my > problem, when I try to use limit in a function, the Postgre doesn't use my > index, then it use sequencial scan. What is the problem ? Without more information we can only guess, but if you're using PL/pgSQL then a cached query plan might be responsible. Here's an excerpt from the PREPARE documentation: In some situations, the query plan produced for a prepared statement will be inferior to the query plan that would have been chosen if the statement had been submitted and executed normally. This is because when the statement is planned and the planner attempts to determine the optimal query plan, the actual values of any parameters specified in the statement are unavailable. PostgreSQL collects statistics on the distribution of data in the table, and can use constant values in a statement to make guesses about the likely result of executing the statement. Since this data is unavailable when planning prepared statements with parameters, the chosen plan may be suboptimal. If you'd like us to take a closer look, then please post a self- contained example, i.e., all SQL statements that somebody could load into an empty database to reproduce the behavior you're seeing. -- Michael Fuhr http://www.fuhr.org/~mfuhr/ From pgsql-bugs-owner@postgresql.org Fri Jul 29 11:05:44 2005 X-Original-To: pgsql-bugs-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 00B8052A1D for ; Fri, 29 Jul 2005 11:05:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 47813-08 for ; Fri, 29 Jul 2005 14:05:35 +0000 (GMT) Received: from wolff.to (wolff.to [66.93.197.194]) by svr1.postgresql.org (Postfix) with SMTP id 3CCF3529F1 for ; Fri, 29 Jul 2005 11:05:34 -0300 (ADT) Received: (qmail 23844 invoked by uid 500); 29 Jul 2005 14:06:42 -0000 Date: Fri, 29 Jul 2005 09:06:42 -0500 From: Bruno Wolff III To: Magno Leite Cc: pgsql-bugs@postgresql.org, pgsql-performance@postgresql.org Subject: Re: BUG #1797: Problem using Limit in a function, seqscan Message-ID: <20050729140642.GA23650@wolff.to> Mail-Followup-To: pgsql-performance@postgresql.org, Bruno Wolff III , Magno Leite References: <20050729125245.1166CF0B14@svr2.postgresql.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050729125245.1166CF0B14@svr2.postgresql.org> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/250 X-Sequence-Number: 12413 On Fri, Jul 29, 2005 at 13:52:45 +0100, Magno Leite wrote: > > Description: Problem using Limit in a function, seqscan > > I looked for about this problem in BUG REPORT but I can't find. This is my > problem, when I try to use limit in a function, the Postgre doesn't use my > index, then it use sequencial scan. What is the problem ? You haven't described the problem well enough to allow us to help you and you posted it to the wrong list. This should be discussed on the performance list, not the bug list. It would help if you showed us the query you are running and run it outside of the function with EXPLAIN ANALYSE and show us that output. Depending on what that output shows, we may ask you other questions. From pgsql-performance-owner@postgresql.org Fri Jul 29 11:12:58 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id A4011529A2 for ; Fri, 29 Jul 2005 11:12:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 47206-10 for ; Fri, 29 Jul 2005 14:12:54 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 44E18529A5 for ; Fri, 29 Jul 2005 11:12:53 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6TECkuX016615; Fri, 29 Jul 2005 10:12:47 -0400 (EDT) To: "Luke Lonergan" Cc: "Andrew McMillan" , "Matthew Schumacher" , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin In-reply-to: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> References: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> Comments: In-reply-to "Luke Lonergan" message dated "Fri, 29 Jul 2005 03:01:07 -0400" Date: Fri, 29 Jul 2005 10:12:46 -0400 Message-ID: <16614.1122646366@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/470 X-Sequence-Number: 13711 "Luke Lonergan" writes: > I guess we see the real culprit here. Anyone surprised it's the WAL? You have not proved that at all. I haven't had time to look at Matthew's problem, but someone upthread implied that it was doing a separate transaction for each word. If so, collapsing that to something more reasonable (say one xact per message) would probably help a great deal. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Jul 29 13:45:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 309E152A23 for ; Fri, 29 Jul 2005 13:45:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 87426-03 for ; Fri, 29 Jul 2005 16:45:15 +0000 (GMT) Received: from mail2.aeccom.com (port-212-202-101-158.static.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id 42B6A52A01 for ; Fri, 29 Jul 2005 13:45:13 -0300 (ADT) Received: from [192.168.2.12] (port-83-236-156-26.static.qsc.de [83.236.156.26]) by mail2.aeccom.com (Postfix) with ESMTP id DA7AE10B; Fri, 29 Jul 2005 18:45:12 +0200 (CEST) Message-ID: <42EA5D18.9020204@aeccom.com> Date: Fri, 29 Jul 2005 18:45:12 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH User-Agent: Mozilla Thunderbird 1.0.6 (X11/20050716) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Cc: Sven Geisler Subject: Performance problems on 4/8way Opteron (dualcore) HP DL585 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.063 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/471 X-Sequence-Number: 13712 Hi, does anybody have expierence with this machine (4x 875 dual core Opteron CPUs)? We run RHEL 3.0, 32bit and under high load it is a drag. We mostly run memory demanding queries. Context switches are pretty much around 20.000 on the average, no cs spikes when we run many processes in parallel. Actually we only see two processes in running state! When there are only a few processes running context switches go much higher. At the moment we are much slower that with a 4way XEON box (DL580). We are running 8.0.3 compiled with -mathlon flags. Regards, Dirk From pgsql-performance-owner@postgresql.org Fri Jul 29 13:46:22 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5A2AF52A23 for ; Fri, 29 Jul 2005 13:46:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 88356-01 for ; Fri, 29 Jul 2005 16:46:17 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 654D8529F3 for ; Fri, 29 Jul 2005 13:46:11 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [63.195.55.98] (account josh@agliodbs.com HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7670117; Fri, 29 Jul 2005 09:48:24 -0700 From: Josh Berkus Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin Date: Fri, 29 Jul 2005 09:47:03 -0700 User-Agent: KMail/1.8 Cc: "Luke Lonergan" , "Andrew McMillan" , "Matthew Schumacher" References: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> In-Reply-To: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507290947.03700.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.012 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/472 X-Sequence-Number: 13713 Luke, > work_mem = 131072 # min 64, size in KB Incidentally, this is much too high for an OLTP application, although I don't think this would have affected the test. > shared_buffers = 16000 # min 16, at least max_connections*2, 8KB > each checkpoint_segments = 128 # in logfile segments, min 1, 16MB > each effective_cache_size = 750000 # typically 8KB each > fsync=false # turns forced synchronization on or off Try changing: wal_buffers = 256 and try Bruce's stop full_page_writes patch. > I guess we see the real culprit here. Anyone surprised it's the WAL? Nope. On high-end OLTP stuff, it's crucial that the WAL have its own dedicated disk resource. Also, running a complex stored procedure for each and every word in each e-mail is rather deadly ... with the e-mail traffic our server at Globix receives, for example, that would amount to running it about 1,000 times a minute. It would be far better to batch this, somehow, maybe using temp tables. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Jul 29 13:50:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id B1DE75281A for ; Fri, 29 Jul 2005 13:50:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 86050-10 for ; Fri, 29 Jul 2005 16:49:54 +0000 (GMT) Received: from mail1.fluidhosting.com (mail1.fluidhosting.com [204.14.90.61]) by svr1.postgresql.org (Postfix) with SMTP id 584A852B1D for ; Fri, 29 Jul 2005 13:49:53 -0300 (ADT) Received: (qmail 8574 invoked by uid 399); 29 Jul 2005 16:49:52 -0000 Received: from 194-185-112-82.f5.ngi.it (HELO ?10.20.20.4?) (194.185.112.82) by mail1.fluidhosting.com with SMTP; 29 Jul 2005 16:49:52 -0000 Message-ID: <42EA5E64.6050209@beccati.com> Date: Fri, 29 Jul 2005 18:50:44 +0200 From: Matteo Beccati User-Agent: Mozilla Thunderbird 1.0.2 (Windows/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: ltree <@ operator selectivity causes very slow plan Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/473 X-Sequence-Number: 13714 Hi, I'm happily using ltree since a long time, but I'm recently having troubles because of ltree <@ operator selectivity that is causing very bad planner choices. An example of slow query is: SELECT batch_id, b.t_stamp AS t_stamp, objects, CASE WHEN sent IS NULL THEN gw_batch_sent(b.batch_id) ELSE sent END AS sent FROM gw_users u JOIN gw_batches b USING (u_id) WHERE u.tree <@ '1041' AND b.t_stamp >= 'today'::date - '7 days'::interval AND b.t_stamp < 'today' ORDER BY t_stamp DESC; I've posted the EXPLAIN ANALYZE output here for better readability: http://rafb.net/paste/results/NrCDMs50.html As you may see, disabling nested loops makes the query lightning fast. The problem is caused by the fact that most of the records of gw_users match the "u.tree <@ '1041'" condition: SELECT COUNT(*) FROM gw_users; count ------- 5012 SELECT COUNT(*) FROM gw_users WHERE tree <@ '1041'; count ------- 4684 Is there anything I can do apart from disabling nested loops? Best regards -- Matteo Beccati http://phpadsnew.com http://phppgads.com From pgsql-performance-owner@postgresql.org Fri Jul 29 14:11:24 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6E32B52B51 for ; Fri, 29 Jul 2005 14:11:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 91578-10 for ; Fri, 29 Jul 2005 17:11:20 +0000 (GMT) Received: from mail.Mi8.com (unknown [63.240.6.42]) by svr1.postgresql.org (Postfix) with ESMTP id 5E44852A23 for ; Fri, 29 Jul 2005 14:11:19 -0300 (ADT) Received: from 172.16.1.110 by mail.Mi8.com with ESMTP (- GW01 Welcome to Mi8 Corporation www.Mi8.com); Fri, 29 Jul 2005 13:11:10 -0400 X-Server-Uuid: F1A2E19A-84E4-48DD-8F48-B475613F58B2 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP01.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Fri, 29 Jul 2005 13:11:11 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Fri, 29 Jul 2005 13:11:10 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 29 Jul 2005 10:11:10 -0700 Subject: Re: Performance problems testing with Spamassassin From: "Luke Lonergan" To: "Alvaro Herrera" Cc: "Andrew McMillan" , "Matthew Schumacher" , pgsql-performance@postgresql.org Message-ID: In-Reply-To: <20050729132319.GA13680@alvh.no-ip.org> MIME-Version: 1.0 X-OriginalArrivalTime: 29 Jul 2005 17:11:11.0273 (UTC) FILETIME=[849F4990:01C59460] X-WSS-ID: 6EF4BCA421S3436249-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.552 tagged_above=0 required=5 tests=AWL, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/474 X-Sequence-Number: 13715 Alvaro, On 7/29/05 6:23 AM, "Alvaro Herrera" wrote: > On Fri, Jul 29, 2005 at 03:01:07AM -0400, Luke Lonergan wrote: > >> I guess we see the real culprit here. Anyone surprised it's the WAL? > > So what? Are you planning to suggest people to turn fsync=false? That's not the conclusion I made, no. I was pointing out that fsync has a HUGE impact on his problem, which implies something to do with the I/O sync operations. Black box bottleneck hunt approach #12. > With fsync off, there's no > work _at all_ going on, not just the WAL -- heap/index file fsync at > checkpoint is also skipped. This is no good. OK - so that's what Tom is pointing out, that fsync impacts more than WAL. However, finding out that fsync/no fsync makes a 400% difference in speed for this problem is interesting and relevant, no? - Luke From pgsql-performance-owner@postgresql.org Fri Jul 29 14:14:00 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id CCA6352936 for ; Fri, 29 Jul 2005 14:13:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92264-06 for ; Fri, 29 Jul 2005 17:13:52 +0000 (GMT) Received: from mail.Mi8.com (d01gw01.mi8.com [63.240.6.47]) by svr1.postgresql.org (Postfix) with ESMTP id 2C23A5282C for ; Fri, 29 Jul 2005 14:13:51 -0300 (ADT) Received: from 172.16.1.112 by mail.Mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D1)); Fri, 29 Jul 2005 13:13:41 -0400 X-Server-Uuid: 241911D6-425B-44B9-A073-E3FE0F8FC774 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by D01SMTP02.Mi8.com with Microsoft SMTPSVC(6.0.3790.211); Fri, 29 Jul 2005 13:13:41 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Fri, 29 Jul 2005 13:13:41 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 29 Jul 2005 10:13:41 -0700 Subject: Re: Performance problems testing with Spamassassin From: "Luke Lonergan" To: "Tom Lane" Cc: "Andrew McMillan" , "Matthew Schumacher" , pgsql-performance@postgresql.org Message-ID: In-Reply-To: <16614.1122646366@sss.pgh.pa.us> MIME-Version: 1.0 X-OriginalArrivalTime: 29 Jul 2005 17:13:41.0685 (UTC) FILETIME=[DE465650:01C59460] X-WSS-ID: 6EF4BC4F2B48612553-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.577 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/475 X-Sequence-Number: 13716 Tom, On 7/29/05 7:12 AM, "Tom Lane" wrote: > "Luke Lonergan" writes: >> I guess we see the real culprit here. Anyone surprised it's the WAL? > > You have not proved that at all. As Alvaro pointed out, fsync has impact on more than WAL, so good point. Interesting that fsync has such a huge impact on this situation though. - Luke From pgsql-performance-owner@postgresql.org Fri Jul 29 14:38:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6558452A20 for ; Fri, 29 Jul 2005 14:38:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 96143-07 for ; Fri, 29 Jul 2005 17:38:23 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 7802E52A23 for ; Fri, 29 Jul 2005 14:38:22 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6THpFCp012026 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 29 Jul 2005 09:51:15 -0800 Message-ID: <42EA6966.8090402@aptalaska.net> Date: Fri, 29 Jul 2005 09:37:42 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Andrew McMillan Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> In-Reply-To: <1122616211.5691.340.camel@lamb.mcmillan.net.nz> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/477 X-Sequence-Number: 13718 Andrew McMillan wrote: > On Thu, 2005-07-28 at 16:13 -0800, Matthew Schumacher wrote: > >>Ok, I finally got some test data together so that others can test >>without installing SA. >> >>The schema and test dataset is over at >>http://www.aptalaska.net/~matt.s/bayes/bayesBenchmark.tar.gz >> >>I have a pretty fast machine with a tuned postgres and it takes it about >>2 minutes 30 seconds to load the test data. Since the test data is the >>bayes information on 616 spam messages than comes out to be about 250ms >>per message. While that is doable, it does add quite a bit of overhead >>to the email system. > > > On my laptop this takes: > > real 1m33.758s > user 0m4.285s > sys 0m1.181s > > One interesting effect is the data in bayes_vars has a huge number of > updates and needs vacuum _frequently_. After the run a vacuum full > compacts it down from 461 pages to 1 page. > > Regards, > Andrew. > I wonder why your laptop is so much faster. My 2 min 30 sec test was done on a dual xeon with a LSI megaraid with 128MB cache and writeback caching turned on. Here are my memory settings: shared_buffers = 16384 work_mem = 32768 maintenance_work_mem = 65536 I tried higher values before I came back to these but it didn't help my performance any. I should also mention that this is a production database server that was servicing other queries when I ran this test. How often should this table be vacuumed, every 5 minutes? Also, this test goes a bit faster with sync turned off, if mysql isn't using sync that would be why it's so much faster. Anyone know what the default for mysql is? Thanks, schu From pgsql-performance-owner@postgresql.org Fri Jul 29 14:35:26 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 62230529FF for ; Fri, 29 Jul 2005 14:35:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 99330-01 for ; Fri, 29 Jul 2005 17:35:14 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id AAF7052936 for ; Fri, 29 Jul 2005 14:35:13 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7670303; Fri, 29 Jul 2005 10:37:27 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Dennis Bjorklund Subject: Re: Performance problems testing with Spamassassin 3.1.0 Date: Fri, 29 Jul 2005 10:38:04 -0700 User-Agent: KMail/1.8 Cc: Matthew Schumacher , pgsql-performance@postgresql.org References: In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Message-Id: <200507291038.04230.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.049 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/476 X-Sequence-Number: 13717 Dennis, > =A0 =A0 =A0 EXCEPTION WHEN unique_violation THEN I seem to remember that catching an exception in a PL/pgSQL procedure was a= =20 large performance cost. It'd be better to do UPDATE ... IF NOT FOUND. =2D-=20 =2D-Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Jul 29 14:43:34 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0E1F352A23 for ; Fri, 29 Jul 2005 14:43:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 96143-09 for ; Fri, 29 Jul 2005 17:43:23 +0000 (GMT) Received: from davinci.ethosmedia.com (server227.ethosmedia.com [209.128.84.227]) by svr1.postgresql.org (Postfix) with ESMTP id 109D952848 for ; Fri, 29 Jul 2005 14:43:22 -0300 (ADT) X-EthosMedia-Virus-Scanned: no infections found Received: from [64.81.245.111] (account josh@agliodbs.com HELO temoku.sf.agliodbs.com) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 7670339; Fri, 29 Jul 2005 10:45:33 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Re: Performance problems on 4/8way Opteron (dualcore) HP DL585 Date: Fri, 29 Jul 2005 10:46:10 -0700 User-Agent: KMail/1.8 Cc: Dirk =?iso-8859-1?q?Lutzeb=E4ck?= , Sven Geisler References: <42EA5D18.9020204@aeccom.com> In-Reply-To: <42EA5D18.9020204@aeccom.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200507291046.10318.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.049 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/478 X-Sequence-Number: 13719 Dirk, > does anybody have expierence with this machine (4x 875 dual core Opteron > CPUs)? Nope. I suspect that you may be the first person to report in on dual-cores. There may be special compile issues with dual-cores that we've not yet encountered. > We run RHEL 3.0, 32bit and under high load it is a drag. We > mostly run memory demanding queries. Context switches are pretty much > around 20.000 on the average, no cs spikes when we run many processes in > parallel. Actually we only see two processes in running state! When > there are only a few processes running context switches go much higher. > At the moment we are much slower that with a 4way XEON box (DL580). Um, that was a bit incoherent. Are you seeing a CS storm or aren't you? -- --Josh Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Jul 29 14:54:35 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 53CB952986 for ; Fri, 29 Jul 2005 14:54:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 98850-10 for ; Fri, 29 Jul 2005 17:54:22 +0000 (GMT) Received: from mail.Mi8.com (unknown [63.240.6.46]) by svr1.postgresql.org (Postfix) with ESMTP id 9D59852936 for ; Fri, 29 Jul 2005 14:54:20 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- GW05 Welcome to Mi8 Corporation www.Mi8.com); Fri, 29 Jul 2005 13:54:18 -0400 X-Server-Uuid: E0C866E6-C6CD-48B5-AE61-E57E73CF3CC7 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 29 Jul 2005 13:54:12 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Fri, 29 Jul 2005 13:54:11 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 29 Jul 2005 10:54:08 -0700 Subject: Re: [PATCHES] COPY FROM performance improvements From: "Luke Lonergan" To: "Bruce Momjian" Cc: "Mark Wong" , "Joshua D. Drake" , "Patrick Welche" , pgsql-performance@postgresql.org, maryedie@osdl.org Message-ID: In-Reply-To: <200507291237.j6TCb0A10838@candle.pha.pa.us> MIME-Version: 1.0 X-OriginalArrivalTime: 29 Jul 2005 17:54:12.0426 (UTC) FILETIME=[871BFAA0:01C59466] X-WSS-ID: 6EF4B2CB1M816182486-03-01 Content-Type: multipart/mixed; boundary=B_3205479250_19305824 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.551 tagged_above=0 required=5 tests=AWL, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/479 X-Sequence-Number: 13720 > This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. --B_3205479250_19305824 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Bruce, On 7/29/05 5:37 AM, "Bruce Momjian" wrote: > Where is the most recent version of the COPY patch? My direct e-mails aren't getting to you, they are trapped in a spam filter on your end, so you didn't get my e-mail with the patch! I've attached it here, sorry to the list owner for the patch inclusion / off-topic. - Luke --B_3205479250_19305824 Content-Type: application/octet-stream; name=copy_parse_improvements_V15.patch.gz Content-Disposition: attachment; filename=copy_parse_improvements_V15.patch.gz Content-Transfer-Encoding: base64 H4sICCKt6UIAA2NvcHlfcGFyc2VfaW1wcm92ZW1lbnRzX1YxNS5wYXRjaADsPGt32zayn5Vf AftuY8mmbEmW5NdN9ziy3GrXtVPZaXe36eGhKUjmhiJVPuy4bf77nQdAghQV223TdPfWp1Uk EBgM5o3BgKNgIt8dijhyd64d960MJjtuOJ87wSSGL4v7bffZi1//92w8uBRTz5eHYmcRhf+W bgLgb+MoDJOdxSz+wd/5AAbW7bNIJpEnb71gJiL4J/bCQLS3O929ZxNvOhVNVzQj+i0U0pub mx9aVK3dEn9LfdFptXqi0z5s7x729kWzBX81httsNj8MoJcDaLUOW3uH3QMGgHObf4TLvtXu CvolhDA+xOhkeH41Oh0Njq9GF+fP1qCtJsRfXoVxMovk5ddnh+IRBGJiEDY7rb0dWF22qGTm i+G7hfhLPmnzt/ojaDtEK1pg1vZxFthXC+zvdPZFq3fY2j/s7f4eC6ziaK9l9XrMUlw//uxr AsB//zORUy+QYnR5Mbg6Pqu7DVGv4+fnL8RGa6Mhnj8X9Pt/4ffeRqNhjIER3xyfvR7yIPho 0pBnW1mPwcWrf9ovX5/al6N/DUW/19vt87Q7iupjuYhkLIMkFsmNFKglMoKfIg7TyJU7Exkn wnViGYs7KQIpJyIJxV0YRffCuQ7TRDhJ5bL7u9bevpbk92IA/DkBWEfZ9E2c/jJxEim8YOK5 ThJGIr4J71B5724cQCEJFwuYEMeOpTM5TkC5r9NE1hs0GgRKJPfQQ06FDNJ51nEs49RP4OlP 8H/t/GL81fGZfXx1Nbbw9/D8xL44tc9G50P6/fr8ajj+anR+fDU8sU9Hw7MTaH1fAnYEbfCf IlutRDYQxWY4bfpI8kRGcy+g1SByIpxSFy9YpIkSE1FEexj6V/CbxKPftfoHLB5VVD3Ys9qt Ti5N+/vWQSZNMRDTczU4IUPfxnmOarWdTTG8OMvwIVwYETXGC5Ka63uwIFsGbjgBHhzhqEjO Q2BQ7E3kRizcGydyXFig0J2WgMQyupVREYgfuo5fGLKlhyDEGiLq3nzX+R4wxf5XQK7re5j2 1vFTEDxFwbYAknZwWfQ0LgC6DqG5tAI7DPz7I+CZSKIUODBVHXJUvFg4IggDgPOIP/AU6WIR RgnIJK+zRAcSbMQ1lsKJpPh3CsozBaxlFMHnXMaxM5OxBaMlOSDbC2x6ZgOFfDRqBYLyorDf NUhUdM+cBKQ9UDvBbfT8r6uMT7vVtdqdvtZDk9sIFgU2CBksCS+I4zUsaxnnAl5uGKBVAN6J TYLjJEngzCVx200jMiDQthJO0QZdEcEU7Fsn8pxrH3oiBdOYDY5zG3oTwBuBI7lDIP4NKCfP 4Lg35C9wTrYQAqIHF6aUk22Rt9rX6RTJF0mCC7EBDl22L4wWsgTnvgn9CQlgnNDc15LDCwcR AvCvITSKQIqiOUg5AEbu3HkwNg5BsgjULArvaBkgQV6CywOV+lFaAnwXgg7QuNIYXugCpg7d ncU0AkHRa2VQuOA4vY7lD6miMuMcAybnqKsJms6lJQG+SRoFKO+LEPgvI6YYImWSB7R84iQg XCk6gjVlsNY0dy6JBKNgGp5gr8LQoyJXL7255zuRf2+RaBHtkZRsMe/gqzKKLHlM1Yxtou4l seJDGsWw6KkngQ+EfIxoI5RAvksMqwSt15I5c32/TIPGdiZvamb33kUsYgAegUTDeBqcI0io wYxhtgYLpR84kgCbGRw8LFkDg6+AICJHLu7Wm6QgIrmQ0pJigl+SX5wJGFr/NleBG+dWVvJV IcwrYrfNiE28CAJoZAD0ZLlwHQLngE1O0K4aljBGU4t4p0EgkQloXPxw5rmKDeF84ct3XnK/ 3TDiuQMM6NolF/TpDcwWyuGWDjEKdgXHsHjLiaLawolipAKNKLkn7oIM+a4YTW2J9vfkXYjH 7LEQ9sab1kYBClKDJwR7D9soZcZpZm4X1L7s0CCusOdO9FZGRyWkNq/lDEzEUjOMKDYTHJjF noSBVA4WQkMvvoHVK3XDpaMIacoqpJbQIR14BBxDd8LoMb51M4v1vIAgzL3JBEAQBgWXigQe XH6D/86dZAnFGjjUH1Iwg2UK+E6c2HdObMuYqFMF92r4j6tVgAWMQ28NYd8t2jpaP7Q5EFEB WIJvWCIVsqApfdz6C9KwHNhES3MPxg/MqyLOR81uzvv7++Y/3fIqt/zLPDI5Y/GnN/4tvPF/ hiOuDP13Ya+oMw+ZZyU5RuRPo3BeH0sfWkHhIulb4swDqm+iq03nPny3lPOgfYb6AcNjC2dl dzORvje31I8g9X0wUkBA1dmNb+15OJG6A1lm/YONJ8Kq1dTUoESutIMwIUiASKwAoZ6B90Su N45y2adnmhNn8LDOoEVhImWmzYH8YDmvYMYdemlGk7FAgAU+sCZKaQKxGdG/Cu9NL8Yxj5ga fNrTZ699iK66yyMRzCQEzIiRSnlJzM8pBLMDUn46D4BPljidzyK0PcA7YM40zKatXYCcJfcL L4TYyoGVwMDdDraAQKyePBPPbMaLNFFcZTtvmQR6cCiRddXooliSIMJfgYqMKMsldTNnZKHF Kb+QyTGpTbxSpSB6RZ1awhjw8BJmawhhpi+DWXJTz6062v1GtrUSI/QRgYNpgfk8DTBtBpOB yUwDl7+ZcTTNcAkxISL5Us68oF5QaeRmgHoGM2Aw3+51rPZeOaH0320zrsAp1NHt2wnnldCQ Fyf84HgUsVXDK5A3YWWELZiCmDB6wBho+XzAQJCJYjZnD1HGTPLHVVxcBrBJ7jCcTmOJNGaY c3txcx8j2SNoO4Vgy17M7Nyn0qBHLPlp5q+otyamReL82nX/6mV/bKP6OCcDo3CupxC4gnKZ xzP+/jTjTzTjxW3diUwgLkQjMgwmTzQi+UbtI/qEyriydwA+YmVciZB52QV5i4v6UOg+gB7M XndlHyDQxXQc3pm4VtoV4CFuhOrUsIk7J9om4KrwBxLZHIe5GTWM0MBh2t1mh2j8fBhO6w3Y CkylLcMpHkGVYLyS8q0JpLmE3EkYSOzFGq/EcuG5b9PFypWPMHCqc/h06xTUjBsVetwvw38F qHYfQbX7FaCgMQcF/UxSmDtJAESHDjg0SnYwc5Nn33AfPI1CkEju5dEOmLMItHECPgczOeFw Y79ltQ/2VoUbn0iUDJF4hCQ9RlL+sIylJBglLsGsBLj9TBPPj5dzoJv4HOMS3mWZ3FCWCaRZ GTCwfY2lvCSOByu+enhbf+noL7sPQ0SMllxbJcgCpI8hz1Wmcne/Y+0e9PJ6DjFK8OxvHsPe /s4DEJjn8X3MQt3wYb0YnWMj7/GBL5y/8u85MerwWfy2UTZx/vIQOnGKiHMhVPSCJ5uLhe9J Sr5wnjPzB+UcEgrEM/Fo8wldf0I3G995iXsj6pShx0IBfFKjRzVEk6sPTkdnw0MOHqbo1uoZ UA3QEsgpBIKlP8Qf6OxNUYFAk/InDeXcWa/ECzrXVd2vAfZb3sV0W12r2+nkVSZ/FLLj6QIf Sow57XfIeTM+E4EV8bk2TXgnI9DK1MUE0DRFVJB4PJpSYFmelxOD22WmQqen8nQLeEp6AuTU UQhQuaWijacz3IRSYn7b5P9vxPwqHey22la31c9LJrqdfau729PSwYF1XUaRC7vO+nA8Hlyc DO3Bxfn5cICVSPbp8ejs9XjYyELLGh45zeNZfT0N5LsFhHDA7uHFKWauVYUBiEQgievrjYZe 1HsksEmUrRcZDTipny4mDuWKsTwgDZIKmWAYeouyyWnHOBRTJ+IwLydHkS3nw2/tU2AM9bi7 AZKKup5dfC5aWGi0pj1WJSU7B0Q4Zc0YA4K/4W4c8qGEinEyTGp8UKRVzA0jzoRmVTKsKJmF NceVGJ1tzDhlflRCYWqgcOp4fgEU0A3rNpC7F+Ocj1WM//r1cPxPe3B8PhieDU+A6SQzPZCZ frcoM59s6fkW4XcjQqU8wEagd7BfkAfnFmc1xFo9WPxgoyaD0hDurNb4S/sB+kKjtbbUVDNA U5aLitzIxTf0GLFVMYgkuvmCn2S6V9KM91jkxUVdhsFsaoOZx/G6jAzDQtxWqkqwKmPYRGPY rDCG2AZDYSkzmbh109w1TedVGHpxdkIKmw8GKsJ4pGBdj0RDic9eoAFqcBvjQIJwEYDAOeLN tlmWBk7KlzPYL6LNAgeHB2mLKExCN/S5qq5CXJpPspS699MtZZO5tZos2o4VlpoGsTcLADgH fK5mEvyZHvC5C2FqW8+ivIzezbEeIZ0BQ91F+rEsPGPYtSJ2CmHYBTipn5gsy0Hl3alzBeNK Wg9NSt9Z6N5nJYhUOmnuOMkL4CEixfLA0vUFPMEDzXWBydRtGsLjnCkV8OEJqz5KNUGBFhba 9ZZV2PM0TmyGMQ0xerIgnPIhPCGO4mE0oysn5nQFGHSWKj3wb5FInLfs6RBVzblwOtVnutKZ M4Q6kIr3yCi5SJsGVgL6EsNvL0FZyoeUek8d4F9jOysXXVL14p79j6XrSJg/lf3/tbKbG7Tm gymkTHxLS1URkw5JYOO8kYDugnwKJyAWsSA8WcxxorV88qI4fgv7kxBnunMCOs3HjqiwoJpN DFnxmBx+Utkr1/1EMhdJwg/0p7g7MCUGU59TKtGACfhsPp8C9jfJBphEH03jPWKSw35YBXFh H1hXxYoyvNmSVPlqjEE4lgMbTv8KvYKJN6mmlMVFxJmxuyMieUk2PsbKDPVjR+26eHlrlZq0 tsS3NV7f2q/mG0MwQrvm51wt0myqGPa9Xv7TGVdaGdKS6NlcrlbK896TkLbj94D1XJY360YR Y3d/3+q1dgvx/ScMZXmLV9gnciz7W24SnxIqs9xW7T7eP6soF3saA6r2Fgf9fetgv5XtLdQ0 uPxLSZKY16JlJWgLGTXzMzhdn7WtpMcLvCQ/j6k/L9TaMqfKXXSVET/Vv2xVciQpq4EBhsoo lm9QoBefoRWwSw/qDLB0VwK6K9d1DVo7LPQl6NoqQ8fMELNIKF+keJvZS+h4/vrs7ChrZDHN mukGyd6uddDK0yEfidL/EUTewmXjJ1ZZ3Ui0SByOVtwbQbOkqi172YNYDSZFxBrJIEyMSyMO huil6rR4m8fsUPZNO+DiUvBGV43Sc2zQX31hX/5tdHlYaHk5+qJXbPni5d+LDa+/HJR7tPdb uy1uLFOWrs9kAcqWYRO2zPDnQ0MV47YE5b7wX29aX9MlHrQsPEZWdbQvxPqbN+tFicb8e4J1 hWZd7sfRhsrDznarZ7Xb7d1CliMLdMH+QjiNTL6LvITyyjTTm/XP4jfrh+Kz+bqRaMFHeICs M4LkDamYtF4s/Ec7yV34qVaK/AFrKH7iFssPY/YHkTq13q6Bf0UTTWfl4Ggh5MOH0lEdjymv fAkeSJ6F7lsFCDrOwuQIpR7XowfHN7TMrODWhxEQPiYeFaiDFvBhHgiSBR8HBS/6cUj1WCpt /dEIVSlju12Qsd1e25SxnzQt0Ns6qCWbYZpA5EXe9yiLJGs1oxlEmQpNwMK9xGGv6qfKEw9g R9+uP8fShMw5f8d1LxDTtb8vpANV2CD4/l2WvtbBoRPH6VyTkWoKaVuPrKUTsWspISgMnRhN Xp6ALB57fnM8xksUdQN7vL4KzV+ejC//pevROnttCz72TZn6haTJsqj/PSSqlqYDkKZuq5tL E+0+jVs0fJKiyIjXGOgCDRXg+/JWBf0OXoqBvWfq4x6bipOyMLy62KUqosLMQuRihvtdUl+n XPhnMZegi8+wwoUBfxYfKkOQGwGsF0SkASlEWm0D+UKR+qHWk9e6lu76GSZCRbW0w+fVk3zt djqgervtXL5+T2J9Yjo9SKJK+eq2+njjq52fqAEBd4GK3ZahpUjGilju558V2ktR3AtRitTU ed9Pai+D8SiwdpGlCAN5R+ThyAn+NoseAPZT2W8gOupwAwwk3s7S2rqCNcWI9AHeLDPkwxxZ QdR9VNqD3AVceBOkYkC2yIaYm3Ah00UWj69GcyyDd6ixjYoaKddAV6FqtHU2omfVzOV61MA1 4WPpc7lhZP6kDkO6oE/QZUx39SHaAn+ZyOE76aZJGFEHiJ+xkgSDNWwfYDVG5ODdkHpD77FB ZrooOB+43Q4UANnqtUzZ6nZw1G5+9jZ8t4gGzA1ASvFF7YzxSoVGYggkwr7UoENCrGtTBvcr OQ+jew0JXLYChSvke3CFHkwOVH7VMNA3tnPpOFK3rSh9jDQza0cpDObEcrGykkjHsoc7Cnwo 6KESbeZaEELMHEkbRZtAYU3iQEJfmMdNo7wisJakixMIqWEhuooRtzzQEtVBJFmqETOM7blr 83MqZV3hUfqtPavdN2Pg92ZaALB/FckFbnewhMLBHczx6dVwDDsHbzaDvf62IvkxHj1ccSNV Hn6dyui+bka0XIXehUiy1zW8fmlC/BQPT8vdHjt19doxAOnvGfUDQI4+kqNjvMSjpt5U8ELU udJ3s6Eui9WLxcKUsPsRCxuonwocSHFxsE4JPTAWuzU4T2RKGEIgsXkQAPRqGEsXX+FZTBSG c1IWR7w6Hh9/ZQ//MRxwgMM6FM5BpblkmJSb98han9Di3zixofvVZQTt/gGQdK/dycQpV6Bt vOLohSkuRb0XgR/YYHxczkZVtEPv56YWZnKJFTl0G4uSqMlNpEp++G4Am3DnzrnXiQtcgnFt 4BlVt1AeNbfgW1uczFTWtXD9SJXqqiJdUcieqZKLNRxn1M2wdsdvvYUNyuirNCObfz8EXCZo /WGiUQCs8LCYWzOuNvhyOPi7fXoxtkf4DpPx61dXl0qkm8tIN3nQDlYfgbios7eoSdMKCaZS W0AVkVK3V6A22MEwu3V2BWoiUokOWoj9PIz6aCxVmo9JJA/I8aPMr3JjQWCerMqUn/haz3MO a4qnbKi5HDz3kshZ865w4QnuQvHfov2GLiDx8KWet1M9aU1JocqjGcVa2f1VB3QoDd6i3afH jC84HhhTzy6eW2JDbFill/ksWQPKvFVeVf8+i3sU/oYvyRZIkhGEAp8oraHn+hrmNELzgDIM tuB0qFDNroSbxSr6urlJPGyfhLlGqUIzvERQLDXDh/Sxo/LWZRpl1W/5reU1FbivqWpeauB5 ShOYB4QGfQvkYlrSyBULwWeAHmqRQkLd3o351iv4F6qmVQnCIqIIly7vA9QMBXPK7EUBTBDD huiDI7oFGiRewK+1oeQ6b3OpbJDWirsU82wpoiP/4cXpNrdrQNchxB30UB3X2k4wsVESShAt 8UPqJSqxES4QNDhfDQbJpL7n/MgpjwVmP/+sC8zwa5lVvDnRVlJT/uNbyo9gEyvd3l4bwoa9 bj+PoujqRmbHsOZChRHobdXVdExN6rmVWeBOlmhZ4jHBhR6mrjxtBBurx2lToshSUCm1axYq T2FcKVUpjhfCeNPWkT4WV3sTvrejWpdCVzUfUuTYTVJHl7uysGW3vucUkIvCeaNST23fxV91 9udDjvmw6LaRyBaRumHikp2xolbQ+8hgk486TRjNpROgdmxEdF4qt4U61MSXO9GIKV/nZxik o9nl+JjyhXgV3cFjuzCd3dDpLvzS+1qusZGTDML1PYK1hAcLUTfzc/oYV9wBDuknBD94Iz8b 78G8nPTkBWYkRDYTGZ8/L26ZYVPe0mW32XFgvr9HZ2BDyIdqF2fbdaG361xmcE31SPjKDmJP 8axYvbohj6LyO3Hqalf1nUaylM/1TebnxgU0NT2ip1r1trxc4kJxC2679zpGyvqPq5NZGiQi S2W8TkO7niwToqAXby5+ALXCdqBK63+FftYKIUKupgY3y7dpjRtwS3q7YhhdmTXGZdfkytrM mjCtfHUNRRe0mia+ziSSczz0owY6aN82QFxg1QPeJbAQGro0L6b87r1MtDflMwHwV/Sqouwd QzkQVFoiI8VcOP22eInFFQyRYtDMNYNas8kxANA5I/IghJ2th1cA6X0w+UIAqYmcYsnVLdsI Y/BdXsfCu2eCxvlN36G3ySyZCiOxx95Zs6Nig1TjXKIgecr9fv5OI7FsW7bySQobMdWuk4G1 WtFOQZhEwpkd4KCG8ItMOPQKwrvtXD+M4Uiq71p5gLzUQb3vRN2C4IcGaoVDTkUJDsyylvf6 y0OWM4OgnSY8tbXj3BJPsqy1Wj7YsK/6KvCydV1x3Tc3uL/IylZHQr0W3iDtV553mQlrPBGG ZaxnpyuFME+f41xc43uCR5P6Cb1mpXieA1298mKAtAOuQOCdQDqvM6H+r71vb2orSfL9W/4U JW5cI4GEQdgG2+3ucNt4h1g37jX0ztwwDuKgB6itB6sjGXN7/N1v5asqq04dPTB299wY704D Up069cjKzMrHL+vKkYN0rl6n1Uo32/KY+kLk5M8vXp05PzVoor+8OOHsgp09sJ/vhZ7T77kg qRXx5PMnrEoJ2UDi8ZOWJhuOVdKRdeYNKD8aPGiWWw7gEZkkYBiO75bzvln5Doy2Zs97w3hE gEitIfstewWfmwEiFdnL2hSec8tEraCB9x8qv6STRRT+Z8/cmMZHI4OIXeAsoAxaZklpIqwV NvyTmKuW9Udb7iM5+7BJrJNXA6XcsetogxwxLLs/7gHn3h/2lQwFXYk9Vs45ogjuyFKypTbU T94PP6DxmZ39dbVM6InTZpuKD1OcozmCBpFgbwWwCCJ7AQhIKJNK4MkN57kpggV7GVi2qrdY VqNiqVW+EDB7nEiH0KooirqqjwL57xau0iL9ehkF2ylz9t5ALeC3AJDFbm/dMTTidQ8BYmZP +wqLpxjPMShMpILDafX6LkMEeiVHKycFivF6Yog2UkoijiRi5JBwfVh119ghob4do4Q4JJRK vEuJEaI+y0MsjmeBsDa3H5tffqUorsRQ/XPfjbFqvUkgDUVt8hTlD0Pg+4awW7U2lmCDh4DO cTntF3D+1826pyx2pQY+1nhvfRvJV/2KE5SUi/vgj9pvhSFqkdbgAt/0aZOsCcwooq1+yUKF Ax8nBI6J1xOGx+O99liBW/Kw6+T4qtvuAxxJlnefkrFxnW2IsAGZ+b/dybjJAgJhF/GmdA1B 564TuEdRgr3lAGkQeLDJ2C+fQWTzaCyRGT3XhdyAusOr6Y0b5gMf4e/ZuhKSBStIFa0gq+XJ lGl7JmbvhB2LhEiJWXiBc+kztEg+hzgdxbK3/9Dy1P15nvX91r4lkl0VFVTpDTpnFED+PMSH 0EnYAaKFvb2ph56b5o47xUTugSMgutsJg6cRP7Ej3t9WcTffYjgV7YkIU2kjF8jcEacXFIL2 9sOgvUpBAy9DfNpORZPdB52bQz5AYU+1gSYKAcpRoG/R3En2zSLcs1J3d4N1XKjTp6xn+ztW lO/v7M0ju8ewSnvBKpGtDPlpOSJWYQ731cLAw4UGflXo62hltLLJKITFV3g1B2nE8X1hzj+B qc48BSkgjZKMlpZndweO3ON5ywP2x/0nwfLQWzvd3jC7et//gKYJeZ3yA5kj8EpQckU2uil6 58Uvn3TMi1nPvu9XWDIImEH82I2ryZDPxH0Jrmn+2G1/np7h2p51bcszu4X5++0PMlG4z+8/ CoI0l5pFeBVaOCF10fnW80ru1hOg9yctzUO/FLblxQiinK/FFAVh60pgol/JXTrJy/QcwhGu zqAlflDjoJyGEasyLmZooAXDEeoKCUvS32x36J86tpyo36lhrw11xv0tC6ptcLQMmgfz6azX c2Ve0ML0afyRcLn/ByJmjLj1lQHcB0kdY/7Cybjmw6n8sL3nzitMRD97T+AUaEiOL/MIZZkl jmjlz1/pcAq3XPRoVl+39mkaB7H25JFG+YD4rkHW7tLK4vULf8kBuli0fDhox1MrUU9kYWEh oElDxMnPaIptcE62XxjAq8BkG8po8/yLvef0NnkR6W2D5o+TDrDd5o/0hAiqOPwwCGmUAWlf rQzh7X82AIR+0lXvhM1pY7QjA/R3R1BejJxBfT/5vA+wQWdIXP2RvRxNawgbiL2oW7PXOXlA dhb9s6PZ8BAKINl+f0RNE87EE/AWP9nV1+RvvQ/RrW/xnhSuet9kb6Jh3WKfCuO82/1Kn6NH +wBt9zg8RxSf+O7t383h0fHBuxPPB/T+vXh3iIOSLzm2oGGiBfPDLUZkQpp4w7QHXcsgKTvZ lMVZJWK1ZAmWYzD0RgoI7vIcj09enBz8cnB0YmiBARxrGk4Wp3q8zFT1a/5mN9tuueWLM8sh w4DPRKznwahDkZ6aroykDBHb10lESgDQOQTT/JO96Bx+y33cFC/YV7m4Fji4yt1bX/jIYcYS RuOoYBzpnxpJq3mRPu4RvLNu8vq54CL21mJylIuA2zTOG1Z2TUOf2hcj0UK6mSIWDhYMz4LE BX7r81CBX7i+xvJnQ43uVueDpzzvlKhXLDwpm0HWHdvIBH12mUOUZo5Pthut1mPnybknOeRV wsZjMyPWk1BFKUhXynIMlbHizxeBsDtxqWH4pIJDlGu7hZ/KW9D2w4gyQMNIveB0CQG6MOYG ZSUaxdx31I13zQgAX1C9zTZauj7BPYDgq0q8/4TLDroPCGhSJzLHacJgqYqTaDgsFOFs1O/D c0g88H//jr+HwCo5VOSDTXX+6PYnPyCBgI5DaBsmFT7b8JEfOqBWWWxpTdrMp8Tq3TZh5Cqw GgJwhr84Ntv37G08rjf8Be908LHvlX5zX0BBposRqBU5Wy85KVnFh9iXIzQFog3Bdd/SCr9H xTzKm+UVbbYXFyck8cTwa/5+54PEvzID8Sweq7XYqwEaMnXgrZYOwYcx19ffaZbvMgAgIoMD QISa1Hmy1x6IOBi3sUJUB42zxEEW5OvTTGAekoEDh2OW8yRc5GCQ4e+42KG4QzncFLPK21Ku qYPo2TBSPoG5qZ1OgEs8sN+ejurwK/ekinQi1CsHoNROt+pbNEyjXgreF0NFCLBYmJUL19nE XgPB52rP9nl/0Lcbcd0lFFF+EJrllvgvCSmB6UfC+Xzon33hz6rpmNCw7OkOxypjBJQeRFyR sGfP9ixVthOlaLifzrgLlSLtG65HRWreqsSzlkiSBtSWAl579AYEVN+erP8JvI6+tuUVB0Ta YclLQVBTzgh0AS1p8N6mDgmyH7vdK+qiO8LgR5Q5sgBj8xHu9xS+DF1wYTLCY83CwYxB9Zlu BbNhGJLxbBKEXMKIkIam7vsG9k/5uh1/su1ruCuHzwgPc9lhDd6gX4vIORgF5hKAIWwU7E30 PIVVkepjly1zvXMHQ3sa+ojh5oZtMHEeTp89BR/tKn60asTxy8NDTVFwavhIAcHWnj3TuSve oI5IZ+yERP+Tg6QiwxkyKncolcVc2cSdUSb0zyvLmz3xfHcD5atr75xQ0RdENG4cGT2IIHMK W7JnfNgfdGBnJv32x5stiItgosJeSMSguACI3UFOR4Gpuo24deivWocoPCBzK+DAPWgJyI8F CW7Gmf3YGSJyTMcXF2CVo3feIEW4QsgNqXvseoGzPcpnk67g7TBNQQ5V1naQPJk7dHyIhG5d R3yYQIMeWurrW+Kw/Vids91lR5mfNhEvkJZeFiYtPBT0sia/kwqRtbs+gM+4GzWa/kVs379v EsIqzoGpRgX0XD/4LIs8AJAN2klnSkeoGl+hT3VSXTSAQEAEwdkIhySVB32wJlWfG553OxBB QxwN5MKb17ieOdCR60PqbPURQM8y96lVnnuT7AKuF00SQlPAGrE/gbay84GUv/ZkAYke0IZv aJ2xXvvyha9JnWY6i2/O7Fh/smJ7go6G09F6vS6WkDDIkj4NjmO10Dv0E5xuaBa/8ehN5DjD vkwhlOLuw4YGGJQOruLJpJ9dSD08YM1E1yBP1uJHLyEuYO23HNBDTk8np2sUncAluePOtnSU jPbN3/10ZiMWTaXzASG7YE7cBVa5RA6xxOTYaB7aU+EE2NUZgc+eSuphCBkn+FzSFREaCHBT mjp+O/rPo7d/P8LY3QKpFkmH7hotFjoOX7MQ2dUi8hytF2K7AphDAO9HzE8uuWk5rsr8l510 43LDiqIdvM9cvQhT1mFjTidUms+KjtORAuRILYeas/Y045EpBh+VB4WtTmZfdW6+4uS4sM4g ruXbTGzxCVo4s9ucn6pE5Tj8M6IPZXgmbfQCIFzs1Btog6C7oD9bKKRdyp16GJT401EjQk+U 9yhKLtC+WOnLKD0R1VOJ9LV50mG0WDpYeVR69KsFqfEdhYbPlFpeWIxikhfrzZ8gJFYd/3zC Tk2kEhBFRDvAI2M41ECbqIU2huds2HU7rMjodF3n3FYQHKDFo1B6XkQbSSlBxhK/A9gocYfR 2iwJE3eRcRSZusokR1Imc+6lQoKi165vrScOwrJvgDxiqkWCd2KIdgNl1k2U65n7juccRC+O Su58FaW0a9F7l7HLntATBhe0ScAtZogZRw4bQI5CPr0ZdAMKdvcD0mS/83ihasPsaqrlIB+m 1BJ7oGk3YjojVVnuZpKpuFClFeMIVxi7H1pC+w8HCSBN/IIksa3SWrW1u/d9578CrRmfXMWE QSYxB5QuhtBdjfPHBuNsdGMviHBf9wm/8PzpFnhe2WjnemJfFw4UM/4qtX/84x92jDfnXf/s eXc6RZ8epucBsv0U+styMU//JCpomPahUDGfa9zneqz6sucM9wb5ppyHJKeteL4X8VQyVMNm Co54OoiyMk8OlCk9Bl2FU3sl70BoQAQ9TSaNSiAs+AEYClABPtMd2idOUTWjnGo0boP1hxWJ wY1fxOzqylWdA9carMEvWXuigGobJO8ia9hSD7Y15AA+BlZDSNuOEWadK6dhoO4CRAWB4bs3 G+hIfOmCooiVXTFlSOx+zqgIOZmN9ePy1mH/4nKKK+gsXN0bRgOPzI9bhgae6CZHC5rYUp0B DfvtjNGaIoMYSbX2wlCuMc0APA/0via+2m8gTpYsLNr7EnnJAq0kJ4+ImK/QB0bwwQ63Fj+L /Wp2sCKP0Lr6u31o55n53fxgyI9mft/cjFSAUgFcNLuqUxnpK3w90GfJiZ8V6K0SKC30Djl6 1DedHDBnWeqxPyAOHXwufJzxnezzF6d/oiF0yaEnLtjfJ9rZ5UAPEZzf6bjgmlUQTIU9NK7I 2KdpAJN1IaDP0/EZdVirhU5M2ypwioVBv8YEof6COvpk/2GjpSrIhf7p1/1RJ/cO6pODf5zQ 5Q/NwfZO2NcpEhydgdb/MRRpoU767NDxXhEuCQd24z6EF1H5WQag9v1TnZWsI/3kvmgdlD6B EIcRmVI7+FQT4zgwlRDtogCVVN9iqArqQzLMxfEEePI06vNuTyKtri/Hg65PTATY22l35LxS 1JOOR0E0iglPChn6Mh7yZapxi6OcTE3o+UN8/qq4yi0zHOVtcgkv5yRPebm1czp23n4Xvyn8 15D3NHJ2FGCxayCSIBiB3ax1DX+VdHpHrlYqRSzbT+QGmpwlKZTAUo0Bm0GFH6PQtUpthsoL RTsD7aQPyFgOCiCr/SZEFlUx3RsxdBeB7QJQFO0Jg+C//iQ6fu6sGarOYKeJYEYi94mbUhZC Cpy+YRaARZVkmojxyoOLhZHNbv3DnCk7RzjZoG1TPAUVdhr3vBbLE4Oz6DzH/uSzB3dHPck+ VLciuTl6g77YthiLuRekZrBxhRBqc26etAEPxEaGODkMyIYwWAgdBiSsx5NG8BJkLAWNFcjw B4SacYm4YHaW4FiO1we6DSzabcDTOoOvXHeq4InXupHZ2oVxw7S9++VqgQOYX4OkBIdF/Gy+ E7Sa9LpAnPKZ1tA3/AxhPduX73c+lHhPKYltjKpeNsWKA9I3DQRdhyQruuIqdzhc9Di6XYN0 WeWU4sTJxcRPY6ZqSvyQxikDxA/5GFuqDyxFqsUvwGUEW1P4lo+PqkAgaBo+NaCkBzprVaMM uu6w1UQjzUh7QW9yxrgrsCdYOgyscuQWQKKiT12QR4P7Ym9f3s0mtgXyZdEMqK966AXkixca f4KzI1I4lMDqaNTsN4QaFxRjlr1hatr+0FCAd80AT65ex1xHwLnSZ0okqFRScBCDTXcem+o4 PluWZcrIpP/IaBdYF380OwEz9+IYHcFYXmsRvDPeF0Vl4Tg9YD+n27SM+hS6XpXjOTAdFglX 2Lq2TEVzaLl7tHcM8ZtcDaR5I3BGqbJTo5V599Oh7Og+vePXj2WR+bB4mLTdlF/nT54GdlKn rkQURrcY5ZWTqGCyg4ucABkIZ27kSc+jdcicvFZBD4dvdF+KWIw3UYURvHF89qrfbXddPHLt ajwFjQLNBJlno/UwIIvPWcP1B7xZVWpyIF0Bh4CDvCWPPCicRj7wTcGL3IRs99sfPtkcjy+5 +dwE6rPwb/9ufuXKDtEHYQUdivHJ0GTIic6wCGghDcVjzb+4XuohfrBBghr7dLLYe9FWkmhe c8TtxyH6rmI5Jx87OVcNCDp0Hj3gmCNQVRpsykoMV9NvgCSQ9EYi/C3jCVptMW/QWmjVHe8I l7gKjEgahA2hxuQNQ3xf5TrizlwT4IJtFXyVqK1pYvrxuRc8zrxQZAWFk/dSV1rCkx/MBDP4 6fzGRwWpkcilqhVGxsiCmaDVaocWBGT3T+7hP1JUh0UCVDAS4pCQ9KfQM9xDF9kVbqHnMpob xgQRdJfoKuTWbnYerRaPRr0STzIbYSSplyH8D2tZFWavJRQo7lhyzyqPs/NBFFZnlSP8tmzm 8QK26iUirXyFyKiEUozGAQGqiXFYboOLxr34MXxx26DWOpigJmqiN8YswctDoGu5tQ9kcF0M 2KVzifzw3vceGqj9TYGOLYu5nMMISOSxNPF3Mx9L6K446nqhFSe6JIdjn0ulbrw82LTqoaAn BWNbuyY1c0ItGm+mrEPjPQVMY2AM4fc3PEChu7EEB7sWDJaHgT5nxy4j/aaIRV3qvY18BvbG euMoAm4EMGLBWEVOqxAaHZmU6T9MKWGSlGVoxOL5hsG9FAwBJdbT+cbT6KLu7F86S+IOTKea tRRNp3JPgTdVI209vsPjCHvopBTzF8Gjg2NHmUUDraiQ7jAnzS38Vqc8LKk/2Vk0BNfd/i4e ab5SYie0b/4GjupDYP3rkOnP8aqk1U8LRkcGSnkSPDgWFz4zoNk0a6dba2rzFbZmhFgulQVF eQusIFgOkFzBukCsnX17eBVW1WlAjb6tU4CkkZOk31nU94P+X75bsv/JbftfegaT286BTZtP RcPxTVx9AJbc3GWw1FHZZ/iIpFNUcpd5T8ZeX3ftaLgmYL90zKiEFaVuYjJItq24J+FL/q7E 0wGBR/92dCzl6Kga2plyRwdg0JX5OYL0QHF6VEudHnM9F2Fa3b2qif8l8tJu60WpupIXt3KL lDFIT9DI40pS+NIZfzzhO0nvm5Pdl85WvL13pbqcd6V6O+9KiKpNrMKJ78jBEsrtP9HBspq1 if6b4nFahgeb9Fd3wVQWG/D+7Yj5WkeMXkc9qG/jinFf3s5yJQ8taaNyF9jiRUnuhyVmZ/56 jqtF9Z4+dMp+rA7oXOeNv1AH0Xm3cODwhiCeEmQgruq8kR26K+eNFf5lvhuWJMyk28s5cwKb +Hd056idRG9IIvev1I0SPlnwBQW84Rs4gxTFpl0nESe4re/ElNL+0gItZcUoQKqQccE7TrJJ 0XWigFXKHCcqUvPfzpElnSNqxQIVD9OCWcVzrEfJQ5cLGlF7qV0JKNqSVgfxbICv5VMoKy19 h/JtoW3ekXYZSKq+2mrvX0Wbz5TDpw9DQ4umNlwudPksldCw0DcUmnvneofu0j1UJnuLwrdo C17aP1TmICrBR/wa79Ct3UMpFWshCaZ4nY6OTtqTQ8OxDgD29KTNw54yygzESd807ddCI24y muuOjbjfMf41hdfUaj1+1Gi19iNQyFVsuMYJvMCGG34c23DNcsw8YamVTdGIeEeAKAL3+zzr YclgAYJwN1hGEB8CFMBFV7Dc5thsjTakCWCTMV9cvDCoLBWyd9nLfbs/zAaqoG5mr3ifM/m8 07/oT6MI2XsAlzA1WL4ZW722F7i/dT8TBdinKRbWGIZzxj5q8DnRPg/NfmDl5DqsKcQ3tx62 7H4+3NsPQe2++X5uBt+GMER3udPLGuSX2dzva5D/tzl+5XeE5njfYo4t/q9vib97HmL/e4dc JC0l9vcard2HCh5bwjfHUMhygj1CD9k6aHDbbprgcGi6f/dCEECaKiY1cVSAKzzSgFKtEACP hYcY+gYT8XJUUciQKoB/WE1FUva5+AR+BklghouD1CTZCtUy2C+07vQQZOq6Tv2p6ivSH6LD M8lSV1vGShzhQaKOIn7YBJUn6olSA7tNGT2CPl+hZj+JAAs3Jg6xEDMCxsrehO/wqIXQK2gf T/HJSkVVt3haiafObQ6OXp29fX325vDo4GlFaYv08kPL5/oIVAYHr0E2cloyNE34REcZxhhN dUQB4UZsMJi+TATgF21rgljsdEEIYO3kkZ1Xl8HRxNqMqZyCvRgtfLTiXDkERjwCzbrNeMbw WpUAh24hGqTM6BouVedyNc2nTU1bXXvaLU0SMtRl37LwYXZD/WAYjvYeoBaOmMvENvSmKoI3 obOI9LZ71UTJE8RT5gZcHUh/VCzmZFwnjHe5ITWdEMNxw1fv+8PlwDBQJP2Ob4ErL/4iXhVy QqF5+czJ8Uiyq4bAfwsfkmkCYSid5I5Km6ZwD4Mvg2iA8JuiMsnvKAUj3HDZd+osyINFWDU/ qZK5h4Ytbvrj81DjLgq22Lb1Pupic/ODirWAazDtURA+7ecS1GTWLwuxQRSOQwQHstwMAmPB 8nNwapAPv6b7Ccp3K2ueBn/v2L9Jg93fbrT2HquKiSvIGjqtr8RRht5gexjQBnF+g8cYpchU obvg+Z2Mr1kC/DrO8z6GyvGDlPMKviK4ydtlAt0Hvc+ua3uBFUlEbxYMGaR5QAjEQwkoxQRJ 69mnkxfC8juCfglcEjjNrA1aSm82EBi7aNz3xIk+GI8uGG+y4LzHVF2dskR4NpYdMsOd3Jjs AstDjbydjg0DMRtjn3fohFzJ4w1MaRPZhXNWkylB7gxFdlXwdidxVzV+bTv/FGfeeZBZsCEj 3Apl2RGfYCUdn/uXA5w1OgmFVvOHyN+q5miU9+89tvbV0ajWnmKIt8IVXBJWcBGq4FeBClYd J4zZYQEdCfGNAAUH5o4jczbcyIibANvxX6Cvx9Abos93ohi2SnG2gU24fEpR+pQy3xbYfgEj cME89T8ddxoRC7DeD24lCenmOrthE0wiFDmJ46a+8+s2WY+/2IkXdFFUcAJKS3+VfFfpLvmX yc+yDZu7Y8ktE/qkXVOGZv4wTjKNorjwKgVlNDFT3Zdt3DK+3iQB9ULBzQ4FCThjD2Eoo+jJ yUBSg77QCd62HNKqZ1JPscF4/Hn2CbiX8+qI8AIfqX8/WOJAgaSmeR9kmb4mMLoESMHxbAr2 Fq7ZnpsuwumD7Dsfs08dIeftLCbZjR+y1A0V4Hr53s6VvE+dLYAKDttck0ulPZh1ur4b+tYH UJx3L/oUU8ERFL4+Kb3EfozXg9TU6ekcRXVG/L9nJaYqcUrh9jPcDvSJ9Vl3AFYN18ghSOaa /QnRaR0/Lr7dYYXNhsn89XnbXWuynAzErMq8HiM0CITWPaWXYWx3bzyuZY3zRrvRaXQRSpve LZ+jwQpceXrt7FWIbkFw8yFQHITtwG0UH5DfUX39yp+a91an227A/z169AF2JpOS92zIw6mt ZWs8KySfXIYhISG4T9sNs3a+BmpKaxvHvtbFvx49aiBarlQAVfR2OIr2ke3CNWlcj3bRqVTq 5o0xHhlD2GeUzTYhJ7TVLGe5+AP0qvEEkGquu4T3TIcHD1dIatlIUxoG4HzqUo0bhA3CPxHV +CVfNwFAxg4b90FClwYI6kL14+2m6mV4QWkktg8oQYlmhqFVSehEk7klg9szaqedWVtm1JtQ AA7fhcVDjI4zdAbTcuZ04saTvp2UVXqQptCE4KcFTg+5UUMhLTsZ742Cmp9m2h8CXvOI0KPt ooFmCibqcUatZ6OPMgRHH/LYb/mM/dXUTh94gOQGi0KoNtdmowHYAK+7tNwuDaZDd686E2vX ztDq5J8AxhkXpyvWoeG5Hfq452eVAz3Yl9lJjeEb34ZhfOxa433D9gy2I9wD+li2gDq7AGwo QPWZjewX09mIUKTtzSLrdChtgeJmPwNkPRtpYIX1pv/HYHxuNwPg3Pvt/lUGlZ4g08QKBDzD 44t++6lqH98J7ZXMronmkQX+qOIAhKtCwA7fZ3Kw9Uxc5BRBK6Nr8kJfNTAkwFt5+AT23Mb8 /fDkb+bt4atjYjM8OkQMpwjL/LLfg3N6BS5Od0uifuxzngj16jhuUTrblETwzwbsJDlt3BMO DixIiXtRMPEnqNJ5L2EeoqrRC2xEUk16vt2ognetaaEGtm+fN8wbK2TMhq4yHT29EVadpg7D ytNW+tgPLs78rPEhF+5cZoXCC5Q333r+QFplGrvFX8E4JlglmNHTcof1EUdAcmcoaeidrNUY Fe4K2wqCM8dYMmi6leylO+rM7YNoAYfund3sEJby2BMIMEKrGUZg2g/MJLs2uDkc7aQukAKy X+xknE9dL5ryoCeW1M7kqZ8eAlaUvFrqsYOqf95FCUEqpH4C2dnZpAvVPKEkN0C0zYau/hN/ zHEIGKlKlQ9lZYLOkGPLdgBkFR1C4RRy/DjMAbi7SIViLzh9DO7oji6s2IYAXS043KOs4hYG M25PoWIodYLah/fL2O/clV6tv7s1XXY/47Pyd7EH7dcJ+9HEbVvBn7of/BrZDHhUAeqe9vP0 s/TL5S5jsrDb8izeXBZLid2F0/+yOxhg4sFswk8yi+Qloz6AN4S6LIn9mCqxtYpG9xB5lkR8 2iGFifnugWa4Id0oHb8OcsWCqu9xqljM4t1Tfkqok19jeTOSOAjh6CUOv9zzi3lBitpMbH/H pvZEXlp2XvMMte4GByfGtkFpdQaAwPYh/ha+4BZNiUdTx6TMOO3OQLRCzF7iT5lf+I+jHTpu oz1xgsVcInnvEAvp4ujunKyd+Y1js1SsXfyQMDjzNT5iLia0TdsFSc5dBZ7WhDMrWBUQqDW/ jw02sKvY3Oi1znEdoKvo0Lj/eHtiTt6CY8G8fW3AsWB+/u3164N3yWBUpIigq2+BQ71C6flg hPPJMU2Q7mGQ2uhfpyIdEqtay+tyL6NqnsiIYQvb2aA9G1CO2OphgwFJbz6PNw7S1X10uzoR ERXp86QeSSQTr/DmndXfvBO/2a9qGJ+CoY9AP2IAmOjKTDWpd4QUFi5ryASeRxqgjPj+ffnN XjJHhaCOYjx20K2dSEN13Ai+dRQPgOPuD1WLfV3ZTJVhT7cwsfVUK6LUJuL4UXu/rhz52ZWo SLsdDZUn6IPHhmP4HkIuYzTZYkiTt5jFy6b2u2EcXcRHMOaQm8/DBd60NEqJ7R/7Vz4gnOuh wb8NFH7oXLUXQwfiWWBFgxHeKezxBnO65kjpHBmwJsH1GD22KlaGAx0ocYqNWPa6zb4lebqX 9Qdwde3MJmJXUOFbdlUuupQvBWW/uS0ozaoLvulRfDSmZl7Ta+gBCo2wl+acCkV2c3vVPX/Q Vj1grImr/xQGBW6hYrUO97Yb5SzLx/YP1QWM74JhG9FxE3RCyhmPghluvmWyy8utrS3VC91h x/auCNwvw4iJ2UhqFd1QPUyCoj6HoQzHnyCIYTweqV6ckX1OmsUc8fI1eOrDPt1CkAxAZ2ML 3una/85P19ZCPIEju0z2iCD/eT9Ez0HzRxAj9vO6BoQPAqhp/dMZRiHEw+u3v1nh+8K8Onhz +Msh1EN9+84cHL988etBEhLmry2DMRzeqy4uBkClgmTqelwEl1kkxEukeMUfekzCIlkuxjNM blN3crzG4eVKCZmCZHTTaBp12b6zF4nmuvBN6o0JYYqBRhTzJcK0TIyuIEeVIHWSVI1Sd7KE xAxFZonMDA9QWmrqtRD4WWW0QFucv4CJ9yZe+mXkrXqT4nycG8TClZ7AVAWs6MjmisGYoAMa ZnMH4+mtrAu4J9OI4u1ufMUoZLXqsUz21B9fE53cdUSWPip2oyDujyxEGEwoK6jGtPrNseIu j05KP0ueMg98UnqfS2qzipuMisayIlchax0MBsxtG56UN3W6qrS6bruPOKAMtJXnSiVPKci1 sp2qR+pyyV6kfARLJODeWmVz75ZQJ1yepcOdkqG2uw/3G62HD3VCBj/yKOricfT33vpTtRZS A257ZzeJDEbbAnYqn2mEf9t9ePvy5L9fvPntoNb2U00GjKWu7+FrFgeMffCvwHccHtvXv3hj 360QxlSH8QHXmVkyAShcbn74wezW7SYmZ7PcfHxCjR7AanMqzkql6QTdzp/Z4rmpfr+oP774 Tvyv/jeYDHR532zv7u2poIdUAhVR2mdHeUBpXJD+9PPu6zKddD65+EWQqHX8uXiNPYAih7d/ dvHt0INaarXOIOZpJYtR8/JgcCbmbssqpyLY7VVmqQlpziQjcppPTSExPQRimrce1fnkpKno c6/nB+3IaB49nXvORTFt5+vP5jTvxc17c5uP4uajuc0ncfPJ3ObTuPl0bvNPcfNPieb+cCml B2xoVk8du1B9yP/MPuqKx3jNWD9dV0+5OjjSp0vx8sFHcyuOBKHXruyISskihVri+cFjFcTz B/kT4hbg8HBKt5ZQ66bRUecUTu4Cz1jV5gwxry4DaiBOQVTs++Fx0l1+CPVs1zEmHG/TOZIM BhX/6ZQm/10QU8qGgmKMOgVsxTVGlkl/kXKyWLMdo28X58OYA/nMVceikjrkY4fIggF19avV EC8m3eP/ekOFJqgy/DnWcs+nMLw1u2yjTjbprNn13+puYVjS2C4UDGqG1hLsiiK29/Z2G60n Dx+qnMPVVRbnc9P6B+lTivsk9c9WxK/9sSGwmNGY0prReQcR0aAlYlA1Ey8ljVDiX0+xug2+ pnicmUpw1QhFOw22jB2DAhyyYD/fmvxaEOuFBUivwG5BsykfVCgkisNaaVyLVQ68DNi+pCtW MyiTv+uCbZSH1NfovaXqoYVrilBW1xkSy8Re3hXUiLJR7dZvL+MT2+cHVpNflxHwXmBvxlsn 3USyXbdEhF+7IWunn9coUs4eOueOFoCSsFukoXmy8lxzB35oBd1APbO0gqCeWVpLUM8srSqo Z5bWF9Qzn1TYcojlAQE1mAD1NGRSzDZ8RlHhgNIXql9Xe8A9W0R+lof5m+hp1Q6r94mdQQC9 WIFBi0lbVe27CQ7zXHrZjGhRfj5wliYAsxCzEtoRGxg+x0MBQ7vwIBhCbhSxLmXORI/MUq/V 7ym8ZZEtk2weJbrhEaN2gGNi0u9IYF5s1pGIQ8LXsDxXK4pjD2lBBhRK/CS7rHwT9xjopxyB GMfFYVjoyIw/dSeIuCl9uXorqhMHpCZrJaQCllkCC0cf8nhku5lxcJGLEqE+MGxvA5XhDa8c Y1aUClV2c1HxSAESPu/OhgsSFqNQEIOCqPJAm2Dx4kdWskACB3dPBnZCJSKCRgsiQIJWIXxD 6FoO8U+czf2B2xdVWiWqEKgjQChw0zV1qArvuleDzG4cXFpc5zntMLqvMa6O0nIdihf7t4Ck 354cPCWcJfV0zX5sxE3Wh0hf6b1dx/DSCb1VcDbNAYW/m5qLM1z/53r9KX/bND9jYN1TY3rG 7FT+iT9P7Y+WMf+s2N93XcsXSDjcErDG4BeDTU+3VVPkXJhT6g8CFpX0f/4QxUqq77jaJAl5 52wQc7tr9gFctuIPKCHRxENMrQ6Dha9xnFVyL04sMedoK43i4lW86qtsOhve8/GqZFv1Sc1Y LBwdXWejccO8Hl5M4HppNnqWgnr2k7eWWU1vrvrjK8tQht6NFqYwe3zc3Val0ht0zhxyF45A qmviJ7j83MS4Sp2H8CyX6oRl5Y8Pxr0a2xmTHsDV3H+qDrzz80G9Wo7F5yLwdT8KP87nprnD O19yDZV7Zg2nXGfj/pdCVz8IXdzlhPojynWn4vTwHldduRqldweZ3Q0HGo2WASRPyDGVhMtk PN+8cL6yaD68qI8G2eSimzbty/p4LGMmAbucWXiA4sZ/AXoxWNEtWi1/EDbjRZOvAoQe+B9k vQLPVAlFEGm1nstZJ+ONYOaInuJyzl/PRpiaDJ20anyMnXJvGeKv5EukhZ0N/RbUg2Zvz3+3 8z3suHaeC9TrerQnEypcggmzptPvYKEJTuAOAnVopI5nKiKpPg+Edp0gFsP920zv3+GRvfAe vjr7+fDoxbv/c/bu4Nd3B8cHRycvTg7fHrkpqXMCpdkhu4aXUwIkhhlXa+eZFWxJmwW0qU2F FBNm+wVaVi7angvJ+ZexHiUdYA/RAbbXEgeYof/cDTKLWQKZxdSml7OcCa7Padlc5YSB/DAq YOteE01gt0ZwaS6B4NKcj+BCX/92dHLw7hdLoScHryyjOXjzCi5blKLqQMMzB1KeyZSQm6tZ HI6g4v0qODA1yDBk9Lz+oFunfgJ0GLVId4AO07wjdJjmHaHDNFdChzFpdJh7SwDEQFGB5TFi AuwFdzcK04RuAR+TgIwxBXwGj5KA1uEnD/carSePHCLdX/g8l2yPSs5K7Ux+F1sT7c1XJWh9 bX7W12x5mqXvWRJ4tO1YegomiGGLi6n17qs8uy5+V1A/vwW6UHMOulBzHrqQidGFQLefjy5k zJLYPHijVxbD5ZB5HAirmY20FFAxd2KAR1y/AujH1yAS+ZeT9cJKjWAMIE2u+Av8BCPviIs8 3m3sbj/cdT6muyKhzdtlSW2unDW3eUd5VRF45VL5U9/itrUot2luGGoiBFV7hbefObfpygcI NHL8DwS4EaxERydOKsVZTH+K4BceOwW0vrrzGx9b0v1NbRFb01GtPYSBOzwUEeAKp/vIHTjD qaM4zhQ/9eD0hRhT/NQP3aOgU/PknWv5W7O7P/p7s1Ks4Xah2Ym7b+kRpZIevs3YFobIe4/Y 4gB5zew3jfg7/mS+f6vgj6Xpv3or+q/eFf1Xk/RfdfRf/J5irPU/t0yOSikw2F03ppeUqxOW UiqNwyFLFjzhSnuEDNq5gt2brdJiL25RVDfGKss7S8KM/fon0iVKsiVWTZa4da6Em/y8RIdk msOy0etLAGDapSM3VPfr9JukyvyotW1V5taeqMzsmUq5YwHLy31GKnkBHn/5AK8osiMRRMjf hJNX2Pqgr+1uP96x+ppHcfy241926Nrjutqk0pu0a+81j3ceq1jtpPJZpBSY1u2C7pomkqxN ozXwoinIXsPlBrMit24uHarHg5rDqdOBezD2O+DVzYTHpMmBe+F3vCe0InMC90za9lp0iJm7 dYg51hg5xgz6vuCT4bjjCz0UTTXLusn+5b1k1aW9ZOav6CUrmimWcZXdxn5RaBBWZ5jjOFMc SDnEiAfFzrPw/cEDzRQtNYtL31xp6Zure9BKFjE4KIm1LPrS3H6u7EqrlrjSdmsxAyhzpbnB 1YO28/xpQUM84rqVZSj1ujJzreRwa3qHW4HA2OsWLPcqG7/I9daMD98i1xtNMSEMYs9bM6rR YI7Bx4HeLjtyOMSjKcFWWEUOIqncPBuk1KvMdhdgZo8Wq0Z7Txq7Ow8fqXDpu9MavMKAjZOK Dk8uJeE8urSbkg8aA3yLwQ2ZpSzXgj+ngBidz65gO6mbHqEEAOIaHYot82Igyfn5VJyV9F3u XC6ME4jIo7iG1FsaCK0ASF1wm1jxiVAvK5jlEyb3+XJ2Hoq+RwdDNYZKyy2T+OkeBG3LPsam MYfupfl+GoTrXrVE2pgSo/hdWAT5jf/LgxxSoW9APQMXC5SBZhscQYD5Ujs0kNug/fBLL8bo cNVl0HyprRqvo/EYQABYXnP74iGACsg/wa33D3+gaK1SqC8mwsDBinKUwJ5Cu1mckKqGKa/m o12GE1LECKEqfXZ1xNQgCpGzY8wBFviDG+APoxP221h2tcfFVwDVznLGNgQTCFSqCyQE/Lp2 BkW5APNcusIi8aQwIT7j6OY6u6nzt2Ln1isuW9n0y4IrvhhDQHX3zZZ8h+OxyHBdvW1ulISv lFm4jMMS8OMsoAdEdqyvS2oqyWf6BuIxecne395u2P+w8xAl6G7LStCH9j6kjAt+wKPZMKcb 3CZETdlBbsIgAZuWQlrQCwBwjLN8Oh4aHoqLo2HrXI/VM/i7fTmp1beonxe2oZV3LASnGvGW WnJ0AFUeRk6I0Gn0tMQCGMseGLNmCrp3UJXRFTjGKlAYLNvP6Xn+AtmnrCHKUOCyvAbXEG7t nqNHePCHHKJjNTt8issjC7pbxn70Xh+iIhx4Ij1LfF3KRsjLpBJqaiYI/ZVLBWUaAo+jR3gP WOLQ1Un0DnlWKTAAPCilCHUsqIuPIwxOyqbKky/1/zAgAwJeANVZAQznUAmvPwBac1PCxEoz tIpfH//0zQHyGkC1zV7z3L7jxfHLw0MVBrLl6IphoC0fAh53dXEmoz0bnsOpZQQLoReWVBCG gj34+Hy1QgKdjX9VNEQH1nju2R9Y8qPtwKkZTVdiTeg5JAJuDzyaArtmWNJc4Mn1VjNFEBIv v7xSQXJ79fbg+Gj9BHG7GYSSRAr0G+wxOLpYJdsMxa5WvnIOacCislw9hF1s6GAL2hKcCx5o MkpTcUy30EiGWE6iUPYPHTzksmOX7TnJk8JGxb02TFhyc6Oe12UQ5Jd0sfrP7BixKoVVFIFJ mx+MpG2Ad9p+YKUFvoT9TV89CPZqib2jZt9P9e7r5idfRzQ3T0mHwUfQbeTcd/zosDsE/pY3 YCNI2m0K84w3Uqqbl+4j/SZgiviXVDtfdosrf4k95uWdu8uLt/mPoi8UHuS9knQqMkvPr5/i +qFnSwuoUF/pAiq6j+qi9/v4CPh3R0Qr/74UqRHUVWbfub3OwrUQ2PH5Df0k7/kfCzYlffYg qeL/n9344n3aX/xxiXoxTqf39YzhsdJuVQDyUrwk0q6OBf5oB4RIy8u03HhUuzA4hSHoQdRg FLIoOBNSm1EHA2Q9kMzNUDRLJKdWJBkGnnqhEthz5VHBPiB8rL0jv7RWZFppumzvMJHB7635 FFrYCnoMxeyCnfD///8AC+cnr9E3AQA= --B_3205479250_19305824-- From pgsql-performance-owner@postgresql.org Fri Jul 29 15:30:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 173C852831 for ; Fri, 29 Jul 2005 15:30:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 08997-05 for ; Fri, 29 Jul 2005 18:29:55 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id B70BA52833 for ; Fri, 29 Jul 2005 15:29:53 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050729182954m9200g4ujse>; Fri, 29 Jul 2005 18:29:54 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id ADDE556021; Fri, 29 Jul 2005 13:29:53 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 51AF756006; Fri, 29 Jul 2005 13:29:49 -0500 (CDT) Message-ID: <42EA759A.60807@arbash-meinel.com> Date: Fri, 29 Jul 2005 13:29:46 -0500 From: John Arbash Meinel User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: josh@agliodbs.com Cc: Dennis Bjorklund , Matthew Schumacher , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <200507291038.04230.josh@agliodbs.com> In-Reply-To: <200507291038.04230.josh@agliodbs.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig541C882B4147AB26BC6EC16A" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.147 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, UPPERCASE_25_50 X-Spam-Level: X-Archive-Number: 200507/480 X-Sequence-Number: 13721 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig541C882B4147AB26BC6EC16A Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Josh Berkus wrote: >Dennis, > > > >> EXCEPTION WHEN unique_violation THEN >> >> > >I seem to remember that catching an exception in a PL/pgSQL procedure was a >large performance cost. It'd be better to do UPDATE ... IF NOT FOUND. > > > Actually, he was doing an implicit UPDATE IF NOT FOUND in that he was doing: UPDATE IF found THEN return; INSERT EXCEPT ... So really, the exception should never be triggered. John =:-> --------------enig541C882B4147AB26BC6EC16A Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC6nWcJdeBCYSNAAMRAvvHAJwPBkfZaj3d2C8AOyUIh1RoJlRCsQCeOFe7 1Dh31ZIEpB4BDhNBSQ3iIgw= =j02d -----END PGP SIGNATURE----- --------------enig541C882B4147AB26BC6EC16A-- From pgsql-performance-owner@postgresql.org Fri Jul 29 15:56:07 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 5A3C45293C for ; Fri, 29 Jul 2005 15:56:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 14457-05 for ; Fri, 29 Jul 2005 18:55:56 +0000 (GMT) Received: from gghcwest.com (adsl-71-128-90-172.dsl.pltn13.pacbell.net [71.128.90.172]) by svr1.postgresql.org (Postfix) with ESMTP id CC3C25281E for ; Fri, 29 Jul 2005 15:55:54 -0300 (ADT) Received: from toonses.gghcwest.com (toonses.gghcwest.com [192.168.168.115]) by gghcwest.com (8.12.10/8.12.9) with ESMTP id j6TItimd028952; Fri, 29 Jul 2005 11:55:47 -0700 Received: from jwb by toonses.gghcwest.com with local (Exim 4.50) id 1Dya1H-0001nm-2f; Fri, 29 Jul 2005 11:55:43 -0700 Subject: Re: Performance problems on 4/8way Opteron (dualcore) HP From: "Jeffrey W. Baker" To: josh@agliodbs.com Cc: pgsql-performance@postgresql.org, Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= , Sven Geisler In-Reply-To: <200507291046.10318.josh@agliodbs.com> References: <42EA5D18.9020204@aeccom.com> <200507291046.10318.josh@agliodbs.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Date: Fri, 29 Jul 2005 11:55:42 -0700 Message-Id: <1122663342.6814.5.camel@toonses.gghcwest.com> Mime-Version: 1.0 X-Mailer: Evolution 2.3.6 X-Virus-Scanned: by amavisd-new at hub.org X-Archive-Number: 200507/481 X-Sequence-Number: 13722 On Fri, 2005-07-29 at 10:46 -0700, Josh Berkus wrote: > Dirk, > > > does anybody have expierence with this machine (4x 875 dual core Opteron > > CPUs)? I'm using dual 275s without problems. > Nope. I suspect that you may be the first person to report in on > dual-cores. There may be special compile issues with dual-cores that > we've not yet encountered. Doubtful. However you could see improvements using recent Linux kernel code. There have been some patches for optimizing scheduling and memory allocations. However, if you are running this machine in 32-bit mode, why did you bother paying $14,000 for your CPUs? You will get FAR better performance in 64-bit mode. 64-bit mode will give you 30-50% better performance on PostgreSQL loads, in my experience. Also, if I remember correctly, the 32-bit x86 kernel doesn't understand Opteron NUMA topology, so you may be seeing poor memory allocation decisions. -jwb > > We run RHEL 3.0, 32bit and under high load it is a drag. We > > mostly run memory demanding queries. Context switches are pretty much > > around 20.000 on the average, no cs spikes when we run many processes in > > parallel. Actually we only see two processes in running state! When > > there are only a few processes running context switches go much higher. > > At the moment we are much slower that with a 4way XEON box (DL580). > > Um, that was a bit incoherent. Are you seeing a CS storm or aren't you? > From pgsql-performance-owner@postgresql.org Fri Jul 29 16:02:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id F123D5298C for ; Fri, 29 Jul 2005 16:02:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 16292-08 for ; Fri, 29 Jul 2005 19:02:42 +0000 (GMT) Received: from mail.Mi8.com (d01gw01.mi8.com [63.240.6.47]) by svr1.postgresql.org (Postfix) with ESMTP id CC35E52941 for ; Fri, 29 Jul 2005 16:02:40 -0300 (ADT) Received: from 172.16.1.118 by mail.Mi8.com with ESMTP (- Welcome to Mi8 Corporation www.Mi8.com (D1)); Fri, 29 Jul 2005 15:02:31 -0400 X-Server-Uuid: 241911D6-425B-44B9-A073-E3FE0F8FC774 Received: from MI8NYCMAIL06.Mi8.com ([172.16.1.175]) by d01smtp03.Mi8.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 29 Jul 2005 15:02:31 -0400 Received: from 67.103.45.218 ([67.103.45.218]) by MI8NYCMAIL06.Mi8.com ( [172.16.1.219]) via Exchange Front-End Server mi8owa.mi8.com ( [172.16.1.104]) with Microsoft Exchange Server HTTP-DAV ; Fri, 29 Jul 2005 15:02:30 -0500 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 29 Jul 2005 12:02:30 -0700 Subject: Re: Performance problems testing with Spamassassin From: "Luke Lonergan" To: "Tom Lane" , "Matthew Schumacher" Cc: pgsql-performance@postgresql.org Message-ID: In-Reply-To: <20875.1122531559@sss.pgh.pa.us> MIME-Version: 1.0 X-OriginalArrivalTime: 29 Jul 2005 19:02:31.0074 (UTC) FILETIME=[12180820:01C59470] X-WSS-ID: 6EF4A2CD2B48673350-01-01 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.576 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO, RCVD_NUMERIC_HELO X-Spam-Level: * X-Archive-Number: 200507/482 X-Sequence-Number: 13723 Tom, On 7/27/05 11:19 PM, "Tom Lane" wrote: > Matthew Schumacher writes: >> After playing with various indexes and what not I simply am unable to >> make this procedure perform any better. Perhaps someone on the list can >> spot the bottleneck and reveal why this procedure isn't performing that >> well or ways to make it better. > > There's not anything obviously wrong with that procedure --- all of the > updates are on primary keys, so one would expect reasonably efficient > query plans to get chosen. Perhaps it'd be worth the trouble to build > the server with profiling enabled and get a gprof trace to see where the > time is going. Yes - that would be excellent. We've used oprofile recently at Mark Wong's suggestion, which doesn't require rebuilding the source. - Luke From pgsql-performance-owner@postgresql.org Fri Jul 29 16:17:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E2579529E7 for ; Fri, 29 Jul 2005 16:17:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 19748-04 for ; Fri, 29 Jul 2005 19:16:54 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 8EDDE52917 for ; Fri, 29 Jul 2005 16:16:53 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6TJUQ9k003626 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Fri, 29 Jul 2005 11:30:27 -0800 Message-ID: <42EA80A4.7070705@aptalaska.net> Date: Fri, 29 Jul 2005 11:16:52 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin References: In-Reply-To: X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: unavailable X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/483 X-Sequence-Number: 13724 Ok, Here is something new, when I take my data.sql file and add a begin and commit at the top and bottom, the benchmark is a LOT slower? My understanding is that it should be much faster because fsync isn't called until the commit instead of on every sql command. I must be missing something here. schu From pgsql-performance-owner@postgresql.org Fri Jul 29 16:22:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 926D952AB8 for ; Fri, 29 Jul 2005 16:21:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 21752-07 for ; Fri, 29 Jul 2005 19:21:54 +0000 (GMT) Received: from mx1.neopolitan.us (mx1.neopolitan.us [65.87.16.224]) by svr1.postgresql.org (Postfix) with ESMTP id 0CADD52AC2 for ; Fri, 29 Jul 2005 16:21:53 -0300 (ADT) Received: from [65.87.16.98] (HELO [10.0.0.221]) by mx1.neopolitan.us (CommuniGate Pro SMTP 4.2.10) with ESMTP id 10168090; Fri, 29 Jul 2005 12:21:54 -0700 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Fri, 29 Jul 2005 12:21:53 -0700 Subject: Re: Performance problems on 4/8way Opteron (dualcore) From: "J. Andrew Rogers" To: , Cc: Dirk Lutzeb=?ISO-8859-1?B?5A==?=ck , Sven Geisler Message-ID: In-Reply-To: <200507291046.10318.josh@agliodbs.com> Mime-version: 1.0 Content-type: text/plain; charset="US-ASCII" Content-transfer-encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.312 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, DNS_FROM_RFC_POST X-Spam-Level: * X-Archive-Number: 200507/484 X-Sequence-Number: 13725 On 7/29/05 10:46 AM, "Josh Berkus" wrote: >> does anybody have expierence with this machine (4x 875 dual core Opteron >> CPUs)? > > Nope. I suspect that you may be the first person to report in on > dual-cores. There may be special compile issues with dual-cores that > we've not yet encountered. There was recently a discussion of similar types of problems on a couple of the supercomputing lists, regarding surprisingly substandard performance from large dual-core opteron installations. The problem as I remember it boiled down to the Linux kernel handling memory/process management very badly on large dual core systems -- pathological NUMA behavior. However, this problem has apparently been fixed in Linux v2.6.12+, and using the more recent kernel on large dual core systems generated *massive* performance improvements on these systems for the individuals with this issue. Using the patched kernel, one gets the performance most people were expecting. The v2.6.12+ kernels are a bit new, but they contain a very important performance patch for systems like the one above. It would definitely be worth testing if possible. J. Andrew Rogers From pgsql-performance-owner@postgresql.org Fri Jul 29 16:24:48 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D231652933 for ; Fri, 29 Jul 2005 16:24:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 21892-04 for ; Fri, 29 Jul 2005 19:24:35 +0000 (GMT) Received: from boutiquenumerique.com (boutiquenumerique.com [82.67.9.10]) by svr1.postgresql.org (Postfix) with ESMTP id 8FD9352A80 for ; Fri, 29 Jul 2005 16:24:33 -0300 (ADT) Received: (qmail 1771 invoked from network); 29 Jul 2005 21:24:39 +0200 Received: from unknown (HELO localhost) (boutiquenumerique-lists@192.168.0.4) by boutiquenumerique.com with SMTP; 29 Jul 2005 21:24:39 +0200 To: "Matthew Schumacher" , "Andrew McMillan" Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> Message-ID: Date: Fri, 29 Jul 2005 21:24:32 +0200 From: PFC Organization: =?iso-8859-15?Q?La_Boutique_Num=E9rique?= Content-Type: text/plain; format=flowed; delsp=yes; charset=iso-8859-15 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit In-Reply-To: <42EA6966.8090402@aptalaska.net> User-Agent: Opera M2/8.0 (Linux, build 1095) X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.005 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/485 X-Sequence-Number: 13726 > Also, this test goes a bit faster with sync turned off, if mysql isn't > using sync that would be why it's so much faster. Anyone know what the > default for mysql is? For InnoDB I think it's like Postgres (only slower) ; for MyISAM it's no fsync, no transactions, no crash tolerance of any kind, and it's not a default value (in the sense that you could tweak it) it's just the way it's coded. From pgsql-performance-owner@postgresql.org Fri Jul 29 17:20:12 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id DEA02529F3 for ; Fri, 29 Jul 2005 17:20:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34528-09 for ; Fri, 29 Jul 2005 20:20:05 +0000 (GMT) Received: from mail1.catalyst.net.nz (godel.catalyst.net.nz [202.78.240.40]) by svr1.postgresql.org (Postfix) with ESMTP id 808D7529FA for ; Fri, 29 Jul 2005 17:20:03 -0300 (ADT) Received: from 222-152-153-141.jetstream.xtra.co.nz ([222.152.153.141] helo=lamb.mcmillan.net.nz) by mail1.catalyst.net.nz with esmtpsa (TLS-1.0:DHE_RSA_AES_256_CBC_SHA:32) (Exim 4.50) id 1DybKt-0004Kk-4e; Sat, 30 Jul 2005 08:20:03 +1200 Received: from lamb.mcmillan.net.nz (lamb.mcmillan.net.nz [127.0.0.1]) by lamb.mcmillan.net.nz (Postfix) with ESMTP id 0B4387A0056; Sat, 30 Jul 2005 08:19:31 +1200 (NZST) Subject: Re: Performance problems testing with Spamassassin 3.1.0 From: Andrew McMillan To: Matthew Schumacher Cc: pgsql-performance@postgresql.org In-Reply-To: <42EA6966.8090402@aptalaska.net> References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="=-DNVOu4hRQ9hQ5n4gHR3j" Date: Sat, 30 Jul 2005 08:19:31 +1200 Message-Id: <1122668371.5691.352.camel@lamb.mcmillan.net.nz> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/486 X-Sequence-Number: 13727 --=-DNVOu4hRQ9hQ5n4gHR3j Content-Type: text/plain Content-Transfer-Encoding: quoted-printable On Fri, 2005-07-29 at 09:37 -0800, Matthew Schumacher wrote: > >=20 > > On my laptop this takes: > >=20 > > real 1m33.758s > > user 0m4.285s > > sys 0m1.181s > >=20 > > One interesting effect is the data in bayes_vars has a huge number of > > updates and needs vacuum _frequently_. After the run a vacuum full > > compacts it down from 461 pages to 1 page. > >=20 >=20 > I wonder why your laptop is so much faster. My 2 min 30 sec test was > done on a dual xeon with a LSI megaraid with 128MB cache and writeback > caching turned on. I only do development stuff on my laptop, and all of my databases are reconstructable from copies, etc... so I turn off fsync in this case. > How often should this table be vacuumed, every 5 minutes? I would be tempted to vacuum after each e-mail, in this case. > Also, this test goes a bit faster with sync turned off, if mysql isn't > using sync that would be why it's so much faster. Anyone know what the > default for mysql is? It depends on your table type for MySQL. For the data in question (i.e. bayes scoring) it would seem that not much would be lost if you did have to restore your data from a day old backup, so perhaps fsync=3Dfalse is OK for this particular application. Regards, Andrew McMillan. ------------------------------------------------------------------------- Andrew @ Catalyst .Net .NZ Ltd, PO Box 11-053, Manners St, Wellington WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St DDI: +64(4)803-2201 MOB: +64(272)DEBIAN OFFICE: +64(4)499-2267 What we wish, that we readily believe. -- Demosthenes ------------------------------------------------------------------------- --=-DNVOu4hRQ9hQ5n4gHR3j Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQBC6o9TjJA0f48GgBIRAvYxAJ0bPLVgHixIXfve5oiOg+k2aFv0+QCfTqh6 4bza1zu2EkvnPnd0PAJFz+c= =ICIU -----END PGP SIGNATURE----- --=-DNVOu4hRQ9hQ5n4gHR3j-- From pgsql-performance-owner@postgresql.org Fri Jul 29 17:31:02 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 887D1529C0 for ; Fri, 29 Jul 2005 17:30:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 38606-04 for ; Fri, 29 Jul 2005 20:30:58 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 05C9F52938 for ; Fri, 29 Jul 2005 17:30:57 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6TKhsEn019125 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Fri, 29 Jul 2005 12:43:55 -0800 Message-ID: <42EA91DC.9030306@aptalaska.net> Date: Fri, 29 Jul 2005 12:30:20 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Andrew McMillan Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> In-Reply-To: <1122668371.5691.352.camel@lamb.mcmillan.net.nz> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/487 X-Sequence-Number: 13728 Andrew McMillan wrote: > > For the data in question (i.e. bayes scoring) it would seem that not > much would be lost if you did have to restore your data from a day old > backup, so perhaps fsync=false is OK for this particular application. > > Regards, > Andrew McMillan. > Restoring from the previous days backup is plenty exceptable in this application, so is it possible to turn fsync off, but only for this database, or would I need to start another psql instance? schu From pgsql-performance-owner@postgresql.org Fri Jul 29 18:02:15 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 941D652AE1 for ; Fri, 29 Jul 2005 18:02:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 45261-01 for ; Fri, 29 Jul 2005 21:02:07 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 79788529C0 for ; Fri, 29 Jul 2005 18:02:05 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6TL1vHT019406; Fri, 29 Jul 2005 17:01:58 -0400 (EDT) To: Andrew McMillan Cc: Matthew Schumacher , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 In-reply-to: <1122668371.5691.352.camel@lamb.mcmillan.net.nz> References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> Comments: In-reply-to Andrew McMillan message dated "Sat, 30 Jul 2005 08:19:31 +1200" Date: Fri, 29 Jul 2005 17:01:57 -0400 Message-ID: <19405.1122670917@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/488 X-Sequence-Number: 13729 Andrew McMillan writes: > On Fri, 2005-07-29 at 09:37 -0800, Matthew Schumacher wrote: >> How often should this table be vacuumed, every 5 minutes? > I would be tempted to vacuum after each e-mail, in this case. Perhaps the bulk of the transient states should be done in a temp table, and only write into a real table when you're done? Dropping a temp table is way quicker than vacuuming it ... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Jul 29 18:48:13 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 6836F52A1B for ; Fri, 29 Jul 2005 18:48:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 51084-09 for ; Fri, 29 Jul 2005 21:48:01 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 3DEC752831 for ; Fri, 29 Jul 2005 18:48:00 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6TM1ZEQ003150 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Fri, 29 Jul 2005 14:01:35 -0800 Message-ID: <42EAA410.9010700@aptalaska.net> Date: Fri, 29 Jul 2005 13:48:00 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> In-Reply-To: <19405.1122670917@sss.pgh.pa.us> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: unavailable X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests= X-Spam-Level: X-Archive-Number: 200507/489 X-Sequence-Number: 13730 Ok, here is where I'm at, I reduced the proc down to this: CREATE FUNCTION update_token (_id INTEGER, _token BYTEA, _spam_count INTEGER, _ham_count INTEGER, _atime INTEGER) RETURNS VOID AS $$ BEGIN LOOP UPDATE bayes_token SET spam_count = spam_count + _spam_count, ham_count = ham_count + _ham_count, atime = _atime WHERE id = _id AND token = _token; IF found THEN RETURN; END IF; INSERT INTO bayes_token VALUES (_id, _token, _spam_count, _ham_count, _atime); IF FOUND THEN UPDATE bayes_vars SET token_count = token_count + 1 WHERE id = _id; IF NOT FOUND THEN RAISE EXCEPTION 'unable to update token_count in bayes_vars'; return FALSE; END IF; RETURN; END IF; RETURN; END LOOP; END; $$ LANGUAGE plpgsql; All it's doing is trying the update before the insert to get around the problem of not knowing which is needed. With only 2-3 of the queries implemented I'm already back to running about the same speed as the original SA proc that is going to ship with SA 3.1.0. All of the queries are using indexes so at this point I'm pretty convinced that the biggest problem is the sheer number of queries required to run this proc 200 times for each email (once for each token). I don't see anything that could be done to make this much faster on the postgres end, it's looking like the solution is going to involve cutting down the number of queries some how. One thing that is still very puzzling to me is why this runs so much slower when I put the data.sql in a transaction. Obviously transactions are acting different when you call a proc a zillion times vs an insert query. Anyway, if anyone else has any ideas I'm all ears, but at this point it's looking like raw query speed is needed for this app and while I don't care for mysql as a database, it does have the speed going for it. schu From pgsql-performance-owner@postgresql.org Fri Jul 29 20:19:08 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7A7AF52A23 for ; Fri, 29 Jul 2005 20:19:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 67627-08 for ; Fri, 29 Jul 2005 23:18:57 +0000 (GMT) Received: from netbox.unitech.com.ar (unknown [200.32.92.34]) by svr1.postgresql.org (Postfix) with ESMTP id 8CDFD529FA for ; Fri, 29 Jul 2005 20:18:54 -0300 (ADT) Received: from dariop (dariop.unitech.com.ar [192.168.1.237]) by netbox.unitech.com.ar (8.11.6/8.11.6) with ESMTP id j6U0IBE25796; Fri, 29 Jul 2005 20:18:12 -0400 Received: from 127.0.0.1 (AVG SMTP 7.0.338 [267.9.7]); Fri, 29 Jul 2005 20:18:56 -0300 From: "Dario" To: "Chris Travers" Cc: Subject: Re: Left joining against two empty tables makes a query Date: Fri, 29 Jul 2005 20:18:56 -0300 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) In-Reply-To: <42E9BD43.6060203@travelamericas.com> X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4952.2800 Importance: Normal X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.071 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/490 X-Sequence-Number: 13731 Sorry for my english. May I ask? (i'm still learning postgresql). Isn't outer join forcing "join order"? The planner will resolve a, then ac in order to resolve left join previously and will not be able to choose the customer_id filter (more selective)... AFAIK (not too far :-)) this will be the join order, even if projects and deparments are not empty, no matter how much statistical info you (the engine) have (has). Workaround: You should probably try to use a subquery to allow planner to choose join order (as long as you can modify source code :-O ). You know project and department are empty now so... SELECT aa.accno, aa.description, aa.link, aa.category, aa.project_id, aa.department, p.projectnumber, d.description from ( SELECT c.accno, c.description, c.link, c.category, ac.project_id, a.department_id AS department FROM chart c JOIN acc_trans ac ON (ac.chart_id = c.id) JOIN ar a ON (a.id = ac.trans_id) WHERE a.customer_id = 11373 AND a.id IN ( SELECT max(id) FROM ar WHERE customer_id = 11373) ) aa LEFT JOIN project p ON (aa.project_id = p.id) LEFT JOIN department d ON (d.id = aa.department) Doubt of it. I rewrite it at first sight. Long life, little spam and prosperity. -----Mensaje original----- De: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org]En nombre de Chris Travers Enviado el: viernes, 29 de julio de 2005 2:23 Para: Gnanavel S CC: Chris Travers; pgsql-performance@postgresql.org Asunto: Re: [PERFORM] Left joining against two empty tables makes a query > > Secondly, the project table has *never* had anything in it. So where > are these numbers coming from? > > > pg_statistics I very much doubt that. I was unable to locate any rows in pg_statistic where the pg_class.oid for either table matched any row's starelid. Tom's argument that this is behavior by design makes sense. I assumed that something like that had to be going on, otherwise there would be nowhere for the numbers to come from. I.e. if there never were any rows in the table, then if pg_statistic is showing 1060 rows, we have bigger problems than a bad query plan. I hope however that eventually tables which are truly empty can be treated intelligently sometime in the future in Left Joins. Otherwise this limits the usefulness of out of the box solutions which may have functionality that we don't use. Such solutions can then kill the database performance quite easily. Chris Travers Metatron Technology Consulting ---------------------------(end of broadcast)--------------------------- TIP 1: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to majordomo@postgresql.org so that your message can get through to the mailing list cleanly From pgsql-performance-owner@postgresql.org Fri Jul 29 21:41:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 9FBF052A8C for ; Fri, 29 Jul 2005 21:41:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 84202-10 for ; Sat, 30 Jul 2005 00:41:48 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id 4974252A89 for ; Fri, 29 Jul 2005 21:41:47 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id 150E135F9DE for ; Sat, 30 Jul 2005 00:41:51 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 90067-01; Sat, 30 Jul 2005 00:41:34 +0000 (GMT) Received: from [10.0.0.7] (63-228-143-68.phnx.qwest.net [63.228.143.68]) by veggie-fuel.com (Postfix) with ESMTP id 20B9735F9D6; Sat, 30 Jul 2005 00:41:35 +0000 (GMT) Subject: Re: Performance problems testing with Spamassassin From: Karim Nassar Reply-To: karim.nassar@acm.org To: Josh Berkus Cc: Matthew Schumacher , Andrew McMillan , Luke Lonergan , pgsql-performance@postgresql.org In-Reply-To: <200507290947.03700.josh@agliodbs.com> References: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> <200507290947.03700.josh@agliodbs.com> Content-Type: text/plain Date: Fri, 29 Jul 2005 17:39:41 -0700 Message-Id: <1122683983.11869.150.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.027 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/491 X-Sequence-Number: 13732 On Fri, 2005-07-29 at 09:47 -0700, Josh Berkus wrote: > Try changing: > wal_buffers = 256 > > and try Bruce's stop full_page_writes patch. > > > I guess we see the real culprit here. Anyone surprised it's the WAL? > > Nope. On high-end OLTP stuff, it's crucial that the WAL have its own > dedicated disk resource. > > Also, running a complex stored procedure for each and every word in each > e-mail is rather deadly ... with the e-mail traffic our server at Globix > receives, for example, that would amount to running it about 1,000 times a > minute. Is this a real-world fix? Seems to me that Spam Assassin runs on a plethora of mail servers, and optimizing his/her/my/your pg config doesn't solve the root problem: there are thousands of (seemingly) high-overhead function calls being executed. > It would be far better to batch this, somehow, maybe using temp > tables. Agreed. On my G4 laptop running the default configured Ubuntu Linux postgresql 7.4.7 package, it took 43 minutes for Matthew's script to run (I ran it twice just to be sure). In my spare time over the last day, I created a brute force perl script that took under 6 minutes. Am I on to something, or did I just optimize for *my* system? http://ccl.cens.nau.edu/~kan4/files/k-bayesBenchmark.tar.gz kan4@slap-happy:~/k-bayesBenchmark$ time ./test.pl <-- snip db creation stuff --> 17:18:44 -- START 17:19:37 -- AFTER TEMP LOAD : loaded 120596 records 17:19:46 -- AFTER bayes_token INSERT : inserted 49359 new records into bayes_token 17:19:50 -- AFTER bayes_vars UPDATE : updated 1 records 17:23:37 -- AFTER bayes_token UPDATE : updated 47537 records DONE real 5m4.551s user 0m29.442s sys 0m3.925s I am sure someone smarter could optimize further. Anyone with a super-spifty machine wanna see if there is an improvement here? -- Karim Nassar From pgsql-performance-owner@postgresql.org Sat Jul 30 04:57:55 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 8C4B052BAA for ; Sat, 30 Jul 2005 04:57:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 85800-03 for ; Sat, 30 Jul 2005 07:57:44 +0000 (GMT) Received: from floppy.pyrenet.fr (floppy.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3F92E528C7 for ; Sat, 30 Jul 2005 04:57:43 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id 39AF230B42; Sat, 30 Jul 2005 10:05:38 +0200 (MET DST) From: William Yu X-Newsgroups: pgsql.performance Subject: Re: Performance problems on 4/8way Opteron (dualcore) Date: Sat, 30 Jul 2005 00:57:38 -0700 Organization: Hub.Org Networking Services Lines: 20 Message-ID: References: <200507291046.10318.josh@agliodbs.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en In-Reply-To: To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.018 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/492 X-Sequence-Number: 13733 I've been running 2x265's on FC4 64-bit (2.6.11-1+) and it's been running perfect. With NUMA enabled, it runs incrementally faster than NUMA off. Performance is definitely better than the 2x244s they replaced -- how much faster, I can't measure since I don't have the transaction volume to compare to previous benchmarks. I do see more consistently low response times though, can run apache also on the server for faster HTML generation times and top seems to show in general twice as much CPU power idle on average (25% per 265 core versus 50% per 244.) I haven't investigated the 2.6.12+ kernel updates yet -- I probably will do our development servers first to give it a test. > The problem as I remember it boiled down to the Linux kernel handling > memory/process management very badly on large dual core systems -- > pathological NUMA behavior. However, this problem has apparently been fixed > in Linux v2.6.12+, and using the more recent kernel on large dual core > systems generated *massive* performance improvements on these systems for > the individuals with this issue. Using the patched kernel, one gets the > performance most people were expecting. From pgsql-performance-owner@postgresql.org Sat Jul 30 05:46:44 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 78CA952C16 for ; Sat, 30 Jul 2005 05:46:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 94638-05 for ; Sat, 30 Jul 2005 08:46:32 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 48C4652BF8 for ; Sat, 30 Jul 2005 05:46:31 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6U908jK000928 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sat, 30 Jul 2005 01:00:09 -0800 Message-ID: <42EB3E63.6040300@aptalaska.net> Date: Sat, 30 Jul 2005 00:46:27 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: karim.nassar@acm.org Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin References: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> <200507290947.03700.josh@agliodbs.com> <1122683983.11869.150.camel@localhost.localdomain> In-Reply-To: <1122683983.11869.150.camel@localhost.localdomain> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.01 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/493 X-Sequence-Number: 13734 Karim Nassar wrote: > > kan4@slap-happy:~/k-bayesBenchmark$ time ./test.pl > <-- snip db creation stuff --> > 17:18:44 -- START > 17:19:37 -- AFTER TEMP LOAD : loaded 120596 records > 17:19:46 -- AFTER bayes_token INSERT : inserted 49359 new records into bayes_token > 17:19:50 -- AFTER bayes_vars UPDATE : updated 1 records > 17:23:37 -- AFTER bayes_token UPDATE : updated 47537 records > DONE > > real 5m4.551s > user 0m29.442s > sys 0m3.925s > > > I am sure someone smarter could optimize further. > > Anyone with a super-spifty machine wanna see if there is an improvement > here? > There is a great improvement in loading the data. While I didn't load it on my server, my test box shows significant gains. It seems that the only thing your script does different is separate the updates from inserts so that an expensive update isn't called when we want to insert. The other major difference is the 'IN' and 'MOT IN' syntax which looks to be much faster than trying everything as an update before inserting. While these optimizations seem to make a huge difference in loading the token data, the real life scenario is a little different. You see, the database keeps track of the number of times each token was found in ham or spam, so that when we see a new message we can parse it into tokens then compare with the database to see how likely the messages is spam based on the statistics of tokens we have already learned on. Since we would want to commit this data after each message, the number of tokens processed at one time would probably only be a few hundred, most of which are probably updates after we have trained on a few thousand emails. I apologize if my crude benchmark was misleading, it was meant to simulate the sheer number of inserts/updates the database may go though in an environment that didn't require people to load spamassassin and start training on spam. I'll do some more testing on Monday, perhaps grouping even 200 tokens at a time using your method will yield significant gains, but probably not as dramatic as it does using my loading benchmark. I post more when I have a chance to look at this in more depth. Thanks, schu From pgsql-performance-owner@postgresql.org Sat Jul 30 07:36:13 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 55E0052965 for ; Sat, 30 Jul 2005 07:36:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 15216-07 for ; Sat, 30 Jul 2005 10:36:10 +0000 (GMT) Received: from veggie-fuel.com (veggie-fuel.com [200.46.208.238]) by svr1.postgresql.org (Postfix) with ESMTP id E8B3A52AC6 for ; Sat, 30 Jul 2005 07:36:09 -0300 (ADT) Received: from localhost (unknown [200.46.204.144]) by veggie-fuel.com (Postfix) with ESMTP id A1DFB35F9E5 for ; Sat, 30 Jul 2005 10:36:11 +0000 (GMT) Received: from veggie-fuel.com ([200.46.208.238]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 17010-07; Sat, 30 Jul 2005 10:35:58 +0000 (GMT) Received: from [10.0.0.7] (63-228-143-68.phnx.qwest.net [63.228.143.68]) by veggie-fuel.com (Postfix) with ESMTP id 1E6BB35F9DE; Sat, 30 Jul 2005 10:35:59 +0000 (GMT) Subject: Re: Performance problems testing with Spamassassin From: Karim Nassar Reply-To: karim.nassar@acm.org To: Matthew Schumacher Cc: pgsql-performance@postgresql.org In-Reply-To: <42EB3E63.6040300@aptalaska.net> References: <3E37B936B592014B978C4415F90D662D0C4221@MI8NYCMAIL06.Mi8.com> <200507290947.03700.josh@agliodbs.com> <1122683983.11869.150.camel@localhost.localdomain> <42EB3E63.6040300@aptalaska.net> Content-Type: text/plain Date: Sat, 30 Jul 2005 03:34:00 -0700 Message-Id: <1122719640.7426.173.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.2.1.1 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.026 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/494 X-Sequence-Number: 13735 On Sat, 2005-07-30 at 00:46 -0800, Matthew Schumacher wrote: > I'll do some more testing on Monday, perhaps grouping even 200 tokens at > a time using your method will yield significant gains, but probably not > as dramatic as it does using my loading benchmark. In that case, some of the clauses could be simplified further since we know that we are dealing with only one user. I don't know what that will get us, since postgres is so damn clever. I suspect that the aggregate functions will be more efficient when you do this, since the temp table will be much smaller, but I am only guessing at this point. If you need to support a massive initial data load, further time savings are to be had by doing COPY instead of 126,000 inserts. Please do keep us updated. Thanking all the gods and/or developers for spamassassin, -- Karim Nassar From pgsql-performance-owner@postgresql.org Sat Jul 30 15:29:17 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D9463529DD for ; Sat, 30 Jul 2005 15:29:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 20485-01 for ; Sat, 30 Jul 2005 18:29:02 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id CC5AD52AA5 for ; Sat, 30 Jul 2005 15:28:59 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id j6UISrlI027898; Sat, 30 Jul 2005 14:28:53 -0400 (EDT) To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 In-reply-to: <42EAA410.9010700@aptalaska.net> References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> Comments: In-reply-to Matthew Schumacher message dated "Fri, 29 Jul 2005 13:48:00 -0800" Date: Sat, 30 Jul 2005 14:28:53 -0400 Message-ID: <27897.1122748133@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.006 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/495 X-Sequence-Number: 13736 Matthew Schumacher writes: > One thing that is still very puzzling to me is why this runs so much > slower when I put the data.sql in a transaction. Obviously transactions > are acting different when you call a proc a zillion times vs an insert > query. I looked into this a bit. It seems that the problem when you wrap the entire insertion series into one transaction is associated with the fact that the test does so many successive updates of the single row in bayes_vars. (VACUUM VERBOSE at the end of the test shows it cleaning up 49383 dead versions of the one row.) This is bad enough when it's in separate transactions, but when it's in one transaction, none of those dead row versions can be marked "fully dead" yet --- so for every update of the row, the unique-key check has to visit every dead version to make sure it's dead in the context of the current transaction. This makes the process O(N^2) in the number of updates per transaction. Which is bad enough if you just want to do one transaction per message, but it's intolerable if you try to wrap the whole bulk-load scenario into one transaction. I'm not sure that we can do anything to make this a lot smarter, but in any case, the real problem is to not do quite so many updates of bayes_vars. How constrained are you as to the format of the SQL generated by SpamAssassin? In particular, could you convert the commands generated for a single message into a single statement? I experimented with passing all the tokens for a given message as a single bytea array, as in the attached, and got almost a factor of 4 runtime reduction on your test case. BTW, it's possible that this is all just a startup-transient problem: once the database has been reasonably well populated, one would expect new tokens to be added infrequently, and so the number of updates to bayes_vars ought to drop off. regards, tom lane Revised insertion procedure: CREATE or replace FUNCTION put_tokens (_id INTEGER, _tokens BYTEA[], _spam_count INTEGER, _ham_count INTEGER, _atime INTEGER) RETURNS VOID AS $$ declare _token bytea; new_tokens integer := 0; BEGIN for i in array_lower(_tokens,1) .. array_upper(_tokens,1) LOOP _token := _tokens[i]; UPDATE bayes_token SET spam_count = spam_count + _spam_count, ham_count = ham_count + _ham_count, atime = _atime WHERE id = _id AND token = _token; IF not found THEN INSERT INTO bayes_token VALUES (_id, _token, _spam_count, _ham_count, _atime); new_tokens := new_tokens + 1; END IF; END LOOP; if new_tokens > 0 THEN UPDATE bayes_vars SET token_count = token_count + new_tokens WHERE id = _id; IF NOT FOUND THEN RAISE EXCEPTION 'unable to update token_count in bayes_vars'; END IF; END IF; RETURN; END; $$ LANGUAGE plpgsql; Typical input: select put_tokens(1,'{"\\125\\42\\80\\167\\166","\\38\\153\\220\\93\\190","\\68\\7\\112\\52\\224","\\51\\14\\78\\155\\49","\\73\\245\\15\\221\\43","\\96\\179\\108\\197\\121","\\123\\97\\220\\173\\247","\\55\\132\\243\\51\\65","\\238\\36\\129\\75\\181","\\145\\253\\196\\106\\90","\\119\\0\\51\\127\\236","\\229\\35\\181\\222\\3","\\163\\1\\191\\220\\79","\\232\\97\\152\\207\\26","\\111\\146\\81\\182\\250","\\47\\141\\12\\76\\45","\\252\\97\\168\\243\\222","\\24\\157\\202\\45\\24","\\230\\207\\30\\46\\115","\\106\\45\\182\\94\\136","\\45\\66\\245\\41\\103","\\108\\126\\171\\154\\210","\\64\\90\\1\\184\\145","\\242\\78\\150\\104\\213","\\214\\\\134\\7\\179\\150","\\249\\12\\247\\164\\74","\\234\\35\\93\\118\\102","\\5\\152\\152\\219\\188","\\99\\186\\172\\56\\241","\\99\\220\\62\\240\\148","\\106\\12\\199\\33\\177","\\34\\74\\190\\192\\186","\\219\\127\\145\\132\\203","\\240\\113\\128\\160\\46","\\83\\5\\239\\206\\221","\\245\\253\\219\\83\\250","\\1\\53\\126\\56\\129","\\206\\1! 30\\97\\246\\47","\\217\\57\\185\\37\\202","\\235\\10\\74\\224\\150","\\80\\151\\70\\52\\96","\\126\\49\\156\\162\\93","\\243\\120\\218\\226\\49","\\251\\132\\118\\47\\221","\\241\\160\\120\\146\\198","\\183\\32\\161\\223\\178","\\80\\205\\77\\57\\2","\\121\\231\\13\\71\\218","\\71\\143\\184\\88\\185","\\163\\96\\119\\211\\142","\\20\\143\\90\\91\\211","\\179\\228\\212\\15\\22","\\243\\35\\149\\9\\55","\\140\\149\\99\\233\\241","\\164\\246\\101\\147\\107","\\202\\70\\218\\40\\114","\\39\\36\\186\\46\\84","\\58\\116\\44\\237\\2","\\80\\204\\185\\47\\105","\\64\\227\\29\\108\\222","\\173\\115\\56\\91\\52","\\102\\39\\157\\252\\64","\\133\\9\\89\\207\\62","\\27\\2\\230\\227\\201","\\163\\45\\123\\160\\129","\\170\\131\\168\\107\\198","\\236\\253\\0\\43\\228","\\44\\255\\93\\197\\136","\\64\\122\\42\\230\\126","\\207\\222\\104\\27\\239","\\26\\240\\78\\73\\45","\\225\\107\\181\\246\\160","\\231\\72\\243\\36\\159","\\248\\60\\14\\67\\145","\\21\\161\\247\\43\\198","\\81\\243\\19! 1\\168\\18","\\237\\227\\23\\40\\140","\\60\\90\\96\\168\\201","\\211\ \107\\181\\46\\38","\\178\\129\\212\\16\\254","\\85\\177\\246\\29\\221","\\182\\123\\178\\157\\9","\\154\\159\\180\\116\\89","\\80\\136\\196\\242\\161","\\185\\110\\90\\163\\157","\\163\\191\\229\\13\\42","\\11\\119\\205\\160\\223","\\75\\216\\70\\223\\6","\\130\\48\\154\\145\\51","\\62\\104\\212\\72\\3","\\247\\105\\51\\64\\136","\\17\\96\\45\\40\\77","\\52\\1\\252\\53\\121","\\68\\195\\58\\103\\91","\\135\\131\\100\\4\\0","\\131\\129\\44\\193\\194","\\47\\234\\101\\143\\26","\\206\\\\134\\32\\154\\0","\\17\\41\\177\\34\\178","\\145\\127\\114\\231\\216","\\19\\172\\6\\39\\126","\\237\\233\\121\\43\\119","\\201\\167\\167\\67\\233","\\88\\159\\102\\50\\117","\\100\\133\\107\\190\\133","\\169\\146\\178\\120\\106"}',1,0,1088628232); select put_tokens(1,'{"\\196\\75\\30\\153\\73","\\73\\245\\15\\221\\43","\\14\\7\\116\\254\\162","\\244\\161\\139\\59\\16","\\214\\226\\238\\196\\30","\\209\\14\\131\\231\\30","\\41\\\\134\\176\\195\\166","\\70\\206\\48\\38\\33","\\247\\131\\136\\80\\31","\\4\\85\\5\\167\\214","\\246\\106\\225\\106\\242","\\28\\0\\229\\160\\90","\\127\\209\\58\\120\\83","\\12\\52\\52\\147\\95","\\255\\115\\21\\5\\68","\\244\\152\\121\\76\\20","\\19\\128\\183\\248\\181","\\140\\91\\18\\127\\208","\\93\\9\\62\\196\\247","\\248\\200\\31\\207\\108","\\44\\216\\247\\15\\195","\\59\\189\\9\\237\\142","\\1\\14\\10\\221\\68","\\163\\155\\122\\223\\104","\\97\\5\\105\\55\\137","\\184\\211\\162\\23\\247","\\239\\249\\83\\68\\54","\\67\\207\\180\\186\\234","\\99\\78\\237\\211\\180","\\200\\11\\32\\179\\50","\\95\\105\\18\\60\\253","\\207\\102\\227\\94\\84","\\71\\143\\184\\88\\185","\\13\\181\\75\\24\\192","\\188\\241\\141\\99\\242","\\139\\124\\248\\130\\4","\\25\\110\\149\\63\\114","\\21\\162\\199\\1! 29\\199","\\164\\246\\101\\147\\107","\\198\\202\\223\\58\\197","\\181\\10\\41\\25\\130","\\71\\163\\116\\239\\170","\\46\\170\\238\\142\\89","\\176\\120\\106\\103\\228","\\39\\228\\25\\38\\170","\\114\\79\\121\\18\\222","\\178\\105\\98\\61\\39","\\90\\61\\12\\23\\135","\\176\\118\\81\\65\\66","\\55\\104\\57\\198\\150","\\206\\251\\224\\128\\41","\\29\\158\\68\\146\\164","\\248\\60\\14\\67\\145","\\210\\220\\161\\10\\254","\\72\\81\\151\\213\\68","\\25\\236\\210\\197\\128","\\72\\37\\208\\227\\54","\\242\\24\\6\\88\\26","\\128\\197\\20\\5\\211","\\98\\105\\71\\42\\180","\\91\\43\\72\\84\\104","\\205\\254\\174\\65\\141","\\222\\194\\126\\204\\164","\\233\\153\\37\\148\\226","\\32\\195\\22\\153\\87","\\194\\97\\220\\251\\18","\\151\\201\\148\\52\\147","\\205\\55\\0\\226\\58","\\172\\12\\50\\0\\140","\\56\\32\\43\\9\\45","\\18\\174\\50\\162\\126","\\138\\150\\12\\72\\189","\\49\\230\\150\\210\\48","\\2\\140\\64\\104\\32","\\14\\174\\41\\196\\121","\\100\\195\\116\\130\\101","\! \222\\45\\94\\39\\64","\\178\\203\\221\\63\\94","\\26\\188\\157\\\\134 \\52","\\119\\0\\51\\127\\236","\\88\\32\\224\\142\\164","\\111\\146\\81\\182\\250","\\12\\177\\151\\83\\13","\\113\\27\\173\\162\\19","\\158\\216\\41\\236\\226","\\16\\88\\\\134\\180\\112","\\43\\32\\16\\77\\238","\\136\\93\\210\\172\\63","\\251\\214\\30\\40\\146","\\156\\27\\198\\60\\170","\\185\\29\\172\\30\\68","\\202\\83\\59\\228\\252","\\219\\127\\145\\132\\203","\\1\\223\\97\\229\\127","\\113\\83\\123\\167\\140","\\99\\1\\116\\56\\165","\\143\\224\\239\\1\\173","\\49\\186\\156\\51\\92","\\246\\224\\70\\245\\137","\\235\\10\\74\\224\\150","\\43\\88\\245\\14\\103","\\88\\128\\232\\142\\254","\\251\\132\\118\\47\\221","\\36\\7\\142\\234\\98","\\130\\126\\199\\170\\126","\\133\\23\\51\\253\\234","\\249\\89\\242\\87\\86","\\102\\243\\47\\193\\211","\\140\\18\\140\\164\\248","\\179\\228\\212\\15\\22","\\168\\155\\243\\169\\191","\\117\\37\\139\\241\\230","\\155\\11\\254\\171\\200","\\196\\159\\253\\223\\15","\\93\\207\\154\\106\\135","\\11\\255\\28\\123\\125","\\239\\9\\226! \\59\\198","\\191\\204\\230\\61\\39","\\175\\204\\181\\113\\\\134","\\64\\227\\29\\108\\222","\\169\\173\\194\\83\\40","\\212\\93\\170\\169\\12","\\249\\55\\232\\182\\208","\\75\\175\\181\\248\\246","\\108\\95\\114\\215\\138","\\220\\37\\59\\207\\197","\\45\\146\\43\\76\\81","\\166\\231\\20\\9\\189","\\27\\126\\81\\92\\75","\\66\\168\\119\\100\\196","\\229\\9\\196\\165\\250","\\83\\186\\103\\184\\46","\\85\\177\\246\\29\\221","\\140\\159\\53\\211\\157","\\214\\193\\192\\217\\109","\\10\\5\\64\\97\\157","\\92\\137\\120\\70\\55","\\235\\45\\181\\44\\98","\\150\\56\\132\\207\\19","\\67\\95\\161\\39\\122","\\109\\65\\145\\170\\79","\\\\134\\28\\90\\39\\33","\\226\\177\\240\\202\\157","\\1\\57\\50\\6\\240","\\249\\240\\222\\56\\161","\\110\\136\\88\\85\\249","\\82\\27\\239\\51\\211","\\114\\223\\252\\83\\189","\\129\\216\\251\\218\\80","\\247\\36\\101\\90\\229","\\209\\73\\221\\46\\11","\\242\\12\\120\\117\\\\134","\\146\\198\\57\\177\\49","\\212\\57\\9\\240\\216","\\215\\151\\2! 16\\59\\75","\\47\\132\\161\\165\\54","\\113\\4\\77\\241\\150","\\217\ \184\\149\\53\\124","\\152\\111\\25\\231\\104","\\42\\185\\112\\250\\156","\\39\\131\\14\\140\\189","\\148\\169\\158\\251\\150","\\184\\142\\204\\122\\179","\\19\\189\\181\\105\\116","\\116\\77\\22\\135\\50","\\236\\231\\60\\132\\229","\\200\\63\\76\\232\\9","\\32\\20\\168\\87\\45","\\99\\129\\99\\165\\29","\\2\\208\\66\\228\\105","\\99\\194\\194\\229\\17","\\85\\250\\55\\51\\114","\\200\\165\\249\\77\\72","\\5\\91\\178\\157\\24","\\245\\253\\219\\83\\250","\\166\\103\\181\\196\\34","\\227\\149\\148\\105\\157","\\95\\44\\15\\251\\98","\\183\\32\\161\\223\\178","\\120\\236\\145\\158\\78","\\244\\4\\92\\233\\112","\\189\\231\\124\\92\\19","\\112\\132\\8\\49\\157","\\160\\243\\244\\94\\104","\\150\\176\\139\\251\\157","\\176\\193\\155\\175\\144","\\161\\208\\145\\92\\92","\\77\\122\\94\\69\\182","\\77\\13\\131\\29\\27","\\92\\9\\178\\204\\254","\\177\\4\\154\\211\\63","\\62\\4\\242\\1\\78","\\4\\129\\113\\205\\164","\\168\\95\\68\\89\\38","\\173\\115\\56\\91\\52","\\212\\161\\1! 59\\148\\179","\\133\\9\\89\\207\\62","\\242\\51\\168\\130\\86","\\154\\199\\208\\84\\2","\\160\\215\\250\\104\\22","\\45\\252\\143\\149\\204","\\48\\50\\91\\39\\243","\\94\\168\\48\\202\\122","\\238\\38\\180\\135\\142","\\234\\59\\24\\148\\2","\\237\\227\\23\\40\\140","\\7\\114\\176\\80\\123","\\204\\170\\0\\60\\65","\\217\\202\\249\\158\\182","\\82\\170\\45\\96\\86","\\118\\11\\123\\51\\216","\\192\\130\\153\\88\\59","\\219\\53\\146\\88\\198","\\203\\114\\182\\147\\145","\\158\\140\\239\\104\\247","\\179\\86\\111\\146\\65","\\192\\168\\51\\125\\183","\\8\\251\\77\\143\\231","\\237\\229\\173\\221\\29","\\69\\178\\247\\196\\175","\\114\\33\\237\\189\\119","\\220\\44\\144\\93\\98","\\241\\38\\138\\127\\252","\\66\\218\\237\\199\\157","\\132\\240\\212\\221\\48","\\180\\41\\157\\84\\37","\\203\\180\\58\\113\\136","\\156\\39\\111\\181\\34","\\16\\202\\216\\183\\55","\\154\\51\\122\\201\\45","\\218\\112\\47\\206\\142","\\189\\141\\110\\230\\132","\\80\\167\\61\\103\\247","\\186\! \15\\121\\27\\167","\\103\\163\\217\\19\\220","\\173\\116\\86\\7\\249" ,"\\25\\37\\98\\35\\127","\\44\\92\\200\\89\\84","\\171\\129\\106\\249\\38","\\24\\147\\77\\\\134\\62","\\254\\184\\72\\159\\91","\\221\\13\\18\\153\\154","\\109\\232\\79\\169\\176","\\152\\103\\190\\50\\18","\\51\\71\\217\\22\\76","\\105\\109\\7\\77\\198","\\250\\121\\163\\49\\73","\\138\\204\\\\134\\247\\116","\\130\\38\\156\\36\\27","\\20\\83\\86\\113\\124","\\40\\63\\161\\157\\76","\\205\\99\\150\\109\\249","\\111\\174\\57\\169\\238","\\106\\169\\245\\170\\240","\\32\\10\\53\\160\\76","\\226\\0\\58\\9\\22","\\63\\83\\21\\3\\205","\\212\\141\\249\\177\\102","\\197\\226\\42\\202\\130","\\70\\40\\85\\176\\2","\\3\\16\\133\\118\\91","\\232\\48\\176\\209\\77","\\20\\149\\0\\2\\144","\\50\\87\\138\\108\\149","\\13\\78\\64\\211\\245","\\15\\158\\123\\62\\103","\\239\\68\\210\\175\\197","\\247\\216\\7\\211\\5","\\112\\100\\135\\210\\101","\\47\\26\\118\\254\\62","\\123\\7\\143\\206\\114","\\184\\43\\252\\56\\194","\\55\\16\\219\\\\134\\201","\\170\\128\\224\\160\\251","\\180\\10! 8\\182\\255\\118","\\164\\155\\151\\195\\67","\\116\\56\\163\\249\\92","\\250\\207\\75\\244\\104","\\122\\219\\25\\49\\17","\\16\\61\\66\\50\\32","\\15\\223\\166\\\\134\\188","\\16\\221\\48\\159\\124","\\163\\66\\245\\19\\190","\\52\\177\\137\\57\\104","\\137\\158\\143\\12\\73","\\175\\156\\252\\243\\165","\\18\\119\\\\134\\198\\209","\\179\\60\\37\\63\\136","\\68\\117\\75\\163\\27","\\234\\108\\150\\93\\15","\\209\\159\\154\\221\\138","\\70\\215\\50\\36\\255","\\237\\64\\91\\125\\54","\\17\\41\\177\\34\\178","\\19\\241\\29\\12\\34","\\\\134\\151\\112\\6\\214","\\63\\61\\146\\243\\60"}',1,0,1088649636); For testing purposes, here is the Perl script I used to make the revised input from the original test script: #!/usr/bin/perl -w $lastat = 0; while (<>) { if (m/^select put_token\((\d+),'(.*)',(\d+),(\d+),(\d+)\);$/) { $d1 = $1; $data = $2; $d3 = $3; $d4 = $4; $d5 = $5; $data =~ s/\\/\\\\/g; if ($d5 == $lastat) { $last2 = $last2 . ',"' . $data . '"'; } else { if ($lastat) { print "select put_tokens($last1,'{$last2}',$last3,$last4,$lastat);\n"; } $last1 = $d1; $last2 = '"' . $data . '"'; $last3 = $d3; $last4 = $d4; $lastat = $d5; } } else { print ; } } print "select put_tokens($last1,'{$last2}',$last3,$last4,$lastat);\n"; BTW, the data quoting is probably wrong here, but I suppose that can be fixed easily. From pgsql-performance-owner@postgresql.org Sat Jul 30 18:06:57 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 2865052AFD for ; Sat, 30 Jul 2005 18:06:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 49659-01 for ; Sat, 30 Jul 2005 21:06:51 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id DAECB52AC9 for ; Sat, 30 Jul 2005 18:06:50 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6ULKa1q008031 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sat, 30 Jul 2005 13:20:37 -0800 Message-ID: <42EBEBE9.4020504@aptalaska.net> Date: Sat, 30 Jul 2005 13:06:49 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> <27897.1122748133@sss.pgh.pa.us> In-Reply-To: <27897.1122748133@sss.pgh.pa.us> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: unavailable X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/496 X-Sequence-Number: 13737 Tom Lane wrote: > I looked into this a bit. It seems that the problem when you wrap the > entire insertion series into one transaction is associated with the fact > that the test does so many successive updates of the single row in > bayes_vars. (VACUUM VERBOSE at the end of the test shows it cleaning up > 49383 dead versions of the one row.) This is bad enough when it's in > separate transactions, but when it's in one transaction, none of those > dead row versions can be marked "fully dead" yet --- so for every update > of the row, the unique-key check has to visit every dead version to make > sure it's dead in the context of the current transaction. This makes > the process O(N^2) in the number of updates per transaction. Which is > bad enough if you just want to do one transaction per message, but it's > intolerable if you try to wrap the whole bulk-load scenario into one > transaction. > > I'm not sure that we can do anything to make this a lot smarter, but > in any case, the real problem is to not do quite so many updates of > bayes_vars. > > How constrained are you as to the format of the SQL generated by > SpamAssassin? In particular, could you convert the commands generated > for a single message into a single statement? I experimented with > passing all the tokens for a given message as a single bytea array, > as in the attached, and got almost a factor of 4 runtime reduction > on your test case. > > BTW, it's possible that this is all just a startup-transient problem: > once the database has been reasonably well populated, one would expect > new tokens to be added infrequently, and so the number of updates to > bayes_vars ought to drop off. > > regards, tom lane > The spamassassins bayes code calls the _put_token method in the storage module a loop. This means that the storage module isn't called once per message, but once per token. I'll look into modifying it to so that the bayes code passes a hash of tokens to the storage module where they can loop or in the case of the pgsql module pass an array of tokens to a procedure where we loop and use temp tables to make this much more efficient. I don't have much time this weekend to toss at this, but will be looking at it on Monday. Thanks, schu From pgsql-performance-owner@postgresql.org Sun Jul 31 02:27:24 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id E6381528DC for ; Sun, 31 Jul 2005 02:27:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54321-04 for ; Sun, 31 Jul 2005 05:27:18 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 50DF1528C6 for ; Sun, 31 Jul 2005 02:27:17 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050731052716m9100g65jae>; Sun, 31 Jul 2005 05:27:16 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 232DE56021; Sun, 31 Jul 2005 00:27:16 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 6A78E56006; Sun, 31 Jul 2005 00:27:09 -0500 (CDT) Message-ID: <42EC612A.1080908@arbash-meinel.com> Date: Sun, 31 Jul 2005 00:27:06 -0500 From: John Arbash Meinel User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> In-Reply-To: <42EAA410.9010700@aptalaska.net> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigFEA353E5A7E31AF4F37B27E4" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.046 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/497 X-Sequence-Number: 13738 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigFEA353E5A7E31AF4F37B27E4 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Matthew Schumacher wrote: >All it's doing is trying the update before the insert to get around the >problem of not knowing which is needed. With only 2-3 of the queries >implemented I'm already back to running about the same speed as the >original SA proc that is going to ship with SA 3.1.0. > >All of the queries are using indexes so at this point I'm pretty >convinced that the biggest problem is the sheer number of queries >required to run this proc 200 times for each email (once for each token). > >I don't see anything that could be done to make this much faster on the >postgres end, it's looking like the solution is going to involve cutting >down the number of queries some how. > >One thing that is still very puzzling to me is why this runs so much >slower when I put the data.sql in a transaction. Obviously transactions >are acting different when you call a proc a zillion times vs an insert >query. > > Well, I played with adding a COMMIT;BEGIN; statement to your exact test every 1000 lines. And this is what I got: Unmodified: real 17m53.587s user 0m6.204s sys 0m3.556s With BEGIN/COMMIT: real 1m53.466s user 0m5.203s sys 0m3.211s So I see the potential for improvement almost 10 fold by switching to transactions. I played with the perl script (and re-implemented it in python), and for the same data as the perl script, using COPY instead of INSERT INTO means 5s instead of 33s. I also played around with adding VACUUM ANALYZE every 10 COMMITS, which brings the speed to: real 1m41.258s user 0m5.394s sys 0m3.212s And doing VACUUM ANALYZE every 5 COMMITS makes it: real 1m46.403s user 0m5.597s sys 0m3.244s I'm assuming the slowdown is because of the extra time spent vacuuming. Overall performance might still be improving, since you wouldn't actually be inserting all 100k rows at once. Just to complete the reference, the perl version runs as: 10:44:02 -- START 10:44:35 -- AFTER TEMP LOAD : loaded 120596 records 10:44:39 -- AFTER bayes_token INSERT : inserted 49359 new records into bayes_token 10:44:41 -- AFTER bayes_vars UPDATE : updated 1 records 10:46:42 -- AFTER bayes_token UPDATE : updated 47537 records DONE My python version runs as: 00:22:54 -- START 00:23:00 -- AFTER TEMP LOAD : loaded 120596 records 00:23:03 -- AFTER bayes_token INSERT : inserted 49359 new records into bayes_token 00:23:06 -- AFTER bayes_vars UPDATE : updated 1 records 00:25:04 -- AFTER bayes_token UPDATE : updated 47537 records DONE The python is effectively just a port of the perl code (with many lines virtually unchanged), and really the only performance difference is that the initial data load is much faster with a COPY. This is all run on Ubuntu, with postgres 7.4.7, and a completely unchanged postgresql.conf. (But the machine is a dual P4 2.4GHz, with 3GB of RAM). John =:-> >Anyway, if anyone else has any ideas I'm all ears, but at this point >it's looking like raw query speed is needed for this app and while I >don't care for mysql as a database, it does have the speed going for it. > >schu > >---------------------------(end of broadcast)--------------------------- >TIP 3: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faq > > > --------------enigFEA353E5A7E31AF4F37B27E4 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC7GEsJdeBCYSNAAMRAun5AKCqn5Y8W4IzbkEpE+sDSjnd1txoIgCg16zP SwD0k//d6yrTIoy4RThIh5I= =hEBe -----END PGP SIGNATURE----- --------------enigFEA353E5A7E31AF4F37B27E4-- From pgsql-performance-owner@postgresql.org Sun Jul 31 02:32:13 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 67BE1528DC for ; Sun, 31 Jul 2005 02:32:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 54320-05 for ; Sun, 31 Jul 2005 05:32:06 +0000 (GMT) Received: from sccmmhc92.asp.att.net (sccmmhc92.asp.att.net [204.127.203.212]) by svr1.postgresql.org (Postfix) with ESMTP id BD071528C9 for ; Sun, 31 Jul 2005 02:32:04 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc92.asp.att.net (sccmmhc92) with ESMTP id <20050731053159m9200g3vsfe>; Sun, 31 Jul 2005 05:32:04 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 5F0D556021; Sun, 31 Jul 2005 00:31:59 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id 361AC56006; Sun, 31 Jul 2005 00:31:55 -0500 (CDT) Message-ID: <42EC624A.2020308@arbash-meinel.com> Date: Sun, 31 Jul 2005 00:31:54 -0500 From: John Arbash Meinel User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Matthew Schumacher Cc: Tom Lane , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> <27897.1122748133@sss.pgh.pa.us> <42EBEBE9.4020504@aptalaska.net> In-Reply-To: <42EBEBE9.4020504@aptalaska.net> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig900729736F9A409B13DA6E47" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.046 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/498 X-Sequence-Number: 13739 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig900729736F9A409B13DA6E47 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Matthew Schumacher wrote: >Tom Lane wrote: > > > >>I looked into this a bit. It seems that the problem when you wrap the >>entire insertion series into one transaction is associated with the fact >>that the test does so many successive updates of the single row in >>bayes_vars. (VACUUM VERBOSE at the end of the test shows it cleaning up >>49383 dead versions of the one row.) This is bad enough when it's in >>separate transactions, but when it's in one transaction, none of those >>dead row versions can be marked "fully dead" yet --- so for every update >>of the row, the unique-key check has to visit every dead version to make >>sure it's dead in the context of the current transaction. This makes >>the process O(N^2) in the number of updates per transaction. Which is >>bad enough if you just want to do one transaction per message, but it's >>intolerable if you try to wrap the whole bulk-load scenario into one >>transaction. >> >>I'm not sure that we can do anything to make this a lot smarter, but >>in any case, the real problem is to not do quite so many updates of >>bayes_vars. >> >>How constrained are you as to the format of the SQL generated by >>SpamAssassin? In particular, could you convert the commands generated >>for a single message into a single statement? I experimented with >>passing all the tokens for a given message as a single bytea array, >>as in the attached, and got almost a factor of 4 runtime reduction >>on your test case. >> >>BTW, it's possible that this is all just a startup-transient problem: >>once the database has been reasonably well populated, one would expect >>new tokens to be added infrequently, and so the number of updates to >>bayes_vars ought to drop off. >> >> regards, tom lane >> >> >> > >The spamassassins bayes code calls the _put_token method in the storage >module a loop. This means that the storage module isn't called once per >message, but once per token. > > Well, putting everything into a transaction per email might make your pain go away. If you saw the email I just sent, I modified your data.sql file to add a "COMMIT;BEGIN" every 1000 selects, and I saw a performance jump from 18 minutes down to less than 2 minutes. Heck, on my machine, the advanced perl version takes more than 2 minutes to run. It is actually slower than the data.sql with commit statements. >I'll look into modifying it to so that the bayes code passes a hash of >tokens to the storage module where they can loop or in the case of the >pgsql module pass an array of tokens to a procedure where we loop and >use temp tables to make this much more efficient. > > Well, you could do that. Or you could just have the bayes code issue "BEGIN;" when it starts processing an email, and a "COMMIT;" when it finishes. From my testing, you will see an enormous speed improvement. (And you might consider including a fairly frequent VACUUM ANALYZE) >I don't have much time this weekend to toss at this, but will be looking >at it on Monday. > > Good luck, John =:-> >Thanks, > >schu > >---------------------------(end of broadcast)--------------------------- >TIP 5: don't forget to increase your free space map settings > > > --------------enig900729736F9A409B13DA6E47 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC7GJKJdeBCYSNAAMRAhehAJ0b7M1ZFaA3lkvkgxsDgnZrAPc8oACgksHm dOogxce0sXl1qIFC6MMwDXo= =Pb5C -----END PGP SIGNATURE----- --------------enig900729736F9A409B13DA6E47-- From pgsql-performance-owner@postgresql.org Sun Jul 31 07:11:24 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C44C952835 for ; Sun, 31 Jul 2005 07:11:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 09876-01 for ; Sun, 31 Jul 2005 10:11:17 +0000 (GMT) Received: from mail2.aeccom.com (port-212-202-101-158.static.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id F04B95280C for ; Sun, 31 Jul 2005 07:11:15 -0300 (ADT) Received: from [192.168.2.132] (openvpn02.core.aeccom.com [192.168.2.132]) by mail2.aeccom.com (Postfix) with ESMTP id C4EF8EC; Sun, 31 Jul 2005 12:11:15 +0200 (CEST) Message-ID: <42ECA3B6.3000607@aeccom.com> Date: Sun, 31 Jul 2005 12:11:02 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Reply-To: lutzeb@aeccom.com User-Agent: Mozilla Thunderbird 0.9 (X11/20041103) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Jeffrey W. Baker" Cc: josh@agliodbs.com, pgsql-performance@postgresql.org, Sven Geisler Subject: Re: Performance problems on 4/8way Opteron (dualcore) HP References: <42EA5D18.9020204@aeccom.com> <200507291046.10318.josh@agliodbs.com> <1122663342.6814.5.camel@toonses.gghcwest.com> In-Reply-To: <1122663342.6814.5.camel@toonses.gghcwest.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/499 X-Sequence-Number: 13740 Hi Jeff, which box are you running precisely and which OS/kernel? We need to run 32bit because we need failover to 32 bit XEON system (DL580). If this does not work out we probably need to switch to 64 bit (dump/restore) and run a nother 64bit failover box too. Regards, Dirk Jeffrey W. Baker wrote: > On Fri, 2005-07-29 at 10:46 -0700, Josh Berkus wrote: > >>Dirk, >> >> >>>does anybody have expierence with this machine (4x 875 dual core Opteron >>>CPUs)? > > > I'm using dual 275s without problems. > > >>Nope. I suspect that you may be the first person to report in on >>dual-cores. There may be special compile issues with dual-cores that >>we've not yet encountered. > > > Doubtful. However you could see improvements using recent Linux kernel > code. There have been some patches for optimizing scheduling and memory > allocations. > > However, if you are running this machine in 32-bit mode, why did you > bother paying $14,000 for your CPUs? You will get FAR better > performance in 64-bit mode. 64-bit mode will give you 30-50% better > performance on PostgreSQL loads, in my experience. Also, if I remember > correctly, the 32-bit x86 kernel doesn't understand Opteron NUMA > topology, so you may be seeing poor memory allocation decisions. > > -jwb > > >>>We run RHEL 3.0, 32bit and under high load it is a drag. We >>>mostly run memory demanding queries. Context switches are pretty much >>>around 20.000 on the average, no cs spikes when we run many processes in >>>parallel. Actually we only see two processes in running state! When >>>there are only a few processes running context switches go much higher. >>>At the moment we are much slower that with a 4way XEON box (DL580). >> >>Um, that was a bit incoherent. Are you seeing a CS storm or aren't you? >> From pgsql-performance-owner@postgresql.org Sun Jul 31 07:12:53 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 798ED52871 for ; Sun, 31 Jul 2005 07:12:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 04661-08 for ; Sun, 31 Jul 2005 10:12:46 +0000 (GMT) Received: from mail2.aeccom.com (port-212-202-101-158.static.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id EAC295286E for ; Sun, 31 Jul 2005 07:12:43 -0300 (ADT) Received: from [192.168.2.132] (openvpn02.core.aeccom.com [192.168.2.132]) by mail2.aeccom.com (Postfix) with ESMTP id D66D2EC; Sun, 31 Jul 2005 12:12:44 +0200 (CEST) Message-ID: <42ECA410.7020804@aeccom.com> Date: Sun, 31 Jul 2005 12:12:32 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Reply-To: lutzeb@aeccom.com User-Agent: Mozilla Thunderbird 0.9 (X11/20041103) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "J. Andrew Rogers" Cc: josh@agliodbs.com, pgsql-performance@postgresql.org, Sven Geisler Subject: Re: Performance problems on 4/8way Opteron (dualcore) HP References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.05 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/500 X-Sequence-Number: 13741 Anybody knows if RedHat is already supporting this patch on an enterprise version? Regards, Dirk J. Andrew Rogers wrote: > On 7/29/05 10:46 AM, "Josh Berkus" wrote: > >>>does anybody have expierence with this machine (4x 875 dual core Opteron >>>CPUs)? >> >>Nope. I suspect that you may be the first person to report in on >>dual-cores. There may be special compile issues with dual-cores that >>we've not yet encountered. > > > > There was recently a discussion of similar types of problems on a couple of > the supercomputing lists, regarding surprisingly substandard performance > from large dual-core opteron installations. > > The problem as I remember it boiled down to the Linux kernel handling > memory/process management very badly on large dual core systems -- > pathological NUMA behavior. However, this problem has apparently been fixed > in Linux v2.6.12+, and using the more recent kernel on large dual core > systems generated *massive* performance improvements on these systems for > the individuals with this issue. Using the patched kernel, one gets the > performance most people were expecting. > > The v2.6.12+ kernels are a bit new, but they contain a very important > performance patch for systems like the one above. It would definitely be > worth testing if possible. > > > J. Andrew Rogers > > From pgsql-performance-owner@postgresql.org Sun Jul 31 12:29:29 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C88B55284A for ; Sun, 31 Jul 2005 12:29:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 92472-07 for ; Sun, 31 Jul 2005 15:29:18 +0000 (GMT) Received: from mx1.neopolitan.us (mx1.neopolitan.us [65.87.16.224]) by svr1.postgresql.org (Postfix) with ESMTP id B84345288D for ; Sun, 31 Jul 2005 12:29:16 -0300 (ADT) Received: from [64.173.27.113] (HELO [192.168.2.45]) by mx1.neopolitan.us (CommuniGate Pro SMTP 4.2.10) with ESMTP id 10180932 for pgsql-performance@postgresql.org; Sun, 31 Jul 2005 08:29:15 -0700 User-Agent: Microsoft-Entourage/11.1.0.040913 Date: Sun, 31 Jul 2005 08:29:15 -0700 Subject: Re: Performance problems on 4/8way Opteron (dualcore) From: "J. Andrew Rogers" To: Message-ID: In-Reply-To: Mime-version: 1.0 Content-type: text/plain; charset="US-ASCII" Content-transfer-encoding: 7bit X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=1.312 tagged_above=0 required=5 tests=AWL, DNS_FROM_RFC_ABUSE, DNS_FROM_RFC_POST X-Spam-Level: * X-Archive-Number: 200507/501 X-Sequence-Number: 13742 On 7/30/05 12:57 AM, "William Yu" wrote: > I haven't investigated the 2.6.12+ kernel updates yet -- I probably will > do our development servers first to give it a test. The kernel updates make the NUMA code dual-core aware, which apparently makes a big difference in some cases but not in others. It makes some sense, since multi-processor multi-core machines will have two different types of non-locality instead of just one that need to be managed. Prior to the v2.6.12 patches, a dual-core dual-proc machine was viewed as a quad-proc machine. The closest thing to a supported v2.6.12 kernel that I know of is FC4, which is not really supported in the enterprise sense of course. J. Andrew Rogers From pgsql-performance-owner@postgresql.org Sun Jul 31 12:52:50 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 7541F528A8 for ; Sun, 31 Jul 2005 12:52:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 96355-10 for ; Sun, 31 Jul 2005 15:52:29 +0000 (GMT) Received: from sccmmhc91.asp.att.net (sccmmhc91.asp.att.net [204.127.203.211]) by svr1.postgresql.org (Postfix) with ESMTP id 0EE8552838 for ; Sun, 31 Jul 2005 12:52:27 -0300 (ADT) Received: from juju.arbash-meinel.com ([12.214.18.81]) by sccmmhc91.asp.att.net (sccmmhc91) with ESMTP id <20050731155226m9100g5rcne>; Sun, 31 Jul 2005 15:52:26 +0000 Received: by juju.arbash-meinel.com (Postfix, from userid 505) id 8485055FFF; Sun, 31 Jul 2005 10:52:25 -0500 (CDT) Received: from [192.168.1.11] (liliana.arbash-meinel.com [192.168.1.11]) by juju.arbash-meinel.com (Postfix) with ESMTP id B990655F9F; Sun, 31 Jul 2005 10:52:18 -0500 (CDT) Message-ID: <42ECF3AE.6060809@arbash-meinel.com> Date: Sun, 31 Jul 2005 10:52:14 -0500 From: John Arbash Meinel User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050404) X-Accept-Language: en-us, en MIME-Version: 1.0 To: John Arbash Meinel Cc: Matthew Schumacher , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> <42EC612A.1080908@arbash-meinel.com> In-Reply-To: <42EC612A.1080908@arbash-meinel.com> X-Enigmail-Version: 0.92.0.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigE3AA658B8D7F7F80FBA76421" X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.046 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/502 X-Sequence-Number: 13743 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigE3AA658B8D7F7F80FBA76421 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit John Arbash Meinel wrote: >Matthew Schumacher wrote: > > > >>All it's doing is trying the update before the insert to get around the >>problem of not knowing which is needed. With only 2-3 of the queries >>implemented I'm already back to running about the same speed as the >>original SA proc that is going to ship with SA 3.1.0. >> >>All of the queries are using indexes so at this point I'm pretty >>convinced that the biggest problem is the sheer number of queries >>required to run this proc 200 times for each email (once for each token). >> >>I don't see anything that could be done to make this much faster on the >>postgres end, it's looking like the solution is going to involve cutting >>down the number of queries some how. >> >>One thing that is still very puzzling to me is why this runs so much >>slower when I put the data.sql in a transaction. Obviously transactions >>are acting different when you call a proc a zillion times vs an insert >>query. >> >> >> >> >Well, I played with adding a COMMIT;BEGIN; statement to your exact test >every 1000 lines. And this is what I got: > > Just for reference, I also tested this on my old server, which is a dual Celeron 450 with 256M ram. FC4 and Postgres 8.0.3 Unmodified: real 54m15.557s user 0m24.328s sys 0m14.200s With Transactions every 1000 selects, and vacuum every 5000: real 8m36.528s user 0m16.585s sys 0m12.569s With Transactions every 1000 selects, and vacuum every 10000: real 7m50.748s user 0m16.183s sys 0m12.489s On this machine vacuum is more expensive, since it doesn't have as much ram. Anyway, on this machine, I see approx 7x improvement. Which I think is probably going to satisfy your spamassassin needs. John =:-> PS> Looking forward to having a spamassassin that can utilize my favorite db. Right now, I'm not using a db backend because it wasn't worth setting up mysql. >Unmodified: >real 17m53.587s >user 0m6.204s >sys 0m3.556s > >With BEGIN/COMMIT: >real 1m53.466s >user 0m5.203s >sys 0m3.211s > >So I see the potential for improvement almost 10 fold by switching to >transactions. I played with the perl script (and re-implemented it in >python), and for the same data as the perl script, using COPY instead of >INSERT INTO means 5s instead of 33s. > >I also played around with adding VACUUM ANALYZE every 10 COMMITS, which >brings the speed to: > >real 1m41.258s >user 0m5.394s >sys 0m3.212s > >And doing VACUUM ANALYZE every 5 COMMITS makes it: >real 1m46.403s >user 0m5.597s >sys 0m3.244s > >I'm assuming the slowdown is because of the extra time spent vacuuming. >Overall performance might still be improving, since you wouldn't >actually be inserting all 100k rows at once. > > ... >This is all run on Ubuntu, with postgres 7.4.7, and a completely >unchanged postgresql.conf. (But the machine is a dual P4 2.4GHz, with >3GB of RAM). > >John >=:-> > > --------------enigE3AA658B8D7F7F80FBA76421 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFC7POyJdeBCYSNAAMRAiUjAJ9x+Q961uI0MY/Og+GK6ZgnAPsz7gCgmuuZ MpgRe2BiJP60c44Qcph6BRY= =mjjm -----END PGP SIGNATURE----- --------------enigE3AA658B8D7F7F80FBA76421-- From pgsql-performance-owner@postgresql.org Sun Jul 31 13:51:21 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 842A35286F for ; Sun, 31 Jul 2005 13:51:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 37578-09 for ; Sun, 31 Jul 2005 16:51:10 +0000 (GMT) Received: from larry.aptalaska.net (larry.aptalaska.net [64.186.96.3]) by svr1.postgresql.org (Postfix) with ESMTP id 6861752833 for ; Sun, 31 Jul 2005 13:51:09 -0300 (ADT) Received: from [192.168.98.9] (rdbck-static-287.palmer.mtaonline.net [64.4.232.33]) (authenticated bits=0) by larry.aptalaska.net (8.13.4/8.13.4) with ESMTP id j6VH54R5026900 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Sun, 31 Jul 2005 09:05:04 -0800 Message-ID: <42ED017A.2050207@aptalaska.net> Date: Sun, 31 Jul 2005 08:51:06 -0800 From: Matthew Schumacher User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <42E80C28.2040608@aptalaska.net> <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> <42EC612A.1080908@arbash-meinel.com> <42ECF3AE.6060809@arbash-meinel.com> In-Reply-To: <42ECF3AE.6060809@arbash-meinel.com> X-Enigmail-Version: 0.91.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-Spam-Bayes: Learn: ham X-Scanned-By: MIMEDefang 2.52 on 64.186.96.3 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.009 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/503 X-Sequence-Number: 13744 Ok, here is the current plan. Change the spamassassin API to pass a hash of tokens into the storage module, pass the tokens to the proc as an array, start a transaction, load the tokens into a temp table using copy, select the tokens distinct into the token table for new tokens, update the token table for known tokens, then commit. This solves the following problems: 1. Each email is a transaction instead of each token. 2. The update statement is only called when we really need an update which avoids all of those searches. 3. The looping work is done inside the proc instead of perl calling a method a zillion times per email. I'm not sure how vacuuming will be done yet, if we vacuum once per email that may be too often, so I may do that every 5 mins in cron. schu From pgsql-performance-owner@postgresql.org Sun Jul 31 14:10:20 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 85B9E52B23 for ; Sun, 31 Jul 2005 14:10:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 68524-01 for ; Sun, 31 Jul 2005 17:10:14 +0000 (GMT) Received: from flake.decibel.org (flake.decibel.org [67.100.216.10]) by svr1.postgresql.org (Postfix) with ESMTP id 9596A52B19 for ; Sun, 31 Jul 2005 14:10:13 -0300 (ADT) Received: by flake.decibel.org (Postfix, from userid 1001) id BEE3015252; Sun, 31 Jul 2005 12:10:12 -0500 (CDT) Date: Sun, 31 Jul 2005 12:10:12 -0500 From: "Jim C. Nasby" To: Matthew Schumacher Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 Message-ID: <20050731171012.GS60019@decibel.org> References: <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> <42EC612A.1080908@arbash-meinel.com> <42ECF3AE.6060809@arbash-meinel.com> <42ED017A.2050207@aptalaska.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <42ED017A.2050207@aptalaska.net> X-Operating-System: FreeBSD 4.11-RELEASE-p10 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.9i X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/504 X-Sequence-Number: 13745 On Sun, Jul 31, 2005 at 08:51:06AM -0800, Matthew Schumacher wrote: > Ok, here is the current plan. > > Change the spamassassin API to pass a hash of tokens into the storage > module, pass the tokens to the proc as an array, start a transaction, > load the tokens into a temp table using copy, select the tokens distinct > into the token table for new tokens, update the token table for known > tokens, then commit. You might consider: UPDATE tokens FROM temp_table (this updates existing records) INSERT INTO tokens SELECT ... FROM temp_table WHERE NOT IN (SELECT ... FROM tokens) This way you don't do an update to newly inserted tokens, which helps keep vacuuming needs in check. > This solves the following problems: > > 1. Each email is a transaction instead of each token. > 2. The update statement is only called when we really need an update > which avoids all of those searches. > 3. The looping work is done inside the proc instead of perl calling a > method a zillion times per email. > > I'm not sure how vacuuming will be done yet, if we vacuum once per email > that may be too often, so I may do that every 5 mins in cron. I would suggest leaving an option to have SA vacuum every n emails, since some people may not want to mess with cron, etc. I suspect that pg_autovacuum would be able to keep up with things pretty well, though. -- Jim C. Nasby, Database Consultant decibel@decibel.org Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Sun Jul 31 15:19:28 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C7D5E528FF for ; Sun, 31 Jul 2005 15:19:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 07626-10 for ; Sun, 31 Jul 2005 18:19:15 +0000 (GMT) Received: from moutng.kundenserver.de (moutng.kundenserver.de [212.227.126.187]) by svr1.postgresql.org (Postfix) with ESMTP id C9CF652825 for ; Sun, 31 Jul 2005 15:19:13 -0300 (ADT) Received: from p548F13E4.dip0.t-ipconnect.de [84.143.19.228] (helo=pse.dyndns.org) by mrelayeu.kundenserver.de with ESMTP (Nemesis), id 0MKwtQ-1DzIP31j2N-0006Hv; Sun, 31 Jul 2005 20:19:13 +0200 Received: from pse1 ([192.168.0.3]) by pse.dyndns.org with esmtp (Exim 4.44) id 1DzIP1-0001z5-Vg; Sun, 31 Jul 2005 20:19:12 +0200 Message-ID: <42ED161F.1020408@pse-consulting.de> Date: Sun, 31 Jul 2005 20:19:11 +0200 From: Andreas Pflug User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Jim C. Nasby" Cc: Matthew Schumacher , pgsql-performance@postgresql.org Subject: Re: Performance problems testing with Spamassassin 3.1.0 References: <1122532023.25574.1.camel@localhost.localdomain> <42E9749F.6000709@aptalaska.net> <1122616211.5691.340.camel@lamb.mcmillan.net.nz> <42EA6966.8090402@aptalaska.net> <1122668371.5691.352.camel@lamb.mcmillan.net.nz> <19405.1122670917@sss.pgh.pa.us> <42EAA410.9010700@aptalaska.net> <42EC612A.1080908@arbash-meinel.com> <42ECF3AE.6060809@arbash-meinel.com> <42ED017A.2050207@aptalaska.net> <20050731171012.GS60019@decibel.org> In-Reply-To: <20050731171012.GS60019@decibel.org> X-Enigmail-Version: 0.89.5.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Provags-ID: kundenserver.de abuse@kundenserver.de login:0ce7ee5c3478b8d72edd8e05ccd40b70 X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.272 tagged_above=0 required=5 tests=AWL, FORGED_RCVD_HELO X-Spam-Level: X-Archive-Number: 200507/505 X-Sequence-Number: 13746 Jim C. Nasby wrote: > On Sun, Jul 31, 2005 at 08:51:06AM -0800, Matthew Schumacher wrote: > >>Ok, here is the current plan. >> >>Change the spamassassin API to pass a hash of tokens into the storage >>module, pass the tokens to the proc as an array, start a transaction, >>load the tokens into a temp table using copy, select the tokens distinct >>into the token table for new tokens, update the token table for known >>tokens, then commit. > > > You might consider: > UPDATE tokens > FROM temp_table (this updates existing records) > > INSERT INTO tokens > SELECT ... > FROM temp_table > WHERE NOT IN (SELECT ... FROM tokens) > > This way you don't do an update to newly inserted tokens, which helps > keep vacuuming needs in check. The subselect might be quite a big set, so avoiding a full table scan and materialization by DELETE temp_table WHERE key IN (select key FROM tokens JOIN temp_table); INSERT INTO TOKENS SELECT * FROM temp_table; or INSERT INTO TOKENS SELECT temp_table.* FROM temp_table LEFT JOIN tokens USING (key) WHERE tokens.key IS NULL might be an additional win, assuming that only a small fraction of tokens is inserted and updated. Regards, Andreas From pgsql-performance-owner@postgresql.org Sun Jul 31 18:12:16 2005 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id C8DF0529B6 for ; Sun, 31 Jul 2005 18:12:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 37050-03 for ; Sun, 31 Jul 2005 21:12:05 +0000 (GMT) Received: from floppy.pyrenet.fr (news.pyrenet.fr [194.116.145.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6C8CF529AA for ; Sun, 31 Jul 2005 18:12:03 -0300 (ADT) Received: by floppy.pyrenet.fr (Postfix, from userid 106) id BB3BB3095A; Sun, 31 Jul 2005 23:20:15 +0200 (MET DST) From: William Yu X-Newsgroups: pgsql.performance Subject: Re: Performance problems on 4/8way Opteron (dualcore) Date: Sun, 31 Jul 2005 14:11:58 -0700 Organization: Hub.Org Networking Services Lines: 30 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317) X-Accept-Language: en-us, en In-Reply-To: To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.017 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200507/506 X-Sequence-Number: 13747 A 4xDC would be far more sensitive to poor NUMA code than 2xDC so I'm not surprised I don't see performance issues on our 2xDC w/ < 2.6.12. J. Andrew Rogers wrote: > On 7/30/05 12:57 AM, "William Yu" wrote: > >>I haven't investigated the 2.6.12+ kernel updates yet -- I probably will >>do our development servers first to give it a test. > > > > The kernel updates make the NUMA code dual-core aware, which apparently > makes a big difference in some cases but not in others. It makes some > sense, since multi-processor multi-core machines will have two different > types of non-locality instead of just one that need to be managed. Prior to > the v2.6.12 patches, a dual-core dual-proc machine was viewed as a quad-proc > machine. > > The closest thing to a supported v2.6.12 kernel that I know of is FC4, which > is not really supported in the enterprise sense of course. > > > J. Andrew Rogers > > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster >