From pgsql-performance-owner@postgresql.org Mon Dec 1 06:14:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B886FD1D3FB for ; Mon, 1 Dec 2003 10:14:33 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71598-08 for ; Mon, 1 Dec 2003 06:14:04 -0400 (AST) Received: from mail.networkprograms.com (unknown [203.190.139.254]) by svr1.postgresql.org (Postfix) with ESMTP id 254D7D1D40B for ; Mon, 1 Dec 2003 06:14:00 -0400 (AST) Received: from KAMALR ([192.9.203.78]) by mail.networkprograms.com (8.12.9/8.12.9) with SMTP id hB1AIDfb018881; Mon, 1 Dec 2003 15:48:19 +0530 Message-ID: <01d601c3b7f4$61897f10$4ecb09c0@KAMALR> Reply-To: "Kamalraj Singh Madhan" From: "Kamalraj Singh Madhan" To: "Jason Tishler" Cc: References: <4c0ccf4c27ff.4c27ff4c0ccf@jhmimail.jhmi.edu> <3FC54A82.2040605@familyhealth.com.au> <20031127032831.GA4836@gp.word-to-the-wise.com> <3FC58097.8090803@familyhealth.com.au> <20031127051230.GA6212@gp.word-to-the-wise.com> <20031128050417.GA14227@gp.word-to-the-wise.com> <3FC6DAB8.4080106@familyhealth.com.au> <20031128203700.GA19831@gp.word-to-the-wise.com> <3FC9BC76.2030308@familyhealth.com.au> Subject: Dump restoration via archive files Date: Mon, 1 Dec 2003 15:47:47 +0530 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4807.1700 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300 X-Scanned-By: MIMEDefang 2.35 X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=REFERENCES X-Spam-Level: X-Archive-Number: 200312/1 X-Sequence-Number: 4861 Hi, I'am taking dump of a huge database and do not want the restoration of that dump to take a lot of time as is the case when you take the dump in text files. I want to take the dump as an archive file and get it restored in very less time. I'am not able to figure out what is the command for taking dump of a database in a archive file. Kindly help it's urgent. thanks and regards Kamalraj Singh From pgsql-performance-owner@postgresql.org Mon Dec 1 15:58:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DA9CBD1D47A for ; Mon, 1 Dec 2003 13:02:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96749-03 for ; Mon, 1 Dec 2003 09:01:53 -0400 (AST) Received: from mailout02.sul.t-online.com (mailout02.sul.t-online.com [194.25.134.17]) by svr1.postgresql.org (Postfix) with ESMTP id C0256D1C526 for ; Mon, 1 Dec 2003 09:01:49 -0400 (AST) Received: from fwd09.aul.t-online.de by mailout02.sul.t-online.com with smtp id 1AQngV-0001ZW-09; Mon, 01 Dec 2003 14:01:51 +0100 Received: from router.azrael.de (rPrT4+Z1oeP6G5n6d-8sO+R3M5AcCy9RHbb3u+fHd1wyh19Ib0mScg@[80.141.228.135]) by fmrl09.sul.t-online.com with esmtp id 1AQngB-0lzJB20; Mon, 1 Dec 2003 14:01:31 +0100 Received: from azrael.azrael.de (azrael.azrael.de [192.168.202.18]) by router.azrael.de (8.11.4/8.11.4) with ESMTP id hB1D1Lr09223 for ; Mon, 1 Dec 2003 14:01:22 +0100 Date: Mon, 1 Dec 2003 14:01:42 +0100 From: Evil Azrael Reply-To: Evil Azrael X-Priority: 3 (Normal) Message-ID: <1024180491.20031201140142@evilazrael.ath.cx> To: pgsql-performance@postgresql.org Subject: Various Questions MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Seen: false X-ID: rPrT4+Z1oeP6G5n6d-8sO+R3M5AcCy9RHbb3u+fHd1wyh19Ib0mScg@t-dialin.net X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/16 X-Sequence-Number: 4876 Hi! I have 4 question which probably someone can answer. 1) I have a transaction during which no data was modified, does it make a difference whether i send COMMIT or ROLLBACK? The effect is the same, but what�s about the speed? 2) Is there any general rule when the GEQO will start using an index? Does he consider the number of tuples in the table or the number of data pages? Or is it even more complex even if you don�t tweak the cost setting for the GEQO? 3) Makes it sense to add a index to a table used for logging? I mean the table can grow rather large due to many INSERTs, but is also seldom queried. Does the index slowdown noticable INSERTs? 4) Temporary tables will always be rather slow as they can�t gain from ANALYZE runs, correct? Thanx in advance for any answer Christoph Nelles -- Mit freundlichen Gr�ssen Evil Azrael mailto:root@evilazrael.ath.cx From pgsql-performance-owner@postgresql.org Mon Dec 1 09:10:16 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9E242D1C526 for ; Mon, 1 Dec 2003 13:09:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92420-10 for ; Mon, 1 Dec 2003 09:08:53 -0400 (AST) Received: from mailout06.sul.t-online.com (mailout06.sul.t-online.com [194.25.134.19]) by svr1.postgresql.org (Postfix) with ESMTP id 356A7D1B8B6 for ; Mon, 1 Dec 2003 09:08:31 -0400 (AST) Received: from fwd00.aul.t-online.de by mailout06.sul.t-online.com with smtp id 1AQnmz-00020G-09; Mon, 01 Dec 2003 14:08:33 +0100 Received: from router.azrael.de (Gn9LogZaoeRMfi7Ubrs5eT8Z1KxX37dlF9bZHjeioEU4MIUXBw0X6Z@[80.141.228.135]) by fmrl00.sul.t-online.com with esmtp id 1AQnmB-1tmhpg0; Mon, 1 Dec 2003 14:07:43 +0100 Received: from azrael.azrael.de (azrael.azrael.de [192.168.202.18]) by router.azrael.de (8.11.4/8.11.4) with ESMTP id hB1D7Sr09265 for ; Mon, 1 Dec 2003 14:07:34 +0100 Date: Mon, 1 Dec 2003 14:07:50 +0100 From: Evil Azrael X-Priority: 3 (Normal) Message-ID: <924548300.20031201140750@evilazrael.de> To: pgsql-performance@postgresql.org Subject: Various Questions MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Seen: false X-ID: Gn9LogZaoeRMfi7Ubrs5eT8Z1KxX37dlF9bZHjeioEU4MIUXBw0X6Z@t-dialin.net X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/2 X-Sequence-Number: 4862 Hi! I have 4 question which probably someone can answer. 1) I have a transaction during which no data was modified, does it make a difference whether i send COMMIT or ROLLBACK? The effect is the same, but what�s about the speed? 2) Is there any general rule when the GEQO will start using an index? Does he consider the number of tuples in the table or the number of data pages? Or is it even more complex even if you don�t tweak the cost setting for the GEQO? 3) Makes it sense to add a index to a table used for logging? I mean the table can grow rather large due to many INSERTs, but is also seldom queried. Does the index slowdown noticable INSERTs? 4) Temporary tables will always be rather slow as they can�t gain from ANALYZE runs, correct? Thanx in advance for any answer Christoph Nelles -- Mit freundlichen Gr�ssen Evil Azrael mailto:evilazrael@evilazrael.de From pgsql-performance-owner@postgresql.org Mon Dec 1 09:27:20 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A7C5DD1D388 for ; Mon, 1 Dec 2003 13:27:06 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 95997-07 for ; Mon, 1 Dec 2003 09:26:38 -0400 (AST) Received: from smtp-send.myrealbox.com (smtp-send.myrealbox.com [192.108.102.143]) by svr1.postgresql.org (Postfix) with ESMTP id 43D90D1B46D for ; Mon, 1 Dec 2003 09:26:34 -0400 (AST) Received: from daithan.intranet.pspl.co.in shridhar_daithankar@smtp-send.myrealbox.com [202.54.11.72] by smtp-send.myrealbox.com with NetMail SMTP Agent $Revision: 3.44 $ on Novell NetWare via secured & encrypted transport (TLS); Mon, 01 Dec 2003 06:26:37 -0700 From: Shridhar Daithankar To: pgsql-performance@postgresql.org Subject: Re: Various Questions Date: Mon, 1 Dec 2003 18:56:03 +0530 User-Agent: KMail/1.5.4 References: <924548300.20031201140750@evilazrael.de> In-Reply-To: <924548300.20031201140750@evilazrael.de> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Message-Id: <200312011856.04009.shridhar_daithankar@myrealbox.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/3 X-Sequence-Number: 4863 On Monday 01 December 2003 18:37, Evil Azrael wrote: > 1) I have a transaction during which no data was modified, does it > make a difference whether i send COMMIT or ROLLBACK? The effect is the > same, but what=B4s about the speed? It should not matter. Both commit and rollback should take same amount of= =20 time.. > 2) Is there any general rule when the GEQO will start using an index? > Does he consider the number of tuples in the table or the number of > data pages? Or is it even more complex even if you don=B4t tweak the > cost setting for the GEQO? I thought GEQO was triggered by numebr of join clauses. That is what GEQO c= ost=20 indicates. It is not triggered by number of tuples in any table etc. But correct me if I am wrong. > 3) Makes it sense to add a index to a table used for logging? I mean > the table can grow rather large due to many INSERTs, but is also > seldom queried. Does the index slowdown noticable INSERTs? Yes. It does make a lot of difference. If the table is very seldom queried,= =20 you can probably create the index before querying and drop it later. Howeve= r=20 even this will cost a seq. scan of table and can be heavy on performance.. Take your pick Shridhar From pgsql-performance-owner@postgresql.org Mon Dec 1 09:45:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EC158D1B46D for ; Mon, 1 Dec 2003 13:45:00 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 98413-10 for ; Mon, 1 Dec 2003 09:44:33 -0400 (AST) Received: from www.postgresql.com (www.postgresql.com [200.46.204.209]) by svr1.postgresql.org (Postfix) with ESMTP id 3E5ECD1BC62 for ; Mon, 1 Dec 2003 09:44:30 -0400 (AST) Received: from mist.cti.unav.es (mist.cti.unav.es [159.237.12.28]) by www.postgresql.com (Postfix) with ESMTP id 8EBEDCF72E0 for ; Mon, 1 Dec 2003 09:44:28 -0400 (AST) Received: from norvelle.net ([159.237.106.162]) by mist.cti.unav.es (8.12.10/8.12.10) with ESMTP id hB1DePnF029510 for ; Mon, 1 Dec 2003 14:40:25 +0100 Date: Mon, 1 Dec 2003 14:40:30 +0100 Mime-Version: 1.0 (Apple Message framework v552) Content-Type: multipart/alternative; boundary=Apple-Mail-18--362965852 Subject: My indexes aren't being used (according to EXPLAIN) From: Erik Norvelle To: pgsql-performance@postgresql.org Message-Id: X-Mailer: Apple Mail (2.552) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/4 X-Sequence-Number: 4864 --Apple-Mail-18--362965852 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Greetings: Apologies if this question has already been answered, but I was unable to locate a prior answer in the archives... I have a table with approximately 10 million records, called "indethom", and with an INTEGER column called "clavis" which is set up as a primary key. When I try to perform a select on the table, restricting the result to only the first 100 records, PostgreSQL performs a sequence scan, rather than an index scan (as shown by using EXPLAIN). Needless to say the sequence scan takes forever. Is there some way to get PostgreSQL to use my wonderful indexes? Have I somehow built the indexes incorrectly or something? Here's the description of the table: ====================== PSQL Output Snip ========================= it=> \d indethom Table "public.indethom" Column | Type | Modifiers ---------------+-----------------------+----------- numeoper | smallint | not null nomeoper | character(3) | not null ... (numerous columns skipped) ... verbum | character varying(22) | not null poslinop | integer | not null posverli | smallint | not null posverop | integer | not null clavis | integer | not null articref | integer | sectref | integer | query_counter | integer | Indexes: indethom_pkey primary key btree (clavis), indethom_articulus_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b), indethom_sectio_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b, refere4a, refere4b), it_clavis_ndx btree (clavis), verbum_ndx btree (verbum) it=> explain select * from indethom where clavis < 25; QUERY PLAN ---------------------------------------------------------------------- Seq Scan on indethom (cost=0.00..1336932.65 rows=3543991 width=236) Filter: (clavis < 25) (2 rows) ================== End Snip ===================== Feel free to point me to any FAQ or previous message that already answers this question. Thanks in advance! -Erik Norvelle --Apple-Mail-18--362965852 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Courier NewGreetings: Apologies if this question has already been answered, but I was unable to locate a prior answer in the archives... I have a table with approximately 10 million records, called "indethom", and with an INTEGER column called "clavis" which is set up as a primary key. When I try to perform a select on the table, restricting the result to only the first 100 records, PostgreSQL performs a sequence scan, rather than an index scan (as shown by using EXPLAIN). Needless to say the sequence scan takes forever. Is there some way to get PostgreSQL to use my wonderful indexes? Have I somehow built the indexes incorrectly or something? Here's the description of the table: ====================== PSQL Output Snip ========================= it=> \d indethom Table "public.indethom" Column | Type | Modifiers ---------------+-----------------------+----------- numeoper | smallint | not null nomeoper | character(3) | not null ... (numerous columns skipped) ... verbum | character varying(22) | not null poslinop | integer | not null posverli | smallint | not null posverop | integer | not null clavis | integer | not null articref | integer | sectref | integer | query_counter | integer | Indexes: indethom_pkey primary key btree (clavis), indethom_articulus_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b), indethom_sectio_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b, refere4a, refere4b), it_clavis_ndx btree (clavis), verbum_ndx btree (verbum) it=> explain select * from indethom where clavis << 25; QUERY PLAN ---------------------------------------------------------------------- Seq Scan on indethom (cost=0.00..1336932.65 rows=3543991 width=236) Filter: (clavis << 25) (2 rows) ================== End Snip ===================== Feel free to point me to any FAQ or previous message that already answers this question. Thanks in advance! -Erik Norvelle --Apple-Mail-18--362965852-- From pgsql-performance-owner@postgresql.org Mon Dec 1 10:01:57 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 45B17D1B43E for ; Mon, 1 Dec 2003 14:01:27 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03769-09 for ; Mon, 1 Dec 2003 10:00:58 -0400 (AST) Received: from mail.libertyrms.com (unknown [209.167.124.227]) by svr1.postgresql.org (Postfix) with ESMTP id 046E8D1C941 for ; Mon, 1 Dec 2003 10:00:55 -0400 (AST) Received: from [10.1.2.130] (helo=dba2) by mail.libertyrms.com with esmtp (Exim 3.22 #3 (Debian)) id 1AQobi-0005yQ-00 for ; Mon, 01 Dec 2003 09:00:58 -0500 Received: by dba2 (Postfix, from userid 1019) id 667C8CC8A; Mon, 1 Dec 2003 09:00:58 -0500 (EST) Date: Mon, 1 Dec 2003 09:00:58 -0500 From: Andrew Sullivan To: pgsql-performance@postgresql.org Subject: Re: Various Questions Message-ID: <20031201140058.GA4107@libertyrms.info> Mail-Followup-To: Andrew Sullivan , pgsql-performance@postgresql.org References: <924548300.20031201140750@evilazrael.de> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <924548300.20031201140750@evilazrael.de> User-Agent: Mutt/1.5.4i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/5 X-Sequence-Number: 4865 On Mon, Dec 01, 2003 at 02:07:50PM +0100, Evil Azrael wrote: > 1) I have a transaction during which no data was modified, does it > make a difference whether i send COMMIT or ROLLBACK? The effect is the > same, but what�s about the speed? It makes no difference. > 2) Is there any general rule when the GEQO will start using an index? > Does he consider the number of tuples in the table or the number of > data pages? Or is it even more complex even if you don�t tweak the > cost setting for the GEQO? GEQO is not what causes indexscans. You're thinking of the planner/optimiser. Generally, the optimiser decides what the optimum plan is to deliver a query. This involves a complicated set of rules. The real important question is, "Am I really getting the fastest plan?" You can find out that with EXPLAIN ANALYSE. If you want to know more about what makes a good plan, I'd start by reading the docs, and then by reading the comments in the source code. > 3) Makes it sense to add a index to a table used for logging? I mean > the table can grow rather large due to many INSERTs, but is also > seldom queried. Does the index slowdown noticable INSERTs? It does, but you might find that it's worth it. If it is seldom queried, but you really need the results and the result set is a small % of the table, then you're probably wise to pay the cost of the index at insert, update, and VACUUM because doing a seqscan on a large table to get one or two rows will destroy all your buffers. > 4) Temporary tables will always be rather slow as they can�t gain from > ANALYZE runs, correct? No, you can ANALYSE them yourself. Of course, you'll need an index unless you plan to read the whole table. Note that, if you use temp tables a lot, you need to be sure to vacuum at least pg_class and pg_attribute more frequently than you might have thought. A -- ---- Andrew Sullivan 204-4141 Yonge Street Afilias Canada Toronto, Ontario Canada M2P 2A8 +1 416 646 3304 x110 From pgsql-performance-owner@postgresql.org Mon Dec 1 10:04:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5054CD1B43E for ; Mon, 1 Dec 2003 14:04:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05586-03 for ; Mon, 1 Dec 2003 10:04:25 -0400 (AST) Received: from mail.libertyrms.com (unknown [209.167.124.227]) by svr1.postgresql.org (Postfix) with ESMTP id 87932D1BC5C for ; Mon, 1 Dec 2003 10:04:20 -0400 (AST) Received: from [10.1.2.130] (helo=dba2) by mail.libertyrms.com with esmtp (Exim 3.22 #3 (Debian)) id 1AQoep-00061c-00 for ; Mon, 01 Dec 2003 09:04:11 -0500 Received: by dba2 (Postfix, from userid 1019) id EF081CC8A; Mon, 1 Dec 2003 09:04:10 -0500 (EST) Date: Mon, 1 Dec 2003 09:04:10 -0500 From: Andrew Sullivan To: pgsql-performance@postgresql.org Subject: Re: My indexes aren't being used (according to EXPLAIN) Message-ID: <20031201140410.GC4107@libertyrms.info> Mail-Followup-To: Andrew Sullivan , 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/1.5.4i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/6 X-Sequence-Number: 4866 On Mon, Dec 01, 2003 at 02:40:30PM +0100, Erik Norvelle wrote: > > it=> explain select * from indethom where clavis < 25; What's the percentage of the table where clavis < 25? Have you ANALYSEd recently? What does the pg_stats view tell you about this table? > Feel free to point me to any FAQ or previous message that already > answers this question. Thanks in advance! This is a pretty common sort of problem. See the archives of this list for several fairly recent discussions of these sorts of problems. A -- ---- Andrew Sullivan 204-4141 Yonge Street Afilias Canada Toronto, Ontario Canada M2P 2A8 +1 416 646 3304 x110 From pgsql-performance-owner@postgresql.org Mon Dec 1 10:18:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 674ACD1D8B5 for ; Mon, 1 Dec 2003 14:17:30 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05776-07 for ; Mon, 1 Dec 2003 10:17:02 -0400 (AST) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 7DB07D1D5AD for ; Mon, 1 Dec 2003 10:16:54 -0400 (AST) Received: (qmail 6347 invoked from network); 1 Dec 2003 14:16:55 -0000 Received: from localhost (HELO squeegit) (threshar@127.0.0.1) by localhost with SMTP; 1 Dec 2003 14:16:55 -0000 Date: Mon, 1 Dec 2003 09:16:34 -0500 From: Jeff To: "Kamalraj Singh Madhan" Cc: jason@tishler.net, pgsql-performance@postgresql.org Subject: Re: Dump restoration via archive files Message-Id: <20031201091634.6f19f28a.threshar@torgo.978.org> In-Reply-To: <01d601c3b7f4$61897f10$4ecb09c0@KAMALR> References: <4c0ccf4c27ff.4c27ff4c0ccf@jhmimail.jhmi.edu> <3FC54A82.2040605@familyhealth.com.au> <20031127032831.GA4836@gp.word-to-the-wise.com> <3FC58097.8090803@familyhealth.com.au> <20031127051230.GA6212@gp.word-to-the-wise.com> <20031128050417.GA14227@gp.word-to-the-wise.com> <3FC6DAB8.4080106@familyhealth.com.au> <20031128203700.GA19831@gp.word-to-the-wise.com> <3FC9BC76.2030308@familyhealth.com.au> <01d601c3b7f4$61897f10$4ecb09c0@KAMALR> X-Mailer: Sylpheed version 0.9.7 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/7 X-Sequence-Number: 4867 On Mon, 1 Dec 2003 15:47:47 +0530 "Kamalraj Singh Madhan" wrote: > Hi, > I'am taking dump of a huge database and do not want the > restoration of > that dump to take a lot of time as is the case when you take the dump > in text files. I want to take the dump as an archive file and get it > restored in very less time. I'am not able to figure out what is the > command for taking dump of a database in a archive file. Kindly help > it's urgent. > Fast backups are an area PG needs work in. Currently, PG has no 'archive file backup'. You do have the following options to get around this: 1. Take big db offline, copy $PGDATA. Has a restore time of how long it takes to copy $PGDATA (And optionally untar/gzip), bring db back online 2. If you are using an LVM, take a snapshot and copy the data. Like #1, it also has a "0" restore time. 3. If you are using a pg_dump generated dump, be sure to really jack up your sort_mem - this will be a HUGE benefit when creating indexes & if you are using 7.4, adding the foriegn keys. Also turning off fsync (Don't forget to turn it back on after your restore!) cna give you some nice speed increases. 4. If you are not using 7.4 and using pg_dump, there isn't much you can do about adding foreign keys going stupidly slow :( -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Mon Dec 1 10:29:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BDE34D1B437 for ; Mon, 1 Dec 2003 14:29:49 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11618-04 for ; Mon, 1 Dec 2003 10:29:22 -0400 (AST) Received: from penguin.goodinassociates.com (unknown [63.150.225.202]) by svr1.postgresql.org (Postfix) with ESMTP id B73FCD1B438 for ; Mon, 1 Dec 2003 10:29:17 -0400 (AST) Received: from [192.168.14.183] (bluejay.goodinassociates.com [192.168.14.183]) by penguin.goodinassociates.com (8.12.8/linuxconf) with ESMTP id hB1ET3na005958; Mon, 1 Dec 2003 08:29:06 -0600 Subject: Re: cross table indexes or something? From: Jeremiah Jahn To: Hannu Krosing Cc: postgres performance In-Reply-To: <1069885949.3176.57.camel@fuji.krosing.net> References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <1069885949.3176.57.camel@fuji.krosing.net> Content-Type: text/plain Message-Id: <1070288942.22346.27.camel@bluejay.goodinassociates.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 (1.4.5-7) Date: Mon, 01 Dec 2003 08:29:03 -0600 Content-Transfer-Encoding: 7bit X-RAVMilter-Version: 8.4.4(snapshot 20030410) (penguin.goodinassociates.com) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/8 X-Sequence-Number: 4868 On Wed, 2003-11-26 at 16:32, Hannu Krosing wrote: > Jeremiah Jahn kirjutas K, 26.11.2003 kell 22:14: > > I was wondering if there is something I can do that would act similar to > > a index over more than one table. > > > > I have about 3 million people in my DB at the moment, they all have > > roles, and many of them have more than one name. > > > > for example, a Judge will only have one name, but a Litigant could have > > multiple aliases. Things go far to slow when I do a query on a judge > > named smith. > > If you dont need all the judges named smith you could try to use LIMIT. Unfortunately I do need all of the judges named smith. > > Have you run ANALYZE ? Why does DB think that there is only one judge > with name like SMITH% ? I've attached the Analyze below. I have no idea why the db thinks there is only 1 judge named simth. Is there some what I can inform the DB about this. In actuality, there aren't any judges named smith at the moment, but there are 22K people named smith. > > ------------- > Hannu > > P.S. > Always send EXPLAIN ANALYZE output if asking for advice on [PERFORM] EXPLAIN ANALYZE select distinct actor.actor_id,court.id,court.name,role_class_code,full_name from actor,identity,court,event,event_actor where role_class_code = 'Judge' and full_name like 'SMITH%' and identity.actor_id = actor.actor_id and identity.court_ori = actor.court_ori and actor.court_ori = court.id and actor.actor_id = event_actor.actor_id and event_actor.event_id = event.event_id and event_date_time > '20021126' order by full_name; QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- Unique (cost=686.42..686.44 rows=1 width=92) (actual time=111923.877..111923.877 rows=0 loops=1) -> Sort (cost=686.42..686.43 rows=1 width=92) (actual time=111923.873..111923.873 rows=0 loops=1) Sort Key: identity.full_name, actor.actor_id, court.id, court.name, actor.role_class_code -> Nested Loop (cost=8.45..686.41 rows=1 width=92) (actual time=111923.836..111923.836 rows=0 loops=1) -> Nested Loop (cost=8.45..680.57 rows=1 width=144) (actual time=109958.426..111157.822 rows=2449 loops=1) -> Hash Join (cost=8.45..9.62 rows=1 width=117) (actual time=109945.754..109945.896 rows=6 loops=1) Hash Cond: (("outer".id)::text = ("inner".court_ori)::text) -> Seq Scan on court (cost=0.00..1.10 rows=10 width=34) (actual time=0.015..0.048 rows=10 loops=1) -> Hash (cost=8.45..8.45 rows=1 width=109) (actual time=109940.161..109940.161 rows=0 loops=1) -> Nested Loop (cost=0.00..8.45 rows=1 width=109) (actual time=10.367..109940.079 rows=7 loops=1) Join Filter: (("outer".court_ori)::text = ("inner".court_ori)::text) -> Index Scan using name_speed on identity (cost=0.00..3.01 rows=1 width=59) (actual time=10.202..238.497 rows=22436 loops=1) Index Cond: (((full_name)::text >= 'SMITH'::character varying) AND ((full_name)::text < 'SMITI'::character varying)) Filter: ((full_name)::text ~~ 'SMITH%'::text) -> Index Scan using actor_speed on actor (cost=0.00..5.42 rows=1 width=50) (actual time=4.883..4.883 rows=0 loops=22436) Index Cond: (("outer".actor_id)::text = (actor.actor_id)::text) Filter: ((role_class_code)::text = 'Judge'::text) -> Index Scan using event_actor_speed on event_actor (cost=0.00..655.59 rows=1229 width=73) (actual time=11.815..198.759 rows=408 loops=6) Index Cond: ((event_actor.actor_id)::text = ("outer".actor_id)::text) -> Index Scan using event_pkey on event (cost=0.00..5.83 rows=1 width=52) (actual time=0.308..0.308 rows=0 loops=2449) Index Cond: (("outer".event_id)::text = (event.event_id)::text) Filter: (event_date_time > '20021126'::bpchar) Total runtime: 111924.833 ms (23 rows) > > ------------- > Hannu > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster -- Jeremiah Jahn From pgsql-performance-owner@postgresql.org Mon Dec 1 11:26:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 03A6ED1DA87 for ; Mon, 1 Dec 2003 15:11:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20152-08 for ; Mon, 1 Dec 2003 11:11:28 -0400 (AST) Received: from mist.cti.unav.es (mist.cti.unav.es [159.237.12.28]) by svr1.postgresql.org (Postfix) with ESMTP id 4F84CD1B465 for ; Mon, 1 Dec 2003 11:11:03 -0400 (AST) Received: from norvelle.net ([159.237.106.162]) by mist.cti.unav.es (8.12.10/8.12.10) with ESMTP id hB1FB5nF032017 for ; Mon, 1 Dec 2003 16:11:06 +0100 Date: Mon, 1 Dec 2003 16:11:11 +0100 Subject: Re: My indexes aren't being used (according to EXPLAIN) Content-Type: text/plain; charset=US-ASCII; format=flowed Mime-Version: 1.0 (Apple Message framework v552) From: Erik Norvelle To: pgsql-performance@postgresql.org Content-Transfer-Encoding: 7bit In-Reply-To: <20031201140410.GC4107@libertyrms.info> Message-Id: <99468E86-2410-11D8-BDFB-000A9583BF06@norvelle.net> X-Mailer: Apple Mail (2.552) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/10 X-Sequence-Number: 4870 The ANALYSE did the trick... Thanks! Will also read through the archives... -Erik On lunes, dici 1, 2003, at 15:04 Europe/Madrid, Andrew Sullivan wrote: > On Mon, Dec 01, 2003 at 02:40:30PM +0100, Erik Norvelle wrote: >> >> it=> explain select * from indethom where clavis < 25; > > What's the percentage of the table where clavis < 25? Have you > ANALYSEd recently? What does the pg_stats view tell you about this > table? > >> Feel free to point me to any FAQ or previous message that already >> answers this question. Thanks in advance! > > This is a pretty common sort of problem. See the archives of this > list for several fairly recent discussions of these sorts of > problems. > > A > > -- > ---- > Andrew Sullivan 204-4141 Yonge Street > Afilias Canada Toronto, Ontario Canada > M2P 2A8 > +1 416 646 3304 x110 > > > ---------------------------(end of > broadcast)--------------------------- > TIP 6: Have you searched our list archives? > > http://archives.postgresql.org > From pgsql-performance-owner@postgresql.org Mon Dec 1 11:25:56 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 49C5FD1B465 for ; Mon, 1 Dec 2003 15:24:35 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 22872-04 for ; Mon, 1 Dec 2003 11:24:08 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 75C84D1B43E for ; Mon, 1 Dec 2003 11:24:04 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id A7636355D6; Mon, 1 Dec 2003 07:23:53 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id A3F1E354CA; Mon, 1 Dec 2003 07:23:53 -0800 (PST) Date: Mon, 1 Dec 2003 07:23:53 -0800 (PST) From: Stephan Szabo To: Jeff Cc: Kamalraj Singh Madhan , jason@tishler.net, pgsql-performance@postgresql.org Subject: Re: Dump restoration via archive files In-Reply-To: <20031201091634.6f19f28a.threshar@torgo.978.org> Message-ID: <20031201071943.G44545@megazone.bigpanda.com> References: <4c0ccf4c27ff.4c27ff4c0ccf@jhmimail.jhmi.edu> <3FC54A82.2040605@familyhealth.com.au> <20031127032831.GA4836@gp.word-to-the-wise.com> <3FC58097.8090803@familyhealth.com.au> <20031127051230.GA6212@gp.word-to-the-wise.com> <20031128050417.GA14227@gp.word-to-the-wise.com> <3FC6DAB8.4080106@familyhealth.com.au> <20031128203700.GA19831@gp.word-to-the-wise.com> <3FC9BC76.2030308@familyhealth.com.au> <01d601c3b7f4$61897f10$4ecb09c0@KAMALR> <20031201091634.6f19f28a.threshar@torgo.978.org> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/9 X-Sequence-Number: 4869 On Mon, 1 Dec 2003, Jeff wrote: > On Mon, 1 Dec 2003 15:47:47 +0530 > "Kamalraj Singh Madhan" wrote: > > 4. If you are not using 7.4 and using pg_dump, there isn't much you can > do about adding foreign keys going stupidly slow :( You can take a schema dump and a separate data only dump where the latter specifies --disable-triggers which should disable the checks when the data is being added. From pgsql-performance-owner@postgresql.org Mon Dec 1 12:01:54 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5B598D1D5A3 for ; Mon, 1 Dec 2003 16:00:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27262-09 for ; Mon, 1 Dec 2003 12:00:15 -0400 (AST) Received: from anchor-post-32.mail.demon.net (anchor-post-32.mail.demon.net [194.217.242.90]) by svr1.postgresql.org (Postfix) with ESMTP id 89847D1B45C for ; Mon, 1 Dec 2003 11:59:59 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-32.mail.demon.net with esmtp (Exim 3.35 #1) id 1AQqSr-000Gja-0W; Mon, 01 Dec 2003 15:59:57 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 679A816C88; Mon, 1 Dec 2003 15:59:55 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 44A6E1665E; Mon, 1 Dec 2003 15:59:53 +0000 (GMT) From: Richard Huxton To: Jeremiah Jahn , Hannu Krosing Subject: Re: cross table indexes or something? Date: Mon, 1 Dec 2003 15:59:52 +0000 User-Agent: KMail/1.5 Cc: postgres performance References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <1069885949.3176.57.camel@fuji.krosing.net> <1070288942.22346.27.camel@bluejay.goodinassociates.com> In-Reply-To: <1070288942.22346.27.camel@bluejay.goodinassociates.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312011559.52880.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/11 X-Sequence-Number: 4871 On Monday 01 December 2003 14:29, Jeremiah Jahn wrote: > On Wed, 2003-11-26 at 16:32, Hannu Krosing wrote: > > Jeremiah Jahn kirjutas K, 26.11.2003 kell 22:14: > > > I was wondering if there is something I can do that would act similar > > > to a index over more than one table. > > > > > > I have about 3 million people in my DB at the moment, they all have > > > roles, and many of them have more than one name. > > > > > > for example, a Judge will only have one name, but a Litigant could have > > > multiple aliases. Things go far to slow when I do a query on a judge > > > named smith. > > > > If you dont need all the judges named smith you could try to use LIMIT. > > Unfortunately I do need all of the judges named smith. > > > Have you run ANALYZE ? Why does DB think that there is only one judge > > with name like SMITH% ? > > I've attached the Analyze below. I have no idea why the db thinks there > is only 1 judge named simth. Is there some what I can inform the DB > about this. In actuality, there aren't any judges named smith at the > moment, but there are 22K people named smith. It's guessing there's approximately 1. I don't think PG measures cross-correlation of various columns cross-table. If role_class_code on table actor? If so, try: CREATE INDEX test_judge_idx ON actor (actor_id) WHERE role_class_code = 'Judge'; And then similar for the other class-codes (assuming you've not got too many of them). Or even just an index on (actor_id,role_class_code). If role_class_code is on a different table, can you say which one? The problem is clearly this step: > -> Index Scan using actor_speed on > actor (cost=0.00..5.42 rows=1 width=50) (actual time=4.883..4.883 rows=0 > loops=22436) > Index Cond: (("outer".actor_id)::text = > (actor.actor_id)::text) Filter: ((role_class_code)::text = 'Judge'::text) Thats 4.883 * 22436 loops = 109555 milliseconds. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Mon Dec 1 12:26:42 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E2833D1BC54 for ; Mon, 1 Dec 2003 16:25:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33340-04 for ; Mon, 1 Dec 2003 12:25:10 -0400 (AST) Received: from mailhost3.tudelft.nl (mailhost3.tudelft.nl [130.161.180.14]) by svr1.postgresql.org (Postfix) with ESMTP id 8DC79D1B4AA for ; Mon, 1 Dec 2003 12:25:09 -0400 (AST) Received: from 127.0.0.1 (localhost [127.0.0.1]) by rav.antivirus (Postfix) with SMTP id 19872263E; Mon, 1 Dec 2003 17:25:08 +0100 (MET) Received: from listserv.tudelft.nl (witlab.tudelft.nl [130.161.180.33]) by mailhost3.tudelft.nl (Postfix) with ESMTP id 0190927D5; Mon, 1 Dec 2003 17:25:08 +0100 (MET) Received: from acm (x193056-2.shuis-s.tudelft.nl [145.94.193.58]) by listserv.tudelft.nl (8.12.10/8.12.8) with ESMTP id hB1GOwQQ025391; Mon, 1 Dec 2003 17:25:07 +0100 (MET) From: "Arjen van der Meijden" To: "'Jeremiah Jahn'" , "'Hannu Krosing'" Cc: "'postgres performance'" Subject: Re: cross table indexes or something? Date: Mon, 1 Dec 2003 17:24:43 +0100 Message-ID: <002901c3b827$a48aeff0$3ac15e91@acm> 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, Build 10.0.2627 In-Reply-To: <1070288942.22346.27.camel@bluejay.goodinassociates.com> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0 tests=BAYES_30, FORGED_RCVD_TRAIL, IN_REP_TO, QUOTED_EMAIL_TEXT X-Spam-Level: X-Archive-Number: 200312/12 X-Sequence-Number: 4872 > Jeremiah Jahn wrote: > > > Have you run ANALYZE ? Why does DB think that there is only > one judge > > with name like SMITH% ? > I've attached the Analyze below. I have no idea why the db > thinks there is only 1 judge named simth. Is there some what > I can inform the DB about this. In actuality, there aren't > any judges named smith at the moment, but there are 22K > people named smith. > I think you're mistaking the command EXPLAIN ANALYZE for the command ANALYZE. Have you actually run the command ANALYZE or perhaps even better if you haven't vacuumed before: VACUUM FULL ANALYZE If you have no idea what vacuum is, check the manual. If you've already run such a VACUUM/ANALYZE-command, then ignore this message :) Best regards, Arjen van der Meijden From pgsql-performance-owner@postgresql.org Mon Dec 1 15:58:44 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A482BD1B432 for ; Mon, 1 Dec 2003 17:15:23 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 43158-03 for ; Mon, 1 Dec 2003 13:14:52 -0400 (AST) Received: from pcs-1.paccomsys.com (208.225.nwc.net [207.151.225.208]) by svr1.postgresql.org (Postfix) with ESMTP id 7BA2BD1B46B for ; Mon, 1 Dec 2003 13:14:50 -0400 (AST) Received: from musicreports.com (mail.musicreports.com [64.161.179.34]) by pcs-1.paccomsys.com (8.12.5/8.12.5) with ESMTP id hB1Gs574031370; Mon, 1 Dec 2003 08:54:06 -0800 Message-ID: <3FCB7708.1020305@musicreports.com> Date: Mon, 01 Dec 2003 09:14:48 -0800 From: Roger Ging User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030708 Debian/1.3-4.lindows43 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Followup - expression (functional) index use in joins References: <3FC4F360.2090609@paccomsys.com> <200311261912.01286.dev@archonet.com> <3FC51B2C.2030700@paccomsys.com> <18733.1069891765@sss.pgh.pa.us> In-Reply-To: <18733.1069891765@sss.pgh.pa.us> Content-Type: multipart/alternative; boundary="------------060907050500000000060500" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/17 X-Sequence-Number: 4877 This is a multi-part message in MIME format. --------------060907050500000000060500 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Tom, Turning enable_hashjoin off made the query run as it had on v7.3. We have worked around this by changing the index from a function call to a direct index on a new column with the results of the function maintained by a trigger. Would there be performance issues from leaving enable_hashjoin off, or do you recomend enabling it, and working around function calls in indices? See results below. I was not sure if I was supposed to reply-all, or just to the list. Sorry if the protocol is incorrect. ppl=# explain analyse select title from music.program p ppl-# join music.logfile l on ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no) ppl-# where l.air_date = '01/30/2001' ppl-# and l.station = 'KABC'; QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------------------------------- Hash Join (cost=69.89..19157.06 rows=2322 width=28) (actual time=500.905..1473.748 rows=242 loops=1) Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text) -> Seq Scan on program p (cost=0.00..16888.98 rows=173998 width=40) (actual time=98.371..532.184 rows=173998 loops=1) -> Hash (cost=69.84..69.84 rows=17 width=9) (actual time=65.817..65.817 rows=0 loops=1) -> Index Scan using idx_logfile_station_air_date on logfile l (cost=0.00..69.84 rows=17 width=9) (actual time=24.499..65.730 rows=32 loops=1) Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone)) Total runtime: 1474.067 ms (7 rows) ppl=# set enable_mergejoin = false; SET ppl=# explain analyse select title from music.program p ppl-# join music.logfile l on ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no) ppl-# where l.air_date = '01/30/2001' ppl-# and l.station = 'KABC'; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------------------- Hash Join (cost=69.89..19157.06 rows=2322 width=28) (actual time=444.834..1428.815 rows=242 loops=1) Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text) -> Seq Scan on program p (cost=0.00..16888.98 rows=173998 width=40) (actual time=105.977..542.870 rows=173998 loops=1) -> Hash (cost=69.84..69.84 rows=17 width=9) (actual time=1.197..1.197 rows=0 loops=1) -> Index Scan using idx_logfile_station_air_date on logfile l (cost=0.00..69.84 rows=17 width=9) (actual time=0.574..1.151 rows=32 loops=1) Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone)) Total runtime: 1429.111 ms (7 rows) ppl=# set enable_hashjoin = false; SET ppl=# explain analyse select title from music.program p ppl-# join music.logfile l on ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no) ppl-# where l.air_date = '01/30/2001' ppl-# and l.station = 'KABC'; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------------------ Nested Loop (cost=0.00..58104.34 rows=2322 width=28) (actual time=0.480..5.357 rows=242 loops=1) -> Index Scan using idx_logfile_station_air_date on logfile l (cost=0.00..69.84 rows=17 width=9) (actual time=0.176..0.754 rows=32 loops=1) Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone)) -> Index Scan using idx_program_mri_id_no_program on program p (cost=0.00..3400.74 rows=870 width=40) (actual time=0.041..0.127 rows=8 loops=32) Index Cond: (("outer".program_id)::text = (music.fn_mri_id_no_program(p.mri_id_no))::text) Total runtime: 5.637 ms (6 rows) Tom Lane wrote: >Roger Ging writes: > > >>Ran vacuum analyse on both program and logfile tables. Estimates are >>more in line with reality now, >> >> > >And they are what now? You really can't expect to get useful help here >when you're being so miserly with the details ... > >FWIW, I suspect you could force 7.4 to generate 7.3's plan by setting >enable_mergejoin to off (might have to also set enable_hashjoin to off, >if it then tries for a hash join). 7.3 could not even consider those >join types in this example, while 7.4 can. The interesting question >from my perspective is why the planner is guessing wrong about the >relative costs of the plans. EXPLAIN ANALYZE results with each type of >join forced would be useful to look at. > > regards, tom lane > >---------------------------(end of broadcast)--------------------------- >TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > > --------------060907050500000000060500 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Tom,

Turning enable_hashjoin off made the query run as it had on v7.3.  We have worked around this by changing the index from a function call to a direct index on a new column with the results of the function maintained by a trigger.  Would there be performance issues from leaving enable_hashjoin off, or do you recomend enabling it, and working around function calls in indices?

See results below.  I was not sure if I was supposed to reply-all, or just to the list.  Sorry if the protocol is incorrect.



ppl=# explain analyse select title from music.program p
ppl-# join music.logfile l on
ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no)
ppl-# where l.air_date = '01/30/2001'
ppl-# and l.station = 'KABC';
                                                                       QUERY PLAN
---------------------------------------------------------------------------------------------------------------------------------------------------------
 Hash Join  (cost=69.89..19157.06 rows=2322 width=28) (actual time=500.905..1473.748 rows=242 loops=1)
   Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text)
   ->  Seq Scan on program p  (cost=0.00..16888.98 rows=173998 width=40) (actual time=98.371..532.184 rows=173998 loops=1)
   ->  Hash  (cost=69.84..69.84 rows=17 width=9) (actual time=65.817..65.817 rows=0 loops=1)
         ->  Index Scan using idx_logfile_station_air_date on logfile l  (cost=0.00..69.84 rows=17 width=9) (actual time=24.499..65.730 rows=32 loops=1)
               Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone))
 Total runtime: 1474.067 ms
(7 rows)

ppl=# set enable_mergejoin = false;
SET
ppl=# explain analyse select title from music.program p
ppl-# join music.logfile l on
ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no)
ppl-# where l.air_date = '01/30/2001'
ppl-# and l.station = 'KABC';
                                                                      QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------------------------------------
 Hash Join  (cost=69.89..19157.06 rows=2322 width=28) (actual time=444.834..1428.815 rows=242 loops=1)
   Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text)
   ->  Seq Scan on program p  (cost=0.00..16888.98 rows=173998 width=40) (actual time=105.977..542.870 rows=173998 loops=1)
   ->  Hash  (cost=69.84..69.84 rows=17 width=9) (actual time=1.197..1.197 rows=0 loops=1)
         ->  Index Scan using idx_logfile_station_air_date on logfile l  (cost=0.00..69.84 rows=17 width=9) (actual time=0.574..1.151 rows=32 loops=1)
               Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone))
 Total runtime: 1429.111 ms
(7 rows)

ppl=# set enable_hashjoin = false;
SET
ppl=# explain analyse select title from music.program p
ppl-# join music.logfile l on
ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no)
ppl-# where l.air_date = '01/30/2001'
ppl-# and l.station = 'KABC';
                                                                      QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------------------
 Nested Loop  (cost=0.00..58104.34 rows=2322 width=28) (actual time=0.480..5.357 rows=242 loops=1)
   ->  Index Scan using idx_logfile_station_air_date on logfile l  (cost=0.00..69.84 rows=17 width=9) (actual time=0.176..0.754 rows=32 loops=1)
         Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone))
   ->  Index Scan using idx_program_mri_id_no_program on program p  (cost=0.00..3400.74 rows=870 width=40) (actual time=0.041..0.127 rows=8 loops=32)
         Index Cond: (("outer".program_id)::text = (music.fn_mri_id_no_program(p.mri_id_no))::text)
 Total runtime: 5.637 ms
(6 rows)


Tom Lane wrote:
Roger Ging <rging@paccomsys.com> writes:
  
Ran vacuum analyse on both program and logfile tables.  Estimates are 
more in line with reality now,
    

And they are what now?  You really can't expect to get useful help here
when you're being so miserly with the details ...

FWIW, I suspect you could force 7.4 to generate 7.3's plan by setting
enable_mergejoin to off (might have to also set enable_hashjoin to off,
if it then tries for a hash join).  7.3 could not even consider those
join types in this example, while 7.4 can.  The interesting question
from my perspective is why the planner is guessing wrong about the
relative costs of the plans.  EXPLAIN ANALYZE results with each type of
join forced would be useful to look at.

			regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

               http://www.postgresql.org/docs/faqs/FAQ.html
--------------060907050500000000060500-- From pgsql-performance-owner@postgresql.org Mon Dec 1 13:31:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B1797D1D61F for ; Mon, 1 Dec 2003 17:31:40 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 43508-08 for ; Mon, 1 Dec 2003 13:31:10 -0400 (AST) Received: from pcs-1.paccomsys.com (208.225.nwc.net [207.151.225.208]) by svr1.postgresql.org (Postfix) with ESMTP id CB4E5D1B486 for ; Mon, 1 Dec 2003 13:31:07 -0400 (AST) Received: from paccomsys.com (mail.musicreports.com [64.161.179.34]) by pcs-1.paccomsys.com (8.12.5/8.12.5) with ESMTP id hB1HAN74031563 for ; Mon, 1 Dec 2003 09:10:24 -0800 Message-ID: <3FCB7ADA.8020101@paccomsys.com> Date: Mon, 01 Dec 2003 09:31:06 -0800 From: Roger Ging User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030708 Debian/1.3-4.lindows43 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Followup - expression (functional) index use in joins References: <3FC4F360.2090609@paccomsys.com> <200311261912.01286.dev@archonet.com> <3FC51B2C.2030700@paccomsys.com> <18733.1069891765@sss.pgh.pa.us> In-Reply-To: <18733.1069891765@sss.pgh.pa.us> Content-Type: multipart/alternative; boundary="------------080104020701070107090204" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/13 X-Sequence-Number: 4873 This is a multi-part message in MIME format. --------------080104020701070107090204 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Turning enable_hashjoin off made the query run as it had on v7.3. We have worked around this by changing the index from a function call to a direct index on a new column with the results of the function maintained by a trigger. Would there be performance issues from leaving enable_hashjoin off, or do you recomend enabling it, and working around function calls in indices? See results below. ppl=# explain analyse select title from music.program p ppl-# join music.logfile l on ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no) ppl-# where l.air_date = '01/30/2001' ppl-# and l.station = 'KABC'; QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------------------------------- Hash Join (cost=69.89..19157.06 rows=2322 width=28) (actual time=500.905..1473.748 rows=242 loops=1) Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text) -> Seq Scan on program p (cost=0.00..16888.98 rows=173998 width=40) (actual time=98.371..532.184 rows=173998 loops=1) -> Hash (cost=69.84..69.84 rows=17 width=9) (actual time=65.817..65.817 rows=0 loops=1) -> Index Scan using idx_logfile_station_air_date on logfile l (cost=0.00..69.84 rows=17 width=9) (actual time=24.499..65.730 rows=32 loops=1) Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone)) Total runtime: 1474.067 ms (7 rows) ppl=# set enable_mergejoin = false; SET ppl=# explain analyse select title from music.program p ppl-# join music.logfile l on ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no) ppl-# where l.air_date = '01/30/2001' ppl-# and l.station = 'KABC'; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------------------- Hash Join (cost=69.89..19157.06 rows=2322 width=28) (actual time=444.834..1428.815 rows=242 loops=1) Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text) -> Seq Scan on program p (cost=0.00..16888.98 rows=173998 width=40) (actual time=105.977..542.870 rows=173998 loops=1) -> Hash (cost=69.84..69.84 rows=17 width=9) (actual time=1.197..1.197 rows=0 loops=1) -> Index Scan using idx_logfile_station_air_date on logfile l (cost=0.00..69.84 rows=17 width=9) (actual time=0.574..1.151 rows=32 loops=1) Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone)) Total runtime: 1429.111 ms (7 rows) ppl=# set enable_hashjoin = false; SET ppl=# explain analyse select title from music.program p ppl-# join music.logfile l on ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no) ppl-# where l.air_date = '01/30/2001' ppl-# and l.station = 'KABC'; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------------------ Nested Loop (cost=0.00..58104.34 rows=2322 width=28) (actual time=0.480..5.357 rows=242 loops=1) -> Index Scan using idx_logfile_station_air_date on logfile l (cost=0.00..69.84 rows=17 width=9) (actual time=0.176..0.754 rows=32 loops=1) Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone)) -> Index Scan using idx_program_mri_id_no_program on program p (cost=0.00..3400.74 rows=870 width=40) (actual time=0.041..0.127 rows=8 loops=32) Index Cond: (("outer".program_id)::text = (music.fn_mri_id_no_program(p.mri_id_no))::text) Total runtime: 5.637 ms (6 rows) Tom Lane wrote: >Roger Ging writes: > > >>Ran vacuum analyse on both program and logfile tables. Estimates are >>more in line with reality now, >> >> > >And they are what now? You really can't expect to get useful help here >when you're being so miserly with the details ... > >FWIW, I suspect you could force 7.4 to generate 7.3's plan by setting >enable_mergejoin to off (might have to also set enable_hashjoin to off, >if it then tries for a hash join). 7.3 could not even consider those >join types in this example, while 7.4 can. The interesting question >from my perspective is why the planner is guessing wrong about the >relative costs of the plans. EXPLAIN ANALYZE results with each type of >join forced would be useful to look at. > > regards, tom lane > >---------------------------(end of broadcast)--------------------------- >TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > > Tom Lane wrote: >Roger Ging writes: > > >>Ran vacuum analyse on both program and logfile tables. Estimates are >>more in line with reality now, >> >> > >And they are what now? You really can't expect to get useful help here >when you're being so miserly with the details ... > >FWIW, I suspect you could force 7.4 to generate 7.3's plan by setting >enable_mergejoin to off (might have to also set enable_hashjoin to off, >if it then tries for a hash join). 7.3 could not even consider those >join types in this example, while 7.4 can. The interesting question >from my perspective is why the planner is guessing wrong about the >relative costs of the plans. EXPLAIN ANALYZE results with each type of >join forced would be useful to look at. > > regards, tom lane > >---------------------------(end of broadcast)--------------------------- >TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > > --------------080104020701070107090204 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit
Turning enable_hashjoin off made the query run as it had on v7.3.  We have worked around this by changing the index from a function call to a direct index on a new column with the results of the function maintained by a trigger.  Would there be performance issues from leaving enable_hashjoin off, or do you recomend enabling it, and working around function calls in indices?

See results below.



ppl=# explain analyse select title from music.program p
ppl-# join music.logfile l on
ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no)
ppl-# where l.air_date = '01/30/2001'
ppl-# and l.station = 'KABC';
                                                                       QUERY PLAN
---------------------------------------------------------------------------------------------------------------------------------------------------------
 Hash Join  (cost=69.89..19157.06 rows=2322 width=28) (actual time=500.905..1473.748 rows=242 loops=1)
   Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text)
   ->  Seq Scan on program p  (cost=0.00..16888.98 rows=173998 width=40) (actual time=98.371..532.184 rows=173998 loops=1)
   ->  Hash  (cost=69.84..69.84 rows=17 width=9) (actual time=65.817..65.817 rows=0 loops=1)
         ->  Index Scan using idx_logfile_station_air_date on logfile l  (cost=0.00..69.84 rows=17 width=9) (actual time=24.499..65.730 rows=32 loops=1)
               Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone))
 Total runtime: 1474.067 ms
(7 rows)

ppl=# set enable_mergejoin = false;
SET
ppl=# explain analyse select title from music.program p
ppl-# join music.logfile l on
ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no)
ppl-# where l.air_date = '01/30/2001'
ppl-# and l.station = 'KABC';
                                                                      QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------------------------------------
 Hash Join  (cost=69.89..19157.06 rows=2322 width=28) (actual time=444.834..1428.815 rows=242 loops=1)
   Hash Cond: ((music.fn_mri_id_no_program("outer".mri_id_no))::text = ("inner".program_id)::text)
   ->  Seq Scan on program p  (cost=0.00..16888.98 rows=173998 width=40) (actual time=105.977..542.870 rows=173998 loops=1)
   ->  Hash  (cost=69.84..69.84 rows=17 width=9) (actual time=1.197..1.197 rows=0 loops=1)
         ->  Index Scan using idx_logfile_station_air_date on logfile l  (cost=0.00..69.84 rows=17 width=9) (actual time=0.574..1.151 rows=32 loops=1)
               Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone))
 Total runtime: 1429.111 ms
(7 rows)

ppl=# set enable_hashjoin = false;
SET
ppl=# explain analyse select title from music.program p
ppl-# join music.logfile l on
ppl-# l.program_id = music.fn_mri_id_no_program(p.mri_id_no)
ppl-# where l.air_date = '01/30/2001'
ppl-# and l.station = 'KABC';
                                                                      QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------------------
 Nested Loop  (cost=0.00..58104.34 rows=2322 width=28) (actual time=0.480..5.357 rows=242 loops=1)
   ->  Index Scan using idx_logfile_station_air_date on logfile l  (cost=0.00..69.84 rows=17 width=9) (actual time=0.176..0.754 rows=32 loops=1)
         Index Cond: (((station)::text = 'KABC'::text) AND (air_date = '2001-01-30 00:00:00'::timestamp without time zone))
   ->  Index Scan using idx_program_mri_id_no_program on program p  (cost=0.00..3400.74 rows=870 width=40) (actual time=0.041..0.127 rows=8 loops=32)
         Index Cond: (("outer".program_id)::text = (music.fn_mri_id_no_program(p.mri_id_no))::text)
 Total runtime: 5.637 ms
(6 rows)


Tom Lane wrote:
Roger Ging <rging@paccomsys.com> writes:
  
Ran vacuum analyse on both program and logfile tables.  Estimates are 
more in line with reality now,
    

And they are what now?  You really can't expect to get useful help here
when you're being so miserly with the details ...

FWIW, I suspect you could force 7.4 to generate 7.3's plan by setting
enable_mergejoin to off (might have to also set enable_hashjoin to off,
if it then tries for a hash join).  7.3 could not even consider those
join types in this example, while 7.4 can.  The interesting question
from my perspective is why the planner is guessing wrong about the
relative costs of the plans.  EXPLAIN ANALYZE results with each type of
join forced would be useful to look at.

			regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

               http://www.postgresql.org/docs/faqs/FAQ.html


Tom Lane wrote:
Roger Ging <rging@paccomsys.com> writes:
  
Ran vacuum analyse on both program and logfile tables.  Estimates are 
more in line with reality now,
    

And they are what now?  You really can't expect to get useful help here
when you're being so miserly with the details ...

FWIW, I suspect you could force 7.4 to generate 7.3's plan by setting
enable_mergejoin to off (might have to also set enable_hashjoin to off,
if it then tries for a hash join).  7.3 could not even consider those
join types in this example, while 7.4 can.  The interesting question
from my perspective is why the planner is guessing wrong about the
relative costs of the plans.  EXPLAIN ANALYZE results with each type of
join forced would be useful to look at.

			regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

               http://www.postgresql.org/docs/faqs/FAQ.html
--------------080104020701070107090204-- From pgsql-performance-owner@postgresql.org Mon Dec 1 14:05:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 32057D1C968 for ; Mon, 1 Dec 2003 18:05:35 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49990-09 for ; Mon, 1 Dec 2003 14:05:04 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 1914FD1D9DC for ; Mon, 1 Dec 2003 14:05:03 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB1I5119023865; Mon, 1 Dec 2003 13:05:01 -0500 (EST) To: Roger Ging Cc: pgsql-performance@postgresql.org Subject: Re: Followup - expression (functional) index use in joins In-reply-to: <3FCB7ADA.8020101@paccomsys.com> References: <3FC4F360.2090609@paccomsys.com> <200311261912.01286.dev@archonet.com> <3FC51B2C.2030700@paccomsys.com> <18733.1069891765@sss.pgh.pa.us> <3FCB7ADA.8020101@paccomsys.com> Comments: In-reply-to Roger Ging message dated "Mon, 01 Dec 2003 09:31:06 -0800" Date: Mon, 01 Dec 2003 13:05:01 -0500 Message-ID: <23864.1070301901@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/14 X-Sequence-Number: 4874 Roger Ging writes: > See results below. Thanks for the report. It seems the issue is that the estimate for the number of matching rows is way off (870 vs 8): > -> Index Scan using idx_program_mri_id_no_program on program p > (cost=0.00..3400.74 rows=870 width=40) (actual time=0.041..0.127 rows=8 > loops=32) which discourages the planner from using a nestloop. I'm not sure we can do much about this in the short term. There's been some discussion of keeping statistics about the values of functional indexes, which would allow a better estimate to be made in this situation; but that won't happen before 7.5 at the earliest. > Turning enable_hashjoin off made the query run as it had on v7.3. We > have worked around this by changing the index from a function call to a > direct index on a new column with the results of the function maintained > by a trigger. Would there be performance issues from leaving > enable_hashjoin off, or do you recomend enabling it, and working around > function calls in indices? Turning enable_hashjoin off globally would be a *really bad* idea IMHO. The workaround with a derived column seems okay, though certainly a pain in the neck. Can you manage to turn off enable_hashjoin just for this one query? That might be the best short-term workaround. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Dec 1 15:48:51 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E65FCD1D59B for ; Mon, 1 Dec 2003 19:48:50 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69074-01 for ; Mon, 1 Dec 2003 15:48:21 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id C4AE1D1BC54 for ; Mon, 1 Dec 2003 15:48:19 -0400 (AST) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4006945; Mon, 01 Dec 2003 11:49:10 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Jeremiah Jahn , Hannu Krosing Subject: Re: cross table indexes or something? Date: Mon, 1 Dec 2003 11:47:51 -0800 User-Agent: KMail/1.4.3 Cc: postgres performance References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <1069885949.3176.57.camel@fuji.krosing.net> <1070288942.22346.27.camel@bluejay.goodinassociates.com> In-Reply-To: <1070288942.22346.27.camel@bluejay.goodinassociates.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200312011147.51359.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/15 X-Sequence-Number: 4875 Jeremiah, > I've attached the Analyze below. I have no idea why the db thinks there > is only 1 judge named simth. Is there some what I can inform the DB > about this. In actuality, there aren't any judges named smith at the > moment, but there are 22K people named smith. No, Hannu meant that you may need to run the following command: ANALYZE actor; ... to update the database statistics on the actors table. That is a maintainence task that needs to be run periodically. If that doesn't fix the bad plan, then the granularity of statistics on the full_name column needs updating; I suggest: ALTER TABLE actor ALTER COLUMN full_name SET STATISTICS 100; ANALYZE actor; And if it's still choosing a slow nested loop, up the stats to 250. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Mon Dec 1 17:44:31 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ED911D1D3A3 for ; Mon, 1 Dec 2003 21:44:27 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 94773-08 for ; Mon, 1 Dec 2003 17:43:58 -0400 (AST) Received: from live (64-35-136-226.gohighspeed.com [64.35.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id F1981D1C9F0 for ; Mon, 1 Dec 2003 17:43:46 -0400 (AST) Received: from [192.168.254.101] (helo=89glass.com) by live with esmtp (Exim 3.35 #1 (Debian)) id 1AQvpN-0007xh-00 for ; Mon, 01 Dec 2003 13:43:33 -0800 Message-ID: <3FCBB639.10409@89glass.com> Date: Mon, 01 Dec 2003 13:44:25 -0800 From: Jared Carr User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: A question on the query planner Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/18 X-Sequence-Number: 4878 I am currently working on optimizing some fairly time consuming queries on a decently large dataset. The Following is the query in question. SELECT z.lat, z.lon, z.city, z.state, q.date_time, c.make, c.model, c.year FROM quotes AS q, zips AS z, cars AS c WHERE z.zip = q.zip AND c.car_id = q.car_id AND z.state != 'AA' AND z.state != 'AE' AND z.state != 'AP' AND z.state = 'WA' ORDER BY date_time; The tables are as follows. Table "public.cars" Column | Type | Modifiers ---------------+-----------------------+---------------------------------------- car_id | character varying(10) | not null default ''::character varying nags_glass_id | character varying(7) | not null default ''::character varying make | character varying(30) | not null default ''::character varying model | character varying(30) | not null default ''::character varying year | character varying(4) | not null default ''::character varying style | character varying(30) | not null default ''::character varying price | double precision | not null default (0)::double precision Indexes: "cars_pkey" primary key, btree (car_id) "cars_car_id_btree_index" btree (car_id) "make_cars_index" btree (make) "model_cars_index" btree (model) "year_cars_index" btree ("year") Table "public.quotes" Column | Type | Modifiers -------------------+-----------------------------+--------------------------------------------------------------------- quote_id | bigint | not null default nextval('quotes_quote_id_seq'::text) visitor_id | bigint | not null default (0)::bigint date_time | timestamp without time zone | not null default '0001-01-01 00:00:00'::timestamp without time zone car_id | character varying(10) | not null default ''::character varying email | text | not null default ''::text zip | character varying(5) | not null default ''::character varying current_referrer | text | not null default ''::text original_referrer | text | not null default ''::text Indexes: "quotes_pkey" primary key, btree (quote_id) "car_id_quotes_index" btree (car_id) "visitor_id_quotes_index" btree (visitor_id) "zip_quotes_index" btree (zip) Table "public.zips" Column | Type | Modifiers --------+-----------------------+--------------------------------------------------- zip_id | bigint | not null default nextval('zips_zip_id_seq'::text) zip | character varying(5) | not null default ''::character varying city | character varying(28) | not null default ''::character varying state | character varying(2) | not null default ''::character varying lat | character varying(10) | not null default ''::character varying lon | character varying(10) | not null default ''::character varying Indexes: "zips_pkey" primary key, btree (zip_id) "zip_zips_index" btree (zip) "zips_state_btree_index" btree (state) The above query with the default setting of 10 for default_statistics_target runs as follows (From Explain Analyze) --------------------------------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=58064.16..58074.20 rows=4015 width=80) (actual time=2415.060..2421.421 rows=4539 loops=1) Sort Key: q.date_time -> Merge Join (cost=57728.02..57823.84 rows=4015 width=80) (actual time=2254.056..2345.013 rows=4539 loops=1) Merge Cond: ("outer"."?column7?" = "inner"."?column5?") -> Sort (cost=56880.61..56890.65 rows=4015 width=62) (actual time=2054.353..2062.189 rows=4693 loops=1) Sort Key: (q.car_id)::text -> Hash Join (cost=1403.91..56640.29 rows=4015 width=62) (actual time=8.479..1757.126 rows=10151 loops=1) Hash Cond: (("outer".zip)::text = ("inner".zip)::text) -> Seq Scan on quotes q (cost=0.00..10657.42 rows=336142 width=27) (actual time=0.062..657.015 rows=336166 loops=1) -> Hash (cost=1402.63..1402.63 rows=511 width=52) (actual time=8.273..8.273 rows=0 loops=1) -> Index Scan using zips_state_btree_index on zips z (cost=0.00..1402.63 rows=511 width=52) (actual time=0.215..6.877 rows=718 loops=1) Index Cond: ((state)::text = 'WA'::text) Filter: (((state)::text <> 'AA'::text) AND ((state)::text <> 'AE'::text) AND ((state)::text <> 'AP'::text)) -> Sort (cost=847.41..870.91 rows=9401 width=37) (actual time=199.172..216.354 rows=11922 loops=1) Sort Key: (c.car_id)::text -> Seq Scan on cars c (cost=0.00..227.01 rows=9401 width=37) (actual time=0.104..43.523 rows=9401 loops=1) Total runtime: 2427.937 ms If I set enable_seqscan=off I get the following QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=122108.52..122118.62 rows=4039 width=80) (actual time=701.002..707.442 rows=4541 loops=1) Sort Key: q.date_time -> Nested Loop (cost=0.00..121866.59 rows=4039 width=80) (actual time=0.648..624.134 rows=4541 loops=1) -> Nested Loop (cost=0.00..102256.36 rows=4039 width=62) (actual time=0.374..381.440 rows=10153 loops=1) -> Index Scan using zips_state_btree_index on zips z (cost=0.00..1413.31 rows=514 width=52) (actual time=0.042..9.043 rows=718 loops=1) Index Cond: ((state)::text = 'WA'::text) Filter: (((state)::text <> 'AA'::text) AND ((state)::text <> 'AE'::text) AND ((state)::text <> 'AP'::text)) -> Index Scan using zip_quotes_index on quotes q (cost=0.00..195.59 rows=48 width=27) (actual time=0.039..0.426 rows=14 loops=718) Index Cond: (("outer".zip)::text = (q.zip)::text) -> Index Scan using cars_car_id_btree_index on cars c (cost=0.00..4.84 rows=1 width=37) (actual time=0.015..0.017 rows=0 loops=10153) Index Cond: ((c.car_id)::text = ("outer".car_id)::text) Total runtime: 711.375 ms I can also get a similar plan if I disable both Hash Joins and Merge Joins. Furthermore I can get some additional speedup without turning off sequence scans if I set the value of default_statistics_target = 1000 then the runtime will be around 1200 otoh if I set default_statistics_target = 100 then the runtime will be around 12000. So, my question is is there any way to get the query planner to recognize the potential performance increase available by using the indexes that are set up without specifically turning off sequential scans before I run this query every time? Thanks for the help. Jared From pgsql-performance-owner@postgresql.org Mon Dec 1 18:25:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id AD5D0D1D916 for ; Mon, 1 Dec 2003 22:24:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03195-04 for ; Mon, 1 Dec 2003 18:24:25 -0400 (AST) Received: from jhuml1.jhmi.edu (jhuml1.jhmi.edu [162.129.234.20]) by svr1.postgresql.org (Postfix) with ESMTP id 0D31AD1D383 for ; Mon, 1 Dec 2003 18:24:22 -0400 (AST) Received: from jhuml1.jhmi.edu (jhuml1.jhmi.edu [162.129.234.20]) by jhuml1.jhmi.edu (PMDF V6.2-X17 #30839) with SMTP id <0HP800CBLM7D7X@jhuml1.jhmi.edu> for pgsql-performance@postgresql.org; Mon, 01 Dec 2003 17:24:23 -0500 (EST) Received: from jhuml1.jhmi.edu ([162.129.234.20]) by jhuml1.jhmi.edu (SAVSMTP 3.1.0.29) with SMTP id M2003120117242205892 for ; Mon, 01 Dec 2003 17:24:22 -0500 Received: from jhmimail.jhmi.edu (jhem2.jhmi.edu [162.129.8.23]) by jhuml1.jhmi.edu (PMDF V6.2-X17 #30839) with ESMTP id <0HP800CQRM8M7X@jhuml1.jhmi.edu> for pgsql-performance@postgresql.org; Mon, 01 Dec 2003 17:24:22 -0500 (EST) Received: from [162.129.178.60] by jhmimail.jhmi.edu (mshttpd); Mon, 01 Dec 2003 22:27:51 +0000 (GMT) Date: Mon, 01 Dec 2003 22:27:51 +0000 (GMT) From: LIANHE SHAO Subject: Is clustering possible to enhance the performance? To: pgsql-performance Message-id: <6b5ba36baf6d.6baf6d6b5ba3@jhmimail.jhmi.edu> MIME-version: 1.0 X-Mailer: iPlanet Messenger Express 5.2 HotFix 1.17 (built Jun 23 2003) Content-type: text/plain; charset=us-ascii Content-language: en Content-transfer-encoding: 7bit Content-disposition: inline X-Accept-Language: en X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/19 X-Sequence-Number: 4879 Hello, I am wondering if it is possible to use several machine as cluster to boost the slow queries. Is that possible? Anybody have tried that before? Initially, I was thinking to use dual CPUS instead of one. but it is not correct because pgsql is not multi-threaded. Any suggestions are welcome and appreciated, Regards, William From pgsql-performance-owner@postgresql.org Mon Dec 1 18:38:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9A84FD1D92C for ; Mon, 1 Dec 2003 22:37:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03971-10 for ; Mon, 1 Dec 2003 18:37:32 -0400 (AST) Received: from candle.pha.pa.us (unknown [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 99B49D1DA62 for ; Mon, 1 Dec 2003 18:37:28 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hB1MbVF23686; Mon, 1 Dec 2003 17:37:31 -0500 (EST) From: Bruce Momjian Message-Id: <200312012237.hB1MbVF23686@candle.pha.pa.us> Subject: Re: Is clustering possible to enhance the performance? In-Reply-To: <6b5ba36baf6d.6baf6d6b5ba3@jhmimail.jhmi.edu> To: LIANHE SHAO Date: Mon, 1 Dec 2003 17:37:31 -0500 (EST) Cc: pgsql-performance X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/20 X-Sequence-Number: 4880 LIANHE SHAO wrote: > Hello, > I am wondering if it is possible to use several > machine as cluster to boost the slow queries. Is > that possible? Anybody have tried that before? > > Initially, I was thinking to use dual CPUS instead > of one. but it is not correct because pgsql is not > multi-threaded. Dual cpu's allow multiple backends to use different cpu's, but a single session can't use more than one cpu. -- 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 Mon Dec 1 20:46:01 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 70173D1B4B1 for ; Tue, 2 Dec 2003 00:45:57 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20740-10 for ; Mon, 1 Dec 2003 20:45:31 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 67E86D1B44D for ; Mon, 1 Dec 2003 20:45:28 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB20jT19008533; Mon, 1 Dec 2003 19:45:30 -0500 (EST) To: Jared Carr Cc: pgsql-performance@postgresql.org Subject: Re: A question on the query planner In-reply-to: <3FCBB639.10409@89glass.com> References: <3FCBB639.10409@89glass.com> Comments: In-reply-to Jared Carr message dated "Mon, 01 Dec 2003 13:44:25 -0800" Date: Mon, 01 Dec 2003 19:45:29 -0500 Message-ID: <8532.1070325929@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/21 X-Sequence-Number: 4881 Jared Carr writes: > I am currently working on optimizing some fairly time consuming queries > on a decently large dataset. It doesn't look that large from here ;-). I'd suggest experimenting with reducing random_page_cost, since at least for your test query it sure looks like everything is in RAM. In theory random_page_cost = 1.0 is the correct setting for all-in-RAM cases. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Dec 2 11:55:36 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 99E24D1D23B for ; Tue, 2 Dec 2003 15:53:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 42575-03 for ; Tue, 2 Dec 2003 11:53:31 -0400 (AST) Received: from mist.cti.unav.es (mist.cti.unav.es [159.237.12.28]) by svr1.postgresql.org (Postfix) with ESMTP id ECE61D1D30D for ; Tue, 2 Dec 2003 11:53:09 -0400 (AST) Received: from norvelle.net ([159.237.106.162]) by mist.cti.unav.es (8.12.10/8.12.10) with ESMTP id hB2FrCnF006389 for ; Tue, 2 Dec 2003 16:53:12 +0100 Date: Tue, 2 Dec 2003 16:53:16 +0100 Mime-Version: 1.0 (Apple Message framework v552) Content-Type: multipart/alternative; boundary=Apple-Mail-22--268600658 Subject: Update performance ... is 200, 000 updates per hour what I should expect? From: Erik Norvelle To: pgsql-performance@postgresql.org Message-Id: X-Mailer: Apple Mail (2.552) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/22 X-Sequence-Number: 4882 --Apple-Mail-22--268600658 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; delsp=yes; charset=ISO-8859-1; format=flowed Folks: I=B4m running a query which is designed to generate a foreign key for a=20= =20 table of approx. 10 million records (I've mentioned this in an earlier=20= =20 posting). The table is called "indethom", and each row contains a=20=20 single word from the works of St. Thomas Aquinas, along with=20=20 grammatical data about the word form, and (most importantly for my=20=20 current problem) a set of columns identifying the particular=20=20 work/section/paragraph that the word appears in. This database is completely non-normalized, and I'm working on=20=20 performing some basic normalization, beginning with creating a table=20=20 called "s2.sectiones" which (naturally) contains a complete listing of=20= =20 all of the sections of all the works of St. Thomas. I will then=20=20 eliminate this information from the original "indethom" table,=20=20 replacing it with the foreign key I am currently generating. ** My question has to do with whether or not I am getting maximal speed=20= =20 out of PostgreSQL, or whether I need to perform further optimizations.=20= =20=20 I am currently getting about 200,000 updates per hour, and updating the=20= =20 entire 10 million rows thus requires 50 hours, which seems a bit much. Here's the query I am running: update indethom set query_counter =3D nextval('s2.query_counter_seq'), -- Just= =20=20 for keeping track of how fast the query is running sectref =3D (select clavis from s2.sectiones where s2.sectiones.nomeoper =3D indethom.nomeoper and s2.sectiones.refere1a =3D indethom.refere1a and=20=20 s2.sectiones.refere1b =3D indethom.refere1b and s2.sectiones.refere2a =3D indethom.refere2a and=20=20 s2.sectiones.refere2b =3D indethom.refere2b and s2.sectiones.refere3a =3D indethom.refere3a and=20=20 s2.sectiones.refere3b =3D indethom.refere3b and s2.sectiones.refere4a =3D indethom.refere4a and=20=20 s2.sectiones.refere4b =3D indethom.refere4b); Here=B4s the query plan: QUERY PLAN ------------------------------------------------------------------------=20 ------------- Seq Scan on indethom (cost=3D0.00..1310352.72 rows=3D10631972 width=3D21= 2) SubPlan -> Index Scan using sectiones_ndx on sectiones (cost=3D0.00..6.03= =20=20 rows=3D1 width=3D4) Index Cond: ((nomeoper =3D $0) AND (refere1a =3D $1) AND=20=20 (refere1b =3D $2) AND (refere2a =3D $3) AND (refere2b =3D $4) AND (refere3a= =3D=20=20 $5) AND (refere3b =3D $6) AND (refere4a =3D $7) AND (refere4b =3D $8)) (4 rows) Note: I have just performed a VACUUM ANALYZE on the indethom table, as=20= =20 suggested by this listserve. Here's the structure of the s2.sectiones table: it=3D> \d s2.sectiones Table "s2.sectiones" Column | Type | Modifiers ----------+--------------+----------- nomeoper | character(3) | refere1a | character(2) | refere1b | character(2) | refere2a | character(2) | refere2b | character(2) | refere3a | character(2) | refere3b | character(2) | refere4a | character(2) | refere4b | character(2) | clavis | integer | Indexes: sectiones_ndx btree (nomeoper, refere1a, refere1b, refere2a,=20=20 refere2b, refere3a, refere3b, refere4a, refere4b) Finally, here is the structure of indethom (some non-relevant columns=20=20 not shown): it=3D> \d indethom Table "public.indethom" Column | Type | Modifiers ---------------+-----------------------+----------- numeoper | smallint | not null nomeoper | character(3) | not null editcrit | character(1) | refere1a | character(2) | refere1b | character(2) | refere2a | character(2) | refere2b | character(2) | refere3a | character(2) | refere3b | character(2) | refere4a | character(2) | refere4b | character(2) | refere5a | character(2) | not null refere5b | smallint | not null referen6 | smallint | not null ... several columns skipped ... verbum | character varying(22) | not null ... other columns skipped ... poslinop | integer | not null posverli | smallint | not null posverop | integer | not null clavis | integer | not null articref | integer | sectref | integer | query_counter | integer | Indexes: indethom_pkey primary key btree (clavis), indethom_articulus_ndx btree (nomeoper, refere1a, refere1b,=20=20 refere2a, refere2b, refere3a, refere3b), indethom_sectio_ndx btree (nomeoper, refere1a, refere1b,=20=20 refere2a, refere2b, refere3a, refere3b, refere4a, refere4b), verbum_ndx btree (verbum) Thanks for your assistance! -Erik Norvelle --Apple-Mail-22--268600658 Content-Transfer-Encoding: quoted-printable Content-Type: text/enriched; charset=ISO-8859-1 Courier NewFolks: I=B4m running a query which is designed to generate a foreign key for a table of approx. 10 million records (I've mentioned this in an earlier posting). The table is called "indethom", and each row contains a single word from the works of St. Thomas Aquinas, along with grammatical data about the word form, and (most importantly for my current problem) a set of columns identifying the particular work/section/paragraph that the word appears in. This database is completely non-normalized, and I'm working on performing some basic normalization, beginning with creating a table called "s2.sectiones" which (naturally) contains a complete listing of all of the sections of all the works of St. Thomas. I will then eliminate this information from the original "indethom" table, replacing it with the foreign key I am currently generating. ** My question has to do with whether or not I am getting maximal speed out of PostgreSQL, or whether I need to perform further optimizations. I am currently getting about 200,000 updates per hour, and updating the entire 10 million rows thus requires 50 hours, which seems a bit much. Here's the query I am running:=20 update indethom=20 set query_counter =3D nextval('s2.query_counter_seq'), -- Just for keeping track of how fast the query is running sectref =3D (select clavis from s2.sectiones where=20 s2.sectiones.nomeoper =3D indethom.nomeoper=20 and s2.sectiones.refere1a =3D indethom.refere1a and s2.sectiones.refere1b =3D indethom.refere1b=20 and s2.sectiones.refere2a =3D indethom.refere2a and s2.sectiones.refere2b =3D indethom.refere2b=20 and s2.sectiones.refere3a =3D indethom.refere3a and s2.sectiones.refere3b =3D indethom.refere3b=20 and s2.sectiones.refere4a =3D indethom.refere4a and s2.sectiones.refere4b =3D indethom.refere4b); Here=B4s the query plan: QUERY PLAN=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20=20=20=20=20=20=20 ---------------------------------------------------------------------------= ---------- Seq Scan on indethom (cost=3D0.00..1310352.72 rows=3D10631972 width=3D212) SubPlan -> Index Scan using sectiones_ndx on sectiones (cost=3D0.00..6.03 rows=3D1 width=3D4) Index Cond: ((nomeoper =3D $0) AND (refere1a =3D $1) AND (refere1b =3D $2) AND (refere2a =3D $3) AND (refere2b =3D $4) AND (refere3a =3D $5) AND (refere3b =3D $6) AND (refere4a =3D $7) AND (refere4b =3D $8)) (4 rows) Note: I have just performed a VACUUM ANALYZE on the indethom table, as suggested by this listserve. Here's the structure of the s2.sectiones table: it=3D> \d s2.sectiones Table "s2.sectiones" Column | Type | Modifiers=20 ----------+--------------+----------- nomeoper | character(3) |=20 refere1a | character(2) |=20 refere1b | character(2) |=20 refere2a | character(2) |=20 refere2b | character(2) |=20 refere3a | character(2) |=20 refere3b | character(2) |=20 refere4a | character(2) |=20 refere4b | character(2) |=20 clavis | integer |=20 Indexes: sectiones_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b, refere4a, refere4b) Finally, here is the structure of indethom (some non-relevant columns not shown): it=3D> \d indethom Table "public.indethom" Column | Type | Modifiers=20 ---------------+-----------------------+----------- numeoper | smallint | not null nomeoper | character(3) | not null editcrit | character(1) |=20 refere1a | character(2) |=20 refere1b | character(2) |=20 refere2a | character(2) |=20 refere2b | character(2) |=20 refere3a | character(2) |=20 refere3b | character(2) |=20 refere4a | character(2) |=20 refere4b | character(2) |=20 refere5a | character(2) | not null refere5b | smallint | not null referen6 | smallint | not null ... several columns skipped ... verbum | character varying(22) | not null ... other columns skipped ... poslinop | integer | not null posverli | smallint | not null posverop | integer | not null clavis | integer | not null articref | integer |=20 sectref | integer |=20 query_counter | integer |=20 Indexes: indethom_pkey primary key btree (clavis), indethom_articulus_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b), indethom_sectio_ndx btree (nomeoper, refere1a, refere1b, refere2a, refere2b, refere3a, refere3b, refere4a, refere4b), verbum_ndx btree (verbum) Thanks for your assistance! -Erik Norvelle --Apple-Mail-22--268600658-- From pgsql-performance-owner@postgresql.org Tue Dec 2 12:30:39 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 190C9D1D93E for ; Tue, 2 Dec 2003 16:29:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50317-01 for ; Tue, 2 Dec 2003 12:29:14 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 31C85D1D92C for ; Tue, 2 Dec 2003 12:29:10 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 4BB53354FA; Tue, 2 Dec 2003 08:29:15 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 4A16A354F8; Tue, 2 Dec 2003 08:29:15 -0800 (PST) Date: Tue, 2 Dec 2003 08:29:15 -0800 (PST) From: Stephan Szabo To: Erik Norvelle Cc: pgsql-performance@postgresql.org Subject: Re: Update performance ... is 200,000 updates per hour In-Reply-To: Message-ID: <20031202082100.T87630@megazone.bigpanda.com> References: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=X-UNKNOWN Content-Transfer-Encoding: QUOTED-PRINTABLE X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/23 X-Sequence-Number: 4883 On Tue, 2 Dec 2003, Erik Norvelle wrote: > ** My question has to do with whether or not I am getting maximal speed > out of PostgreSQL, or whether I need to perform further optimizations. > I am currently getting about 200,000 updates per hour, and updating the > entire 10 million rows thus requires 50 hours, which seems a bit much. Well, it doesn't entirely surprise me much given the presumably 10 million iterations of the index scan that it's doing. Explain analyze output (even over a subset of the indethom table by adding a where clause) would probably help to get better info. I'd suggest seeing if something like: update indethom set query_counter=3D...,sectref=3Ds.clavis FROM s2.sectiones s where s2.sectiones.nomeoper =3D indethom.nomeoper and ...; tries a join that might give a better plan. > Here's the query I am running: > update indethom > set query_counter =3D nextval('s2.query_counter_seq'), -- Just > for keeping track of how fast the query is running > sectref =3D (select clavis from s2.sectiones where > s2.sectiones.nomeoper =3D indethom.nomeoper > and s2.sectiones.refere1a =3D indethom.refere1a and > s2.sectiones.refere1b =3D indethom.refere1b > and s2.sectiones.refere2a =3D indethom.refere2a and > s2.sectiones.refere2b =3D indethom.refere2b > and s2.sectiones.refere3a =3D indethom.refere3a and > s2.sectiones.refere3b =3D indethom.refere3b > and s2.sectiones.refere4a =3D indethom.refere4a and > s2.sectiones.refere4b =3D indethom.refere4b); > > Here=B4s the query plan: > QUERY PLAN > ------------------------------------------------------------------------ > ------------- > Seq Scan on indethom (cost=3D0.00..1310352.72 rows=3D10631972 width=3D= 212) > SubPlan > -> Index Scan using sectiones_ndx on sectiones (cost=3D0.00..6.03 > rows=3D1 width=3D4) > Index Cond: ((nomeoper =3D $0) AND (refere1a =3D $1) AND > (refere1b =3D $2) AND (refere2a =3D $3) AND (refere2b =3D $4) AND (refere= 3a =3D > $5) AND (refere3b =3D $6) AND (refere4a =3D $7) AND (refere4b =3D $8)) > (4 rows) From pgsql-performance-owner@postgresql.org Tue Dec 2 12:34:28 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3C47BD1D92D for ; Tue, 2 Dec 2003 16:33:06 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48233-08 for ; Tue, 2 Dec 2003 12:32:38 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id F3696D1D6B9 for ; Tue, 2 Dec 2003 12:32:33 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB2GWV19013346; Tue, 2 Dec 2003 11:32:31 -0500 (EST) To: Erik Norvelle Cc: pgsql-performance@postgresql.org Subject: Re: Update performance ... is 200, 000 updates per hour what I should expect? In-reply-to: References: Comments: In-reply-to Erik Norvelle message dated "Tue, 02 Dec 2003 16:53:16 +0100" Date: Tue, 02 Dec 2003 11:32:31 -0500 Message-ID: <13345.1070382751@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/24 X-Sequence-Number: 4884 Erik Norvelle writes: > update indethom > set query_counter =3D nextval('s2.query_counter_seq'), -- Just= > =20=20 > for keeping track of how fast the query is running > sectref =3D (select clavis from s2.sectiones where > s2.sectiones.nomeoper =3D indethom.nomeoper > and s2.sectiones.refere1a =3D indethom.refere1a and=20=20 > s2.sectiones.refere1b =3D indethom.refere1b > and s2.sectiones.refere2a =3D indethom.refere2a and=20=20 > s2.sectiones.refere2b =3D indethom.refere2b > and s2.sectiones.refere3a =3D indethom.refere3a and=20=20 > s2.sectiones.refere3b =3D indethom.refere3b > and s2.sectiones.refere4a =3D indethom.refere4a and=20=20 > s2.sectiones.refere4b =3D indethom.refere4b); This is effectively forcing a nestloop-with-inner-indexscan join. You might be better off with update indethom set query_counter = nextval('s2.query_counter_seq'), sectref = sectiones.clavis from s2.sectiones where s2.sectiones.nomeoper = indethom.nomeoper and s2.sectiones.refere1a = indethom.refere1a and s2.sectiones.refere1b = indethom.refere1b and s2.sectiones.refere2a = indethom.refere2a and s2.sectiones.refere2b = indethom.refere2b and s2.sectiones.refere3a = indethom.refere3a and s2.sectiones.refere3b = indethom.refere3b and s2.sectiones.refere4a = indethom.refere4a and s2.sectiones.refere4b = indethom.refere4b; regards, tom lane From pgsql-performance-owner@postgresql.org Tue Dec 2 12:41:52 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 95624D1D722 for ; Tue, 2 Dec 2003 16:41:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49804-08 for ; Tue, 2 Dec 2003 12:40:53 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id E4B61D1DAF2 for ; Tue, 2 Dec 2003 12:40:47 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 1D99E36CF3; Tue, 2 Dec 2003 11:40:52 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1ARDa0-00046L-00; Tue, 02 Dec 2003 11:40:52 -0500 To: Erik Norvelle Cc: pgsql-performance@postgresql.org Subject: Re: Update performance ... is 200, 000 updates per hour what I should expect? References: In-Reply-To: From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 02 Dec 2003 11:40:51 -0500 Message-ID: <87u14j9nqk.fsf@stark.dyndns.tv> Lines: 54 User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3 MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/25 X-Sequence-Number: 4885 Erik Norvelle writes: > Here's the query I am running: > update indethom > set query_counter = nextval('s2.query_counter_seq'), -- Just for keeping track of how fast the query is running > sectref = (select clavis from s2.sectiones where > s2.sectiones.nomeoper = indethom.nomeoper > and s2.sectiones.refere1a = indethom.refere1a and s2.sectiones.refere1b = indethom.refere1b > and s2.sectiones.refere2a = indethom.refere2a and s2.sectiones.refere2b = indethom.refere2b > and s2.sectiones.refere3a = indethom.refere3a and s2.sectiones.refere3b = indethom.refere3b > and s2.sectiones.refere4a = indethom.refere4a and s2.sectiones.refere4b = indethom.refere4b); > > Here�s the query plan: > QUERY PLAN > ------------------------------------------------------------------------------------- > Seq Scan on indethom (cost=0.00..1310352.72 rows=10631972 width=212) > SubPlan > -> Index Scan using sectiones_ndx on sectiones (cost=0.00..6.03 rows=1 width=4) > Index Cond: ((nomeoper = $0) AND (refere1a = $1) AND (refere1b = $2) AND (refere2a = $3) AND (refere2b = $4) AND (refere3a = $5) AND (refere3b = $6) AND (refere4a = $7) AND (refere4b = $8)) > (4 rows) Firstly, you might try running "vacuum full" on both tables. If there are tons of extra dead records that are left-over they could be slowing down the update. This isn't the fastest possible plan but it's pretty good. You might be able to get it somewhat faster using the non-standard "from" clause on the update statement. update indethom set sectref = clavis from sectiones where sectiones.nomeoper = indethom.nomeoper and sectiones.refere1a = indethom.refere1a and sectiones.refere1b = indethom.refere1b and sectiones.refere2a = indethom.refere2a and sectiones.refere2b = indethom.refere2b and sectiones.refere3a = indethom.refere3a and sectiones.refere3b = indethom.refere3b and sectiones.refere4a = indethom.refere4a and sectiones.refere4b = indethom.refere4b This might be able to use a merge join which will take longer to get started because it has to sort both tables, but might finish faster. You might also try just paring the index down to just the two or three most useful columns. Is it common that something matches refere1a and refere1b but doesn't match the remaining? A 8-column index is a lot of overhead. I'm not sure how much that effects lookup times but it might be substantial. -- greg From pgsql-performance-owner@postgresql.org Tue Dec 2 13:18:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C1B3AD1D8BB for ; Tue, 2 Dec 2003 17:17:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 54908-09 for ; Tue, 2 Dec 2003 13:16:54 -0400 (AST) Received: from bramble.mmrd.com (unknown [65.217.53.66]) by svr1.postgresql.org (Postfix) with ESMTP id 871EED1C96D for ; Tue, 2 Dec 2003 13:16:49 -0400 (AST) Received: from thorn.mmrd.com (thorn.mmrd.com [172.25.10.100]) by bramble.mmrd.com (8.12.8/8.12.8) with ESMTP id hB2GPqcM003471; Tue, 2 Dec 2003 11:25:53 -0500 Received: from gnvex001.mmrd.com (gnvex001.mmrd.com [192.168.3.55]) by thorn.mmrd.com (8.11.6/8.11.6) with ESMTP id hB2HGLl13424; Tue, 2 Dec 2003 12:16:22 -0500 Received: from camel.mmrd.com ([172.25.5.213]) by gnvex001.mmrd.com with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2657.72) id XT87Z43S; Tue, 2 Dec 2003 12:16:19 -0500 Subject: Re: A question on the query planner From: Robert Treat To: Jared Carr Cc: pgsql-performance@postgresql.org In-Reply-To: <3FCBB639.10409@89glass.com> References: <3FCBB639.10409@89glass.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 Date: 02 Dec 2003 12:16:21 -0500 Message-Id: <1070385381.24915.8650.camel@camel> Mime-Version: 1.0 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/26 X-Sequence-Number: 4886 On Mon, 2003-12-01 at 16:44, Jared Carr wrote: > I am currently working on optimizing some fairly time consuming queries > on a decently large > dataset. > > The Following is the query in question. > > SELECT z.lat, z.lon, z.city, z.state, q.date_time, c.make, c.model, c.year > FROM quotes AS q, zips AS z, cars AS c > WHERE > z.zip = q.zip AND > c.car_id = q.car_id AND > z.state != 'AA' AND > z.state != 'AE' AND > z.state != 'AP' AND > z.state = 'WA' > ORDER BY date_time; > This wont completely solve your problem, but z.state = 'WA' would seem to be mutually exclusive of the != AA|AE|AP. While it's not much, it is extra overhead there doesn't seem to be any need for... Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Tue Dec 2 13:55:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 019A4D1D921 for ; Tue, 2 Dec 2003 17:55:36 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 64484-04 for ; Tue, 2 Dec 2003 13:55:10 -0400 (AST) Received: from live (64-35-136-226.gohighspeed.com [64.35.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id 77905D1D59B for ; Tue, 2 Dec 2003 13:55:04 -0400 (AST) Received: from [192.168.254.101] (helo=89glass.com) by live with esmtp (Exim 3.35 #1 (Debian)) id 1AREjd-00041X-00; Tue, 02 Dec 2003 09:54:53 -0800 Message-ID: <3FCCD225.8030609@89glass.com> Date: Tue, 02 Dec 2003 09:55:49 -0800 From: Jared Carr User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Robert Treat Cc: pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> In-Reply-To: <1070385381.24915.8650.camel@camel> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/27 X-Sequence-Number: 4887 Robert Treat wrote: >On Mon, 2003-12-01 at 16:44, Jared Carr wrote: > > >>I am currently working on optimizing some fairly time consuming queries >>on a decently large >>dataset. >> >>The Following is the query in question. >> >>SELECT z.lat, z.lon, z.city, z.state, q.date_time, c.make, c.model, c.year >> FROM quotes AS q, zips AS z, cars AS c >> WHERE >> z.zip = q.zip AND >> c.car_id = q.car_id AND >> z.state != 'AA' AND >> z.state != 'AE' AND >> z.state != 'AP' AND >> z.state = 'WA' >> ORDER BY date_time; >> >> >> > >This wont completely solve your problem, but z.state = 'WA' would seem >to be mutually exclusive of the != AA|AE|AP. While it's not much, it is >extra overhead there doesn't seem to be any need for... > >Robert Treat > > That is an excellent point, unfortunately it doesn't change the query plan at all. Furthermore noticed that in the following query plan it is doing the sequential scan on quotes first, and then doing the sequential on zips. IMHO this should be the other way around, since the result set for zips is considerably smaller especially give that we are using a where clause to limit the number of items returned from zips, so it would seem that it would be faster to scan zips then join onto quotes, but perhaps it needs to do the sequential scan on both regardless. Of course still there is the holy grail of getting it to actually use the indexes. :P QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=57812.71..57822.86 rows=4058 width=80) (actual time=2522.826..2529.237 rows=4581 loops=1) Sort Key: q.date_time -> Merge Join (cost=57473.20..57569.50 rows=4058 width=80) (actual time=2360.656..2451.987 rows=4581 loops=1) Merge Cond: ("outer"."?column7?" = "inner"."?column5?") -> Sort (cost=56625.79..56635.93 rows=4058 width=62) (actual time=2077.209..2085.095 rows=4735 loops=1) Sort Key: (q.car_id)::text -> Hash Join (cost=1088.19..56382.58 rows=4058 width=62) (actual time=86.111..1834.682 rows=10193 loops=1) Hash Cond: (("outer".zip)::text = ("inner".zip)::text) -> Seq Scan on quotes q (cost=0.00..10664.25 rows=336525 width=27) (actual time=0.098..658.905 rows=336963 loops=1) -> Hash (cost=1086.90..1086.90 rows=516 width=52) (actual time=85.798..85.798 rows=0 loops=1) -> Seq Scan on zips z (cost=0.00..1086.90 rows=516 width=52) (actual time=79.532..84.151 rows=718 loops=1) Filter: ((state)::text = 'WA'::text) -> Sort (cost=847.41..870.91 rows=9401 width=37) (actual time=282.896..300.082 rows=11950 loops=1) Sort Key: (c.car_id)::text -> Seq Scan on cars c (cost=0.00..227.01 rows=9401 width=37) (actual time=0.102..43.516 rows=9401 loops=1) From pgsql-admin-owner@postgresql.org Tue Dec 2 14:12:25 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 14559D1D59B; Tue, 2 Dec 2003 18:12:18 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66932-04; Tue, 2 Dec 2003 14:11:51 -0400 (AST) Received: from THOR.goeci.com (thor.goeci.com [66.28.220.99]) by svr1.postgresql.org (Postfix) with ESMTP id 60207D1D930; Tue, 2 Dec 2003 14:11:45 -0400 (AST) Received: by THOR.goeci.com with Internet Mail Service (5.5.2653.19) id ; Tue, 2 Dec 2003 13:11:46 -0500 Message-ID: <2D92FEBFD3BE1346A6C397223A8DD3FC0924C9@THOR.goeci.com> From: Murthy Kambhampaty To: "'CLeon@phs.org'" Cc: pgsql-performance@postgresql.org, pgsql-admin@postgresql.org, "'linux-xfs@oss.sgi.com'" Subject: Re: [linux-lvm] RE: [PERFORM] backup/restore - another Date: Tue, 2 Dec 2003 13:11:45 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/12 X-Sequence-Number: 11520 xfs_freeze is a userspace program included in the xfsprogs rpm. It does run on Redhat 7.3 (the SGI supplied kernels and userspace for RedHat 7.3 are somewhat dated; I'd suggest patching the 2.4.21 kernel with XFS 1.3.1 patches and upgrading the userspace programs from the SRPMS). Post to the linux-xfs mailing list if you need further guidance (lots of people seem to still run XFS on Redhat 7.3). > -----Original Message----- > From: CLeon@phs.org [mailto:CLeon@phs.org] > Sent: Thursday, October 30, 2003 12:28 PM > To: linux-lvm@sistina.com; tgl@sss.pgh.pa.us > Cc: threshar@torgo.978.org; josh@agliodbs.com; markw@osdl.org; > pgsql-performance@postgresql.org; pgsql-admin@postgresql.org > Subject: Re: [linux-lvm] RE: [ADMIN] [PERFORM] backup/restore > - another > > > Does xfs_freeze work on red hat 7.3? > > Cynthia Leon > > -----Original Message----- > From: Murthy Kambhampaty [mailto:murthy.kambhampaty@goeci.com] > Sent: Friday, October 17, 2003 11:34 AM > To: 'Tom Lane'; Murthy Kambhampaty > Cc: 'Jeff'; Josh Berkus; markw@osdl.org; > pgsql-performance@postgresql.org; linux-lvm@sistina.com; > pgsql-admin@postgresql.org > Subject: [linux-lvm] RE: [ADMIN] [PERFORM] backup/restore - another > area. > > > Friday, October 17, 2003 12:05, Tom Lane > [mailto:tgl@sss.pgh.pa.us] wrote: > > >Murthy Kambhampaty writes: > >> ... The script handles situations > >> where (i) the XFS filesystem containing $PGDATA has an > >external log and (ii) > >> the postmaster log ($PGDATA/pg_xlog) is written to a > >filesystem different > >> than the one containing the $PGDATA folder. > > > >It does? How exactly can you ensure snapshot consistency between > >data files and XLOG if they are on different filesystem > > Say, you're setup looks something like this: > > mount -t xfs /dev/VG1/LV_data /home/pgdata > mount -t xfs /dev/VG1/LV_xlog /home/pgdata/pg_xlog > > When you want to take the filesystem backup, you do: > > Step 1: > xfs_freeze -f /dev/VG1/LV_xlog > xfs_freeze -f /dev/VG1/LV_data > This should finish any checkpoints that were in > progress, and not > start any new ones > till you unfreeze. (writes to an xfs_frozen filesystem > wait for the > xfs_freeze -u, > but reads proceed; see text from xfs_freeze manpage in postcript > below.) > > > Step2: > create snapshots of /dev/VG1/LV_xlog and /dev/VG1/LV_xlog > > Step 3: > xfs_freeze -u /dev/VG1/LV_data > xfs_freeze -u /dev/VG1/LV_xlog > Unfreezing in this order should assure that checkpoints > resume where > they left off, then log writes commence. > > > Step4: > mount the snapshots taken in Step2 somewhere; e.g. /mnt/snap_data and > /mnt/snap_xlog. Copy (or rsync or whatever) /mnt/snap_data to > /mnt/pgbackup/ > and /mnt/snap_xlog to /mnt/pgbackup/pg_xlog. Upon completion, > /mnt/pgbackup/ > looks to the postmaster like /home/pgdata would if the server > had crashed at > the moment that Step1 was initiated. As I understand it, > during recovery > (startup) the postmaster will roll the database forward to this point, > "checkpoint-ing" all the transactions that made it into the > log before the > crash. > > Step5: > remove the snapshots created in Step2. > > The key is > (i) xfs_freeze allows you to "quiesce" any filesystem at any > point in time > and, if I'm not mistaken, the order (LIFO) in which you > freeze and unfreeze > the two filesystems: freeze $PGDATA/pg_xlog then $PGDATA; > unfreeze $PGDATA > then $PGDATA/pg_xlog. > (ii) WAL recovery assures consistency after a (file)sytem crash. > > Presently, the test server for my backup scripts is set-up > this way, and the > backup works flawlessly, AFAICT. (Note that the backup script starts a > postmaster on the filesystem copy each time, so you get early > warning of > problems. Moreover the data in the "production" and "backup" > copies are > tested and found to be identical. > > Comments? Any suggestions for additional tests? > > Thanks, > Murthy > > PS: From the xfs_freeze manpage: > "xfs_freeze suspends and resumes access to an XFS filesystem (see > xfs(5)). > > xfs_freeze halts new access to the filesystem and creates a > stable image > on disk. xfs_freeze is intended to be used with volume managers and > hardware RAID devices that support the creation of snapshots. > > The mount-point argument is the pathname of the directory where the > filesystem is mounted. The filesystem must be mounted to be > frozen (see > mount(8)). > > The -f flag requests the specified XFS filesystem to be > frozen from new > modifications. When this is selected, all ongoing transactions in the > filesystem are allowed to complete, new write system calls are halted, > other calls which modify the filesystem are halted, and all > dirty data, > metadata, and log information are written to disk. Any process > attempting to write to the frozen filesystem will block > waiting for the > filesystem to be unfrozen. > > Note that even after freezing, the on-disk filesystem can contain > information on files that are still in the process of unlinking. These > files will not be unlinked until the filesystem is unfrozen or a clean > mount of the snapshot is complete. > > The -u option is used to un-freeze the filesystem and allow operations > to continue. Any filesystem modifications that were blocked by the > freeze are unblocked and allowed to complete." > > _______________________________________________ > linux-lvm mailing list > linux-lvm@sistina.com > http://lists.sistina.com/mailman/listinfo/linux-lvm > read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/ > > ============================================================== > ================ > --- PRESBYTERIAN HEALTHCARE SERVICES DISCLAIMER --- > > This message originates from Presbyterian Healthcare Services > or one of its > affiliated organizations. It contains information, which may > be confidential > or privileged, and is intended only for the individual or > entity named above. > It is prohibited for anyone else to disclose, copy, > distribute or use the > contents of this message. All personal messages express views > solely of the > sender, which are not to be attributed to Presbyterian > Healthcare Services or > any of its affiliated organizations, and may not be > distributed without this > disclaimer. If you received this message in error, please notify us > immediately at postmaster@phs.org. > ============================================================== > ================ > > > ---------------------------(end of > broadcast)--------------------------- > TIP 9: 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 Dec 2 15:01:39 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 492EDD1BC5C for ; Tue, 2 Dec 2003 19:01:15 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77317-06 for ; Tue, 2 Dec 2003 15:00:49 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 27689D1B445 for ; Tue, 2 Dec 2003 15:00:44 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 172AD36E82; Tue, 2 Dec 2003 13:59:57 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1ARFka-0004fq-00; Tue, 02 Dec 2003 13:59:56 -0500 To: Jared Carr Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> In-Reply-To: <3FCCD225.8030609@89glass.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 02 Dec 2003 13:59:56 -0500 Message-ID: <87d6b79har.fsf@stark.dyndns.tv> Lines: 34 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 postgresql.org X-Archive-Number: 200312/28 X-Sequence-Number: 4888 Jared Carr writes: > Furthermore noticed that in the following query plan it is doing the > sequential scan on quotes first, and then doing the sequential on zips. IMHO > this should be the other way around, since the result set for zips is > considerably smaller especially give that we are using a where clause to > limit the number of items returned from zips, so it would seem that it would > be faster to scan zips then join onto quotes, but perhaps it needs to do the > sequential scan on both regardless. >-> Hash Join (cost=1088.19..56382.58 rows=4058 width=62) (actual time=86.111..1834.682 rows=10193 loops=1) > Hash Cond: (("outer".zip)::text = ("inner".zip)::text) > -> Seq Scan on quotes q (cost=0.00..10664.25 rows=336525 width=27) (actual time=0.098..658.905 rows=336963 loops=1) > -> Hash (cost=1086.90..1086.90 rows=516 width=52) (actual time=85.798..85.798 rows=0 loops=1) > -> Seq Scan on zips z (cost=0.00..1086.90 rows=516 width=52) (actual time=79.532..84.151 rows=718 loops=1) > Filter: ((state)::text = 'WA'::text) You're misreading it. Hash join is done by reading in one table into a hash table, then reading the other table looking up entries in the hash table. The zips are being read into the hash table which is appropriate if it's the smaller table. > Of course still there is the holy grail of getting it to actually use > the indexes. :P > Merge Cond: ("outer"."?column7?" = "inner"."?column5?") Well it looks like you have something strange going on. What data type is car_id in each table? -- greg From pgsql-performance-owner@postgresql.org Tue Dec 2 15:08:54 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DF2CED1B445 for ; Tue, 2 Dec 2003 19:08:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 79577-05 for ; Tue, 2 Dec 2003 15:08:28 -0400 (AST) Received: from penguin.goodinassociates.com (unknown [63.150.225.202]) by svr1.postgresql.org (Postfix) with ESMTP id 5F920D1B437 for ; Tue, 2 Dec 2003 15:08:22 -0400 (AST) Received: from [192.168.14.183] (bluejay.goodinassociates.com [192.168.14.183]) by penguin.goodinassociates.com (8.12.8/linuxconf) with ESMTP id hB2J8Dna011647 for ; Tue, 2 Dec 2003 13:08:13 -0600 Subject: Re: cross table indexes or something? From: Jeremiah Jahn To: postgres performance In-Reply-To: <200312011147.51359.josh@agliodbs.com> References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <1069885949.3176.57.camel@fuji.krosing.net> <1070288942.22346.27.camel@bluejay.goodinassociates.com> <200312011147.51359.josh@agliodbs.com> Content-Type: text/plain Message-Id: <1070392092.22346.43.camel@bluejay.goodinassociates.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 (1.4.5-7) Date: Tue, 02 Dec 2003 13:08:13 -0600 Content-Transfer-Encoding: 7bit X-RAVMilter-Version: 8.4.4(snapshot 20030410) (penguin.goodinassociates.com) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/29 X-Sequence-Number: 4889 Thanks to all, I had already run analyze. But the STATISTICS setting seems to have worked. I'm just not sure what it did..? Would anyone care to explain. On Mon, 2003-12-01 at 13:47, Josh Berkus wrote: > Jeremiah, > > > I've attached the Analyze below. I have no idea why the db thinks there > > is only 1 judge named simth. Is there some what I can inform the DB > > about this. In actuality, there aren't any judges named smith at the > > moment, but there are 22K people named smith. > > No, Hannu meant that you may need to run the following command: > > ANALYZE actor; > > ... to update the database statistics on the actors table. That is a > maintainence task that needs to be run periodically. > > If that doesn't fix the bad plan, then the granularity of statistics on the > full_name column needs updating; I suggest: > > ALTER TABLE actor ALTER COLUMN full_name SET STATISTICS 100; > ANALYZE actor; > > And if it's still choosing a slow nested loop, up the stats to 250. -- Jeremiah Jahn From pgsql-performance-owner@postgresql.org Tue Dec 2 15:13:52 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E3A99D1B4BC for ; Tue, 2 Dec 2003 19:13:51 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81600-03 for ; Tue, 2 Dec 2003 15:13:26 -0400 (AST) Received: from live (64-35-136-226.gohighspeed.com [64.35.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id F31FCD1BC54 for ; Tue, 2 Dec 2003 15:13:20 -0400 (AST) Received: from [192.168.254.101] (helo=89glass.com) by live with esmtp (Exim 3.35 #1 (Debian)) id 1ARFxb-0004TO-00; Tue, 02 Dec 2003 11:13:23 -0800 Message-ID: <3FCCE48B.10406@89glass.com> Date: Tue, 02 Dec 2003 11:14:19 -0800 From: Jared Carr User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Greg Stark Cc: pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> In-Reply-To: <87d6b79har.fsf@stark.dyndns.tv> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/30 X-Sequence-Number: 4890 Greg Stark wrote: > >> Merge Cond: ("outer"."?column7?" = "inner"."?column5?") >> >> > >Well it looks like you have something strange going on. What data type is >car_id in each table? > > > car_id is a varchar(10) in both tables. From pgsql-performance-owner@postgresql.org Tue Dec 2 15:28:57 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3438CD1B475 for ; Tue, 2 Dec 2003 19:28:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 84577-01 for ; Tue, 2 Dec 2003 15:28:29 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 35932D1B4B3 for ; Tue, 2 Dec 2003 15:28:24 -0400 (AST) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4012391; Tue, 02 Dec 2003 11:29:18 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Jeremiah Jahn , postgres performance Subject: Re: cross table indexes or something? Date: Tue, 2 Dec 2003 11:27:52 -0800 User-Agent: KMail/1.4.3 References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <200312011147.51359.josh@agliodbs.com> <1070392092.22346.43.camel@bluejay.goodinassociates.com> In-Reply-To: <1070392092.22346.43.camel@bluejay.goodinassociates.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200312021127.52839.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/31 X-Sequence-Number: 4891 Jeremiah, > Thanks to all, I had already run analyze. But the STATISTICS setting > seems to have worked. I'm just not sure what it did..? Would anyone care > to explain. The STATISTICS setting improves the granularity of statistics kept by the query planner on that column; increasing the granularity (i.e. more random samples) can significantly improve things in cases where you have data whose distribution is significantly skewed. Certainly whenever you see the query planner using a slow nestloop becuase of a bad row-return estimate, it is one of the first things to try. Its drawbacks are 4-fold: 1) to keep it working, you will probably need to run ANALZYE more often than you have been; 2) these ANALYZEs will take longer, and have the annoying side effect of flooring your CPU while they do; 3) You will have to be sure that your vacuum plan includes vacuuming the pg_statistic table as the database superuser, as that table will be getting updated more often. 4) Currently, pg_dump does *not* back up statistics settings. So you will need to save a script which does this in preparation for having to restore your database. Which is why the stats are set low by default. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Tue Dec 2 16:11:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 402B6D1B4B5 for ; Tue, 2 Dec 2003 20:11:40 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89757-07 for ; Tue, 2 Dec 2003 16:11:14 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 486D8D1B432 for ; Tue, 2 Dec 2003 16:11:08 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 3F34937430; Tue, 2 Dec 2003 15:11:03 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1ARGrP-0005xe-00; Tue, 02 Dec 2003 15:11:03 -0500 To: Jared Carr Cc: Greg Stark , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> In-Reply-To: <3FCCE48B.10406@89glass.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 02 Dec 2003 15:11:02 -0500 Message-ID: <871xrn9e09.fsf@stark.dyndns.tv> Lines: 18 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 postgresql.org X-Archive-Number: 200312/32 X-Sequence-Number: 4892 Jared Carr writes: > Greg Stark wrote: > > > > >> Merge Cond: ("outer"."?column7?" = "inner"."?column5?") > >> > > > >Well it looks like you have something strange going on. What data type is > > car_id in each table? > car_id is a varchar(10) in both tables. Well for some reason it's being cast to a text to do the merge. What version of postgres is this btw? The analyzes look like 7.4? -- greg From pgsql-performance-owner@postgresql.org Tue Dec 2 16:28:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5DDB5D1B452 for ; Tue, 2 Dec 2003 20:28:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88959-10 for ; Tue, 2 Dec 2003 16:28:28 -0400 (AST) Received: from live (64-35-136-226.gohighspeed.com [64.35.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id EF7A3D1B470 for ; Tue, 2 Dec 2003 16:28:21 -0400 (AST) Received: from [192.168.254.101] (helo=89glass.com) by live with esmtp (Exim 3.35 #1 (Debian)) id 1ARH89-0004qB-00; Tue, 02 Dec 2003 12:28:21 -0800 Message-ID: <3FCCF61E.90307@89glass.com> Date: Tue, 02 Dec 2003 12:29:18 -0800 From: Jared Carr User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Greg Stark Cc: pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> In-Reply-To: <871xrn9e09.fsf@stark.dyndns.tv> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/33 X-Sequence-Number: 4893 Greg Stark wrote: >Jared Carr writes: > > > >>Greg Stark wrote: >> >> >> >>>> Merge Cond: ("outer"."?column7?" = "inner"."?column5?") >>>> >>>> >>>> >>>Well it looks like you have something strange going on. What data type is >>>car_id in each table? >>> >>> >>car_id is a varchar(10) in both tables. >> >> > >Well for some reason it's being cast to a text to do the merge. > >What version of postgres is this btw? The analyzes look like 7.4? > > > Yes, this is 7.4. From pgsql-performance-owner@postgresql.org Tue Dec 2 16:38:23 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 30F3BD1B4B4 for ; Tue, 2 Dec 2003 20:38:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 94230-05 for ; Tue, 2 Dec 2003 16:37:56 -0400 (AST) Received: from yertle.kcilink.com (yertle.kcilink.com [216.194.193.105]) by svr1.postgresql.org (Postfix) with ESMTP id 1DB69D1D458 for ; Tue, 2 Dec 2003 16:37:50 -0400 (AST) Received: by yertle.kcilink.com (Postfix, from userid 100) id 4CE7C2178C; Tue, 2 Dec 2003 15:37:40 -0500 (EST) From: Vivek Khera MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <16332.63508.196365.801418@yertle.int.kciLink.com> Date: Tue, 2 Dec 2003 15:37:40 -0500 To: pgsql-performance@postgresql.org Subject: autovacuum daemon stops doing work after about an hour X-Mailer: VM 7.17 under 21.4 (patch 14) "Reasonable Discussion" XEmacs Lucid X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/34 X-Sequence-Number: 4894 I took advantage of last weekend to upgrade from 7.2.4 to 7.4.0 on a new faster box. Now I'm trying to implement pg_autovacuum. It seems to work ok, but after about an hour or so, it does nothing. The process still is running, but nothing is sent to the log file. I'm running the daemon as distributed with PG 7.4 release as follows: pg_autovacuum -d4 -V 0.15 -A 1 -U postgres -L /var/tmp/autovacuum.log -D the last few lines of the log are: [2003-12-02 11:43:58 AM] VACUUM ANALYZE "public"."msg_recipients" [2003-12-02 12:24:33 PM] select relfilenode,reltuples,relpages from pg_class where relfilenode=18588239 [2003-12-02 12:24:33 PM] table name: vkmlm."public"."msg_recipients" [2003-12-02 12:24:33 PM] relfilenode: 18588239; relisshared: 0 [2003-12-02 12:24:33 PM] reltuples: 9; relpages: 529132 [2003-12-02 12:24:33 PM] curr_analyze_count: 1961488; cur_delete_count: 1005040 [2003-12-02 12:24:33 PM] ins_at_last_analyze: 1961488; del_at_last_vacuum: 1005040 [2003-12-02 12:24:33 PM] insert_threshold: 509; delete_threshold 1001 [2003-12-02 12:24:33 PM] Performing: VACUUM ANALYZE "public"."user_list" [2003-12-02 12:24:33 PM] VACUUM ANALYZE "public"."user_list" [2003-12-02 12:43:19 PM] select relfilenode,reltuples,relpages from pg_class where relfilenode=18588202 [2003-12-02 12:43:19 PM] table name: vkmlm."public"."user_list" [2003-12-02 12:43:19 PM] relfilenode: 18588202; relisshared: 0 [2003-12-02 12:43:19 PM] reltuples: 9; relpages: 391988 [2003-12-02 12:43:19 PM] curr_analyze_count: 1159843; cur_delete_count: 1118540 [2003-12-02 12:43:19 PM] ins_at_last_analyze: 1159843; del_at_last_vacuum: 1118540 [2003-12-02 12:43:19 PM] insert_threshold: 509; delete_threshold 1001 Then it just sits there. I started it at 11:35am, and it is now 3:30pm. I did the same last night at about 10:58pm, and it ran and did work until 11:57pm, then sat there until I killed/restarted pg_autovacuum this morning at 11:35. The process is not using any CPU time. I just killed/restarted it and it found work to do on my busy tables which I'd expect. I'm running Postgres 7.4 release on FreeBSD 4.9-RELEASE. -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Vivek Khera, Ph.D. Khera Communications, Inc. Internet: khera@kciLink.com Rockville, MD +1-240-453-8497 AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/ From pgsql-performance-owner@postgresql.org Tue Dec 2 17:58:14 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 110BED1D591 for ; Tue, 2 Dec 2003 21:58:09 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06610-04 for ; Tue, 2 Dec 2003 17:57:44 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 2D7C2D1D58F for ; Tue, 2 Dec 2003 17:57:34 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 507401E13; Tue, 2 Dec 2003 16:57:27 -0500 (EST) To: Josh Berkus Cc: Jeremiah Jahn , postgres performance Subject: Re: cross table indexes or something? From: Neil Conway In-Reply-To: <200312021127.52839.josh@agliodbs.com> (Josh Berkus's message of "Tue, 2 Dec 2003 11:27:52 -0800") References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <200312011147.51359.josh@agliodbs.com> <1070392092.22346.43.camel@bluejay.goodinassociates.com> <200312021127.52839.josh@agliodbs.com> Date: Tue, 02 Dec 2003 16:57:27 -0500 Message-ID: <87llpu288o.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/35 X-Sequence-Number: 4895 Josh Berkus writes: > 1) to keep it working, you will probably need to run ANALZYE more > often than you have been; I'm not sure why this would be the case -- can you elaborate? > 4) Currently, pg_dump does *not* back up statistics settings. Yes, it does. -Neil From pgsql-performance-owner@postgresql.org Tue Dec 2 18:32:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DA75FD1B590 for ; Tue, 2 Dec 2003 22:32:36 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04544-08 for ; Tue, 2 Dec 2003 18:32:12 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 61304D1B4BD for ; Tue, 2 Dec 2003 18:32:05 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 15416372CB; Tue, 2 Dec 2003 17:32:12 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1ARJ3z-0006Vz-00; Tue, 02 Dec 2003 17:32:11 -0500 To: Jared Carr Cc: Greg Stark , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> In-Reply-To: <3FCCF61E.90307@89glass.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 02 Dec 2003 17:32:11 -0500 Message-ID: <87ptf697h0.fsf@stark.dyndns.tv> Lines: 62 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 postgresql.org X-Archive-Number: 200312/36 X-Sequence-Number: 4896 Jared Carr writes: > Greg Stark wrote: > > > Well it looks like you have something strange going on. What data type is > > car_id in each table? > > > car_id is a varchar(10) in both tables. Huh. The following shows something strange. It seems joining on two varchars no longer works well. Instead the optimizer has to convert both columns to text. I know some inter-type comparisons were removed a while ago, but I would not have thought that would effect varchar-varchar comparisons. I think this is pretty bad. test=# create table a (x varchar primary key); NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index "a_pkey" for table "a" CREATE TABLE test=# create table b (x varchar primary key); NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index "b_pkey" for table "b" CREATE TABLE test=# select * from a,b where a.x=b.x; x | x ---+--- (0 rows) test=# explain select * from a,b where a.x=b.x; QUERY PLAN ------------------------------------------------------------------ Merge Join (cost=139.66..159.67 rows=1001 width=64) Merge Cond: ("outer"."?column2?" = "inner"."?column2?") -> Sort (cost=69.83..72.33 rows=1000 width=32) Sort Key: (a.x)::text -> Seq Scan on a (cost=0.00..20.00 rows=1000 width=32) -> Sort (cost=69.83..72.33 rows=1000 width=32) Sort Key: (b.x)::text -> Seq Scan on b (cost=0.00..20.00 rows=1000 width=32) (8 rows) test=# create table a2 (x text primary key); NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index "a2_pkey" for table "a2" CREATE TABLE test=# create table b2 (x text primary key); NOTICE: CREATE TABLE / PRIMARY KEY will create implicit index "b2_pkey" for table "b2" CREATE TABLE test=# explain select * from a2,b2 where a2.x=b2.x; QUERY PLAN ------------------------------------------------------------------- Hash Join (cost=22.50..57.51 rows=1001 width=64) Hash Cond: ("outer".x = "inner".x) -> Seq Scan on a2 (cost=0.00..20.00 rows=1000 width=32) -> Hash (cost=20.00..20.00 rows=1000 width=32) -> Seq Scan on b2 (cost=0.00..20.00 rows=1000 width=32) (5 rows) -- greg From pgsql-performance-owner@postgresql.org Tue Dec 2 18:44:04 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 65E16D1B4D7 for ; Tue, 2 Dec 2003 22:43:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 13860-02 for ; Tue, 2 Dec 2003 18:43:34 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 42F3CD1B495 for ; Tue, 2 Dec 2003 18:43:27 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 1DD4936F21; Tue, 2 Dec 2003 17:43:35 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1ARJF1-0006ZC-00; Tue, 02 Dec 2003 17:43:35 -0500 To: Greg Stark Cc: Jared Carr , Greg Stark , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> In-Reply-To: <87ptf697h0.fsf@stark.dyndns.tv> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 02 Dec 2003 17:43:34 -0500 Message-ID: <87k75e96y1.fsf@stark.dyndns.tv> Lines: 34 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 postgresql.org X-Archive-Number: 200312/37 X-Sequence-Number: 4897 Greg Stark writes: > Huh. The following shows something strange. Worse, with enable_hashjoin off it's even more obvious something's broken: test=# set enable_hashjoin = off; SET test=# explain select * from a,b where a.x=b.x; QUERY PLAN ------------------------------------------------------------------ Merge Join (cost=139.66..159.67 rows=1001 width=64) Merge Cond: ("outer"."?column2?" = "inner"."?column2?") -> Sort (cost=69.83..72.33 rows=1000 width=32) Sort Key: (a.x)::text -> Seq Scan on a (cost=0.00..20.00 rows=1000 width=32) -> Sort (cost=69.83..72.33 rows=1000 width=32) Sort Key: (b.x)::text -> Seq Scan on b (cost=0.00..20.00 rows=1000 width=32) (8 rows) test=# explain select * from a2,b2 where a2.x=b2.x; QUERY PLAN ----------------------------------------------------------------------------- Merge Join (cost=0.00..63.04 rows=1001 width=64) Merge Cond: ("outer".x = "inner".x) -> Index Scan using a2_pkey on a2 (cost=0.00..24.00 rows=1000 width=32) -> Index Scan using b2_pkey on b2 (cost=0.00..24.00 rows=1000 width=32) (4 rows) -- greg From pgsql-performance-owner@postgresql.org Tue Dec 2 19:18:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2C8DDD1C951 for ; Tue, 2 Dec 2003 23:18:04 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 09502-10 for ; Tue, 2 Dec 2003 19:17:39 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 17647D1B465 for ; Tue, 2 Dec 2003 19:17:32 -0400 (AST) Received: from [66.219.92.2] (HELO temoku) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4013451; Tue, 02 Dec 2003 15:18:29 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Neil Conway Subject: Re: cross table indexes or something? Date: Tue, 2 Dec 2003 15:04:28 -0800 User-Agent: KMail/1.4.3 Cc: Jeremiah Jahn , postgres performance References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <200312021127.52839.josh@agliodbs.com> <87llpu288o.fsf@mailbox.samurai.com> In-Reply-To: <87llpu288o.fsf@mailbox.samurai.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200312021504.28020.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/38 X-Sequence-Number: 4898 Neil, > > 1) to keep it working, you will probably need to run ANALZYE more > > often than you have been; >=20 > I'm not sure why this would be the case -- can you elaborate? For the more granular stats to be useful, they have to be accurate; otherwi= se=20 you'll go back to a nestloop as soon as the query planner encounters a valu= e=20 that it doens't think is in the table at all.=20 >=20 > > 4) Currently, pg_dump does *not* back up statistics settings. >=20 > Yes, it does. Oh, good. Was this a 7.4 improvement? I missed that in the changelogs .= ... --=20 -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Tue Dec 2 19:37:42 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 99D50D1B465 for ; Tue, 2 Dec 2003 23:37:40 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21727-07 for ; Tue, 2 Dec 2003 19:37:16 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id C7C32D1B437 for ; Tue, 2 Dec 2003 19:37:09 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 113FD1EA3; Tue, 2 Dec 2003 18:37:17 -0500 (EST) To: josh@agliodbs.com Cc: Jeremiah Jahn , postgres performance Subject: Re: cross table indexes or something? From: Neil Conway In-Reply-To: <200312021504.28020.josh@agliodbs.com> (Josh Berkus's message of "Tue, 2 Dec 2003 15:04:28 -0800") References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <200312021127.52839.josh@agliodbs.com> <87llpu288o.fsf@mailbox.samurai.com> <200312021504.28020.josh@agliodbs.com> Date: Tue, 02 Dec 2003 18:37:17 -0500 Message-ID: <87d6b623ma.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/39 X-Sequence-Number: 4899 Josh Berkus writes: > Oh, good. Was this a 7.4 improvement? No, it was in 7.3 -Neil From pgsql-performance-owner@postgresql.org Tue Dec 2 19:51:18 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4771DD1B465 for ; Tue, 2 Dec 2003 23:51:13 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 22485-06 for ; Tue, 2 Dec 2003 19:50:49 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id DDC00D1B4AA for ; Tue, 2 Dec 2003 19:50:41 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB2Nol19025477; Tue, 2 Dec 2003 18:50:47 -0500 (EST) To: Greg Stark Cc: Jared Carr , pgsql-performance@postgresql.org Subject: Re: A question on the query planner In-reply-to: <87ptf697h0.fsf@stark.dyndns.tv> References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> Comments: In-reply-to Greg Stark message dated "02 Dec 2003 17:32:11 -0500" Date: Tue, 02 Dec 2003 18:50:47 -0500 Message-ID: <25476.1070409047@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/40 X-Sequence-Number: 4900 Greg Stark writes: > Huh. The following shows something strange. It seems joining on two varchars > no longer works well. Instead the optimizer has to convert both columns to > text. Define "no longer works well". varchar doesn't have its own comparison operators anymore, but AFAIK that makes no difference. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Dec 2 21:33:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1D1CED1DA62 for ; Wed, 3 Dec 2003 01:33:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34464-04 for ; Tue, 2 Dec 2003 21:33:19 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 098B2D1D400 for ; Tue, 2 Dec 2003 21:33:12 -0400 (AST) Received: from stark.dyndns.tv (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 44A4F36D1B; Tue, 2 Dec 2003 20:33:16 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.dyndns.tv with smtp (Exim 3.36 #1 (Debian)) id 1ARLtE-00072x-00; Tue, 02 Dec 2003 20:33:16 -0500 To: Tom Lane Cc: Greg Stark , Jared Carr , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> In-Reply-To: <25476.1070409047@sss.pgh.pa.us> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 02 Dec 2003 20:33:15 -0500 Message-ID: <87znea7kis.fsf@stark.dyndns.tv> Lines: 34 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 postgresql.org X-Archive-Number: 200312/41 X-Sequence-Number: 4901 Tom Lane writes: > Define "no longer works well". varchar doesn't have its own comparison > operators anymore, but AFAIK that makes no difference. Well it seems to completely bar the use of a straight merge join between two index scans: test=# set enable_seqscan = off; SET test=# explain select * from a,b where a.x=b.x; QUERY PLAN --------------------------------------------------------------------------- Nested Loop (cost=100000000.00..100002188.86 rows=1001 width=64) -> Seq Scan on a (cost=100000000.00..100000020.00 rows=1000 width=32) -> Index Scan using b_pkey on b (cost=0.00..2.16 rows=1 width=32) Index Cond: (("outer".x)::text = (b.x)::text) (4 rows) test=# explain select * from a2,b2 where a2.x=b2.x; QUERY PLAN ----------------------------------------------------------------------------- Merge Join (cost=0.00..63.04 rows=1001 width=64) Merge Cond: ("outer".x = "inner".x) -> Index Scan using a2_pkey on a2 (cost=0.00..24.00 rows=1000 width=32) -> Index Scan using b2_pkey on b2 (cost=0.00..24.00 rows=1000 width=32) (4 rows) -- greg From pgsql-performance-owner@postgresql.org Tue Dec 2 21:46:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9C528D1D3FB for ; Wed, 3 Dec 2003 01:46:51 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34819-06 for ; Tue, 2 Dec 2003 21:46:28 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id C2B90D1D400 for ; Tue, 2 Dec 2003 21:46:19 -0400 (AST) Received: from familyhealth.com.au (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (8.12.9p1/8.12.9) with ESMTP id hB31jNoD000629; Wed, 3 Dec 2003 09:45:23 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <3FCD4170.6060507@familyhealth.com.au> Date: Wed, 03 Dec 2003 09:50:40 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Josh Berkus Cc: Jeremiah Jahn , postgres performance Subject: Re: cross table indexes or something? References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <200312011147.51359.josh@agliodbs.com> <1070392092.22346.43.camel@bluejay.goodinassociates.com> <200312021127.52839.josh@agliodbs.com> In-Reply-To: <200312021127.52839.josh@agliodbs.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/42 X-Sequence-Number: 4902 > 4) Currently, pg_dump does *not* back up statistics settings. Is this a TODO? Chris From pgsql-performance-owner@postgresql.org Tue Dec 2 22:00:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A37DBD1D362 for ; Wed, 3 Dec 2003 02:00:55 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31599-07 for ; Tue, 2 Dec 2003 22:00:32 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id A3B92D1D94A for ; Tue, 2 Dec 2003 22:00:23 -0400 (AST) Received: from familyhealth.com.au (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (8.12.9p1/8.12.9) with ESMTP id hB31xLoD000771; Wed, 3 Dec 2003 09:59:22 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <3FCD44B7.8000105@familyhealth.com.au> Date: Wed, 03 Dec 2003 10:04:39 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Christopher Kings-Lynne Cc: Josh Berkus , Jeremiah Jahn , postgres performance Subject: Re: cross table indexes or something? References: <1069877651.22346.13.camel@bluejay.goodinassociates.com> <200312011147.51359.josh@agliodbs.com> <1070392092.22346.43.camel@bluejay.goodinassociates.com> <200312021127.52839.josh@agliodbs.com> <3FCD4170.6060507@familyhealth.com.au> In-Reply-To: <3FCD4170.6060507@familyhealth.com.au> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/43 X-Sequence-Number: 4903 >> 4) Currently, pg_dump does *not* back up statistics settings. > > > Is this a TODO? Oops - sorry I thought you meant 'pg_dump does not back up statistics'. Probably still should be a TODO :) Chris From pgsql-performance-owner@postgresql.org Tue Dec 2 22:23:24 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0F6BDD1B439 for ; Wed, 3 Dec 2003 02:23:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 39445-07 for ; Tue, 2 Dec 2003 22:22:57 -0400 (AST) Received: from necsin-fw.nec.com.sg (necsin-fw.nec.com.sg [203.127.255.1]) by svr1.postgresql.org (Postfix) with ESMTP id EF097D1CCCD for ; Tue, 2 Dec 2003 22:22:47 -0400 (AST) Received: from kato.nec.com.sg (kato.nec.com.sg [203.127.254.2]) by necsin-fw.nec.com.sg with ESMTP id hB32MjrE027562 for ; Wed, 3 Dec 2003 10:22:46 +0800 (SGT) Received: from necsind1.nec.com.sg ([203.127.252.220]) by kato.nec.com.sg with ESMTP id hB32Mju2007470 for ; Wed, 3 Dec 2003 10:22:45 +0800 (SGT) Subject: Minimum hardware requirements for Postgresql db To: pgsql-performance@postgresql.org X-Mailer: Lotus Notes Release 5.0.5 September 22, 2000 Message-ID: From: CHEWTC@ap.nec.com.sg Date: Wed, 3 Dec 2003 10:22:51 +0800 X-MIMETrack: Serialize by Router on NECSIND1/WTC/NECSIN(Release 5.0.12 |February 13, 2003) at 12/03/2003 10:22:51 AM MIME-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=BAYES_30, NO_REAL_NAME X-Spam-Level: X-Archive-Number: 200312/44 X-Sequence-Number: 4904 Dear all We would be recommending to our ct. on the use of Postgresql db as compared to MS SQL Server. We are targetting to use Redhat Linux ES v2.1, Postgresql v7.3.4 and Postgresql ODBC 07.03.0100. We would like to know the minimum specs required for our below target. The minimum specs is referring to no. of CPU, memory, harddisk capacity, RAID technology etc. And also the Postgresql parameters and configuration to run such a system. 1) We will be running 2 x Postgresql db in the machine. 2) Total number of connections to be around 100. The connections from the clients machines will be in ODBC and socket connections. 3) Estimated number of transactions to be written into the Postgresql db is around 15000 records per day. The growth rate in terms of number of connections is around 10% per year and the data retention is kept on average at least for 18 months for the 2 databases. Are there any reference books or sites that I can tap on for the above requirement? Thank you, REgards. From pgsql-performance-owner@postgresql.org Tue Dec 2 23:53:57 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5D173D1B486 for ; Wed, 3 Dec 2003 03:53:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 45763-06 for ; Tue, 2 Dec 2003 23:53:30 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 05D65D1B44D for ; Tue, 2 Dec 2003 23:53:20 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB33rR19026700; Tue, 2 Dec 2003 22:53:27 -0500 (EST) To: Greg Stark Cc: Jared Carr , pgsql-performance@postgresql.org Subject: Re: A question on the query planner In-reply-to: <87znea7kis.fsf@stark.dyndns.tv> References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> Comments: In-reply-to Greg Stark message dated "02 Dec 2003 20:33:15 -0500" Date: Tue, 02 Dec 2003 22:53:27 -0500 Message-ID: <26699.1070423607@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/45 X-Sequence-Number: 4905 Greg Stark writes: > Tom Lane writes: >> Define "no longer works well". > Well it seems to completely bar the use of a straight merge join between two > index scans: Hmmm ... [squints] ... it's not supposed to do that ... [digs] ... yeah, there's something busted here. Will get back to you ... regards, tom lane From pgsql-performance-owner@postgresql.org Wed Dec 3 00:58:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C711BD1C941 for ; Wed, 3 Dec 2003 04:58:35 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 65370-02 for ; Wed, 3 Dec 2003 00:58:05 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id E40EFD1C526 for ; Wed, 3 Dec 2003 00:58:04 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hB34w0bI026629 for ; Wed, 3 Dec 2003 04:58:00 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hB34iPBu024920 for pgsql-performance@postgresql.org; Wed, 3 Dec 2003 04:44:25 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Minimum hardware requirements for Postgresql db Date: Tue, 02 Dec 2003 23:44:21 -0500 Organization: cbbrowne Computing Inc Lines: 86 Message-ID: References: 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/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:uax1B1VcxmD25T2Hq2Lp1p3MbCc= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/46 X-Sequence-Number: 4906 After takin a swig o' Arrakan spice grog, CHEWTC@ap.nec.com.sg belched out: > We would be recommending to our ct. on the use of Postgresql db as > compared to MS SQL Server. We are targetting to use Redhat Linux ES > v2.1, Postgresql v7.3.4 and Postgresql ODBC 07.03.0100. > > We would like to know the minimum specs required for our below > target. The minimum specs is referring to no. of CPU, memory, > harddisk capacity, RAID technology etc. And also the Postgresql > parameters and configuration to run such a system. > > 1) We will be running 2 x Postgresql db in the machine. > > 2) Total number of connections to be around 100. The connections > from the clients machines will be in ODBC and socket connections. > > 3) Estimated number of transactions to be written into the > Postgresql db is around 15000 records per day. > > The growth rate in terms of number of connections is around 10% per > year and the data retention is kept on average at least for 18 > months for the 2 databases. > > Are there any reference books or sites that I can tap on for the > above requirement? Perhaps the best reference on detailed performance information is the "General Bits" documents. These don't point particularly at minimal hardware requirements, but rather at how to configure the DBMS to best reflect what hardware you have. But there's some degree to which you can work backwards... If you'll need to support 100 concurrent connections, then minimum shared_buffers is 200, which implies 1600K of RAM required for shared buffers. 100 connections probably implies around 100MB of memory for the backend processes to support the connections. That all points to the notion that you'd more than probably get half-decent performance if you had a mere 256MB of RAM, which is about $50 worth these days. None of it sounds terribly challenging; 15K records per day is 625 records per hour which represents an INSERT every 6 seconds. Even if that has to fit into an 8 hour day, that's still not a high number of transactions per second. That _sounds like_ an application that could work on old, obsolete hardware. I would imagine that my old Intel Pentium Pro 200 might cope with the load, in much the way that that server is more than capable of supporting a web server that would serve a local workgroup. (I only have 64MB of RAM on that box, which would be a mite low, but it's an _ancient_ server...) The only thing that makes me a little suspicious that there's something funny about the prescription is your indication of having 100 concurrent users, which is really rather heavyweight in comparison with the comparatively tiny number of transactions. Is this for some sort of "departmental application"? Where there's a lot of manual data entry, so that each user would generate a transaction every 3-4 minutes? That actually sounds about right... Let me suggest that the "cost driver" in this will _not_ be the cost of the hardware to support the database itself; it will instead be in having redundant hardware and backup hardware to ensure reliability. It would seem likely that just about any sort of modern hardware would be pretty adequate to the task. You can hardly _buy_ a system with less than Gigahertz-speed CPUs, 40GB of disk, and 256MB of RAM. Upgrade to have 2 SCSI disks, 512MB (or more, which is better) of RAM, and the cost of a suitable system still won't be outrageous. Double it, buying a standby server, and the cost still oughtn't be real scary. And if the application is important, you _should_ have a standby server, irrespective of what software you might be running. -- (reverse (concatenate 'string "moc.enworbbc" "@" "enworbbc")) http://www3.sympatico.ca/cbbrowne/x.html Rules of the Evil Overlord #199. "I will not make alliances with those more powerful than myself. Such a person would only double-cross me in my moment of glory. I will make alliances with those less powerful than myself. I will then double-cross them in their moment of glory." From pgsql-performance-owner@postgresql.org Wed Dec 3 13:49:49 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E40E0D1D58F for ; Wed, 3 Dec 2003 17:49:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68173-09 for ; Wed, 3 Dec 2003 13:49:15 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id D198BD1B8EA for ; Wed, 3 Dec 2003 13:49:11 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB3Hn719002705; Wed, 3 Dec 2003 12:49:07 -0500 (EST) To: Greg Stark , Jared Carr , pgsql-performance@postgresql.org Subject: Re: A question on the query planner In-reply-to: <26699.1070423607@sss.pgh.pa.us> References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> <26699.1070423607@sss.pgh.pa.us> Comments: In-reply-to Tom Lane message dated "Tue, 02 Dec 2003 22:53:27 -0500" Date: Wed, 03 Dec 2003 12:49:06 -0500 Message-ID: <2704.1070473746@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/47 X-Sequence-Number: 4907 > Hmmm ... [squints] ... it's not supposed to do that ... The attached patch seems to make it better. regards, tom lane Index: src/backend/optimizer/path/costsize.c =================================================================== RCS file: /cvsroot/pgsql-server/src/backend/optimizer/path/costsize.c,v retrieving revision 1.115 diff -c -r1.115 costsize.c *** src/backend/optimizer/path/costsize.c 5 Oct 2003 22:44:25 -0000 1.115 --- src/backend/optimizer/path/costsize.c 3 Dec 2003 17:40:58 -0000 *************** *** 1322,1327 **** --- 1322,1331 ---- float4 *numbers; int nnumbers; + /* Ignore any binary-compatible relabeling */ + if (var && IsA(var, RelabelType)) + var = (Var *) ((RelabelType *) var)->arg; + /* * Lookup info about var's relation and attribute; if none available, * return default estimate. Index: src/backend/optimizer/path/pathkeys.c =================================================================== RCS file: /cvsroot/pgsql-server/src/backend/optimizer/path/pathkeys.c,v retrieving revision 1.53 diff -c -r1.53 pathkeys.c *** src/backend/optimizer/path/pathkeys.c 4 Aug 2003 02:40:00 -0000 1.53 --- src/backend/optimizer/path/pathkeys.c 3 Dec 2003 17:40:58 -0000 *************** *** 25,36 **** #include "optimizer/tlist.h" #include "optimizer/var.h" #include "parser/parsetree.h" #include "parser/parse_func.h" #include "utils/lsyscache.h" #include "utils/memutils.h" ! static PathKeyItem *makePathKeyItem(Node *key, Oid sortop); static List *make_canonical_pathkey(Query *root, PathKeyItem *item); static Var *find_indexkey_var(Query *root, RelOptInfo *rel, AttrNumber varattno); --- 25,37 ---- #include "optimizer/tlist.h" #include "optimizer/var.h" #include "parser/parsetree.h" + #include "parser/parse_expr.h" #include "parser/parse_func.h" #include "utils/lsyscache.h" #include "utils/memutils.h" ! static PathKeyItem *makePathKeyItem(Node *key, Oid sortop, bool checkType); static List *make_canonical_pathkey(Query *root, PathKeyItem *item); static Var *find_indexkey_var(Query *root, RelOptInfo *rel, AttrNumber varattno); *************** *** 41,50 **** * create a PathKeyItem node */ static PathKeyItem * ! makePathKeyItem(Node *key, Oid sortop) { PathKeyItem *item = makeNode(PathKeyItem); item->key = key; item->sortop = sortop; return item; --- 42,70 ---- * create a PathKeyItem node */ static PathKeyItem * ! makePathKeyItem(Node *key, Oid sortop, bool checkType) { PathKeyItem *item = makeNode(PathKeyItem); + /* + * Some callers pass expressions that are not necessarily of the same + * type as the sort operator expects as input (for example when dealing + * with an index that uses binary-compatible operators). We must relabel + * these with the correct type so that the key expressions will be seen + * as equal() to expressions that have been correctly labeled. + */ + if (checkType) + { + Oid lefttype, + righttype; + + op_input_types(sortop, &lefttype, &righttype); + if (exprType(key) != lefttype) + key = (Node *) makeRelabelType((Expr *) key, + lefttype, -1, + COERCE_DONTCARE); + } + item->key = key; item->sortop = sortop; return item; *************** *** 70,78 **** { Expr *clause = restrictinfo->clause; PathKeyItem *item1 = makePathKeyItem(get_leftop(clause), ! restrictinfo->left_sortop); PathKeyItem *item2 = makePathKeyItem(get_rightop(clause), ! restrictinfo->right_sortop); List *newset, *cursetlink; --- 90,100 ---- { Expr *clause = restrictinfo->clause; PathKeyItem *item1 = makePathKeyItem(get_leftop(clause), ! restrictinfo->left_sortop, ! false); PathKeyItem *item2 = makePathKeyItem(get_rightop(clause), ! restrictinfo->right_sortop, ! false); List *newset, *cursetlink; *************** *** 668,674 **** } /* OK, make a sublist for this sort key */ ! item = makePathKeyItem(indexkey, sortop); cpathkey = make_canonical_pathkey(root, item); /* --- 690,696 ---- } /* OK, make a sublist for this sort key */ ! item = makePathKeyItem(indexkey, sortop, true); cpathkey = make_canonical_pathkey(root, item); /* *************** *** 785,791 **** tle->resdom->restypmod, 0); outer_item = makePathKeyItem((Node *) outer_var, ! sub_item->sortop); /* score = # of mergejoin peers */ score = count_canonical_peers(root, outer_item); /* +1 if it matches the proper query_pathkeys item */ --- 807,814 ---- tle->resdom->restypmod, 0); outer_item = makePathKeyItem((Node *) outer_var, ! sub_item->sortop, ! true); /* score = # of mergejoin peers */ score = count_canonical_peers(root, outer_item); /* +1 if it matches the proper query_pathkeys item */ *************** *** 893,899 **** PathKeyItem *pathkey; sortkey = get_sortgroupclause_expr(sortcl, tlist); ! pathkey = makePathKeyItem(sortkey, sortcl->sortop); /* * The pathkey becomes a one-element sublist, for now; --- 916,922 ---- PathKeyItem *pathkey; sortkey = get_sortgroupclause_expr(sortcl, tlist); ! pathkey = makePathKeyItem(sortkey, sortcl->sortop, true); /* * The pathkey becomes a one-element sublist, for now; *************** *** 937,943 **** { oldcontext = MemoryContextSwitchTo(GetMemoryChunkContext(restrictinfo)); key = get_leftop(restrictinfo->clause); ! item = makePathKeyItem(key, restrictinfo->left_sortop); restrictinfo->left_pathkey = make_canonical_pathkey(root, item); MemoryContextSwitchTo(oldcontext); } --- 960,966 ---- { oldcontext = MemoryContextSwitchTo(GetMemoryChunkContext(restrictinfo)); key = get_leftop(restrictinfo->clause); ! item = makePathKeyItem(key, restrictinfo->left_sortop, false); restrictinfo->left_pathkey = make_canonical_pathkey(root, item); MemoryContextSwitchTo(oldcontext); } *************** *** 945,951 **** { oldcontext = MemoryContextSwitchTo(GetMemoryChunkContext(restrictinfo)); key = get_rightop(restrictinfo->clause); ! item = makePathKeyItem(key, restrictinfo->right_sortop); restrictinfo->right_pathkey = make_canonical_pathkey(root, item); MemoryContextSwitchTo(oldcontext); } --- 968,974 ---- { oldcontext = MemoryContextSwitchTo(GetMemoryChunkContext(restrictinfo)); key = get_rightop(restrictinfo->clause); ! item = makePathKeyItem(key, restrictinfo->right_sortop, false); restrictinfo->right_pathkey = make_canonical_pathkey(root, item); MemoryContextSwitchTo(oldcontext); } Index: src/backend/utils/cache/lsyscache.c =================================================================== RCS file: /cvsroot/pgsql-server/src/backend/utils/cache/lsyscache.c,v retrieving revision 1.108 diff -c -r1.108 lsyscache.c *** src/backend/utils/cache/lsyscache.c 4 Oct 2003 18:22:59 -0000 1.108 --- src/backend/utils/cache/lsyscache.c 3 Dec 2003 17:40:58 -0000 *************** *** 465,470 **** --- 465,493 ---- } /* + * op_input_types + * + * Returns the left and right input datatypes for an operator + * (InvalidOid if not relevant). + */ + void + op_input_types(Oid opno, Oid *lefttype, Oid *righttype) + { + HeapTuple tp; + Form_pg_operator optup; + + tp = SearchSysCache(OPEROID, + ObjectIdGetDatum(opno), + 0, 0, 0); + if (!HeapTupleIsValid(tp)) /* shouldn't happen */ + elog(ERROR, "cache lookup failed for operator %u", opno); + optup = (Form_pg_operator) GETSTRUCT(tp); + *lefttype = optup->oprleft; + *righttype = optup->oprright; + ReleaseSysCache(tp); + } + + /* * op_mergejoinable * * Returns the left and right sort operators corresponding to a Index: src/include/utils/lsyscache.h =================================================================== RCS file: /cvsroot/pgsql-server/src/include/utils/lsyscache.h,v retrieving revision 1.82 diff -c -r1.82 lsyscache.h *** src/include/utils/lsyscache.h 4 Oct 2003 18:22:59 -0000 1.82 --- src/include/utils/lsyscache.h 3 Dec 2003 17:41:00 -0000 *************** *** 40,45 **** --- 40,46 ---- extern bool opclass_is_hash(Oid opclass); extern RegProcedure get_opcode(Oid opno); extern char *get_opname(Oid opno); + extern void op_input_types(Oid opno, Oid *lefttype, Oid *righttype); extern bool op_mergejoinable(Oid opno, Oid *leftOp, Oid *rightOp); extern void op_mergejoin_crossops(Oid opno, Oid *ltop, Oid *gtop, RegProcedure *ltproc, RegProcedure *gtproc); From pgsql-performance-owner@postgresql.org Wed Dec 3 14:41:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3F618D1C976 for ; Wed, 3 Dec 2003 18:41:57 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73995-09 for ; Wed, 3 Dec 2003 14:41:27 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 46C2BD1B4E1 for ; Wed, 3 Dec 2003 14:41:22 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hB3IdZDW011329; Wed, 3 Dec 2003 11:39:35 -0700 (MST) Date: Wed, 3 Dec 2003 11:22:47 -0700 (MST) From: "scott.marlowe" To: Cc: Subject: Re: Minimum hardware requirements for Postgresql db In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/49 X-Sequence-Number: 4909 On Wed, 3 Dec 2003 CHEWTC@ap.nec.com.sg wrote: > Dear all > > We would be recommending to our ct. on the use of Postgresql db as compared > to MS SQL Server. We are targetting to use Redhat Linux ES v2.1, Postgresql > v7.3.4 and Postgresql ODBC 07.03.0100. > > We would like to know the minimum specs required for our below target. The > minimum specs is referring to no. of CPU, memory, harddisk capacity, RAID > technology etc. And also the Postgresql parameters and configuration to run > such a system. > > 1) We will be running 2 x Postgresql db in the machine. > > 2) Total number of connections to be around 100. The connections from the > clients machines will be in ODBC and socket connections. > > 3) Estimated number of transactions to be written into the Postgresql db is > around 15000 records per day. > > > The growth rate in terms of number of connections is around 10% per year > and the data retention is kept on average at least for 18 months for the 2 > databases. > > Are there any reference books or sites that I can tap on for the above > requirement? Like another poster pointed out, this is a walk in the park for postgresql. My workstation (1.1GHz celeron, 40 gig IDE drive, 512 Meg memory) could handle this load while still being my workstation. :-) From pgsql-performance-owner@postgresql.org Wed Dec 3 14:27:33 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 58B54D1B438 for ; Wed, 3 Dec 2003 18:27:24 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77358-02 for ; Wed, 3 Dec 2003 14:26:53 -0400 (AST) Received: from live (64-35-136-226.gohighspeed.com [64.35.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id E60DFD1D30D for ; Wed, 3 Dec 2003 14:26:51 -0400 (AST) Received: from [192.168.254.101] (helo=89glass.com) by live with esmtp (Exim 3.35 #1 (Debian)) id 1ARbhc-0001Gw-00; Wed, 03 Dec 2003 10:26:20 -0800 Message-ID: <3FCE2B08.6090606@89glass.com> Date: Wed, 03 Dec 2003 10:27:20 -0800 From: Jared Carr User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: Greg Stark , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> <26699.1070423607@sss.pgh.pa.us> <2704.1070473746@sss.pgh.pa.us> In-Reply-To: <2704.1070473746@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 postgresql.org X-Archive-Number: 200312/48 X-Sequence-Number: 4908 Tom Lane wrote: >>Hmmm ... [squints] ... it's not supposed to do that ... >> >> > >The attached patch seems to make it better. > > > The patch definitely makes things more consistent...unfortunately it is more consistent toward the slower execution times. Of course I am looking at this simply from a straight performance standpoint and not a viewpoint of what *should* be happening. At any rate here are the query plans with the various settings. Default Settings: QUERY PLAN -------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=15290.20..15300.34 rows=4058 width=80) (actual time=2944.650..2951.292 rows=4672 loops=1) Sort Key: q.date_time -> Hash Join (cost=13529.79..15046.99 rows=4058 width=80) (actual time=2678.033..2873.475 rows=4672 loops=1) Hash Cond: (("outer".car_id)::text = ("inner".car_id)::text) -> Seq Scan on cars c (cost=0.00..227.01 rows=9401 width=37) (actual time=19.887..50.971 rows=9401 loops=1) -> Hash (cost=13475.65..13475.65 rows=4058 width=62) (actual time=2643.377..2643.377 rows=0 loops=1) -> Hash Join (cost=1088.19..13475.65 rows=4058 width=62) (actual time=86.739..2497.558 rows=10284 loops=1) Hash Cond: (("outer".zip)::text = ("inner".zip)::text) -> Seq Scan on quotes q (cost=0.00..10664.25 rows=336525 width=27) (actual time=0.223..1308.561 rows=340694 loops=1) -> Hash (cost=1086.90..1086.90 rows=516 width=52) (actual time=84.329..84.329 rows=0 loops=1) -> Seq Scan on zips z (cost=0.00..1086.90 rows=516 width=52) (actual time=78.363..82.901 rows=718 loops=1) Filter: ((state)::text = 'WA'::text) Total runtime: 2955.366 ms SET enable_seqscan=false; QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=103557.82..103567.97 rows=4058 width=80) (actual time=1015.122..1021.750 rows=4673 loops=1) Sort Key: q.date_time -> Merge Join (cost=102734.94..103314.61 rows=4058 width=80) (actual time=802.908..941.520 rows=4673 loops=1) Merge Cond: ("outer"."?column7?" = ("inner".car_id)::text) -> Sort (cost=102734.94..102745.08 rows=4058 width=62) (actual time=802.112..812.755 rows=4827 loops=1) Sort Key: (q.car_id)::text -> Nested Loop (cost=0.00..102491.73 rows=4058 width=62) (actual time=148.535..555.653 rows=10285 loops=1) -> Index Scan using zip_zips_index on zips z (cost=0.00..1272.69 rows=516 width=52) (actual time=148.243..155.577 rows=718 loops=1) Filter: ((state)::text = 'WA'::text) -> Index Scan using zip_quotes_index on quotes q (cost=0.00..195.55 rows=48 width=27) (actual time=0.042..0.454 rows=14 loops=718) Index Cond: (("outer".zip)::text = (q.zip)::text) -> Index Scan using cars_car_id_btree_index on cars c (cost=0.00..506.87 rows=9401 width=37) (actual time=0.220..46.910 rows=12019 loops=1) Total runtime: 1027.339 ms There is still a 3x decrease in execution time here, but it is overall slower than before the patch was applied. SET enable_mergejoin = false; AND SET enable_seqscan = false; QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=104586.15..104596.29 rows=4058 width=80) (actual time=887.719..894.358 rows=4673 loops=1) Sort Key: q.date_time -> Hash Join (cost=102545.88..104342.94 rows=4058 width=80) (actual time=593.710..815.541 rows=4673 loops=1) Hash Cond: (("outer".car_id)::text = ("inner".car_id)::text) -> Index Scan using cars_car_id_btree_index on cars c (cost=0.00..506.87 rows=9401 width=37) (actual time=0.182..37.306 rows=9401 loops=1) -> Hash (cost=102491.73..102491.73 rows=4058 width=62) (actual time=593.040..593.040 rows=0 loops=1) -> Nested Loop (cost=0.00..102491.73 rows=4058 width=62) (actual time=146.647..551.975 rows=10285 loops=1) -> Index Scan using zip_zips_index on zips z (cost=0.00..1272.69 rows=516 width=52) (actual time=146.378..153.767 rows=718 loops=1) Filter: ((state)::text = 'WA'::text) -> Index Scan using zip_quotes_index on quotes q (cost=0.00..195.55 rows=48 width=27) (actual time=0.044..0.464 rows=14 loops=718) Index Cond: (("outer".zip)::text = (q.zip)::text) Total runtime: 898.438 ms Again a decrease in execution time. On the other hand: SET enable_hasdjoin=false; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------------------ Sort (cost=62829.86..62840.00 rows=4058 width=80) (actual time=11368.025..11374.629 rows=4673 loops=1) Sort Key: q.date_time -> Merge Join (cost=62006.97..62586.65 rows=4058 width=80) (actual time=11188.371..11295.156 rows=4673 loops=1) Merge Cond: (("outer".car_id)::text = "inner"."?column7?") -> Index Scan using cars_car_id_btree_index on cars c (cost=0.00..506.87 rows=9401 width=37) (actual time=0.167..37.728 rows=9401 loops=1) -> Sort (cost=62006.97..62017.12 rows=4058 width=62) (actual time=11187.581..11196.343 rows=4827 loops=1) Sort Key: (q.car_id)::text -> Merge Join (cost=60037.99..61763.76 rows=4058 width=62) (actual time=10893.572..10975.658 rows=10285 loops=1) Merge Cond: ("outer"."?column6?" = "inner"."?column4?") -> Sort (cost=1110.15..1111.44 rows=516 width=52) (actual time=86.679..87.166 rows=718 loops=1) Sort Key: (z.zip)::text -> Seq Scan on zips z (cost=0.00..1086.90 rows=516 width=52) (actual time=79.023..83.921 rows=718 loops=1) Filter: ((state)::text = 'WA'::text) -> Sort (cost=58927.84..59769.15 rows=336525 width=27) (actual time=9848.479..10319.275 rows=340426 loops=1) Sort Key: (q.zip)::text -> Seq Scan on quotes q (cost=0.00..10664.25 rows=336525 width=27) (actual time=0.227..2171.917 rows=340740 loops=1) Total runtime: 11408.120 ms Which really is not that surprising. And Finally: set enable_hashjoin=false; enable_seqscan=false; QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=103557.82..103567.97 rows=4058 width=80) (actual time=1206.168..1212.880 rows=4673 loops=1) Sort Key: q.date_time -> Merge Join (cost=102734.94..103314.61 rows=4058 width=80) (actual time=809.448..949.110 rows=4673 loops=1) Merge Cond: ("outer"."?column7?" = ("inner".car_id)::text) -> Sort (cost=102734.94..102745.08 rows=4058 width=62) (actual time=808.660..819.317 rows=4827 loops=1) Sort Key: (q.car_id)::text -> Nested Loop (cost=0.00..102491.73 rows=4058 width=62) (actual time=151.457..559.886 rows=10285 loops=1) -> Index Scan using zip_zips_index on zips z (cost=0.00..1272.69 rows=516 width=52) (actual time=151.179..158.375 rows=718 loops=1) Filter: ((state)::text = 'WA'::text) -> Index Scan using zip_quotes_index on quotes q (cost=0.00..195.55 rows=48 width=27) (actual time=0.042..0.455 rows=14 loops=718) Index Cond: (("outer".zip)::text = (q.zip)::text) -> Index Scan using cars_car_id_btree_index on cars c (cost=0.00..506.87 rows=9401 width=37) (actual time=0.213..47.307 rows=12019 loops=1) Total runtime: 1218.459 ms Anyway, thanks for the attention to this issue. And I hope that this helps some. Jared From pgsql-performance-owner@postgresql.org Tue Dec 9 15:57:27 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 19B5AD1B438 for ; Wed, 3 Dec 2003 18:30:24 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69540-10 for ; Wed, 3 Dec 2003 14:29:53 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 28EC2D1B432 for ; Wed, 3 Dec 2003 14:29:52 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ARbl2-0000uE-00; Wed, 03 Dec 2003 20:29:52 +0200 Date: Wed, 3 Dec 2003 20:29:52 +0200 To: pgsql-performance@postgresql.org Subject: Slow UPDATE, INSERT OK Message-ID: <20031203182952.GA3097@alcaron.ee> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/161 X-Sequence-Number: 5021 Hello! I am relative newcomer to SQL and PostgreSQL world, so please forgive me if this question is stupid. I am experiencing strange behaviour, where simple UPDATE of one field is very slow, compared to INSERT into table with multiple indexes. I have two tables - one with raw data records (about 24000), where one field contains status information (varchar(10)). First table has no indexes, only primary key (recid). Second table contains processed records - some fields are same as first table, others are calculated during processing. Records are processed by Python script, which uses PyPgSQL for PostgreSQL access. Processing is done by selecting all records from table1 where status matches certain criteria (import). Each record is processed and results are inserted into table2, after inserting status field on same record in table1 is updated with new value (done). Update statement itself is extremely simple: "update table1 set status = 'done' where recid = ..." Most interesting is, that insert takes 0.004 seconds in average, but update takes 0.255 seconds in average. Processing of 24000 records took around 1 hour 20 minutes. Then i changed processing logic not to update every record in table1 after processing. Instead i did insert recid value into temporary table and updated records in table1 after all records were processed and inserted into table2: UPDATE table1 SET Status = 'done' WHERE recid IN (SELECT recid FROM temptable) This way i got processing time of 24000 records down to about 16 minutes. About 13 minutes from this took last UPDATE statement. Why is UPDATE so slow compared to INSERT? I would expect more or less similar performance, or slower on insert since table2 has four indexes in addition to primary key, table1 has only primary key, which is used on update. Am i doing something wrong or is this normal? I am using PostgreSQL 7.3.4, Debian/GNU Linux 3.0 (Woody), kernel 2.4.21, Python 2.3.2, PyPgSQL 2.4 -- Ivar Zarans From pgsql-performance-owner@postgresql.org Wed Dec 3 15:24:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 48DBAD1B445 for ; Wed, 3 Dec 2003 19:24:06 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85877-03 for ; Wed, 3 Dec 2003 15:23:35 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 50678D1B446 for ; Wed, 3 Dec 2003 15:23:34 -0400 (AST) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 23E5236CB8; Wed, 3 Dec 2003 14:23:33 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1ARcaz-00021S-00; Wed, 03 Dec 2003 14:23:33 -0500 To: "scott.marlowe" Cc: , Subject: Re: Minimum hardware requirements for Postgresql db References: In-Reply-To: From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 03 Dec 2003 14:23:32 -0500 Message-ID: <87smk166yz.fsf@stark.dyndns.tv> Lines: 26 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 postgresql.org X-Archive-Number: 200312/50 X-Sequence-Number: 4910 "scott.marlowe" writes: > > 3) Estimated number of transactions to be written into the Postgresql db is > > around 15000 records per day. > > > > The growth rate in terms of number of connections is around 10% per year > > and the data retention is kept on average at least for 18 months for the 2 > > databases. > Like another poster pointed out, this is a walk in the park for > postgresql. My workstation (1.1GHz celeron, 40 gig IDE drive, 512 Meg > memory) could handle this load while still being my workstation. Well there's some info missing. Like what would you actually be _doing_ with these data? 15,000 inserts per day is nothing. But after 18 months that's over 5M records not including the 10% growth rate. 5M records isn't really all that much but it's enough that it's possible to write slow queries against it. If you're doing big batch updates or complex reports against the data that will be more interesting than the inserts. -- greg From pgsql-performance-owner@postgresql.org Wed Dec 3 15:32:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 87851D1B4E0 for ; Wed, 3 Dec 2003 19:32:06 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 87175-03 for ; Wed, 3 Dec 2003 15:31:35 -0400 (AST) Received: from localhost.localdomain (dsl093-038-087.pdx1.dsl.speakeasy.net [66.93.38.87]) by svr1.postgresql.org (Postfix) with ESMTP id 22D19D1BC5C for ; Wed, 3 Dec 2003 15:31:34 -0400 (AST) Received: from commandprompt.com (tacodog [127.0.0.1]) by localhost.localdomain (8.12.8/8.12.8) with ESMTP id hB3JVQFj027627 for ; Wed, 3 Dec 2003 11:31:27 -0800 Message-ID: <3FCE3A0E.8050206@commandprompt.com> Date: Wed, 03 Dec 2003 11:31:26 -0800 From: Al Hulaton User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031014 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Minimum hardware requirements for Postgresql db References: In-Reply-To: Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/51 X-Sequence-Number: 4911 > 1) We will be running 2 x Postgresql db in the machine. > > 2) Total number of connections to be around 100. The connections from the > clients machines will be in ODBC and socket connections. > > 3) Estimated number of transactions to be written into the Postgresql db is > around 15000 records per day. Assuming this server will be dedicated to PostgreSQL only, the needs outlined above are modest. As was pointed out in other posts, a simple sub-ghz machine with 512mb of ram is more than enough, but I'd slap on a gig only because RAM is cheaper now. If the database on this server is crucial, I'd look at setting up a UPS, RAID (at this level, even software-based RAID will do fine, RAID 5 preferably) and investing in a backup/replicator solution. -- Best, Al Hulaton | Sr. Account Engineer | Command Prompt, Inc. 503.667.4564 | ahulaton@commandprompt.com Home of Mammoth Replicator for PostgreSQL Managed PostgreSQL, Linux services and consulting Read and Search O'Reilly's 'Practical PostgreSQL' at http://www.commandprompt.com From pgsql-performance-owner@postgresql.org Wed Dec 3 15:33:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2D9FAD1B445 for ; Wed, 3 Dec 2003 19:33:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88160-01 for ; Wed, 3 Dec 2003 15:32:51 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 3681FD1B4D6 for ; Wed, 3 Dec 2003 15:32:50 -0400 (AST) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id E4EAF37372; Wed, 3 Dec 2003 14:32:50 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1ARcjy-00022F-00; Wed, 03 Dec 2003 14:32:50 -0500 To: Jared Carr Cc: Tom Lane , Greg Stark , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> <26699.1070423607@sss.pgh.pa.us> <2704.1070473746@sss.pgh.pa.us> <3FCE2B08.6090606@89glass.com> In-Reply-To: <3FCE2B08.6090606@89glass.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 03 Dec 2003 14:32:50 -0500 Message-ID: <87k75d66jh.fsf@stark.dyndns.tv> Lines: 17 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 postgresql.org X-Archive-Number: 200312/52 X-Sequence-Number: 4912 Jared Carr writes: > The patch definitely makes things more consistent...unfortunately it is more > consistent toward the slower execution times. Of course I am looking at this > simply from a straight performance standpoint and not a viewpoint of what > *should* be happening. At any rate here are the query plans with the various > settings. The optimizer seems to be at least considering reasonable plans now. It seems from the estimates that you need to rerun analyze. You might try "vacuum full analyze" to be sure. Also, you might try raising effective_cache_size and/or lowering random_page_size (it looks like something around 2 might help). -- greg From pgsql-performance-owner@postgresql.org Wed Dec 3 17:23:35 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EB37FD1B470 for ; Wed, 3 Dec 2003 21:23:33 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04772-02 for ; Wed, 3 Dec 2003 17:23:03 -0400 (AST) Received: from web.opentransfer.com (unknown [69.49.227.40]) by svr1.postgresql.org (Postfix) with ESMTP id D2511D1D0AE for ; Wed, 3 Dec 2003 17:22:45 -0400 (AST) Received: (from httpd@localhost) by web.opentransfer.com (8.11.6/8.11.6) id hB3LMRX17342 for pgsql-performance@postgresql.org; Wed, 3 Dec 2003 15:22:27 -0600 X-Authentication-Warning: web.opentransfer.com: httpd set sender to erik@norvelle.net using -f Received: from 212.21.224.151 ( [212.21.224.151]) as user erik@norvelle.net@69.49.238.2 by 69.49.227.23 with HTTP; Wed, 3 Dec 2003 15:22:27 -0600 Message-ID: <1070486547.3fce54135a65d@69.49.227.23> Date: Wed, 3 Dec 2003 15:22:27 -0600 From: erik@norvelle.net To: pgsql-performance@postgresql.org Subject: Re: Update performance ... Recommended configuration changes? MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit User-Agent: Internet Messaging Program (IMP) 3.1 X-Originating-IP: 212.21.224.151 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/53 X-Sequence-Number: 4913 Thanks to Greg Stark, Tom Lane and Stephan Szabo for their advice on rewriting my query... the revised query plan claims it should only take about half the time my original query did. Now for a somewhat different question: How might I improve my DB performance by adjusting the various parameters in postgresql.conf and kernel config? Again, TKA. Here's what I've currently got (hardware, kernel config. and postgresql.conf) Hardware: Mac iBook, G3 900Mhz, 640MB memory (This is my research machine :p ) OS: OS X 10.2.6 Postgresql version: 7.3.2 Kernel Config: sysctl -w kern.sysv.shmmax=4194304 sysctl -w kern.sysv.shmmin=1 sysctl -w kern.sysv.shmmni=32 sysctl -w kern.sysv.shmseg=8 sysctl -w kern.sysv.shmall=1024 ========================= Snip of postgresql.conf ================= # # Shared Memory Size # shared_buffers = 128 # min max_connections*2 or 16, 8KB each max_fsm_relations = 2000 # min 10, fsm is free space map, ~40 bytes max_fsm_pages = 20000 # min 1000, fsm is free space map, ~6 bytes max_locks_per_transaction = 128 # min 10 wal_buffers = 16 # min 4, typically 8KB each # # Non-shared Memory Sizes # sort_mem = 65535 # min 64, size in KB vacuum_mem = 8192 # min 1024, size in KB # # Write-ahead log (WAL) # #checkpoint_segments = 3 # in logfile segments, min 1, 16MB each #checkpoint_timeout = 300 # range 30-3600, in seconds # #commit_delay = 0 # range 0-100000, in microseconds #commit_siblings = 5 # range 1-1000 # fsync = false #wal_sync_method = fsync # the default varies across platforms: # # fsync, fdatasync, open_sync, or open_datasync #wal_debug = 0 # range 0-16 ========================== End Snip ======================= Saludos, Erik Norvelle From pgsql-performance-owner@postgresql.org Wed Dec 3 17:33:42 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1C9C9D1B4B3 for ; Wed, 3 Dec 2003 21:33:40 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03934-09 for ; Wed, 3 Dec 2003 17:33:10 -0400 (AST) Received: from mail.hive.nj2.inquent.com (mc.carriermail.com [205.178.180.9]) by svr1.postgresql.org (Postfix) with SMTP id 5A272D1D306 for ; Wed, 3 Dec 2003 17:32:51 -0400 (AST) Received: (qmail 2744 invoked from network); 3 Dec 2003 21:33:22 -0000 Received: from unknown (HELO ?192.168.1.199?) (134.22.70.148) by 205.178.180.9 with SMTP; 3 Dec 2003 21:33:22 -0000 Subject: Re: Update performance ... Recommended configuration From: Rod Taylor To: erik@norvelle.net Cc: Postgresql Performance In-Reply-To: <1070486547.3fce54135a65d@69.49.227.23> References: <1070486547.3fce54135a65d@69.49.227.23> Content-Type: text/plain Message-Id: <1070487190.92431.57.camel@jester> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Wed, 03 Dec 2003 16:33:10 -0500 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/54 X-Sequence-Number: 4914 > shared_buffers = 128 # min max_connections*2 or 16, 8KB each Try 1500. > sort_mem = 65535 # min 64, size in KB I'd pull this in. You only have 640MB ram, which means about 8 large sorts to swap. How about 16000? > fsync = false I presume you understand the risks involved with this setting and dataloss. From pgsql-performance-owner@postgresql.org Wed Dec 3 17:41:18 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 49BA0D1CCCF for ; Wed, 3 Dec 2003 21:41:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06226-06 for ; Wed, 3 Dec 2003 17:40:48 -0400 (AST) Received: from swan.mail.pas.earthlink.net (swan.mail.pas.earthlink.net [207.217.120.123]) by svr1.postgresql.org (Postfix) with ESMTP id 57283D1BC5C for ; Wed, 3 Dec 2003 17:40:46 -0400 (AST) Received: from dpc6682164065.direcpc.com ([66.82.164.65] helo=earthlink.net) by swan.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1AReji-0002nu-00 for pgsql-performance@postgresql.org; Wed, 03 Dec 2003 13:40:44 -0800 Message-ID: <3FCE5855.5040403@earthlink.net> Date: Wed, 03 Dec 2003 16:40:37 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Subject: Has anyone run on the new G5 yet Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/55 X-Sequence-Number: 4915 To all, We are building a data warehouse composed of essentially click stream data. The DB is growing fairly quickly as to be expected, currently at 90GB for one months data. The idea is to keep 6 months detailed data on line and then start aggregating older data to summary tables. We have 2 fact tables currently, one with about 68 million rows and the other with about 210 million rows. Numerous dimension tables ranging from a dozen rows to millions. We are currently running on a Dell 2650 with 2 Xeon 2.8 processors in hyper-threading mode, 4GB of ram, and 5 SCSI drives in a RAID 0, Adaptec PERC3/Di, configuration. I believe they are 10k drives. Files system is EXT3. We are running RH9 Linux kernel 2.4.20-20.9SMP with bigmem turned on. This box is used only for the warehouse. All the ETL work is done on this machine as well. DB version is postgreSQL 7.4. We are running into issues with IO saturation obviously. Since this thing is only going to get bigger we are looking for some advice on how to accommodate DB's of this size. First question is do we gain anything by moving the RH Enterprise version of Linux in terms of performance, mainly in the IO realm as we are not CPU bound at all? Second and more radical, has anyone run postgreSQL on the new Apple G5 with an XRaid system? This seems like a great value combination. Fast CPU, wide bus, Fibre Channel IO, 2.5TB all for ~17k. I keep see references to terabyte postgreSQL installations, I was wondering if anyone on this list is in charge of one of those and can offer some advice on how to position ourselves hardware wise. Thanks. --sean From pgsql-performance-owner@postgresql.org Wed Dec 3 17:54:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 83918D1B486 for ; Wed, 3 Dec 2003 21:54:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 12089-01 for ; Wed, 3 Dec 2003 17:54:26 -0400 (AST) Received: from swan.mail.pas.earthlink.net (swan.mail.pas.earthlink.net [207.217.120.123]) by svr1.postgresql.org (Postfix) with ESMTP id 9F683D1B432 for ; Wed, 3 Dec 2003 17:54:24 -0400 (AST) Received: from dpc6682164065.direcpc.com ([66.82.164.65] helo=earthlink.net) by swan.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1AReww-0006mH-00; Wed, 03 Dec 2003 13:54:25 -0800 Message-ID: <3FCE5B85.9000406@earthlink.net> Date: Wed, 03 Dec 2003 16:54:13 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Cc: Sean Shanny Subject: Re: Has anyone run on the new G5 yet References: <3FCE5855.5040403@earthlink.net> In-Reply-To: <3FCE5855.5040403@earthlink.net> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/56 X-Sequence-Number: 4916 I should also add that we have already done a ton of tuning based on the archives of this list so we are not starting from scratch here. Thanks. --sean Sean Shanny wrote: > To all, > > We are building a data warehouse composed of essentially click stream > data. The DB is growing fairly quickly as to be expected, currently > at 90GB for one months data. The idea is to keep 6 months detailed > data on line and then start aggregating older data to summary tables. > We have 2 fact tables currently, one with about 68 million rows and > the other with about 210 million rows. Numerous dimension tables > ranging from a dozen rows to millions. > > We are currently running on a Dell 2650 with 2 Xeon 2.8 processors in > hyper-threading mode, 4GB of ram, and 5 SCSI drives in a RAID 0, > Adaptec PERC3/Di, configuration. I believe they are 10k drives. > Files system is EXT3. We are running RH9 Linux kernel 2.4.20-20.9SMP > with bigmem turned on. This box is used only for the warehouse. All > the ETL work is done on this machine as well. DB version is > postgreSQL 7.4. > > We are running into issues with IO saturation obviously. Since this > thing is only going to get bigger we are looking for some advice on > how to accommodate DB's of this size. > > First question is do we gain anything by moving the RH Enterprise > version of Linux in terms of performance, mainly in the IO realm as we > are not CPU bound at all? Second and more radical, has anyone run > postgreSQL on the new Apple G5 with an XRaid system? This seems like > a great value combination. Fast CPU, wide bus, Fibre Channel IO, > 2.5TB all for ~17k. > > I keep see references to terabyte postgreSQL installations, I was > wondering if anyone on this list is in charge of one of those and can > offer some advice on how to position ourselves hardware wise. > > Thanks. > > --sean > > > ---------------------------(end of broadcast)--------------------------- > TIP 3: 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 Wed Dec 3 18:13:01 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DD708D1B554 for ; Wed, 3 Dec 2003 22:12:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11205-04 for ; Wed, 3 Dec 2003 18:12:29 -0400 (AST) Received: from main.wiredfool.com (dsl231-055-035.sea1.dsl.speakeasy.net [216.231.55.35]) by svr1.postgresql.org (Postfix) with ESMTP id 995E8D1B432 for ; Wed, 3 Dec 2003 18:12:11 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 9CF302728E; Wed, 3 Dec 2003 14:07:10 -0800 (PST) Received: from main.wiredfool.com ([127.0.0.1]) by localhost (cabbage [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 06031-08; Wed, 3 Dec 2003 14:07:08 -0800 (PST) Received: from [IPv6:::1] (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 832EB2728D; Wed, 3 Dec 2003 14:07:08 -0800 (PST) In-Reply-To: <3FCE5855.5040403@earthlink.net> References: <3FCE5855.5040403@earthlink.net> Mime-Version: 1.0 (Apple Message framework v606) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit Cc: "pgsql-performance@postgresql.org" From: Eric Soroos Subject: Re: Has anyone run on the new G5 yet Date: Wed, 3 Dec 2003 14:12:08 -0800 To: Sean Shanny X-Mailer: Apple Mail (2.606) X-Virus-Scanned: by amavisd-new-20030616-p3 (Debian) at main.wiredfool.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/57 X-Sequence-Number: 4917 Sean > Second and more radical, has anyone run postgreSQL on the new Apple > G5 with an XRaid system? This seems like a great value combination. > Fast CPU, wide bus, Fibre Channel IO, 2.5TB all for ~17k. > > I keep see references to terabyte postgreSQL installations, I was > wondering if anyone on this list is in charge of one of those and can > offer some advice on how to position ourselves hardware wise. From my (admittedly low end) OSX experience, you just don't have the filesystem options on OSX that you have on linux, from the noatime mount, filesystem types, and the raid options. I also feel that the software stack is a bit more mature and tested on the linux side of things. I doubt that the g5 hardware is that much faster than what you have right now. The raid hardware might be a good deal for you even on a linux platform. There are reports of it 'just working' with x86 linux hardware. eric From pgsql-performance-owner@postgresql.org Wed Dec 3 18:32:57 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BF5EDD1B486 for ; Wed, 3 Dec 2003 22:32:49 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19449-01 for ; Wed, 3 Dec 2003 18:32:20 -0400 (AST) Received: from bramble.mmrd.com (unknown [65.217.53.66]) by svr1.postgresql.org (Postfix) with ESMTP id 6D3D6D1B479 for ; Wed, 3 Dec 2003 18:32:17 -0400 (AST) Received: from thorn.mmrd.com (thorn.mmrd.com [172.25.10.100]) by bramble.mmrd.com (8.12.8/8.12.8) with ESMTP id hB3Lg7cM016428 for ; Wed, 3 Dec 2003 16:42:07 -0500 Received: from gnvex001.mmrd.com (gnvex001.mmrd.com [192.168.3.55]) by thorn.mmrd.com (8.11.6/8.11.6) with ESMTP id hB3MWJl14844 for ; Wed, 3 Dec 2003 17:32:19 -0500 Received: from camel.mmrd.com ([172.25.5.213]) by gnvex001.mmrd.com with SMTP (Microsoft Exchange Internet Mail Service Version 5.5.2657.72) id XT87Z8D2; Wed, 3 Dec 2003 17:32:16 -0500 Subject: sequence overhead From: Robert Treat To: pgsql-performance@postgresql.org Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 Date: 03 Dec 2003 17:32:18 -0500 Message-Id: <1070490739.25244.10530.camel@camel> Mime-Version: 1.0 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/58 X-Sequence-Number: 4918 Just wondering if anyone has done any testing on the amount of overhead for insert you might gain by adding a serial column to a table. I'm thinking of adding a few to some tables that get an average of 30 - 40 inserts per second, sometimes bursting over 100 inserts per second and wondering if there will be any noticeable impact. Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Wed Dec 3 18:35:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B485ED1B518 for ; Wed, 3 Dec 2003 22:35:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 10759-08 for ; Wed, 3 Dec 2003 18:35:16 -0400 (AST) Received: from redhotpenguin.com (c-24-7-81-19.client.comcast.net [24.7.81.19]) by svr1.postgresql.org (Postfix) with ESMTP id 38D99D1B486 for ; Wed, 3 Dec 2003 18:35:13 -0400 (AST) Received: (qmail 14413 invoked by uid 506); 3 Dec 2003 22:35:14 -0000 Received: from fred@redhotpenguin.com by poster.redhotpenguin.com by uid 502 with qmail-scanner-1.16 (clamscan: 0.65. spamassassin: 2.44. Clear:SA:0(-1.1/5.0):. Processed in 1.006262 secs); 03 Dec 2003 22:35:14 -0000 Received: from c-24-7-81-19.client.comcast.net (HELO harpua) (fred@redhotpenguin.com@24.7.81.19) by 192.168.0.2 with RC4-MD5 encrypted SMTP; 3 Dec 2003 22:35:13 -0000 From: "Fred Moyer" To: "'Sean Shanny'" , Subject: Re: Has anyone run on the new G5 yet Date: Wed, 3 Dec 2003 14:35:45 -0800 Message-ID: <000001c3b9ed$cb1bf2a0$0300a8c0@harpua> 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, Build 10.0.2616 In-Reply-To: <3FCE5B85.9000406@earthlink.net> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/59 X-Sequence-Number: 4919 > We are running into issues with IO saturation obviously. Since this > thing is only going to get bigger we are looking for some advice on > how to accommodate DB's of this size. > Second and more radical, has anyone run > postgreSQL on the new Apple G5 with an XRaid system? This seems like > a great value combination. Fast CPU, wide bus, Fibre Channel IO, > 2.5TB all for ~17k. If you are going for I/O performance you are best off with one of the Xserve competitors listed at http://www.apple.com/xserve/raid/. The Xserve is based on IDE drives which have a lower seek time (say 8.9 ms) compared to scsi (3.6 ms for seagate cheetah). For small random read/write operations (like databases) this will give you a noticable improvement in performance over ide drives. Also make sure to get as many drives as possible, more spindles equals better I/O performance. > I keep see references to terabyte postgreSQL installations, I was > wondering if anyone on this list is in charge of one of those and can > offer some advice on how to position ourselves hardware wise. I've gone to about half terabyte size and all I can say is you should plan for at least one quarter to one half a rack of drivespace (assuming 14 drives per 4u that's 42 to 84 drives). Do yourself a favor and get more rather than less, you will really appreciate it. I averaged about 2 mb/s average per drive via the raid controller stats on 14 drive array during I/O bound seek and update operations in 2 raid 10 arrays (half xlogs and half data). That comes out to around 2 hours for a terabyte with 70 drives assuming a constant scaling. You may be able to get more or less depending on your setup and query workload. > Thanks. > > --sean > > > ---------------------------(end of > broadcast)--------------------------- > TIP 3: 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 > ---------------------------(end of broadcast)--------------------------- TIP 5: Have you checked our extensive FAQ? http://www.postgresql.org/docs/faqs/FAQ.html From pgsql-performance-owner@postgresql.org Wed Dec 3 19:18:23 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 769D3D1B47F for ; Wed, 3 Dec 2003 23:18:19 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 22389-07 for ; Wed, 3 Dec 2003 19:17:50 -0400 (AST) Received: from candle.pha.pa.us (unknown [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id D5A41D1B479 for ; Wed, 3 Dec 2003 19:17:46 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hB3NHbY27219; Wed, 3 Dec 2003 18:17:37 -0500 (EST) From: Bruce Momjian Message-Id: <200312032317.hB3NHbY27219@candle.pha.pa.us> Subject: Re: A question on the query planner In-Reply-To: <26699.1070423607@sss.pgh.pa.us> To: Tom Lane Date: Wed, 3 Dec 2003 18:17:37 -0500 (EST) Cc: Greg Stark , Jared Carr , pgsql-performance@postgresql.org X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/60 X-Sequence-Number: 4920 Tom Lane wrote: > Greg Stark writes: > > Tom Lane writes: > >> Define "no longer works well". > > > Well it seems to completely bar the use of a straight merge join between two > > index scans: > > Hmmm ... [squints] ... it's not supposed to do that ... [digs] ... yeah, > there's something busted here. Will get back to you ... LOL, but I am not sure why. :-) -- 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 Wed Dec 3 19:22:54 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 37422D1B47E for ; Wed, 3 Dec 2003 23:22:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21780-07 for ; Wed, 3 Dec 2003 19:22:23 -0400 (AST) Received: from live (64-35-136-226.gohighspeed.com [64.35.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id 2B7DAD1B446 for ; Wed, 3 Dec 2003 19:22:20 -0400 (AST) Received: from [192.168.254.101] (helo=89glass.com) by live with esmtp (Exim 3.35 #1 (Debian)) id 1ARgK1-0002hW-00; Wed, 03 Dec 2003 15:22:17 -0800 Message-ID: <3FCE700E.5070206@89glass.com> Date: Wed, 03 Dec 2003 15:21:50 -0800 From: Jared Carr User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Greg Stark Cc: Tom Lane , pgsql-performance@postgresql.org Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> <26699.1070423607@sss.pgh.pa.us> <2704.1070473746@sss.pgh.pa.us> <3FCE2B08.6090606@89glass.com> <87k75d66jh.fsf@stark.dyndns.tv> In-Reply-To: <87k75d66jh.fsf@stark.dyndns.tv> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/61 X-Sequence-Number: 4921 Greg Stark wrote: >Jared Carr writes: > > > >>The patch definitely makes things more consistent...unfortunately it is more >>consistent toward the slower execution times. Of course I am looking at this >>simply from a straight performance standpoint and not a viewpoint of what >>*should* be happening. At any rate here are the query plans with the various >>settings. >> >> > >The optimizer seems to be at least considering reasonable plans now. It seems >from the estimates that you need to rerun analyze. You might try "vacuum full >analyze" to be sure. > >Also, you might try raising effective_cache_size and/or lowering >random_page_size (it looks like something around 2 might help). > > > Yep, I had forgotten to run vacuum since I had patched it :P. The overall performance is definitely better, I will go ahead and tweak the server settings and see what I can get. Thanks again for all the help. From pgsql-performance-owner@postgresql.org Wed Dec 3 19:57:49 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 23741D1B453 for ; Wed, 3 Dec 2003 23:57:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 29740-03 for ; Wed, 3 Dec 2003 19:57:17 -0400 (AST) Received: from sardegna.mbigroup.it (unknown [151.8.40.3]) by svr1.postgresql.org (Postfix) with ESMTP id 7AD92D1B465 for ; Wed, 3 Dec 2003 19:57:13 -0400 (AST) Received: from sicilia.pisa.mbigroup.it (localhost [127.0.0.1]) by sardegna.mbigroup.it (8.11.6/8.8.7) with ESMTP id hB3Nobx25281; Thu, 4 Dec 2003 00:50:37 +0100 Received: from bigfoot.com (canarie.pisa.mbigroup.it [192.168.9.30]) by sicilia.pisa.mbigroup.it (8.11.6/8.11.6) with ESMTP id hB3NoV302709; Thu, 4 Dec 2003 00:50:31 +0100 Message-ID: <3FCE76B3.5040502@bigfoot.com> Date: Thu, 04 Dec 2003 00:50:11 +0100 From: Gaetano Mendola User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031202 Thunderbird/0.4RC1 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Cc: Sean Shanny Subject: Re: Has anyone run on the new G5 yet References: <3FCE5855.5040403@earthlink.net> In-Reply-To: <3FCE5855.5040403@earthlink.net> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/63 X-Sequence-Number: 4923 Sean Shanny wrote: > We are currently running on a Dell 2650 with 2 Xeon 2.8 processors in > hyper-threading mode, 4GB of ram, and 5 SCSI drives in a RAID 0, Adaptec > PERC3/Di, configuration. I believe they are 10k drives. Files system > is EXT3. We are running RH9 Linux kernel 2.4.20-20.9SMP with bigmem > turned on. This box is used only for the warehouse. All the ETL work > is done on this machine as well. DB version is postgreSQL 7.4. Are you experiencing improvment using the hyper-threading ? Regards Gaetano Mendola From pgsql-performance-owner@postgresql.org Wed Dec 3 19:56:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 71D31D1B455 for ; Wed, 3 Dec 2003 23:56:04 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28098-03 for ; Wed, 3 Dec 2003 19:55:36 -0400 (AST) Received: from sardegna.mbigroup.it (unknown [151.8.40.3]) by svr1.postgresql.org (Postfix) with ESMTP id EB67AD1B479 for ; Wed, 3 Dec 2003 19:55:32 -0400 (AST) Received: from sicilia.pisa.mbigroup.it (localhost [127.0.0.1]) by sardegna.mbigroup.it (8.11.6/8.8.7) with ESMTP id hB3NtPp25341; Thu, 4 Dec 2003 00:55:25 +0100 Received: from bigfoot.com (canarie.pisa.mbigroup.it [192.168.9.30]) by sicilia.pisa.mbigroup.it (8.11.6/8.11.6) with ESMTP id hB3NtJ303048; Thu, 4 Dec 2003 00:55:19 +0100 Message-ID: <3FCE77D3.3010704@bigfoot.com> Date: Thu, 04 Dec 2003 00:54:59 +0100 From: Gaetano Mendola User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031202 Thunderbird/0.4RC1 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Cc: Tom Lane Subject: Re: A question on the query planner References: <3FCBB639.10409@89glass.com> <1070385381.24915.8650.camel@camel> <3FCCD225.8030609@89glass.com> <87d6b79har.fsf@stark.dyndns.tv> <3FCCE48B.10406@89glass.com> <871xrn9e09.fsf@stark.dyndns.tv> <3FCCF61E.90307@89glass.com> <87ptf697h0.fsf@stark.dyndns.tv> <25476.1070409047@sss.pgh.pa.us> <87znea7kis.fsf@stark.dyndns.tv> <26699.1070423607@sss.pgh.pa.us> <2704.1070473746@sss.pgh.pa.us> In-Reply-To: <2704.1070473746@sss.pgh.pa.us> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/62 X-Sequence-Number: 4922 Tom Lane wrote: >>Hmmm ... [squints] ... it's not supposed to do that ... > > > The attached patch seems to make it better. I guess is too late for 7.3.5. :-( Any chance for 7.4.1 ? Regards Gaetano Mendola From pgsql-performance-owner@postgresql.org Wed Dec 3 20:25:12 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A9D95D1B453 for ; Thu, 4 Dec 2003 00:25:09 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 32760-01 for ; Wed, 3 Dec 2003 20:24:40 -0400 (AST) Received: from swan.mail.pas.earthlink.net (swan.mail.pas.earthlink.net [207.217.120.123]) by svr1.postgresql.org (Postfix) with ESMTP id BEF29D1B441 for ; Wed, 3 Dec 2003 20:24:37 -0400 (AST) Received: from dpc6682164065.direcpc.com ([66.82.164.65] helo=earthlink.net) by swan.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1ARhIG-0002BZ-00; Wed, 03 Dec 2003 16:24:35 -0800 Message-ID: <3FCE7EB5.7030809@earthlink.net> Date: Wed, 03 Dec 2003 19:24:21 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Gaetano Mendola Cc: "pgsql-performance@postgresql.org" Subject: Re: Has anyone run on the new G5 yet References: <3FCE5855.5040403@earthlink.net> <3FCE76B3.5040502@bigfoot.com> In-Reply-To: <3FCE76B3.5040502@bigfoot.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/64 X-Sequence-Number: 4924 Gaetano, I don't believe we have ever run the system without it turned on. Another switch to fiddle with. :-) --sean Gaetano Mendola wrote: > Sean Shanny wrote: > >> We are currently running on a Dell 2650 with 2 Xeon 2.8 processors in >> hyper-threading mode, 4GB of ram, and 5 SCSI drives in a RAID 0, >> Adaptec PERC3/Di, configuration. I believe they are 10k drives. >> Files system is EXT3. We are running RH9 Linux kernel 2.4.20-20.9SMP >> with bigmem turned on. This box is used only for the warehouse. All >> the ETL work is done on this machine as well. DB version is >> postgreSQL 7.4. > > > Are you experiencing improvment using the hyper-threading ? > > > Regards > Gaetano Mendola > > From pgsql-performance-owner@postgresql.org Wed Dec 3 23:10:51 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 048FCD1C979 for ; Thu, 4 Dec 2003 03:10:48 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53790-04 for ; Wed, 3 Dec 2003 23:10:20 -0400 (AST) Received: from necsin-fw.nec.com.sg (necsin-fw.nec.com.sg [203.127.255.1]) by svr1.postgresql.org (Postfix) with ESMTP id B23BCD1B477 for ; Wed, 3 Dec 2003 23:10:09 -0400 (AST) Received: from kato.nec.com.sg (kato.nec.com.sg [203.127.254.2]) by necsin-fw.nec.com.sg with ESMTP id hB439srE006979; Thu, 4 Dec 2003 11:09:54 +0800 (SGT) Received: from necsind1.nec.com.sg ([203.127.252.220]) by kato.nec.com.sg with ESMTP id hB439ru2025635; Thu, 4 Dec 2003 11:09:54 +0800 (SGT) Subject: Re: Minimum hardware requirements for Postgresql db To: Christopher Browne , pgsql-performance@postgresql.org X-Mailer: Lotus Notes Release 5.0.5 September 22, 2000 Message-ID: From: CHEWTC@ap.nec.com.sg Date: Thu, 4 Dec 2003 11:09:59 +0800 X-MIMETrack: Serialize by Router on NECSIND1/WTC/NECSIN(Release 5.0.12 |February 13, 2003) at 12/04/2003 11:10:00 AM MIME-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/65 X-Sequence-Number: 4925 Dear all Sorry for my mistake on the 15000 recs per day. In fact, this server is planned as a OLTP database server for a retailer. Our intention is either to setup 1 or 2 Postgresql db in the server. The proper sizing info for the 1st Postgresql db should be: No. of item master : 200,000 (This item master grows at 0.5% daily). No. of transactions from Point-of-Sales machines: 25,000 Plus other tables, the total sizing that I estimated is 590,000 records daily. The 2nd Postgresql db will be used by end users on client machines linked via ODBC, doing manual data entry. This will house the item master, loyalty card master and other Historical data to be kept for at least 1.5 years. Therefore total sizing for this db is around 165,000,000 recs at any time. In summary, the single machine must be able to take up around 100 users connections via both socket and ODBC. And house the above number of records. Thank you, REgards. Christopher Browne To: pgsql-performance@postgresql.org Sent by: cc: pgsql-performance-owner@pos Subject: Re: [PERFORM] Minimum hardware requirements for Postgresql db tgresql.org 03/12/2003 12:44 PM After takin a swig o' Arrakan spice grog, CHEWTC@ap.nec.com.sg belched out: > We would be recommending to our ct. on the use of Postgresql db as > compared to MS SQL Server. We are targetting to use Redhat Linux ES > v2.1, Postgresql v7.3.4 and Postgresql ODBC 07.03.0100. > > We would like to know the minimum specs required for our below > target. The minimum specs is referring to no. of CPU, memory, > harddisk capacity, RAID technology etc. And also the Postgresql > parameters and configuration to run such a system. > > 1) We will be running 2 x Postgresql db in the machine. > > 2) Total number of connections to be around 100. The connections > from the clients machines will be in ODBC and socket connections. > > 3) Estimated number of transactions to be written into the > Postgresql db is around 15000 records per day. > > The growth rate in terms of number of connections is around 10% per > year and the data retention is kept on average at least for 18 > months for the 2 databases. > > Are there any reference books or sites that I can tap on for the > above requirement? Perhaps the best reference on detailed performance information is the "General Bits" documents. These don't point particularly at minimal hardware requirements, but rather at how to configure the DBMS to best reflect what hardware you have. But there's some degree to which you can work backwards... If you'll need to support 100 concurrent connections, then minimum shared_buffers is 200, which implies 1600K of RAM required for shared buffers. 100 connections probably implies around 100MB of memory for the backend processes to support the connections. That all points to the notion that you'd more than probably get half-decent performance if you had a mere 256MB of RAM, which is about $50 worth these days. None of it sounds terribly challenging; 15K records per day is 625 records per hour which represents an INSERT every 6 seconds. Even if that has to fit into an 8 hour day, that's still not a high number of transactions per second. That _sounds like_ an application that could work on old, obsolete hardware. I would imagine that my old Intel Pentium Pro 200 might cope with the load, in much the way that that server is more than capable of supporting a web server that would serve a local workgroup. (I only have 64MB of RAM on that box, which would be a mite low, but it's an _ancient_ server...) The only thing that makes me a little suspicious that there's something funny about the prescription is your indication of having 100 concurrent users, which is really rather heavyweight in comparison with the comparatively tiny number of transactions. Is this for some sort of "departmental application"? Where there's a lot of manual data entry, so that each user would generate a transaction every 3-4 minutes? That actually sounds about right... Let me suggest that the "cost driver" in this will _not_ be the cost of the hardware to support the database itself; it will instead be in having redundant hardware and backup hardware to ensure reliability. It would seem likely that just about any sort of modern hardware would be pretty adequate to the task. You can hardly _buy_ a system with less than Gigahertz-speed CPUs, 40GB of disk, and 256MB of RAM. Upgrade to have 2 SCSI disks, 512MB (or more, which is better) of RAM, and the cost of a suitable system still won't be outrageous. Double it, buying a standby server, and the cost still oughtn't be real scary. And if the application is important, you _should_ have a standby server, irrespective of what software you might be running. -- (reverse (concatenate 'string "moc.enworbbc" "@" "enworbbc")) http://www3.sympatico.ca/cbbrowne/x.html Rules of the Evil Overlord #199. "I will not make alliances with those more powerful than myself. Such a person would only double-cross me in my moment of glory. I will make alliances with those less powerful than myself. I will then double-cross them in their moment of glory." ---------------------------(end of broadcast)--------------------------- TIP 9: 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 Dec 4 01:41:40 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0BF8FD1B4A7 for ; Thu, 4 Dec 2003 05:41:35 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73588-04 for ; Thu, 4 Dec 2003 01:41:04 -0400 (AST) Received: from mta11.srv.hcvlny.cv.net (mta11.srv.hcvlny.cv.net [167.206.5.86]) by svr1.postgresql.org (Postfix) with ESMTP id 03288D1B579 for ; Thu, 4 Dec 2003 01:41:03 -0400 (AST) Received: from [192.168.1.10] (ool-4352919e.dyn.optonline.net [67.82.145.158]) by mta11.srv.hcvlny.cv.net (iPlanet Messaging Server 5.2 HotFix 1.16 (built May 14 2003)) with ESMTP id <0HPC00JY4VOM7B@mta11.srv.hcvlny.cv.net> for pgsql-performance@postgresql.org; Thu, 04 Dec 2003 00:38:47 -0500 (EST) Date: Thu, 04 Dec 2003 00:38:46 -0500 From: "Matthew T. O'Connor" Subject: Re: autovacuum daemon stops doing work after about an hour In-reply-to: <16332.63508.196365.801418@yertle.int.kciLink.com> To: Vivek Khera Cc: pgsql-performance@postgresql.org Message-id: <1070516326.1455.11.camel@zedora.zeut.net> MIME-version: 1.0 X-Mailer: Ximian Evolution 1.4.5 (1.4.5-7) Content-type: text/plain Content-transfer-encoding: 7BIT References: <16332.63508.196365.801418@yertle.int.kciLink.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/66 X-Sequence-Number: 4926 On Tue, 2003-12-02 at 15:37, Vivek Khera wrote: > Now I'm trying to implement pg_autovacuum. It seems to work ok, but > after about an hour or so, it does nothing. The process still is > running, but nothing is sent to the log file. > > I'm running the daemon as distributed with PG 7.4 release as follows: > > pg_autovacuum -d4 -V 0.15 -A 1 -U postgres -L /var/tmp/autovacuum.log -D > > the last few lines of the log are: > > [2003-12-02 11:43:58 AM] VACUUM ANALYZE "public"."msg_recipients" > [2003-12-02 12:24:33 PM] select relfilenode,reltuples,relpages from pg_class where relfilenode=18588239 > [2003-12-02 12:24:33 PM] table name: vkmlm."public"."msg_recipients" > [2003-12-02 12:24:33 PM] relfilenode: 18588239; relisshared: 0 > [2003-12-02 12:24:33 PM] reltuples: 9; relpages: 529132 > [2003-12-02 12:24:33 PM] curr_analyze_count: 1961488; cur_delete_count: 1005040 > [2003-12-02 12:24:33 PM] ins_at_last_analyze: 1961488; del_at_last_vacuum: 1005040 > [2003-12-02 12:24:33 PM] insert_threshold: 509; delete_threshold 1001 > [2003-12-02 12:24:33 PM] Performing: VACUUM ANALYZE "public"."user_list" > [2003-12-02 12:24:33 PM] VACUUM ANALYZE "public"."user_list" > [2003-12-02 12:43:19 PM] select relfilenode,reltuples,relpages from pg_class where relfilenode=18588202 > [2003-12-02 12:43:19 PM] table name: vkmlm."public"."user_list" > [2003-12-02 12:43:19 PM] relfilenode: 18588202; relisshared: 0 > [2003-12-02 12:43:19 PM] reltuples: 9; relpages: 391988 > [2003-12-02 12:43:19 PM] curr_analyze_count: 1159843; cur_delete_count: 1118540 > [2003-12-02 12:43:19 PM] ins_at_last_analyze: 1159843; del_at_last_vacuum: 1118540 > [2003-12-02 12:43:19 PM] insert_threshold: 509; delete_threshold 1001 > > Then it just sits there. I started it at 11:35am, and it is now > 3:30pm. Weird.... Alphabetically speaking, is vkmlm."public"."user_list" be the last table in the last schema in the last database? You are running with -d4, so you would get a message about going to sleep shortly after dealing with the last table, but you didn't get the sleep message, so I don't think the problem is that pg_autovacuum is sleeping for an inordinate amount time. > I did the same last night at about 10:58pm, and it ran and did work until > 11:57pm, then sat there until I killed/restarted pg_autovacuum this > morning at 11:35. The process is not using any CPU time. > > I just killed/restarted it and it found work to do on my busy tables > which I'd expect. when you kill it, do you get a core file? Could you do a backtrace and see where pg_autovacuum is hung up? > I'm running Postgres 7.4 release on FreeBSD 4.9-RELEASE. I don't run FreeBSD, so I haven't tested with FreeBSD. Recently Craig Boston reported and submitted a patch for a crash on FreeBSD, but that doesn't sound like your problem. Could be some other type of platform dependent problem. From pgsql-general-owner@postgresql.org Thu Dec 4 16:30:58 2003 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B4599D1DE21; Thu, 4 Dec 2003 06:22:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 76284-10; Thu, 4 Dec 2003 02:22:13 -0400 (AST) Received: from smtp.tor.pathcom.com (smtp.tor.pathcom.com [209.250.128.26]) by svr1.postgresql.org (Postfix) with ESMTP id 19CC1D1B49E; Thu, 4 Dec 2003 02:22:10 -0400 (AST) Received: from sinope.axxent.ca (v47.wall.tor.axxent.ca [209.250.131.207]) by smtp.tor.pathcom.com (8.12.9/8.12.6) with ESMTP id hB46MAKu018309; Thu, 4 Dec 2003 01:22:10 -0500 (EST) Received: by sinope.inside.pathcom.com with Internet Mail Service (5.5.2653.19) id ; Thu, 4 Dec 2003 01:22:10 -0500 Message-ID: From: "Passynkov, Vadim" To: "'pgsql-general@postgresql.org'" Cc: "'pgsql-performance@postgresql.org'" Subject: Async Query Processing on Solaris Date: Thu, 4 Dec 2003 01:22:08 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/260 X-Sequence-Number: 53654 I am using Asynchronous Query Processing interface from libpq library. And I got some strange results on Solaris My test select query is 'SELECT * from pg_user;' and I use select system synchronous I/O multiplexer in 'C' The first test sends 10000 select queries using 10 nonblocking connections to database ( PQsendQuery ). The second test sends the same 10000 select queries using 1 connection ( PQexec ). On FreeBSD there is a huge difference between the async and the sync tests. The async test is much faster than sync test. On Solaris there is no speed difference between async and sync test, actually async test is even slower than sync test. Q. Why ? On FreeBSD: /usr/bin/time ./PgAsyncManager async async test start ... 10000 done 9.46 real 3.48 user 1.25 sys /usr/bin/time ./PgAsyncManager sync sync test start ... 10000 done 22.64 real 3.35 user 1.24 sys On Solaris: /usr/bin/time ./PgAsyncManager async async test start ... 10000 done real 20.6 user 2.1 sys 0.4 /usr/bin/time ./PgAsyncManager sync sync test start ... 10000 done real 18.4 user 1.1 sys 0.5 From pgsql-performance-owner@postgresql.org Thu Dec 4 04:21:44 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DE28BD1B488 for ; Thu, 4 Dec 2003 08:21:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 87163-10 for ; Thu, 4 Dec 2003 04:21:12 -0400 (AST) Received: from monsoon.he.net (monsoon.he.net [64.62.221.2]) by svr1.postgresql.org (Postfix) with SMTP id 32634D1B483 for ; Thu, 4 Dec 2003 04:21:10 -0400 (AST) Received: from [10.5.73.162] ([64.4.254.252]) by monsoon.he.net for ; Thu, 4 Dec 2003 00:21:04 -0800 Subject: Re: Has anyone run on the new G5 yet From: Paul Tuckfield To: pgsql-performance@postgresql.org Cc: shannyconsulting@earthlink.net Content-Type: text/plain Organization: Message-Id: <1070526289.4666.35.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-4) Date: 04 Dec 2003 00:24:50 -0800 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/67 X-Sequence-Number: 4927 (hope I'm posting this correctly) You wrote: >First question is do we gain anything by moving the RH Enterprise >version of Linux in terms of performance, mainly in the IO realm as we >are not CPU bound at all? Second and more radical, has anyone run >postgreSQL on the new Apple G5 with an XRaid system? This seems like a >great value combination. Fast CPU, wide bus, Fibre Channel IO, 2.5TB >all for ~17k. Wow, funny coincidence: I've got a pair of dual xeons w. 8G + 14disk fcal arrays, and an xserve with an XRaid that I've been screwing around with. If you have specific tests you'd like to see, let me know. --- so, for the truly IO bound, here's my recent messin' around summary: In the not-so-structured tests I've done, I've been disappointed with Redhat AS 2.1. IO thruput. I've had difficulty driving a lot of IO thru my dual fcal channels: I can only get one going at 60M/sec, and when I drive IO to the second, I still see only about 60M/sec combined. and when I does get that high it uses about 30% CPU on a dual xeon hyperthreaded box, all in sys (by vmstat). something very wrong there, and the only thing I can conclude is that I'm serializing in the driver somehow (qla2200 driver), thus parallel channels do the same as one, and interrupt madness drives the cpu up just to do this contentious IO. This contrasts with the Redhat 9 I just installed on a similar box, that got 170M/sec on 2 fcal channels, and the expected 5-6% cpu. The above testing was dd straight from /dev/rawX devices, so no buffer cache confusion there. Also had problems getting the Redhat AS to bind to my newer qla2300 adapters at all, whereas they bound fine under RH9. Redhat makes the claim of finer grained locks/semaphores in the qla and AIC drivers in RH AS, but my tests seem to show that the 2 fcal ports were serializing against eachother in the kernel under RH AS, and not so under RH9. Maybe I'm useing the wrong driver under AS. eh. so sort story long, it seems like you're better of with RH9. But again, before you lay out serious coin for xserve or others, if you have specific tests you want to see, I'll take a little time to contrast w. exserve. One of the xeons also has an aic7x scsi controler w 4 drives so It might match your rig better. I also did some token testing on the xserve I have which I believe may only have one processor (how do you tell on osX?) and the xraid has 5 spindles in it. I did a cursory build of postgres on it and also a io test (to the filesystem) and saw about 90M/sec. Dunno if it has dual paths (if you guys know how to tell, let me know) Biggest problem I've had in the past w. linux in general is that it seems to make poor VM choices under heavy filesystem IO. I don't really get exactly where it's going wrong , but I've had numerous experiences on older systems where bursty IO would seem to cause paging on the box (pageout of pieces of the oracle SGA shared memory) which is a performance disaseter. It seems to happen even when the shared memory was sized reasonably below the size of physical ram, presumably because linux is too aggressive in allocating filesystem cache (?) anyway, it seems to make decisions based on desire for zippy workstation performance and gets burned on thruput on database servers. I'm guessing this may be an issue for you , when doing heavy IO. Thing is, it'll show like you're IO bound kindof because you're thrashing. From pgsql-general-owner@postgresql.org Thu Dec 4 08:48:49 2003 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D4739D1B534; Thu, 4 Dec 2003 12:48:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27040-06; Thu, 4 Dec 2003 08:48:13 -0400 (AST) Received: from smtp.tor.pathcom.com (smtp.tor.pathcom.com [209.250.128.26]) by svr1.postgresql.org (Postfix) with ESMTP id AD63ED1B457; Thu, 4 Dec 2003 08:48:09 -0400 (AST) Received: from sinope.axxent.ca (v47.wall.tor.axxent.ca [209.250.131.207]) by smtp.tor.pathcom.com (8.12.9/8.12.6) with ESMTP id hB4CmD5g007083; Thu, 4 Dec 2003 07:48:13 -0500 (EST) Received: by sinope.inside.pathcom.com with Internet Mail Service (5.5.2653.19) id ; Thu, 4 Dec 2003 07:48:13 -0500 Message-ID: From: "Passynkov, Vadim" To: "'pgsql-general@postgresql.org'" Cc: "'pgsql-performance@postgresql.org'" Subject: Async Query Processing on Solaris Date: Thu, 4 Dec 2003 07:48:12 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/234 X-Sequence-Number: 53628 I am using Asynchronous Query Processing interface from libpq library. And I got some strange results on Solaris My test select query is 'SELECT * from pg_user;' and I use select system synchronous I/O multiplexer in 'C' The first test sends 10000 select queries using 10 nonblocking connections to database ( PQsendQuery ). The second test sends the same 10000 select queries using 1 connection ( PQexec ). On FreeBSD there is a huge difference between the async and the sync tests. The async test is much faster than sync test. On Solaris there is no speed difference between async and sync test, actually async test is even slower than sync test. Q. Why ? On FreeBSD: /usr/bin/time ./PgAsyncManager async async test start ... 10000 done 9.46 real 3.48 user 1.25 sys /usr/bin/time ./PgAsyncManager sync sync test start ... 10000 done 22.64 real 3.35 user 1.24 sys On Solaris: /usr/bin/time ./PgAsyncManager async async test start ... 10000 done real 20.6 user 2.1 sys 0.4 /usr/bin/time ./PgAsyncManager sync sync test start ... 10000 done real 18.4 user 1.1 sys 0.5 From pgsql-performance-owner@postgresql.org Thu Dec 4 09:28:48 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 18A67D1B499 for ; Thu, 4 Dec 2003 13:28:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 32634-05 for ; Thu, 4 Dec 2003 09:28:17 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 3B5D9D1B48D for ; Thu, 4 Dec 2003 09:28:14 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hB4DSDbI091964 for ; Thu, 4 Dec 2003 13:28:13 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hB4DCKCM089904 for pgsql-performance@postgresql.org; Thu, 4 Dec 2003 13:12:20 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Minimum hardware requirements for Postgresql db Date: Thu, 04 Dec 2003 08:10:50 -0500 Organization: cbbrowne Computing Inc Lines: 88 Message-ID: References: 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/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:YmtsyZ3pDDni9fqhiYAH2g4BzLU= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/69 X-Sequence-Number: 4929 CHEWTC@ap.nec.com.sg wrote: > Sorry for my mistake on the 15000 recs per day. It was useful for us to pick at that a bit; it was certainly looking a mite suspicious. > In fact, this server is planned as a OLTP database server for a retailer. > Our intention is either to setup 1 or 2 Postgresql db in the server. > > The proper sizing info for the 1st Postgresql db should be: > > No. of item master : 200,000 > (This item master grows at 0.5% daily). > > No. of transactions from Point-of-Sales machines: 25,000 > Plus other tables, the total sizing that I estimated is 590,000 > records daily. So that's more like 7 TPS, with, more than likely, a peak load several times that. > The 2nd Postgresql db will be used by end users on client machines linked > via ODBC, doing manual data entry. > This will house the item master, loyalty card master and other Historical > data to be kept for at least 1.5 years. > > Therefore total sizing for this db is around 165,000,000 recs at any time. FYI, it is useful to plan for purging the old data from the very beginning; if you don't, things can get ugly :-(. > In summary, the single machine must be able to take up around 100 > users connections via both socket and ODBC. And house the above > number of records. Based on multiplying the load by 40, we certainly move from "pedestrian hardware where anything will do" to something requiring more exotic hardware. - You _definitely_ want a disk array, with a bunch of SCSI disks. - You _definitely_ will want some form of RAID controller with battery-backed cache. - You probably want multiple CPUs. - You almost certainly will want a second (and maybe third) complete redundant system that you replicate data to. - The thing that will have _wild_ effects on whether this is enough, or whether you need to go for something even _more_ exotic (e.g. - moving to big iron UNIX(tm), whether that be Solaris, AIX, or HP/UX) is the issue of how heavily the main database gets hit by queries. If "all" it is doing is consolidating transactions, and there is little query load from the POS systems, that is a very different level of load from what happens if it is also servicing pricing queries. Performance will get _destroyed_, regardless of how heavy the iron is, if you hit the OLTP system with a lot of transaction reports. You'll want a secondary replicated system to draw that load off. Evaluating whether it needs to be "big" hardware or "really enormous" hardware is not realistic based on what you have said. There are _big_ variations possible based notably on: 1. What kind of query load does the OLTP server have to serve up? If the answer is "lots," then everything gets more expensive. 2. How was the database schema and the usage of the clients designed? How well it is done will have a _large_ impact on how many TPS the system can cope with. You'll surely need to do some prototyping, and be open to possibilities such as that you'll need to consider alternative OSes. On Intel/AMD hardware, it may be worth considering FreeBSD; it may also be needful to consider "official UNIX(tm)" hardware. It would be unrealistic to pretend more certainty... -- (reverse (concatenate 'string "ac.notelrac.teneerf" "@" "454aa")) http://www.ntlug.org/~cbbrowne/nonrdbms.html "Being really good at C++ is like being really good at using rocks to sharpen sticks." -- Thant Tessman From pgsql-admin-owner@postgresql.org Thu Dec 4 10:55:55 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 64D61D1B468 for ; Thu, 4 Dec 2003 14:55:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 44395-09 for ; Thu, 4 Dec 2003 10:55:26 -0400 (AST) Received: from ar-sd.net (unknown [81.196.32.112]) by svr1.postgresql.org (Postfix) with ESMTP id 105B6D1B432 for ; Thu, 4 Dec 2003 10:55:22 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by ar-sd.net (Postfix) with ESMTP id 7C9DD1D021 for ; Thu, 4 Dec 2003 16:58:37 +0200 (EET) Received: from andy (unknown [192.168.0.11]) by ar-sd.net (Postfix) with SMTP id 562D61CCD8 for ; Thu, 4 Dec 2003 16:58:37 +0200 (EET) Message-ID: <014901c3ba76$ffd65650$0b00a8c0@andy> From: "Andrei Bintintan" To: Subject: Index not used. WHY? Date: Thu, 4 Dec 2003 16:57:51 +0200 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0146_01C3BA87.C0BC67F0" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4927.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4927.1200 X-Virus-Scanned: by AMaViS snapshot-20020531 X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.6 tagged_above=0.0 required=5.0 tests=HTML_50_60, HTML_FONT_COLOR_BLUE, HTML_FONT_COLOR_GREEN, HTML_FONT_COLOR_RED, HTML_FONT_COLOR_UNSAFE X-Spam-Level: X-Archive-Number: 200312/35 X-Sequence-Number: 11543 This is a multi-part message in MIME format. ------=_NextPart_000_0146_01C3BA87.C0BC67F0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi,=20 I have the following table: CREATE TABLE public.rights ( id int4 DEFAULT nextval('"rights_id_seq"'::text) NOT NULL,=20 id_user int4 NOT NULL,=20 id_modull int4 NOT NULL,=20 CONSTRAINT rights_pkey PRIMARY KEY (id) )=20 and I created the following indexes: CREATE INDEX right_id_modull_idx ON rights USING btree (id_modull); CREATE INDEX right_id_user_idx ON rights USING btree (id_user); Now the problem: EXPLAIN SELECT * FROM rights r WHERE r.id_modull =3D15 returnes: Seq Scan on rights r (cost=3D0.00..12.30 rows=3D42 width=3D12) Filter: (id_modull =3D 15) EXPLAIN SELECT * FROM rights r WHERE r.id_user =3D15 returnes: Index Scan using right_id_user_idx on rights r (cost=3D0.00..8.35 rows=3D11= width=3D12) Index Cond: (id_user =3D 15) Question: Why the right_id_modull_idx is NOT USED at the 1st query and the = second query the right_id_user_idx index is used.=20 I don't understand this.=20 Thanx in advance. Andy. ------=_NextPart_000_0146_01C3BA87.C0BC67F0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi,
 
I have the following table:

CREATE TABLE public.rights (
id int4 DEFAULT=20 nextval('"rights_id_seq"'::text) NOT NULL,
= id_user int4 NOT NULL,
= id_modull int4 NOT NULL,
= CONSTRAINT rights_pkey PRIMARY KEY (id)
)=20

and I created the following indexes:<= /P>

CREATE INDEX right_id_modull_idx ON=20 rights USING btree (id_modull);
CREATE INDEX right_id_u= ser_idx=20 ON rights USING b= tree=20 (id_user);

Now the problem:

EXPLAIN SELECT * FROM rights r WHERE r.id_modull =3D15
returnes:
Seq Scan on= =20 rights r (cost=3D0.00..12.30 rows=3D42 width=3D12)
Filter: (id_modull = =3D=20 15)

EXPLAIN SELECT * FROM rights r WHERE r.id_user =3D15
returnes:
Index Scan using right_id_user_idx on=20 rights r (cost=3D0.00..8.35 rows=3D11 width=3D12)
= Index=20 Cond: (id_user =3D 15)

Question: Why the right_id_modull_idx is NOT US= ED at=20 the 1st query and the second query the right_id_user_idx index is used.=20

I don't understand this.

Thanx in=20 advance.
Andy.

 

------=_NextPart_000_0146_01C3BA87.C0BC67F0-- From pgsql-admin-owner@postgresql.org Thu Dec 4 11:21:53 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BB9F1D1B4B1; Thu, 4 Dec 2003 15:20:16 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53565-08; Thu, 4 Dec 2003 11:19:49 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id EBAEBD1B468; Thu, 4 Dec 2003 11:19:44 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 3E7E535500; Thu, 4 Dec 2003 07:19:49 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 3CE3935437; Thu, 4 Dec 2003 07:19:49 -0800 (PST) Date: Thu, 4 Dec 2003 07:19:49 -0800 (PST) From: Stephan Szabo To: Andrei Bintintan Cc: pgsql-admin@postgresql.org, pgsql-performance@postgresql.org Subject: Re: Index not used. WHY? In-Reply-To: <014901c3ba76$ffd65650$0b00a8c0@andy> Message-ID: <20031204071432.R66123@megazone.bigpanda.com> References: <014901c3ba76$ffd65650$0b00a8c0@andy> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/37 X-Sequence-Number: 11545 On Thu, 4 Dec 2003, Andrei Bintintan wrote: > Hi, > > I have the following table: > CREATE TABLE public.rights ( > id int4 DEFAULT nextval('"rights_id_seq"'::text) NOT NULL, > id_user int4 NOT NULL, > id_modull int4 NOT NULL, > CONSTRAINT rights_pkey PRIMARY KEY (id) > ) > > and I created the following indexes: > > CREATE INDEX right_id_modull_idx ON rights USING btree (id_modull); > CREATE INDEX right_id_user_idx ON rights USING btree (id_user); > > Now the problem: > > EXPLAIN SELECT * FROM rights r WHERE r.id_modull =15 > returnes: > Seq Scan on rights r (cost=0.00..12.30 rows=42 width=12) > Filter: (id_modull = 15) > > EXPLAIN SELECT * FROM rights r WHERE r.id_user =15 > returnes: > Index Scan using right_id_user_idx on rights r (cost=0.00..8.35 rows=11 width=12) > Index Cond: (id_user = 15) > > Question: Why the right_id_modull_idx is NOT USED at the 1st query and > the second query the right_id_user_idx index is used. As a note, pgsql-performance is a better list for these questions. So, standard questions: How many rows are in the table, what does EXPLAIN ANALYZE show for the queries, if you force index usage (set enable_seqscan=off) on the first what does EXPLAIN ANALYZE show then, have you used ANALYZE/VACUUM ANALYZE recently? From pgsql-performance-owner@postgresql.org Thu Dec 4 12:07:16 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B74CFD1B4CC for ; Thu, 4 Dec 2003 16:06:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 62031-10 for ; Thu, 4 Dec 2003 12:06:27 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 1629AD1B45A for ; Thu, 4 Dec 2003 12:06:27 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB4G6PIt022475 for ; Thu, 4 Dec 2003 08:06:25 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 08:06:25 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 08:06:25 -0800 Subject: tuning questions From: Jack Coates To: pgsql-performance@postgresql.org Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070553983.6498.52.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 08:06:23 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/71 X-Sequence-Number: 4931 Hi, sorry for duplication, I asked this on pgsql-admin first before realizing it wasn't the appropriate list. I'm having trouble optimizing PostgreSQL for an admittedly heinous worst-case scenario load. testbed: dual P3 1.3 GHz box with 2GB RAM two IDE 120G drives on separate channels (DMA on), OS on one, DB on the other, some swap on each (totalling 2.8G). RH Linux 8. I've installed PG 7.3.4 from source (./configure && make && make install) and from PGDG RPMs and can switch back and forth. I also have the 7.4 source but haven't done any testing with it yet aside from starting it and importing some data. The application is on another server, and does this torture test: it builds a large table (~6 million rows in one test, ~18 million in another). Rows are then pulled in chunks of 4 to 6 thousand, acted on, and inserted back into another table (which will of course eventually grow to the full size of the first). The problem is that pulling the 4 to 6 thousand rows puts PostgreSQL into a tail spin: postmaster hammers on CPU anywhere from 90 seconds to five minutes before returning the data. During this time vmstat shows that disk activity is up of course, but it doesn't appear to be with page swapping (free and top and vmstat). Another problem is that performance of the 6 million row job is decent if I stop the job and run a vacuumdb --analyze before letting it continue; is this something that 7.4 will help with? vacuumb --analyze doesn't seem to have much effect on the 18 million row job. I've tweaked shared buffers to 8192, pushed sort memory to 2048, vacuum memory to 8192, and effective cache size to 10000. /proc/sys/kernel/shmmax is set to 1600000000 and /proc/sys/fs/file-max is set to 65536. Ulimit -n 3192. I've read several sites and postings on tuning PG and have tried a number of different theories, but I'm still not getting the architecture of how things work. thanks, -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 12:34:36 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A4873D1B49D for ; Thu, 4 Dec 2003 16:34:32 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70287-04 for ; Thu, 4 Dec 2003 12:34:01 -0400 (AST) Received: from yertle.kcilink.com (yertle.kcilink.com [216.194.193.105]) by svr1.postgresql.org (Postfix) with ESMTP id A30A4D1B4A1 for ; Thu, 4 Dec 2003 12:34:00 -0400 (AST) Received: by yertle.kcilink.com (Postfix, from userid 100) id 3B86E2178B; Thu, 4 Dec 2003 11:33:59 -0500 (EST) From: Vivek Khera MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <16335.25079.106697.7521@yertle.int.kciLink.com> Date: Thu, 4 Dec 2003 11:33:59 -0500 To: pgsql-performance@postgresql.org Subject: Re: autovacuum daemon stops doing work after about an hour In-Reply-To: <1070516326.1455.11.camel@zedora.zeut.net> References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> X-Mailer: VM 7.17 under 21.4 (patch 14) "Reasonable Discussion" XEmacs Lucid X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/72 X-Sequence-Number: 4932 >>>>> "MTO" == Matthew T O'Connor writes: >> Then it just sits there. I started it at 11:35am, and it is now >> 3:30pm. MTO> Weird.... Alphabetically speaking, is vkmlm."public"."user_list" be the MTO> last table in the last schema in the last database? You are running conveniently, yes it is... MTO> with -d4, so you would get a message about going to sleep shortly after MTO> dealing with the last table, but you didn't get the sleep message, so I MTO> don't think the problem is that pg_autovacuum is sleeping for an MTO> inordinate amount time. The only sleep logged was [2003-12-03 04:47:13 PM] 1 All DBs checked in: 84996853 usec, will sleep for 469 secs. Here's all it did on yesterday afternoon's "hour of work": [2003-12-03 04:45:48 PM] Performing: ANALYZE "public"."url_track" [2003-12-03 04:46:27 PM] Performing: ANALYZE "public"."msg_recipients" [2003-12-03 04:46:55 PM] Performing: ANALYZE "public"."deliveries" [2003-12-03 04:46:55 PM] Performing: ANALYZE "public"."user_list" [2003-12-03 04:47:12 PM] Performing: ANALYZE "public"."sessions" [2003-12-03 04:55:02 PM] Performing: ANALYZE "public"."url_track" [2003-12-03 04:55:22 PM] Performing: VACUUM ANALYZE "public"."msg_recipients" [2003-12-03 05:40:11 PM] Performing: VACUUM ANALYZE "public"."user_list" then 18 minutes later, it reported: [2003-12-03 05:58:25 PM] select relfilenode,reltuples,relpages from pg_class where relfilenode=18588202 [2003-12-03 05:58:25 PM] table name: vkmlm."public"."user_list" [2003-12-03 05:58:25 PM] relfilenode: 18588202; relisshared: 0 [2003-12-03 05:58:25 PM] reltuples: 9; relpages: 427920 [2003-12-03 05:58:25 PM] curr_analyze_count: 2559236; cur_delete_count: 2475824 [2003-12-03 05:58:25 PM] ins_at_last_analyze: 2559236; del_at_last_vacuum: 2475824 [2003-12-03 05:58:25 PM] insert_threshold: 509; delete_threshold 1001 and stopped doing anything. MTO> when you kill it, do you get a core file? Could you do a backtrace and MTO> see where pg_autovacuum is hung up? nope. unfortunately my PG libs are without debugging, too. I'll rebuild pg_autovacuum with debugging and run it under gdb so I can see where it gets stuck. I'll report back when I find something. I just wanted to check first if anyone else ran into this. From pgsql-performance-owner@postgresql.org Thu Dec 4 13:00:42 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C53B3D1B444 for ; Thu, 4 Dec 2003 17:00:14 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 74130-08 for ; Thu, 4 Dec 2003 12:59:43 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 0A207D1B446 for ; Thu, 4 Dec 2003 12:59:37 -0400 (AST) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4022070; Thu, 04 Dec 2003 09:00:25 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Jack Coates , pgsql-performance@postgresql.org Subject: Re: tuning questions Date: Thu, 4 Dec 2003 08:59:06 -0800 User-Agent: KMail/1.4.3 References: <1070553983.6498.52.camel@cletus.lyris.com> In-Reply-To: <1070553983.6498.52.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200312040859.06963.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/74 X-Sequence-Number: 4933 Jack, > The application is on another server, and does this torture test: it > builds a large table (~6 million rows in one test, ~18 million in > another). Rows are then pulled in chunks of 4 to 6 thousand, acted on, > and inserted back into another table (which will of course eventually > grow to the full size of the first). >e tweaked shared buffers to 8192, pushed sort memory to 2048, vacuum > memory to 8192, and effective cache size to 10000. > /proc/sys/kernel/shmmax is set to 1600000000 and /proc/sys/fs/file-max > is set to 65536. Ulimit -n 3192. Have you read this? http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html Actually, your situation is not "worst case". For one thing, your process is effectively single-user; this allows you to throw all of your resources at one user. The problem is that your settings have effectively throttled PG at a level appropriate to a many-user and/or multi-purpose system. You need to "open them up". For something involving massive updating/transformation like this, once you've done the basics (see that URL above) the main settings which will affect you are sort_mem and checkpoint_segments, both of which I'd advise jacking way up (test by increments). Raising wal_buffers wouldn't hurt either. Also, give some thought to running VACUUM and/or ANALYZE between segments of your procedure. Particularly if you do updates to many rows of a table and then query based on the changed data, it is vital to run an ANALYZE first, and usually a good idea to run a VACUUM if it was an UPDATE or DELETE and not an INSERT. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Dec 4 13:00:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F26C7D1B49B for ; Thu, 4 Dec 2003 17:00:23 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75349-03 for ; Thu, 4 Dec 2003 12:59:53 -0400 (AST) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 3D5DDD1B464 for ; Thu, 4 Dec 2003 12:59:52 -0400 (AST) Received: (qmail 25797 invoked from network); 4 Dec 2003 16:59:55 -0000 Received: from localhost (HELO squeegit) (threshar@127.0.0.1) by localhost with SMTP; 4 Dec 2003 16:59:55 -0000 Date: Thu, 4 Dec 2003 11:59:32 -0500 From: Jeff To: Jack Coates Cc: pgsql-performance@postgresql.org Subject: Re: tuning questions Message-Id: <20031204115932.154eebed.threshar@torgo.978.org> In-Reply-To: <1070553983.6498.52.camel@cletus.lyris.com> References: <1070553983.6498.52.camel@cletus.lyris.com> X-Mailer: Sylpheed version 0.9.7 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/73 X-Sequence-Number: 4934 On Thu, 04 Dec 2003 08:06:23 -0800 Jack Coates wrote: > testbed: > dual P3 1.3 GHz box with 2GB RAM > two IDE 120G drives on separate channels (DMA on), OS on one, DB on > the other, some swap on each (totalling 2.8G). > RH Linux 8. Side Note: be sure to turn off write caching on those disks or you may have data corruption in the event of a failure > The problem is that pulling the 4 to 6 thousand rows puts PostgreSQL > into a tail spin: postmaster hammers on CPU anywhere from 90 seconds > to five minutes before returning the data. During this time vmstat > shows that disk activity is up of course, but it doesn't appear to be > with page swapping (free and top and vmstat). > Have you tried modifying the app to retrieve the rows in smaller chunks? (use a cursor). this way it only needs to alloate memory to hold say, 100 rows at a time instead of 6000. Also, have you explain analyze'd your queries to make sure PG is picking a good plan to execute? > I've tweaked shared buffers to 8192, pushed sort memory to 2048, > vacuum memory to 8192, and effective cache size to 10000. > /proc/sys/kernel/shmmax is set to 1600000000 and /proc/sys/fs/file-max > is set to 65536. Ulimit -n 3192. you should set effective cache size bigger, especially with 2GB of memory. effective_cache_size tells PG 'about' how much data it cna expect the OS to cache. and.. I'm not sure about your query, but perhaps the sort of those 6000 rows is spilling to disk? If you look in explain analyze you'll see in the "Sort" step(s) it will tell you how many rows and how "wide" they are. If rows * width > sort_mem, it will have to spill the sort to disk, which is slow. If you post query info and explain analyze's we can help optimize the query itself. -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Thu Dec 4 13:28:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8BB0CD1B4C4 for ; Thu, 4 Dec 2003 17:28:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81423-01 for ; Thu, 4 Dec 2003 13:28:21 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id C3F73D1B4C2 for ; Thu, 4 Dec 2003 13:28:20 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hB4HSJbI029103 for ; Thu, 4 Dec 2003 17:28:19 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hB4H6SMY025766 for pgsql-performance@postgresql.org; Thu, 4 Dec 2003 17:06:28 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Has anyone run on the new G5 yet Date: Thu, 04 Dec 2003 09:06:33 -0800 Organization: Hub.Org Networking Services Lines: 12 Message-ID: References: <3FCE5855.5040403@earthlink.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: <3FCE5855.5040403@earthlink.net> To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/76 X-Sequence-Number: 4936 Sean Shanny wrote: > > First question is do we gain anything by moving the RH Enterprise > version of Linux in terms of performance, mainly in the IO realm as we > are not CPU bound at all? Second and more radical, has anyone run > postgreSQL on the new Apple G5 with an XRaid system? This seems like a > great value combination. Fast CPU, wide bus, Fibre Channel IO, 2.5TB > all for ~17k. Seems like a great value but until Apple produces a G5 that supports ECC, I'd pass on them. From pgsql-performance-owner@postgresql.org Thu Dec 4 13:10:29 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id AD265D1B450 for ; Thu, 4 Dec 2003 17:10:25 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75122-07 for ; Thu, 4 Dec 2003 13:09:55 -0400 (AST) Received: from mail.dsvr.co.uk (mail.dsvr.co.uk [212.69.192.9]) by svr1.postgresql.org (Postfix) with ESMTP id 850D8D1B49A for ; Thu, 4 Dec 2003 13:09:54 -0400 (AST) Received: from dsvr.net (eth0.rt1.o1-1.tck.dsvr.net [::ffff:212.69.216.20]) by mail.dsvr.co.uk with esmtp; Thu, 04 Dec 2003 17:09:53 +0000 Message-ID: <3FCF6B2F.3070802@dsvr.net> Date: Thu, 04 Dec 2003 17:13:19 +0000 From: Rob Fielding User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20031119 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: tuning questions References: <1070553983.6498.52.camel@cletus.lyris.com> In-Reply-To: <1070553983.6498.52.camel@cletus.lyris.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/75 X-Sequence-Number: 4935 > > I've tweaked shared buffers to 8192, pushed sort memory to 2048, vacuum > memory to 8192, and effective cache size to 10000. > /proc/sys/kernel/shmmax is set to 1600000000 and /proc/sys/fs/file-max > is set to 65536. Ulimit -n 3192. Your sharedmemory is too high, and not even being used effectivey. Your other settings are too low. Ball park guessing here, but I'd say first read (and understand) this: http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html Then make shared memory about 10-20% available ram, and set: ((shmmax/1024) - ( 14.2 * max_connections ) - 250 ) / 8.2 = shared_buffers decrease random_page_cost to 0.3 and wack up sort mem by 16 times, effective cache size to about 50% RAM (depending on your other settings) and try that for starters. -- Rob Fielding rob@dsvr.net www.dsvr.co.uk Development Designer Servers Ltd From pgsql-performance-owner@postgresql.org Thu Dec 4 13:46:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3D91CD1B43F for ; Thu, 4 Dec 2003 17:46:39 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 83381-02 for ; Thu, 4 Dec 2003 13:46:08 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 8342DD1B441 for ; Thu, 4 Dec 2003 13:46:03 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hB4HhWDW022521; Thu, 4 Dec 2003 10:43:32 -0700 (MST) Date: Thu, 4 Dec 2003 10:26:38 -0700 (MST) From: "scott.marlowe" To: Jack Coates Cc: Subject: Re: tuning questions In-Reply-To: <1070553983.6498.52.camel@cletus.lyris.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/77 X-Sequence-Number: 4937 On Thu, 4 Dec 2003, Jack Coates wrote: > Another problem is that performance of the 6 million row job is decent > if I stop the job and run a vacuumdb --analyze before letting it > continue; is this something that 7.4 will help with? vacuumb --analyze > doesn't seem to have much effect on the 18 million row job. Just to add to what the others have said here, you probably want to run the pg_autovacuum daemon in the background. It comes with 7.4 but will work fine with 7.3. From pgsql-performance-owner@postgresql.org Thu Dec 4 13:58:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 95C70D1B474 for ; Thu, 4 Dec 2003 17:58:09 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85015-03 for ; Thu, 4 Dec 2003 13:57:39 -0400 (AST) Received: from rlx13.zapatec.com (66-117-144-213.zapatec.lmi.net [66.117.144.213]) by svr1.postgresql.org (Postfix) with ESMTP id 8AFFED1B45B for ; Thu, 4 Dec 2003 13:57:38 -0400 (AST) Received: from rlx11.zapatec.com (rlx11.pr.zapatec.com [192.168.1.132]) by rlx13.zapatec.com (Postfix) with ESMTP id E7059A941 for ; Thu, 4 Dec 2003 09:57:38 -0800 (PST) Received: (from dror@localhost) by rlx11.zapatec.com (8.12.3/8.12.3/Submit) id hB4Hvcjh058970 for pgsql-performance@postgresql.org; Thu, 4 Dec 2003 09:57:38 -0800 (PST) (envelope-from dror) Date: Thu, 4 Dec 2003 09:57:38 -0800 From: Dror Matalon To: pgsql-performance@postgresql.org Subject: Re: tuning questions Message-ID: <20031204175738.GD34796@rlx11.zapatec.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <20031204115932.154eebed.threshar@torgo.978.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031204115932.154eebed.threshar@torgo.978.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/78 X-Sequence-Number: 4938 On Thu, Dec 04, 2003 at 11:59:32AM -0500, Jeff wrote: > On Thu, 04 Dec 2003 08:06:23 -0800 > Jack Coates wrote: > > > testbed: > > dual P3 1.3 GHz box with 2GB RAM > > two IDE 120G drives on separate channels (DMA on), OS on one, DB on > > the other, some swap on each (totalling 2.8G). > > RH Linux 8. > > Side Note: be sure to turn off write caching on those disks or you may > have data corruption in the event of a failure I've seen this comment several times from different people. Would someone care to explain how you would get data corruption? I thought that the whole idea of the log is to provide a journal similar to what you get in a journaling file system. In other words, the db writes a series of transactions to the log and marks that "log entry" (don't know the right nomeclature) as valid. When the db crashes, it reads the log, and discards the last "log entry" if it wasn't marked as valid, and "replays" any transactions that haven't been commited ot the db. The end result being that you might loose your last transaction(s) if the db crashes, but nothing ever gets corrupted. So what am I missing in this picture? Regards, Dror -- Dror Matalon Zapatec Inc 1700 MLK Way Berkeley, CA 94709 http://www.fastbuzz.com http://www.zapatec.com From pgsql-performance-owner@postgresql.org Thu Dec 4 14:17:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DA417D1B49B for ; Thu, 4 Dec 2003 18:17:37 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88924-03 for ; Thu, 4 Dec 2003 14:17:07 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id C8DC2D1B444 for ; Thu, 4 Dec 2003 14:17:05 -0400 (AST) Received: from [66.219.92.2] (HELO temoku) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4022445; Thu, 04 Dec 2003 10:17:55 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: "scott.marlowe" , Jack Coates Subject: Re: tuning questions Date: Thu, 4 Dec 2003 10:03:52 -0800 User-Agent: KMail/1.4.3 Cc: References: In-Reply-To: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200312041003.52985.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/81 X-Sequence-Number: 4941 Scott, > Just to add to what the others have said here, you probably want to run= =20 > the pg_autovacuum daemon in the background. It comes with 7.4 but will= =20 > work fine with 7.3.=20=20 I don't recommend using pg_autovacuum with a data transformation task. pg= _av=20 is designed for "regular use" not huge batch tasks. --=20 -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Dec 4 14:08:34 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 71886D1B444 for ; Thu, 4 Dec 2003 18:08:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85878-05 for ; Thu, 4 Dec 2003 14:07:59 -0400 (AST) Received: from indygecko.com (h24-71-76-41.ok.shawcable.net [24.71.76.41]) by svr1.postgresql.org (Postfix) with ESMTP id 0B4D9D1B443 for ; Thu, 4 Dec 2003 14:07:58 -0400 (AST) Received: from indy ([::ffff:192.168.10.10]) by indygecko.com with esmtp; Thu, 04 Dec 2003 10:07:56 -0800 Subject: Re: tuning questions From: Jord Tanner To: Dror Matalon Cc: pgsql-performance@postgresql.org In-Reply-To: <20031204175738.GD34796@rlx11.zapatec.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <20031204115932.154eebed.threshar@torgo.978.org> <20031204175738.GD34796@rlx11.zapatec.com> Message-Id: <1070561276.5586.188.camel@gecko.indygecko.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.4.5 Date: Thu, 04 Dec 2003 10:07:56 -0800 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/79 X-Sequence-Number: 4939 If I understand the problem correctly, the issue is that IDE drives signal that data has been written to disk when they actually are holding the data in the write cache. In the case of a power down (and I remember someone showing some test results confirming this, check the list archive) the data in the drive write cache is lost, resulting in corrupted logs. Anyone else have more details? Jord Tanner On Thu, 2003-12-04 at 09:57, Dror Matalon wrote: > On Thu, Dec 04, 2003 at 11:59:32AM -0500, Jeff wrote: > > On Thu, 04 Dec 2003 08:06:23 -0800 > > Jack Coates wrote: > > > > > testbed: > > > dual P3 1.3 GHz box with 2GB RAM > > > two IDE 120G drives on separate channels (DMA on), OS on one, DB on > > > the other, some swap on each (totalling 2.8G). > > > RH Linux 8. > > > > Side Note: be sure to turn off write caching on those disks or you may > > have data corruption in the event of a failure > > I've seen this comment several times from different people. > Would someone care to explain how you would get data corruption? I > thought that the whole idea of the log is to provide a journal similar > to what you get in a journaling file system. > > In other words, the db writes a series of transactions to the log and marks > that "log entry" (don't know the right nomeclature) as valid. When the db > crashes, it reads the log, and discards the last "log entry" if it wasn't > marked as valid, and "replays" any transactions that haven't been > commited ot the db. The end result being that you might loose your last > transaction(s) if the db crashes, but nothing ever gets corrupted. > > So what am I missing in this picture? > > Regards, > > Dror -- Jord Tanner From pgsql-performance-owner@postgresql.org Thu Dec 4 14:12:28 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A87DBD1B43F for ; Thu, 4 Dec 2003 18:12:23 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 87707-03 for ; Thu, 4 Dec 2003 14:11:53 -0400 (AST) Received: from mail.libertyrms.com (unknown [209.167.124.227]) by svr1.postgresql.org (Postfix) with ESMTP id 4630CD1B468 for ; Thu, 4 Dec 2003 14:11:52 -0400 (AST) Received: from [10.1.2.146] (helo=dba3.int.libertyrms.info) by mail.libertyrms.com with esmtp (Exim 3.22 #3 (Debian)) id 1ARxxA-0003CA-00 for ; Thu, 04 Dec 2003 13:11:52 -0500 Received: by dba3.int.libertyrms.info (Postfix, from userid 1019) id 6889A138F4; Thu, 4 Dec 2003 13:11:52 -0500 (EST) Date: Thu, 4 Dec 2003 13:11:52 -0500 From: Andrew Sullivan To: pgsql-performance@postgresql.org Subject: Re: tuning questions Message-ID: <20031204181152.GO6080@libertyrms.info> Mail-Followup-To: Andrew Sullivan , pgsql-performance@postgresql.org References: <1070553983.6498.52.camel@cletus.lyris.com> <20031204115932.154eebed.threshar@torgo.978.org> <20031204175738.GD34796@rlx11.zapatec.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031204175738.GD34796@rlx11.zapatec.com> User-Agent: Mutt/1.5.4i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/80 X-Sequence-Number: 4940 On Thu, Dec 04, 2003 at 09:57:38AM -0800, Dror Matalon wrote: > > I've seen this comment several times from different people. > Would someone care to explain how you would get data corruption? I > thought that the whole idea of the log is to provide a journal similar > to what you get in a journaling file system. > So what am I missing in this picture? That a journalling file system can _also_ have file corruption if you have write caching enabled and no battery back up. If the drive tells the OS, "Yep! It's all on the disk!" bit it is _not_ actually scribed in the little bitty magnetic patterns -- and at that very moment, the power goes away -- the data that was reported to have been on the disk, but which was actually _not_ on the disk, is no longer anywhere. (Well, except in the past. But time travel was disabled some versions ago. ;-) A -- ---- Andrew Sullivan 204-4141 Yonge Street Afilias Canada Toronto, Ontario Canada M2P 2A8 +1 416 646 3304 x110 From pgsql-performance-owner@postgresql.org Thu Dec 4 14:31:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 07530D1B474 for ; Thu, 4 Dec 2003 18:30:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 86836-10 for ; Thu, 4 Dec 2003 14:30:15 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id C7440D1B483 for ; Thu, 4 Dec 2003 14:30:13 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hB4ITPDW025767; Thu, 4 Dec 2003 11:29:25 -0700 (MST) Date: Thu, 4 Dec 2003 11:12:30 -0700 (MST) From: "scott.marlowe" To: Josh Berkus Cc: Jack Coates , Subject: Re: tuning questions In-Reply-To: <200312041003.52985.josh@agliodbs.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/82 X-Sequence-Number: 4942 On Thu, 4 Dec 2003, Josh Berkus wrote: > Scott, > > > Just to add to what the others have said here, you probably want to run > > the pg_autovacuum daemon in the background. It comes with 7.4 but will > > work fine with 7.3. > > I don't recommend using pg_autovacuum with a data transformation task. pg_av > is designed for "regular use" not huge batch tasks. What bad thing is likely to happen if it's used here? Fire too often or use too much I/O bandwidth? Would that be fixed by the patch being tested to introduce a delay every x pages of vacuuming? From pgsql-performance-owner@postgresql.org Thu Dec 4 14:36:52 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4C724D1B49E for ; Thu, 4 Dec 2003 18:36:36 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88923-07 for ; Thu, 4 Dec 2003 14:36:05 -0400 (AST) Received: from sardegna.mbigroup.it (unknown [151.8.40.3]) by svr1.postgresql.org (Postfix) with ESMTP id 3BFEFD1B444 for ; Thu, 4 Dec 2003 14:36:03 -0400 (AST) Received: from sicilia.pisa.mbigroup.it (localhost [127.0.0.1]) by sardegna.mbigroup.it (8.11.6/8.8.7) with ESMTP id hB4IZxe02358; Thu, 4 Dec 2003 19:35:59 +0100 Received: from bigfoot.com (canarie.pisa.mbigroup.it [192.168.9.30]) by sicilia.pisa.mbigroup.it (8.11.6/8.11.6) with ESMTP id hB4IZs301619; Thu, 4 Dec 2003 19:35:54 +0100 Message-ID: <3FCF7E74.9060604@bigfoot.com> Date: Thu, 04 Dec 2003 19:35:32 +0100 From: Gaetano Mendola User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031202 Thunderbird/0.4RC1 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Cc: Vivek Khera Subject: Re: autovacuum daemon stops doing work after about an hour References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <16335.25079.106697.7521@yertle.int.kciLink.com> In-Reply-To: <16335.25079.106697.7521@yertle.int.kciLink.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/83 X-Sequence-Number: 4943 Vivek Khera wrote: >>>>>>"MTO" == Matthew T O'Connor writes: > > >>>Then it just sits there. I started it at 11:35am, and it is now >>>3:30pm. > > > MTO> Weird.... Alphabetically speaking, is vkmlm."public"."user_list" be the > MTO> last table in the last schema in the last database? You are running > > conveniently, yes it is... > > MTO> with -d4, so you would get a message about going to sleep shortly after > MTO> dealing with the last table, but you didn't get the sleep message, so I > MTO> don't think the problem is that pg_autovacuum is sleeping for an > MTO> inordinate amount time. > > The only sleep logged was > > [2003-12-03 04:47:13 PM] 1 All DBs checked in: 84996853 usec, will sleep for 469 secs. What I seen is: # tail -f auto.log [2003-12-04 07:10:18 PM] reltuples: 72; relpages: 1 [2003-12-04 07:10:18 PM] curr_analyze_count: 72; cur_delete_count: 0 [2003-12-04 07:10:18 PM] ins_at_last_analyze: 72; del_at_last_vacuum: 0 [2003-12-04 07:10:18 PM] insert_threshold: 572; delete_threshold 536 [2003-12-04 07:10:18 PM] table name: empdb."public"."contracts" [2003-12-04 07:10:18 PM] relfilenode: 17784; relisshared: 0 [2003-12-04 07:10:18 PM] reltuples: 347; relpages: 5 [2003-12-04 07:10:18 PM] curr_analyze_count: 347; cur_delete_count: 0 [2003-12-04 07:10:18 PM] ins_at_last_analyze: 347; del_at_last_vacuum: 0 [2003-12-04 07:10:18 PM] insert_threshold: 847; delete_threshold 673 [ 5 minutes of delay ] <----- LOOK THIS [2003-12-04 07:10:18 PM] 503 All DBs checked in: 179396 usec, will sleep for 300 secs. [2003-12-04 07:15:19 PM] 504 All DBs checked in: 98814 usec, will sleep for 300 secs. I think is a good Idea put a fflush after: fprintf(LOGOUTPUT, "[%s] %s\n", timebuffer, logentry); Regards Gaetano Mendola From pgsql-performance-owner@postgresql.org Thu Dec 4 14:58:25 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3E913D1B441 for ; Thu, 4 Dec 2003 18:58:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 93647-03 for ; Thu, 4 Dec 2003 14:57:52 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 43CD0D1B433 for ; Thu, 4 Dec 2003 14:57:51 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ARyff-00048V-00; Thu, 04 Dec 2003 20:57:51 +0200 Date: Thu, 4 Dec 2003 20:57:51 +0200 To: pgsql-performance@postgresql.org Subject: Slow UPADTE, compared to INSERT Message-ID: <20031204185751.GA15893@alcaron.ee> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/84 X-Sequence-Number: 4944 Hello! I am relative newcomer to SQL and PostgreSQL world, so please forgive me if this question is stupid. I am experiencing strange behaviour, where simple UPDATE of one field is very slow, compared to INSERT into table with multiple indexes. I have two tables - one with raw data records (about 24000), where one field contains status information (varchar(10)). First table has no indexes, only primary key (recid). Second table contains processed records - some fields are same as first table, others are calculated during processing. Records are processed by Python script, which uses PyPgSQL for PostgreSQL access. Processing is done by selecting all records from table1 where status matches certain criteria (import). Each record is processed and results are inserted into table2, after inserting status field on same record in table1 is updated with new value (done). Update statement itself is extremely simple: "update table1 set status = 'done' where recid = ..." Most interesting is, that insert takes 0.004 seconds in average, but update takes 0.255 seconds in average. Processing of 24000 records took around 1 hour 20 minutes. Then i changed processing logic not to update every record in table1 after processing. Instead i did insert recid value into temporary table and updated records in table1 after all records were processed and inserted into table2: UPDATE table1 SET Status = 'done' WHERE recid IN (SELECT recid FROM temptable) This way i got processing time of 24000 records down to about 16 minutes. About 13 minutes from this took last UPDATE statement. Why is UPDATE so slow compared to INSERT? I would expect more or less similar performance, or slower on insert since table2 has four indexes in addition to primary key, table1 has only primary key, which is used on update. Am i doing something wrong or is this normal? I am using PostgreSQL 7.3.4, Debian/GNU Linux 3.0 (Woody), kernel 2.4.21, Python 2.3.2, PyPgSQL 2.4 -- Ivar Zarans From pgsql-performance-owner@postgresql.org Thu Dec 4 15:17:26 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BF9C0D1B446 for ; Thu, 4 Dec 2003 19:17:25 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 98924-02 for ; Thu, 4 Dec 2003 15:16:55 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 537A2D1B48E for ; Thu, 4 Dec 2003 15:16:53 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB4JGrIt025290 for ; Thu, 4 Dec 2003 11:16:53 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 11:16:53 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 11:16:53 -0800 Subject: Re: tuning questions From: Jack Coates To: pgsql-performance In-Reply-To: <3FCF6AEB.908@dsvr.net> References: <1070553983.6498.52.camel@cletus.lyris.com> <3FCF6AEB.908@dsvr.net> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070565411.13923.70.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 11:16:51 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/85 X-Sequence-Number: 4945 On Thu, 2003-12-04 at 09:12, Rob Fielding wrote: > > > > I've tweaked shared buffers to 8192, pushed sort memory to 2048, vacuum > > memory to 8192, and effective cache size to 10000. > > /proc/sys/kernel/shmmax is set to 1600000000 and /proc/sys/fs/file-max > > is set to 65536. Ulimit -n 3192. > > Your sharedmemory is too high, and not even being used effectivey. Your > other settings are too low. > > Ball park guessing here, but I'd say first read (and understand) this: > > http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html I've read it many times, understanding is slower :-) > > Then make shared memory about 10-20% available ram, and set: > > ((shmmax/1024) - ( 14.2 * max_connections ) - 250 ) / 8.2 = shared_buffers > > decrease random_page_cost to 0.3 and wack up sort mem by 16 times, > effective cache size to about 50% RAM (depending on your other settings) > and try that for starters. Following this, I've done: 2gb ram = 2,000,000,000 bytes 15 % of that = 300,000,000 bytes divided by 1024 = 292,969 kbytes max_conn * 14.2 = 454 kbytes subtract c4 = 292,514 kbytes subtract 250 = 292,264 kbytes divide by 8.2 = 35,642 shared_buffers performance is unchanged for the 18M job -- pg continues to use ~ 285-300M, system load and memory usage stay the same. I killed that, deleted from the affected tables, inserted a 6M job, and started a vacuumdb --anaylze. It's been running for 20 minutes now... getting the SQL query better optimized for PG is on my todo list, but not something I can do right now -- this application is designed to be cross-platform with MS-SQL, PG, and Oracle so tweaking SQL is a touchy subject. The pgavd conversation is intriguing, but I don't really understand the role of vacuuming. Would this be a correct statement: "PG needs to regularly re-evaluate the database in order to adjust itself?" I'm imagining that it continues to treat the table as a small one until vacuum informs it that the table is now large? -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 15:17:50 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 05124D1B453 for ; Thu, 4 Dec 2003 19:17:49 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 93647-10 for ; Thu, 4 Dec 2003 15:17:19 -0400 (AST) Received: from main.wiredfool.com (dsl231-055-035.sea1.dsl.speakeasy.net [216.231.55.35]) by svr1.postgresql.org (Postfix) with ESMTP id 9476DD1B498 for ; Thu, 4 Dec 2003 15:17:12 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 2573C2728E for ; Thu, 4 Dec 2003 11:12:10 -0800 (PST) Received: from main.wiredfool.com ([127.0.0.1]) by localhost (cabbage [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 25416-02 for ; Thu, 4 Dec 2003 11:12:07 -0800 (PST) Received: from [IPv6:::1] (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 1D9312728D for ; Thu, 4 Dec 2003 11:12:07 -0800 (PST) Mime-Version: 1.0 (Apple Message framework v606) In-Reply-To: <20031204181152.GO6080@libertyrms.info> References: <1070553983.6498.52.camel@cletus.lyris.com> <20031204115932.154eebed.threshar@torgo.978.org> <20031204175738.GD34796@rlx11.zapatec.com> <20031204181152.GO6080@libertyrms.info> Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <745F4B76-268E-11D8-848E-0003930F2A6C@soroos.net> Content-Transfer-Encoding: 7bit From: Eric Soroos Subject: Re: tuning questions Date: Thu, 4 Dec 2003 11:17:08 -0800 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.606) X-Virus-Scanned: by amavisd-new-20030616-p3 (Debian) at main.wiredfool.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/86 X-Sequence-Number: 4946 On Dec 4, 2003, at 10:11 AM, Andrew Sullivan wrote: > On Thu, Dec 04, 2003 at 09:57:38AM -0800, Dror Matalon wrote: >> >> I've seen this comment several times from different people. >> Would someone care to explain how you would get data corruption? I >> thought that the whole idea of the log is to provide a journal similar >> to what you get in a journaling file system. > >> So what am I missing in this picture? > > That a journalling file system can _also_ have file corruption if you > have write caching enabled and no battery back up. If the drive > tells the OS, "Yep! It's all on the disk!" bit it is _not_ actually > scribed in the little bitty magnetic patterns -- and at that very > moment, the power goes away -- the data that was reported to have been > on the disk, but which was actually _not_ on the disk, is no longer > anywhere. (Well, except in the past. But time travel was disabled > some versions ago. ;-) It's not just a theoretical problem. It's happened to me on a laptop drive in the last week or so. I was testing out dbmail by hammering on it on Panther laptop, hfs+ journaling enabled, psql 7.4, latest and greatest. I managed to hang the system hard, requiring a reboot. Psql wouldn't start after the crash, complaining of a damaged relation and helpfully telling me that 'you may need to restore from backup'. No big deal on the data loss, since it was a test/hammering installation. It would have been nice to be able to drop that relation or prune the entire database, but I'm sure that would ultimately run into referential integrity problems. eric From pgsql-performance-owner@postgresql.org Thu Dec 4 15:34:26 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6DD2AD1B445 for ; Thu, 4 Dec 2003 19:34:08 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97909-10 for ; Thu, 4 Dec 2003 15:33:38 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 1DDBFD1B483 for ; Thu, 4 Dec 2003 15:33:35 -0400 (AST) Received: from [66.219.92.2] (HELO temoku) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4022845; Thu, 04 Dec 2003 11:34:25 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Jack Coates , pgsql-performance Subject: Re: tuning questions Date: Thu, 4 Dec 2003 11:20:21 -0800 User-Agent: KMail/1.4.3 References: <1070553983.6498.52.camel@cletus.lyris.com> <3FCF6AEB.908@dsvr.net> <1070565411.13923.70.camel@cletus.lyris.com> In-Reply-To: <1070565411.13923.70.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200312041120.21124.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/88 X-Sequence-Number: 4948 Jack, > Following this, I've done: > 2gb ram > =3D > 2,000,000,000 > bytes This calculation is fun, but I really don't know where you got it from. I= t=20 seems quite baroque. What are you trying to set, exactly? > getting the SQL query better optimized for PG is on my todo list, but > not something I can do right now -- this application is designed to be > cross-platform with MS-SQL, PG, and Oracle so tweaking SQL is a touchy > subject. Well, if you're queries are screwed up, no amount of .conf optimization is= =20 going to help you much. You could criticize that PG is less adept than= =20 some other systems at re-writing "bad queries", and you would be correct.= =20=20 However, there's not much to do about that on existing systems. How about posting some sample code? > The pgavd conversation is intriguing, but I don't really understand the > role of vacuuming. Would this be a correct statement: "PG needs to > regularly re-evaluate the database in order to adjust itself?" I'm > imagining that it continues to treat the table as a small one until > vacuum informs it that the table is now large? Not Vacuum, Analyze. Otherwise correct. Mind you, in "regular use" where= =20 only a small % of the table changes per hour, periodic ANALYZE is fine.=20= =20 However, in "batch data transform" analyze statements need to be keyed to t= he=20 updates and/or imports. BTW, I send a couple of e-mails to the Lyris documentation maintainer about= =20 updating out-of-date information about setting up PostgreSQL. I never got= a=20 response, and I don't think my changes were made. --=20 -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Dec 4 15:24:11 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DAE58D1B482 for ; Thu, 4 Dec 2003 19:24:10 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 99873-02 for ; Thu, 4 Dec 2003 15:23:41 -0400 (AST) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 7515AD1B45B for ; Thu, 4 Dec 2003 15:23:39 -0400 (AST) Received: (qmail 26680 invoked from network); 4 Dec 2003 19:23:45 -0000 Received: from localhost (HELO squeegit) (threshar@127.0.0.1) by localhost with SMTP; 4 Dec 2003 19:23:45 -0000 Date: Thu, 4 Dec 2003 14:23:20 -0500 From: Jeff To: Ivar Zarans Cc: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-Id: <20031204142320.3c021100.threshar@torgo.978.org> In-Reply-To: <20031204185751.GA15893@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> X-Mailer: Sylpheed version 0.9.7 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/87 X-Sequence-Number: 4947 On Thu, 4 Dec 2003 20:57:51 +0200 Ivar Zarans wrote: . > table1 is updated with new value (done). Update statement itself is > extremely simple: "update table1 set status = 'done' where recid = > ..." > > Most interesting is, that insert takes 0.004 seconds in average, but > update takes 0.255 seconds in average. Processing of 24000 records > took around 1 hour 20 minutes. Do you have an index on recid? and did you vacuum analyze after you loaded up the data? > > Then i changed processing logic not to update every record in table1 > after processing. Instead i did insert recid value into temporary > table and updated records in table1 after all records were processed > and inserted into table2: > UPDATE table1 SET Status = 'done' WHERE recid IN (SELECT recid FROM > temptable) > "IN" queries are terribly slow on versions before 7.4 > Why is UPDATE so slow compared to INSERT? I would expect more or less > similar performance, or slower on insert since table2 has four indexes > in addition to primary key, table1 has only primary key, which is used > on update. Am i doing something wrong or is this normal? > Remember, UPDATE has to do all the work of select and more. And if you have 4 indexes those will also add to the time (Since it has to update/add them to the tree) -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Thu Dec 4 15:37:10 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0D875D1B484 for ; Thu, 4 Dec 2003 19:31:07 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 00227-04 for ; Thu, 4 Dec 2003 15:30:37 -0400 (AST) Received: from lorax.kcilink.com (lorax.kciLink.com [206.112.95.1]) by svr1.postgresql.org (Postfix) with ESMTP id 741C2D1B439 for ; Thu, 4 Dec 2003 15:29:59 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id B61F53E80 for ; Thu, 4 Dec 2003 14:29:59 -0500 (EST) Received: from lorax.kcilink.com ([127.0.0.1]) by localhost (lorax.kcilink.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 36778-02-3 for ; Thu, 4 Dec 2003 14:29:59 -0500 (EST) Received: from lorax.kciLink.com (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id 120F73E7E for ; Thu, 4 Dec 2003 14:29:59 -0500 (EST) Received: (from news@localhost) by lorax.kciLink.com (8.12.9p2/8.12.9/Submit) id hB4JTwdo058269 for pgsql-performance@postgresql.org; Thu, 4 Dec 2003 14:29:58 -0500 (EST) (envelope-from news) To: pgsql-performance@postgresql.org Path: not-for-mail From: Vivek Khera Newsgroups: ml.postgres.performance Subject: Re: autovacuum daemon stops doing work after about an hour Date: Thu, 04 Dec 2003 14:29:58 -0500 Organization: Khera Communications, Inc., Rockville, MD Lines: 29 Message-ID: References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> NNTP-Posting-Host: yertle.kcilink.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: lorax.kcilink.com 1070566198 42459 216.194.193.105 (4 Dec 2003 19:29:58 GMT) X-Complaints-To: daemon@kciLink.com NNTP-Posting-Date: Thu, 4 Dec 2003 19:29:58 +0000 (UTC) User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, berkeley-unix) Cancel-Lock: sha1:Gxyqdsi8L4gsC/yxpygcCYyn8qQ= X-Virus-Scanned: by amavisd-new at kciLink.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/89 X-Sequence-Number: 4949 >>>>> "MTO" == Matthew T O'Connor writes: >> I'm running Postgres 7.4 release on FreeBSD 4.9-RELEASE. MTO> I don't run FreeBSD, so I haven't tested with FreeBSD. Recently Craig MTO> Boston reported and submitted a patch for a crash on FreeBSD, but that MTO> doesn't sound like your problem. Could be some other type of platform MTO> dependent problem. Oh lucky me. I think I found it. I compiled with -g -O and ran it under gdb, so the output is line buffered. The last thing it prints out now is this: [2003-12-04 02:11:17 PM] 3 All DBs checked in: -786419782 usec, will sleep for -1272 secs. since sleep() takes an unsigned int as its parameter, we are actually sleeping for 4294966024 seconds == 136 years. I recall reading about the negative time to test the dbs somewhere... I guess I'll get on debugging that. The time keeper in this box is pretty darned accurate otherwise (using ntpd). -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Vivek Khera, Ph.D. Khera Communications, Inc. Internet: khera@kciLink.com Rockville, MD +1-240-453-8497 AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/ From pgsql-performance-owner@postgresql.org Thu Dec 4 15:46:01 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 72C16D1B48E for ; Thu, 4 Dec 2003 19:45:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03342-10 for ; Thu, 4 Dec 2003 15:45:29 -0400 (AST) Received: from lorax.kcilink.com (lorax.kciLink.com [206.112.95.1]) by svr1.postgresql.org (Postfix) with ESMTP id CD8AED1B521 for ; Thu, 4 Dec 2003 15:44:41 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id DF9C73E5B for ; Thu, 4 Dec 2003 14:44:42 -0500 (EST) Received: from lorax.kcilink.com ([127.0.0.1]) by localhost (lorax.kcilink.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 10293-03-5 for ; Thu, 4 Dec 2003 14:44:42 -0500 (EST) Received: from lorax.kciLink.com (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id 33E593E16 for ; Thu, 4 Dec 2003 14:44:42 -0500 (EST) Received: (from news@localhost) by lorax.kciLink.com (8.12.9p2/8.12.9/Submit) id hB4JigwZ033557 for pgsql-performance@postgresql.org; Thu, 4 Dec 2003 14:44:42 -0500 (EST) (envelope-from news) To: pgsql-performance@postgresql.org Path: not-for-mail From: Vivek Khera Newsgroups: ml.postgres.performance Subject: Re: autovacuum daemon stops doing work after about an hour Date: Thu, 04 Dec 2003 14:44:41 -0500 Organization: Khera Communications, Inc., Rockville, MD Lines: 32 Message-ID: References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> NNTP-Posting-Host: yertle.kcilink.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: lorax.kcilink.com 1070567082 32569 216.194.193.105 (4 Dec 2003 19:44:42 GMT) X-Complaints-To: daemon@kciLink.com NNTP-Posting-Date: Thu, 4 Dec 2003 19:44:42 +0000 (UTC) User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, berkeley-unix) Cancel-Lock: sha1:ywPj8067Vdc9BlPRneCXMWnBHFc= X-Virus-Scanned: by amavisd-new at kciLink.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/90 X-Sequence-Number: 4950 >>>>> "MTO" == Matthew T O'Connor writes: MTO> I don't run FreeBSD, so I haven't tested with FreeBSD. Recently Craig MTO> Boston reported and submitted a patch for a crash on FreeBSD, but that some more debugging data: (gdb) print now $2 = {tv_sec = 1070565077, tv_usec = 216477} (gdb) print then $3 = {tv_sec = 1070561568, tv_usec = 668963} (gdb) print diff $4 = -5459981371352 (gdb) print sleep_secs $5 = -1272 so for some reason, instead of calculating 3508547514 as the diff, it got a hugely negative number. I'll bet it has something to do with the compiler... more debugging to follow (without -O compilation...) MTO> ---------------------------(end of broadcast)--------------------------- MTO> TIP 7: don't forget to increase your free space map settings -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Vivek Khera, Ph.D. Khera Communications, Inc. Internet: khera@kciLink.com Rockville, MD +1-240-453-8497 AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/ From pgsql-performance-owner@postgresql.org Thu Dec 4 15:55:33 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0B298D1B535 for ; Thu, 4 Dec 2003 19:55:32 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06898-09 for ; Thu, 4 Dec 2003 15:55:02 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 9F474D1B445 for ; Thu, 4 Dec 2003 15:51:00 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB4JovIt031753; Thu, 4 Dec 2003 11:50:58 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 11:50:57 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 11:50:57 -0800 Subject: Re: tuning questions From: Jack Coates To: josh@agliodbs.com Cc: pgsql-performance In-Reply-To: <200312041120.21124.josh@agliodbs.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <3FCF6AEB.908@dsvr.net> <1070565411.13923.70.camel@cletus.lyris.com> <200312041120.21124.josh@agliodbs.com> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070567455.13923.83.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 11:50:55 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: josh@agliodbs.com,pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/91 X-Sequence-Number: 4951 On Thu, 2003-12-04 at 11:20, Josh Berkus wrote: > Jack, > > > Following this, I've done: > > 2gb ram > > = > > 2,000,000,000 > > bytes > > This calculation is fun, but I really don't know where you got it from. It > seems quite baroque. What are you trying to set, exactly? Message-ID: <3FCF6AEB.908@dsvr.net> Date: Thu, 04 Dec 2003 17:12:11 +0000 From: Rob Fielding > > getting the SQL query better optimized for PG is on my todo list, but > > not something I can do right now -- this application is designed to be > > cross-platform with MS-SQL, PG, and Oracle so tweaking SQL is a touchy > > subject. > > Well, if you're queries are screwed up, no amount of .conf optimization is > going to help you much. You could criticize that PG is less adept than > some other systems at re-writing "bad queries", and you would be correct. > However, there's not much to do about that on existing systems. > > How about posting some sample code? Tracking that down in CVS and translating from C++ is going to take a while -- is there a way to get PG to log the queries it's receiving? > > > The pgavd conversation is intriguing, but I don't really understand the > > role of vacuuming. Would this be a correct statement: "PG needs to > > regularly re-evaluate the database in order to adjust itself?" I'm > > imagining that it continues to treat the table as a small one until > > vacuum informs it that the table is now large? > > Not Vacuum, Analyze. Otherwise correct. Mind you, in "regular use" where > only a small % of the table changes per hour, periodic ANALYZE is fine. > However, in "batch data transform" analyze statements need to be keyed to the > updates and/or imports. > > BTW, I send a couple of e-mails to the Lyris documentation maintainer about > updating out-of-date information about setting up PostgreSQL. I never got a > response, and I don't think my changes were made. She sits on the other side of the cube wall from me, and if I find a decent config it's going into the manual -- consider this a golden opportunity :-) -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 15:55:50 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 70EC5D1B8F5 for ; Thu, 4 Dec 2003 19:55:49 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08063-03 for ; Thu, 4 Dec 2003 15:55:19 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 316C8D1B4A6 for ; Thu, 4 Dec 2003 15:51:21 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ARzVR-0004Ia-00; Thu, 04 Dec 2003 21:51:21 +0200 Date: Thu, 4 Dec 2003 21:51:21 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031204195121.GA16497@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031204142320.3c021100.threshar@torgo.978.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031204142320.3c021100.threshar@torgo.978.org> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/92 X-Sequence-Number: 4952 On Thu, Dec 04, 2003 at 02:23:20PM -0500, Jeff wrote: > > Most interesting is, that insert takes 0.004 seconds in average, but > > update takes 0.255 seconds in average. Processing of 24000 records > > took around 1 hour 20 minutes. > > Do you have an index on recid? Yes, this is primary key of table1 > and did you vacuum analyze after you loaded up the data? No, this is running as nightly cronjob. All tests were done during one day, so no vacuum was done. > "IN" queries are terribly slow on versions before 7.4 OK, this is useful to know :) > > Why is UPDATE so slow compared to INSERT? I would expect more or less > > similar performance, or slower on insert since table2 has four indexes > > in addition to primary key, table1 has only primary key, which is used > > on update. Am i doing something wrong or is this normal? > Remember, UPDATE has to do all the work of select and more. > > And if you have 4 indexes those will also add to the time (Since it has > to update/add them to the tree) My primary concern is performance difference between INSERT and UPDATE in my first tests. There i did select from table1, fetched record, processed it and inserted into table2. Then updated status of fetched record in table1. Repeated in cycle as long as fetch returned record. Average time for INSERT was 0.004 seconds, average time for UPDATE 0.255 seconds. Update was done as "update table1 set status = 'done' where recid = xxxx". As far as i understand, this type of simple update should be faster, compared to INSERT into table with four indexes, but in my case it is more than 60 times slower. Why?? My second tests were done with temporary table and update query as: "UPDATE table1 SET Status = 'done' WHERE recid IN (SELECT recid FROM temptable)". It is still slower than INSERT, but more or less acceptable. Compared to my first tests overall processing time dropped from 1 hour and 20 minutes to 16 minutes. So, my question remains - why is simple update more than 60 times slower, compared to INSERT? Any ideas? -- Ivar Zarans From pgsql-performance-owner@postgresql.org Thu Dec 4 16:50:54 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 34795D1B441 for ; Thu, 4 Dec 2003 20:28:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14901-06 for ; Thu, 4 Dec 2003 16:28:22 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 76791D1B453 for ; Thu, 4 Dec 2003 16:28:20 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hB4KSKbI057925 for ; Thu, 4 Dec 2003 20:28:20 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hB4JwtAg052657 for pgsql-performance@postgresql.org; Thu, 4 Dec 2003 19:58:55 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Slow UPADTE, compared to INSERT Date: Thu, 04 Dec 2003 11:59:01 -0800 Organization: Hub.Org Networking Services Lines: 16 Message-ID: References: <20031204185751.GA15893@alcaron.ee> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: <20031204185751.GA15893@alcaron.ee> To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/95 X-Sequence-Number: 4955 Ivar Zarans wrote: > > I am experiencing strange behaviour, where simple UPDATE of one field is > very slow, compared to INSERT into table with multiple indexes. I have > two tables - one with raw data records (about 24000), where one field In Postgres and any other DB that uses MVCC (multi-version concurrency), UPDATES will always be slower than INSERTS. With MVCC, what the DB does is makes a copy of the record, updates that record and then invalidates the previous record. This allows maintains a consistent view for anybody who's reading the DB and also avoids the requirement of row locks. If you have to use UPDATE, make sure (1) your UPDATE WHERE clause is properly indexed and (2) you are running ANALYZE/VACUUM periodically so the query planner can optimize for your UPDATE statements. From pgsql-performance-owner@postgresql.org Thu Dec 4 16:55:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 280A2D1B44C for ; Thu, 4 Dec 2003 20:24:13 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14240-08 for ; Thu, 4 Dec 2003 16:23:42 -0400 (AST) Received: from anchor-post-31.mail.demon.net (anchor-post-31.mail.demon.net [194.217.242.89]) by svr1.postgresql.org (Postfix) with ESMTP id D73B4D1B445 for ; Thu, 4 Dec 2003 16:23:40 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-31.mail.demon.net with esmtp (Exim 3.35 #1) id 1AS00j-0001qN-0V; Thu, 04 Dec 2003 20:23:41 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 3B0F517336; Thu, 4 Dec 2003 20:23:39 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id A787C16C60; Thu, 4 Dec 2003 20:23:37 +0000 (GMT) From: Richard Huxton To: Ivar Zarans , pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Date: Thu, 4 Dec 2003 20:23:36 +0000 User-Agent: KMail/1.5 References: <20031204185751.GA15893@alcaron.ee> <20031204142320.3c021100.threshar@torgo.978.org> <20031204195121.GA16497@alcaron.ee> In-Reply-To: <20031204195121.GA16497@alcaron.ee> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312042023.37050.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/97 X-Sequence-Number: 4957 On Thursday 04 December 2003 19:51, Ivar Zarans wrote: > > My second tests were done with temporary table and update query as: > "UPDATE table1 SET Status = 'done' WHERE recid IN (SELECT recid FROM > temptable)". It is still slower than INSERT, but more or less > acceptable. Compared to my first tests overall processing time dropped > from 1 hour and 20 minutes to 16 minutes. Ah - it's probably not the update but the IN. You can rewrite it using PG's non-standard FROM: UPDATE t1 SET status='done' FROM t_tmp WHERE t1.rec_id = t_tmp.rec_id; Now that doesn't explain why the update is taking so long. One fifth of a second is extremely slow. Are you certain that the index is being used? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Dec 4 16:53:29 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0C370D1B443 for ; Thu, 4 Dec 2003 20:27:57 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14628-06 for ; Thu, 4 Dec 2003 16:27:26 -0400 (AST) Received: from anchor-post-30.mail.demon.net (anchor-post-30.mail.demon.net [194.217.242.88]) by svr1.postgresql.org (Postfix) with ESMTP id 29C31D1B476 for ; Thu, 4 Dec 2003 16:27:25 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-30.mail.demon.net with esmtp (Exim 3.35 #1) id 1AS04L-0002XU-0U; Thu, 04 Dec 2003 20:27:25 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 924F117336; Thu, 4 Dec 2003 20:27:24 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id CEEFF17331; Thu, 4 Dec 2003 20:27:22 +0000 (GMT) From: Richard Huxton To: Jack Coates , josh@agliodbs.com Subject: Re: tuning questions Date: Thu, 4 Dec 2003 20:27:22 +0000 User-Agent: KMail/1.5 Cc: pgsql-performance References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041120.21124.josh@agliodbs.com> <1070567455.13923.83.camel@cletus.lyris.com> In-Reply-To: <1070567455.13923.83.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312042027.22223.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/96 X-Sequence-Number: 4956 On Thursday 04 December 2003 19:50, Jack Coates wrote: > > I'm trying to set Postgres's shared memory usage in a fashion that > allows it to return requested results quickly. Unfortunately, none of > these changes allow PG to use more than a little under 300M RAM. > vacuumdb --analyze is now taking an inordinate amount of time as well > (40 minutes and counting), so that change needs to be rolled back. You don't want PG to use all your RAM, it's designed to let the underlying OS do a lot of caching for it. Probably worth having a look at vmstat/iostat and see if it's saturating on I/O. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Dec 4 16:44:09 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 29F7AD1B478 for ; Thu, 4 Dec 2003 20:38:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 17654-01 for ; Thu, 4 Dec 2003 16:38:23 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id B2D08D1B44C for ; Thu, 4 Dec 2003 16:38:20 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB4KblIt007660; Thu, 4 Dec 2003 12:37:47 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 12:37:47 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 12:37:47 -0800 Subject: Re: tuning questions From: Jack Coates To: Richard Huxton Cc: josh@agliodbs.com, pgsql-performance In-Reply-To: <200312042027.22223.dev@archonet.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041120.21124.josh@agliodbs.com> <1070567455.13923.83.camel@cletus.lyris.com> <200312042027.22223.dev@archonet.com> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070570264.13923.88.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 12:37:45 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: dev@archonet.com, josh@agliodbs.com, pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/93 X-Sequence-Number: 4953 On Thu, 2003-12-04 at 12:27, Richard Huxton wrote: > On Thursday 04 December 2003 19:50, Jack Coates wrote: > > > > I'm trying to set Postgres's shared memory usage in a fashion that > > allows it to return requested results quickly. Unfortunately, none of > > these changes allow PG to use more than a little under 300M RAM. > > vacuumdb --analyze is now taking an inordinate amount of time as well > > (40 minutes and counting), so that change needs to be rolled back. > > You don't want PG to use all your RAM, it's designed to let the underlying OS > do a lot of caching for it. Probably worth having a look at vmstat/iostat and > see if it's saturating on I/O. latest changes: shared_buffers = 35642 max_fsm_relations = 1000 max_fsm_pages = 10000 wal_buffers = 64 sort_mem = 32768 vacuum_mem = 32768 effective_cache_size = 10000 /proc/sys/kernel/shmmax = 500000000 IO is active, but hardly saturated. CPU load is hefty though, load average is at 4 now. procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 0 2 1 2808 11436 39616 1902988 0 0 240 896 765 469 2 11 87 0 2 1 2808 11432 39616 1902988 0 0 244 848 768 540 4 3 93 0 2 1 2808 11432 39616 1902984 0 0 204 876 788 507 3 4 93 0 2 1 2808 11432 39616 1902984 0 0 360 416 715 495 4 1 96 0 2 1 2808 11432 39616 1902984 0 0 376 328 689 441 2 1 97 0 2 0 2808 11428 39616 1902976 0 0 464 360 705 479 2 1 97 0 2 1 2808 11428 39616 1902976 0 0 432 380 718 547 3 1 97 0 2 1 2808 11428 39616 1902972 0 0 440 372 742 512 1 3 96 0 2 1 2808 11428 39616 1902972 0 0 416 364 711 504 3 1 96 0 2 1 2808 11424 39616 1902972 0 0 456 492 743 592 2 1 97 0 2 1 2808 11424 39616 1902972 0 0 440 352 707 494 2 1 97 0 2 1 2808 11424 39616 1902972 0 0 456 360 709 494 2 2 97 0 2 1 2808 11436 39616 1902968 0 0 536 516 807 708 3 2 94 -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 16:47:11 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6921BD1B49A for ; Thu, 4 Dec 2003 20:47:07 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19863-01 for ; Thu, 4 Dec 2003 16:46:37 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 8211FD1C9ED for ; Thu, 4 Dec 2003 16:43:25 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1AS0Jq-0004Y1-00; Thu, 04 Dec 2003 22:43:26 +0200 Date: Thu, 4 Dec 2003 22:43:26 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031204204326.GA17083@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031204142320.3c021100.threshar@torgo.978.org> <20031204195121.GA16497@alcaron.ee> <200312042023.37050.dev@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312042023.37050.dev@archonet.com> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/94 X-Sequence-Number: 4954 On Thu, Dec 04, 2003 at 08:23:36PM +0000, Richard Huxton wrote: > Ah - it's probably not the update but the IN. You can rewrite it using PG's > non-standard FROM: > > UPDATE t1 SET status='done' FROM t_tmp WHERE t1.rec_id = t_tmp.rec_id; Thanks for the hint. I'll try this. > Now that doesn't explain why the update is taking so long. One fifth of a > second is extremely slow. Are you certain that the index is being used? Explain shows following output: explain update table1 set status = 'PROC' where recid = '199901'; Index Scan using table1_pkey on table1 (cost=0.00..6.01 rows=1 width=198) Index Cond: (recid = 199901::bigint) (2 rows) -- Ivar Zarans From pgsql-performance-owner@postgresql.org Thu Dec 4 17:00:03 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A0396D1B443 for ; Thu, 4 Dec 2003 20:53:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20708-05 for ; Thu, 4 Dec 2003 16:53:15 -0400 (AST) Received: from dezeut.zeut.net (ool-4352919e.dyn.optonline.net [67.82.145.158]) by svr1.postgresql.org (Postfix) with ESMTP id BBAB6D1B47A for ; Thu, 4 Dec 2003 16:52:57 -0400 (AST) Received: from dezeut.zeut.net (localhost.localdomain [127.0.0.1]) by dezeut.zeut.net (8.12.8/8.12.8) with ESMTP id hB4Kqp7N021686; Thu, 4 Dec 2003 15:52:51 -0500 Received: (from apache@localhost) by dezeut.zeut.net (8.12.8/8.12.8/Submit) id hB4KqpDP021684; Thu, 4 Dec 2003 15:52:51 -0500 X-Authentication-Warning: dezeut.zeut.net: apache set sender to matthew@zeut.net using -f Received: from 66.106.27.14 (SquirrelMail authenticated user dbmailtest) by matth.zeut.net with HTTP; Thu, 4 Dec 2003 15:52:51 -0500 (EST) Message-ID: <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> Date: Thu, 4 Dec 2003 15:52:51 -0500 (EST) Subject: Re: autovacuum daemon stops doing work after about an hour From: "Matthew T. O'Connor" To: In-Reply-To: References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> X-Priority: 3 Importance: Normal Cc: X-Mailer: SquirrelMail (version 1.2.11) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/98 X-Sequence-Number: 4958 >>>>>> "MTO" == Matthew T O'Connor writes: > > MTO> I don't run FreeBSD, so I haven't tested with FreeBSD. Recently > Craig MTO> Boston reported and submitted a patch for a crash on FreeBSD, > but that > > some more debugging data: > > (gdb) print now > $2 = {tv_sec = 1070565077, tv_usec = 216477} > (gdb) print then > $3 = {tv_sec = 1070561568, tv_usec = 668963} > (gdb) print diff > $4 = -5459981371352 > (gdb) print sleep_secs > $5 = -1272 > > so for some reason, instead of calculating 3508547514 as the diff, it > got a hugely negative number. > > I'll bet it has something to do with the compiler... more debugging to > follow (without -O compilation...) Could this be the recently reported bug where time goes backwards on FreeBSD? Can anyone who knows more about this problem chime in, I know it was recently discussed on Hackers. The simple fix is to just make sure it's a positive number. If not, then just sleep for some small positive amount. I can make a patch for this, probably sometime this weekend. Thanks for tracking this down. From pgsql-performance-owner@postgresql.org Thu Dec 4 17:29:14 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7DE7FD1B4C3 for ; Thu, 4 Dec 2003 21:29:12 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 26747-10 for ; Thu, 4 Dec 2003 17:28:43 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 8D82BD1B49D for ; Thu, 4 Dec 2003 17:28:40 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hB4LRauH007806; Thu, 4 Dec 2003 14:27:36 -0700 (MST) Date: Thu, 4 Dec 2003 14:10:41 -0700 (MST) From: "scott.marlowe" To: Jack Coates Cc: Richard Huxton , , pgsql-performance Subject: Re: tuning questions In-Reply-To: <1070570264.13923.88.camel@cletus.lyris.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/103 X-Sequence-Number: 4963 On Thu, 4 Dec 2003, Jack Coates wrote: > On Thu, 2003-12-04 at 12:27, Richard Huxton wrote: > > On Thursday 04 December 2003 19:50, Jack Coates wrote: > > > > > > I'm trying to set Postgres's shared memory usage in a fashion that > > > allows it to return requested results quickly. Unfortunately, none of > > > these changes allow PG to use more than a little under 300M RAM. > > > vacuumdb --analyze is now taking an inordinate amount of time as well > > > (40 minutes and counting), so that change needs to be rolled back. > > > > You don't want PG to use all your RAM, it's designed to let the underlying OS > > do a lot of caching for it. Probably worth having a look at vmstat/iostat and > > see if it's saturating on I/O. > > latest changes: > shared_buffers = 35642 > max_fsm_relations = 1000 > max_fsm_pages = 10000 > wal_buffers = 64 > sort_mem = 32768 > vacuum_mem = 32768 > effective_cache_size = 10000 > > /proc/sys/kernel/shmmax = 500000000 > > IO is active, but hardly saturated. CPU load is hefty though, load > average is at 4 now. Postgresql is busily managing a far too large shared buffer. Let the kernel do that. Postgresql's shared buffers should be bug enough to hold as much of the current working set as it can, up to about 25% or so of the servers memory, or 512Meg, whichever comes first. Unless a single query will actually use all of the buffer at once, you're not likely to see an improvement. Also, your effective cache size is really small. On a typical Postgresql server with 2 gigs of ram, you'll have about 1 to 1.5 gigs as kernel cache and buffer, and if it's dedicated to postgresql, then the effective cache setting for 1 gig would be 131072 (assuming 8k pages). If you're updating a lot of tuples without vacuums, you'll likely want to up your fsm settings. Note you can change things like sort_mem, effective_cache_size and random_page_cost on the fly (but not buffers, they're allocated at startup, nor fsm, they are as well.) so, if you're gonna have one huge honkin query that needs to sort a hundred megs at a time, but you'd rather not up your sort memory that high (sort mem is PER SORT, not per backend or per database, so it can get out of hand quickly) then you can just set sort_mem=128000; before throwing out the big queries that need all the sort. From pgsql-performance-owner@postgresql.org Thu Dec 4 17:20:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3806AD1B491 for ; Thu, 4 Dec 2003 21:20:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 26242-05 for ; Thu, 4 Dec 2003 17:20:23 -0400 (AST) Received: from yertle.kcilink.com (yertle.kcilink.com [216.194.193.105]) by svr1.postgresql.org (Postfix) with ESMTP id D9481D1B47A for ; Thu, 4 Dec 2003 17:20:20 -0400 (AST) Received: by yertle.kcilink.com (Postfix, from userid 100) id 5AF5A2178A; Thu, 4 Dec 2003 16:20:22 -0500 (EST) From: Vivek Khera MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <16335.42262.196435.193875@yertle.int.kciLink.com> Date: Thu, 4 Dec 2003 16:20:22 -0500 To: "Matthew T. O'Connor" Cc: Subject: Re: autovacuum daemon stops doing work after about an hour In-Reply-To: <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> X-Mailer: VM 7.17 under 21.4 (patch 14) "Reasonable Discussion" XEmacs Lucid X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/99 X-Sequence-Number: 4959 >>>>> "MTO" == Matthew T O'Connor writes: MTO> Could this be the recently reported bug where time goes backwards on MTO> FreeBSD? Can anyone who knows more about this problem chime in, I know it MTO> was recently discussed on Hackers. Time does not go backwards -- the now and then variables are properly incrementing in time as you see from the debugging output. The error appears to be with the computation of the "diff". It is either a C programming error, or a compiler error. I'm not a C "cop" so I can't tell you which it is. Witness this program, below, compiled as "cc -g -o t t.c" and the output here: % ./t seconds = 3509 seconds1 = 3509000000 useconds = -452486 stepped diff = 3508547514 seconds2 = -785967296 seconds3 = 3509000000 diff = -786419782 long long diff = 3508547514 % apperantly, if you compute (now.tv_sec - then.tv_sec) * 1000000 all at once, it overflows since the RHS is all computed using longs rather than long longs. Fix is to cast at least one of the values to long long on the RHS, as in the computation of seconds3 below. compare that to the computation of seconds2 and you'll see that this is the cause. I'd be curious to see the output of this program on other platforms and other compilers. I'm using gcc 2.95.4 as shipped with FreeBSD 4.8+. That all being said, you should never sleep less than the base time, and never for more than a max amount, perhaps 1 hour? --cut here-- #include #include int main() { struct timeval now, then; long long diff = 0; long long seconds, seconds1, seconds2, seconds3, useconds; now.tv_sec = 1070565077L; now.tv_usec = 216477L; then.tv_sec = 1070561568L; then.tv_usec = 668963L; seconds = now.tv_sec - then.tv_sec; printf("seconds = %lld\n",seconds); seconds1 = seconds * 1000000; printf("seconds1 = %lld\n",seconds1); useconds = now.tv_usec - then.tv_usec; printf("useconds = %lld\n",useconds); diff = seconds1 + useconds; printf("stepped diff = %lld\n",diff); /* this appears to be the culprit... it should be same as seconds1 */ seconds2 = (now.tv_sec - then.tv_sec) * 1000000; printf("seconds2 = %lld\n",seconds2); /* seems we need to cast long's to long long's for this computation */ seconds3 = ((long long)now.tv_sec - (long long)then.tv_sec) * 1000000; printf("seconds3 = %lld\n",seconds3); diff = (now.tv_sec - then.tv_sec) * 1000000 + (now.tv_usec - then.tv_usec); printf ("diff = %lld\n",diff); diff = ((long long)now.tv_sec - (long long)then.tv_sec) * 1000000 + (now.tv_usec - then.tv_usec); printf ("long long diff = %lld\n",diff); exit(0); } --cut here-- From pgsql-performance-owner@postgresql.org Thu Dec 4 17:22:40 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5A0AFD1B4AE for ; Thu, 4 Dec 2003 21:22:38 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27632-01 for ; Thu, 4 Dec 2003 17:22:09 -0400 (AST) Received: from yertle.kcilink.com (yertle.kcilink.com [216.194.193.105]) by svr1.postgresql.org (Postfix) with ESMTP id 921C3D1B47A for ; Thu, 4 Dec 2003 17:22:07 -0400 (AST) Received: by yertle.kcilink.com (Postfix, from userid 100) id 60C142178A; Thu, 4 Dec 2003 16:22:09 -0500 (EST) From: Vivek Khera MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <16335.42369.252957.290956@yertle.int.kciLink.com> Date: Thu, 4 Dec 2003 16:22:09 -0500 To: "Matthew T. O'Connor" Cc: Subject: Re: autovacuum daemon stops doing work after about an hour In-Reply-To: <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> X-Mailer: VM 7.17 under 21.4 (patch 14) "Reasonable Discussion" XEmacs Lucid X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/100 X-Sequence-Number: 4960 Actually, you can simplify the fix thusly: diff = (long long)(now.tv_sec - then.tv_sec) * 1000000 + (now.tv_usec - then.tv_usec); From pgsql-performance-owner@postgresql.org Thu Dec 4 17:25:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 15FA5D1B4C1 for ; Thu, 4 Dec 2003 21:25:14 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 26873-09 for ; Thu, 4 Dec 2003 17:24:44 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 1A3DDD1B4B9 for ; Thu, 4 Dec 2003 17:24:42 -0400 (AST) Received: from [66.219.92.2] (HELO temoku) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4023430; Thu, 04 Dec 2003 13:25:33 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Jack Coates , Richard Huxton Subject: Re: tuning questions Date: Thu, 4 Dec 2003 13:24:37 -0800 User-Agent: KMail/1.4.3 Cc: pgsql-performance References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042027.22223.dev@archonet.com> <1070570264.13923.88.camel@cletus.lyris.com> In-Reply-To: <1070570264.13923.88.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200312041324.37889.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/101 X-Sequence-Number: 4961 Jack, > latest changes: > shared_buffers =3D 35642 This is fine, it's about 14% of available RAM. Though the way you calculat= ed=20 it still confuses me. It's not complicated; it should be between 6% and 1= 5%=20 of available RAM; since you're doing a data-transformation DB, yours should= =20 be toward the high end.=20 > max_fsm_relations =3D 1000 > max_fsm_pages =3D 10000 You want to raise this a whole lot if your data transformations involve lar= ge=20 delete or update batches. I'd suggest running "vacuum analyze verbose"= =20 between steps to see how many dead pages you're accumulating. > wal_buffers =3D 64 > sort_mem =3D 32768 > vacuum_mem =3D 32768 > effective_cache_size =3D 10000 This is way the heck too low. it's supposed to be the size of all availabl= e=20 RAM; I'd set it to 2GB*65% as a start. > IO is active, but hardly saturated. CPU load is hefty though, load > average is at 4 now. Unless you're doing huge statistical aggregates (like radar charts), or hea= vy=20 numerical calculations-by-query, high CPU and idle I/O usually indicates a= =20 really bad query, like badly mismatched data types on a join or unconstrain= ed=20 joins or overblown formatting-by-query. --=20 -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Dec 4 17:26:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EBD27D1B4A6 for ; Thu, 4 Dec 2003 21:26:19 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28396-03 for ; Thu, 4 Dec 2003 17:25:49 -0400 (AST) Received: from lerami.lerctr.org (lerami.lerctr.org [207.158.72.11]) by svr1.postgresql.org (Postfix) with ESMTP id 073D7D1B491 for ; Thu, 4 Dec 2003 17:25:43 -0400 (AST) Received: from lerlaptop-red.iadfw.net ([207.136.3.72]) by lerami.lerctr.org with asmtp (Exim 4.30) id 1AS0yf-0004Vv-85; Thu, 04 Dec 2003 15:25:37 -0600 Date: Thu, 04 Dec 2003 15:25:36 -0600 From: Larry Rosenman To: Vivek Khera , "Matthew T. O'Connor" Cc: pgsql-performance@postgresql.org Subject: Re: autovacuum daemon stops doing work after about an Message-ID: <401840000.1070573136@lerlaptop-red.iadfw.net> In-Reply-To: <16335.42262.196435.193875@yertle.int.kciLink.com> References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> <16335.42262.196435.193875@yertle.int.kciLink.com> X-Mailer: Mulberry/3.1.0 (Linux/x86) X-PGP-Info: All other keys are old/dead. X-PGP-Key: 0x3c49bdd6 X-PGP-Fingerprint: D0D1 3C11 F42F 6B29 FA67 6BF3 AD13 4685 3C49 BDD6 MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="==========1B86D33B13306EE04F08==========" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/102 X-Sequence-Number: 4962 --==========1B86D33B13306EE04F08========== Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: quoted-printable Content-Disposition: inline --On Thursday, December 04, 2003 16:20:22 -0500 Vivek Khera=20 wrote: >>>>>> "MTO" =3D=3D Matthew T O'Connor writes: > > MTO> Could this be the recently reported bug where time goes backwards on > MTO> FreeBSD? Can anyone who knows more about this problem chime in, I > know it MTO> was recently discussed on Hackers. > > > Time does not go backwards -- the now and then variables are properly > incrementing in time as you see from the debugging output. > > The error appears to be with the computation of the "diff". It is > either a C programming error, or a compiler error. I'm not a C "cop" > so I can't tell you which it is. > > Witness this program, below, compiled as "cc -g -o t t.c" and the > output here: > > % ./t > seconds =3D 3509 > seconds1 =3D 3509000000 > useconds =3D -452486 > stepped diff =3D 3508547514 > seconds2 =3D -785967296 > seconds3 =3D 3509000000 > diff =3D -786419782 > long long diff =3D 3508547514 > % > > apperantly, if you compute (now.tv_sec - then.tv_sec) * 1000000 all at > once, it overflows since the RHS is all computed using longs rather > than long longs. Fix is to cast at least one of the values to long > long on the RHS, as in the computation of seconds3 below. compare > that to the computation of seconds2 and you'll see that this is the > cause. > > I'd be curious to see the output of this program on other platforms > and other compilers. I'm using gcc 2.95.4 as shipped with FreeBSD > 4.8+. this is with the UnixWare compiler: $ cc -O -o testvk testvk.c $ ./testvk seconds =3D 3509 seconds1 =3D 3509000000 useconds =3D -452486 stepped diff =3D 3508547514 seconds2 =3D -785967296 seconds3 =3D 3509000000 diff =3D -786419782 long long diff =3D 3508547514 $ I think this is a C bug. > > That all being said, you should never sleep less than the base time, > and never for more than a max amount, perhaps 1 hour? > > > --cut here-- ># include ># include > > int > main() > { > struct timeval now, then; > long long diff =3D 0; > long long seconds, seconds1, seconds2, seconds3, useconds; > > now.tv_sec =3D 1070565077L; > now.tv_usec =3D 216477L; > > then.tv_sec =3D 1070561568L; > then.tv_usec =3D 668963L; > > seconds =3D now.tv_sec - then.tv_sec; > printf("seconds =3D %lld\n",seconds); > seconds1 =3D seconds * 1000000; > printf("seconds1 =3D %lld\n",seconds1); > useconds =3D now.tv_usec - then.tv_usec; > printf("useconds =3D %lld\n",useconds); > > diff =3D seconds1 + useconds; > printf("stepped diff =3D %lld\n",diff); > > /* this appears to be the culprit... it should be same as seconds1 */ > seconds2 =3D (now.tv_sec - then.tv_sec) * 1000000; > printf("seconds2 =3D %lld\n",seconds2); > > /* seems we need to cast long's to long long's for this computation */ > seconds3 =3D ((long long)now.tv_sec - (long long)then.tv_sec) * 1000000; > printf("seconds3 =3D %lld\n",seconds3); > > > diff =3D (now.tv_sec - then.tv_sec) * 1000000 + (now.tv_usec - > then.tv_usec); printf ("diff =3D %lld\n",diff); > > diff =3D ((long long)now.tv_sec - (long long)then.tv_sec) * 1000000 + > (now.tv_usec - then.tv_usec); printf ("long long diff =3D %lld\n",diff); > > exit(0); > } > > > --cut here-- > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) > --=20 Larry Rosenman http://www.lerctr.org/~ler Phone: +1 972-414-9812 E-Mail: ler@lerctr.org US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749 --==========1B86D33B13306EE04F08========== Content-Type: application/pgp-signature Content-Transfer-Encoding: 7bit -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (FreeBSD) iD8DBQE/z6ZQrRNGhTxJvdYRAizdAKCJrroU/PruGlADjJEybSh+IhRHwQCffnpM rZH61B7ilXl1WNXE+fvLmCA= =SdUH -----END PGP SIGNATURE----- --==========1B86D33B13306EE04F08==========-- From pgsql-performance-owner@postgresql.org Thu Dec 4 17:37:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DC2BED1B4A6 for ; Thu, 4 Dec 2003 21:37:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28408-09 for ; Thu, 4 Dec 2003 17:37:13 -0400 (AST) Received: from yertle.kcilink.com (yertle.kcilink.com [216.194.193.105]) by svr1.postgresql.org (Postfix) with ESMTP id 0FDDCD1B484 for ; Thu, 4 Dec 2003 17:37:11 -0400 (AST) Received: by yertle.kcilink.com (Postfix, from userid 100) id 5D1622178A; Thu, 4 Dec 2003 16:37:12 -0500 (EST) From: Vivek Khera MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <16335.43272.273411.653445@yertle.int.kciLink.com> Date: Thu, 4 Dec 2003 16:37:12 -0500 To: pgsql-performance@postgresql.org Subject: Re: autovacuum daemon stops doing work after about an In-Reply-To: <401840000.1070573136@lerlaptop-red.iadfw.net> References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <2348.66.106.27.14.1070571171.squirrel@matth.zeut.net> <16335.42262.196435.193875@yertle.int.kciLink.com> <401840000.1070573136@lerlaptop-red.iadfw.net> X-Mailer: VM 7.17 under 21.4 (patch 14) "Reasonable Discussion" XEmacs Lucid X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/104 X-Sequence-Number: 4964 >>>>> "LR" == Larry Rosenman writes: >> I'd be curious to see the output of this program on other platforms >> and other compilers. I'm using gcc 2.95.4 as shipped with FreeBSD >> 4.8+. LR> this is with the UnixWare compiler: LR> $ cc -O -o testvk testvk.c LR> $ ./testvk LR> seconds = 3509 LR> seconds1 = 3509000000 LR> useconds = -452486 LR> stepped diff = 3508547514 LR> seconds2 = -785967296 LR> seconds3 = 3509000000 LR> diff = -786419782 LR> long long diff = 3508547514 LR> $ LR> I think this is a C bug. Upon further reflection, I think so to. The entire RHS is long's so the arithmetic is done in longs, then assigned to a long long when done (after things have overflowed). Forcing any one of the RHS values to be long long causes the arithmetic to all be done using long longs, and then you get the numbers you expect. I think you only notice this in autovacuum when it takes a long time to complete the work, like my example of about 3500 seconds. From pgsql-performance-owner@postgresql.org Thu Dec 4 18:13:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6EBA7D1B4A7 for ; Thu, 4 Dec 2003 22:13:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 35109-02 for ; Thu, 4 Dec 2003 18:13:13 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 8D7CDD1B443 for ; Thu, 4 Dec 2003 18:13:10 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1AS1ii-0004xy-00; Fri, 05 Dec 2003 00:13:12 +0200 Date: Fri, 5 Dec 2003 00:13:12 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031204221312.GA18874@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031204142320.3c021100.threshar@torgo.978.org> <20031204195121.GA16497@alcaron.ee> <200312042023.37050.dev@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312042023.37050.dev@archonet.com> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/105 X-Sequence-Number: 4965 On Thu, Dec 04, 2003 at 08:23:36PM +0000, Richard Huxton wrote: > Ah - it's probably not the update but the IN. You can rewrite it using PG's > non-standard FROM: > > UPDATE t1 SET status='done' FROM t_tmp WHERE t1.rec_id = t_tmp.rec_id; This was one *very useful* hint! Using this method i got my processing time of 24000 records down to around 3 minutes 10 seconds. Comparing with initial 1 hour 20 minutes and then 16 minutes, this is impressive improvement! > Now that doesn't explain why the update is taking so long. One fifth of a > second is extremely slow. Are you certain that the index is being used? I posted results of "EXPLAIN" in my previous message. Meanwhile i tried to update just one record, using "psql". Also tried out "EXPLAIN ANALYZE". This way i did not see any big delay - total runtime for one update was around 1 msec. I am confused - has slowness of UPDATE something to do with Python and PyPgSQL, since "psql" seems to have no delay whatsoever? Or is this related to using two cursors, one for select results and other for update? Even if this is related to Python or cursors, how am i getting so big speed improvement only by using different query? -- Ivar From pgsql-performance-owner@postgresql.org Thu Dec 4 18:38:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F285DD1B458 for ; Thu, 4 Dec 2003 22:38:01 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 38598-02 for ; Thu, 4 Dec 2003 18:37:33 -0400 (AST) Received: from anchor-post-30.mail.demon.net (anchor-post-30.mail.demon.net [194.217.242.88]) by svr1.postgresql.org (Postfix) with ESMTP id 1E9EDD1B464 for ; Thu, 4 Dec 2003 18:37:31 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-30.mail.demon.net with esmtp (Exim 3.35 #1) id 1AS26G-0002VJ-0U; Thu, 04 Dec 2003 22:37:32 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 4EE0A16CB0; Thu, 4 Dec 2003 22:37:31 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 7483316732; Thu, 4 Dec 2003 22:37:29 +0000 (GMT) From: Richard Huxton To: William Yu , pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Date: Thu, 4 Dec 2003 22:37:28 +0000 User-Agent: KMail/1.5 References: <20031204185751.GA15893@alcaron.ee> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312042237.28378.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/106 X-Sequence-Number: 4966 On Thursday 04 December 2003 19:59, William Yu wrote: > Ivar Zarans wrote: > > I am experiencing strange behaviour, where simple UPDATE of one field is > > very slow, compared to INSERT into table with multiple indexes. I have > > two tables - one with raw data records (about 24000), where one field > > In Postgres and any other DB that uses MVCC (multi-version concurrency), > UPDATES will always be slower than INSERTS. With MVCC, what the DB does > is makes a copy of the record, updates that record and then invalidates > the previous record. [snip] Yes, but he's seeing 0.25secs to update one row - that's something odd. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Dec 4 18:46:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 39AF7D1B482 for ; Thu, 4 Dec 2003 22:46:06 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37999-07 for ; Thu, 4 Dec 2003 18:45:36 -0400 (AST) Received: from anchor-post-33.mail.demon.net (anchor-post-33.mail.demon.net [194.217.242.91]) by svr1.postgresql.org (Postfix) with ESMTP id DD90FD1B46E for ; Thu, 4 Dec 2003 18:45:33 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-33.mail.demon.net with esmtp (Exim 3.35 #1) id 1AS2Dz-000FAX-0X; Thu, 04 Dec 2003 22:45:33 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 9775816CB0; Thu, 4 Dec 2003 22:45:25 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 0442B16C65; Thu, 4 Dec 2003 22:45:22 +0000 (GMT) From: Richard Huxton To: Ivar Zarans , pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Date: Thu, 4 Dec 2003 22:45:21 +0000 User-Agent: KMail/1.5 References: <20031204185751.GA15893@alcaron.ee> <200312042023.37050.dev@archonet.com> <20031204221312.GA18874@alcaron.ee> In-Reply-To: <20031204221312.GA18874@alcaron.ee> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312042245.21899.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/107 X-Sequence-Number: 4967 On Thursday 04 December 2003 22:13, Ivar Zarans wrote: > On Thu, Dec 04, 2003 at 08:23:36PM +0000, Richard Huxton wrote: > > Ah - it's probably not the update but the IN. You can rewrite it using > > PG's non-standard FROM: > > > > UPDATE t1 SET status='done' FROM t_tmp WHERE t1.rec_id = t_tmp.rec_id; > > This was one *very useful* hint! Using this method i got my processing > time of 24000 records down to around 3 minutes 10 seconds. Comparing > with initial 1 hour 20 minutes and then 16 minutes, this is impressive > improvement! Be aware, this is specific to PG - I'm not aware of this construction working on any other DB. Three minutes still doesn't sound brilliant, but that could be tuning issues. > > Now that doesn't explain why the update is taking so long. One fifth of a > > second is extremely slow. Are you certain that the index is being used? > > I posted results of "EXPLAIN" in my previous message. Meanwhile i tried > to update just one record, using "psql". Also tried out "EXPLAIN > ANALYZE". This way i did not see any big delay - total runtime for one > update was around 1 msec. Yep - the explain looked fine. If you run EXPLAIN ANALYSE it will give you timings too (actual timings will be slightly less than reported ones since PG won't be timing/reporting). > I am confused - has slowness of UPDATE something to do with Python and > PyPgSQL, since "psql" seems to have no delay whatsoever? Or is this > related to using two cursors, one for select results and other for > update? Even if this is related to Python or cursors, how am i getting > so big speed improvement only by using different query? Hmm - you didn't mention cursors. If this was a problem with PyPgSQL in general I suspect we'd know about it by now. It could however be some cursor-related issue. In general, you're probably better off trying to do updates/inserts as a single statement and letting PG manage things rather than processing one row at a time. If you've got the time, try putting together a small test-script with some dummy data and see if it's reproducible. I'm sure the other Python users would be interested in seeing where the problem is. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Dec 4 19:00:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 65E47D1B464 for ; Thu, 4 Dec 2003 23:00:18 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 43930-01 for ; Thu, 4 Dec 2003 18:59:49 -0400 (AST) Received: from main.wiredfool.com (dsl231-055-035.sea1.dsl.speakeasy.net [216.231.55.35]) by svr1.postgresql.org (Postfix) with ESMTP id 40A99D1B43F for ; Thu, 4 Dec 2003 18:59:46 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id D8F1F2728E; Thu, 4 Dec 2003 14:54:45 -0800 (PST) Received: from main.wiredfool.com ([127.0.0.1]) by localhost (cabbage [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 26637-09; Thu, 4 Dec 2003 14:54:43 -0800 (PST) Received: from [IPv6:::1] (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 724852728D; Thu, 4 Dec 2003 14:54:43 -0800 (PST) In-Reply-To: <1070570264.13923.88.camel@cletus.lyris.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041120.21124.josh@agliodbs.com> <1070567455.13923.83.camel@cletus.lyris.com> <200312042027.22223.dev@archonet.com> <1070570264.13923.88.camel@cletus.lyris.com> Mime-Version: 1.0 (Apple Message framework v606) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <8DABDA81-26AD-11D8-8622-0003930F2A6C@soroos.net> Content-Transfer-Encoding: 7bit Cc: pgsql-performance From: Eric Soroos Subject: Re: tuning questions Date: Thu, 4 Dec 2003 14:59:45 -0800 To: Jack Coates X-Mailer: Apple Mail (2.606) X-Virus-Scanned: by amavisd-new-20030616-p3 (Debian) at main.wiredfool.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/108 X-Sequence-Number: 4968 > > IO is active, but hardly saturated. CPU load is hefty though, load > average is at 4 now. > > procs memory swap io > system cpu > r b w swpd free buff cache si so bi bo in cs > us sy id > 0 2 1 2808 11432 39616 1902984 0 0 204 876 788 507 > 3 4 93 You're getting a load average of 4 with 93% idle? That's a reasonable number of context switches, and if the blocks you're reading/writing are discontinous, I could see io saturation rearing it's head. This looks to me like you're starting and killing a lot of processes. Is this thrashing psql connections, or is it one big query? What are your active processes? Your effective cache size looks to be about 1900 megs (+- binary), assuming all of it is pg. eric From pgsql-performance-owner@postgresql.org Thu Dec 4 19:17:24 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9D265D1B4BE for ; Thu, 4 Dec 2003 23:17:18 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 45489-04 for ; Thu, 4 Dec 2003 19:16:50 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 2652AD1B49B for ; Thu, 4 Dec 2003 19:16:47 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB4NGEIt006141; Thu, 4 Dec 2003 15:16:14 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 15:16:13 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 15:16:13 -0800 Subject: Re: tuning questions From: Jack Coates To: josh@agliodbs.com Cc: Richard Huxton , pgsql-performance In-Reply-To: <200312041324.37889.josh@agliodbs.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042027.22223.dev@archonet.com> <1070570264.13923.88.camel@cletus.lyris.com> <200312041324.37889.josh@agliodbs.com> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070579771.18838.187.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 15:16:11 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: josh@agliodbs.com, dev@archonet.com, pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/109 X-Sequence-Number: 4969 On Thu, 2003-12-04 at 13:24, Josh Berkus wrote: > Jack, > > > latest changes: > > shared_buffers = 35642 > > This is fine, it's about 14% of available RAM. Though the way you calculated > it still confuses me. It's not complicated; it should be between 6% and 15% > of available RAM; since you're doing a data-transformation DB, yours should > be toward the high end. > > > max_fsm_relations = 1000 > > max_fsm_pages = 10000 > > You want to raise this a whole lot if your data transformations involve large > delete or update batches. I'd suggest running "vacuum analyze verbose" > between steps to see how many dead pages you're accumulating. This looks really difficult to tune, and based on the load I'm giving it, it looks really important. I've tried the verbose analyze and I've looked at the rules of thumb, neither approach seems good for the pattern of "hammer the system for a day or two, then leave it alone for a week." I'm setting it to 500000 (half of the biggest table size divided by a 6k page size), but I'll keep tweaking this. > > > wal_buffers = 64 > > sort_mem = 32768 > > vacuum_mem = 32768 > > effective_cache_size = 10000 > > This is way the heck too low. it's supposed to be the size of all available > RAM; I'd set it to 2GB*65% as a start. This makes a little bit of difference. I set it to 65% (15869 pages). Now we have some real disk IO: procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 0 3 1 2804 10740 40808 1899856 0 0 26624 0 941 4144 13 24 63 1 2 1 2804 10808 40808 1899848 0 0 21748 60 1143 3655 9 22 69 still high cpu (3-ish load) though, and there's no noticeable improvement in query speed. > > > IO is active, but hardly saturated. CPU load is hefty though, load > > average is at 4 now. > > Unless you're doing huge statistical aggregates (like radar charts), or heavy > numerical calculations-by-query, high CPU and idle I/O usually indicates a > really bad query, like badly mismatched data types on a join or unconstrained > joins or overblown formatting-by-query. Ran that by the programmer responsible for this area and watched the statements go by with tcpdump -X. Looks like really simple stuff to me: select a handful of values, then insert into one table and delete from another. -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 19:20:48 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A4521D1B43F for ; Thu, 4 Dec 2003 23:20:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 42304-09 for ; Thu, 4 Dec 2003 19:20:15 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 28974D1B4A6 for ; Thu, 4 Dec 2003 19:20:12 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB4NKBIt006934; Thu, 4 Dec 2003 15:20:11 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 15:20:10 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 15:20:10 -0800 Subject: Re: tuning questions From: Jack Coates To: Eric Soroos Cc: pgsql-performance In-Reply-To: <8DABDA81-26AD-11D8-8622-0003930F2A6C@soroos.net> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041120.21124.josh@agliodbs.com> <1070567455.13923.83.camel@cletus.lyris.com> <200312042027.22223.dev@archonet.com> <1070570264.13923.88.camel@cletus.lyris.com> <8DABDA81-26AD-11D8-8622-0003930F2A6C@soroos.net> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070580008.13923.193.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 15:20:08 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: eric-psql@soroos.net,pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/110 X-Sequence-Number: 4970 On Thu, 2003-12-04 at 14:59, Eric Soroos wrote: > > > > IO is active, but hardly saturated. CPU load is hefty though, load > > average is at 4 now. > > > > procs memory swap io > > system cpu > > r b w swpd free buff cache si so bi bo in cs > > us sy id > > > 0 2 1 2808 11432 39616 1902984 0 0 204 876 788 507 > > 3 4 93 > > You're getting a load average of 4 with 93% idle? down a bit since my last set of tweaks, but yeah: 3:18pm up 2 days, 3:37, 3 users, load average: 3.42, 3.31, 2.81 66 processes: 65 sleeping, 1 running, 0 zombie, 0 stopped CPU0 states: 2.0% user, 3.4% system, 0.0% nice, 93.4% idle CPU1 states: 1.3% user, 2.3% system, 0.0% nice, 95.2% idle Mem: 2064656K av, 2053896K used, 10760K free, 0K shrd, 40388K buff Swap: 2899716K av, 2800K used, 2896916K free 1896232K cached PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND 23103 root 15 0 1072 1072 840 R 1.3 0.0 0:01 top 23046 postgres 15 0 33364 32M 32220 S 0.5 1.6 0:12 postmaster > > That's a reasonable number of context switches, and if the blocks > you're reading/writing are discontinous, I could see io saturation > rearing it's head. > > This looks to me like you're starting and killing a lot of processes. isn't that by design though? I've been looking at other postgres servers around the company and they seem to act pretty similar under load (none is being pounded to this level, though). > > Is this thrashing psql connections, or is it one big query? What are > your active processes? [root@postgres root]# ps auxw | grep postgres postgres 23042 0.0 0.4 308808 8628 pts/0 S 14:46 0:00 /usr/bin/postmaster -p 5432 postgres 23043 0.0 0.4 309788 8596 pts/0 S 14:46 0:00 postgres: stats buffer process postgres 23044 0.0 0.4 308828 8620 pts/0 S 14:46 0:00 postgres: stats collector process postgres 23046 0.6 1.4 309952 29872 pts/0 R 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 INSERT waiting postgres 23047 1.4 14.7 310424 304240 pts/0 S 14:46 0:21 postgres: lmuser lmdb 10.0.0.2 idle postgres 23048 0.4 14.7 310044 304368 pts/0 S 14:46 0:07 postgres: lmuser lmdb 10.0.0.2 idle postgres 23049 0.0 0.5 309820 10352 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23050 0.0 0.6 310424 13352 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23051 0.0 0.6 309940 12992 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23052 0.0 0.5 309880 11916 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23053 0.0 0.6 309924 12872 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23054 0.0 0.6 310012 13460 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23055 0.0 0.5 309932 12284 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23056 2.0 14.7 309964 304072 pts/0 S 14:46 0:30 postgres: lmuser lmdb 10.0.0.2 idle postgres 23057 2.4 14.7 309916 304104 pts/0 S 14:46 0:37 postgres: lmuser lmdb 10.0.0.2 idle postgres 23058 0.0 0.6 310392 13168 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23059 0.5 14.7 310424 304072 pts/0 S 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 idle postgres 23060 0.0 0.6 309896 13212 pts/0 S 14:46 0:00 postgres: lmuser lmdb 10.0.0.2 idle postgres 23061 0.5 1.4 309944 29832 pts/0 R 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 INSERT postgres 23062 0.6 1.4 309936 29832 pts/0 S 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 INSERT waiting postgres 23063 0.6 1.4 309944 30028 pts/0 S 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 INSERT waiting postgres 23064 0.6 1.4 309944 29976 pts/0 S 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 INSERT waiting postgres 23065 1.4 14.7 310412 304112 pts/0 S 14:46 0:21 postgres: lmuser lmdb 216.91.56.200 idle postgres 23066 0.5 1.4 309944 29496 pts/0 S 14:46 0:08 postgres: lmuser lmdb 216.91.56.200 INSERT waiting postgres 23067 0.5 1.4 310472 30040 pts/0 D 14:46 0:09 postgres: lmuser lmdb 216.91.56.200 idle postgres 23068 0.6 1.4 309936 30104 pts/0 R 14:46 0:09 postgres: lmuser lmdb 216.91.56.200 INSERT waiting postgres 23069 0.5 1.4 309936 29716 pts/0 S 14:46 0:09 postgres: lmuser lmdb 216.91.56.200 INSERT waiting postgres 23070 0.6 1.4 309944 29744 pts/0 S 14:46 0:09 postgres: lmuser lmdb 10.0.0.2 INSERT waiting ten-ish stay idle all the time, the inserts go to update when the big select is done and rows get moved from the active to the completed table. > Your effective cache size looks to be about 1900 megs (+- binary), > assuming all of it is pg. > > eric > -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 19:48:31 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EC60AD1B49B for ; Thu, 4 Dec 2003 23:48:27 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48633-07 for ; Thu, 4 Dec 2003 19:47:59 -0400 (AST) Received: from anchor-post-32.mail.demon.net (anchor-post-32.mail.demon.net [194.217.242.90]) by svr1.postgresql.org (Postfix) with ESMTP id C1012D1B458 for ; Thu, 4 Dec 2003 19:47:55 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-32.mail.demon.net with esmtp (Exim 3.35 #1) id 1AS3CP-000F3J-0W; Thu, 04 Dec 2003 23:47:57 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 9B17016C5B; Thu, 4 Dec 2003 23:47:56 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id ED8A515A6F; Thu, 4 Dec 2003 23:47:54 +0000 (GMT) From: Richard Huxton To: Jack Coates , josh@agliodbs.com Subject: Re: tuning questions Date: Thu, 4 Dec 2003 23:47:54 +0000 User-Agent: KMail/1.5 Cc: pgsql-performance References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041324.37889.josh@agliodbs.com> <1070579771.18838.187.camel@cletus.lyris.com> In-Reply-To: <1070579771.18838.187.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312042347.54461.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/111 X-Sequence-Number: 4971 On Thursday 04 December 2003 23:16, Jack Coates wrote: > > > > effective_cache_size = 10000 > > > > This is way the heck too low. it's supposed to be the size of all > > available RAM; I'd set it to 2GB*65% as a start. > > This makes a little bit of difference. I set it to 65% (15869 pages). That's still only about 127MB (15869 * 8KB). > Now we have some real disk IO: > procs memory swap io > system cpu > r b w swpd free buff cache si so bi bo in cs us > sy id > 0 3 1 2804 10740 40808 1899856 0 0 26624 0 941 4144 According to this your cache is currently 1,899,856 KB which in 8KB blocks is 237,482 - be frugal and say effective_cache_size = 200000 (or even 150000 if the trace above isn't typical). -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Dec 4 20:33:17 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 73A8CD1B4A6 for ; Fri, 5 Dec 2003 00:33:15 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 54066-09 for ; Thu, 4 Dec 2003 20:32:47 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 4A9D8D1B47A for ; Thu, 4 Dec 2003 20:32:43 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB50W8It019837; Thu, 4 Dec 2003 16:32:08 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Thu, 4 Dec 2003 16:32:08 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Thu, 04 Dec 2003 16:32:08 -0800 Subject: Re: tuning questions From: Jack Coates To: Richard Huxton Cc: josh@agliodbs.com, pgsql-performance In-Reply-To: <200312042347.54461.dev@archonet.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041324.37889.josh@agliodbs.com> <1070579771.18838.187.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070584326.18838.235.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Thu, 04 Dec 2003 16:32:06 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: dev@archonet.com, josh@agliodbs.com, pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/112 X-Sequence-Number: 4972 On Thu, 2003-12-04 at 15:47, Richard Huxton wrote: > On Thursday 04 December 2003 23:16, Jack Coates wrote: > > > > > > effective_cache_size = 10000 > > > > > > This is way the heck too low. it's supposed to be the size of all > > > available RAM; I'd set it to 2GB*65% as a start. > > > > This makes a little bit of difference. I set it to 65% (15869 pages). > > That's still only about 127MB (15869 * 8KB). yeah, missed the final digit when I copied it into the postgresql.conf :-( Just reloaded with 158691 pages. > > > Now we have some real disk IO: > > procs memory swap io > > system cpu > > r b w swpd free buff cache si so bi bo in cs us > > sy id > > 0 3 1 2804 10740 40808 1899856 0 0 26624 0 941 4144 > > According to this your cache is currently 1,899,856 KB which in 8KB blocks is > 237,482 - be frugal and say effective_cache_size = 200000 (or even 150000 if > the trace above isn't typical). d'oh, just realized what you're telling me here. /me smacks forehead. Let's try effective_cache of 183105... (75%). Starting both servers, waiting for big fetch to start, and... procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 0 0 0 2800 11920 40532 1906516 0 0 0 0 521 8 0 0 100 0 1 0 2800 11920 40532 1906440 0 0 356 52 611 113 1 3 97 0 1 0 2800 11920 40532 1906424 0 0 20604 0 897 808 1 18 81 0 1 0 2800 11920 40532 1906400 0 0 26112 0 927 820 1 13 87 0 1 0 2800 11920 40532 1906384 0 0 26112 0 923 812 1 12 87 0 1 0 2800 11920 40532 1906372 0 0 24592 0 921 805 1 13 87 0 1 0 2800 11920 40532 1906368 0 0 3248 48 961 1209 0 4 96 0 1 0 2800 11920 40532 1906368 0 0 2600 0 845 1631 0 2 98 0 1 0 2800 11920 40532 1906364 0 0 2728 0 871 1714 0 2 98 better in vmstat... but the query doesn't work any better unfortunately. The frustrating thing is, we also have a UP P3-500 with 512M RAM and two IDE drives with the same PG install which is doing okay with this load -- still half the speed of MS-SQL2K, but usable. I'm at a loss. -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Thu Dec 4 21:45:52 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 099C9D1B456 for ; Fri, 5 Dec 2003 01:45:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 63663-05 for ; Thu, 4 Dec 2003 21:45:16 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 2FF3FD1B446 for ; Thu, 4 Dec 2003 21:45:11 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1AS51u-0005YQ-00; Fri, 05 Dec 2003 03:45:14 +0200 Date: Fri, 5 Dec 2003 03:45:14 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031205014514.GA21079@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <200312042023.37050.dev@archonet.com> <20031204221312.GA18874@alcaron.ee> <200312042245.21899.dev@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312042245.21899.dev@archonet.com> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/113 X-Sequence-Number: 4973 On Thu, Dec 04, 2003 at 10:45:21PM +0000, Richard Huxton wrote: > If you've got the time, try putting together a small test-script with some > dummy data and see if it's reproducible. I'm sure the other Python users > would be interested in seeing where the problem is. Tried with test-script, but this functioned normally (Murphy's law!). Then tweaked postrgesql.conf and switched on debugging options. Results show (in my opinion) that Python has nothing to do with slow UPDATE. Timing from postgresql itself shows duration of 0.29 sec. === postgres[21247]: [2707] DEBUG: StartTransactionCommand postgres[21247]: [2708-1] LOG: query: postgres[21247]: [2708-2] UPDATE postgres[21247]: [2708-3] imp_cdr_200311 postgres[21247]: [2708-4] SET postgres[21247]: [2708-5] Status = 'SKIP' postgres[21247]: [2708-6] WHERE postgres[21247]: [2708-7] ImpRecID = '202425' ... Skipped rewritten parse tree ... postgres[21247]: [2710-1] LOG: plan: postgres[21247]: [2710-2] { INDEXSCAN postgres[21247]: [2710-3] :startup_cost 0.00 postgres[21247]: [2710-4] :total_cost 6.01 postgres[21247]: [2710-5] :rows 1 postgres[21247]: [2710-6] :width 199 postgres[21247]: [2710-7] :qptargetlist ( ... Skipped target list ... postgres[21247]: [2711] DEBUG: CommitTransactionCommand postgres[21247]: [2712] LOG: duration: 0.292529 sec === Any suggestions for further investigation? -- Ivar Zarans From pgsql-performance-owner@postgresql.org Thu Dec 4 22:08:15 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CF9D0D1B457 for ; Fri, 5 Dec 2003 02:07:55 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66891-05 for ; Thu, 4 Dec 2003 22:07:28 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id E91A3D1B43A for ; Thu, 4 Dec 2003 22:07:24 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1AS5NQ-0005bY-00; Fri, 05 Dec 2003 04:07:28 +0200 Date: Fri, 5 Dec 2003 04:07:28 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031205020728.GA21440@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <200312042023.37050.dev@archonet.com> <20031204221312.GA18874@alcaron.ee> <200312042245.21899.dev@archonet.com> <20031205014514.GA21079@alcaron.ee> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031205014514.GA21079@alcaron.ee> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/114 X-Sequence-Number: 4974 I have played around with explain and explain analyze and noticed one interesting oddity: === explain UPDATE table1 SET status = 'SKIP' WHERE recid = 196641; Seq Scan on table1 (cost=0.00..16709.97 rows=1 width=199) Filter: (recid = 196641) === explain UPDATE table1 SET status = 'SKIP' WHERE recid = '196641'; Index Scan using table1_pkey on table1 (cost=0.00..6.01 rows=1 width=199) Index Cond: (recid = 196641::bigint) === explain UPDATE table1 SET status = 'SKIP' WHERE recid = 196641::bigint; Index Scan using table1_pkey on table1 (cost=0.00..6.01 rows=1 width=199) Index Cond: (recid = 196641::bigint) === Why first example, where recid is given as numeric constant, is using sequential scan, but second example, where recid is given as string constant works with index scan, as expected? Third example shows, that numeric constant must be typecasted in order to function properly. Is this normal behaviour of fields with bigint type? -- Ivar Zarans From pgsql-performance-owner@postgresql.org Thu Dec 4 22:14:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 931C2D1B470 for ; Fri, 5 Dec 2003 02:14:57 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 64493-09 for ; Thu, 4 Dec 2003 22:14:30 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 6EF60D1B472 for ; Thu, 4 Dec 2003 22:14:25 -0400 (AST) Received: from familyhealth.com.au (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (8.12.9p1/8.12.9) with ESMTP id hB52E0oD057144; Fri, 5 Dec 2003 10:14:01 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <3FCFEA58.7090505@familyhealth.com.au> Date: Fri, 05 Dec 2003 10:15:52 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Ivar Zarans Cc: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT References: <20031204185751.GA15893@alcaron.ee> <200312042023.37050.dev@archonet.com> <20031204221312.GA18874@alcaron.ee> <200312042245.21899.dev@archonet.com> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> In-Reply-To: <20031205020728.GA21440@alcaron.ee> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/115 X-Sequence-Number: 4975 > Why first example, where recid is given as numeric constant, is using > sequential scan, but second example, where recid is given as string > constant works with index scan, as expected? Third example shows, that > numeric constant must be typecasted in order to function properly. > > Is this normal behaviour of fields with bigint type? Yes, it's a known performance problem in PostgreSQL 7.4 and below. I believe it's been fixed in 7.5 CVS already. Chris From pgsql-performance-owner@postgresql.org Fri Dec 5 00:53:36 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6354ED1B449 for ; Fri, 5 Dec 2003 04:53:01 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 87664-03 for ; Fri, 5 Dec 2003 00:52:30 -0400 (AST) Received: from main.wiredfool.com (dsl231-055-035.sea1.dsl.speakeasy.net [216.231.55.35]) by svr1.postgresql.org (Postfix) with ESMTP id 096C4D1B432 for ; Fri, 5 Dec 2003 00:52:29 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 03F9E2728E; Thu, 4 Dec 2003 20:47:25 -0800 (PST) Received: from main.wiredfool.com ([127.0.0.1]) by localhost (cabbage [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 27978-09; Thu, 4 Dec 2003 20:47:22 -0800 (PST) Received: from [IPv6:::1] (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 6F1312728D; Thu, 4 Dec 2003 20:47:22 -0800 (PST) In-Reply-To: <1070584326.18838.235.camel@cletus.lyris.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041324.37889.josh@agliodbs.com> <1070579771.18838.187.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> <1070584326.18838.235.camel@cletus.lyris.com> Mime-Version: 1.0 (Apple Message framework v606) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit Cc: pgsql-performance From: Eric Soroos Subject: Re: tuning questions Date: Thu, 4 Dec 2003 20:52:22 -0800 To: Jack Coates X-Mailer: Apple Mail (2.606) X-Virus-Scanned: by amavisd-new-20030616-p3 (Debian) at main.wiredfool.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/116 X-Sequence-Number: 4976 > > d'oh, just realized what you're telling me here. /me smacks forehead. > Let's try effective_cache of 183105... (75%). Starting both servers, > waiting for big fetch to start, and... > > procs memory swap io > system cpu > r b w swpd free buff cache si so bi bo in cs us > sy id > 0 0 0 2800 11920 40532 1906516 0 0 0 0 521 8 > 0 0 100 > 0 1 0 2800 11920 40532 1906440 0 0 356 52 611 113 > 1 3 97 > 0 1 0 2800 11920 40532 1906424 0 0 20604 0 897 808 > 1 18 81 > 0 1 0 2800 11920 40532 1906400 0 0 26112 0 927 820 > 1 13 87 > 0 1 0 2800 11920 40532 1906384 0 0 26112 0 923 812 > 1 12 87 > 0 1 0 2800 11920 40532 1906372 0 0 24592 0 921 805 > 1 13 87 > 0 1 0 2800 11920 40532 1906368 0 0 3248 48 961 1209 > 0 4 96 > 0 1 0 2800 11920 40532 1906368 0 0 2600 0 845 1631 > 0 2 98 > 0 1 0 2800 11920 40532 1906364 0 0 2728 0 871 1714 > 0 2 98 > > better in vmstat... but the query doesn't work any better > unfortunately. Your io now looks like you're getting a few seconds of continuous read, and then you're getting into maxing out random reads. These look about right for a single ide drive. > The frustrating thing is, we also have a UP P3-500 with 512M RAM and > two > IDE drives with the same PG install which is doing okay with this load > -- still half the speed of MS-SQL2K, but usable. I'm at a loss. I wonder if you're doing table scans. From the earlier trace, it looked like you have a few parallel select/process/insert processes going. If that's the case, you might be getting a big sequential scan at first, then at some point you have enough selects going that it wtarts looking more like random access. Can you run one of the selects from the psql console and see how fast it runs? Do your inserts have any foreign key relations? One thing you might try is to shut down the postmaster and move the pg_clog and pg_xlog directories to the other drive, and leave symlinks pointing back. That should help your insert performance by putting the wal on a seperate drive from the table data. It will really help if you wind up having uncached read and write access at the same time. You also might gain by using software raid 0 (with large stripe size, 512k or so) across both drives, but if you don't have the appropriate paritions in there now it's going to be a bunch of work. eric From pgsql-performance-owner@postgresql.org Fri Dec 5 04:09:16 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7561CD1B460 for ; Fri, 5 Dec 2003 08:09:13 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03299-10 for ; Fri, 5 Dec 2003 04:08:43 -0400 (AST) Received: from ar-sd.net (unknown [81.196.32.112]) by svr1.postgresql.org (Postfix) with ESMTP id 81837D1B474 for ; Fri, 5 Dec 2003 04:08:40 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by ar-sd.net (Postfix) with ESMTP id 262561C8D4; Fri, 5 Dec 2003 10:11:48 +0200 (EET) Received: from andy (unknown [192.168.0.11]) by ar-sd.net (Postfix) with SMTP id 055C917723; Fri, 5 Dec 2003 10:11:48 +0200 (EET) Message-ID: <00c901c3bb07$592bbe30$0b00a8c0@andy> From: "Andrei Bintintan" To: "Stephan Szabo" Cc: References: <014901c3ba76$ffd65650$0b00a8c0@andy> <20031204071432.R66123@megazone.bigpanda.com> Subject: Re: [ADMIN] Index not used. WHY? Date: Fri, 5 Dec 2003 10:11:11 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4927.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4927.1200 X-Virus-Scanned: by AMaViS snapshot-20020531 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/117 X-Sequence-Number: 4977 There are around 700 rows in this table. If I set enable_seqscan=off then the index is used and I also used Vacuum Analyze recently. I find it strange because the number of values of id_user and id_modull are somehow in the same distribution and when I search the table the id_user index is used but the id_modull index is not used. Does somehow postgre know that a seq scan runs faster in this case as a index scan? Should I erase this index? I have to say that the data's in this table are not changed offen, but there are a LOT of joins made with this table. Best regards. Andy. ----- Original Message ----- From: "Stephan Szabo" To: "Andrei Bintintan" Cc: ; Sent: Thursday, December 04, 2003 5:19 PM Subject: Re: [ADMIN] Index not used. WHY? > > On Thu, 4 Dec 2003, Andrei Bintintan wrote: > > > Hi, > > > > I have the following table: > > CREATE TABLE public.rights ( > > id int4 DEFAULT nextval('"rights_id_seq"'::text) NOT NULL, > > id_user int4 NOT NULL, > > id_modull int4 NOT NULL, > > CONSTRAINT rights_pkey PRIMARY KEY (id) > > ) > > > > and I created the following indexes: > > > > CREATE INDEX right_id_modull_idx ON rights USING btree (id_modull); > > CREATE INDEX right_id_user_idx ON rights USING btree (id_user); > > > > Now the problem: > > > > EXPLAIN SELECT * FROM rights r WHERE r.id_modull =15 > > returnes: > > Seq Scan on rights r (cost=0.00..12.30 rows=42 width=12) > > Filter: (id_modull = 15) > > > > EXPLAIN SELECT * FROM rights r WHERE r.id_user =15 > > returnes: > > Index Scan using right_id_user_idx on rights r (cost=0.00..8.35 rows=11 width=12) > > Index Cond: (id_user = 15) > > > > Question: Why the right_id_modull_idx is NOT USED at the 1st query and > > the second query the right_id_user_idx index is used. > > As a note, pgsql-performance is a better list for these questions. > > So, standard questions: > > How many rows are in the table, what does EXPLAIN ANALYZE show for the > queries, if you force index usage (set enable_seqscan=off) on the first > what does EXPLAIN ANALYZE show then, have you used ANALYZE/VACUUM ANALYZE > recently? > From pgsql-performance-owner@postgresql.org Fri Dec 5 04:40:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0F5C4D1B47A for ; Fri, 5 Dec 2003 08:40:37 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14504-02 for ; Fri, 5 Dec 2003 04:40:07 -0400 (AST) Received: from smtp-send.myrealbox.com (smtp-send.myrealbox.com [192.108.102.143]) by svr1.postgresql.org (Postfix) with ESMTP id D8495D1B475 for ; Fri, 5 Dec 2003 04:40:04 -0400 (AST) Received: from myrealbox.com shridhar_daithankar@smtp-send.myrealbox.com [202.54.11.72] by smtp-send.myrealbox.com with NetMail SMTP Agent $Revision: 3.45 $ on Novell NetWare via secured & encrypted transport (TLS); Fri, 05 Dec 2003 01:40:07 -0700 Message-ID: <3FD04457.6000504@myrealbox.com> Date: Fri, 05 Dec 2003 14:09:51 +0530 From: Shridhar Daithankar User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Andrei Bintintan Cc: Stephan Szabo , pgsql-performance@postgresql.org Subject: Re: [ADMIN] Index not used. WHY? References: <014901c3ba76$ffd65650$0b00a8c0@andy> <20031204071432.R66123@megazone.bigpanda.com> <00c901c3bb07$592bbe30$0b00a8c0@andy> In-Reply-To: <00c901c3bb07$592bbe30$0b00a8c0@andy> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/118 X-Sequence-Number: 4978 Andrei Bintintan wrote: > There are around 700 rows in this table. > If I set enable_seqscan=off then the index is used and I also used Vacuum > Analyze recently. For 700 rows I think seq. would work best. > > I find it strange because the number of values of id_user and id_modull are > somehow in the same distribution and when I search the table the id_user > index is used but the id_modull index is not used. > > Does somehow postgre know that a seq scan runs faster in this case as a > index scan? Should I erase this index? > I have to say that the data's in this table are not changed offen, but there > are a LOT of joins made with this table. If table is cached then it does not matter. Unless it grows substantially, say to around hundred thousand rows(Note your table is small), idex wouldn't be that useful. Shridhar From pgsql-performance-owner@postgresql.org Fri Dec 5 05:13:33 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E32D0D1B475 for ; Fri, 5 Dec 2003 09:13:30 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 10446-07 for ; Fri, 5 Dec 2003 05:13:01 -0400 (AST) Received: from ecbull20.frec.bull.fr (ecbull20.frec.bull.fr [129.183.4.3]) by svr1.postgresql.org (Postfix) with ESMTP id 0EB03D1B47A for ; Fri, 5 Dec 2003 05:12:57 -0400 (AST) Received: from ecn002.frec.bull.fr (ecn002.frec.bull.fr [129.183.4.6]) by ecbull20.frec.bull.fr (8.9.2/8.9.1) with ESMTP id KAA16934; Fri, 5 Dec 2003 10:13:05 +0100 Received: from BULL.NET ([129.183.148.134]) by ecn002.frec.bull.fr (Lotus Domino Release 5.0.12) with ESMTP id 2003120510163396:112 ; Fri, 5 Dec 2003 10:16:33 +0100 Message-ID: <3FD04C25.307690CC@BULL.NET> Date: Fri, 05 Dec 2003 10:13:09 +0100 From: Thierry Missimilly Organization: BSIS/R&D X-Mailer: Mozilla 4.75 [en] (Windows NT 5.0; U) X-Accept-Language: fr,en MIME-Version: 1.0 To: Jack Coates Cc: pgsql-performance Subject: Re: tuning questions References: <1070553983.6498.52.camel@cletus.lyris.com> <200312041120.21124.josh@agliodbs.com> <1070567455.13923.83.camel@cletus.lyris.com> <200312042027.22223.dev@archonet.com> <1070570264.13923.88.camel@cletus.lyris.com> X-MIMETrack: Itemize by SMTP Server on ECN002/FR/BULL(Release 5.0.12 |February 13, 2003) at 05/12/2003 10:16:34, Serialize by Router on ECN002/FR/BULL(Release 5.0.12 |February 13, 2003) at 05/12/2003 10:16:40, Serialize complete at 05/12/2003 10:16:40 Content-Type: multipart/mixed; boundary="------------A086F02D18D306A929300A18" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/119 X-Sequence-Number: 4979 This is a multi-part message in MIME format. --------------A086F02D18D306A929300A18 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Jack Coates wrote: > > latest changes: > shared_buffers = 35642 > max_fsm_relations = 1000 > max_fsm_pages = 10000 > wal_buffers = 64 > sort_mem = 32768 > vacuum_mem = 32768 > effective_cache_size = 10000 > > /proc/sys/kernel/shmmax = 500000000 > > IO is active, but hardly saturated. CPU load is hefty though, load > average is at 4 now. > > procs memory swap io > system cpu > r b w swpd free buff cache si so bi bo in cs us > sy id > 0 2 1 2808 11436 39616 1902988 0 0 240 896 765 469 > 2 11 87 > 0 2 1 2808 11432 39616 1902988 0 0 244 848 768 540 > 4 3 93 > 0 2 1 2808 11432 39616 1902984 0 0 204 876 788 507 > 3 4 93 > 0 2 1 2808 11432 39616 1902984 0 0 360 416 715 495 > 4 1 96 > 0 2 1 2808 11432 39616 1902984 0 0 376 328 689 441 > 2 1 97 > 0 2 0 2808 11428 39616 1902976 0 0 464 360 705 479 > 2 1 97 > 0 2 1 2808 11428 39616 1902976 0 0 432 380 718 547 > 3 1 97 > 0 2 1 2808 11428 39616 1902972 0 0 440 372 742 512 > 1 3 96 > 0 2 1 2808 11428 39616 1902972 0 0 416 364 711 504 > 3 1 96 > 0 2 1 2808 11424 39616 1902972 0 0 456 492 743 592 > 2 1 97 > 0 2 1 2808 11424 39616 1902972 0 0 440 352 707 494 > 2 1 97 > 0 2 1 2808 11424 39616 1902972 0 0 456 360 709 494 > 2 2 97 > 0 2 1 2808 11436 39616 1902968 0 0 536 516 807 708 > 3 2 94 > Hi Jack, As show by vmstat, your Operating System is spending 96% of its time in Idle. On RedHat 8.0 IA32, Idle means idle and Wait I/O. In your case, i think they are Wait I/O as you are working on 2.8 GB DB with only 2GB RAM, but it should be arround 30%. Your performances whould increase only if User CPU increase otherwise, for exemple if your system swap, only Sys CPU whould increase and your application will stay slow. You can better check your I/O with : iostat 3 1000, and check that the max tps are on the database filesystem. So, all the Postgres tuning you have tried do not change a lot as the bottleneck is your I/O throuput. But, one thing you can check is which parts of Postgres need a lot of I/O. To do that, after shuting down PG, move your database on an other disk (OS disk ?) for exemple /mypg/data and create a symblolic link for /mypg/data/ to $PGDATA/base. Restart PG, and while you execute your application, check with iostat which disk as the max of tps. I bet, it is the disk where the WAL buffer are logged. One more thing about I/O, for an IDE disk, the maximum number of Write Block + Read Block per sec is about 10000 based on the I/O block size is 1 K. That means 10 Mb/s. if you need more, you can try Stripped SCSI disks or RAID0 subsystem disks. Thierry Missimilly > > -- > Jack Coates, Lyris Technologies Applications Engineer > 510-549-4350 x148, jack@lyris.com > "Interoperability is the keyword, uniformity is a dead end." > --Olivier Fourdan > > ---------------------------(end of broadcast)--------------------------- > TIP 3: 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 --------------A086F02D18D306A929300A18 Content-Type: text/x-vcard; charset=us-ascii; name="THIERRY.MISSIMILLY.vcf" Content-Description: Card for Thierry Missimilly Content-Disposition: attachment; filename="THIERRY.MISSIMILLY.vcf" Content-Transfer-Encoding: 7bit begin:vcard n:Missimilly;Thierry tel;fax:+33 (0)4 76 29 78 78 tel;work:+33 (0)4 76 29 74 54 x-mozilla-html:FALSE url:http:\\www.bull.com org:BIS/R&D adr:;;Bull SA, 1, rue de provence - BP 208;ECHIROLLES;;38432;FRANCE version:2.1 email;internet:Thierry.Missimilly@bull.net x-mozilla-cpt:;-18184 fn:Thierry Missimilly end:vcard --------------A086F02D18D306A929300A18-- From pgsql-performance-owner@postgresql.org Fri Dec 5 06:08:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3299BD1B475 for ; Fri, 5 Dec 2003 10:08:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 23188-03 for ; Fri, 5 Dec 2003 06:08:29 -0400 (AST) Received: from anchor-post-31.mail.demon.net (anchor-post-31.mail.demon.net [194.217.242.89]) by svr1.postgresql.org (Postfix) with ESMTP id 4699BD1B47A for ; Fri, 5 Dec 2003 06:08:22 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-31.mail.demon.net with esmtp (Exim 3.35 #1) id 1ASCsp-00003O-0V; Fri, 05 Dec 2003 10:08:23 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 0AD5716A42; Fri, 5 Dec 2003 10:08:21 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 3462315948; Fri, 5 Dec 2003 10:08:20 +0000 (GMT) From: Richard Huxton To: Ivar Zarans , pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Date: Fri, 5 Dec 2003 10:08:20 +0000 User-Agent: KMail/1.5 References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> In-Reply-To: <20031205020728.GA21440@alcaron.ee> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312051008.20317.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/120 X-Sequence-Number: 4980 On Friday 05 December 2003 02:07, Ivar Zarans wrote: > I have played around with explain and explain analyze and noticed one > interesting oddity: [snip] > Why first example, where recid is given as numeric constant, is using > sequential scan, but second example, where recid is given as string > constant works with index scan, as expected? Third example shows, that > numeric constant must be typecasted in order to function properly. > > Is this normal behaviour of fields with bigint type? As Christopher says, normal (albeit irritating). Not sure it applies here - all the examples you've shown me are using the index. Well - I must admit I'm stumped. Unless you have a *lot* of indexes and foreign keys to check, I can't see why it would take so long to update a single row. Can you post the schema for the table? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Fri Dec 5 08:39:20 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 94E06D1B48B for ; Fri, 5 Dec 2003 12:39:14 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 42257-03 for ; Fri, 5 Dec 2003 08:38:45 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 99F67D1B460 for ; Fri, 5 Dec 2003 08:38:41 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ASFEJ-00079B-00; Fri, 05 Dec 2003 14:38:43 +0200 Date: Fri, 5 Dec 2003 14:38:43 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031205123843.GA27170@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312051008.20317.dev@archonet.com> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/121 X-Sequence-Number: 4981 On Fri, Dec 05, 2003 at 10:08:20AM +0000, Richard Huxton wrote: > > numeric constant must be typecasted in order to function properly. > > > > Is this normal behaviour of fields with bigint type? > > As Christopher says, normal (albeit irritating). Not sure it applies here - > all the examples you've shown me are using the index. I guess i have solved this mystery. Problem appears to be exactly with this - numeric constant representation in query. I am using PyPgSQL for PostgreSQL access and making update queries as this: qry = "UPDATE table1 SET status = %s WHERE recid = %s" cursor.execute(qry, status, recid) Execute method of cursor object is supposed to merge "status" and "recid" values into "qry", using proper quoting. When i started to play around with debug information i noticed, that this query used sequential scan for "recid". Then i also noticed, that query, sent to server looked like this: "UPDATE table1 SET status = 'SKIP' WHERE recid = 199901" Sure enough, when i used psql and EXPLAIN on this query, i got query plan with sequential scan. And using recid value as string or typecasted integer gave correct results with index scan. I wrote about this in my previous message. It seems, that PyPgSQL query quoting is not aware of this performance problem (to which Cristopher referred) and final query, sent to server is correct SQL, but not correct, considering PostgreSQL bugs. One more explanation - previously i posted some logs, showing correct query, using index scan, but still taking 0.29 seconds. Reason for this delay is logging itself - it generates enough IO traffic to have impact on query speed. With logging disabled, this query takes around 0.0022 seconds, which is perfectly normal. Finally - what would be correct solution to this problem? Upgrading to 7.5 CVS is not an option :) One possibility is not to use PyPgSQL variable substitution and create every query "by hand" - not very nice solution, since variable substitution and quoting is quite convenient. Second (and better) possibility is to ask PyPgSQL develeopers to take care of PostgreSQL oddities. Any other suggestions? -- Ivar From pgsql-performance-owner@postgresql.org Fri Dec 5 08:50:29 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 247CBD1B48E for ; Fri, 5 Dec 2003 12:50:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 47259-02 for ; Fri, 5 Dec 2003 08:49:53 -0400 (AST) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id 9A015D1B454 for ; Fri, 5 Dec 2003 08:49:48 -0400 (AST) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id hB5CnwOX026523 for ; Fri, 5 Dec 2003 18:19:58 +0530 Received: from persistent.co.in (daithan.intranet.pspl.co.in [192.168.7.161]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id hB5Cnvhe026485; Fri, 5 Dec 2003 18:19:57 +0530 Message-ID: <3FD07EEA.8030007@persistent.co.in> Date: Fri, 05 Dec 2003 18:19:46 +0530 From: Shridhar Daithankar Organization: Persistent Systems Pvt. Ltd. User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Ivar Zarans Cc: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> In-Reply-To: <20031205123843.GA27170@alcaron.ee> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/122 X-Sequence-Number: 4982 Ivar Zarans wrote: > It seems, that PyPgSQL query quoting is not aware of this performance > problem (to which Cristopher referred) and final query, sent to server > is correct SQL, but not correct, considering PostgreSQL bugs. Personally I don't consider a bug but anyways.. You are the one facing problem so I understand.. > Finally - what would be correct solution to this problem? Upgrading to > 7.5 CVS is not an option :) One possibility is not to use PyPgSQL > variable substitution and create every query "by hand" - not very nice > solution, since variable substitution and quoting is quite convenient. > > Second (and better) possibility is to ask PyPgSQL develeopers to take care > of PostgreSQL oddities. > > Any other suggestions? I know zero in python but just guessing.. Will following help? qry = "UPDATE table1 SET status = %s WHERE recid = '%s'" cursor.execute(qry, status, recid) Just a thought.. Shridhar From pgsql-performance-owner@postgresql.org Fri Dec 5 09:13:56 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BC3B6D1B445 for ; Fri, 5 Dec 2003 13:13:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48494-07 for ; Fri, 5 Dec 2003 09:13:26 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id BB5F9D1B465 for ; Fri, 5 Dec 2003 09:13:22 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ASFlt-0007Tk-00; Fri, 05 Dec 2003 15:13:25 +0200 Date: Fri, 5 Dec 2003 15:13:25 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031205131325.GB27524@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3FD07EEA.8030007@persistent.co.in> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/123 X-Sequence-Number: 4983 On Fri, Dec 05, 2003 at 06:19:46PM +0530, Shridhar Daithankar wrote: > >is correct SQL, but not correct, considering PostgreSQL bugs. > > Personally I don't consider a bug but anyways.. You are the one facing > problem so I understand.. Well, if this is not bug, then what is consideration behind this behaviour? BTW, according to Cristopher it is fixed in 7.5 CVS. Why fix it if this is not a bug? :)) One more question - is this "feature" related only to "bigint" fields, or are other datatypes affected as well? > Will following help? > > qry = "UPDATE table1 SET status = %s WHERE recid = '%s'" > cursor.execute(qry, status, recid) Yes, this helps. But then it sort of obsoletes PyPgSQL-s own quoting logic. I would prefer to take care of this all by myself or trust some underlying code to do this for me. And PyPgSQL is quite nice - it checks datatype and acts accordingly. -- Ivar From pgsql-performance-owner@postgresql.org Fri Dec 5 09:24:23 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E65C0D1B459 for ; Fri, 5 Dec 2003 13:24:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52549-02 for ; Fri, 5 Dec 2003 09:23:53 -0400 (AST) Received: from anchor-post-32.mail.demon.net (anchor-post-32.mail.demon.net [194.217.242.90]) by svr1.postgresql.org (Postfix) with ESMTP id 74625D1B44A for ; Fri, 5 Dec 2003 09:23:49 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-32.mail.demon.net with esmtp (Exim 3.35 #1) id 1ASFvv-000Bfj-0W; Fri, 05 Dec 2003 13:23:47 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 2B8CE1789C; Fri, 5 Dec 2003 13:23:46 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 41C731789A; Fri, 5 Dec 2003 13:23:44 +0000 (GMT) From: Richard Huxton To: Shridhar Daithankar , Ivar Zarans Subject: Re: Slow UPADTE, compared to INSERT Date: Fri, 5 Dec 2003 13:23:43 +0000 User-Agent: KMail/1.5 Cc: pgsql-performance@postgresql.org References: <20031204185751.GA15893@alcaron.ee> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> In-Reply-To: <3FD07EEA.8030007@persistent.co.in> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312051323.43966.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/124 X-Sequence-Number: 4984 On Friday 05 December 2003 12:49, Shridhar Daithankar wrote: > Ivar Zarans wrote: > > It seems, that PyPgSQL query quoting is not aware of this performance > > problem (to which Cristopher referred) and final query, sent to server > > is correct SQL, but not correct, considering PostgreSQL bugs. > > Will following help? > > qry = "UPDATE table1 SET status = %s WHERE recid = '%s'" > cursor.execute(qry, status, recid) Better IMHO would be: "UPDATE table1 SET status = %s WHERE recid = %s::int8" PG is very strict regarding types - normally a good thing, but it can hit you unexpectedly in this scenario. The reason is that the literal number is treated as int4, whereas quoted it is marked as type unknown. Unkown gets cast to int8, whereas int4 gets left as-is. If you want to know why int4 doesn't get promoted to int8 automatically, browse the hackers list for the last couple of years. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Fri Dec 5 09:52:10 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D9E95D1B45B for ; Fri, 5 Dec 2003 13:52:08 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55940-03 for ; Fri, 5 Dec 2003 09:51:41 -0400 (AST) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id F0DC9D1B457 for ; Fri, 5 Dec 2003 09:51:36 -0400 (AST) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id hB5Dpp1x032718 for ; Fri, 5 Dec 2003 19:21:51 +0530 Received: from persistent.co.in (daithan.intranet.pspl.co.in [192.168.7.161]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id hB5Dpnhe032677; Fri, 5 Dec 2003 19:21:50 +0530 Message-ID: <3FD08D6A.4080207@persistent.co.in> Date: Fri, 05 Dec 2003 19:21:38 +0530 From: Shridhar Daithankar Organization: Persistent Systems Pvt. Ltd. User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Ivar Zarans Cc: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> <20031205131325.GB27524@alcaron.ee> In-Reply-To: <20031205131325.GB27524@alcaron.ee> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/125 X-Sequence-Number: 4985 Ivar Zarans wrote: > On Fri, Dec 05, 2003 at 06:19:46PM +0530, Shridhar Daithankar wrote: > > >>>is correct SQL, but not correct, considering PostgreSQL bugs. >>Personally I don't consider a bug but anyways.. You are the one facing >>problem so I understand.. > Well, if this is not bug, then what is consideration behind this > behaviour? BTW, according to Cristopher it is fixed in 7.5 CVS. > Why fix it if this is not a bug? :)) This is not a bug. It is just that people find it confusing when postgresql planner consider seemingly same type as different. e.g. treating int8 as different than int4. Obvious thinking is they should be same. But given postgresql's flexibility with create type, it is difficult to promote. AFAIK, the fix in CVS is to make indexes operatable with seemingly compatible types. Which does not change the fact that postgresql can not upgrade data types on it's own. Write good queries which adhere to strict data typing. It is better to understand anyway. > One more question - is this "feature" related only to "bigint" fields, > or are other datatypes affected as well? Every data type is affected. int2 will not use a int4 index and so on. >>Will following help? >> >>qry = "UPDATE table1 SET status = %s WHERE recid = '%s'" >>cursor.execute(qry, status, recid) > > > Yes, this helps. But then it sort of obsoletes PyPgSQL-s own quoting > logic. I would prefer to take care of this all by myself or trust some > underlying code to do this for me. And PyPgSQL is quite nice - it > checks datatype and acts accordingly. Well, then pypgsql should be upgraded to query the pg catalogd to find exact type of column. But that would be too cumbersome I guess. Shridhar From pgsql-performance-owner@postgresql.org Fri Dec 5 11:35:34 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D3C5AD1B459 for ; Fri, 5 Dec 2003 15:35:27 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73756-03 for ; Fri, 5 Dec 2003 11:35:00 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 20755D1B43A for ; Fri, 5 Dec 2003 11:34:56 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 867E3355D4; Fri, 5 Dec 2003 07:34:45 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 85311355A1; Fri, 5 Dec 2003 07:34:45 -0800 (PST) Date: Fri, 5 Dec 2003 07:34:45 -0800 (PST) From: Stephan Szabo To: Andrei Bintintan Cc: pgsql-performance@postgresql.org Subject: Re: [ADMIN] Index not used. WHY? In-Reply-To: <00c901c3bb07$592bbe30$0b00a8c0@andy> Message-ID: <20031205072359.X5253@megazone.bigpanda.com> References: <014901c3ba76$ffd65650$0b00a8c0@andy> <20031204071432.R66123@megazone.bigpanda.com> <00c901c3bb07$592bbe30$0b00a8c0@andy> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/126 X-Sequence-Number: 4986 On Fri, 5 Dec 2003, Andrei Bintintan wrote: > There are around 700 rows in this table. > If I set enable_seqscan=off then the index is used and I also used Vacuum > Analyze recently. > > I find it strange because the number of values of id_user and id_modull are > somehow in the same distribution and when I search the table the id_user > index is used but the id_modull index is not used. It was guessing that one would return 11 rows and the other 42 which is why one used the index and the other wouldn't. If those numbers aren't realistic, you may want to raise the statistics target for the columns (see ALTER TABLE) and re-run analyze. > Does somehow postgre know that a seq scan runs faster in this case as a > index scan? Should I erase this index? It's making an educated guess. When you're doing an index scan, it needs to read through the index and then get matching rows from the table. However, because those reads from the table are in a potentially random order, there's usually a higher cost associated with those reads than if the table was read in order (barring cases where you know your database should always stay cached in disk cache, etc...). If there's say 50 pages in the entire table, a sequence scan does 50 sequential page reads and is checking all those tuples. If you're getting say 42 rows through an index, you're first reading through the index, and then getting pages in a random order from the table where depends on the distribution of values throughout the table. There's a variable in the configuration, random_page_cost which controls the ratio of cost between a sequential read and a random one (defaulting to 4). From pgsql-performance-owner@postgresql.org Fri Dec 5 12:48:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 260B0D1B43A for ; Fri, 5 Dec 2003 16:48:19 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85689-07 for ; Fri, 5 Dec 2003 12:47:47 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 6B46FD1B46D for ; Fri, 5 Dec 2003 12:47:46 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ASJ7H-0007wW-00; Fri, 05 Dec 2003 18:47:43 +0200 Date: Fri, 5 Dec 2003 18:47:43 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031205164743.GA30510@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> <200312051323.43966.dev@archonet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312051323.43966.dev@archonet.com> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/127 X-Sequence-Number: 4987 On Fri, Dec 05, 2003 at 01:23:43PM +0000, Richard Huxton wrote: > Better IMHO would be: "UPDATE table1 SET status = %s WHERE recid = %s::int8" Thanks for the hint! > unexpectedly in this scenario. The reason is that the literal number is > treated as int4, whereas quoted it is marked as type unknown. Unkown gets > cast to int8, whereas int4 gets left as-is. This explains a lot. Thanks! BTW, is this mentioned somewhere in PostgreSQL documentation? I can't remember anything on this subject. Maybe i just somehow skipped it... -- Ivar From pgsql-performance-owner@postgresql.org Fri Dec 5 12:53:32 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BBEEBD1B43F for ; Fri, 5 Dec 2003 16:53:26 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 86410-01 for ; Fri, 5 Dec 2003 12:52:55 -0400 (AST) Received: from uhh.alcaron.ee (uhh.alcaron.ee [194.204.62.169]) by svr1.postgresql.org (Postfix) with ESMTP id 85074D1B445 for ; Fri, 5 Dec 2003 12:52:54 -0400 (AST) Received: from iff by uhh.alcaron.ee with local id 1ASJCH-0007xz-00; Fri, 05 Dec 2003 18:52:53 +0200 Date: Fri, 5 Dec 2003 18:52:53 +0200 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT Message-ID: <20031205165253.GB30510@alcaron.ee> References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> <20031205131325.GB27524@alcaron.ee> <3FD08D6A.4080207@persistent.co.in> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3FD08D6A.4080207@persistent.co.in> User-Agent: Mutt/1.3.28i From: Ivar Zarans X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/128 X-Sequence-Number: 4988 On Fri, Dec 05, 2003 at 07:21:38PM +0530, Shridhar Daithankar wrote: > planner consider seemingly same type as different. e.g. treating int8 as > different than int4. Obvious thinking is they should be same. But given > postgresql's flexibility with create type, it is difficult to promote. OK, this makes sense and explains a lot. Thanks! > Well, then pypgsql should be upgraded to query the pg catalogd to find > exact type of column. But that would be too cumbersome I guess. Yes, so it seems. Time to rewrite my queries :) Thanks again for help and explanations! -- Ivar From pgsql-performance-owner@postgresql.org Fri Dec 5 13:06:28 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C1942D1B481 for ; Fri, 5 Dec 2003 17:06:18 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89713-03 for ; Fri, 5 Dec 2003 13:05:48 -0400 (AST) Received: from fep02-mail.bloor.is.net.cable.rogers.com (fep02-mail.bloor.is.net.cable.rogers.com [66.185.86.72]) by svr1.postgresql.org (Postfix) with ESMTP id 3C2F7D1B445 for ; Fri, 5 Dec 2003 13:05:47 -0400 (AST) Received: from rogers.com ([65.50.20.132]) by fep02-mail.bloor.is.net.cable.rogers.com (InterMail vM.5.01.05.12 201-253-122-126-112-20020820) with ESMTP id <20031205170439.CYCP481016.fep02-mail.bloor.is.net.cable.rogers.com@rogers.com>; Fri, 5 Dec 2003 12:04:39 -0500 Message-ID: <3FD0BAE9.7050801@rogers.com> Date: Fri, 05 Dec 2003 12:05:45 -0500 From: "Mike C. Fletcher" User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20030925 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Ivar Zarans Cc: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> In-Reply-To: <20031205123843.GA27170@alcaron.ee> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Authentication-Info: Submitted using SMTP AUTH PLAIN at fep02-mail.bloor.is.net.cable.rogers.com from [65.50.20.132] using ID at Fri, 5 Dec 2003 12:04:39 -0500 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/129 X-Sequence-Number: 4989 I just spent 2 days tracking this error down in my own code, actually. What I wound up doing is having the two places where I generate the queries (everything in my system goes through those two points, as I'm using a middleware layer) check values used as identifying fields for the presence of a bigint, and if one exists, replaces it with a wrapper that does the coerced-string representation: class Wrap: def __init__( self, value ): self.value = value def __str__( self ): return "'%s'::bigint"%(self.value,) __repr__ = __str__ value = Wrap(value) Just doing that for the indexing/identifying values ATM. pyPgSQL will back up to using simple repr for the object (rather than raising an error as it would if you were using a formatted string), but will otherwise treat it as a regular value for quoting and the like, so no other modifications to the code required. By no means an elegant fix, but since your post (well, the resulting thread) managed to solve my problem, figured I should at least tell everyone thanks and how I worked around the problem. You wouldn't want this kind of hack down in the pyPgSQL level I would think, as it's DB-version specific. I suppose you could alter the __repr__ of the PgInt8 class/type to always use the string or coerced form, but it seems wrong to me. I'm actually hesitant to include it in our own middleware layer, but oh well, it does seem to be necessary for even somewhat reasonable performance. BTW, my case was a largish (88,000 record) table with a non-unique bigint key, explain on update shows sequential search, while with 'int'::bigint goes to index search. Using pyPgSQL as the interface to 7.3.4 and 7.3.3. Enjoy, Mike Ivar Zarans wrote: >On Fri, Dec 05, 2003 at 10:08:20AM +0000, Richard Huxton wrote: > > ... >I am using PyPgSQL for PostgreSQL access and making update queries as this: > > ... >It seems, that PyPgSQL query quoting is not aware of this performance >problem (to which Cristopher referred) and final query, sent to server >is correct SQL, but not correct, considering PostgreSQL bugs. > > ... >Finally - what would be correct solution to this problem? Upgrading to >7.5 CVS is not an option :) One possibility is not to use PyPgSQL >variable substitution and create every query "by hand" - not very nice >solution, since variable substitution and quoting is quite convenient. > >Second (and better) possibility is to ask PyPgSQL develeopers to take care >of PostgreSQL oddities. > >Any other suggestions? > > _______________________________________ Mike C. Fletcher Designer, VR Plumber, Coder http://members.rogers.com/mcfletch/ From pgsql-performance-owner@postgresql.org Fri Dec 5 13:12:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 00838D1B445 for ; Fri, 5 Dec 2003 17:12:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89407-03 for ; Fri, 5 Dec 2003 13:12:09 -0400 (AST) Received: from web01-imail.rogers.com (web01-imail.bloor.is.net.cable.rogers.com [66.185.86.75]) by svr1.postgresql.org (Postfix) with ESMTP id B8D6DD1B454 for ; Fri, 5 Dec 2003 13:12:08 -0400 (AST) Received: from rogers.com ([65.50.20.132]) by web01-imail.rogers.com (InterMail vM.5.01.05.12 201-253-122-126-112-20020820) with ESMTP id <20031205171225.ZBLS73768.web01-imail.rogers.com@rogers.com> for ; Fri, 5 Dec 2003 12:12:25 -0500 Message-ID: <3FD0BC67.5080300@rogers.com> Date: Fri, 05 Dec 2003 12:12:07 -0500 From: "Mike C. Fletcher" User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20030925 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT References: <20031204185751.GA15893@alcaron.ee> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> <200312051323.43966.dev@archonet.com> <20031205164743.GA30510@alcaron.ee> In-Reply-To: <20031205164743.GA30510@alcaron.ee> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Authentication-Info: Submitted using SMTP AUTH PLAIN at web01-imail.rogers.com from [65.50.20.132] using ID at Fri, 5 Dec 2003 12:12:25 -0500 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/130 X-Sequence-Number: 4990 Ivar Zarans wrote: >On Fri, Dec 05, 2003 at 01:23:43PM +0000, Richard Huxton wrote: > > >>Better IMHO would be: "UPDATE table1 SET status = %s WHERE recid = %s::int8" >> >> > >Thanks for the hint! > > Which makes the wrapper class need: def __str__( self ): return "%s::int8"%(self.value,) Enjoy, Mike _______________________________________ Mike C. Fletcher Designer, VR Plumber, Coder http://members.rogers.com/mcfletch/ From pgsql-performance-owner@postgresql.org Fri Dec 5 13:27:20 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2236DD1B43A for ; Fri, 5 Dec 2003 17:27:14 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89434-05 for ; Fri, 5 Dec 2003 13:26:43 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id E993ED1B465 for ; Fri, 5 Dec 2003 13:26:41 -0400 (AST) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4027472; Fri, 05 Dec 2003 09:27:31 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Jack Coates , Richard Huxton Subject: Re: tuning questions Date: Fri, 5 Dec 2003 09:26:05 -0800 User-Agent: KMail/1.4.3 Cc: pgsql-performance References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> <1070584326.18838.235.camel@cletus.lyris.com> In-Reply-To: <1070584326.18838.235.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200312050926.05155.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/131 X-Sequence-Number: 4991 Jack, > The frustrating thing is, we also have a UP P3-500 with 512M RAM and two > IDE drives with the same PG install which is doing okay with this load > -- still half the speed of MS-SQL2K, but usable. I'm at a loss. Overall, I'm really getting the feeling that this procedure was optimized for Oracle and/or MSSQL and is hitting some things that aren't such a good idea for PostgreSQL. I highly suggest that you try using log_duration and log_statement (and in 7.4 log_min_duration_statement) to try to locate which particular statements are taking the longest. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Dec 5 13:29:26 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C14D5D1B442 for ; Fri, 5 Dec 2003 17:29:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89372-08 for ; Fri, 5 Dec 2003 13:28:47 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 802FED1B43A for ; Fri, 5 Dec 2003 13:28:46 -0400 (AST) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id B086336FFD; Fri, 5 Dec 2003 12:28:30 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.dyndns.tv ident=foobar) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1ASJkk-0001zl-00; Fri, 05 Dec 2003 12:28:30 -0500 To: Ivar Zarans Cc: pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> <20031205131325.GB27524@alcaron.ee> In-Reply-To: <20031205131325.GB27524@alcaron.ee> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 05 Dec 2003 12:28:30 -0500 Message-ID: <87ad672myp.fsf@stark.dyndns.tv> Lines: 20 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 postgresql.org X-Archive-Number: 200312/132 X-Sequence-Number: 4992 Ivar Zarans writes: > > qry = "UPDATE table1 SET status = %s WHERE recid = '%s'" > > cursor.execute(qry, status, recid) > > Yes, this helps. But then it sort of obsoletes PyPgSQL-s own quoting > logic. I would prefer to take care of this all by myself or trust some > underlying code to do this for me. And PyPgSQL is quite nice - it > checks datatype and acts accordingly. You should tell the PyPgSQL folk to use the new binary protocol for parameters so that there are no quoting issues at all. But if it's going to interpolate strings into the query then pyPgSQL really ought to be doing '%s' as above even for numbers. This lets postgres decide what the optimal datatype is based on what you're comparing it to. Skipping the quotes will only cause headaches. -- greg From pgsql-performance-owner@postgresql.org Fri Dec 5 15:29:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A1EBDD1B459 for ; Fri, 5 Dec 2003 19:29:05 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15559-07 for ; Fri, 5 Dec 2003 15:28:35 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id D7CACD1B454 for ; Fri, 5 Dec 2003 15:28:33 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hB5JSRbI075028 for ; Fri, 5 Dec 2003 19:28:27 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hB5J6CkP027493 for pgsql-performance@postgresql.org; Fri, 5 Dec 2003 19:06:12 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Slightly OT -- Xeon versus Opteron Comparison Date: Fri, 05 Dec 2003 11:06:25 -0800 Organization: Hub.Org Networking Services Lines: 11 Message-ID: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/133 X-Sequence-Number: 4993 Ace's Hardware has put together a fairly comprehensive comparison between Xeon & Opteron platforms running server apps. Unfortunately, only MySQL "data mining" benchmarks as the review crew doesn't have that much experience with OLTP-type systems but I'm gonna try to convince them to add the ODSL DB benchmarks assuming they work fairly well with Postgres. Read up the goodies here: http://www.aceshardware.com/read.jsp?id=60000275 From pgsql-performance-owner@postgresql.org Fri Dec 5 17:00:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D8D59D1B457 for ; Fri, 5 Dec 2003 21:00:57 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 29329-05 for ; Fri, 5 Dec 2003 17:00:28 -0400 (AST) Received: from jhuml2.jhmi.edu (jhuml2.jhmi.edu [162.129.234.21]) by svr1.postgresql.org (Postfix) with ESMTP id 57BE2D1B482 for ; Fri, 5 Dec 2003 17:00:25 -0400 (AST) Received: from jhuml2.jhmi.edu (jhuml2.jhmi.edu [162.129.234.21]) by jhuml2.jhmi.edu (PMDF V6.2-X17 #30840) with SMTP id <0HPF00EJAWJ081@jhuml2.jhmi.edu> for pgsql-performance@postgresql.org; Fri, 05 Dec 2003 15:59:59 -0500 (EST) Received: from jhuml2.jhmi.edu ([162.129.234.21]) by jhuml2.jhmi.edu (SAVSMTP 3.1.0.29) with SMTP id M2003120515595921657 for ; Fri, 05 Dec 2003 15:59:59 -0500 Received: from jhmimail.jhmi.edu (jhem2.jhmi.edu [162.129.8.23]) by jhuml2.jhmi.edu (PMDF V6.2-X17 #30840) with ESMTP id <0HPF00I0WWZZ3O@jhuml2.jhmi.edu> for pgsql-performance@postgresql.org; Fri, 05 Dec 2003 15:59:59 -0500 (EST) Received: from [162.129.178.57] by jhmimail.jhmi.edu (mshttpd); Fri, 05 Dec 2003 13:03:42 -0800 Date: Fri, 05 Dec 2003 13:03:42 -0800 From: LIANHE SHAO Subject: query using cpu nearly 100%, why? To: pgsql-performance Message-id: <19be88196606.19660619be88@jhmimail.jhmi.edu> MIME-version: 1.0 X-Mailer: iPlanet Messenger Express 5.2 HotFix 1.17 (built Jun 23 2003) Content-type: text/plain; charset=us-ascii Content-language: en Content-transfer-encoding: 7bit Content-disposition: inline X-Accept-Language: en X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/134 X-Sequence-Number: 4994 Hello, I use php as front-end to query our database. When I use System Monitor to check the usage of cpu and memory, I noticed that the cpu very easily gets up to 100%. Is that normal? if not, could someone points out possible reason? I am using linux7.3, pgsql 7.3.4, 1G Memory and 2GHz CPU. Regards, William From pgsql-performance-owner@postgresql.org Fri Dec 5 17:52:25 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5B8DFD1B433 for ; Fri, 5 Dec 2003 21:52:23 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37296-04 for ; Fri, 5 Dec 2003 17:51:54 -0400 (AST) Received: from noao.edu (noao.edu [140.252.1.54]) by svr1.postgresql.org (Postfix) with ESMTP id 1F52ED1B43A for ; Fri, 5 Dec 2003 17:51:52 -0400 (AST) Received: from weaver.tuc.noao.edu ([140.252.38.8] verified) by noao.edu (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 9794474; Fri, 05 Dec 2003 14:51:48 -0700 Received: from weaver.tuc.noao.edu (localhost.localdomain [127.0.0.1]) by weaver.tuc.noao.edu (8.12.8/8.12.5) with ESMTP id hB5Lpm7Q018261; Fri, 5 Dec 2003 14:51:48 -0700 Received: (from swampler@localhost) by weaver.tuc.noao.edu (8.12.8/8.12.8/Submit) id hB5LpmxV018259; Fri, 5 Dec 2003 14:51:48 -0700 X-Authentication-Warning: weaver.tuc.noao.edu: swampler set sender to swampler@noao.edu using -f Subject: Help tracking down problem with inserts slowing down... From: Steve Wampler Reply-To: swampler@noao.edu To: Postgres-performance Content-Type: text/plain Content-Transfer-Encoding: 7bit Organization: National Solar Observatory Message-Id: <1070661108.20063.86.camel@weaver.tuc.noao.edu> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Fri, 05 Dec 2003 14:51:48 -0700 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/135 X-Sequence-Number: 4995 I need some help tracking down a sudden, massive slowdown in inserts in one of our databases. PG: 7.2.3 (RedHat 8.0) Background. We currently run nearly identical systems at two sites: Site A is a 'lab' site used for development, Site B is a production site. The databases in question have identical structure: A simple table with 4 columns with a trigger function on inserts (which checks to see if the entry already exists, and if so, changes the insert into an update...) A simple view with 4 columns into the above table. All access is through jdbc (JDK 1.3.1, jdbc 7.1-1.3), postgresql.conf's are identical. The two sites were performing at comparable speeds until a few days ago, when we deleted several million records from each database and then did a vacuum full; analyze on both. Now inserts at Site B are several orders of magnitude slower than at Site A. The odd thing is that Site B's DB now has only 60,000 records while Site A's is up around 3 million. Inserts at A average 63ms, inserts at B are now up at 4.5 seconds! EXPLAIN doesn't show any difference between the two. Can someone suggest ways to track this down? I don't know much about postgresql internals/configuration. Thanks! Steve -- Steve Wampler -- swampler@noao.edu The gods that smiled on your birth are now laughing out loud. From pgsql-performance-owner@postgresql.org Fri Dec 5 19:39:36 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 79B24D1B496 for ; Fri, 5 Dec 2003 23:39:34 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 46967-09 for ; Fri, 5 Dec 2003 19:39:06 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 5B260D1B480 for ; Fri, 5 Dec 2003 19:39:03 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 0BAAF1E36; Fri, 5 Dec 2003 18:38:49 -0500 (EST) To: swampler@noao.edu Cc: Postgres-performance Subject: Re: Help tracking down problem with inserts slowing From: Neil Conway In-Reply-To: <1070661108.20063.86.camel@weaver.tuc.noao.edu> (Steve Wampler's message of "Fri, 05 Dec 2003 14:51:48 -0700") References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> Date: Fri, 05 Dec 2003 18:38:47 -0500 Message-ID: <87fzfy96ns.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/136 X-Sequence-Number: 4996 Steve Wampler writes: > PG: 7.2.3 (RedHat 8.0) You're using PG 7.2.3 with the PG 7.1 JDBC driver; FWIW, upgrading to newer software is highly recommended. > The two sites were performing at comparable speeds until a few days > ago, when we deleted several million records from each database and > then did a vacuum full; analyze on both. Now inserts at Site B are > several orders of magnitude slower than at Site A. Two thoughts: (1) Can you confirm that the VACUUM FULL on site B actually removed all the tuples you intended it to remove? Concurrent transactions can limit the amount of data that VACUUM FULL is able to reclaim. If you run contrib/pgstattuple (or compare the database's disk consumption with the number of live rows in it), you should be able to tell. (2) Look at the EXPLAIN for the SELECTs generated by the ON INSERT trigger -- is there any difference between site A and B? -Neil From pgsql-performance-owner@postgresql.org Fri Dec 5 21:23:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0EA5ED1B488 for ; Sat, 6 Dec 2003 01:23:37 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 57911-07 for ; Fri, 5 Dec 2003 21:23:09 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 65888D1B46D for ; Fri, 5 Dec 2003 21:23:05 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB61MiIt030448 for ; Fri, 5 Dec 2003 17:22:44 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Fri, 5 Dec 2003 17:22:44 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Fri, 05 Dec 2003 17:22:44 -0800 Subject: Re: tuning questions From: Jack Coates To: pgsql-performance In-Reply-To: <200312050926.05155.josh@agliodbs.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> <1070584326.18838.235.camel@cletus.lyris.com> <200312050926.05155.josh@agliodbs.com> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070673761.13542.534.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Fri, 05 Dec 2003 17:22:42 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/137 X-Sequence-Number: 4997 On Fri, 2003-12-05 at 09:26, Josh Berkus wrote: > Jack, > > > The frustrating thing is, we also have a UP P3-500 with 512M RAM and two > > IDE drives with the same PG install which is doing okay with this load > > -- still half the speed of MS-SQL2K, but usable. I'm at a loss. > > Overall, I'm really getting the feeling that this procedure was optimized for > Oracle and/or MSSQL and is hitting some things that aren't such a good idea > for PostgreSQL. I highly suggest that you try using log_duration and > log_statement (and in 7.4 log_min_duration_statement) to try to locate which > particular statements are taking the longest. I'll definitely buy that as round two of optimization, but round one is still "it's faster on the slower server." hdparm -I is identical between the boxes, filesystem structure layout is identical, disk organization isn't identical, but far worse: the UP low ram box has PG on /dev/hdb, ew. Predictably, vmstat shows low numbers... but steady numbers. dev is the box which goes fast, and I was wrong, it's actually a 2GHz P4. rufus is the box which goes slow. During the big fetch: dev bi sits around 2000 blocks for twenty seconds while bo is around 50 blocks, then bo jumps to 800 or so while the data is returned, then we're done. rufus bi starts at 16000 blocks, then drops steadily while bo climbs. After a minute or so, bi stabilizes at 4096 blocks, then bo bursts to return the data. Then the next fetch starts, and it's bi of 500, bo of 300 for several minutes. These observations certainly all point to Eric and Thierry's recommendations to better organize the filesystem and get faster disks.. except that the dev box gets acceptable performance. So, I've dug into postgresql.conf on dev and rufus, and here's what I found: RUFUS how much ram do you have? 75% converted to 8K pages of that for effective_cache 15% of that or 512M, whichever is larger, converted to 8K pages for shared_buffers 15% of that converted to 8K pages for vacuum_mem how many messages will you send between vacuums? divide that by 2 and divide by 6 for max_fsm_pages DEV how much ram do you have? 48% converted to 8K pages of that for effective_cache 6.5% of that or 512M, whichever is larger, converted to 8K pages for shared_buffers 52% of that converted to 8K pages for vacuum_mem max_fsm_pages untouched on this box. I adjusted rufus's configuration to match those percentages, but left max_fsm_pages dialed up to 500000. Now Rufus's vmstat shows much better behavior: bi 12000 blocks gradually sloping down to 3000 during the big select, bo steady until it's ready to return. As more jobs come in, we see overlap areas where bi is 600-ish and bo is 200-ish, but they only last a few tens of seconds. The big selects are still a lot slower than they are on the smaller database and overall performance is still unacceptable. Next I dialed max_fsm_pages back down to 10000 -- no change. Hm, maybe it's been too long since the last vacuumdb --analyze, let's give it another. hdparm -Tt shows that disk performance is crappo on rufus, half what it is on dev -- and freaking dev is using 16 bit IO! This is a motherboard IDE controller issue. South Bridge: VIA vt8233 Revision: ISA 0x0 IDE 0x6 That's it, I'm throwing out this whole test series and starting over with different hardware. Database server is now a dual 2GHz Xeon with 2GB RAM & 2940UW SCSI, OS and PG's logs on 36G drive, PG data on 9GB drive. Data is importing now and I'll restart the tests tonight. -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Fri Dec 5 22:55:25 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7EB48D1B445 for ; Sat, 6 Dec 2003 02:55:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 65049-06 for ; Fri, 5 Dec 2003 22:54:54 -0400 (AST) Received: from lakemtao04.cox.net (lakemtao04.cox.net [68.1.17.241]) by svr1.postgresql.org (Postfix) with ESMTP id D3F88D1B43E for ; Fri, 5 Dec 2003 22:54:49 -0400 (AST) Received: from [192.168.0.13] ([68.105.167.134]) by lakemtao04.cox.net (InterMail vM.5.01.06.05 201-253-122-130-105-20030824) with ESMTP id <20031206025451.PPKC19895.lakemtao04.cox.net@[192.168.0.13]>; Fri, 5 Dec 2003 21:54:51 -0500 From: Robert Treat To: swampler@noao.edu, Postgres-performance Subject: Re: Help tracking down problem with inserts slowing down... Date: Fri, 5 Dec 2003 21:54:52 -0500 User-Agent: KMail/1.5 References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> In-Reply-To: <1070661108.20063.86.camel@weaver.tuc.noao.edu> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312052154.52223.xzilla@users.sourceforge.net> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/138 X-Sequence-Number: 4998 On Friday 05 December 2003 16:51, Steve Wampler wrote: > I need some help tracking down a sudden, massive slowdown > in inserts in one of our databases. > > PG: 7.2.3 (RedHat 8.0) > > Background. We currently run nearly identical systems > at two sites: Site A is a 'lab' site used for development, > Site B is a production site. > > The databases in question have identical structure: > > A simple table with 4 columns with a trigger function > on inserts (which checks to see if the entry already > exists, and if so, changes the insert into an update...) > A simple view with 4 columns into the above table. > > All access is through jdbc (JDK 1.3.1, jdbc 7.1-1.3), > postgresql.conf's are identical. > > The two sites were performing at comparable speeds until > a few days ago, when we deleted several million records > from each database and then did a vacuum full; analyze > on both. Now inserts at Site B are several orders of > magnitude slower than at Site A. The odd thing is that > Site B's DB now has only 60,000 records while Site A's is > up around 3 million. Inserts at A average 63ms, inserts > at B are now up at 4.5 seconds! > > EXPLAIN doesn't show any difference between the two. > > Can someone suggest ways to track this down? I don't know > much about postgresql internals/configuration. > What does explain analyze show for the insert query? Are there FK and/or Indexes involved here? Did you you reindex? A vacuum verbose could give you a good indication if you need to reindex, compare the # of pages in the index with the # in the table. Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Sat Dec 6 09:19:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8B79AD1B480 for ; Sat, 6 Dec 2003 13:19:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31461-07 for ; Sat, 6 Dec 2003 09:19:26 -0400 (AST) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id F403AD1B442 for ; Sat, 6 Dec 2003 09:19:21 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hB6DJL916645; Sat, 6 Dec 2003 08:19:21 -0500 (EST) From: Bruce Momjian Message-Id: <200312061319.hB6DJL916645@candle.pha.pa.us> Subject: Re: Has anyone run on the new G5 yet In-Reply-To: <1070526289.4666.35.camel@localhost.localdomain> To: Paul Tuckfield Date: Sat, 6 Dec 2003 08:19:21 -0500 (EST) Cc: pgsql-performance@postgresql.org, shannyconsulting@earthlink.net X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/139 X-Sequence-Number: 4999 Paul Tuckfield wrote: > Biggest problem I've had in the past w. linux in general is that it > seems to make poor VM choices under heavy filesystem IO. I don't really > get exactly where it's going wrong , but I've had numerous experiences > on older systems where bursty IO would seem to cause paging on the box > (pageout of pieces of the oracle SGA shared memory) which is a > performance disaseter. It seems to happen even when the shared memory > was sized reasonably below the size of physical ram, presumably because > linux is too aggressive in allocating filesystem cache (?) anyway, it > seems to make decisions based on desire for zippy workstation > performance and gets burned on thruput on database servers. I'm > guessing this may be an issue for you , when doing heavy IO. Thing is, > it'll show like you're IO bound kindof because you're thrashing. This is not surprising. There has always been an issue with dynamic buffer cache systems contending with memory used by processes. It takes a long time to get the balance right, and still there might be cases where it gets things wrong. Isn't there a Linux option to lock shared memory in to RAM? If so, we should document this in our manuals, but right now, there is no mention of it. -- 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 Sat Dec 6 13:09:36 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 153F8D1B436 for ; Sat, 6 Dec 2003 17:09:35 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49392-10 for ; Sat, 6 Dec 2003 13:09:03 -0400 (AST) Received: from mail.facnd.com (unknown [66.173.100.50]) by svr1.postgresql.org (Postfix) with ESMTP id 1CA1FD1B45A for ; Sat, 6 Dec 2003 13:09:03 -0400 (AST) Received: by mail.facnd.com (Postfix on SuSE Linux 7.3 (i386), from userid 553) id A581F1EF2; Sat, 6 Dec 2003 11:09:16 -0600 (CST) Received: from rob (unknown [192.1.1.100]) by mail.facnd.com (Postfix on SuSE Linux 7.3 (i386)) with ESMTP id 738301EDE for ; Sat, 6 Dec 2003 11:09:09 -0600 (CST) Reply-To: From: "Rob Sell" To: Subject: Pgbench results Date: Sat, 6 Dec 2003 11:09:05 -0600 Organization: Fargo Assembly Company MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.5510 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Thread-Index: AcO8G6dYoeSXmZ47QHWWkdmaMLnjsQ== Message-Id: <20031206170910.738301EDE@mail.facnd.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=2.2 tagged_above=0.0 required=5.0 tests=FORGED_MUA_OUTLOOK, MISSING_OUTLOOK_NAME X-Spam-Level: ** X-Archive-Number: 200312/140 X-Sequence-Number: 5000 Greetings all, I'm wondering is there a website where people can submit their pgbench results along with their hardware and configuration's? If so where are they at? I have yet to find any. I think this could be a very useful tool not only for people looking at setting up a new server but for people trying to tune their db... Thanks Rob From pgsql-performance-owner@postgresql.org Sat Dec 6 13:47:17 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EAE83D1B46C for ; Sat, 6 Dec 2003 17:47:15 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52940-04 for ; Sat, 6 Dec 2003 13:46:44 -0400 (AST) Received: from anchor-post-35.mail.demon.net (anchor-post-35.mail.demon.net [194.217.242.85]) by svr1.postgresql.org (Postfix) with ESMTP id 25689D1B445 for ; Sat, 6 Dec 2003 13:46:39 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-35.mail.demon.net with esmtp (Exim 3.35 #1) id 1ASgVp-0009La-0Z; Sat, 06 Dec 2003 17:46:38 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id B44BC1740C; Sat, 6 Dec 2003 17:46:36 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 239D517333; Sat, 6 Dec 2003 17:46:34 +0000 (GMT) From: Richard Huxton To: LIANHE SHAO , pgsql-performance Subject: Re: query using cpu nearly 100%, why? Date: Sat, 6 Dec 2003 17:46:32 +0000 User-Agent: KMail/1.5 References: <19be88196606.19660619be88@jhmimail.jhmi.edu> In-Reply-To: <19be88196606.19660619be88@jhmimail.jhmi.edu> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312061746.33037.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/141 X-Sequence-Number: 5001 On Friday 05 December 2003 21:03, LIANHE SHAO wrote: > Hello, > I use php as front-end to query our database. When I use System Monitor to > check the usage of cpu and memory, I noticed that the cpu very easily gets > up to 100%. Is that normal? if not, could someone points out possible > reason? The idea is that CPU should go to 100% when there's work to be done, and drop off when the system is idle. There's nothing to be gained with having the CPU at 50% and taking twice as long to perform a task. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Sat Dec 6 15:42:19 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 15685D1B441 for ; Sat, 6 Dec 2003 19:42:19 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66287-03 for ; Sat, 6 Dec 2003 15:41:49 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 24950D1B481 for ; Sat, 6 Dec 2003 15:41:47 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 1133C1E12; Sat, 6 Dec 2003 14:41:41 -0500 (EST) To: LIANHE SHAO Cc: pgsql-performance Subject: Re: query using cpu nearly 100%, why? From: Neil Conway In-Reply-To: <19be88196606.19660619be88@jhmimail.jhmi.edu> (LIANHE SHAO's message of "Fri, 05 Dec 2003 13:03:42 -0800") References: <19be88196606.19660619be88@jhmimail.jhmi.edu> Date: Sat, 06 Dec 2003 14:41:36 -0500 Message-ID: <8765gt91jj.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/142 X-Sequence-Number: 5002 LIANHE SHAO writes: > Hello, I use php as front-end to query our database. When I use > System Monitor to check the usage of cpu and memory, I noticed that > the cpu very easily gets up to 100%. Is that normal? if not, could > someone points out possible reason? You haven't given us nearly enough information about the problem to allow us to provide any meaningful advice. -Neil From pgsql-performance-owner@postgresql.org Sat Dec 6 15:54:37 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5EC36D1B442 for ; Sat, 6 Dec 2003 19:54:36 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68107-04 for ; Sat, 6 Dec 2003 15:54:06 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 86D96D1B468 for ; Sat, 6 Dec 2003 15:54:04 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 7C0281EBB; Sat, 6 Dec 2003 14:54:04 -0500 (EST) To: Shridhar Daithankar Cc: Ivar Zarans , pgsql-performance@postgresql.org Subject: Re: Slow UPADTE, compared to INSERT From: Neil Conway In-Reply-To: <3FD08D6A.4080207@persistent.co.in> (Shridhar Daithankar's message of "Fri, 05 Dec 2003 19:21:38 +0530") References: <20031204185751.GA15893@alcaron.ee> <20031205014514.GA21079@alcaron.ee> <20031205020728.GA21440@alcaron.ee> <200312051008.20317.dev@archonet.com> <20031205123843.GA27170@alcaron.ee> <3FD07EEA.8030007@persistent.co.in> <20031205131325.GB27524@alcaron.ee> <3FD08D6A.4080207@persistent.co.in> Date: Sat, 06 Dec 2003 14:54:03 -0500 Message-ID: <871xrh90ys.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/143 X-Sequence-Number: 5003 Shridhar Daithankar writes: > This is not a bug. It is just that people find it confusing when > postgresql planner consider seemingly same type as different. It certainly is a bug, or at least a deficiency: PostgreSQL planner *could* use the index to process the query, but the planner doesn't consider doing so. The fact that it isn't able to do the necessary type coercion is the *cause* of the bug, not a defence for this behavior. > AFAIK, the fix in CVS is to make indexes operatable with seemingly > compatible types. Which does not change the fact that postgresql can > not upgrade data types on it's own. I'm not sure what you mean by that. In any case, I just checked, and it does seem Tom has fixed this in CVS: template1=# create table abc (b int8); CREATE TABLE template1=# set enable_seqscan = false; SET template1=# create index abc_b_idx on abc (b); CREATE INDEX template1=# explain select * from abc where b = 4; QUERY PLAN ---------------------------------------------------------------------- Index Scan using abc_b_idx on abc (cost=0.00..17.07 rows=5 width=8) Index Cond: (b = 4) (2 rows) Cool! -Neil From pgsql-performance-owner@postgresql.org Sun Dec 7 10:28:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 37CCED1B8C0 for ; Sun, 7 Dec 2003 14:28:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 59777-05 for ; Sun, 7 Dec 2003 10:28:17 -0400 (AST) Received: from noao.edu (noao.edu [140.252.1.54]) by svr1.postgresql.org (Postfix) with ESMTP id 05921D1B518 for ; Sun, 7 Dec 2003 10:28:13 -0400 (AST) Received: from weaver.tuc.noao.edu ([140.252.38.8] verified) by noao.edu (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 9807948; Sun, 07 Dec 2003 07:28:17 -0700 Received: from weaver.tuc.noao.edu (localhost.localdomain [127.0.0.1]) by weaver.tuc.noao.edu (8.12.8/8.12.5) with ESMTP id hB7ESG7Q008423; Sun, 7 Dec 2003 07:28:16 -0700 Received: (from swampler@localhost) by weaver.tuc.noao.edu (8.12.8/8.12.8/Submit) id hB7ESGx1008421; Sun, 7 Dec 2003 07:28:16 -0700 Date: Sun, 7 Dec 2003 07:28:16 -0700 From: Steve Wampler To: Robert Treat Cc: swampler@noao.edu, Postgres-performance Subject: Re: Help tracking down problem with inserts slowing down... Message-ID: <20031207142816.GA8321@weaver.tuc.noao.edu> References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> <200312052154.52223.xzilla@users.sourceforge.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312052154.52223.xzilla@users.sourceforge.net> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/144 X-Sequence-Number: 5004 On Fri, Dec 05, 2003 at 09:54:52PM -0500, Robert Treat wrote: > On Friday 05 December 2003 16:51, Steve Wampler wrote: > > I need some help tracking down a sudden, massive slowdown > > in inserts in one of our databases. > > > > PG: 7.2.3 (RedHat 8.0) > > > > Background. We currently run nearly identical systems > > at two sites: Site A is a 'lab' site used for development, > > Site B is a production site. > > > > The databases in question have identical structure: > > > > A simple table with 4 columns with a trigger function > > on inserts (which checks to see if the entry already > > exists, and if so, changes the insert into an update...) > > A simple view with 4 columns into the above table. > > > > All access is through jdbc (JDK 1.3.1, jdbc 7.1-1.3), > > postgresql.conf's are identical. > > > > The two sites were performing at comparable speeds until > > a few days ago, when we deleted several million records > > from each database and then did a vacuum full; analyze > > on both. Now inserts at Site B are several orders of > > magnitude slower than at Site A. The odd thing is that > > Site B's DB now has only 60,000 records while Site A's is > > up around 3 million. Inserts at A average 63ms, inserts > > at B are now up at 4.5 seconds! > > > > EXPLAIN doesn't show any difference between the two. > > > > Can someone suggest ways to track this down? I don't know > > much about postgresql internals/configuration. > > > > What does explain analyze show for the insert query? > > Are there FK and/or Indexes involved here? Did you you reindex? > A vacuum verbose could give you a good indication if you need to reindex, > compare the # of pages in the index with the # in the table. Thanks Robert! It looks like reindex did the trick. Now I have a general question - what are the relationships between: vacuum, analyze, reindex, and dropping/recreating the indices? That is, which is the following is 'best' (or is there a different ordering that is better)?: (1) vacuum analyze reindex (2) vacuum reindex analyze (3) drop indices vacuum create indices analyze (4) drop indices vacuum analyze create indices And, is reindex equivalent to dropping, then recreating the indices? [it appears to be "no", from what I've just seen, but I don't know...] Thanks! Steve -- Steve Wampler -- swampler@noao.edu The gods that smiled on your birth are now laughing out loud. From pgsql-performance-owner@postgresql.org Sun Dec 7 10:53:04 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6631DD1B518 for ; Sun, 7 Dec 2003 14:53:02 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 61222-06 for ; Sun, 7 Dec 2003 10:52:35 -0400 (AST) Received: from noao.edu (noao.edu [140.252.1.54]) by svr1.postgresql.org (Postfix) with ESMTP id 75BF6D1B48F for ; Sun, 7 Dec 2003 10:52:31 -0400 (AST) Received: from weaver.tuc.noao.edu ([140.252.38.8] verified) by noao.edu (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 9808033; Sun, 07 Dec 2003 07:52:36 -0700 Received: from weaver.tuc.noao.edu (localhost.localdomain [127.0.0.1]) by weaver.tuc.noao.edu (8.12.8/8.12.5) with ESMTP id hB7EqZ7Q008814; Sun, 7 Dec 2003 07:52:35 -0700 Received: (from swampler@localhost) by weaver.tuc.noao.edu (8.12.8/8.12.8/Submit) id hB7EqZOc008812; Sun, 7 Dec 2003 07:52:35 -0700 Date: Sun, 7 Dec 2003 07:52:35 -0700 From: Steve Wampler To: Steve Wampler Cc: Robert Treat , Postgres-performance Subject: Re: Help tracking down problem with inserts slowing down... Message-ID: <20031207145235.GA8734@weaver.tuc.noao.edu> References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> <200312052154.52223.xzilla@users.sourceforge.net> <20031207142816.GA8321@weaver.tuc.noao.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031207142816.GA8321@weaver.tuc.noao.edu> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/145 X-Sequence-Number: 5005 On Fri, Dec 05, 2003 at 09:54:52PM -0500, Robert Treat wrote: >... > A vacuum verbose could give you a good indication if you need to reindex, > compare the # of pages in the index with the # in the table. Hmmm, I have a feeling that's not as obvious as I thought... I can't identify the index (named 'id_index') in the output of vacuum verbose. The closest I can find is: NOTICE: --Relation pg_index-- NOTICE: Pages 2: Changed 0, Empty 0; Tup 56: Vac 0, Keep 0, UnUsed 42. Total CPU 0.00s/0.00u sec elapsed 0.00 sec. Which probably isn't correct, right (the name doesn't seem to match)? The table's entry is: NOTICE: --Relation attributes_table-- NOTICE: Pages 639: Changed 0, Empty 0; Tup 52846: Vac 0, Keep 0, UnUsed 48. Total CPU 0.00s/0.01u sec elapsed 0.01 sec. Thanks! Steve -- Steve Wampler -- swampler@noao.edu The gods that smiled on your birth are now laughing out loud. From pgsql-performance-owner@postgresql.org Sun Dec 7 12:54:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 398F3D1B49F for ; Sun, 7 Dec 2003 16:54:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68515-07 for ; Sun, 7 Dec 2003 12:53:49 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 2BBD4D1B4D1 for ; Sun, 7 Dec 2003 12:53:48 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB7Gqb19022389; Sun, 7 Dec 2003 11:52:37 -0500 (EST) To: Steve Wampler Cc: Robert Treat , Postgres-performance Subject: Re: Help tracking down problem with inserts slowing down... In-reply-to: <20031207145235.GA8734@weaver.tuc.noao.edu> References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> <200312052154.52223.xzilla@users.sourceforge.net> <20031207142816.GA8321@weaver.tuc.noao.edu> <20031207145235.GA8734@weaver.tuc.noao.edu> Comments: In-reply-to Steve Wampler message dated "Sun, 07 Dec 2003 07:52:35 -0700" Date: Sun, 07 Dec 2003 11:52:37 -0500 Message-ID: <22388.1070815957@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/146 X-Sequence-Number: 5006 Steve Wampler writes: > Hmmm, I have a feeling that's not as obvious as I thought... I can't > identify the index (named 'id_index') in the output of vacuum verbose. In 7.2, the index reports look like Index %s: Pages %u; Tuples %.0f. and should appear in the part of the printout that deals with their owning table. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Dec 7 13:29:27 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5B694D1B4B9 for ; Sun, 7 Dec 2003 17:29:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70674-08 for ; Sun, 7 Dec 2003 13:28:51 -0400 (AST) Received: from noao.edu (noao.edu [140.252.1.54]) by svr1.postgresql.org (Postfix) with ESMTP id D8D1CD1B8C3 for ; Sun, 7 Dec 2003 13:28:50 -0400 (AST) Received: from weaver.tuc.noao.edu ([140.252.38.8] verified) by noao.edu (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 9808796; Sun, 07 Dec 2003 10:28:50 -0700 Received: from weaver.tuc.noao.edu (localhost.localdomain [127.0.0.1]) by weaver.tuc.noao.edu (8.12.8/8.12.5) with ESMTP id hB7HSo7Q011516; Sun, 7 Dec 2003 10:28:50 -0700 Received: (from swampler@localhost) by weaver.tuc.noao.edu (8.12.8/8.12.8/Submit) id hB7HSoow011514; Sun, 7 Dec 2003 10:28:50 -0700 X-Authentication-Warning: weaver.tuc.noao.edu: swampler set sender to swampler@noao.edu using -f Subject: Re: Help tracking down problem with inserts slowing From: Steve Wampler Reply-To: swampler@noao.edu To: Neil Conway Cc: Postgres-performance In-Reply-To: <87fzfy96ns.fsf@mailbox.samurai.com> References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> <87fzfy96ns.fsf@mailbox.samurai.com> Content-Type: text/plain Content-Transfer-Encoding: 7bit Organization: National Solar Observatory Message-Id: <1070818130.20063.145.camel@weaver.tuc.noao.edu> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Sun, 07 Dec 2003 10:28:50 -0700 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/147 X-Sequence-Number: 5007 On Fri, 2003-12-05 at 16:38, Neil Conway wrote: > > (1) Can you confirm that the VACUUM FULL on site B actually > removed all the tuples you intended it to remove? Concurrent > transactions can limit the amount of data that VACUUM FULL is > able to reclaim. If you run contrib/pgstattuple (or compare > the database's disk consumption with the number of live rows > in it), you should be able to tell. Hmmm, I installed 7.2.3 from RPMs, but the contrib package seems to be missing the pgstattuple library code. (According to the readme, I should do: $ make $ make install $ psql -e -f /usr/local/pgsql/share/contrib/pgstattuple.sql test but the first two lines don't make sense with the binary rpm distribution and trying the last line as (for my world): ->psql -e -f /usr/share/pgsql/contrib/pgstattuple.sql farm.devel.configdb yields: DROP FUNCTION pgstattuple(NAME); psql:/usr/share/pgsql/contrib/pgstattuple.sql:1: ERROR: RemoveFunction: function 'pgstattuple(name)' does not exist CREATE FUNCTION pgstattuple(NAME) RETURNS FLOAT8 AS '$libdir/pgstattuple', 'pgstattuple' LANGUAGE 'c' WITH (isstrict); psql:/usr/share/pgsql/contrib/pgstattuple.sql:4: ERROR: stat failed on file '$libdir/pgstattuple': No such file or directory I don't need this right now (a reindex seems to have fixed our problem for now...), but it sounds like it would be useful in the future. Thanks! Steve -- Steve Wampler -- swampler@noao.edu The gods that smiled on your birth are now laughing out loud. From pgsql-performance-owner@postgresql.org Sun Dec 7 17:16:44 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 28EEAD1B464 for ; Sun, 7 Dec 2003 21:16:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 00464-03 for ; Sun, 7 Dec 2003 17:16:13 -0400 (AST) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 8395FD1B4A1 for ; Sun, 7 Dec 2003 17:16:10 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hB7LGBG27161; Sun, 7 Dec 2003 16:16:11 -0500 (EST) From: Bruce Momjian Message-Id: <200312072116.hB7LGBG27161@candle.pha.pa.us> Subject: Re: autovacuum daemon stops doing work after about an In-Reply-To: <16335.43272.273411.653445@yertle.int.kciLink.com> To: Vivek Khera Date: Sun, 7 Dec 2003 16:16:11 -0500 (EST) Cc: pgsql-performance@postgresql.org X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/148 X-Sequence-Number: 5008 This has been fixed and will be in 7.4.1. --------------------------------------------------------------------------- Vivek Khera wrote: > >>>>> "LR" == Larry Rosenman writes: > > >> I'd be curious to see the output of this program on other platforms > >> and other compilers. I'm using gcc 2.95.4 as shipped with FreeBSD > >> 4.8+. > LR> this is with the UnixWare compiler: > LR> $ cc -O -o testvk testvk.c > LR> $ ./testvk > LR> seconds = 3509 > LR> seconds1 = 3509000000 > LR> useconds = -452486 > LR> stepped diff = 3508547514 > LR> seconds2 = -785967296 > LR> seconds3 = 3509000000 > LR> diff = -786419782 > LR> long long diff = 3508547514 > LR> $ > > LR> I think this is a C bug. > > Upon further reflection, I think so to. The entire RHS is long's so > the arithmetic is done in longs, then assigned to a long long when > done (after things have overflowed). Forcing any one of the RHS > values to be long long causes the arithmetic to all be done using long > longs, and then you get the numbers you expect. > > I think you only notice this in autovacuum when it takes a long time > to complete the work, like my example of about 3500 seconds. > > ---------------------------(end of broadcast)--------------------------- > TIP 3: 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 > -- 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 Sun Dec 7 20:47:32 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 43765D1B436 for ; Mon, 8 Dec 2003 00:47:28 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 17543-09 for ; Sun, 7 Dec 2003 20:47:00 -0400 (AST) Received: from pns.mm.eutelsat.org (pns.mm.eutelsat.org [194.214.173.227]) by svr1.postgresql.org (Postfix) with ESMTP id 78E3DD1B46F for ; Sun, 7 Dec 2003 20:46:56 -0400 (AST) Received: from nts-03.mm.eutelsat.org (localhost [127.0.0.1]) by pns.mm.eutelsat.org (8.11.6/linuxconf) with ESMTP id hB80oCs01675 for ; Mon, 8 Dec 2003 01:50:12 +0100 Received: from bigfoot.com (accesspoint.mm.eutelsat.org [194.214.173.4]) by nts-03.mm.eutelsat.org (8.11.6/linuxconf) with ESMTP id hB80ZRR16821 for ; Mon, 8 Dec 2003 01:35:27 +0100 Message-ID: <3FD3C9DA.5040805@bigfoot.com> Date: Mon, 08 Dec 2003 01:46:18 +0100 From: Gaetano Mendola User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031206 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Subject: Re: autovacuum daemon stops doing work after about an hour References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <16335.25079.106697.7521@yertle.int.kciLink.com> <3FCF7E74.9060604@bigfoot.com> In-Reply-To: <3FCF7E74.9060604@bigfoot.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/149 X-Sequence-Number: 5009 Gaetano Mendola wrote: > Vivek Khera wrote: > >>>>>>> "MTO" == Matthew T O'Connor writes: >> >> >> >>>> Then it just sits there. I started it at 11:35am, and it is now >>>> 3:30pm. >> >> >> >> MTO> Weird.... Alphabetically speaking, is vkmlm."public"."user_list" >> be the >> MTO> last table in the last schema in the last database? You are running >> >> conveniently, yes it is... >> >> MTO> with -d4, so you would get a message about going to sleep shortly >> after >> MTO> dealing with the last table, but you didn't get the sleep >> message, so I >> MTO> don't think the problem is that pg_autovacuum is sleeping for an >> MTO> inordinate amount time. >> >> The only sleep logged was >> >> [2003-12-03 04:47:13 PM] 1 All DBs checked in: 84996853 usec, will >> sleep for 469 secs. > > > What I seen is: > > > # tail -f auto.log > [2003-12-04 07:10:18 PM] reltuples: 72; relpages: 1 > [2003-12-04 07:10:18 PM] curr_analyze_count: 72; cur_delete_count: 0 > [2003-12-04 07:10:18 PM] ins_at_last_analyze: 72; del_at_last_vacuum: 0 > [2003-12-04 07:10:18 PM] insert_threshold: 572; delete_threshold 536 > [2003-12-04 07:10:18 PM] table name: empdb."public"."contracts" > [2003-12-04 07:10:18 PM] relfilenode: 17784; relisshared: 0 > [2003-12-04 07:10:18 PM] reltuples: 347; relpages: 5 > [2003-12-04 07:10:18 PM] curr_analyze_count: 347; cur_delete_count: 0 > [2003-12-04 07:10:18 PM] ins_at_last_analyze: 347; del_at_last_vacuum: 0 > [2003-12-04 07:10:18 PM] insert_threshold: 847; delete_threshold 673 > > > [ 5 minutes of delay ] <----- LOOK THIS > > > [2003-12-04 07:10:18 PM] 503 All DBs checked in: 179396 usec, will sleep > for 300 secs. > [2003-12-04 07:15:19 PM] 504 All DBs checked in: 98814 usec, will sleep > for 300 secs. > > I think is a good Idea put a fflush after: > > fprintf(LOGOUTPUT, "[%s] %s\n", timebuffer, logentry); Was I wrong ? If you are watching in tail the log believeme is really annoying. Regards Gaetano Mendola From pgsql-performance-owner@postgresql.org Sun Dec 7 21:29:30 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 77DDCD1C951 for ; Mon, 8 Dec 2003 01:29:28 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 22383-10 for ; Sun, 7 Dec 2003 21:29:03 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id E7B63D1B48B for ; Sun, 7 Dec 2003 21:28:59 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hB81SxbI091483 for ; Mon, 8 Dec 2003 01:28:59 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hB81JZKN090568 for pgsql-performance@postgresql.org; Mon, 8 Dec 2003 01:19:35 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: autovacuum daemon stops doing work after about an hour Date: Sun, 07 Dec 2003 20:17:33 -0500 Organization: cbbrowne Computing Inc Lines: 15 Message-ID: References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <16335.25079.106697.7521@yertle.int.kciLink.com> <3FCF7E74.9060604@bigfoot.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/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:ooK8UmYiF4cDsjXv4BcRCfE2D7g= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/150 X-Sequence-Number: 5010 The world rejoiced as mendola@bigfoot.com (Gaetano Mendola) wrote: > I think is a good Idea put a fflush after: > > fprintf(LOGOUTPUT, "[%s] %s\n", timebuffer, logentry); I thought I had put fflush()es at all the interesting locations... Apparently it was an error to not go to the effort of making sure it worked well on FreeBSD. (It was on my list, but I never got the Round Tuits...) There's an AMD-64 box coming in soon, targeted at FreeBSD, so that should change... -- let name="cbbrowne" and tld="acm.org" in name ^ "@" ^ tld;; http://www3.sympatico.ca/cbbrowne/linux.html What would a chair look like, if your knees bent the other way? From pgsql-performance-owner@postgresql.org Mon Dec 8 01:28:03 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 40957D1B469 for ; Mon, 8 Dec 2003 05:28:02 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 51692-10 for ; Mon, 8 Dec 2003 01:27:31 -0400 (AST) Received: from mta2.srv.hcvlny.cv.net (mta2.srv.hcvlny.cv.net [167.206.5.68]) by svr1.postgresql.org (Postfix) with ESMTP id 4F6D4D1B436 for ; Mon, 8 Dec 2003 01:27:30 -0400 (AST) Received: from zeut.net (ool-4352919e.dyn.optonline.net [67.82.145.158]) by mta2.srv.hcvlny.cv.net (iPlanet Messaging Server 5.2 HotFix 1.16 (built May 14 2003)) with ESMTP id <0HPK00I889TOJ8@mta2.srv.hcvlny.cv.net> for pgsql-performance@postgresql.org; Mon, 08 Dec 2003 00:27:25 -0500 (EST) Date: Mon, 08 Dec 2003 00:27:29 -0500 From: "Matthew T. O'Connor" Subject: Re: autovacuum daemon stops doing work after about an hour In-reply-to: To: Christopher Browne Cc: pgsql-performance@postgresql.org Message-id: <3FD40BC1.7050906@zeut.net> 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/5.0 (X11; U; Linux i686; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <16335.25079.106697.7521@yertle.int.kciLink.com> <3FCF7E74.9060604@bigfoot.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/151 X-Sequence-Number: 5011 Christopher Browne wrote: >The world rejoiced as mendola@bigfoot.com (Gaetano Mendola) wrote: > > >>I think is a good Idea put a fflush after: >> >>fprintf(LOGOUTPUT, "[%s] %s\n", timebuffer, logentry); >> >> > >I thought I had put fflush()es at all the interesting locations... > > I just looked through the code, I think there are fflush()es at all but one interesting locations. The last log_entry call before sleeping doesn't have an fflush call after it. I'll submit a patch that adds it. >Apparently it was an error to not go to the effort of making sure it >worked well on FreeBSD. (It was on my list, but I never got the Round >Tuits...) There's an AMD-64 box coming in soon, targeted at FreeBSD, >so that should change... > > Yeah, FreeBSD testing would have been nice, but I don't have access to any FreeBSD boxes so..... From pgsql-performance-owner@postgresql.org Tue Dec 9 16:00:17 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 02C09D1B445 for ; Mon, 8 Dec 2003 14:04:48 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 26196-04 for ; Mon, 8 Dec 2003 10:04:19 -0400 (AST) Received: from smtp02do.de.uu.net (smtp02do.de.uu.net [192.76.144.69]) by svr1.postgresql.org (Postfix) with ESMTP id D29F0D1B4B1 for ; Mon, 8 Dec 2003 10:04:14 -0400 (AST) Received: from bommel.kecam-han.de ([193.99.158.1]) by smtp02do.de.uu.net (8.9.3p2/5.5.5) with ESMTP id PAA22793 for ; Mon, 8 Dec 2003 15:04:17 +0100 (MET) Received: from mailrelay.kecam-han.de (ldap [10.9.1.54]) by bommel.kecam-han.de (8.11.6+Sun/8.9.1) with ESMTP id hB8E4c609689 for ; Mon, 8 Dec 2003 15:04:38 +0100 (MET) Received: from ke-elektronik.de (localhost [127.0.0.1]) by mailrelay.kecam-han.de (8.12.2/8.12.2) with ESMTP id hB8E3ZUr019106 for ; Mon, 8 Dec 2003 15:03:55 +0100 (MET) Message-ID: <3FD484B6.EF39A92F@ke-elektronik.de> Date: Mon, 08 Dec 2003 15:03:34 +0100 From: Hartmut Raschick Organization: ke Kommunikations-Elektronik GmbH X-Mailer: Mozilla 4.79 [en] (X11; U; SunOS 5.6 sun4u) X-Accept-Language: de, ru, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: TRUNCATE veeeery slow compared to DELETE in 7.4 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/162 X-Sequence-Number: 5022 has anyone else noticed a huge difference in "DELETE TABLE " vs. "TRUNCATE " starting w/postgres 7.4? putting aside details (num rows, indexes....): ca. 300 tables (already empty if desired...) ALL to be emptied (via batch file). here's a small "time pgsql -f kill_all" output: DELETE: 1) 0.03u 0.04s 0:02.46 2.8% (already empty) 2) 0.05u 0.06s 0:01.19 9.2% (already empty) TRUNCATE: 1) 0.10u 0.06s 6:58.66 0.0% (already empty, compile runnig simult.) 2) 0.10u 0.02s 2:51.71 0.0% (already empty) lovely, innit? settings in 7.4 (wal, shm...) are as for 7.3.x unless dead or (in their 7.4 default version) even higher. glimpsing at the quantify output (of the truncate version) it looks as if this is "for (i = 0; i < all; i++)" whereas (from exec. time) delete does "\rm -rf" is this a pay-off for autocommit gone away? a conspiracy? ...what am i saying... we are using TRUNCATE btw, because someone once noted that this was "good style", saying: "yes, i want to empty the whole thing", not: "oops! forgot the where-clause, sorry for your table!" well, enlight me, please! P.S.: Grammarians dispute - and the case is still before the courts. - Horace, Epistles (Ars Poetica) -- Hartmut "Hardy" Raschick / Dept. t2 ke Kommunikations-Elektronik GmbH Wohlenberstr. 3, 30179 Hannover Phone: ++49 (0)511 6747-564 Fax: ++49 (0)511 6747-340 e-Mail: hartmut.raschick@ke-elektronik.de http://www.ke-elektronik.de From pgsql-performance-owner@postgresql.org Mon Dec 8 11:15:32 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E3166D1B4BA for ; Mon, 8 Dec 2003 15:15:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31612-10 for ; Mon, 8 Dec 2003 11:15:02 -0400 (AST) Received: from noao.edu (noao.edu [140.252.1.54]) by svr1.postgresql.org (Postfix) with ESMTP id A752AD1B449 for ; Mon, 8 Dec 2003 11:14:54 -0400 (AST) Received: from weaver.tuc.noao.edu ([140.252.38.8] verified) by noao.edu (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 9815943; Mon, 08 Dec 2003 08:14:53 -0700 Received: from weaver.tuc.noao.edu (localhost.localdomain [127.0.0.1]) by weaver.tuc.noao.edu (8.12.8/8.12.5) with ESMTP id hB8FEr7Q001182; Mon, 8 Dec 2003 08:14:53 -0700 Received: (from swampler@localhost) by weaver.tuc.noao.edu (8.12.8/8.12.8/Submit) id hB8FErhP001180; Mon, 8 Dec 2003 08:14:53 -0700 X-Authentication-Warning: weaver.tuc.noao.edu: swampler set sender to swampler@noao.edu using -f Subject: Re: Help tracking down problem with inserts slowing From: Steve Wampler Reply-To: swampler@noao.edu To: Tom Lane Cc: Robert Treat , Postgres-performance In-Reply-To: <22388.1070815957@sss.pgh.pa.us> References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> <200312052154.52223.xzilla@users.sourceforge.net> <20031207142816.GA8321@weaver.tuc.noao.edu> <20031207145235.GA8734@weaver.tuc.noao.edu> <22388.1070815957@sss.pgh.pa.us> Content-Type: text/plain Content-Transfer-Encoding: 7bit Organization: National Solar Observatory Message-Id: <1070896493.20063.159.camel@weaver.tuc.noao.edu> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Mon, 08 Dec 2003 08:14:53 -0700 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/152 X-Sequence-Number: 5012 On Sun, 2003-12-07 at 09:52, Tom Lane wrote: > Steve Wampler writes: > > Hmmm, I have a feeling that's not as obvious as I thought... I can't > > identify the index (named 'id_index') in the output of vacuum verbose. > > In 7.2, the index reports look like > Index %s: Pages %u; Tuples %.0f. > and should appear in the part of the printout that deals with their > owning table. Thanks, Tom. Are there any reasons why it would not appear?: ------------------------------------------------------------- farm.devel.configdb=# vacuum verbose attributes_table; NOTICE: --Relation attributes_table-- NOTICE: Pages 1389: Changed 0, Empty 0; Tup 111358: Vac 0, Keep 0, UnUsed 51. Total CPU 0.00s/0.02u sec elapsed 0.03 sec. NOTICE: --Relation pg_toast_1743942-- NOTICE: Pages 0: Changed 0, Empty 0; Tup 0: Vac 0, Keep 0, UnUsed 0. Total CPU 0.00s/0.00u sec elapsed 0.00 sec. VACUUM farm.devel.configdb=# \d attributes_table Table "attributes_table" Column | Type | Modifiers --------+--------------------------+--------------- id | character varying(64) | not null name | character varying(64) | not null units | character varying(32) | value | text | time | timestamp with time zone | default now() Indexes: id_index Primary key: attributes_table_pkey Triggers: trigger_insert --------------------------------------------------------------- The odd thing is that I could have sworn it appeared yesterday... -- Steve Wampler -- swampler@noao.edu The gods that smiled on your birth are now laughing out loud. From pgsql-performance-owner@postgresql.org Mon Dec 8 11:36:18 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 79985D1B4BA for ; Mon, 8 Dec 2003 15:36:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 35649-09 for ; Mon, 8 Dec 2003 11:35:50 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 3E177D1B481 for ; Mon, 8 Dec 2003 11:35:46 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB8FZn19009619; Mon, 8 Dec 2003 10:35:50 -0500 (EST) To: swampler@noao.edu Cc: Robert Treat , Postgres-performance Subject: Re: Help tracking down problem with inserts slowing In-reply-to: <1070896493.20063.159.camel@weaver.tuc.noao.edu> References: <1070661108.20063.86.camel@weaver.tuc.noao.edu> <200312052154.52223.xzilla@users.sourceforge.net> <20031207142816.GA8321@weaver.tuc.noao.edu> <20031207145235.GA8734@weaver.tuc.noao.edu> <22388.1070815957@sss.pgh.pa.us> <1070896493.20063.159.camel@weaver.tuc.noao.edu> Comments: In-reply-to Steve Wampler message dated "Mon, 08 Dec 2003 08:14:53 -0700" Date: Mon, 08 Dec 2003 10:35:49 -0500 Message-ID: <9618.1070897749@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/153 X-Sequence-Number: 5013 Steve Wampler writes: > Thanks, Tom. Are there any reasons why it would not appear?: Oh, I shoulda read the code more carefully. I was looking at the bottom of lazy_scan_index, where the printout is done, and failed to notice the test at the top: /* * If the index is not partial, skip the scan, and just assume it has * the same number of tuples as the heap. */ So for ordinary indexes, nothing will appear unless vacuum has actual work to do (that is, it recycled at least one dead tuple in the table). Short answer: update or delete some row in the table, and then try vacuum verbose. Alternatively, you can just look at the pg_class row for the index. relpages and reltuples will contain the info you are after ... and they are certainly up to date at this point ;-) regards, tom lane From pgsql-performance-owner@postgresql.org Mon Dec 8 13:44:29 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D58F8D1B46F for ; Mon, 8 Dec 2003 17:44:27 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 65976-05 for ; Mon, 8 Dec 2003 13:43:56 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 3D613D1B491 for ; Mon, 8 Dec 2003 13:43:54 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB8HhlIt007581 for ; Mon, 8 Dec 2003 09:43:47 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Mon, 8 Dec 2003 09:43:47 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Mon, 08 Dec 2003 09:43:47 -0800 Subject: Re: tuning questions From: Jack Coates To: pgsql-performance In-Reply-To: <1070673761.13542.534.camel@cletus.lyris.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> <1070584326.18838.235.camel@cletus.lyris.com> <200312050926.05155.josh@agliodbs.com> <1070673761.13542.534.camel@cletus.lyris.com> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070905424.16088.63.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Mon, 08 Dec 2003 09:43:45 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/154 X-Sequence-Number: 5014 On Fri, 2003-12-05 at 17:22, Jack Coates wrote: ... > That's it, I'm throwing out this whole test series and starting over > with different hardware. Database server is now a dual 2GHz Xeon with > 2GB RAM & 2940UW SCSI, OS and PG's logs on 36G drive, PG data on 9GB > drive. Data is importing now and I'll restart the tests tonight. Sorry to reply at myself, but thought I'd note that the performance is practically unchanged by moving to better hardware and separating logs and data onto different spindles. Although the disks are twice as fast by hdparm -Tt, their behavior as shown by iostat and vmstat is little different between dual and dev (single P4-2GHz/512MB/(2)IDE drives). Dual is moderately faster than my first, IDE-based testbed (about 8%), but still only 30% as fast as the low-powered dev. I've been running vacuumdb --analyze and/or vaccuumdb --full between each config change, and I also let the job run all weekend. Saturday it got --analyze every three hours or so, Sunday it got --analyze once in the morning. None of these vacuumdb's are making any difference. Theories at this point, in no particular order: a) major differences between my 7.3.4 from source (compiled with no options) and dev's 7.3.2-1PGDG RPMs. Looking at the spec file doesn't reveal anything glaring to me, but is there something I'm missing? b) major differences between my kernel 2.4.18-14smp (RH8) and dev's kernel 2.4.18-3 (RH7.3). c) phase of the moon. While SQL optimization is likely to improve performance across the board, it doesn't explain the differences between these two systems and I'd like to avoid it as a theory until the fast box can perform as well as the slow box. Any ideas? Thanks in advance, -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Mon Dec 8 14:41:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4EB45D1BA62 for ; Mon, 8 Dec 2003 18:41:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 74864-07 for ; Mon, 8 Dec 2003 14:41:15 -0400 (AST) Received: from lorax.kcilink.com (lorax.kciLink.com [206.112.95.1]) by svr1.postgresql.org (Postfix) with ESMTP id 8F7E4D1B4C1 for ; Mon, 8 Dec 2003 14:41:14 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id 6B2C53E55 for ; Mon, 8 Dec 2003 13:41:14 -0500 (EST) Received: from lorax.kcilink.com ([127.0.0.1]) by localhost (lorax.kcilink.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 97237-03 for ; Mon, 8 Dec 2003 13:41:13 -0500 (EST) Received: from lorax.kciLink.com (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id A5E463E28 for ; Mon, 8 Dec 2003 13:41:13 -0500 (EST) Received: (from news@localhost) by lorax.kciLink.com (8.12.9p2/8.12.9/Submit) id hB8IfDT7082051 for pgsql-performance@postgresql.org; Mon, 8 Dec 2003 13:41:13 -0500 (EST) (envelope-from news) To: pgsql-performance@postgresql.org Path: not-for-mail From: Vivek Khera Newsgroups: ml.postgres.performance Subject: Re: autovacuum daemon stops doing work after about an hour Date: Mon, 08 Dec 2003 13:41:12 -0500 Organization: Khera Communications, Inc., Rockville, MD Lines: 20 Message-ID: References: <16332.63508.196365.801418@yertle.int.kciLink.com> <1070516326.1455.11.camel@zedora.zeut.net> <16335.25079.106697.7521@yertle.int.kciLink.com> <3FCF7E74.9060604@bigfoot.com> <3FD40BC1.7050906@zeut.net> NNTP-Posting-Host: yertle.kcilink.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: lorax.kcilink.com 1070908873 32027 216.194.193.105 (8 Dec 2003 18:41:13 GMT) X-Complaints-To: daemon@kciLink.com NNTP-Posting-Date: Mon, 8 Dec 2003 18:41:13 +0000 (UTC) User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, berkeley-unix) Cancel-Lock: sha1:OM58FMOrWScQove6ySmsaM8yG9c= X-Virus-Scanned: by amavisd-new at kciLink.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/155 X-Sequence-Number: 5015 >>>>> "MTO" == Matthew T O'Connor writes: MTO> Yeah, FreeBSD testing would have been nice, but I don't have access to MTO> any FreeBSD boxes so..... FWIW, with the fflush() added after that sleep, and the fix to the long long computation of sleep time to keep it from overflowing, pg_autovacuum has been working flawlessly on my FreeBSD 4.9 + PG 7.4.0 production server. I'm just still playing with tuning pg_autovacuum to keep it from vacuuming my busy tables *too* often. Just a question: will my test program show negative sleep 'diff' on your linux box? I can't imagine that it would give different results than on freebsd. -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Vivek Khera, Ph.D. Khera Communications, Inc. Internet: khera@kciLink.com Rockville, MD +1-240-453-8497 AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/ From pgsql-performance-owner@postgresql.org Mon Dec 8 15:21:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 53D80D1B45A for ; Mon, 8 Dec 2003 19:21:38 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77846-05 for ; Mon, 8 Dec 2003 15:21:10 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 5BCDDD1B448 for ; Mon, 8 Dec 2003 15:21:09 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hB8JJw19022186; Mon, 8 Dec 2003 14:19:59 -0500 (EST) To: Jack Coates Cc: pgsql-performance Subject: Re: tuning questions In-reply-to: <1070905424.16088.63.camel@cletus.lyris.com> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> <1070584326.18838.235.camel@cletus.lyris.com> <200312050926.05155.josh@agliodbs.com> <1070673761.13542.534.camel@cletus.lyris.com> <1070905424.16088.63.camel@cletus.lyris.com> Comments: In-reply-to Jack Coates message dated "Mon, 08 Dec 2003 09:43:45 -0800" Date: Mon, 08 Dec 2003 14:19:58 -0500 Message-ID: <22185.1070911198@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/156 X-Sequence-Number: 5016 Jack Coates writes: > Theories at this point, in no particular order: > a) major differences between my 7.3.4 from source (compiled with no > options) and dev's 7.3.2-1PGDG RPMs. Looking at the spec file doesn't > reveal anything glaring to me, but is there something I'm missing? There are quite a few performance-related patches between 7.3.2 and 7.3.4. Most of them should be in 7.3.4's favor but there are some places where we had to take a performance hit in order to have a suitably low-risk fix for a bug. You haven't told us enough about the problem to know if any of those cases apply, though. AFAIR you have not actually showed either the slow query or EXPLAIN ANALYZE results for it on the two boxes ... regards, tom lane From pgsql-performance-owner@postgresql.org Tue Dec 9 04:32:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CA514D1B454 for ; Tue, 9 Dec 2003 08:32:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 83357-10 for ; Tue, 9 Dec 2003 04:32:26 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id BA86BD1B438 for ; Tue, 9 Dec 2003 04:32:24 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 10DDD1F93; Tue, 9 Dec 2003 03:32:25 -0500 (EST) To: stephen farrell Cc: pgsql-performance@postgresql.org Subject: Re: Problem with insert into select... From: Neil Conway In-Reply-To: <3FBD72D5.1040202@almaden.ibm.com> (stephen farrell's message of "Thu, 20 Nov 2003 18:05:09 -0800") References: <3FBD2D8B.8020509@almaden.ibm.com> <7439.1069369120@sss.pgh.pa.us> <3FBD72D5.1040202@almaden.ibm.com> Date: Tue, 09 Dec 2003 03:32:24 -0500 Message-ID: <87smjuqtlz.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/157 X-Sequence-Number: 5017 stephen farrell writes: > With the indexes created it worked. It took about 4 hours, but it > inserted all of the records. Has this been satisfactorily resolved? If not, can you post an EXPLAIN ANALYZE for the failing query, as Tom asked earlier? -Neil From pgsql-performance-owner@postgresql.org Tue Dec 9 12:58:49 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 86080D1B46E for ; Tue, 9 Dec 2003 16:58:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70646-03 for ; Tue, 9 Dec 2003 12:58:13 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by svr1.postgresql.org (Postfix) with ESMTP id 1F065D1B4AE for ; Tue, 9 Dec 2003 12:58:12 -0400 (AST) Received: from luna.lyris.com (luna.lyris.com [216.91.57.116]) by luna.lyris.com (8.12.8/8.12.8) with ESMTP id hB9GvtIt032473; Tue, 9 Dec 2003 08:57:55 -0800 Received: from luna.lyris.com ([216.91.57.116]) by luna.lyris.com (MailMonitor for SMTP v1.2.2 ) ; Tue, 9 Dec 2003 08:57:55 -0800 (PST) Received: from cletus.lyris.com (cletus.lyris.com [216.91.56.155]) by luna.lyris.net with SMTP (MailShield v2.04 - LINUX Jul 17 2001 16:58:31); Tue, 09 Dec 2003 08:57:55 -0800 Subject: Re: tuning questions From: Jack Coates To: Tom Lane Cc: pgsql-performance In-Reply-To: <22185.1070911198@sss.pgh.pa.us> References: <1070553983.6498.52.camel@cletus.lyris.com> <200312042347.54461.dev@archonet.com> <1070584326.18838.235.camel@cletus.lyris.com> <200312050926.05155.josh@agliodbs.com> <1070673761.13542.534.camel@cletus.lyris.com> <1070905424.16088.63.camel@cletus.lyris.com> <22185.1070911198@sss.pgh.pa.us> Content-Type: text/plain Organization: Lyris Technologies, Inc. Message-Id: <1070989072.16079.19.camel@cletus.lyris.com> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.4-9mdk Date: Tue, 09 Dec 2003 08:57:53 -0800 Content-Transfer-Encoding: 7bit X-SMTP-HELO: cletus.lyris.com X-SMTP-MAIL-FROM: jack@lyris.com X-SMTP-RCPT-TO: tgl@sss.pgh.pa.us,pgsql-performance@postgresql.org X-SMTP-PEER-INFO: cletus.lyris.com [216.91.56.155] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/158 X-Sequence-Number: 5018 On Mon, 2003-12-08 at 11:19, Tom Lane wrote: > Jack Coates writes: > > Theories at this point, in no particular order: > > > a) major differences between my 7.3.4 from source (compiled with no > > options) and dev's 7.3.2-1PGDG RPMs. Looking at the spec file doesn't > > reveal anything glaring to me, but is there something I'm missing? > > There are quite a few performance-related patches between 7.3.2 and > 7.3.4. Most of them should be in 7.3.4's favor but there are some > places where we had to take a performance hit in order to have a > suitably low-risk fix for a bug. You haven't told us enough about > the problem to know if any of those cases apply, though. AFAIR > you have not actually showed either the slow query or EXPLAIN ANALYZE > results for it on the two boxes ... > > regards, tom lane Right, because re-architecture of a cross-platform query makes sense if performance is bad on all systems, but is questionable activity when performance is fine on some systems and lousy on others. Hence my statement that while SQL optimization is certainly something we want to do for across-the-board performance increase, I wanted to focus on other issues for troubleshooting this problem. I will be back to ask about data access models later :-) I ended up going back to a default postgresql.conf and reapplying the various tunings one-by-one. Turns out that while setting fsync = false had little effect on the slow IDE box, it had a drastic effect on this faster SCSI box and performance is quite acceptable now (aside from the expected falloff of about 30% after the first twenty minutes, which I believe comes from growing and shrinking tables without vacuumdb --analyzing). -- Jack Coates, Lyris Technologies Applications Engineer 510-549-4350 x148, jack@lyris.com "Interoperability is the keyword, uniformity is a dead end." --Olivier Fourdan From pgsql-performance-owner@postgresql.org Tue Dec 9 13:08:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8C7D2D1B454 for ; Tue, 9 Dec 2003 17:08:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69676-06 for ; Tue, 9 Dec 2003 13:08:11 -0400 (AST) Received: from bayswater1.ymogen.net (host-154-240-27-217.pobox.net.uk [217.27.240.154]) by svr1.postgresql.org (Postfix) with ESMTP id 7D0D3D1B47D for ; Tue, 9 Dec 2003 13:08:10 -0400 (AST) Received: from solent (82-68-95-1.dsl.in-addr.zen.co.uk [82.68.95.1]) by bayswater1.ymogen.net (Postfix) with ESMTP id 006C896DE6; Tue, 9 Dec 2003 17:07:54 +0000 (GMT) From: "Matt Clark" To: "Jack Coates" , "Tom Lane" Cc: "pgsql-performance" Subject: Re: tuning questions Date: Tue, 9 Dec 2003 17:07:53 -0000 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: <1070989072.16079.19.camel@cletus.lyris.com> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/159 X-Sequence-Number: 5019 > I ended up going back to a default postgresql.conf and reapplying the > various tunings one-by-one. Turns out that while setting fsync = false > had little effect on the slow IDE box, it had a drastic effect on this > faster SCSI box and performance is quite acceptable now (aside from the > expected falloff of about 30% after the first twenty minutes, which I > believe comes from growing and shrinking tables without vacuumdb > --analyzing). Hmm. I wonder if that could be related to the issue where many IDE drives have write-caching enabled. With the write cache enabled fsyncs are nearly immediate, so setting fsync=false makes little difference... From pgsql-performance-owner@postgresql.org Tue Dec 9 13:36:48 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 46DDDD1B48E for ; Tue, 9 Dec 2003 17:36:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75503-09 for ; Tue, 9 Dec 2003 13:36:13 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 46C4ED1B487 for ; Tue, 9 Dec 2003 13:36:12 -0400 (AST) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4049825; Tue, 09 Dec 2003 09:37:02 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Jack Coates , Tom Lane Subject: Re: tuning questions Date: Tue, 9 Dec 2003 09:35:04 -0800 User-Agent: KMail/1.4.3 Cc: pgsql-performance References: <1070553983.6498.52.camel@cletus.lyris.com> <22185.1070911198@sss.pgh.pa.us> <1070989072.16079.19.camel@cletus.lyris.com> In-Reply-To: <1070989072.16079.19.camel@cletus.lyris.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200312090935.04658.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/160 X-Sequence-Number: 5020 Jack, > Right, because re-architecture of a cross-platform query makes sense if > performance is bad on all systems, but is questionable activity when > performance is fine on some systems and lousy on others. Hence my > statement that while SQL optimization is certainly something we want to > do for across-the-board performance increase, I wanted to focus on other > issues for troubleshooting this problem. I will be back to ask about > data access models later :-) Yes, but an EXPLAIN ANALYZE will also help show issues like sorts running out of memory, etc. Really, we don't currently have enough information to do more than speculate; it's like trying to repair a car engine wearing a blindfold. Particularly since it's possible that there are only 1 or 2 "bad queries" which are messing everything else up. For that matter, it would really help to know: -- How many simulatneous connections are running update queries during this process? -- How about some sample VACUUM VERBOSE results for the intra-process vacuums? > I ended up going back to a default postgresql.conf and reapplying the > various tunings one-by-one. Turns out that while setting fsync = false > had little effect on the slow IDE box, it had a drastic effect on this > faster SCSI box and performance is quite acceptable now (aside from the > expected falloff of about 30% after the first twenty minutes, which I > believe comes from growing and shrinking tables without vacuumdb > --analyzing). Well, that brings 2 things immediately to mind: 1) That may improve performance, but it does mean that if your machine loses power you *will* be restoring from backup. It's risky to do. 2) Your IDE system has write-caching enabled. Once again, this is a nice performmance boost, if you don't mind database corruption in a power-out. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Tue Dec 16 17:07:50 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E9048D1B48F for ; Tue, 9 Dec 2003 21:21:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 23084-03 for ; Tue, 9 Dec 2003 17:21:23 -0400 (AST) Received: from www.smart-start-hosting.com (unknown [38.118.153.169]) by svr1.postgresql.org (Postfix) with ESMTP id 8CE43D1B461 for ; Tue, 9 Dec 2003 17:21:20 -0400 (AST) Received: from shadovitz (localhost [127.0.0.1]) by localhost (8.10.2/8.10.2) with ESMTP id hB9LEim13571; Tue, 9 Dec 2003 13:14:44 -0800 From: "David Shadovitz" To: pgsql-performance@postgresql.org Cc: david@shadovitz.com Subject: Why is VACUUM ANALYZE so slow? Date: Tue, 9 Dec 2003 14:14:44 -0700 Message-Id: <20031209211444.M99946@www.shadovitz.com> X-Mailer: Open WebMail 1.81 20021203 X-OriginatingIP: 206.135.121.30 (david@shadovitz) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/232 X-Sequence-Number: 5092 I'm running PG 7.2.2 on RH Linux 8.0. I'd like to know why "VACUUM ANALYZE
" is extemely slow (hours) for certain tables. Here's what the log file shows when I run this command on my "employees" table, which has just 5 columns and 55 records: VACUUM ANALYZE employees DEBUG: --Relation employees-- DEBUG: index employees_pkey: Pages 2; Tuples 55: Deleted 0. CPU 0.00s/0.00u sec elapsed 0.00 sec. DEBUG: index emp_dept_id_idx: Pages 2; Tuples 55: Deleted 0. CPU 0.00s/0.00u sec elapsed 0.00 sec. DEBUG: index emp_emp_num_idx: Pages 2; Tuples 55: Deleted 0. CPU 0.00s/0.00u sec elapsed 0.00 sec. DEBUG: recycled transaction log file 00000000000000CC DEBUG: geqo_main: using edge recombination crossover [ERX] (When I get a chance I will enable timestamping of log file entries.) Thanks for any insight. Please reply to me personally (david@shadovitz.com) as well as to the list. -David From pgsql-performance-owner@postgresql.org Tue Dec 9 19:21:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C724CD1B481 for ; Tue, 9 Dec 2003 23:21:05 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 38678-08 for ; Tue, 9 Dec 2003 19:20:37 -0400 (AST) Received: from jhuml1.jhmi.edu (jhuml1.jhmi.edu [162.129.234.20]) by svr1.postgresql.org (Postfix) with ESMTP id 6A890D1B45A for ; Tue, 9 Dec 2003 19:20:34 -0400 (AST) Received: from jhuml1.jhmi.edu (jhuml1.jhmi.edu [162.129.234.20]) by jhuml1.jhmi.edu (PMDF V6.2-X17 #30839) with SMTP id <0HPN00G0UHZ4HW@jhuml1.jhmi.edu> for pgsql-performance@postgresql.org; Tue, 09 Dec 2003 18:20:20 -0500 (EST) Received: from jhuml1.jhmi.edu ([162.129.234.20]) by jhuml1.jhmi.edu (SAVSMTP 3.1.0.29) with SMTP id M2003120918202007799 for ; Tue, 09 Dec 2003 18:20:20 -0500 Received: from jhmimail.jhmi.edu (jhem2.jhmi.edu [162.129.8.23]) by jhuml1.jhmi.edu (PMDF V6.2-X17 #30839) with ESMTP id <0HPN00HGBI5W7F@jhuml1.jhmi.edu> for pgsql-performance@postgresql.org; Tue, 09 Dec 2003 18:20:20 -0500 (EST) Received: from [162.129.178.60] by jhmimail.jhmi.edu (mshttpd); Tue, 09 Dec 2003 23:24:19 +0000 (GMT) Date: Tue, 09 Dec 2003 23:24:19 +0000 (GMT) From: LIANHE SHAO Subject: Index problem or function problem? To: pgsql-performance Message-id: <4595eb458230.4582304595eb@jhmimail.jhmi.edu> MIME-version: 1.0 X-Mailer: iPlanet Messenger Express 5.2 HotFix 1.17 (built Jun 23 2003) Content-type: text/plain; charset=us-ascii Content-language: en Content-transfer-encoding: 7bit Content-disposition: inline X-Accept-Language: en X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/165 X-Sequence-Number: 5025 Hello, Today I met a very strange query problem, which I spend several hours on it but have no clue. To make thing clear, let me write somewhat in detail. I have two almost exactly same queries, except that one is: lower(annotation) = lower (chip), another is: annotation = chip. While the first one can get result in less 10 seconds, the second one will hange for more that 5 minutes. What a big differents !! I checked the indexes, there are both index for lower() and without lower(). I even droped these indexes and recreated them, then use vacuum analyze, reindex, but thing does not change. the query plan give quite different paths. Could somebody give any clues where difference comes from? Thanks a lot. The first query, which get results in less than 10 seconds PGA=> explain select ei.expid, er.geneid, er.sampleid, ei.annotation, si.samplename, ei.title as exp_name, aaa.chip, aaa.sequence_derived_from as accession_number, aaa.gene_symbol, aaa.title as gene_function, er.exprs, er.mas5exprs from expressiondata er, experimentinfo ei, sampleinfo si, affy_array_annotation aaa where exists (select distinct ei.expid from experimentinfo) and lower(ei.annotation) = lower (aaa.chip) and (lower (aaa.title) like '%mif%' or lower(aaa.sequence_description) like '%mif%') and exists (select distinct ei.annotation from experimentinfo) and ei.expid = er.expid and er.expid = si.expid and er.sampleid = si.sampleid and er.geneid = aaa.probeset_id order by si.sampleid limit 20; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------- ------------------- Limit (cost=24289.05..24289.10 rows=19 width=256) -> Sort (cost=24289.05..24289.10 rows=19 width=256) Sort Key: si.sampleid -> Hash Join (cost=6.11..24288.64 rows=19 width=256) Hash Cond: ("outer".expid = "inner".expid) Join Filter: ("outer".sampleid = "inner".sampleid) -> Nested Loop (cost=0.00..24278.66 rows=27 width=217) Join Filter: ("outer".expid = "inner".expid) -> Nested Loop (cost=0.00..18378.77 rows=45 width=180) -> Seq Scan on experimentinfo ei (cost=0.00..374.50 rows=5 width=99) Filter: ((subplan) AND (subplan)) SubPlan -> Unique (cost=8.67..8.78 rows=2 width=0) -> Sort (cost=8.67..8.72 rows=21 width=0) Sort Key: $0 -> Seq Scan on experimentinfo (cost=0.00..8.21 rows=21 width=0) -> Unique (cost=8.67..8.78 rows=2 width=0) -> Sort (cost=8.67..8.72 rows=21 width=0) Sort Key: $1 -> Seq Scan on experimentinfo (cost=0.00..8.21 rows=21 width=0) -> Index Scan using affy_array_annotation_lower_chip_idx on affy_array_annotation aaa (cost=0.00..3429.2 4 rows=9 width=81) Index Cond: (lower(("outer".annotation)::text) = lower((aaa.chip)::text)) Filter: ((lower(title) ~~ '%mif%'::text) OR (lower(sequence_description) ~~ '%mif%'::text)) -> Index Scan using expressiondata_geneid_idx on expressiondata er (cost=0.00..130.96 rows=34 width=37) Index Cond: (er.geneid = "outer".probeset_id) -> Hash (cost=4.55..4.55 rows=155 width=39) -> Seq Scan on sampleinfo si (cost=0.00..4.55 rows=155 width=39) (27 rows) ===================== The second query, which hangs. PGA=> explain select ei.expid, er.geneid, er.sampleid, ei.annotation, si.samplename, ei.title as exp_name, aaa.chip, aaa.sequence_derived_from as accession_number, aaa.gene_symbol, aaa.title as gene_function, er.exprs, er.mas5exprs from expressiondata er, experimentinfo ei, sampleinfo si, affy_array_annotation aaa where exists (select distinct ei.expid from experimentinfo) and ei.annotation = aaa.chip and (lower (aaa.title) like '%mif%' or lower(aaa.sequence_description) like '%mif%') and exists (select distinct ei.annotation from experimentinfo) and ei.expid = er.expid and er.expid = si.expid and er.sampleid = si.sampleid and er.geneid = aaa.probeset_id order by si.sampleid limit 20; QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------- Limit (cost=157127.91..157128.38 rows=20 width=256) -> Merge Join (cost=157127.91..157137.33 rows=401 width=256) Merge Cond: (("outer".sampleid = "inner".sampleid) AND ("outer".expid = "inner".expid)) -> Sort (cost=157117.73..157119.11 rows=553 width=217) Sort Key: er.sampleid, er.expid -> Merge Join (cost=154417.78..157092.52 rows=553 width=217) Merge Cond: (("outer".annotation = "inner".chip) AND ("outer".geneid = "inner".probeset_id)) -> Sort (cost=96501.38..97830.62 rows=531694 width=136) Sort Key: ei.annotation, er.geneid -> Nested Loop (cost=0.00..20188.81 rows=531694 width=136) -> Seq Scan on experimentinfo ei (cost=0.00..374.50 rows=5 width=99) Filter: ((subplan) AND (subplan)) SubPlan -> Unique (cost=8.67..8.78 rows=2 width=0) -> Sort (cost=8.67..8.72 rows=21 width=0) Sort Key: $0 -> Seq Scan on experimentinfo (cost=0.00..8.21 rows=21 width=0) -> Unique (cost=8.67..8.78 rows=2 width=0) -> Sort (cost=8.67..8.72 rows=21 width=0) Sort Key: $1 -> Seq Scan on experimentinfo (cost=0.00..8.21 rows=21 width=0) -> Index Scan using expressiondata_expid_idx on expressiondata er (cost=0.00..2508.21 rows=101275 width=37) Index Cond: ("outer".expid = er.expid) -> Sort (cost=57916.40..57920.67 rows=1710 width=81) Sort Key: aaa.chip, aaa.probeset_id -> Seq Scan on affy_array_annotation aaa (cost=0.00..57824.60 rows=1710 width=81) Filter: ((lower(title) ~~ '%mif%'::text) OR (lower(sequence_description) ~~ '%mif%'::text)) -> Sort (cost=10.19..10.58 rows=155 width=39) Sort Key: si.sampleid, si.expid -> Seq Scan on sampleinfo si (cost=0.00..4.55 rows=155 width=39) (30 rows) ================= The related tables: Table "public.experimentinfo" Column | Type | Modifiers ---------------+------------------------+----------- expid | integer | name | character varying(128) | lab | character varying(128) | contact | character varying(128) | title | character varying(128) | abstract | text | nsamples | integer | disease_type | character varying(32) | annotation | character varying(32) | Indexes: experimetininfo_annotation_idx btree (annotation), experimetininfo_lower_annotation_idx btree (lower(annotation)), expinfo btree (expid) Table "public.affy_array_annotation" Column | Type | Modifiers -----------------------------------+------------------------+----------- chip | character varying(32) | not null organism | character varying(24) | annotation_date | character varying(24) | sequence_type | character varying(24) | sequence_source | character varying(32) | sequence_derived_from | character varying(32) | sequence_description | text | sequence_id | text | transcript_id | character varying(32) | group_id | character varying(64) | title | text | gene_symbol | character varying(64) | Indexes: affy_array_annotation_chip_idx btree (chip), affy_array_annotation_idx_gene_symbol btree (gene_symbol), affy_array_annotation_idx_locuslink btree (locuslink), affy_array_annotation_idx_omim btree (omim), affy_array_annotation_idx_pfam btree (pfam), affy_array_annotation_idx_sequence_derived_from btree (sequence_derived_from), affy_array_annotation_idx_sequence_description btree (sequence_description), affy_array_annotation_idx_title btree (title), affy_array_annotation_lower_chip_idx btree (lower(chip)), affy_array_annotation_lower_gene_symbol_idx btree (lower(gene_symbol)), affy_array_annotation_lower_probeset_id_idx btree (lower(probeset_id)), affy_array_annotation_lower_sequence_description_idx btree (lower(sequence_description)), affy_array_annotation_lower_title_idx btree (lower(title)), affy_array_annotation_pkey btree (probeset_id, chip), affy_array_annotation_probeset_id_idx btree (probeset_id), Regards, William From pgsql-performance-owner@postgresql.org Wed Dec 10 01:56:48 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 471FED1B494 for ; Wed, 10 Dec 2003 05:56:47 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 87267-08 for ; Wed, 10 Dec 2003 01:56:16 -0400 (AST) Received: from linda-1.paradise.net.nz (bm-1a.paradise.net.nz [202.0.58.20]) by svr1.postgresql.org (Postfix) with ESMTP id C2A78D1B4AE for ; Wed, 10 Dec 2003 01:56:14 -0400 (AST) Received: from smtp-3.paradise.net.nz (smtp-3a.paradise.net.nz [202.0.32.196]) by linda-1.paradise.net.nz (Paradise.net.nz) with ESMTP id <0HPO00BD80HQSW@linda-1.paradise.net.nz> for pgsql-performance@postgresql.org; Wed, 10 Dec 2003 18:56:14 +1300 (NZDT) Received: from paradise.net.nz (203-79-100-38.adsl.paradise.net.nz [203.79.100.38]) by smtp-3.paradise.net.nz (Postfix) with ESMTP id 05519ADF50 for ; Wed, 10 Dec 2003 18:56:14 +1300 (NZDT) Date: Wed, 10 Dec 2003 18:56:38 +1300 From: Mark Kirkwood Subject: Solaris Performance (Again) To: pgsql-performance@postgresql.org Message-id: <3FD6B596.8090803@paradise.net.nz> MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=ISO-8859-1 Content-transfer-encoding: 7bit X-Accept-Language: en-us, en User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031008 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/166 X-Sequence-Number: 5026 This is a well-worn thread title - apologies, but these results seemed interesting, and hopefully useful in the quest to get better performance on Solaris: I was curious to see if the rather uninspiring pgbench performance obtained from a Sun 280R (see General: ATA Disks and RAID controllers for database servers) could be improved if more time was spent tuning. With the help of a fellow workmate who is a bit of a Solaris guy, we decided to have a go. The major performance killer appeared to be mounting the filesystem with the logging option. The next most significant seemed to be the choice of sync_method for Pg - the default (open_datasync), which we initially thought should be the best - appears noticeably slower than fdatasync. We also tried changing some of the tuneable filesystem options using tunefs - without any measurable effect. Are Pg/Solaris folks running with logging on and sync_method default out there ? - or have most of you been through this already ? Pgbench Results (no. clients and transactions/s ) : Setup 1: filesystem mounted with logging No. tps ----------- 1 17 2 17 4 22 8 22 16 28 32 32 64 37 Setup 2: filesystem mounted without logging No. tps ----------- 1 48 2 55 4 57 8 62 16 65 32 82 64 95 Setup 3 : filesystem mounted without logging, Pg sync_method = fdatasync No. tps ----------- 1 89 2 94 4 95 8 93 16 99 32 115 64 122 Note : The Pgbench runs were conducted using -s 10 and -t 1000 -c 1->64, 2 - 3 runs of each setup were performed (averaged figures shown). Mark From pgsql-performance-owner@postgresql.org Wed Dec 10 09:54:24 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 17654D1B4C4 for ; Wed, 10 Dec 2003 13:54:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50470-02 for ; Wed, 10 Dec 2003 09:53:53 -0400 (AST) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 73D61D1B4CC for ; Wed, 10 Dec 2003 09:53:49 -0400 (AST) Received: (qmail 42480 invoked from network); 10 Dec 2003 13:53:47 -0000 Received: from localhost (HELO squeegit) (threshar@127.0.0.1) by localhost with SMTP; 10 Dec 2003 13:53:47 -0000 Date: Wed, 10 Dec 2003 08:53:23 -0500 From: Jeff To: Mark Kirkwood Cc: pgsql-performance@postgresql.org Subject: Re: Solaris Performance (Again) Message-Id: <20031210085323.749a549f.threshar@torgo.978.org> In-Reply-To: <3FD6B596.8090803@paradise.net.nz> References: <3FD6B596.8090803@paradise.net.nz> X-Mailer: Sylpheed version 0.9.7 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/167 X-Sequence-Number: 5027 On Wed, 10 Dec 2003 18:56:38 +1300 Mark Kirkwood wrote: > The major performance killer appeared to be mounting the filesystem > with the logging option. The next most significant seemed to be the > choice of sync_method for Pg - the default (open_datasync), which we > initially thought should be the best - appears noticeably slower than > fdatasync. > Some interesting stuff, I'll have to play with it. Currently I'm pleased with my solaris performance. What version of PG? If it is before 7.4 PG compiles with _NO_ optimization by default and was a huge part of the slowness of PG on solaris. -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Wed Dec 10 13:20:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7D016D1B4AC for ; Wed, 10 Dec 2003 17:20:05 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 84869-05 for ; Wed, 10 Dec 2003 13:19:34 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 22996D1B456 for ; Wed, 10 Dec 2003 13:19:33 -0400 (AST) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 4054923; Wed, 10 Dec 2003 09:20:09 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Hartmut Raschick , pgsql-performance@postgresql.org Subject: Re: TRUNCATE veeeery slow compared to DELETE in 7.4 Date: Wed, 10 Dec 2003 09:18:02 -0800 User-Agent: KMail/1.4.3 References: <3FD484B6.EF39A92F@ke-elektronik.de> In-Reply-To: <3FD484B6.EF39A92F@ke-elektronik.de> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200312100918.02501.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/168 X-Sequence-Number: 5028 Hartmut, > DELETE: > 1) 0.03u 0.04s 0:02.46 2.8% (already empty) > 2) 0.05u 0.06s 0:01.19 9.2% (already empty) > > TRUNCATE: > 1) 0.10u 0.06s 6:58.66 0.0% (already empty, compile runnig simult.) > 2) 0.10u 0.02s 2:51.71 0.0% (already empty) How about some times for a full table? Incidentally, I believe that TRUNCATE has always been slightly slower than DROP TABLE. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Wed Dec 10 15:16:25 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C881DD1B469 for ; Wed, 10 Dec 2003 19:16:24 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 00769-09 for ; Wed, 10 Dec 2003 15:15:54 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id D5CBCD1B4AD for ; Wed, 10 Dec 2003 15:15:52 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 9D3471E15; Wed, 10 Dec 2003 14:15:36 -0500 (EST) To: Mark Kirkwood Cc: pgsql-performance@postgresql.org Subject: Re: Solaris Performance (Again) From: Neil Conway In-Reply-To: <3FD6B596.8090803@paradise.net.nz> (Mark Kirkwood's message of "Wed, 10 Dec 2003 18:56:38 +1300") References: <3FD6B596.8090803@paradise.net.nz> Date: Wed, 10 Dec 2003 14:15:35 -0500 Message-ID: <87smjs4h7s.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/169 X-Sequence-Number: 5029 Mark Kirkwood writes: > Note : The Pgbench runs were conducted using -s 10 and -t 1000 -c > 1->64, 2 - 3 runs of each setup were performed (averaged figures > shown). FYI, the pgbench docs state: NOTE: scaling factor should be at least as large as the largest number of clients you intend to test; else you'll mostly be measuring update contention. -Neil From pgsql-performance-owner@postgresql.org Wed Dec 10 15:30:43 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A7277D1B4AC for ; Wed, 10 Dec 2003 19:30:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08440-01 for ; Wed, 10 Dec 2003 15:30:12 -0400 (AST) Received: from honorio.sinectis.com.ar (honorio.sinectis.com.ar [216.244.192.201]) by svr1.postgresql.org (Postfix) with ESMTP id 8780DD1B45E for ; Wed, 10 Dec 2003 15:30:09 -0400 (AST) Received: by honorio.sinectis.com.ar (Postfix, from userid 99) id BF23C6C8A6; Wed, 10 Dec 2003 16:29:58 -0300 (GMT+3) Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 X-Mailer: Sinectis Webmail 5.6.16-1.4.4 From: nbarraza@uolsinectis.com.ar To: pgsql-performance@postgresql.org Cc: nbarraza@boldt.com.ar Reply-To: nbarraza@uolsinectis.com.ar Message-Id: <20031210192958.BF23C6C8A6@honorio.sinectis.com.ar> Date: Wed, 10 Dec 2003 16:29:58 -0300 (GMT+3) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/170 X-Sequence-Number: 5030 I have some problems on performance using postgresql v. 7.3.2 running on Linux RedHat 9. An update involving several rows (about 500000) on a table having 2800000 tuples takes in the order of 6 minutes. It is more than it takes on other plataforms (SqlServer, FOX). I think that there�s something wrong on my configuration. I�ve already adjusted some parameters as I could understand memory and disk usage. Next, I send a description of parameters changed in postgresql.conf, a scheme of the table, and an EXPLAIN ANALYZE of the command. The hardware configuration is a Pentium III 1 Ghz, 512 MB of memory, and an SCSI drive of 20 GB. Following goes the description: -- Values changed in postgresql.conf tcpip_socket = true max_connections = 64 shared_buffers = 4096 wal_buffers = 100 vacuum_mem = 16384 vacuum_mem = 16384 sort_mem = 32168 checkpoint_segments = 8 effective_cache_size = 10000 -- -- PostgreSQL database dump -- \connect - nestor SET search_path = public, pg_catalog; -- -- TOC entry 2 (OID 22661417) -- Name: jugadas; Type: TABLE; Schema: public; Owner: nestor -- CREATE TABLE jugadas ( fecha_ju character(8), hora_ju character(4), juego character(2), juego_vta character(2), sorteo_p character(5), sorteo_v character(5), nro_servidor character(1), ticket character(9), terminal character(4), sistema character(1), agente character(5), subagente character(3), operador character(2), importe character(7), anulada character(1), icode character(15), codseg character(15), tipo_moneda character(1), apuesta character(100), extraido character(1) ); -- -- TOC entry 4 (OID 25553754) -- Name: key_jug_1; Type: INDEX; Schema: public; Owner: nestor -- CREATE UNIQUE INDEX key_jug_1 ON jugadas USING btree (juego, juego_vta, sorteo_p, nro_servidor, ticket); boss=# explain analyze update jugadas set extraido = 'S' where juego = '03' and juego_vta = '03' and sorteo_p = '89353' and extraido = 'N'; QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------- Seq Scan on jugadas (cost=0.00..174624.96 rows=70061 width=272) (actual time=21223.88..51858.07 rows=517829 loops=1) Filter: ((juego = '03'::bpchar) AND (juego_vta = '03'::bpchar) AND (sorteo_p = '89353'::bpchar) AND (extraido = 'N'::bpchar)) Total runtime: 291167.36 msec (3 rows) boss=# show enable_seqscan; enable_seqscan ---------------- on (1 row) ************* FORCING INDEX SCAN *********************************** boss=# set enable_seqscan = false; SET boss=# explain analyze update jugadas set extraido = 'N' where juego = '03' and juego_vta = '03' and sorteo_p = '89353' and extraido = 'S'; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------- Index Scan using key_jug_1 on jugadas (cost=0.00..597959.76 rows=98085 width=272) (actual time=9.93..39947.93 rows=517829 loops=1) Index Cond: ((juego = '03'::bpchar) AND (juego_vta = '03'::bpchar) AND (sorteo_p = '89353'::bpchar)) Filter: (extraido = 'S'::bpchar) Total runtime: 335280.56 msec (4 rows) boss=# Thank you in advance for any help. Nestor From pgsql-performance-owner@postgresql.org Wed Dec 10 15:55:02 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 104A5D1B449 for ; Wed, 10 Dec 2003 19:55:01 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11628-03 for ; Wed, 10 Dec 2003 15:54:30 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 1B03DD1B4D4 for ; Wed, 10 Dec 2003 15:54:29 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBAJsE19008001; Wed, 10 Dec 2003 14:54:14 -0500 (EST) To: Josh Berkus Cc: Hartmut Raschick , pgsql-performance@postgresql.org Subject: Re: TRUNCATE veeeery slow compared to DELETE in 7.4 In-reply-to: <200312100918.02501.josh@agliodbs.com> References: <3FD484B6.EF39A92F@ke-elektronik.de> <200312100918.02501.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Wed, 10 Dec 2003 09:18:02 -0800" Date: Wed, 10 Dec 2003 14:54:14 -0500 Message-ID: <8000.1071086054@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/171 X-Sequence-Number: 5031 Josh Berkus writes: > Incidentally, I believe that TRUNCATE has always been slightly slower than > DROP TABLE. Well, it would be: it has to delete the original files and then create new ones. I imagine the time to create new, empty indexes is the bulk of the time Hartmut is measuring. (Remember that an "empty" index has at least one page in it, the metadata page, for all of our index types, so there is some actual I/O involved to do this.) It does not bother me that TRUNCATE takes nonzero time; it's intended to be used in situations where DELETE would take huge amounts of time (especially after you factor in the subsequent VACUUM activity). The fact that DELETE takes near-zero time on a zero-length table is not very relevant. regards, tom lane From pgsql-performance-owner@postgresql.org Wed Dec 10 16:35:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 19D2BD1B498 for ; Wed, 10 Dec 2003 20:35:55 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 16293-07 for ; Wed, 10 Dec 2003 16:35:25 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 5D136D1B468 for ; Wed, 10 Dec 2003 16:35:23 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 03E9835305; Wed, 10 Dec 2003 12:35:06 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id F3F07352A2; Wed, 10 Dec 2003 12:35:05 -0800 (PST) Date: Wed, 10 Dec 2003 12:35:05 -0800 (PST) From: Stephan Szabo To: nbarraza@uolsinectis.com.ar Cc: pgsql-performance@postgresql.org, nbarraza@boldt.com.ar Subject: Re: In-Reply-To: <20031210192958.BF23C6C8A6@honorio.sinectis.com.ar> Message-ID: <20031210123323.S707@megazone.bigpanda.com> References: <20031210192958.BF23C6C8A6@honorio.sinectis.com.ar> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=X-UNKNOWN Content-Transfer-Encoding: QUOTED-PRINTABLE X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/172 X-Sequence-Number: 5032 On Wed, 10 Dec 2003 nbarraza@uolsinectis.com.ar wrote: > I have some problems on performance using postgresql v. 7.3.2 running on > Linux RedHat 9. An update involving several rows (about 500000) on a > table having 2800000 tuples takes in the order of 6 minutes. It is more > than it takes on other plataforms (SqlServer, FOX). I think that there=B4s > something wrong on my configuration. I=B4ve already adjusted some > parameters as I could understand memory and disk usage. Next, I send a > description of parameters changed in postgresql.conf, a scheme of the > table, and an EXPLAIN ANALYZE of the command. The hardware configuration > is a Pentium III 1 Ghz, 512 MB of memory, and an SCSI drive of 20 GB. > Following goes the description: > -- Values changed in postgresql.conf > CREATE TABLE jugadas ( > fecha_ju character(8), > hora_ju character(4), > juego character(2), > juego_vta character(2), > sorteo_p character(5), > sorteo_v character(5), > nro_servidor character(1), > ticket character(9), > terminal character(4), > sistema character(1), > agente character(5), > subagente character(3), > operador character(2), > importe character(7), > anulada character(1), > icode character(15), > codseg character(15), > tipo_moneda character(1), > apuesta character(100), > extraido character(1) > ); Are there any tables that reference this one or other triggers? If so, what do the tables/contraints/triggers involve look like? I'm guessing there might be given the difference in the actual time numbers to the total runtime on the explain analyze. From pgsql-performance-owner@postgresql.org Thu Dec 11 00:22:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7D7DBD1B430 for ; Thu, 11 Dec 2003 04:22:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 72947-04 for ; Thu, 11 Dec 2003 00:22:13 -0400 (AST) Received: from homer.lsd.di.uminho.pt (gsd.di.uminho.pt [193.136.20.132]) by svr1.postgresql.org (Postfix) with SMTP id C2BE7D1B462 for ; Thu, 11 Dec 2003 00:22:06 -0400 (AST) Received: (qmail 31367 invoked from network); 11 Dec 2003 04:18:20 -0000 Received: from alfranio.lsd.di.uminho.pt (HELO lsd.di.uminho.pt) (192.168.2.144) by mailer.lsd.di.uminho.pt with SMTP; 11 Dec 2003 04:18:20 -0000 Message-ID: <3FD7EEE8.4080904@lsd.di.uminho.pt> Date: Thu, 11 Dec 2003 04:13:28 +0000 From: Alfranio Correia Junior User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030225 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Performance problems with a higher number of clients Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/173 X-Sequence-Number: 5033 Hello, I am facing a problem trying to put 500 concurrent users accessing a postgresql instance. Basically, the machine begins to do a lot i/o... swap area increases more and more... The vmstat began with 9200 (swpd) and after 20 minutes it was like that: VMSTAT: procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 2 29 1 106716 9576 7000 409876 32 154 5888 1262 616 1575 8 12 80 0 29 1 107808 9520 6896 409904 60 220 5344 1642 662 1510 9 15 76 0 89 1 108192 9528 6832 410184 172 138 6810 1750 693 2466 11 16 73 0 27 1 108192 9900 6824 409852 14 112 4488 1294 495 862 2 9 88 8 55 1 108452 9552 6800 410284 26 12 6266 1082 651 2284 8 11 81 5 78 2 109220 8688 6760 410816 148 534 6318 1632 683 1230 6 13 81 The application that I am trying to running mimmics the tpc-c benchmark... Actually, I am simulating the tpc-c workload without considering screens and other details. The only interesting is on the database workload proposed by the benchmark and its distributions. The machine is a dual-processor pentium III, with 1GB, external storage device. It runs Linux version 2.4.21-dt1 (root@dupond) (gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-113)) #7 SMP Mon Apr 21 19:43:17 GMT 2003, Postgresql 7.5devel. Postgresql configuration: effective_cache_size = 35000 shared_buffers = 5000 random_page_cost = 2 cpu_index_tuple_cost = 0.0005 sort_mem = 10240 I would like to know if this behaivor is normal considering the number of clients, the workload and the database size (7.8 GB) ? Or if there is something that I can change to get better results. Best regards, Alfranio Junior. From pgsql-performance-owner@postgresql.org Thu Dec 11 02:04:30 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8AEAAD1B4BD for ; Thu, 11 Dec 2003 06:04:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 76711-09 for ; Thu, 11 Dec 2003 02:04:02 -0400 (AST) Received: from linda-2.paradise.net.nz (bm-2a.paradise.net.nz [202.0.58.21]) by svr1.postgresql.org (Postfix) with ESMTP id 5D4BFD1B481 for ; Thu, 11 Dec 2003 02:04:01 -0400 (AST) Received: from smtp-3.paradise.net.nz (smtp-3a.paradise.net.nz [202.0.32.196]) by linda-2.paradise.net.nz (Paradise.net.nz) with ESMTP id <0HPP005JQVID8M@linda-2.paradise.net.nz> for pgsql-performance@postgresql.org; Thu, 11 Dec 2003 19:03:50 +1300 (NZDT) Received: from paradise.net.nz (218-101-14-73.paradise.net.nz [218.101.14.73]) by smtp-3.paradise.net.nz (Postfix) with ESMTP id B082FADF83; Thu, 11 Dec 2003 19:03:49 +1300 (NZDT) Date: Thu, 11 Dec 2003 19:04:15 +1300 From: Mark Kirkwood Subject: Re: Solaris Performance (Again) In-reply-to: <20031210085323.749a549f.threshar@torgo.978.org> To: Jeff Cc: pgsql-performance@postgresql.org Message-id: <3FD808DF.9080504@paradise.net.nz> MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=us-ascii Content-transfer-encoding: 7bit X-Accept-Language: en-us, en User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031008 References: <3FD6B596.8090803@paradise.net.nz> <20031210085323.749a549f.threshar@torgo.978.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/174 X-Sequence-Number: 5034 Good point - It is Pg 7.4beta1 , compiled with CFLAGS += -O2 -funroll-loops -fexpensive-optimizations Jeff wrote: > >What version of PG? > >If it is before 7.4 PG compiles with _NO_ optimization by default and >was a huge part of the slowness of PG on solaris. > > > > From pgsql-performance-owner@postgresql.org Thu Dec 11 02:09:53 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BC5E6D1B432 for ; Thu, 11 Dec 2003 06:09:51 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 83134-04 for ; Thu, 11 Dec 2003 02:09:24 -0400 (AST) Received: from linda-1.paradise.net.nz (bm-1a.paradise.net.nz [202.0.58.20]) by svr1.postgresql.org (Postfix) with ESMTP id C4B20D1B46C for ; Thu, 11 Dec 2003 02:09:22 -0400 (AST) Received: from smtp-3.paradise.net.nz (smtp-3a.paradise.net.nz [202.0.32.196]) by linda-1.paradise.net.nz (Paradise.net.nz) with ESMTP id <0HPP00KVTVRLD1@linda-1.paradise.net.nz> for pgsql-performance@postgresql.org; Thu, 11 Dec 2003 19:09:22 +1300 (NZDT) Received: from paradise.net.nz (218-101-14-73.paradise.net.nz [218.101.14.73]) by smtp-3.paradise.net.nz (Postfix) with ESMTP id 959BDADFE0; Thu, 11 Dec 2003 19:09:21 +1300 (NZDT) Date: Thu, 11 Dec 2003 19:09:47 +1300 From: Mark Kirkwood Subject: Re: Solaris Performance (Again) In-reply-to: <87smjs4h7s.fsf@mailbox.samurai.com> To: Neil Conway Cc: pgsql-performance@postgresql.org Message-id: <3FD80A2B.2010601@paradise.net.nz> MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=us-ascii Content-transfer-encoding: 7bit X-Accept-Language: en-us, en User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031008 References: <3FD6B596.8090803@paradise.net.nz> <87smjs4h7s.fsf@mailbox.samurai.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/175 X-Sequence-Number: 5035 yes - originally I was going to stop at 8 clients, but once the bit was between the teeth....If I get another box to myself I will try -s 50 or 100 and see what that shows up. cheers Mark Neil Conway wrote: > FYI, the pgbench docs state: > > NOTE: scaling factor should be at least as large as the largest > number of clients you intend to test; else you'll mostly be > measuring update contention. > >-Neil > > > From pgsql-performance-owner@postgresql.org Thu Dec 11 02:37:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2F44AD1B481 for ; Thu, 11 Dec 2003 06:37:12 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85303-07 for ; Thu, 11 Dec 2003 02:36:42 -0400 (AST) Received: from smtp-send.myrealbox.com (smtp-send.myrealbox.com [192.108.102.143]) by svr1.postgresql.org (Postfix) with ESMTP id 6CE0CD1B47A for ; Thu, 11 Dec 2003 02:36:41 -0400 (AST) Received: from myrealbox.com shridhar_daithankar@smtp-send.myrealbox.com [202.54.11.72] by smtp-send.myrealbox.com with NetMail SMTP Agent $Revision: 3.46 $ on Novell NetWare via secured & encrypted transport (TLS); Wed, 10 Dec 2003 23:36:39 -0700 Message-ID: <3FD81067.6010305@myrealbox.com> Date: Thu, 11 Dec 2003 12:06:23 +0530 From: Shridhar Daithankar User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Alfranio Correia Junior Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems with a higher number of clients References: <3FD7EEE8.4080904@lsd.di.uminho.pt> In-Reply-To: <3FD7EEE8.4080904@lsd.di.uminho.pt> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/176 X-Sequence-Number: 5036 Alfranio Correia Junior wrote: > Postgresql configuration: > > effective_cache_size = 35000 > shared_buffers = 5000 > random_page_cost = 2 > cpu_index_tuple_cost = 0.0005 > sort_mem = 10240 Lower sort mem to say 2000-3000, up shared buffers to 10K and up effective cache size to around 65K. That should make it behave bit better. I guess tuning sort mem alone would give you performance you are expecting.. Tune them one by one. HTH Shridhar From pgsql-performance-owner@postgresql.org Thu Dec 11 09:29:20 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 01509D1B462 for ; Thu, 11 Dec 2003 13:29:16 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34383-04 for ; Thu, 11 Dec 2003 09:28:49 -0400 (AST) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 601D4D1B49E for ; Thu, 11 Dec 2003 09:28:45 -0400 (AST) Received: (qmail 2973 invoked from network); 11 Dec 2003 13:28:51 -0000 Received: from localhost (HELO squeegit) (threshar@127.0.0.1) by localhost with SMTP; 11 Dec 2003 13:28:51 -0000 Date: Thu, 11 Dec 2003 08:28:26 -0500 From: Jeff To: Alfranio Correia Junior Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems with a higher number of clients Message-Id: <20031211082826.7173d5e0.threshar@torgo.978.org> In-Reply-To: <3FD7EEE8.4080904@lsd.di.uminho.pt> References: <3FD7EEE8.4080904@lsd.di.uminho.pt> X-Mailer: Sylpheed version 0.9.7 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/177 X-Sequence-Number: 5037 On Thu, 11 Dec 2003 04:13:28 +0000 Alfranio Correia Junior wrote: > r b w swpd free buff cache si so bi bo in cs > us sy id > 2 29 1 106716 9576 7000 409876 32 154 5888 1262 616 1575 > 8 12 80 On linux I've found as soon as it has to swap its oh-so-wonderful VM brings the machine to a screeching halt. > sort_mem = 10240 > Here's a big problem This gives _EACH SORT_ 10MB (No more, no less) to play with. 10MB * 500 connections == 5000MB in one case.. Some queries may have more sort steps. It is possible 1 connection could be using 30-40MB of sort_mem. You'll need to bring that value down to prevent swapping. If you have a few "common" queries that are run a lot check out hte explain analyze. You can see about how much sort_mem you'll need. Look in the sort step. it should tell you the width and the # of rows. Multiply those. That is sort of how much memory you'll need (I'd round it up a bit) If under normal workload your DB is swapping you have problems. You'll need to either tune your config or get bigger hardware. You may want to also consider an OS that deals with that situation a bit better. good luck. -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Thu Dec 11 10:59:08 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4316AD1BB91 for ; Thu, 11 Dec 2003 14:55:31 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 51326-01 for ; Thu, 11 Dec 2003 10:55:03 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 6A8E8D1C4E2 for ; Thu, 11 Dec 2003 10:54:43 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBBEsH19013179; Thu, 11 Dec 2003 09:54:18 -0500 (EST) To: Alfranio Correia Junior Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems with a higher number of clients In-reply-to: <3FD7EEE8.4080904@lsd.di.uminho.pt> References: <3FD7EEE8.4080904@lsd.di.uminho.pt> Comments: In-reply-to Alfranio Correia Junior message dated "Thu, 11 Dec 2003 04:13:28 +0000" Date: Thu, 11 Dec 2003 09:54:17 -0500 Message-ID: <13178.1071154457@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/178 X-Sequence-Number: 5038 Alfranio Correia Junior writes: > I am facing a problem trying to put 500 concurrent users accessing > a postgresql instance. I think you're going to need to buy more RAM. 1Gb of RAM means there is a maximum of 2Mb available per Postgres process before you start to go into swap hell --- in practice a lot less, since you have to allow for other things like the kernel and other applications. AFAIR TPC-C doesn't involve any complex queries, so it's possible you could run it with only 1Mb of workspace per process, but not when you've configured > sort_mem = 10240 That's ten times more than your configuration can possibly support. (I don't recall whether TPC-C uses any queries that would sort, so it's possible this setting isn't affecting you; but if you are doing any sorts then it's killing you.) Bottom line is you probably need more RAM. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Dec 11 12:01:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 12CEFD1DF71 for ; Thu, 11 Dec 2003 16:00:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 59848-08 for ; Thu, 11 Dec 2003 12:00:21 -0400 (AST) Received: from hotmail.com (bay7-f88.bay7.hotmail.com [64.4.11.88]) by svr1.postgresql.org (Postfix) with ESMTP id B625AD1DF95 for ; Thu, 11 Dec 2003 12:00:20 -0400 (AST) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Thu, 11 Dec 2003 08:00:19 -0800 Received: from 200.24.104.55 by by7fd.bay7.hotmail.msn.com with HTTP; Thu, 11 Dec 2003 16:00:19 GMT X-Originating-IP: [200.24.104.55] X-Originating-Email: [mileruiz@hotmail.com] X-Sender: mileruiz@hotmail.com From: "sandra ruiz" To: pgsql-performance@postgresql.org Subject: hints in Postgres? Date: Thu, 11 Dec 2003 11:00:19 -0500 Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; format=flowed Message-ID: X-OriginalArrivalTime: 11 Dec 2003 16:00:19.0890 (UTC) FILETIME=[E066B920:01C3BFFF] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/179 X-Sequence-Number: 5039 Hi list, I need to know if there is anything like hints of Oracle in Postgres..otherwise..I wish to find a way to force a query plan to use the indexes or tell the optimizer things like "optimize based in statistics", "I want to define the order of the a join" , "optimize based on a execution plan that I consider the best" ... thanks. _________________________________________________________________ Las mejores tiendas, los precios mas bajos, entregas en todo el mundo, YupiMSN Compras: http://latam.msn.com/compras/ From pgsql-performance-owner@postgresql.org Thu Dec 11 12:22:53 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 070CED1B4A1 for ; Thu, 11 Dec 2003 16:22:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66373-09 for ; Thu, 11 Dec 2003 12:22:21 -0400 (AST) Received: from kix.fsv.cvut.cz (Kix.FSV.CVUT.CZ [147.32.129.84]) by svr1.postgresql.org (Postfix) with ESMTP id 656D1D1DF30 for ; Thu, 11 Dec 2003 12:22:20 -0400 (AST) Received: from localhost (stehule@localhost) by kix.fsv.cvut.cz (8.11.6/8.11.6) with ESMTP id hBBGMDY09791; Thu, 11 Dec 2003 17:22:13 +0100 Date: Thu, 11 Dec 2003 17:22:13 +0100 (CET) From: Pavel Stehule To: sandra ruiz Cc: pgsql-performance@postgresql.org Subject: Re: hints in Postgres? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/180 X-Sequence-Number: 5040 hello maybe http://www.gtsm.com/oscon2003/toc.html http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html bye Pavel On Thu, 11 Dec 2003, sandra ruiz wrote: > Hi list, > > I need to know if there is anything like hints of Oracle in > Postgres..otherwise..I wish to find a way to force a query plan to use the > indexes or tell the optimizer things like "optimize based in statistics", "I > want to define the order of the a join" , "optimize based on a execution > plan that I consider the best" ... > > thanks. > > _________________________________________________________________ > Las mejores tiendas, los precios mas bajos, entregas en todo el mundo, > YupiMSN Compras: http://latam.msn.com/compras/ > > > ---------------------------(end of broadcast)--------------------------- > TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > From pgsql-performance-owner@postgresql.org Thu Dec 11 13:00:27 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C792DD1B438 for ; Thu, 11 Dec 2003 17:00:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 72502-05 for ; Thu, 11 Dec 2003 12:59:52 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id D6147D1B46E for ; Thu, 11 Dec 2003 12:59:51 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBBGxobK005184 for ; Thu, 11 Dec 2003 16:59:51 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBBGaZG5002490 for pgsql-performance@postgresql.org; Thu, 11 Dec 2003 16:36:35 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: hints in Postgres? Date: Thu, 11 Dec 2003 11:31:45 -0500 Organization: cbbrowne Computing Inc Lines: 24 Message-ID: References: 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/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:y+0IPFvfAM3U+PlNhp3qRQc1Qx0= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/182 X-Sequence-Number: 5042 Quoth mileruiz@hotmail.com ("sandra ruiz"): > I need to know if there is anything like hints of Oracle in > Postgres..otherwise..I wish to find a way to force a query plan to use > the indexes or tell the optimizer things like "optimize based in > statistics", "I want to define the order of the a join" , "optimize > based on a execution plan that I consider the best" ... It is commonly considered a MISFEATURE of Oracle that it forces you to tweak all of those sorts of 'knobs.' The approach taken with PostgreSQL is to use problems discovered to try to improve the quality of the query optimizer. It is usually clever enough to do a good job, and if it can be improved to automatically notice that "better" plan, then that is a better thing than imposing the burden of tuning each query on you. Tom Lane is "Doctor Optimization," and if you look at past discussion threads of this sort, you'll see that he tends to rather strongly oppose the introduction of "hints." -- select 'aa454' || '@' || 'freenet.carleton.ca'; http://www3.sympatico.ca/cbbrowne/linux.html As of next Monday, COMSAT will be flushed in favor of a string and two tin cans. Please update your software. From pgsql-performance-owner@postgresql.org Tue Dec 16 17:08:02 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F3AF2D1B8B6 for ; Thu, 11 Dec 2003 16:51:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 72502-03 for ; Thu, 11 Dec 2003 12:51:14 -0400 (AST) Received: from dycon.com (titus.dycon.com [12.13.144.17]) by svr1.postgresql.org (Postfix) with SMTP id 5DA30D1D1F2 for ; Thu, 11 Dec 2003 12:51:13 -0400 (AST) Received: (qmail 25373 invoked by uid 1015); 11 Dec 2003 16:44:36 -0000 Received: from unknown (HELO ulanji.com) (12.13.144.25) by 0 with SMTP; 11 Dec 2003 16:44:36 -0000 Message-ID: <3FD89D72.8080508@ulanji.com> Date: Thu, 11 Dec 2003 11:38:10 -0500 From: "Sean P. Thomas" User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030312 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Optimizing FK & PK performance... Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/234 X-Sequence-Number: 5095 I am working on migrating to postgres and had some questions regarding optimization that I could not find references in the documentation: 1. Is there any performance difference for declaring a primary or foreign key a column or table contraint? From the documentation, which way is faster and/or scales better: CREATE TABLE distributors ( did integer, name varchar(40), PRIMARY KEY(did) ); CREATE TABLE distributors ( did integer PRIMARY KEY, name varchar(40) ); 2. Is DEFERRABLE and INITIALLY IMMEDIATE or INITIALLY DEFERRABLE perferred for performance? We generally have very small transactions (web app) but we utilize a model of: view (limit scope for security) -> rules -> before triggers (validate permissions and to set proper permissions) -> tables. I know there were some issues with deferring that was fixed but does it benefit performance or cause any reliability issues? Thank you for your assistance and let me know if I can offer additional information. --spt From pgsql-performance-owner@postgresql.org Thu Dec 11 12:46:02 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C7FA8D1B4B2 for ; Thu, 11 Dec 2003 16:46:00 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68329-10 for ; Thu, 11 Dec 2003 12:45:30 -0400 (AST) Received: from tino.sinectis.com.ar (tino.sinectis.com.ar [216.244.192.232]) by svr1.postgresql.org (Postfix) with ESMTP id A6345D1B480 for ; Thu, 11 Dec 2003 12:45:29 -0400 (AST) Received: by tino.sinectis.com.ar (Postfix, from userid 99) id CFE346C3BE; Thu, 11 Dec 2003 13:45:17 -0300 (GMT+3) Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 X-Mailer: Sinectis Webmail 5.6.16-1.4.4 From: nbarraza@uolsinectis.com.ar To: pgsql-performance@postgresql.org Subject: Command Reply-To: nbarraza@uolsinectis.com.ar Message-Id: <20031211164517.CFE346C3BE@tino.sinectis.com.ar> Date: Thu, 11 Dec 2003 13:45:17 -0300 (GMT+3) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/181 X-Sequence-Number: 5041 show From pgsql-performance-owner@postgresql.org Thu Dec 11 16:20:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A27DDD1B456 for ; Thu, 11 Dec 2003 20:20:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05943-02 for ; Thu, 11 Dec 2003 16:19:52 -0400 (AST) Received: from visionlink.org (mail.visionlink.org [208.139.207.159]) by svr1.postgresql.org (Postfix) with SMTP id EB0FBD1D0AE for ; Thu, 11 Dec 2003 16:19:49 -0400 (AST) Received: from [65.100.170.142] (HELO [192.168.1.201]) by visionlink.org (Stalker SMTP Server 1.8b8) with ESMTP id S.0002302965 for ; Thu, 11 Dec 2003 13:20:53 -0700 Mime-Version: 1.0 X-Sender: bohmer@mail.visionlink.org Message-Id: Date: Thu, 11 Dec 2003 13:19:42 -0700 To: pgsql-performance@postgresql.org From: Jeff Bohmer Subject: Hardware suggestions for Linux/PGSQL server Content-Type: text/plain; charset="us-ascii" ; format="flowed" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/183 X-Sequence-Number: 5043 Hi everyone, I want to pick your brains for hardware suggestions about a Linux-based PostgreSQL 7.4 server. It will be a dedicated DB server backing our web sites and hit by application servers (which do connection pooling). I've hopefully provided all relevant information below. Any thoughts, comments or suggestions are welcome. Our current server and database: Mac OS X Server 10.2.8 single 1.25GHz G4 2 GB 333MHz RAM 7200 rpm SCSI drive for OS, logs 15k rpm SCSI drive for data PostgreSQL 7.3.4 1 database, 1.1 GB in size, growing by ~15 MB / week 60 tables, 1 schema, largest is 1m rows, 1 at 600k, 3 at 100k Peak traffic: 500 UPDATEs, INSERTs and DELETEs / minute 6000 SELECTs / minutes 90 connections Performance is fine most of the time, but not during peak loads. We're never swapping and disk IO during the SELECT peaks is hardly anything (under 3MB/sec). I think UPDATE peaks might be saturating disk IO. Normally, most queries finish in under .05 seconds. Some take 2-3 seconds. During peaks, the fast queries are just OK and the slower ones take too long (like over 8 seconds). We're moving to Linux from OS X for improved stability and more hardware options. We need to do this soon. The current server is max'd out at 2GB RAM and I'm afraid might start swapping in a month. Projected database/traffic in 12 months: Database size will be at least 2.5 GB Largest table still 1m rows, but 100k tables will grow to 250k Will be replicated to a suitable standby slave machine Peak traffic: 2k UPDATEs, INSERTs, DELETEs / minute 20k SELECTs / minute 150 - 200 connections We're willing to shell out extra bucks to get something that will undoubtedly handle the projected peak load in 12 months with excellent performance. But we're not familiar with PG's performance on Linux and don't like to waste money. I've been thinking of this (overkill? not enough?): 2 Intel 32-bit CPUs Lowest clock speed chip for the fastest available memory bus 4 GB RAM (maybe we only need 3 GB to start with?) SCSI RAID 1 for OS For PostgreSQL data and logs ... 15k rpm SCSI disks RAID 5, 7 disks, 256MB battery-backed write cache (Should we save $ and get a 4-disk RAID 10 array?) I wonder about the 32bit+bigmem vs. 64bit question. At what database size will we need more than 4GB RAM? We'd like to always have enough RAM to cache the entire database. While 64bit is in our long-term future, we're willing to stick with 32bit Linux until 64bit Linux on Itanium/Opteron and 64bit PostgreSQL "settle in" to proven production-quality. TIA, - Jeff -- Jeff Bohmer VisionLink, Inc. _________________________________ 303.402.0170 www.visionlink.org _________________________________ People. Tools. Change. Community. From pgsql-performance-owner@postgresql.org Thu Dec 11 16:28:51 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0AABAD1B456 for ; Thu, 11 Dec 2003 20:28:50 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06229-05 for ; Thu, 11 Dec 2003 16:28:20 -0400 (AST) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id CA045D1D0AE for ; Thu, 11 Dec 2003 16:28:17 -0400 (AST) Received: (qmail 10562 invoked by uid 500); 11 Dec 2003 20:27:55 -0000 Date: Thu, 11 Dec 2003 14:27:55 -0600 From: Bruno Wolff III To: sandra ruiz Cc: pgsql-performance@postgresql.org Subject: Re: hints in Postgres? Message-ID: <20031211202755.GB9844@wolff.to> Mail-Followup-To: sandra ruiz , 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/1.5.4i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/184 X-Sequence-Number: 5044 On Thu, Dec 11, 2003 at 11:00:19 -0500, sandra ruiz wrote: > Hi list, > > I need to know if there is anything like hints of Oracle in > Postgres..otherwise..I wish to find a way to force a query plan to use the > indexes or tell the optimizer things like "optimize based in statistics", > "I want to define the order of the a join" , "optimize based on a execution > plan that I consider the best" ... There are a few things you can do. You can explicitly fix the join order using INNER JOIN (in 7.4 you have to set a GUC variable for this to force join order). You can disable specific plan types (though sequential just becomes very expensive as sometimes there is no other way to do things). You can set tuning values to properly express the relative cost of things like CPU time, sequential disk reads and random disk reads. These are done by setting GUC variables either in the postgres config file or using SET commands. They are per backend so some queries can be done using one set of values while others going on at the same time use different values. From pgsql-performance-owner@postgresql.org Thu Dec 11 17:00:25 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4F5CCD1B45A for ; Thu, 11 Dec 2003 21:00:24 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11733-04 for ; Thu, 11 Dec 2003 16:59:54 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 2E7E5D1B456 for ; Thu, 11 Dec 2003 16:59:52 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBBKxobI038025 for ; Thu, 11 Dec 2003 20:59:51 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBBKo3Me037153 for pgsql-performance@postgresql.org; Thu, 11 Dec 2003 20:50:03 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Hardware suggestions for Linux/PGSQL server Date: Thu, 11 Dec 2003 12:50:06 -0800 Organization: Hub.Org Networking Services Lines: 40 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/185 X-Sequence-Number: 5045 Jeff Bohmer wrote: > We're willing to shell out extra bucks to get something that will > undoubtedly handle the projected peak load in 12 months with excellent > performance. But we're not familiar with PG's performance on Linux and > don't like to waste money. Properly tuned, PG on Linux runs really nice. A few people have mentioned the VM swapping algorithm on Linux is semi-dumb. I get around that problem by having a ton of memory and almost no swap. > I've been thinking of this (overkill? not enough?): > 2 Intel 32-bit CPUs > Lowest clock speed chip for the fastest available memory bus > 4 GB RAM (maybe we only need 3 GB to start with?) > SCSI RAID 1 for OS > For PostgreSQL data and logs ... > 15k rpm SCSI disks > RAID 5, 7 disks, 256MB battery-backed write cache > (Should we save $ and get a 4-disk RAID 10 array?) > > I wonder about the 32bit+bigmem vs. 64bit question. At what database > size will we need more than 4GB RAM? With 4GB of RAM, you're already running into bigmem. By default, Linux gives 2GB of address space to programs and 2GB to kernel. I usually see people quote 5%-15% penalty in general for using PAE versus a flat address space. I've seen simple MySQL benchmarks where 64-bit versions run 35%+ faster versus 32-bit+PAE but how that translates to PG, I dunno yet. > We'd like to always have enough RAM to cache the entire database. While > 64bit is in our long-term future, we're willing to stick with 32bit > Linux until 64bit Linux on Itanium/Opteron and 64bit PostgreSQL "settle > in" to proven production-quality. Well if this is the case, you probably should get an Opteron server *now* and just run 32-bit Linux on it until you're sure about the software. No point in buying a Xeon and then throwing the machine away in a year when you decide you need 64-bit for more speed. From pgsql-performance-owner@postgresql.org Thu Dec 11 18:02:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ADD90D1B4AD for ; Thu, 11 Dec 2003 22:02:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20062-05 for ; Thu, 11 Dec 2003 18:02:27 -0400 (AST) Received: from visionlink.org (mail.visionlink.org [208.139.207.159]) by svr1.postgresql.org (Postfix) with SMTP id 3C9AFD1DE9B for ; Thu, 11 Dec 2003 18:02:24 -0400 (AST) Received: from [65.100.170.142] (HELO [192.168.1.201]) by visionlink.org (Stalker SMTP Server 1.8b8) with ESMTP id S.0002303498 for ; Thu, 11 Dec 2003 15:03:25 -0700 Mime-Version: 1.0 X-Sender: bohmer@mail.visionlink.org Message-Id: In-Reply-To: References: Date: Thu, 11 Dec 2003 15:02:11 -0700 To: pgsql-performance@postgresql.org From: Jeff Bohmer Subject: Re: Hardware suggestions for Linux/PGSQL server Content-Type: text/plain; charset="us-ascii" ; format="flowed" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/186 X-Sequence-Number: 5046 >Properly tuned, PG on Linux runs really nice. A few people have >mentioned the VM swapping algorithm on Linux is semi-dumb. I get >around that problem by having a ton of memory and almost no swap. I think we want your approach: enough RAM to avoid swapping altogether. >With 4GB of RAM, you're already running into bigmem. By default, >Linux gives 2GB of address space to programs and 2GB to kernel. It seems I don't fully understand the bigmem situation. I've searched the archives, googled, checked RedHat's docs, etc. But I'm getting conflicting, incomplete and/or out of date information. Does anyone have pointers to bigmem info or configuration for the 2.4 kernel? If Linux is setup with 2GB for kernel and 2GB for user, would that be OK with a DB size of 2-2.5 GB? I'm figuring the kernel will cache most/all of the DB in it's 2GB and there's 2GB left for PG processes. Where does PG's SHM buffers live, kernel or user? (I don't plan on going crazy with buffers, but will guess we'd need about 128MB, 256MB at most.) >I usually see people quote 5%-15% penalty in general for using PAE >versus a flat address space. I've seen simple MySQL benchmarks where >64-bit versions run 35%+ faster versus 32-bit+PAE but how that >translates to PG, I dunno yet. > >>We'd like to always have enough RAM to cache the entire database. >>While 64bit is in our long-term future, we're willing to stick with >>32bit Linux until 64bit Linux on Itanium/Opteron and 64bit >>PostgreSQL "settle in" to proven production-quality. > >Well if this is the case, you probably should get an Opteron server >*now* and just run 32-bit Linux on it until you're sure about the >software. No point in buying a Xeon and then throwing the machine >away in a year when you decide you need 64-bit for more speed. That's a good point. I had forgotten about the option to run 32bit on an Operton. If we had 3GB or 4GB initially on an Opteron, we'd need bigmem for 32bit Linux, right? This might work nicely since we'd factor in the penalty from PAE for now and have the performance boost from moving to 64bit available on demand. Not having to build another DB server in a year would also be nice. FYI, we need stability first and performance second. Thank you, - Jeff -- Jeff Bohmer VisionLink, Inc. _________________________________ 303.402.0170 www.visionlink.org _________________________________ People. Tools. Change. Community. From pgsql-performance-owner@postgresql.org Thu Dec 11 19:08:08 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1BD25D1B480 for ; Thu, 11 Dec 2003 23:08:07 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28357-10 for ; Thu, 11 Dec 2003 19:07:38 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id EB134D1D455 for ; Thu, 11 Dec 2003 19:07:30 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hBBN6k8q016183; Thu, 11 Dec 2003 16:06:46 -0700 (MST) Date: Thu, 11 Dec 2003 15:48:58 -0700 (MST) From: "scott.marlowe" To: Jeff Bohmer Cc: Subject: Re: Hardware suggestions for Linux/PGSQL server In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/187 X-Sequence-Number: 5047 Just one more piece of advice, you might want to look into a good battery backed cache hardware RAID controller. They work quite well for heavily updated databases. The more drives you throw at the RAID array the faster it will be. From pgsql-performance-owner@postgresql.org Thu Dec 11 20:00:27 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 88934D1DF1A for ; Fri, 12 Dec 2003 00:00:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 30808-09 for ; Thu, 11 Dec 2003 19:59:54 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id AA479D1D455 for ; Thu, 11 Dec 2003 19:59:51 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBBNxpbI061868 for ; Thu, 11 Dec 2003 23:59:51 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBBNWfcJ058513 for pgsql-performance@postgresql.org; Thu, 11 Dec 2003 23:32:41 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Hardware suggestions for Linux/PGSQL server Date: Thu, 11 Dec 2003 15:32:47 -0800 Organization: Hub.Org Networking Services Lines: 44 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/188 X-Sequence-Number: 5048 Jeff Bohmer wrote: > It seems I don't fully understand the bigmem situation. I've searched > the archives, googled, checked RedHat's docs, etc. But I'm getting > conflicting, incomplete and/or out of date information. Does anyone > have pointers to bigmem info or configuration for the 2.4 kernel? Bigmem is the name for Linux's PAE support. > If Linux is setup with 2GB for kernel and 2GB for user, would that be OK > with a DB size of 2-2.5 GB? I'm figuring the kernel will cache most/all > of the DB in it's 2GB and there's 2GB left for PG processes. Where does > PG's SHM buffers live, kernel or user? (I don't plan on going crazy > with buffers, but will guess we'd need about 128MB, 256MB at most.) PG's SHM buffers live in user. Whether Linux's OS caches lives in user or kernel, I think it's in kernel and I remember reading a max of ~950KB w/o bigmem which means your 3.5GB of available OS memory will definitely have to be swapped in and out of kernel space using PAE. >> Well if this is the case, you probably should get an Opteron server >> *now* and just run 32-bit Linux on it until you're sure about the >> software. No point in buying a Xeon and then throwing the machine away >> in a year when you decide you need 64-bit for more speed. > > That's a good point. I had forgotten about the option to run 32bit on > an Operton. If we had 3GB or 4GB initially on an Opteron, we'd need > bigmem for 32bit Linux, right? > > This might work nicely since we'd factor in the penalty from PAE for now > and have the performance boost from moving to 64bit available on > demand. Not having to build another DB server in a year would also be > nice. > > FYI, we need stability first and performance second. We ordered a 2x Opteron server the moment the CPU was released and it's been perfect -- except for one incident where the PCI riser card had drifted out of the PCI slot due to the heavy SCSI cables connected to the card. I think most of the Opteron server MBs are pretty solid but you want extra peace-of-mind, you could get a server from Newisys as they pack in a cartload of extra monitoring features. From pgsql-performance-owner@postgresql.org Thu Dec 11 21:17:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 75486D1B513 for ; Fri, 12 Dec 2003 01:17:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 40621-10 for ; Thu, 11 Dec 2003 21:17:16 -0400 (AST) Received: from p15135922.pureserver.info (willfork.com [217.160.217.214]) by svr1.postgresql.org (Postfix) with ESMTP id 0C4DFD1B507 for ; Thu, 11 Dec 2003 21:17:12 -0400 (AST) Received: from pD9EB7C08.dip.t-dialin.net (pD9EB7C08.dip.t-dialin.net [217.235.124.8]) (authenticated bits=0) by p15135922.pureserver.info (8.12.6/8.12.6) with ESMTP id hBC1H6Am011618 for ; Fri, 12 Dec 2003 02:17:09 +0100 Subject: Measuring execution time for sql called from PL/pgSQL From: Aram Kananov To: "pgsql-performance@postgresql.org" Content-Type: text/plain Organization: Message-Id: <1071191825.5163.164.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 12 Dec 2003 02:17:06 +0100 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/189 X-Sequence-Number: 5049 Hi, I've got very slow insert performance on some table which has trigger based on complex PL/pgSQL function. Apparently insert is slow due some slow sql inside that function, since CPU load is very high and disk usage is low during insert. I run Red Hat 9 Anthlon 2.6 1GB ram Fast IDE Disk Setting following in postgres.conf apparently doesn't help: log_statement = true log_duration = true since it logs only sql issued by client. It logs only once per session the sql text but during call to the PL/pgSQL function, but of course no duration. Due the complexity of PL/pgSQL function trying to step by step see the execution plans is very time consuming. Q1) Is there any way to see which statements are called for PL/pgSQL and their duration? I've tried to measure the duration of sql with printing out "localtimestamp" but for some reason during the same pg/plsql call it returns the same value: Example: Following gets and prints out the localtimestamp value in the loop create or replace function foobar() returns integer as ' declare v timestamp; begin loop select localtimestamp into v; raise notice ''Timestamp: %'', v; end loop; return null; end; ' language 'plpgsql' ; and as result of "select foobar();" i constantly get the same value: NOTICE: Timestamp: 2003-12-12 01:51:35.768053 NOTICE: Timestamp: 2003-12-12 01:51:35.768053 NOTICE: Timestamp: 2003-12-12 01:51:35.768053 NOTICE: Timestamp: 2003-12-12 01:51:35.768053 NOTICE: Timestamp: 2003-12-12 01:51:35.768053 Q2) what i do wrong here and what is the "Proper Way" to measure execution time of sql called inside PG/plSQL. Thanks in advance WBR -- Aram From pgsql-performance-owner@postgresql.org Tue Dec 16 17:13:09 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 940DAD1B507 for ; Fri, 12 Dec 2003 01:57:48 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48748-06 for ; Thu, 11 Dec 2003 21:57:20 -0400 (AST) Received: from www.smart-start-hosting.com (unknown [38.118.153.169]) by svr1.postgresql.org (Postfix) with ESMTP id 74D23D1B454 for ; Thu, 11 Dec 2003 21:57:16 -0400 (AST) Received: from shadovitz.com (localhost [127.0.0.1]) by localhost (8.10.2/8.10.2) with ESMTP id hBC1oVm20348; Thu, 11 Dec 2003 17:50:31 -0800 From: "David Shadovitz" To: Aram Kananov , "pgsql-performance@postgresql.org" Subject: Re: Measuring execution time for sql called from PL/pgSQL Date: Thu, 11 Dec 2003 18:50:31 -0700 Message-Id: <20031212015031.M16173@www.shadovitz.com> In-Reply-To: <1071191825.5163.164.camel@localhost.localdomain> References: <1071191825.5163.164.camel@localhost.localdomain> X-Mailer: Open WebMail 1.81 20021203 X-OriginatingIP: 216.180.56.31 (david@shadovitz.com) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/236 X-Sequence-Number: 5096 > I've tried to measure the duration of sql with printing out > "localtimestamp" but for some reason during the same pg/plsql call > it returns the same value: Aram, >From http://www.postgresql.org/docs/current/static/functions-datetime.html: There is also the function timeofday(), which for historical reasons returns a text string rather than a timestamp value: SELECT timeofday(); Result: Sat Feb 17 19:07:32.000126 2001 EST It is important to know that CURRENT_TIMESTAMP and related functions return the start time of the current transaction; their values do not change during the transaction. This is considered a feature: the intent is to allow a single transaction to have a consistent notion of the "current" time, so that multiple modifications within the same transaction bear the same time stamp. timeofday() returns the wall-clock time and does advance during transactions. -David From pgsql-performance-owner@postgresql.org Thu Dec 11 22:35:52 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A717BD1B44D for ; Fri, 12 Dec 2003 02:35:50 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50809-02 for ; Thu, 11 Dec 2003 22:35:23 -0400 (AST) Received: from dhcp1.lsd.di.uminho.pt (gsd.di.uminho.pt [193.136.20.132]) by svr1.postgresql.org (Postfix) with ESMTP id 89DB6D1B460 for ; Thu, 11 Dec 2003 22:35:17 -0400 (AST) Received: from lsd.di.uminho.pt (localhost.localdomain [127.0.0.1]) by dhcp1.lsd.di.uminho.pt (8.12.8/8.12.8) with ESMTP id hBC2ZGdS003891; Fri, 12 Dec 2003 02:35:18 GMT Message-ID: <3FD92964.8040709@lsd.di.uminho.pt> Date: Fri, 12 Dec 2003 02:35:16 +0000 From: Alfranio Tavares Correia Junior User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030225 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Performance problems with a higher number of clients References: <3FD7EEE8.4080904@lsd.di.uminho.pt> <13178.1071154457@sss.pgh.pa.us> In-Reply-To: <13178.1071154457@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 postgresql.org X-Archive-Number: 200312/190 X-Sequence-Number: 5050 Thanks for the advices, The performance is a bit better now. Unfortunately, the machine does not allow to put more than 200 - ~250 users without noticing swap hell. I have to face the fact that I don't have enough memory.... I used the following configuration: effective_cache_size = 65000 shared_buffers = 10000 random_page_cost = 2 cpu_index_tuple_cost = 0.0005 sort_mem = 512 - I tested each query to see the amount of space required to sort as Jeff suggested --> nothing above this value I tested the system with 100, 200, 300, 400, 500 and finally 250 users. Until ~250 users the system presents good response time and the swap almost does not exist. During these expirements, I also started psql and tried to run some queries. Unfortunately, even with ~250 users there is one query that takes too long to finish... In fact, I canceled its execution after 5 minutes waiting to see anything. This is the query: select count(distinct(s_i_id)) from stock, order_line where ol_w_id = _xx_ and ol_d_id = _xx_ and ol_o_id between _xx_ and _xx_ and s_w_id = ol_w_id and s_i_id = ol_i_id and s_quantity < _xx_; When the system has no load, after a vacuum -f, I can execute the query and the plan produced is presented as follows: Aggregate (cost=49782.16..49782.16 rows=1 width=4) (actual time=52361.573..52361.574 rows=1 loops=1) -> Nested Loop (cost=0.00..49780.24 rows=768 width=4) (actual time=101.554..52328.913 rows=952 loops=1) -> Index Scan using pk_order_line on order_line o (cost=0.00..15779.32 rows=8432 width=4) (actual time=84.352..151.345 rows=8964 loops=1) Index Cond: ((ol_w_id = 4) AND (ol_d_id = 4) AND (ol_o_id >= 100) AND (ol_o_id <= 1000)) -> Index Scan using pk_stock on stock (cost=0.00..4.02 rows=1 width=4) (actual time=5.814..5.814 rows=0 loops=8964) Index Cond: ((stock.s_w_id = 4) AND (stock.s_i_id = "outer".ol_i_id)) Filter: (s_quantity < 20) Total runtime: 52403.673 ms (8 rows) The talbes are designed as follows: --ROWS ~5000000 CREATE TABLE stock ( s_i_id int NOT NULL , s_w_id int NOT NULL , s_quantity int NULL , s_dist_01 char (24) NULL , s_dist_02 char (24) NULL , s_dist_03 char (24) NULL , s_dist_04 char (24) NULL , s_dist_05 char (24) NULL , s_dist_06 char (24) NULL , s_dist_07 char (24) NULL , s_dist_08 char (24) NULL , s_dist_09 char (24) NULL , s_dist_10 char (24) NULL , s_ytd int NULL , s_order_cnt int NULL , s_remote_cnt int NULL , s_data char (50) NULL ); --ROWS ~15196318 CREATE TABLE order_line ( ol_o_id int NOT NULL , ol_d_id int NOT NULL , ol_w_id int NOT NULL , ol_number int NOT NULL , ol_i_id int NULL , ol_supply_w_id int NULL , ol_delivery_d timestamp NULL , ol_quantity int NULL , ol_amount numeric(6, 2) NULL , ol_dist_info char (24) NULL ); ALTER TABLE stock ADD CONSTRAINT PK_stock PRIMARY KEY ( s_w_id, s_i_id ); ALTER TABLE order_line ADD CONSTRAINT PK_order_line PRIMARY KEY ( ol_w_id, ol_d_id, ol_o_id, ol_number ); CREATE INDEX IX_order_line ON order_line(ol_i_id); Any suggestion ? Tom Lane wrote: >Alfranio Correia Junior writes: > > >>I am facing a problem trying to put 500 concurrent users accessing >>a postgresql instance. >> >> > >I think you're going to need to buy more RAM. 1Gb of RAM means there >is a maximum of 2Mb available per Postgres process before you start >to go into swap hell --- in practice a lot less, since you have to allow >for other things like the kernel and other applications. > >AFAIR TPC-C doesn't involve any complex queries, so it's possible you >could run it with only 1Mb of workspace per process, but not when >you've configured > > > >>sort_mem = 10240 >> >> > >That's ten times more than your configuration can possibly support. >(I don't recall whether TPC-C uses any queries that would sort, so >it's possible this setting isn't affecting you; but if you are doing >any sorts then it's killing you.) > >Bottom line is you probably need more RAM. > > regards, tom lane > >---------------------------(end of broadcast)--------------------------- >TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > > From pgsql-performance-owner@postgresql.org Fri Dec 12 02:36:30 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 971B7D1CAFE for ; Fri, 12 Dec 2003 06:36:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 78740-02 for ; Fri, 12 Dec 2003 02:35:58 -0400 (AST) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id D7545D1CB36 for ; Fri, 12 Dec 2003 02:35:54 -0400 (AST) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id hBC6ZxCW007264 for ; Fri, 12 Dec 2003 12:05:59 +0530 Received: from persistent.co.in (daithan.intranet.pspl.co.in [192.168.7.161]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id hBC6ZuAo007197; Fri, 12 Dec 2003 12:05:56 +0530 Message-ID: <3FD961C5.9070204@persistent.co.in> Date: Fri, 12 Dec 2003 12:05:49 +0530 From: Shridhar Daithankar Organization: Persistent Systems Pvt. Ltd. User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Jeff Bohmer Cc: pgsql-performance@postgresql.org Subject: Re: Hardware suggestions for Linux/PGSQL server References: In-Reply-To: Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/191 X-Sequence-Number: 5051 Jeff Bohmer wrote: >> Well if this is the case, you probably should get an Opteron server >> *now* and just run 32-bit Linux on it until you're sure about the >> software. No point in buying a Xeon and then throwing the machine away >> in a year when you decide you need 64-bit for more speed. > > > That's a good point. I had forgotten about the option to run 32bit on > an Operton. If we had 3GB or 4GB initially on an Opteron, we'd need > bigmem for 32bit Linux, right? > > This might work nicely since we'd factor in the penalty from PAE for now > and have the performance boost from moving to 64bit available on > demand. Not having to build another DB server in a year would also be > nice. FWIW, there are only two pieces of software that need 64bit aware for a typical server job. Kernel and glibc. Rest of the apps can do fine as 32 bits unless you are oracle and insist on outsmarting OS. In fact running 32 bit apps on 64 bit OS has plenty of advantages like effectively using the cache. Unless you need 64bit, going for 64bit software is not advised. Shridhar -- ----------------------------- Shridhar Daithankar LIMS CPE Team Member, PSPL. mailto:shridhar_daithankar@persistent.co.in Phone:- +91-20-5676700 Extn.270 Fax :- +91-20-5676701 ----------------------------- From pgsql-performance-owner@postgresql.org Fri Dec 12 02:50:04 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 366E1D1B8BB; Fri, 12 Dec 2003 06:50:00 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80913-04; Fri, 12 Dec 2003 02:49:30 -0400 (AST) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id C6E5ED1B441; Fri, 12 Dec 2003 02:49:27 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hBC6nQR15608; Fri, 12 Dec 2003 01:49:26 -0500 (EST) From: Bruce Momjian Message-Id: <200312120649.hBC6nQR15608@candle.pha.pa.us> Subject: fsync method checking In-Reply-To: <3FD6B596.8090803@paradise.net.nz> To: Mark Kirkwood Date: Fri, 12 Dec 2003 01:49:26 -0500 (EST) Cc: pgsql-performance@postgresql.org, PostgreSQL-development X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Type: multipart/mixed; boundary=ELM1071211762-12324-0_ Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/192 X-Sequence-Number: 5052 --ELM1071211762-12324-0_ Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII Mark Kirkwood wrote: > This is a well-worn thread title - apologies, but these results seemed > interesting, and hopefully useful in the quest to get better performance > on Solaris: > > I was curious to see if the rather uninspiring pgbench performance > obtained from a Sun 280R (see General: ATA Disks and RAID controllers > for database servers) could be improved if more time was spent > tuning. > > With the help of a fellow workmate who is a bit of a Solaris guy, we > decided to have a go. > > The major performance killer appeared to be mounting the filesystem with > the logging option. The next most significant seemed to be the choice of > sync_method for Pg - the default (open_datasync), which we initially > thought should be the best - appears noticeably slower than fdatasync. I thought the default was fdatasync, but looking at the code it seems the default is open_datasync if O_DSYNC is available. I assume the logic is that we usually do only one write() before fsync(), so open_datasync should be faster. Why do we not use O_FSYNC over fsync(). Looking at the code: #if defined(O_SYNC) #define OPEN_SYNC_FLAG O_SYNC #else #if defined(O_FSYNC) #define OPEN_SYNC_FLAG O_FSYNC #endif #endif #if defined(OPEN_SYNC_FLAG) #if defined(O_DSYNC) && (O_DSYNC != OPEN_SYNC_FLAG) #define OPEN_DATASYNC_FLAG O_DSYNC #endif #endif #if defined(OPEN_DATASYNC_FLAG) #define DEFAULT_SYNC_METHOD_STR "open_datasync" #define DEFAULT_SYNC_METHOD SYNC_METHOD_OPEN #define DEFAULT_SYNC_FLAGBIT OPEN_DATASYNC_FLAG #else #if defined(HAVE_FDATASYNC) #define DEFAULT_SYNC_METHOD_STR "fdatasync" #define DEFAULT_SYNC_METHOD SYNC_METHOD_FDATASYNC #define DEFAULT_SYNC_FLAGBIT 0 #else #define DEFAULT_SYNC_METHOD_STR "fsync" #define DEFAULT_SYNC_METHOD SYNC_METHOD_FSYNC #define DEFAULT_SYNC_FLAGBIT 0 #endif #endif I think the problem is that we prefer O_DSYNC over fdatasync, but do not prefer O_FSYNC over fsync. Running the attached test program shows on BSD/OS 4.3: write 0.000360 write & fsync 0.001391 write, close & fsync 0.001308 open o_fsync, write 0.000924 showing O_FSYNC faster than fsync(). -- 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 --ELM1071211762-12324-0_ Content-Transfer-Encoding: 7bit Content-Type: text/plain Content-Disposition: inline; filename="/wrk/tmp/test_sync.c" /* * test_fsync.c * tests if fsync can be done from another process than the original write */ #include #include #include #include #include #include void die(char *str); void print_elapse(struct timeval start_t, struct timeval elapse_t); int main(int argc, char *argv[]) { struct timeval start_t; struct timeval elapse_t; int tmpfile; char *strout = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"; /* write only */ gettimeofday(&start_t, NULL); if ((tmpfile = open("/var/tmp/test_fsync.out", O_RDWR | O_CREAT)) == -1) die("can't open /var/tmp/test_fsync.out"); write(tmpfile, &strout, 200); close(tmpfile); gettimeofday(&elapse_t, NULL); unlink("/var/tmp/test_fsync.out"); printf("write "); print_elapse(start_t, elapse_t); printf("\n"); /* write & fsync */ gettimeofday(&start_t, NULL); if ((tmpfile = open("/var/tmp/test_fsync.out", O_RDWR | O_CREAT)) == -1) die("can't open /var/tmp/test_fsync.out"); write(tmpfile, &strout, 200); fsync(tmpfile); close(tmpfile); gettimeofday(&elapse_t, NULL); unlink("/var/tmp/test_fsync.out"); printf("write & fsync "); print_elapse(start_t, elapse_t); printf("\n"); /* write, close & fsync */ gettimeofday(&start_t, NULL); if ((tmpfile = open("/var/tmp/test_fsync.out", O_RDWR | O_CREAT)) == -1) die("can't open /var/tmp/test_fsync.out"); write(tmpfile, &strout, 200); close(tmpfile); /* reopen file */ if ((tmpfile = open("/var/tmp/test_fsync.out", O_RDWR | O_CREAT)) == -1) die("can't open /var/tmp/test_fsync.out"); fsync(tmpfile); close(tmpfile); gettimeofday(&elapse_t, NULL); unlink("/var/tmp/test_fsync.out"); printf("write, close & fsync "); print_elapse(start_t, elapse_t); printf("\n"); /* open_fsync, write */ gettimeofday(&start_t, NULL); if ((tmpfile = open("/var/tmp/test_fsync.out", O_RDWR | O_CREAT | O_FSYNC)) == -1) die("can't open /var/tmp/test_fsync.out"); write(tmpfile, &strout, 200); close(tmpfile); gettimeofday(&elapse_t, NULL); unlink("/var/tmp/test_fsync.out"); printf("open o_fsync, write "); print_elapse(start_t, elapse_t); printf("\n"); return 0; } void print_elapse(struct timeval start_t, struct timeval elapse_t) { if (elapse_t.tv_usec < start_t.tv_usec) { elapse_t.tv_sec--; elapse_t.tv_usec += 1000000; } printf("%ld.%06ld", (long) (elapse_t.tv_sec - start_t.tv_sec), (long) (elapse_t.tv_usec - start_t.tv_usec)); } void die(char *str) { fprintf(stderr, "%s", str); exit(1); } --ELM1071211762-12324-0_-- From pgsql-performance-owner@postgresql.org Fri Dec 12 03:47:23 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 52B20D1DE69 for ; Fri, 12 Dec 2003 07:47:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81363-10 for ; Fri, 12 Dec 2003 03:46:51 -0400 (AST) Received: from serwer.skawsoft.com.pl (skawina.eu.org [80.48.213.66]) by svr1.postgresql.org (Postfix) with ESMTP id 8297CD1DE67 for ; Fri, 12 Dec 2003 03:46:48 -0400 (AST) Received: from klaster.net (core-1.citynet.pl [80.48.135.69]) by serwer.skawsoft.com.pl (Postfix) with ESMTP id AF5A02B3CE; Fri, 12 Dec 2003 08:46:31 +0100 (CET) Message-ID: <3FD9725D.9030304@klaster.net> Date: Fri, 12 Dec 2003 08:46:37 +0100 From: Tomasz Myrta User-Agent: Mozilla/5.0 (Windows; U; Win 9x 4.90; pl-PL; rv:1.5) Gecko/20031007 X-Accept-Language: pl, en-us, en MIME-Version: 1.0 To: Aram Kananov Cc: "pgsql-performance@postgresql.org" Subject: Re: Measuring execution time for sql called from PL/pgSQL References: <1071191825.5163.164.camel@localhost.localdomain> In-Reply-To: <1071191825.5163.164.camel@localhost.localdomain> Content-Type: text/plain; charset=ISO-8859-2; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/193 X-Sequence-Number: 5053 Dnia 2003-12-12 02:17, U�ytkownik Aram Kananov napisa�: > select localtimestamp into v; > raise notice ''Timestamp: %'', v; Don't use localtimestamp, now() neither any transaction based time function. They all return the same value among whole transaction. The only time function, which can be used for performance tests is timeofday(). You can read more about time functions in manual. Regards, Tomasz Myrta From pgsql-performance-owner@postgresql.org Fri Dec 12 04:28:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B52BCD1DE81 for ; Fri, 12 Dec 2003 08:28:37 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 83925-10 for ; Fri, 12 Dec 2003 04:28:08 -0400 (AST) Received: from conure.mail.pas.earthlink.net (conure.mail.pas.earthlink.net [207.217.120.54]) by svr1.postgresql.org (Postfix) with ESMTP id 93129D1DE69 for ; Fri, 12 Dec 2003 04:28:06 -0400 (AST) Received: from hsa119.pool029.at101.earthlink.net ([216.249.98.119] helo=shadovitzcmptr) by conure.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 1AUied-0004Zt-00 for pgsql-performance@postgresql.org; Fri, 12 Dec 2003 00:28:07 -0800 Received: by localhost with Microsoft MAPI; Fri, 12 Dec 2003 00:18:41 -0800 Message-ID: <01C3C045.7EDD13C0.david@shadovitz.com> From: David Shadovitz Reply-To: "david@shadovitz.com" To: "'pgsql-performance@postgresql.org'" Subject: Query plan - now what? Date: Fri, 12 Dec 2003 00:18:12 -0800 X-Mailer: Microsoft Internet E-mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/194 X-Sequence-Number: 5054 Well, now that I have the plan for my slow-running query, what do I do? Where should I focus my attention? Thanks. -David Hash Join (cost=16620.59..22331.88 rows=40133 width=266) (actual time=118773.28..580889.01 rows=57076 loops=1) -> Hash Join (cost=16619.49..21628.48 rows=40133 width=249) (actual time=118771.29..535709.47 rows=57076 loops=1) -> Hash Join (cost=16618.41..20724.39 rows=40133 width=240) (actual time=118768.04..432327.82 rows=57076 loops=1) -> Hash Join (cost=16617.34..19920.66 rows=40133 width=223) (actual time=118764.67..340333.78 rows=57076 loops=l) -> Hash Join (cost=16616.14..19217.14 rows=4Ol33 width=214) (actual time=118761.38..258978.8l row=57076 loops=1) -> Merge Join (cost=16615.07..18413.42 rows=40133 width=205) (actual time=118758.74..187180.55 rows=57076 loops=i) -> Index Scan using grf_grf_id_idx on giraffes (cost=O.O0..1115.61 rows=53874 width=8) (actual time=2.37..6802.38 rows=57077 loops=l) -> Sort (cost=l66l5.07..16615.07 rows=18554 width=197) (actual time=118755.11..120261.06 rows=59416 loops=l) -> Hash Join (cost=8126.08..14152.54 rows=18554 width=197) (actual time=50615.72..l09853.7l rows=16310 loops=1) -> Hash Join (cost=8124.39..12690.30 rows=24907 width=179) (actual time=50607.36..86868.58 rows=iSBiS loops=1) -> Hash Join (cost=249.26..2375.23 rows=24907 width=131) (actual time=23476.42..35107.80 rows=16310 loops=l) -> Nested Loop (cost=248.2l..1938.31 rows=24907 width=118) (actual time=23474.70..28155.13 rows=16310 loops=1) -> Seq Scan on zebras (cost=0.00..l.0l rows=l width=14) (actual time=O.64..0.72 rows=1 ioops=1) -> Materialize (cost=1688.23..l688.23 rows=24907 width=104) (actual time=23473.77..23834.26 rows=16310 loops=l) -> Hash Join (cost=248.21..1688.23 rows=24907 width=lO4) (actual time=1199.26..23059.92 rows=16310 loops=l) -> Seq Scan on frogs (cost=0.00..755.07 rows=24907 width=83) (actual time=0.53..4629.58 rows=25702 loops=l) -> Hash (cost=225.57..225.57 rows=9057 width=21) (actual time=1198.0l..1198.01 rows=0 loops=1) -> Seq Scan on tigers (cost=0.00..225.57 rows=9057 width=21) (actual time=0.39..892.67 rows=9927 loops=1) -> Hash (cost=l.O4..1.-4 rows=4 width=13) (actual time=l.07..1.07 rows=0 loops=1) -> Seq Scan on deers (cost=0.0O..1.04 rows=4 width=13) (actual time=0.64..0.95 rows=4 loops=1) -> Hash (cost=4955.28..4955.28 rows=91528 width=48) (actual tlne=27O40.82..27040.82 rows=0 loops=1) -> Seq Scan on warthogs (cost=0.00..4955.28 rows=91528 width=48) (actual time=3.92..24031.27 rows=91528 loops=1) -> Hash (cost=1.55..1.55 rows=55 width=18) (actual time=7.l3..7.13 rows=0 loops=1) -> Seq Scan on monkeys (cost=0.00..l.55 rows=55 width=18) (actual time=0.64..5.38 rows=55 loops=1) -> Hash (cost=l.O5..1.05 rows=S width=9) (actual time=1.16..l.l6 rows=0 loops=1) -> Seq Scan on worms (cost=0.00..1.05 rows=S width=9) (actual time=0.65..1.00 rows=5 loops=1) -> Hash (cost=1.16..1.16 rows=16 width=9) (actual time=l.86..1.86 rows=0 loops=1) -> Seq Scan on lions (cost=0.00..l.16 rows=16 width=9) (actual time=0.lO..1.36 rows=16 loops=1) -> Hash (cost=1.06..1.06 rows=6 width=17) (actual time=1.35..1.35 rows=0 loops=1) -> Seq Scan on dogs (cost=0.00..1.06 rows=6 width=17) (actual time=0.65..1.16 rows=6 loops=l) -> Hash (cost=1.07..1.07 rows=3 width=9) (actual time=1.23..1.23 rows=0 loops=1) -> Seq Scan on parrots (cost=0.00..1.07 rows=3 width=9) (actual time=0.69..1.13 rows=3 loops=1) -> Hash (cost=l.08..1.08 rows=8 width=17) (actual time=0.98..0.98 rows=0 loops=1) -> Seq Scan on rhinos (cost=0.00..1.08 rows=8 width=17) (actual time=0.10..0.73 rows=8 loops=1) Total runtime: 58l341.00 msec From pgsql-performance-owner@postgresql.org Fri Dec 12 05:14:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A25E7D1DE82 for ; Fri, 12 Dec 2003 09:14:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 91701-08 for ; Fri, 12 Dec 2003 05:14:13 -0400 (AST) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id 59238D1B554 for ; Fri, 12 Dec 2003 05:14:10 -0400 (AST) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id hBC9EHe3010031 for ; Fri, 12 Dec 2003 14:44:17 +0530 Received: from persistent.co.in (daithan.intranet.pspl.co.in [192.168.7.161]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id hBC9EEAo009969; Fri, 12 Dec 2003 14:44:15 +0530 Message-ID: <3FD986DD.7000909@persistent.co.in> Date: Fri, 12 Dec 2003 14:44:05 +0530 From: Shridhar Daithankar Organization: Persistent Systems Pvt. Ltd. User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "david@shadovitz.com" Cc: "'pgsql-performance@postgresql.org'" Subject: Re: Query plan - now what? References: <01C3C045.7EDD13C0.david@shadovitz.com> In-Reply-To: <01C3C045.7EDD13C0.david@shadovitz.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/195 X-Sequence-Number: 5055 David Shadovitz wrote: > Well, now that I have the plan for my slow-running query, what do I do? Where > should I focus my attention? Briefly looking over the plan and seeing the estimated v/s actual row mismatch,I can suggest you following. 1. Vacuum(full) the database. Probably you have already done it. 2. Raise statistics_target to 500 or more and reanalyze the table(s) in question. 3. Set enable_hash_join to false, before running the query and see if it helps. HTH Shridhar From pgsql-performance-owner@postgresql.org Fri Dec 12 06:49:12 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CC7D8D1B456 for ; Fri, 12 Dec 2003 10:49:10 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 02299-07 for ; Fri, 12 Dec 2003 06:48:41 -0400 (AST) Received: from smtp02do.de.uu.net (smtp02do.de.uu.net [192.76.144.69]) by svr1.postgresql.org (Postfix) with ESMTP id 27958D1B442 for ; Fri, 12 Dec 2003 06:48:37 -0400 (AST) Received: from bommel.kecam-han.de ([193.99.158.1]) by smtp02do.de.uu.net (8.9.3p2/5.5.5) with ESMTP id LAA07894 for ; Fri, 12 Dec 2003 11:48:38 +0100 (MET) Received: from mailrelay.kecam-han.de (ldap [10.9.1.54]) by bommel.kecam-han.de (8.11.6+Sun/8.9.1) with ESMTP id hBCAmx616545 for ; Fri, 12 Dec 2003 11:48:59 +0100 (MET) Received: from ke-elektronik.de (localhost [127.0.0.1]) by mailrelay.kecam-han.de (8.12.2/8.12.2) with ESMTP id hBCAltUr012470 for ; Fri, 12 Dec 2003 11:48:15 +0100 (MET) Message-ID: <3FD99CDA.8161D80A@ke-elektronik.de> Date: Fri, 12 Dec 2003 11:47:54 +0100 From: Hartmut Raschick Organization: ke Kommunikations-Elektronik GmbH X-Mailer: Mozilla 4.79 [en] (X11; U; SunOS 5.6 sun4u) X-Accept-Language: de, ru, en MIME-Version: 1.0 To: psqlperformace Subject: Re: TRUNCATE veeeery slow compared to DELETE in 7.4 References: <3FD484B6.EF39A92F@ke-elektronik.de> <200312100918.02501.josh@agliodbs.com> <8000.1071086054@sss.pgh.pa.us> Content-Type: multipart/mixed; boundary="------------59882B14BEB9053A486EE494" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/196 X-Sequence-Number: 5056 This is a multi-part message in MIME format. --------------59882B14BEB9053A486EE494 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable for the clearer understanding: this is NOT about TRUNCATE being slow "as such" vs. DELETE, but about a change in the order of a (...) magnitude from 7.3.4 to 7.4... here's some more info, plus test results w/a "full" db: 300 tables, 20000 pieces of modelled hw, so there's one table w/20000 entries, each model has a special table (per type), too; so, entries over all of them sum up to 20000; not all types are present. plus: some types (w/not many instances) have "very special" tables, too, these sometimes w/lots of columns 500-1600... alone on a sun fire-280 w/2 U-IIIi cpu's (well, only need one...): all the time of the test, no vacuum anything was performed, thus - by the book - making things worse... for the DELETE case. 7.4: ---- "full" database: TRUNCATE: 0.03u 0.03s 1:21.40 0.0% DELETE: 0.05u 0.01s 0:04.46 1.3% empty database: TRUNCATE:0.02u 0.05s 1:21.00 0.0% DELETE: 0.04u 0.04s 0:01.32 6.0% now for 7.3.4 database server (same machine, of cause): -------------- "full" database: TRUNCATE: 0.04u 0.04s 0:03.79 2.1% DELETE: 0.03u 0.03s 0:06.51 0.9% empty database: TRUNCATE: 0.04u 0.05s 0:01.51 5.9% DELETE: 0.01u 0.02s 0:01.00 3.0% what can i say... ...please find the attached configs. i reeeeally don't think this can be explained by table/index complexity, it's the _same_ schema and contents for both cases, they both were started w/createdb, they both were filled the same way (by our server prog), there was no vacuum nowhere, test execution order was the same in both cases. P.S.: Mon pessimisme va jusqu'=E0 suspecter la sinc=E9rit=E9 des pessimiste= s. - Jean Rostand (1894-1977), Journal d'un caract=E8re, 1931 --=20 Hartmut "Hardy" Raschick / Dept. t2 ke Kommunikations-Elektronik GmbH Wohlenberstr. 3, 30179 Hannover Phone: ++49 (0)511 6747-564 Fax: ++49 (0)511 6747-340 e-Mail: hartmut.raschick@ke-elektronik.de http://www.ke-elektronik.de --------------59882B14BEB9053A486EE494 Content-Type: text/plain; charset=us-ascii; name="postgresql.conf-7.3.4" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="postgresql.conf-7.3.4" # # PostgreSQL configuration file # ----------------------------- # # This file consists of lines of the form: # # name = value # # (The '=' is optional.) White space may be used. Comments are introduced # with '#' anywhere on a line. The complete list of option names and # allowed values can be found in the PostgreSQL documentation. The # commented-out settings shown in this file represent the default values. # # Any option can also be given as a command line switch to the # postmaster, e.g. 'postmaster -c log_connections=on'. Some options # can be changed at run-time with the 'SET' SQL command. # # This file is read on postmaster startup and when the postmaster # receives a SIGHUP. If you edit the file on a running system, you have # to SIGHUP the postmaster for the changes to take effect, or use # "pg_ctl reload". #======================================================================== # # Connection Parameters # #tcpip_socket = false tcpip_socket = true #ssl = false #max_connections = 32 max_connections = 128 #superuser_reserved_connections = 2 #port = 5432 port = 5433 #hostname_lookup = false #show_source_port = false #unix_socket_directory = '' #unix_socket_group = '' #unix_socket_permissions = 0777 # octal #virtual_host = '' #krb_server_keyfile = '' # # Shared Memory Size # #shared_buffers = 64 # min max_connections*2 or 16, 8KB each shared_buffers = 256 # 2*max_connections, min 16, typically 8KB each #max_fsm_relations = 1000 # min 10, fsm is free space map, ~40 bytes #max_fsm_pages = 10000 # min 1000, fsm is free space map, ~6 bytes max_fsm_pages = 50000 # min 1000, fsm is free space map, ~6 bytes #max_locks_per_transaction = 64 # min 10 #wal_buffers = 8 # min 4, typically 8KB each wal_buffers = 40 # min 4, typically 8KB each # # Non-shared Memory Sizes # #sort_mem = 1024 # min 64, size in KB sort_mem = 16384 # min 64, size in KB #vacuum_mem = 8192 # min 1024, size in KB vacuum_mem = 16384 # min 1024, size in KB # # Write-ahead log (WAL) # #checkpoint_segments = 3 # in logfile segments, min 1, 16MB each checkpoint_segments = 20 # in logfile segments, min 1, 16MB each #checkpoint_timeout = 300 # range 30-3600, in seconds checkpoint_timeout = 600 # range 30-3600, in seconds # #commit_delay = 0 # range 0-100000, in microseconds commit_delay = 100000 # range 0-100000, in microseconds #commit_siblings = 5 # range 1-1000 commit_siblings = 100 # range 1-1000 # #fsync = true fsync = false #wal_sync_method = fsync # the default varies across platforms: wal_sync_method = fdatasync # the default varies across platforms: # # fsync, fdatasync, open_sync, or open_datasync #wal_debug = 0 # range 0-16 # # Optimizer Parameters # #enable_seqscan = true enable_seqscan = false #enable_indexscan = true #enable_tidscan = true #enable_sort = true #enable_nestloop = true #enable_mergejoin = true #enable_hashjoin = true #effective_cache_size = 1000 # typically 8KB each #random_page_cost = 4 # units are one sequential page fetch cost #cpu_tuple_cost = 0.01 # (same) #cpu_index_tuple_cost = 0.001 # (same) #cpu_operator_cost = 0.0025 # (same) #default_statistics_target = 10 # range 1-1000 # # GEQO Optimizer Parameters # #geqo = true #geqo_selection_bias = 2.0 # range 1.5-2.0 #geqo_threshold = 11 #geqo_pool_size = 0 # default based on tables in statement, # range 128-1024 #geqo_effort = 1 #geqo_generations = 0 #geqo_random_seed = -1 # auto-compute seed # # Message display # #server_min_messages = notice # Values, in order of decreasing detail: # debug5, debug4, debug3, debug2, debug1, # info, notice, warning, error, log, fatal, # panic #client_min_messages = notice # Values, in order of decreasing detail: # debug5, debug4, debug3, debug2, debug1, # log, info, notice, warning, error #silent_mode = false #log_connections = false log_connections = true #log_pid = false #log_statement = false #log_duration = false #log_timestamp = false #log_min_error_statement = error # Values in order of increasing severity: # debug5, debug4, debug3, debug2, debug1, # info, notice, warning, error, panic(off) #debug_print_parse = false #debug_print_rewritten = false #debug_print_plan = false #debug_pretty_print = false #explain_pretty_print = true # requires USE_ASSERT_CHECKING #debug_assertions = true # # Syslog # #syslog = 0 # range 0-2 #syslog_facility = 'LOCAL0' #syslog_ident = 'postgres' # # Statistics # #show_parser_stats = false #show_planner_stats = false #show_executor_stats = false #show_statement_stats = false # requires BTREE_BUILD_STATS #show_btree_build_stats = false # # Access statistics collection # #stats_start_collector = true #stats_reset_on_server_start = true #stats_command_string = false #stats_row_level = false #stats_block_level = false # # Lock Tracing # #trace_notify = false # requires LOCK_DEBUG #trace_locks = false #trace_userlocks = false #trace_lwlocks = false #debug_deadlocks = false #trace_lock_oidmin = 16384 #trace_lock_table = 0 # # Misc # #autocommit = true #dynamic_library_path = '$libdir' #search_path = '$user,public' #datestyle = 'iso, us' #timezone = unknown # actually, defaults to TZ environment setting #australian_timezones = false #client_encoding = sql_ascii # actually, defaults to database encoding #authentication_timeout = 60 # 1-600, in seconds #deadlock_timeout = 1000 # in milliseconds #default_transaction_isolation = 'read committed' #max_expr_depth = 10000 # min 10 #max_files_per_process = 1000 # min 25 #password_encryption = true #sql_inheritance = true #transform_null_equals = false #statement_timeout = 0 # 0 is disabled, in milliseconds #db_user_namespace = false # # Locale settings # # (initialized by initdb -- may be changed) LC_MESSAGES = 'C' LC_MONETARY = 'en_US.ISO8859-15' LC_NUMERIC = 'en_US.ISO8859-15' LC_TIME = 'en_US.ISO8859-15' --------------59882B14BEB9053A486EE494 Content-Type: text/plain; charset=us-ascii; name="postgresql.conf-7.4" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="postgresql.conf-7.4" # ----------------------------- # PostgreSQL configuration file # ----------------------------- # # This file consists of lines of the form: # # name = value # # (The '=' is optional.) White space may be used. Comments are introduced # with '#' anywhere on a line. The complete list of option names and # allowed values can be found in the PostgreSQL documentation. The # commented-out settings shown in this file represent the default values. # # Any option can also be given as a command line switch to the # postmaster, e.g. 'postmaster -c log_connections=on'. Some options # can be changed at run-time with the 'SET' SQL command. # # This file is read on postmaster startup and when the postmaster # receives a SIGHUP. If you edit the file on a running system, you have # to SIGHUP the postmaster for the changes to take effect, or use # "pg_ctl reload". #--------------------------------------------------------------------------- # CONNECTIONS AND AUTHENTICATION #--------------------------------------------------------------------------- # - Connection Settings - #tcpip_socket = false tcpip_socket = true max_connections = 128 # note: increasing max_connections costs about 500 bytes of shared # memory per connection slot, in addition to costs from shared_buffers # and max_locks_per_transaction. #superuser_reserved_connections = 2 #port = 5432 port = 5474 #unix_socket_directory = '' #unix_socket_group = '' #unix_socket_permissions = 0777 # octal #virtual_host = '' # what interface to listen on; defaults to any #rendezvous_name = '' # defaults to the computer name # - Security & Authentication - #authentication_timeout = 60 # 1-600, in seconds #ssl = false #password_encryption = true #krb_server_keyfile = '' #db_user_namespace = false #--------------------------------------------------------------------------- # RESOURCE USAGE (except WAL) #--------------------------------------------------------------------------- # - Memory - shared_buffers = 1024 # min 16, at least max_connections*2, 8KB each #sort_mem = 1024 # min 64, size in KB sort_mem = 16384 # min 64, size in KB #vacuum_mem = 8192 # min 1024, size in KB vacuum_mem = 16384 # min 1024, size in KB # - Free Space Map - #max_fsm_pages = 20000 # min max_fsm_relations*16, 6 bytes each max_fsm_pages = 50000 # min max_fsm_relations*16, 6 bytes each #max_fsm_relations = 1000 # min 100, ~50 bytes each # - Kernel Resource Usage - #max_files_per_process = 1000 # min 25 #preload_libraries = '' #--------------------------------------------------------------------------- # WRITE AHEAD LOG #--------------------------------------------------------------------------- # - Settings - #fsync = true # turns forced synchronization on or off fsync = false # turns forced synchronization on or off #wal_sync_method = fsync # the default varies across platforms: wal_sync_method = fdatasync # the default varies across platforms: # fsync, fdatasync, open_sync, or open_datasync #wal_buffers = 8 # min 4, 8KB each wal_buffers = 40 # min 4, 8KB each # - Checkpoints - #checkpoint_segments = 3 # in logfile segments, min 1, 16MB each checkpoint_segments = 20 # in logfile segments, min 1, 16MB each #checkpoint_timeout = 300 # range 30-3600, in seconds checkpoint_timeout = 600 # range 30-3600, in seconds #checkpoint_warning = 30 # 0 is off, in seconds #commit_delay = 0 # range 0-100000, in microseconds commit_delay = 100000 # range 0-100000, in microseconds #commit_siblings = 5 # range 1-1000 commit_siblings = 100 # range 1-1000 #--------------------------------------------------------------------------- # QUERY TUNING #--------------------------------------------------------------------------- # - Planner Method Enabling - #enable_hashagg = true #enable_hashjoin = true #enable_indexscan = true #enable_mergejoin = true #enable_nestloop = true #enable_seqscan = true enable_seqscan = false #enable_sort = true #enable_tidscan = true # - Planner Cost Constants - #effective_cache_size = 1000 # typically 8KB each #random_page_cost = 4 # units are one sequential page fetch cost #cpu_tuple_cost = 0.01 # (same) #cpu_index_tuple_cost = 0.001 # (same) #cpu_operator_cost = 0.0025 # (same) # - Genetic Query Optimizer - #geqo = true #geqo_threshold = 11 #geqo_effort = 1 #geqo_generations = 0 #geqo_pool_size = 0 # default based on tables in statement, # range 128-1024 #geqo_selection_bias = 2.0 # range 1.5-2.0 # - Other Planner Options - #default_statistics_target = 10 # range 1-1000 #from_collapse_limit = 8 #join_collapse_limit = 8 # 1 disables collapsing of explicit JOINs #--------------------------------------------------------------------------- # ERROR REPORTING AND LOGGING #--------------------------------------------------------------------------- # - Syslog - #syslog = 0 # range 0-2; 0=stdout; 1=both; 2=syslog #syslog_facility = 'LOCAL0' #syslog_ident = 'postgres' # - When to Log - #client_min_messages = notice # Values, in order of decreasing detail: # debug5, debug4, debug3, debug2, debug1, # log, info, notice, warning, error #log_min_messages = notice # Values, in order of decreasing detail: # debug5, debug4, debug3, debug2, debug1, # info, notice, warning, error, log, fatal, # panic #log_error_verbosity = default # terse, default, or verbose messages #log_min_error_statement = panic # Values in order of increasing severity: # debug5, debug4, debug3, debug2, debug1, # info, notice, warning, error, panic(off) #log_min_duration_statement = -1 # Log all statements whose # execution time exceeds the value, in # milliseconds. Zero prints all queries. # Minus-one disables. #silent_mode = false # DO NOT USE without Syslog! # - What to Log - #debug_print_parse = false #debug_print_rewritten = false #debug_print_plan = false #debug_pretty_print = false #log_connections = false log_connections = true #log_duration = false #log_pid = false #log_statement = false #log_timestamp = false #log_hostname = false #log_source_port = false #--------------------------------------------------------------------------- # RUNTIME STATISTICS #--------------------------------------------------------------------------- # - Statistics Monitoring - #log_parser_stats = false #log_planner_stats = false #log_executor_stats = false #log_statement_stats = false # - Query/Index Statistics Collector - #stats_start_collector = true #stats_command_string = false #stats_block_level = false #stats_row_level = false #stats_reset_on_server_start = true #--------------------------------------------------------------------------- # CLIENT CONNECTION DEFAULTS #--------------------------------------------------------------------------- # - Statement Behavior - #search_path = '$user,public' # schema names #check_function_bodies = true #default_transaction_isolation = 'read committed' #default_transaction_read_only = false #statement_timeout = 0 # 0 is disabled, in milliseconds # - Locale and Formatting - #datestyle = 'iso, mdy' #timezone = unknown # actually, defaults to TZ environment setting #australian_timezones = false #extra_float_digits = 0 # min -15, max 2 #client_encoding = sql_ascii # actually, defaults to database encoding # These settings are initialized by initdb -- they may be changed lc_messages = 'C' # locale for system error message strings lc_monetary = 'en_US.ISO8859-15' # locale for monetary formatting lc_numeric = 'en_US.ISO8859-15' # locale for number formatting lc_time = 'en_US.ISO8859-15' # locale for time formatting # - Other Defaults - #explain_pretty_print = true #dynamic_library_path = '$libdir' #max_expr_depth = 10000 # min 10 #--------------------------------------------------------------------------- # LOCK MANAGEMENT #--------------------------------------------------------------------------- #deadlock_timeout = 1000 # in milliseconds #max_locks_per_transaction = 64 # min 10, ~260*max_connections bytes each #--------------------------------------------------------------------------- # VERSION/PLATFORM COMPATIBILITY #--------------------------------------------------------------------------- # - Previous Postgres Versions - #add_missing_from = true #regex_flavor = advanced # advanced, extended, or basic #sql_inheritance = true # - Other Platforms & Clients - #transform_null_equals = false --------------59882B14BEB9053A486EE494-- From pgsql-performance-owner@postgresql.org Tue Dec 16 17:15:35 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D2FD8D1B45C; Fri, 12 Dec 2003 12:22:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08050-05; Fri, 12 Dec 2003 08:22:13 -0400 (AST) Received: from smxsat1.smxs.net (smxsat1.smxs.net [213.150.10.1]) by svr1.postgresql.org (Postfix) with ESMTP id DD1F7D1B454; Fri, 12 Dec 2003 08:22:07 -0400 (AST) Received: from m01x1.s-mxs.net [10.3.55.201] by smxsat1.smxs.net over TLS secured channel with XWall v3.28f ; Fri, 12 Dec 2003 13:22:03 +0100 Received: from m0102.s-mxs.net [10.3.55.2] by m01x1.s-mxs.net with XWall v3.28f ; Fri, 12 Dec 2003 13:22:01 +0100 Received: from m0114.s-mxs.net ([10.3.55.14]) by m0102.s-mxs.net with Microsoft SMTPSVC(5.0.2195.5329); Fri, 12 Dec 2003 13:22:01 +0100 X-MimeOLE: Produced By Microsoft Exchange V6.0.6503.0 content-class: urn:content-classes:message Subject: Re: [HACKERS] fsync method checking MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----_=_NextPart_001_01C3C0AA.8B1026A5" Date: Fri, 12 Dec 2003 13:22:00 +0100 Message-ID: <46C15C39FEB2C44BA555E356FBCD6FA40184CF68@m0114.s-mxs.net> X-MS-Has-Attach: yes X-MS-TNEF-Correlator: Thread-Topic: [HACKERS] fsync method checking Thread-Index: AcPAfH+RrTtfj1bcTvGhyP/wCNch4gAJ4azQ From: "Zeugswetter Andreas SB SD" To: "Bruce Momjian" , "Mark Kirkwood" Cc: , "PostgreSQL-development" X-OriginalArrivalTime: 12 Dec 2003 12:22:01.0144 (UTC) FILETIME=[8B5A7380:01C3C0AA] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/238 X-Sequence-Number: 5098 This is a multi-part message in MIME format. ------_=_NextPart_001_01C3C0AA.8B1026A5 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable > Running the attached test program shows on BSD/OS 4.3: >=20 > write 0.000360 > write & fsync 0.001391 I think the "write & fsync" pays for the previous "write" test (same filena= me). > write, close & fsync 0.001308 > open o_fsync, write 0.000924 I have tried to modify the program to more closely resemble WAL=20 writes (all writes to WAL are 8k), the file is usually already open,=20 and test larger (16k) transactions. zeu@a82101002:~> test_sync1 write 0.000625 write & fsync 0.016748 write & fdatasync 0.006650 write, close & fsync 0.017084 write, close & fdatasync 0.006890 open o_dsync, write 0.015997 open o_dsync, one write 0.007128 For the last line xlog.c would need to be modified, but the measurements seem to imply that it is only worth it on platforms that have O_DSYNC but not fdatasync.=20=20 Andreas ------_=_NextPart_001_01C3C0AA.8B1026A5 Content-Type: application/octet-stream; name="test_sync1.c" Content-Transfer-Encoding: base64 Content-Description: test_sync1.c Content-Disposition: attachment; filename="test_sync1.c" LyoNCiAqCXRlc3RfZnN5bmMuYw0KICoJCXRlc3RzIGlmIGZzeW5jIGNhbiBi ZSBkb25lIGZyb20gYW5vdGhlciBwcm9jZXNzIHRoYW4gdGhlIG9yaWdpbmFs IHdyaXRlDQogKi8NCg0KI2luY2x1ZGUgPHN5cy90eXBlcy5oPg0KI2luY2x1 ZGUgPGZjbnRsLmg+DQojaW5jbHVkZSA8c3RkaW8uaD4NCiNpbmNsdWRlIDxz dGRsaWIuaD4NCiNpbmNsdWRlIDx0aW1lLmg+DQojaW5jbHVkZSA8dW5pc3Rk Lmg+DQoNCnZvaWQgZGllKGNoYXIgKnN0cik7DQp2b2lkIHByaW50X2VsYXBz ZShzdHJ1Y3QgdGltZXZhbCBzdGFydF90LCBzdHJ1Y3QgdGltZXZhbCBlbGFw c2VfdCk7DQoNCmludCBtYWluKGludCBhcmdjLCBjaGFyICphcmd2W10pDQp7 DQoJc3RydWN0IHRpbWV2YWwgc3RhcnRfdDsNCglzdHJ1Y3QgdGltZXZhbCBl bGFwc2VfdDsNCglpbnQgdG1wZmlsZSwgaTsNCgljaGFyICpzdHJvdXQgPSAo Y2hhciAqKSBtYWxsb2MoNjU1MzYpOyAgDQoNCglmb3IgKGk9MDsgaTw2NTUz NjsgaSsrKQ0KCQlzdHJvdXRbaV0gPSAnYSc7DQoNCglpZiAoKHRtcGZpbGUg PSBvcGVuKCIvdmFyL3RtcC90ZXN0X2ZzeW5jLm91dCIsIE9fUkRXUiB8IE9f Q1JFQVQsIFNfSVJVU1IgfCBTX0lXVVNSKSkgPT0gLTEpDQoJCWRpZSgiY2Fu J3Qgb3BlbiAvdmFyL3RtcC90ZXN0X2ZzeW5jLm91dCIpOw0KCXdyaXRlKHRt cGZpbGUsIHN0cm91dCwgNjU1MzYpOw0KCWZzeW5jKHRtcGZpbGUpOw0KCWNs b3NlKHRtcGZpbGUpOwkJDQoNCgkvKiB3cml0ZSBvbmx5ICovCQ0KCWlmICgo dG1wZmlsZSA9IG9wZW4oIi92YXIvdG1wL3Rlc3RfZnN5bmMub3V0IiwgT19S RFdSKSkgPT0gLTEpDQoJCWRpZSgiY2FuJ3Qgb3BlbiAvdmFyL3RtcC90ZXN0 X2ZzeW5jLm91dCIpOw0KCWdldHRpbWVvZmRheSgmc3RhcnRfdCwgTlVMTCk7 DQoJd3JpdGUodG1wZmlsZSwgc3Ryb3V0LCA4MTkyKTsNCgl3cml0ZSh0bXBm aWxlLCBzdHJvdXQsIDgxOTIpOw0KCWdldHRpbWVvZmRheSgmZWxhcHNlX3Qs IE5VTEwpOw0KCWNsb3NlKHRtcGZpbGUpOwkJDQoJcHJpbnRmKCJ3cml0ZSAg ICAgICAgICAgICAgICAgICAgICAiKTsNCglwcmludF9lbGFwc2Uoc3RhcnRf dCwgZWxhcHNlX3QpOw0KCXByaW50ZigiXG4iKTsNCg0KCWlmICgodG1wZmls ZSA9IG9wZW4oIi92YXIvdG1wL3Rlc3RfZnN5bmMub3V0IiwgT19SRFdSKSkg PT0gLTEpDQoJCWRpZSgiY2FuJ3Qgb3BlbiAvdmFyL3RtcC90ZXN0X2ZzeW5j Lm91dCIpOw0KCWZzeW5jKHRtcGZpbGUpOw0KCWNsb3NlKHRtcGZpbGUpOwkJ DQoNCgkvKiB3cml0ZSAmIGZzeW5jICovDQoJaWYgKCh0bXBmaWxlID0gb3Bl bigiL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiLCBPX1JEV1IpKSA9PSAtMSkN CgkJZGllKCJjYW4ndCBvcGVuIC92YXIvdG1wL3Rlc3RfZnN5bmMub3V0Iik7 DQoJZ2V0dGltZW9mZGF5KCZzdGFydF90LCBOVUxMKTsNCgl3cml0ZSh0bXBm aWxlLCBzdHJvdXQsIDgxOTIpOw0KCXdyaXRlKHRtcGZpbGUsIHN0cm91dCwg ODE5Mik7DQoJZnN5bmModG1wZmlsZSk7DQoJZ2V0dGltZW9mZGF5KCZlbGFw c2VfdCwgTlVMTCk7DQoJY2xvc2UodG1wZmlsZSk7CQkNCglwcmludGYoIndy aXRlICYgZnN5bmMgICAgICAgICAgICAgICIpOw0KCXByaW50X2VsYXBzZShz dGFydF90LCBlbGFwc2VfdCk7DQoJcHJpbnRmKCJcbiIpOw0KDQoJLyogd3Jp dGUgJiBmZGF0YXN5bmMgKi8NCglpZiAoKHRtcGZpbGUgPSBvcGVuKCIvdmFy L3RtcC90ZXN0X2ZzeW5jLm91dCIsIE9fUkRXUikpID09IC0xKQ0KCQlkaWUo ImNhbid0IG9wZW4gL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiKTsNCglnZXR0 aW1lb2ZkYXkoJnN0YXJ0X3QsIE5VTEwpOw0KCXdyaXRlKHRtcGZpbGUsIHN0 cm91dCwgODE5Mik7DQoJd3JpdGUodG1wZmlsZSwgc3Ryb3V0LCA4MTkyKTsN CglmZGF0YXN5bmModG1wZmlsZSk7DQoJZ2V0dGltZW9mZGF5KCZlbGFwc2Vf dCwgTlVMTCk7DQoJY2xvc2UodG1wZmlsZSk7CQkNCglwcmludGYoIndyaXRl ICYgZmRhdGFzeW5jICAgICAgICAgICIpOw0KCXByaW50X2VsYXBzZShzdGFy dF90LCBlbGFwc2VfdCk7DQoJcHJpbnRmKCJcbiIpOw0KDQoJaWYgKCh0bXBm aWxlID0gb3BlbigiL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiLCBPX1JEV1Ip KSA9PSAtMSkNCgkJZGllKCJjYW4ndCBvcGVuIC92YXIvdG1wL3Rlc3RfZnN5 bmMub3V0Iik7DQoJZnN5bmModG1wZmlsZSk7DQoJY2xvc2UodG1wZmlsZSk7 CQkNCg0KCS8qIHdyaXRlLCBjbG9zZSAmIGZzeW5jICovDQoJaWYgKCh0bXBm aWxlID0gb3BlbigiL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiLCBPX1JEV1Ip KSA9PSAtMSkNCgkJZGllKCJjYW4ndCBvcGVuIDEgL3Zhci90bXAvdGVzdF9m c3luYy5vdXQiKTsNCglnZXR0aW1lb2ZkYXkoJnN0YXJ0X3QsIE5VTEwpOw0K CXdyaXRlKHRtcGZpbGUsIHN0cm91dCwgODE5Mik7DQoJd3JpdGUodG1wZmls ZSwgc3Ryb3V0LCA4MTkyKTsNCgljbG9zZSh0bXBmaWxlKTsNCgkvKiByZW9w ZW4gZmlsZSAqLw0KCWlmICgodG1wZmlsZSA9IG9wZW4oIi92YXIvdG1wL3Rl c3RfZnN5bmMub3V0IiwgT19SRFdSKSkgPT0gLTEpDQoJCWRpZSgiY2FuJ3Qg b3BlbiAyIC92YXIvdG1wL3Rlc3RfZnN5bmMub3V0Iik7DQoJZnN5bmModG1w ZmlsZSk7DQoJY2xvc2UodG1wZmlsZSk7CQkNCglnZXR0aW1lb2ZkYXkoJmVs YXBzZV90LCBOVUxMKTsNCglwcmludGYoIndyaXRlLCBjbG9zZSAmIGZzeW5j ICAgICAgICIpOw0KCXByaW50X2VsYXBzZShzdGFydF90LCBlbGFwc2VfdCk7 DQoJcHJpbnRmKCJcbiIpOw0KDQoJLyogd3JpdGUsIGNsb3NlICYgZmRhdGFz eW5jICovDQoJaWYgKCh0bXBmaWxlID0gb3BlbigiL3Zhci90bXAvdGVzdF9m c3luYy5vdXQiLCBPX1JEV1IpKSA9PSAtMSkNCgkJZGllKCJjYW4ndCBvcGVu IDEgL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiKTsNCglnZXR0aW1lb2ZkYXko JnN0YXJ0X3QsIE5VTEwpOw0KCXdyaXRlKHRtcGZpbGUsIHN0cm91dCwgODE5 Mik7DQoJd3JpdGUodG1wZmlsZSwgc3Ryb3V0LCA4MTkyKTsNCgljbG9zZSh0 bXBmaWxlKTsNCgkvKiByZW9wZW4gZmlsZSAqLw0KCWlmICgodG1wZmlsZSA9 IG9wZW4oIi92YXIvdG1wL3Rlc3RfZnN5bmMub3V0IiwgT19SRFdSKSkgPT0g LTEpDQoJCWRpZSgiY2FuJ3Qgb3BlbiAyIC92YXIvdG1wL3Rlc3RfZnN5bmMu b3V0Iik7DQoJZmRhdGFzeW5jKHRtcGZpbGUpOw0KCWdldHRpbWVvZmRheSgm ZWxhcHNlX3QsIE5VTEwpOw0KCWNsb3NlKHRtcGZpbGUpOwkJDQoJcHJpbnRm KCJ3cml0ZSwgY2xvc2UgJiBmZGF0YXN5bmMgICAiKTsNCglwcmludF9lbGFw c2Uoc3RhcnRfdCwgZWxhcHNlX3QpOw0KCXByaW50ZigiXG4iKTsNCg0KCWlm ICgodG1wZmlsZSA9IG9wZW4oIi92YXIvdG1wL3Rlc3RfZnN5bmMub3V0Iiwg T19SRFdSKSkgPT0gLTEpDQoJCWRpZSgiY2FuJ3Qgb3BlbiAvdmFyL3RtcC90 ZXN0X2ZzeW5jLm91dCIpOw0KCWZzeW5jKHRtcGZpbGUpOw0KCWNsb3NlKHRt cGZpbGUpOwkJDQoNCgkvKiBvcGVuX2RzeW5jLCB3cml0ZSAqLw0KCWlmICgo dG1wZmlsZSA9IG9wZW4oIi92YXIvdG1wL3Rlc3RfZnN5bmMub3V0IiwgT19S RFdSIHwgT19EU1lOQykpID09IC0xKQ0KCQlkaWUoImNhbid0IG9wZW4gL3Zh ci90bXAvdGVzdF9mc3luYy5vdXQiKTsNCglnZXR0aW1lb2ZkYXkoJnN0YXJ0 X3QsIE5VTEwpOw0KCXdyaXRlKHRtcGZpbGUsIHN0cm91dCwgODE5Mik7DQoJ d3JpdGUodG1wZmlsZSwgc3Ryb3V0LCA4MTkyKTsNCglnZXR0aW1lb2ZkYXko JmVsYXBzZV90LCBOVUxMKTsNCgljbG9zZSh0bXBmaWxlKTsNCglwcmludGYo Im9wZW4gb19kc3luYywgd3JpdGUgICAgICAgICIpOw0KCXByaW50X2VsYXBz ZShzdGFydF90LCBlbGFwc2VfdCk7DQoJcHJpbnRmKCJcbiIpOw0KDQoJaWYg KCh0bXBmaWxlID0gb3BlbigiL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiLCBP X1JEV1IpKSA9PSAtMSkNCgkJZGllKCJjYW4ndCBvcGVuIC92YXIvdG1wL3Rl c3RfZnN5bmMub3V0Iik7DQoJZnN5bmModG1wZmlsZSk7DQoJY2xvc2UodG1w ZmlsZSk7CQkNCg0KCS8qIG9wZW5fZHN5bmMsIHdyaXRlICovDQoJaWYgKCh0 bXBmaWxlID0gb3BlbigiL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiLCBPX1JE V1IgfCBPX0RTWU5DKSkgPT0gLTEpDQoJCWRpZSgiY2FuJ3Qgb3BlbiAvdmFy L3RtcC90ZXN0X2ZzeW5jLm91dCIpOw0KCWdldHRpbWVvZmRheSgmc3RhcnRf dCwgTlVMTCk7DQoJd3JpdGUodG1wZmlsZSwgc3Ryb3V0LCAxNjM4NCk7DQoJ Z2V0dGltZW9mZGF5KCZlbGFwc2VfdCwgTlVMTCk7DQoJY2xvc2UodG1wZmls ZSk7DQoJcHJpbnRmKCJvcGVuIG9fZHN5bmMsIG9uZSB3cml0ZSAgICAiKTsN CglwcmludF9lbGFwc2Uoc3RhcnRfdCwgZWxhcHNlX3QpOw0KCXByaW50Zigi XG4iKTsNCg0KCXVubGluaygiL3Zhci90bXAvdGVzdF9mc3luYy5vdXQiKTsN Cg0KCXJldHVybiAwOw0KfQ0KDQp2b2lkIHByaW50X2VsYXBzZShzdHJ1Y3Qg dGltZXZhbCBzdGFydF90LCBzdHJ1Y3QgdGltZXZhbCBlbGFwc2VfdCkNCnsN CglpZiAoZWxhcHNlX3QudHZfdXNlYyA8IHN0YXJ0X3QudHZfdXNlYykNCgl7 DQoJCWVsYXBzZV90LnR2X3NlYy0tOw0KCQllbGFwc2VfdC50dl91c2VjICs9 IDEwMDAwMDA7DQoJfQ0KDQoJcHJpbnRmKCIlbGQuJTA2bGQiLCAobG9uZykg KGVsYXBzZV90LnR2X3NlYyAtIHN0YXJ0X3QudHZfc2VjKSwNCgkJCQkJIChs b25nKSAoZWxhcHNlX3QudHZfdXNlYyAtIHN0YXJ0X3QudHZfdXNlYykpOw0K fQ0KDQp2b2lkIGRpZShjaGFyICpzdHIpDQp7DQoJZnByaW50ZihzdGRlcnIs ICIlcyIsIHN0cik7DQoJZXhpdCgxKTsNCn0NCg== ------_=_NextPart_001_01C3C0AA.8B1026A5-- From pgsql-admin-owner@postgresql.org Fri Dec 12 08:40:45 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 12CB4D1B52A; Fri, 12 Dec 2003 12:40:36 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19260-02; Fri, 12 Dec 2003 08:40:07 -0400 (AST) Received: from host2.hostseguro.com (66-98-192-97.hostseguro.com [66.98.192.97]) by svr1.postgresql.org (Postfix) with ESMTP id 1A18FD1B459; Fri, 12 Dec 2003 08:40:04 -0400 (AST) Received: from cpanel by host2.hostseguro.com with local (Exim 4.24) id 1AUmaN-0002LT-9n; Fri, 12 Dec 2003 10:39:59 -0200 Received: from 200-180-166-037.paemt7003.dsl.brasiltelecom.net.br (200-180-166-037.paemt7003.dsl.brasiltelecom.net.br [200.180.166.37]) by sistemica.info (IMP) with HTTP for ; Fri, 12 Dec 2003 10:39:59 -0200 Message-ID: <1071232799.3fd9b71f3d0bb@sistemica.info> Date: Fri, 12 Dec 2003 10:39:59 -0200 From: Rhaoni Chiu Pereira To: PostgreSQL ADMIN , PostgreSQL Performance Subject: ODBC Driver generates a too big "windows swap file" and it's too slow 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: 200.180.166.37 X-MailScanner-Information: Verificado pelo McAfee VirusScan / Scanned by McAfee VirusScan X-MailScanner: Nao infectado / Found to be clean X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - host2.hostseguro.com X-AntiAbuse: Original Domain - postgresql.org X-AntiAbuse: Originator/Caller UID/GID - [32001 32001] / [47 12] X-AntiAbuse: Sender Address Domain - sistemica.info X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=BAYES_10, RCVD_IN_DSBL, UPPERCASE_25_50, USER_AGENT_IMP X-Spam-Level: X-Archive-Number: 200312/166 X-Sequence-Number: 11674 Hi List, First of all, I tried to subcribe the ODBC list but it seems that the subscription's link is broken ! So here it goes: I have a delphi software use ttable components that converts dbf information to PostgreSQL an Oracle Databases. My problem is that PostgreSQL is too slow, the oracle db makes this convertion in 3.45 min and the Pg db makes int 29 min. The software is the same ( only the database reference is diferent ) , this sotware uses BDE to access the database with oracle native driver and using postgreSQL odbc driver version 5. Both databases are in the same machine ( Pentium 4 1.8Ghz, 384MB RAM DDR ) running RH 9 , Oracle 9i and PostgreSQL 7.3.2- 3. When I ran this conversion I "snorted" the communication between the server and the station to see how it does the sql requests , here it goes: ORACLE : - select owner, object_name, object_type, created from sys.all_objects where object_type in ('TABLE', 'VIEW' ) and owner = 'VENDAS' and object_name = 'FTCOFI00' order by 1 ASC, 2 ASC - select owner, index_name, uniqueness from sys.all_indexes where table_owner = 'VENDAS' and table_name = 'FTCOFI00' order by owner ASC, index_name ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCOFI01' order by column_position ASC - SELECT "EMP" ,"FIL" ,"CODIGO_FISCAL" ,"CODIGO_FISCAL_ESTORNO" ,"DESCRICAO_FISCAL " ,"CODIGO_OPERACIONAL" ,"DESCRICAO_USUARIO" ,"COD_NATIPI" ,"COD_NATIBGE" ,"EXTO _NF1" ,"TEXTO_NF2" ,"NF_NORMALDIF" ,"NF_TRANSFILIAL" ,"COD_FILIAL" ,"COD_LANCTO_ FILIAL" ,"NF_EXPORTACAO_DIRETA" ,"NF_EXPORTACAO_INDIRETA" ,"NF_SIMPREMESSA" ,"NF _DEVOLUCAO" ,"NF_ENTRADA" ,"NF_REPOSICAO" ,"NF_OUTRASERIE" ,"NF_CONSIGNACAO" ,"N F_PRODGRATIS" ,"NF_FATURANTECIP" ,"NF_DIFBASEICM" ,"NF_DIF_VALORICM" ,"NF_DIFBAS EIPI" ,"NF_DIFVALORIPI" ,"NF_DIFPRECO" ,"BLOQ_CREDITO" ,"LIBERA_CREDITO" ,"VER_P ARAM_VENDAS" ,"ENTRA_COBRANCA" ,"BASECALC_VLRBRUTO" ,"DESCNF_REFICM" ,"ALIQICM_I GUALEST" ,"COD_TRIBICM" ,"COD_TRIBIPI" ,"ATUAL_ESTOQUE" ,"ATUAL_FABRICACAO" ,"AT UAL_FATURA" ,"ATUAL_OUTENTR" ,"ATUAL_OUTSAIDA" ,"ATUAL_TRANFIL" ,"ATUAL_SEMIACAB " ,"ATUAL_CARTPED" ,"ATUAL_ENTRSAID" ,"REV_CUSTMEDIO" ,"DIGITAR_FISICO" ,"DIGITA R_FINANCEIRO" ,"USAR_CUSTO_CMU_INFORMAR" ,"GRUPO_FATURAMENTO" ,"TIPO_NF" ,"RESUM O_FISCAL_CODIGO" , "ATUAL_DISTRIB" ,"IMPR_OBS_NF_REG_ES" ,"DIFE_RECEITA" ,"COD_LANCTO" ,"SITUACAO" FROM "FTCOFI00" ORDER BY "EMP" ASC , "FIL" ASC , "CODIGO_FISCAL" ASC - select owner, object_name, object_type, created from sys.all_objects where object_type in ('TABLE', 'VIEW') and owner = 'VENDAS' and object_name = 'FTCLCR00' order by 1 ASC, 2 ASC - select owner, index_name, uniqueness from sys.all_indexes where table_owner = 'VENDAS' and table_name = 'FTCLCR00' order by owner ASC, index_name ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR01' order by column_position ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR02' order by column_position ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR03' order by column_position ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR04' order by column_position ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR05' order by column_position ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR06' order by column_position ASC - select column_name from sys.all_ind_columns where index_owner = 'VENDAS' and index_name = 'FTCLCR07' order by column_position ASC - SELECT "EMP" ,"FIL" ,"TIPO_CADASTRO" ,"CODIGO" ,"RAZAO_SOCIAL" ,"NOME_FANTASIA" ,"EMP_ENDERECO" ,"EMP_NRO" ,"EMP_COMPLEMENTO" ,"EMP_BAIRRO" ,"EMP_CIDADE" ,"EMP_ ESTADO" ,"EMP_CEP" ,"EMP_PAIS" ,"EMP_EAN" ,"COB_ENDERECO" ,"COB_NRO" ,"COB_COMPL EMENTO" ,"COB_BAIRRO" ,"COB_CIDADE" ,"COB_ESTADO" ,"COB_CEP" ,"COB_PAIS" ,"COB_E AN" ,"ENT_ENDERECO" ,"ENT_NRO" ,"ENT_COMPLEMENTO" ,"ENT_BAIRRO" ,"ENT_CIDADE" ," ENT_ESTADO", "ENT_CEP" ,"ENT_PAIS" ,"ENT_EAN" ,"LOJA_EAN" ,"TELEFONE" ,"CELULAR" ,"FAX" ,"EMA IL" ,"SITE" ,"CONTATO_NOME" ,"CONTATO_TELEFONE" ,"CONTATO_EMAIL" ,"CONTATO_DDMM_ ANIV" ,"SITUACAO_CADASTRO" ,"OBSERVACOES" ,"DATA_CADASTRO" ,"DATA_ALTERACAO" ,"T IPO_CONTRIBUINTE" ,"CODIGO_CONTRIBUINTE" ,"TIPO_INSCRICAO" ,"CODIGO_INSCRICAO"," CODIGO_REDE" ,"CODIGO_TIPO_CLIENTE" ,"CODIGO_GRUPO_CLIENTE" ,"CODIGO_SUFRAMA" ," DATA_VALIDADE_SUFRAMA" ,"LIMITE_CREDITO" , "MARCA" ,"CLASSE" ,"BANDEIRA_CLIENTE" ,"CODIGO_TIPO_CREDOR" ,"NOME_REPRESENTANTE " ,"TIPO_CONDICAO_PGTO" ,"PRAZO_PGTO_01" ,"PRAZO_PGTO_02" ,"PRAZO_PGTO_03" ,"COD IGO_MOEDA_COMPRA" ,"FATOR_QUALIDADE" ,"DESPESA_FINANCEIRA" ,"CODIGO_DARF" ,"CODI GO_NATUREZA_RENDIMENTO" ,"CONTA_CORRENTE_BANCO" ,"CONTA_CORRENTE_AGENCIA" ,"CONT A_CORRENTE_NUMERO" ,"FORNECEDOR_SULPLASTIC" ,"SUFRAMA_TRIB_ICM" ,"SUFRAMA_TRIB_I PI" ,"CONTA_CORRENTE_AGENC_DC" , "CONTA_CORRENTE_NUM_DC" ,"CONTA_COR_FORMA_PAGTO" ,"FORMA_CREDITO" ,"SENHA" ,"LIM ITE_CREDITO_PUIG" ,"COD_REPRES" ,"COD_CLIENTE_TEP" ,"EDI_MERCADOR" ,"TIPO_NF" ," BONIFIC_BALCAO" FROM "FTCLCR00" ORDER BY "EMP" ASC , "FIL" ASC , "TIPO_CADASTRO" ASC , "CODIGO" ASC PostgreSQL: - select relname, nspname, relkind from pg_catalog.pg_class, pg_catalog.pg_namespace where relkind in ('r', 'v') and nspname like 'vendas' and relname like 'ftcofi00' and relname !~ '^pg_|^dd_' and pg_namespace.oid = relnamespace order by nspname, relname - select u.nspname, c.relname, a.attname, a.atttypid, t.typname,a.attnum, a.attlen, a.atttypmod, a.attnotnull, c.relhasrules, c.relkind from pg_catalog.pg_namespace u, pg_catalog.pg_class c, pg_catalog.pg_attribute a, pg_catalog.pg_type t where u.oid = c.relnamespace and (not a.attisdropped) and c.oid= a.attrelid and a.atttypid = t.oid and (a.attnum > 0) and c.relname like 'ftcofi00' and u.nspname like 'vendas' order by u.nspname, c.relname, attnum - select u.nspname, c.relname, a.attname, a.atttypid, t.typname,a.attnum, a.attlen, a.atttypmod, a.attnotnull, c.relhasrules, c.relkind from pg_catalog.pg_namespace u, pg_catalog.pg_class c, pg_catalog.pg_attribute a, pg_catalog.pg_type t where u.oid = c.relnamespace and (not a.attisdropped) and c.oid= a.attrelid and a.atttypid = t.oid and (a.attnum > 0) and c.relname = 'ftcofi00'and u.nspname = 'vendas' order by u.nspname, c.relname, attnum - select c.relname, i.indkey, i.indisunique, i.indisclustered, a.amname, c.relhasrules, n.nspname from pg_catalog.pg_index i, pg_catalog.pg_class c, pg_catalog.pg_class d, pg_catalog.pg_am a, pg_catalog.pg_namespace n where d.relname = 'ftcofi00' and n.nspname = 'vendas' and n.oid = d.relnamespace and d.oid = i.indrelid and i.indexrelid = c.oid and c.relam = a.oid order by i.indisprimary desc, i.indisunique, n.nspname, c.relname - SELECT "emp" ,"fil" ,"codigo_fiscal" ,"codigo_fiscal_estorno" ,"descricao_fiscal " ,"codigo_operacional" ,"descricao_usuario" ,"cod_natipi" ,"cod_natibge" ,"text o_nf1" ,"texto_nf2" ,"nf_normaldif" ,"nf_transfilial" ,"cod_filial" ,"cod_lancto _filial" ,"nf_exportacao_direta" ,"nf_exportacao_indireta" ,"nf_simpremessa" ,"n f_devolucao" ,"nf_entrada" ,"nf_reposicao" ,"nf_outraserie" ,"nf_consignacao" ," nf_prodgratis" ,"nf_faturantecip" ,"nf_difbaseicm" ,"nf_dif_valoricm" ,"nf_difba seipi" ,"nf_difvaloripi" ,"nf_difpreco" ,"bloq_credito" ,"libera_credito" ,"ver_ param_vendas" ,"entra_cobranca" ,"basecalc_vlrbruto" ,"descnf_reficm" ,"aliqicm_ igualest" ,"cod_tribicm" ,"cod_tribipi" ,"atual_estoque" ,"atual_fabricacao" ,"a tual_fatura" ,"atual_outentr" ,"atual_outsaida" ,"atual_tranfil" ,"atual_semiaca b" ,"atual_cartped" ,"atual_entrsaid" ,"rev_custmedio" ,"digitar_fisico" ,"digit ar_financeiro","usar_custo_cmu_informar" ,"grupo_faturamento" ,"tipo_nf" ,"resum o_fiscal_codigo" ,"atual_distrib" , "impr_obs_nf_reg_es" ,"difer_receita" ,"cod_lancto" ,"situacao" FROM "vendas"."ftcofi00" ORDER BY "emp" ASC , "fil" ASC , "codigo_fiscal" ASC - select relname, nspname, relkind from pg_catalog.pg_class, pg_catalog.pg_namespace where relkind in ('r', 'v') and nspname like 'vendas' and relname like 'ftclcr00' and relname !~ '^pg_|^dd_' and pg_namespace.oid = relnamespace order by nspname, relname - select u.nspname, c.relname, a.attname, a.atttypid, t.typname,a.attnum, a.attlen, a.atttypmod, a.attnotnull, c.relhasrules, c.relkind from pg_catalog.pg_namespace u, pg_catalog.pg_class c, pg_catalog.pg_attribute a, pg_catalog.pg_type t where u.oid = c.relnamespace and (not a.attisdropped) and c.oid= a.attrelid and a.atttypid = t.oid and (a.attnum > 0) and c.relname like 'ftclcr00' and u.nspname like 'vendas' order by u.nspname, c.relname, attnum - select u.nspname, c.relname, a.attname, a.atttypid, t.typname,a.attnum, a.attlen, a.atttypmod, a.attnotnull, c.relhasrules, c.relkind from pg_catalog.pg_namespace u, pg_catalog.pg_class c, pg_catalog.pg_attribute a, pg_catalog.pg_type t where u.oid = c.relnamespace and (not a.attisdropped) and c.oid= a.attrelid and a.atttypid = t.oid and (a.attnum > 0) and c.relname = 'ftclcr00'and u.nspname = 'vendas' order by u.nspname, c.relname, attnum - select c.relname, i.indkey, i.indisunique, i.indisclustered, a.amname, c.relhasrules, n.nspname from pg_catalog.pg_index i, pg_catalog.pg_class c, pg_catalog.pg_class d, pg_catalog.pg_am a, pg_catalog.pg_namespace n where d.relname = 'ftclcr00' and n.nspname = 'vendas' and n.oid = d.relnamespace and d.oid = i.indrelid and i.indexrelid = c.oid and c.relam = a.oid order by i.indisprimary desc, i.indisunique, n.nspname, c.relname - SELECT "emp" ,"fil" ,"tipo_cadastro" ,"codigo" ,"razao_social","nome_fantasia" , "emp_endereco" ,"emp_nro" ,"emp_complemento" ,"emp_bairro" ,"emp_cidade" ,"emp_e stado" ,"emp_cep" ,"emp_pais","emp_ean" ,"cob_endereco" ,"cob_nro" ,"cob_complem ento" ,"cob_bairro" ,"cob_cidade" ,"cob_estado" ,"cob_cep" ,"cob_pais" ,"cob_ean " ,"ent_endereco" ,"ent_nro" ,"ent_complemento" ,"ent_bairro" ,"ent_cidade" ,"en t_estado" ,"ent_cep" ,"ent_pais" ,"ent_ean" ,"loja_ean" ,"telefone" ,"celular" , "fax" ,"email" ,"site" ,"contato_nome" ,"contato_telefone" ,"contato_email" ,"co ntato_ddmm_aniv" ,"situacao_cadastro" ,"observacoes" ,"data_cadastro" ,"data_alt eracao" ,"tipo_contribuinte" ,"codigo_contribuinte" ,"tipo_inscricao" ,"codigo_i nscricao" ,"codigo_rede" ,"codigo_tipo_cliente" ,"codigo_grupo_cliente" ,"codigo _suframa" ,"data_validade_suframa" ,"limite_credito" ,"marca" ,"classe" ,"bandei ra_cliente" ,"codigo_tipo_credor" ,"nome_representante" ,"tipo_condicao_pgto" ," prazo_pgto_01" ,"prazo_pgto_02" ,"prazo_pgto_03" , "codigo_moeda_compra" ,"fator_qualidade" ,"despesa_financeira" ,"codigo_darf" ," codigo_natureza_rendimento" ,"conta_corrente_banco" ,"conta_corrente_agencia" ," conta_corrente_numero" ,"fornecedor_sulplastic" ,"suframa_trib_icm" ,"suframa_tr ib_ipi" ,"conta_corrente_agenc_dc" ,"conta_corrente_num_dc" ,"conta_cor_forma_pa gto" ,"forma_credito" ,"senha" ,"limite_credito_puig" ,"cod_repres" ,"cod_client e_tep" ,"edi_mercador" ,"tipo_nf" ,"bonific_balcao" FROM "vendas"."ftclcr00" ORDER BY "emp" ASC , "fil" ASC , "tipo_cadastro" ASC , "codigo" ASC So , this snort generated a 3MB file for Oracle and it didn't request a bigger windows swap file but PostgreSQL generated a 153 MB file and I needed a 700 MB windows swap file ( this is unacceptable !!!! ). I tried changing the ttables components to a SQL Query but Pg did it in 49min an Oracle in 29min ( it looks like a index problem but there's no way to force an index in Pqsql ). I don't know whate else to do , and I really want to use PgSQL instead of Oracle but to do this I must PgSQL working in a compatibile time ! Any suggestions ? Atenciosamente, Rhaoni Chiu Pereira Sist�mica Computadores Visite-nos na Web: http://sistemica.info Fone/Fax : +55 51 3328 1122 From pgsql-performance-owner@postgresql.org Fri Dec 12 11:34:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 606FFD1DFA3 for ; Fri, 12 Dec 2003 15:33:57 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 40164-08 for ; Fri, 12 Dec 2003 11:33:30 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 411DDD1E001 for ; Fri, 12 Dec 2003 11:33:26 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBCFXU19019870; Fri, 12 Dec 2003 10:33:30 -0500 (EST) To: "david@shadovitz.com" Cc: "'pgsql-performance@postgresql.org'" Subject: Re: Query plan - now what? In-reply-to: <01C3C045.7EDD13C0.david@shadovitz.com> References: <01C3C045.7EDD13C0.david@shadovitz.com> Comments: In-reply-to David Shadovitz message dated "Fri, 12 Dec 2003 00:18:12 -0800" Date: Fri, 12 Dec 2003 10:33:30 -0500 Message-ID: <19869.1071243210@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/198 X-Sequence-Number: 5058 David Shadovitz writes: > Well, now that I have the plan for my slow-running query, what do I > do? This is not very informative when you didn't show us the query nor the table schemas (column datatypes and the existence of indexes are the important parts). I have a feeling that you might be well advised to fold the multiple tables into one "animals" table, but there's not enough info here to make that recommendation for sure. BTW, what did you do with this, print and OCR it? It's full of the most bizarre typos ... mostly "l" for "1", but others too ... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 12 11:47:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C6A7ED1E139 for ; Fri, 12 Dec 2003 15:47:53 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 42827-04 for ; Fri, 12 Dec 2003 11:47:26 -0400 (AST) Received: from grouse.mail.pas.earthlink.net (grouse.mail.pas.earthlink.net [207.217.120.116]) by svr1.postgresql.org (Postfix) with ESMTP id DB06ED1B4AE for ; Fri, 12 Dec 2003 11:46:50 -0400 (AST) Received: from user-11202lf.dsl.mindspring.com ([66.32.10.175] helo=shadovitzcmptr) by grouse.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 1AUpVF-0001ZS-00; Fri, 12 Dec 2003 07:46:53 -0800 Received: by localhost with Microsoft MAPI; Fri, 12 Dec 2003 07:46:42 -0800 Message-ID: <01C3C084.156718C0.david@shadovitz.com> From: David Shadovitz Reply-To: "david@shadovitz.com" To: "'pgsql-performance@postgresql.org'" Cc: 'Tom Lane' Subject: Re: Query plan - now what? Date: Fri, 12 Dec 2003 07:46:36 -0800 X-Mailer: Microsoft Internet E-mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/199 X-Sequence-Number: 5059 > This is not very informative when you didn't show us the query nor > the table schemas.. > BTW, what did you do with this, print and OCR it? Tom, I work in a classified environment, so I had to sanitize the query plan, print it, and OCR it. I spent a lot of time fixing typos, but I guess at midnight my eyes missed some. This hassle is why I posted neither the query nor the schema. The database is normalized, though, but my use of animal names of couse masks this. If you think that you or anyone else would invest the time, I could post more info. I will also try Shridhar's suggestions on statistics_target and enable_hash_join. Thanks. -David From pgsql-admin-owner@postgresql.org Fri Dec 12 12:13:55 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5B352D1E0C1; Fri, 12 Dec 2003 16:13:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49005-04; Fri, 12 Dec 2003 12:13:09 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id B895AD1DF54; Fri, 12 Dec 2003 12:06:57 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hBCG688q023596; Fri, 12 Dec 2003 09:06:08 -0700 (MST) Date: Fri, 12 Dec 2003 08:48:14 -0700 (MST) From: "scott.marlowe" To: Rhaoni Chiu Pereira Cc: PostgreSQL ADMIN , PostgreSQL Performance Subject: Re: ODBC Driver generates a too big "windows swap file" and In-Reply-To: <1071232799.3fd9b71f3d0bb@sistemica.info> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/170 X-Sequence-Number: 11678 On Fri, 12 Dec 2003, Rhaoni Chiu Pereira wrote: Hi, is there a switch in your pgsql/odbc connector to enable cursors? If so, try turning that on. From pgsql-admin-owner@postgresql.org Fri Dec 12 12:16:12 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 62C40D1DFAC; Fri, 12 Dec 2003 16:16:05 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49681-02; Fri, 12 Dec 2003 12:15:36 -0400 (AST) Received: from ndl1mr1-a-fixed.sancharnet.in (ndl1mr1-a-fixed.sancharnet.in [61.0.0.45]) by svr1.postgresql.org (Postfix) with ESMTP id 5930ED1DFFD; Fri, 12 Dec 2003 12:11:26 -0400 (AST) Received: from conversion-daemon.ndl1mr1-a-fixed.sancharnet.in by ndl1mr1-a-fixed.sancharnet.in (iPlanet Messaging Server 5.2 HotFix 1.17 (built Jun 23 2003)) id <0HPS00701HUPJ5@ndl1mr1-a-fixed.sancharnet.in>; Fri, 12 Dec 2003 21:38:50 +0530 (IST) Received: from sancharnet.in ([61.0.95.114]) by ndl1mr1-a-fixed.sancharnet.in (iPlanet Messaging Server 5.2 HotFix 1.17 (built Jun 23 2003)) with ESMTPA id <0HPS00E3BI6JTR@ndl1mr1-a-fixed.sancharnet.in>; Fri, 12 Dec 2003 21:38:50 +0530 (IST) Date: Fri, 12 Dec 2003 21:43:10 +0530 From: Sai Hertz And Control Systems Subject: Tables Without OIDS and its effect To: pgsql-admin@postgresql.org Cc: pgsql-performance@postgresql.org Reply-To: aspire420@hotpop.com Message-id: <3FD9E916.8050005@sancharnet.in> Organization: Sai Hertz And Control Systems MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=us-ascii Content-transfer-encoding: 7BIT X-Accept-Language: en-us, en User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=BAYES_20, FROM_ENDS_IN_NUMS, RCVD_IN_OSIRUSOFT_COM, RCVD_IN_RFCI, USER_AGENT_MOZILLA_UA X-Spam-Level: X-Archive-Number: 200312/171 X-Sequence-Number: 11679 Dear all , I have created my tables without OIDS now my doubts are : 1. Will this speed up the data insertion process 2. Though I have not written any code in my any of the pgsql functions which depend on OIDS 1. Will without OIDS the functions behave internally differently 2. Will my application break at any point 3. I decided to work with out OIDS because 1. It has a limit of -2147483648 to +2147483647 2 Due to this limitation I would not like to drop recreate my database because it is a bit difficult/dirty process All links and suggestion pertaining to OIDS are most welcome my mail box is at your disposal and dont hassitate to drop a two line comment. ----------------------- My Sys Config: RH 9.0 PostgreSQL 7.3.4 GCC 3.2.2 PHP 4.3.4 ---------------------- Regards, V Kashyap From pgsql-performance-owner@postgresql.org Fri Dec 12 12:59:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C55B4D1B4BB for ; Fri, 12 Dec 2003 16:59:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55639-04 for ; Fri, 12 Dec 2003 12:59:13 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 2CE21D1B4A8 for ; Fri, 12 Dec 2003 12:59:13 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBCGxCU6032463 for ; Fri, 12 Dec 2003 16:59:12 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBCGZEVL029458 for pgsql-performance@postgresql.org; Fri, 12 Dec 2003 16:35:14 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Hardware suggestions for Linux/PGSQL server Date: Fri, 12 Dec 2003 08:35:17 -0800 Organization: Hub.Org Networking Services Lines: 17 Message-ID: References: <3FD961C5.9070204@persistent.co.in> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: <3FD961C5.9070204@persistent.co.in> To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/203 X-Sequence-Number: 5063 Shridhar Daithankar wrote: > > FWIW, there are only two pieces of software that need 64bit aware for a > typical server job. Kernel and glibc. Rest of the apps can do fine as 32 > bits unless you are oracle and insist on outsmarting OS. > > In fact running 32 bit apps on 64 bit OS has plenty of advantages like > effectively using the cache. Unless you need 64bit, going for 64bit > software is not advised. This is a good point. While doing research on this matter a few months back, I saw comments by people testing 64-bit MySQL that some operations would run faster and some slower due to the use of 64-bit datatypes versus 32-bit. The best solution in the end is probably to run 32-bit Postgres under a 64-bit kernel -- unless your DB tends to have a lot of 64-bit datatypes. From pgsql-performance-owner@postgresql.org Fri Dec 12 12:53:51 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 12791D1B481 for ; Fri, 12 Dec 2003 16:53:50 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53512-08 for ; Fri, 12 Dec 2003 12:53:18 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 3EC1BD1B8BE for ; Fri, 12 Dec 2003 12:53:17 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBCGqu19020395; Fri, 12 Dec 2003 11:53:16 -0500 (EST) To: "david@shadovitz.com" Cc: "'pgsql-performance@postgresql.org'" Subject: Re: Query plan - now what? In-reply-to: <01C3C084.156718C0.david@shadovitz.com> References: <01C3C084.156718C0.david@shadovitz.com> Comments: In-reply-to David Shadovitz message dated "Fri, 12 Dec 2003 07:46:36 -0800" Date: Fri, 12 Dec 2003 11:52:56 -0500 Message-ID: <20394.1071247976@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/202 X-Sequence-Number: 5062 David Shadovitz writes: > If you think that you or anyone else would invest the time, I could post more > info. I doubt you will get any useful help if you don't post more info. > I will also try Shridhar's suggestions on statistics_target and > enable_hash_join. It seemed to me that the row estimates were not so far off that I would call it a statistical failure; you can try increasing the stats target but I'm not hopeful about that. My guess is that you will have to look to revising either the query or the whole database structure (ie, merging tables). We'll need the info I asked for before we can make any recommendations, though. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 12 14:05:44 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CCCFAD1E016 for ; Fri, 12 Dec 2003 18:05:36 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 67920-01 for ; Fri, 12 Dec 2003 14:05:05 -0400 (AST) Received: from tino.sinectis.com.ar (tino.sinectis.com.ar [216.244.192.232]) by svr1.postgresql.org (Postfix) with ESMTP id A88A8D1E012 for ; Fri, 12 Dec 2003 14:05:00 -0400 (AST) Received: by tino.sinectis.com.ar (Postfix, from userid 99) id 697FD6C58F; Fri, 12 Dec 2003 15:04:49 -0300 (GMT+3) Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 X-Mailer: Sinectis Webmail 5.6.16-1.4.4 From: nbarraza@uolsinectis.com.ar To: pgsql-performance@postgresql.org Subject: Performance related to size of tables Reply-To: nbarraza@uolsinectis.com.ar Message-Id: <20031212180449.697FD6C58F@tino.sinectis.com.ar> Date: Fri, 12 Dec 2003 15:04:49 -0300 (GMT+3) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/204 X-Sequence-Number: 5064 Hi everyone, I found that performance get worse as the size of a given table increases. I mean, for example I�ve just run some scripts shown in http://www.potentialtech.com/wmoran/postgresql.php I understand that those scripts are designed to see the behavior of postgresql under different filesystems. However, since them generate a lot of I/O activity, I think they can be used to adjust some configuration parameters. In that way, I increased the number of tuples inserted in the initial table to 2000000 and 3000000. What I saw is that the running time goes from 3 min., to 11 min. My question is, is it possible to use that test to tune some parameters?, if the answer is yes, what parameters should I change to get shorter running times? Thanks a lot Nestor From pgsql-performance-owner@postgresql.org Tue Dec 16 17:08:01 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 833B7D1B45A; Fri, 12 Dec 2003 20:55:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92295-10; Fri, 12 Dec 2003 16:55:25 -0400 (AST) Received: from dbl.q-ag.de (dbl.q-ag.de [80.146.160.66]) by svr1.postgresql.org (Postfix) with ESMTP id 6555DD1D8B4; Fri, 12 Dec 2003 16:55:19 -0400 (AST) Received: from colorfullife.com (dbl [127.0.0.1]) by dbl.q-ag.de (8.12.3/8.12.3/Debian-6.6) with ESMTP id hBCKsYN9021076; Fri, 12 Dec 2003 21:54:36 +0100 Message-ID: <3FDA2B0A.1060709@colorfullife.com> Date: Fri, 12 Dec 2003 21:54:34 +0100 From: Manfred Spraul User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4.1) Gecko/20031030 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Bruce Momjian Cc: Mark Kirkwood , pgsql-performance@postgresql.org, PostgreSQL-development Subject: Re: [HACKERS] fsync method checking References: <200312120649.hBC6nQR15608@candle.pha.pa.us> In-Reply-To: <200312120649.hBC6nQR15608@candle.pha.pa.us> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/233 X-Sequence-Number: 5093 Bruce Momjian wrote: > write 0.000360 > write & fsync 0.001391 > write, close & fsync 0.001308 > open o_fsync, write 0.000924 > > That's 1 milliseconds vs. 1.3 milliseconds. Neither value is realistic - I guess the hw cache on and the os doesn't issue cache flush commands. Realistic values are probably 5 ms vs 5.3 ms - 6%, not 30%. How large is the syscall latency with BSD/OS 4.3? One advantage of a seperate write and fsync call is better performance for the writes that are triggered within AdvanceXLInsertBuffer: I'm not sure how often that's necessary, but it's a write while holding both the WALWriteLock and WALInsertLock. If every write contains an implicit sync, that call would be much more expensive than necessary. -- Manfred From pgsql-performance-owner@postgresql.org Fri Dec 12 17:11:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ADE28D1D38F for ; Fri, 12 Dec 2003 21:11:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96734-03 for ; Fri, 12 Dec 2003 17:11:13 -0400 (AST) Received: from linda-2.paradise.net.nz (bm-2a.paradise.net.nz [202.0.58.21]) by svr1.postgresql.org (Postfix) with ESMTP id 8BBAED1BCB8 for ; Fri, 12 Dec 2003 17:11:11 -0400 (AST) Received: from smtp-1.paradise.net.nz (smtp-1a.paradise.net.nz [202.0.32.194]) by linda-2.paradise.net.nz (Paradise.net.nz) with ESMTP id <0HPS00JS3W6N6P@linda-2.paradise.net.nz> for pgsql-performance@postgresql.org; Sat, 13 Dec 2003 10:11:12 +1300 (NZDT) Received: from paradise.net.nz (218-101-14-127.paradise.net.nz [218.101.14.127]) by smtp-1.paradise.net.nz (Postfix) with ESMTP id 188B982768; Sat, 13 Dec 2003 10:11:11 +1300 (NZDT) Date: Sat, 13 Dec 2003 10:11:29 +1300 From: Mark Kirkwood Subject: Re: Performance related to size of tables In-reply-to: <20031212180449.697FD6C58F@tino.sinectis.com.ar> To: nbarraza@uolsinectis.com.ar Cc: pgsql-performance@postgresql.org Message-id: <3FDA2F01.10205@paradise.net.nz> MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 8BIT X-Accept-Language: en-us, en User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.5) Gecko/20031213 References: <20031212180449.697FD6C58F@tino.sinectis.com.ar> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/205 X-Sequence-Number: 5065 If you want to speed up the elapsed times, then the first thing would be to attempt to reduce the IO using some indexes, e.g. on test1(anumber), test2(anumber), test3((anumber%13)), test3((anumber%5)) and test4((anumber%27)) However if you wish to keep hammering the IO then the you would not use any indexes. However elapsed times for operations like: CREATE TABLE test4 AS SELECT ... FROM test1 JOIN test2 ON test1.anumber=test2.anumber; are going to increase non linearly with the size of the source table test1 (unless there are indexes on the anumber columns). I think this particular test is designed as a testbed for measuring IO performance - as opposed to Postgresql performance. regards Mark nbarraza@uolsinectis.com.ar wrote: >Hi everyone, >I found that performance get worse as the size of a given table >increases. I mean, for example I�ve just run some scripts shown in > >http://www.potentialtech.com/wmoran/postgresql.php > >I understand that those scripts are designed to see the behavior of postgresql under different filesystems. However, since them generate >a lot of I/O activity, I think they can be used to adjust some >configuration parameters. In that way, I increased the number of tuples inserted in the initial table to 2000000 and 3000000. What >I saw is that the running time goes from 3 min., to 11 min. My question is, is it possible to use that test to tune >some parameters?, if the answer is yes, what parameters should I change to get shorter running times? > > > > From pgsql-performance-owner@postgresql.org Fri Dec 12 17:29:35 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BBF6BD1B45A; Fri, 12 Dec 2003 21:29:34 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 98506-02; Fri, 12 Dec 2003 17:29:04 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 8CDDBD1B460; Fri, 12 Dec 2003 17:29:02 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBCLSl19022122; Fri, 12 Dec 2003 16:28:47 -0500 (EST) To: Manfred Spraul Cc: Bruce Momjian , Mark Kirkwood , pgsql-performance@postgresql.org, PostgreSQL-development Subject: Re: [HACKERS] fsync method checking In-reply-to: <3FDA2B0A.1060709@colorfullife.com> References: <200312120649.hBC6nQR15608@candle.pha.pa.us> <3FDA2B0A.1060709@colorfullife.com> Comments: In-reply-to Manfred Spraul message dated "Fri, 12 Dec 2003 21:54:34 +0100" Date: Fri, 12 Dec 2003 16:28:47 -0500 Message-ID: <22121.1071264527@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/206 X-Sequence-Number: 5066 Manfred Spraul writes: > One advantage of a seperate write and fsync call is better performance > for the writes that are triggered within AdvanceXLInsertBuffer: I'm not > sure how often that's necessary, but it's a write while holding both the > WALWriteLock and WALInsertLock. If every write contains an implicit > sync, that call would be much more expensive than necessary. Ideally that path isn't taken very often. But I'm currently having a discussion off-list with a CMU student who seems to be seeing a case where it happens a lot. (She reports that both WALWriteLock and WALInsertLock are causes of a lot of process blockages, which seems to mean that a lot of the WAL I/O is being done with both held, which would have to mean that AdvanceXLInsertBuffer is doing the I/O. More when we figure out what's going on exactly...) regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 12 18:29:48 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 89384D1B430 for ; Fri, 12 Dec 2003 22:29:45 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01515-09 for ; Fri, 12 Dec 2003 18:29:16 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id EE927D1BA62 for ; Fri, 12 Dec 2003 18:29:13 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBCMTDU6081572 for ; Fri, 12 Dec 2003 22:29:13 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBCM1uoZ044278 for pgsql-performance@postgresql.org; Fri, 12 Dec 2003 22:01:56 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Update on putting WAL on ramdisk/ Date: Fri, 12 Dec 2003 14:02:03 -0800 Organization: Hub.Org Networking Services Lines: 17 Message-ID: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/207 X-Sequence-Number: 5067 Some arbitrary data processing job WAL on single drive: 7.990 rec/s WAL on 2nd IDE drive: 8.329 rec/s WAL on tmpfs: 13.172 rec/s A huge jump in performance but a bit scary having a WAL that can disappear at any time. I'm gonna workup a rsync script and do some power-off experiments to see how badly it gets mangled. This could be good method though when you're dumping and restore an entire DB. Make a tmpfs mount, restore, shutdown DB and then copy the WAL back to the HD. I checked out the SanDisk IDE FlashDrives. They have a write cycle life of 2 million. I'll explore more expensive solid state drives. From pgsql-performance-owner@postgresql.org Fri Dec 12 18:37:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CEF13D1B8C0 for ; Fri, 12 Dec 2003 22:37:18 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06594-03 for ; Fri, 12 Dec 2003 18:36:50 -0400 (AST) Received: from mta1.sucs.soton.ac.uk (mta1.sucs.soton.ac.uk [152.78.128.140]) by svr1.postgresql.org (Postfix) with ESMTP id C2508D1B4C5 for ; Fri, 12 Dec 2003 18:36:46 -0400 (AST) Received: from russell (cpc1-glfd2-4-0-cust205.glfd.cable.ntl.com [81.99.187.205]) (authenticated bits=0) by mta1.sucs.soton.ac.uk (8.12.10/8.12.10) with ESMTP id hBCMaP3A022304; Fri, 12 Dec 2003 22:36:26 GMT From: "Russell Garrett" To: "William Yu" , Subject: Re: Update on putting WAL on ramdisk/ Date: Fri, 12 Dec 2003 22:36:26 -0000 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 (9.0.2910.0) In-reply-to: X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-ISS-MailScanner: Believed to be clean X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/208 X-Sequence-Number: 5068 > WAL on single drive: 7.990 rec/s > WAL on 2nd IDE drive: 8.329 rec/s > WAL on tmpfs: 13.172 rec/s > > A huge jump in performance but a bit scary having a WAL that can > disappear at any time. I'm gonna workup a rsync script and do some > power-off experiments to see how badly it gets mangled. Surely this is just equivalent to disabling fsync? If you put a WAL on a volatile file system, there's not a whole lot of point in having one at all. -------------------------------------------------------------------- Russ Garrett russ@last.fm http://last.fm From pgsql-performance-owner@postgresql.org Tue Dec 16 17:08:03 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2D6F2D1DFE7 for ; Fri, 12 Dec 2003 22:41:14 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06322-02 for ; Fri, 12 Dec 2003 18:40:44 -0400 (AST) Received: from q.idc-mcs.com (unknown [64.156.199.9]) by svr1.postgresql.org (Postfix) with ESMTP id ADFBED1C919 for ; Fri, 12 Dec 2003 18:40:41 -0400 (AST) 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_01C3C100.F1020EA8" Subject: Excessive rows/tuples seriously degrading query performance Date: Fri, 12 Dec 2003 14:40:28 -0800 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Excessive rows/tuples seriously degrading query performance Thread-Index: AcPBAPD/EOCvMLcRQ9isBa7sLviPMg== From: "Chadwick, Russell" To: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/235 X-Sequence-Number: 5094 This is a multi-part message in MIME format. ------_=_NextPart_001_01C3C100.F1020EA8 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable =20 Hello everyone. Can anyone explain why this table which has never had more than a couple ro= ws in it shows > 500k in the query planner even after running vacuum full. = Its terribly slow to return 2 rows of data. The 2 rows in it are being up= dated a lot but I couldn't find any explanation for this behavior. Anythin= g I could try besides droping db and recreating?=20=20 Thanks - Russ =20 toolshed=3D# explain analyze select * from stock_log_positions ; QUERY PLAN=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20 ---------------------------------------------------------------------------= ---------------------------------------------- Seq Scan on stock_log_positions (cost=3D0.00..10907.77 rows=3D613577 widt= h=3D22) (actual time=3D701.39..701.41 rows=3D2 loops=3D1) Total runtime: 701.54 msec (2 rows) =20 toolshed=3D# vacuum full analyze verbose stock_log_positions; INFO: --Relation public.stock_log_positions-- INFO: Pages 4773: Changed 1, reaped 767, Empty 0, New 0; Tup 613737: Vac 5= 7620, Keep/VTL 613735/613713, UnUsed 20652, MinLen 52, MaxLen 52; Re-using:= Free/Avail. Space 4322596/4322596; EndEmpty/Avail. Pages 0/4773. CPU 9.11s/13.68u sec elapsed 22.94 sec. INFO: Index idx_stock_log_positions_when_log_filename: Pages 9465; Tuples = 613737: Deleted 57620. CPU 1.55s/1.27u sec elapsed 6.69 sec. INFO: Rel stock_log_positions: Pages: 4773 --> 4620; Tuple(s) moved: 59022. CPU 1.00s/4.45u sec elapsed 8.83 sec. INFO: Index idx_stock_log_positions_when_log_filename: Pages 9778; Tuples = 613737: Deleted 2897. CPU 1.32s/0.44u sec elapsed 6.23 sec. INFO: Analyzing public.stock_log_positions VACUUM =20 toolshed=3D# explain analyze select * from stock_log_positions ; QUERY PLAN=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20= =20=20=20 ---------------------------------------------------------------------------= ---------------------------------------------- Seq Scan on stock_log_positions (cost=3D0.00..10757.37 rows=3D613737 widt= h=3D22) (actual time=3D789.21..789.24 rows=3D2 loops=3D1) Total runtime: 789.40 msec (2 rows) =20 toolshed=3D# select * from stock_log_positions ; when_log | filename | position=20 ------------+--------------+---------- 2003-12-11 | ActiveTrader | 0 2003-12-11 | Headlines | 0 (2 rows) ------_=_NextPart_001_01C3C100.F1020EA8 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
 
Hello everyone.
Can anyone explain why this table which ha= s never=20 had more than a couple rows in it shows > 500k in the query planner even= =20 after running vacuum full.  Its terribly slow to return 2 rows of=20 data.  The 2 rows in it are being updated a lot but I couldn't find an= y=20 explanation for this behavior.  Anything I could try besides droping d= b and=20 recreating? 
Thanks - Russ
 
toolshed=3D# explain analyze select * from stock_log_positions=20 ;
           &nbs= p;            &= nbsp;           &nbs= p;            &= nbsp;    =20 QUERY=20 PLAN            = ;            &n= bsp;            = ;            &n= bsp;     =20
-----------------------------------------------------------------------= --------------------------------------------------
 Seq=20 Scan on stock_log_positions  (cost=3D0.00..10907.77 rows=3D613577 widt= h=3D22)=20 (actual time=3D701.39..701.41 rows=3D2 loops=3D1)
 Total runtime: 7= 01.54=20 msec
(2 rows)
 
toolshed=3D# vacuum full analyze verbose stock_log_positions;
INFO:=  =20 --Relation public.stock_log_positions--
INFO:  Pages 4773: Changed = 1,=20 reaped 767, Empty 0, New 0; Tup 613737: Vac 57620, Keep/VTL 613735/613713,= =20 UnUsed 20652, MinLen 52, MaxLen 52; Re-using: Free/Avail. Space 4322596/432= 2596;=20 EndEmpty/Avail. Pages 0/4773.
       = CPU=20 9.11s/13.68u sec elapsed 22.94 sec.
INFO:  Index=20 idx_stock_log_positions_when_log_filename: Pages 9465; Tuples 613737: Delet= ed=20 57620.
        CPU 1.55s/1.27u sec el= apsed=20 6.69 sec.
INFO:  Rel stock_log_positions: Pages: 4773 --> 4620;= =20 Tuple(s) moved: 59022.
        CPU=20 1.00s/4.45u sec elapsed 8.83 sec.
INFO:  Index=20 idx_stock_log_positions_when_log_filename: Pages 9778; Tuples 613737: Delet= ed=20 2897.
        CPU 1.32s/0.44u sec ela= psed=20 6.23 sec.
INFO:  Analyzing public.stock_log_positions
VACUUM
 
toolshed=3D# explain analyze select * from stock_log_positions=20 ;
           &nbs= p;            &= nbsp;           &nbs= p;            &= nbsp;    =20 QUERY=20 PLAN            = ;            &n= bsp;            = ;            &n= bsp;     =20
-----------------------------------------------------------------------= --------------------------------------------------
 Seq=20 Scan on stock_log_positions  (cost=3D0.00..10757.37 rows=3D613737 widt= h=3D22)=20 (actual time=3D789.21..789.24 rows=3D2 loops=3D1)
 Total runtime: 7= 89.40=20 msec
(2 rows)
 
toolshed=3D# select * from stock_log_positions ;
  when_log&nb= sp;=20 |   filename   | position=20
------------+--------------+----------
 2003-12-11 | ActiveTrad= er=20 |        0
 2003-12-11 |=20 Headlines    |        0(2=20 rows)
------_=_NextPart_001_01C3C100.F1020EA8-- From pgsql-performance-owner@postgresql.org Fri Dec 12 18:59:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C72EFD1B488 for ; Fri, 12 Dec 2003 22:59:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08281-06 for ; Fri, 12 Dec 2003 18:59:16 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 2BB05D1B4AD for ; Fri, 12 Dec 2003 18:59:14 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBCMxDU6012627 for ; Fri, 12 Dec 2003 22:59:13 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBCMjO55097997 for pgsql-performance@postgresql.org; Fri, 12 Dec 2003 22:45:24 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Update on putting WAL on ramdisk/ Date: Fri, 12 Dec 2003 14:45:31 -0800 Organization: Hub.Org Networking Services Lines: 18 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/209 X-Sequence-Number: 5069 Russell Garrett wrote: >>WAL on single drive: 7.990 rec/s >>WAL on 2nd IDE drive: 8.329 rec/s >>WAL on tmpfs: 13.172 rec/s >> >>A huge jump in performance but a bit scary having a WAL that can >>disappear at any time. I'm gonna workup a rsync script and do some >>power-off experiments to see how badly it gets mangled. > > > Surely this is just equivalent to disabling fsync? If you put a WAL on a > volatile file system, there's not a whole lot of point in having one at all. These tests were all with fsync off. And no, it's not equivalent to fsync off since the WAL is always written immediately regardless of fsync setting. From pgsql-admin-owner@postgresql.org Fri Dec 12 19:45:56 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1B70BD1B495; Fri, 12 Dec 2003 23:11:09 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08644-05; Fri, 12 Dec 2003 19:10:40 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id E4A7AD1CC78; Fri, 12 Dec 2003 19:10:36 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 6BC5D205E; Fri, 12 Dec 2003 18:10:22 -0500 (EST) To: aspire420@hotpop.com Cc: pgsql-admin@postgresql.org, pgsql-performance@postgresql.org Subject: Re: [PERFORM] Tables Without OIDS and its effect From: Neil Conway In-Reply-To: <3FD9E916.8050005@sancharnet.in> (Sai Hertz And Control Systems's message of "Fri, 12 Dec 2003 21:43:10 +0530") References: <3FD9E916.8050005@sancharnet.in> Date: Fri, 12 Dec 2003 18:10:21 -0500 Message-ID: <87he05pr8i.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/178 X-Sequence-Number: 11685 Sai Hertz And Control Systems writes: > I have created my tables without OIDS now my doubts are : > 1. Will this speed up the data insertion process Slightly. It means that each inserted row will be 4 bytes smaller (on disk), which in turn means you can fit more tuples on a page, and therefore you'll need fewer pages and less disk space. However, I'd be surprised if the performance improvement is very significant. > 2. Though I have not written any code in my any of the pgsql functions > which depend on OIDS > 1. Will without OIDS the functions behave internally differently > 2. Will my application break at any point No. BTW, we intend to phase out the use of OIDs for user tables in the long term. There have been a few threads on -hackers that discuss the plans for doing this. -Neil From pgsql-admin-owner@postgresql.org Sat Dec 13 06:34:37 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7FAAFD1B4B4; Sat, 13 Dec 2003 10:34:34 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 84123-04; Sat, 13 Dec 2003 06:34:05 -0400 (AST) Received: from ndl1mr1-a-fixed.sancharnet.in (ndl1mr1-a-fixed.sancharnet.in [61.0.0.45]) by svr1.postgresql.org (Postfix) with ESMTP id D388DD1B462; Sat, 13 Dec 2003 06:34:01 -0400 (AST) Received: from conversion-daemon.ndl1mr1-a-fixed.sancharnet.in by ndl1mr1-a-fixed.sancharnet.in (iPlanet Messaging Server 5.2 HotFix 1.17 (built Jun 23 2003)) id <0HPT00E01X538I@ndl1mr1-a-fixed.sancharnet.in>; Sat, 13 Dec 2003 15:59:48 +0530 (IST) Received: from sancharnet.in ([61.0.95.123]) by ndl1mr1-a-fixed.sancharnet.in (iPlanet Messaging Server 5.2 HotFix 1.17 (built Jun 23 2003)) with ESMTPA id <0HPT00899X4TZ0@ndl1mr1-a-fixed.sancharnet.in>; Sat, 13 Dec 2003 15:59:48 +0530 (IST) Date: Sat, 13 Dec 2003 16:03:37 +0530 From: Sai Hertz And Control Systems Subject: Re: [PERFORM] Tables Without OIDS and its effect In-reply-to: <87he05pr8i.fsf@mailbox.samurai.com> To: Neil Conway Cc: aspire420@hotpop.com, Postgres Admin List , pgsql-performance@postgresql.org Reply-To: aspire420@hotpop.com Message-id: <3FDAEB01.3040200@sancharnet.in> Organization: Sai Hertz And Control Systems MIME-version: 1.0 Content-type: multipart/alternative; boundary="Boundary_(ID_xtW9LKqOugmqp3bRWinEDw)" X-Accept-Language: en-us, en User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 References: <3FD9E916.8050005@sancharnet.in> <87he05pr8i.fsf@mailbox.samurai.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/179 X-Sequence-Number: 11686 This is a multi-part message in MIME format. --Boundary_(ID_xtW9LKqOugmqp3bRWinEDw) Content-type: text/plain; charset=us-ascii; format=flowed Content-transfer-encoding: 7BIT Hello Neil Conway, We are doing some test on our applications and will let know the community if without OIDS we could gain more speed . >>2. Though I have not written any code in my any of the pgsql functions >>which depend on OIDS >> 1. Will without OIDS the functions behave internally differently >> 2. Will my application break at any point >> >> > >No. > >BTW, we intend to phase out the use of OIDs for user tables in the >long term. There have been a few threads on -hackers that discuss the >plans for doing this. > > This was a relief for us all , but an the same time we have found one incompatibility This incompatibility is with 1. StarOffice 7.0 2. OpenOffice 1.1 and the incompatibility is when I retrieve data into Star SpreadSheet or Open Office SpreadSheet I am greeted with an error field *OID* not found. Though these both are connecting to PostgreSQL 7.3.4 (Linux GCC 3.x) via psqlODBC 07.02.0003 On the Same time WinSQL connects as usual via psqlODBC 07.02.0003 and is working fine. Though this does not effect us a lot since we are using PHP to show and retrieve data We are posting this such that any one relying totally on OpenOffice for data retrieve and display better know this , Our Test config was: -------------------------- Client :- O.S Win XP (No service pack) OpenOffice 1.1 Windows version StarOffice 7.0 Eval Pack psqlODBC 07.02.0003 Server :- OS RH 9.0 kernel-2.4.20-24.9 PostgreSQL 7.3.4 Please if anyone has a different story while using WITHOUT OIDS please let us and every one know . Regards, V Kashyap --Boundary_(ID_xtW9LKqOugmqp3bRWinEDw) Content-type: text/html; charset=us-ascii Content-transfer-encoding: 7BIT Hello  Neil Conway,

We are doing some test on our applications and will let know the community  if without OIDS we could gain
more speed .
2. Though I have not written any code in my any of the pgsql functions
which depend on OIDS
  1. Will without OIDS the functions behave internally differently
  2. Will my application break at any point
    

No.

BTW, we intend to phase out the use of OIDs for user tables in the
long term. There have been a few threads on -hackers that discuss the
plans for doing this.
This was a relief  for us all , but an the same time we have found one incompatibility

This incompatibility is with
1.  StarOffice 7.0
2. OpenOffice 1.1
and the incompatibility is when I retrieve data into Star SpreadSheet or  Open Office SpreadSheet
I am greeted with an error field OID  not found.
Though these both are connecting to PostgreSQL  7.3.4 (Linux GCC 3.x) via psqlODBC  07.02.0003
On the Same time WinSQL connects as usual via psqlODBC  07.02.0003  and is working fine.

Though  this does not  effect us a lot since we are using PHP to show and retrieve data
We are posting this such that any one relying totally on OpenOffice for data  retrieve and display
better know this ,

Our Test config was:
--------------------------
Client :-
O.S           Win XP (No service pack)
OpenOffice    1.1  Windows version
StarOffice      7.0   Eval Pack
psqlODBC 07.02.0003
Server  :-
OS                      RH 9.0 kernel-2.4.20-24.9
PostgreSQL       7.3.4

Please if anyone has a different story  while using  WITHOUT  OIDS please let us and every one know .


Regards,
V Kashyap




--Boundary_(ID_xtW9LKqOugmqp3bRWinEDw)-- From pgsql-performance-owner@postgresql.org Sat Dec 13 14:39:18 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 33968D1DDE9 for ; Sat, 13 Dec 2003 18:39:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 23546-05 for ; Sat, 13 Dec 2003 14:38:54 -0400 (AST) Received: from purple.bdb.fi (purple.bdb.fi [195.197.212.62]) by svr1.postgresql.org (Postfix) with ESMTP id 8DE35D1DD50 for ; Sat, 13 Dec 2003 14:38:47 -0400 (AST) Received: by purple.bdb.fi (Postfix, from userid 101) id 4BF092FA8; Sat, 13 Dec 2003 20:38:53 +0200 (EET) Received: from localhost (localhost [127.0.0.1]) by purple.bdb.fi (Postfix) with ESMTP id 138722F9B for ; Sat, 13 Dec 2003 20:38:52 +0200 (EET) Date: Sat, 13 Dec 2003 20:38:52 +0200 (EET) From: Kari Lavikka To: pgsql-performance@postgresql.org Subject: a lot of problems with pg 7.4 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/212 X-Sequence-Number: 5072 Hi! We have been running a rather busy website using pg 7.3 as the database. Peak hitrate is something like 120 request / second without images and other static stuff. The site is a sort of image gallery for IRC users. I evaluated pg 7.4 on our development server and it looked just fine but performance with production loads seems to be quite poor. Most of performance problems are caused by nonsensical query plans but there's also some strange slowness that I can't locate. I have included the essential tables, columns and indexes that participate to queries in this mail. table rows ------- ---- users 50k image 400k comment 17M Table "public.users" Column | Type | -------------+-----------------------------+ uid | integer | nick | character varying(40) | status | character(1) | Indexes: "users_pkey" primary key, btree (uid) "users_upper_nick" unique, btree (upper((nick)::text)) "users_status" btree (status) Table "public.image" Column | Type | ----------------------+-----------------------------+ image_id | integer | uid | integer | status | character(1) | Indexes: "image_pkey" primary key, btree (image_id) "image_uid_status" btree (uid, status) Table "public.comment" Column | Type | ------------+-----------------------------+ comment_id | integer | image_id | integer | uid_sender | integer | comment | character varying(255) | Indexes: "comment_pkey" primary key, btree (comment_id) "comment_image_id" btree (image_id) "comment_uid_sender" btree (uid_sender) Planner estimates the cost of nested loop to be much higher than hash join _although_ the other side of join consists of only one row (which is found using a unique index). Well, bad estimation. Difference in runtime is huge. galleria=# explain analyze SELECT i.image_id, i.info, i.stamp, i.status, i.t_width, i.t_height, u.nick, u.uid FROM users u INNER JOIN image i ON i.uid = u.uid WHERE upper(u.nick) = upper('Intuitio') AND (i.status = 'd' OR i.status = 'v') AND u.status = 'a' ORDER BY status, stamp DESC; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=21690.23..21694.22 rows=1595 width=64) (actual time=2015.615..2015.637 rows=35 loops=1) Sort Key: i.status, i.stamp -> Hash Join (cost=907.20..21605.38 rows=1595 width=64) (actual time=891.400..2015.464 rows=35 loops=1) Hash Cond: ("outer".uid = "inner".uid) -> Seq Scan on image i (cost=0.00..18207.19 rows=330005 width=54) (actual time=0.012..1607.278 rows=341086 loops=1) Filter: ((status = 'd'::bpchar) OR (status = 'v'::bpchar)) -> Hash (cost=906.67..906.67 rows=213 width=14) (actual time=0.128..0.128 rows=0 loops=1) -> Index Scan using users_upper_nick on users u (cost=0.00..906.67 rows=213 width=14) (actual time=0.120..0.122 rows=1 loops=1) Index Cond: (upper((nick)::text) = 'INTUITIO'::text) Filter: (status = 'a'::bpchar) Total runtime: 2015.756 ms galleria=# set enable_hashjoin = false; SET galleria=# explain analyze SELECT i.image_id, i.info, i.stamp, i.status, i.t_width, i.t_height, u.nick, u.uid FROM users u INNER JOIN image i ON i.uid = u.uid WHERE upper(u.nick) = upper('Intuitio') AND (i.status = 'd' OR i.status = 'v') AND u.status = 'a' ORDER BY status, stamp DESC; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------------- Sort (cost=31090.72..31094.71 rows=1595 width=64) (actual time=5.240..5.267 rows=35 loops=1) Sort Key: i.status, i.stamp -> Nested Loop (cost=0.00..31005.87 rows=1595 width=64) (actual time=4.474..5.082 rows=35 loops=1) -> Index Scan using users_upper_nick on users u (cost=0.00..906.67 rows=213 width=14) (actual time=3.902..3.906 rows=1 loops=1) Index Cond: (upper((nick)::text) = 'INTUITIO'::text) Filter: (status = 'a'::bpchar) -> Index Scan using image_uid_status on image i (cost=0.00..141.03 rows=23 width=54) (actual time=0.537..0.961 rows=35 loops=1) Index Cond: (i.uid = "outer".uid) Filter: ((status = 'd'::bpchar) OR (status = 'v'::bpchar)) Total runtime: 5.479 ms (10 rows) Is there anything to do for this besides forcing hashjoin off? I think there were similar problems with 7.3 Now something specific to 7.4. The following query selects all comments written to user's image. It worked just fine with pg 7.3 but there seems to be a Materialize in a bit strange place. galleria=# explain SELECT s.nick, c.comment, c.private, c.admin, c.parsable, c.uid_sender, c.stamp, i.image_id, c.comment_id FROM users s, comment c, image i WHERE s.uid = c.uid_sender AND s.status = 'a' AND c.visible = 'y' AND c.image_id = i.image_id AND i.image_id = 184239 ORDER BY c.comment_id DESC; QUERY PLAN ---------------------------------------------------------------------------------------------------------------- Sort (cost=1338.43..1339.41 rows=392 width=92) Sort Key: c.comment_id -> Nested Loop (cost=1308.41..1321.54 rows=392 width=92) -> Index Scan using image_pkey on image i (cost=0.00..5.29 rows=2 width=4) Index Cond: (image_id = 184239) -> Materialize (cost=1308.41..1310.37 rows=196 width=92) -> Nested Loop (cost=0.00..1308.41 rows=196 width=92) -> Index Scan using comment_image_id on "comment" c (cost=0.00..60.68 rows=207 width=82) Index Cond: (184239 = image_id) Filter: (visible = 'y'::bpchar) -> Index Scan using users_pkey on users s (cost=0.00..6.02 rows=1 width=14) Index Cond: (s.uid = "outer".uid_sender) Filter: (status = 'a'::bpchar) However, when the joins are written in a different style the plan seems to be just right. galleria=# explain SELECT u.nick, c.comment, c.private, c.admin, c.parsable, c.uid_sender, c.stamp, i.image_id, c.comment_id FROM image i INNER JOIN comment c ON c.image_id = i.image_id INNER JOIN users u ON u.uid = c.uid_sender WHERE c.visible = 'y' AND c.image_id = i.image_id AND i.image_id = 184239 ORDER BY c.comment_id DESC; QUERY PLAN ------------------------------------------------------------------------------------------------------- Sort (cost=17.76..17.76 rows=1 width=92) Sort Key: c.comment_id -> Nested Loop (cost=0.00..17.75 rows=1 width=92) -> Nested Loop (cost=0.00..11.72 rows=1 width=82) -> Index Scan using image_pkey on image i (cost=0.00..5.29 rows=2 width=4) Index Cond: (image_id = 184239) -> Index Scan using comment_image_id on "comment" c (cost=0.00..3.20 rows=1 width=82) Index Cond: ((c.image_id = "outer".image_id) AND (184239 = c.image_id)) Filter: (visible = 'y'::bpchar) -> Index Scan using users_pkey on users u (cost=0.00..6.01 rows=1 width=14) Index Cond: (u.uid = "outer".uid_sender) (11 rows) I happened to look into this query when one of them got stuck. Normally postgres performs tens of these in a second, but after shutting down the web server one them was still running. I gathered some statistics and the runtime was something like half an hour! It was causing pretty much disk io but quite little cpu load. Don't know what it was doing... galleria=# select * from pg_stat_activity where current_query != ''; datid | datname | procpid | usesysid | usename | current_query | query_start -------+----------+---------+----------+----------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+------------------------------- 17144 | galleria | 27849 | 100 | galleria | SELECT s.nick, c.comment, c.private, c.admin, c.parsable, c.uid_sender, c.stamp, i.image_id, c.comment_id FROM users s, comment c, image i WHERE s.uid = c.uid_sender AND s.status = 'a' AND c.visible = 'y' AND c.image_id = i.image_id AND i.image_id = 95406 | 2003-12-08 19:15:10.218859+02 (1 row) PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ Command 29756 tuner 25 0 1212 1212 800 R 42.7 0.0 2:51.03 top 27849 postgres 15 0 783m 783m 780m D 6.2 20.0 0:55.86 postmaster procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 90912 551104 16180 3195612 0 0 2724 0 692 628 5 0 95 0 0 1 90912 548280 16192 3198464 0 0 2864 0 810 689 3 2 95 0 1 0 90912 545644 16192 3201068 0 0 2604 0 686 663 5 1 95 0 0 1 90912 542980 16192 3203712 0 0 2644 0 684 673 3 1 96 0 1 0 90912 540260 16220 3206480 0 0 2780 40 827 684 4 1 95 0 0 1 90912 537724 16224 3209032 0 0 2556 0 613 666 3 0 97 0 0 1 90912 534920 16224 3211840 0 0 2808 0 658 714 6 0 94 0 0 1 90912 532172 16224 3214596 0 0 2756 0 678 769 5 0 95 0 There's some other slowness with 7.4 as well. After running just fine for several hours pg starts to eat a LOT of cpu. Query plans are just like normally and pg_stat_activity shows nothing special. Disconnecting and reconnecting all clients seems to help (restarting web server). hw/sw configuration is something like this: Dual Xeon 2.8GHz with 4GB of memory. RAID is fast enuff. Linux 2.4 (Debian) Postgres is compiled using gcc 3.2, cflags: CFLAGS=-march=pentium4 -O3 -msse2 -mmmx |\__/| ( oo ) Kari Lavikka - tuner@bdb.fi - (050) 380 3808 __ooO( )Ooo_______ _____ ___ _ _ _ _ _ _ _ "" From pgsql-performance-owner@postgresql.org Sat Dec 13 18:57:51 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id AA64CD1B467 for ; Sat, 13 Dec 2003 22:57:49 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48980-02 for ; Sat, 13 Dec 2003 18:57:26 -0400 (AST) Received: from visionlink.org (mail.visionlink.org [208.139.207.159]) by svr1.postgresql.org (Postfix) with SMTP id 4D613D1B465 for ; Sat, 13 Dec 2003 18:57:17 -0400 (AST) Received: from [65.100.170.142] (HELO [192.168.1.201]) by visionlink.org (Stalker SMTP Server 1.8b8) with ESMTP id S.0002315554 for ; Sat, 13 Dec 2003 15:58:15 -0700 Mime-Version: 1.0 X-Sender: bohmer@mail.visionlink.org Message-Id: In-Reply-To: References: Date: Sat, 13 Dec 2003 15:57:04 -0700 To: pgsql-performance@postgresql.org From: Jeff Bohmer Subject: Re: Hardware suggestions for Linux/PGSQL server Content-Type: text/plain; charset="us-ascii" ; format="flowed" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/213 X-Sequence-Number: 5073 >Just one more piece of advice, you might want to look into a good battery >backed cache hardware RAID controller. They work quite well for heavily >updated databases. The more drives you throw at the RAID array the faster >it will be. I've seen this list often recommended such a setup. We'll probably get battery-backed write cache and start out with a 4 disk RAID 10 array. Then add more disks and change RAID 5 if more read performance is needed. Thanks, - Jeff -- Jeff Bohmer VisionLink, Inc. _________________________________ 303.402.0170 www.visionlink.org _________________________________ People. Tools. Change. Community. From pgsql-performance-owner@postgresql.org Sat Dec 13 19:01:05 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F2430D1B4D3 for ; Sat, 13 Dec 2003 23:01:03 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49422-02 for ; Sat, 13 Dec 2003 19:00:40 -0400 (AST) Received: from visionlink.org (mail.visionlink.org [208.139.207.159]) by svr1.postgresql.org (Postfix) with SMTP id D4A84D1B467 for ; Sat, 13 Dec 2003 19:00:31 -0400 (AST) Received: from [65.100.170.142] (HELO [192.168.1.201]) by visionlink.org (Stalker SMTP Server 1.8b8) with ESMTP id S.0002315566 for ; Sat, 13 Dec 2003 16:01:46 -0700 Mime-Version: 1.0 X-Sender: bohmer@mail.visionlink.org Message-Id: In-Reply-To: References: <3FD961C5.9070204@persistent.co.in> Date: Sat, 13 Dec 2003 16:00:32 -0700 To: pgsql-performance@postgresql.org From: Jeff Bohmer Subject: Re: Hardware suggestions for Linux/PGSQL server Content-Type: text/plain; charset="us-ascii" ; format="flowed" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/214 X-Sequence-Number: 5074 >Shridhar Daithankar wrote: >> >>FWIW, there are only two pieces of software that need 64bit aware >>for a typical server job. Kernel and glibc. Rest of the apps can do >>fine as 32 bits unless you are oracle and insist on outsmarting OS. >> >>In fact running 32 bit apps on 64 bit OS has plenty of advantages >>like effectively using the cache. Unless you need 64bit, going for >>64bit software is not advised. > >This is a good point. While doing research on this matter a few >months back, I saw comments by people testing 64-bit MySQL that some >operations would run faster and some slower due to the use of 64-bit >datatypes versus 32-bit. The best solution in the end is probably to >run 32-bit Postgres under a 64-bit kernel -- unless your DB tends to >have a lot of 64-bit datatypes. Thanks Shridhar and William, This advice has been very helpful. I would imagine a lot of folks are, or will soon be looking at 32- vs. 64-bit just for memory reasons and not 64-bit apps. - Jeff -- Jeff Bohmer VisionLink, Inc. _________________________________ 303.402.0170 www.visionlink.org _________________________________ People. Tools. Change. Community. From pgsql-performance-owner@postgresql.org Sun Dec 14 01:42:39 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F08F0D1B482 for ; Sun, 14 Dec 2003 05:42:38 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 79311-06 for ; Sun, 14 Dec 2003 01:42:22 -0400 (AST) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 63E3AD1B442 for ; Sun, 14 Dec 2003 01:42:09 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hBE5gLI21410; Sun, 14 Dec 2003 00:42:21 -0500 (EST) From: Bruce Momjian Message-Id: <200312140542.hBE5gLI21410@candle.pha.pa.us> Subject: Re: Tuning for mid-size server In-Reply-To: <20031021213408.GA12200@libertyrms.info> To: Andrew Sullivan Date: Sun, 14 Dec 2003 00:42:21 -0500 (EST) Cc: pgsql-performance@postgresql.org X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/215 X-Sequence-Number: 5075 Andrew Sullivan wrote: > On Tue, Oct 21, 2003 at 03:11:17PM -0600, scott.marlowe wrote: > > I think where it makes sense is when you have something like a report > > server where the result sets may be huge, but the parellel load is load, > > i.e. 5 or 10 users tossing around 100 Meg or more at time. > > In our case, we were noticing that truss showed an unbelievable > amount of time spent by the postmaster doing open() calls to the OS > (this was on Solaris 7). So we thought, "Let's try a 2G buffer > size." 2G was more than enough to hold the entire data set under > question. Once the buffer started to fill, even plain SELECTs > started taking a long time. The buffer algorithm is just not that > clever, was my conclusion. > > (Standard disclaimer: not a long, controlled test. It's just a bit > of gossip.) I know this is an old email, but have you tested larger shared buffers in CVS HEAD with Jan's new cache replacement policy? -- 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 Sun Dec 14 01:44:18 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3ADF5D1B442 for ; Sun, 14 Dec 2003 05:44:16 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 78697-07 for ; Sun, 14 Dec 2003 01:43:59 -0400 (AST) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id A6447D1B4D5 for ; Sun, 14 Dec 2003 01:43:46 -0400 (AST) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id hBE5i0v21938 for pgsql-performance@postgresql.org; Sun, 14 Dec 2003 00:44:00 -0500 (EST) From: Bruce Momjian Message-Id: <200312140544.hBE5i0v21938@candle.pha.pa.us> Subject: Update performance doc In-Reply-To: To: pgsql-performance@postgresql.org Date: Sun, 14 Dec 2003 00:44:00 -0500 (EST) X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/216 X-Sequence-Number: 5076 I have updated my hardware performance documentation to reflect the findings during the past few months on the performance list: http://candle.pha.pa.us/main/writings/pgsql/hw_performance/index.html Thanks. -- 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 Sun Dec 14 03:24:02 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9DF6BD1D8EB for ; Sun, 14 Dec 2003 07:24:01 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 86498-06 for ; Sun, 14 Dec 2003 03:23:45 -0400 (AST) Received: from flamingo.mail.pas.earthlink.net (flamingo.mail.pas.earthlink.net [207.217.120.232]) by svr1.postgresql.org (Postfix) with ESMTP id 553DDD1BB74 for ; Sun, 14 Dec 2003 03:23:32 -0400 (AST) Received: from hsa070.pool014.at101.earthlink.net ([216.249.85.70] helo=shadovitzcmptr) by flamingo.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 1AVQbM-0004Yv-00; Sat, 13 Dec 2003 23:23:40 -0800 Received: by localhost with Microsoft MAPI; Sat, 13 Dec 2003 23:22:05 -0800 Message-ID: <01C3C1CF.EC044640.david@shadovitz.com> From: David Shadovitz Reply-To: "david@shadovitz.com" To: "'pgsql-performance@postgresql.org'" Cc: 'Tom Lane' Subject: Re: Query plan - now what? Date: Sat, 13 Dec 2003 23:21:36 -0800 X-Mailer: Microsoft Internet E-mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=1.0 tagged_above=0.0 required=5.0 tests=BAYES_30, RCVD_IN_OSIRUSOFT_COM, UPPERCASE_25_50 X-Spam-Level: * X-Archive-Number: 200312/217 X-Sequence-Number: 5077 Here are my query and schema. The ERD is at http://dshadovi.f2o.org/pg_erd.jpg (sorry about its resolution). -David SELECT zbr.zebra_name , dog.dog_name , mnk.monkey_name , wrm.abbreviation || ptr.abbreviation as abbrev2 , whg.warthog_num , whg.color , rhn.rhino_name , der.deer_name , lin.designator , frg.frog_id , frg.sound_id , tgr.tiger_name , frg.leg_length , frg.jump_distance FROM frogs frg , deers der , warthogs whg , rhinos rhn , zebras zbr , dogs dog , monkeys mnk , worms wrm , parrots prt , giraffes grf , lions lin , tigers tgr WHERE 1 = 1 AND frg.deer_id = der.deer_id AND whg.whg_id = frg.frg_id AND frg.rhino_id = rhn.rhino_id AND zbr.zebra_id = dog.zebra_id AND dog.dog_id = mky.dog_id AND mky.dog_id = whg.dog_id AND mky.monkey_num = whg.monkey_num AND whg.worm_id = wrm.worm_id AND whg.parrot_id = prt.parrot_id AND prt.beak = 'L' AND frg.frog_id = grf.frog_id AND grf.lion_id = lin.lion_id AND frg.tiger_id = tgr.tiger_id ; CREATE TABLE zebras ( zebra_id INTEGER NOT NULL, zebra_name VARCHAR(25), PRIMARY KEY (zebra_id), UNIQUE (zebra_name)); CREATE TABLE dogs ( zebra_id INTEGER NOT NULL, dog_id INTEGER NOT NULL, dog_name VARCHAR(25), FOREIGN KEY (zebra_id) REFERENCES zebras (zebra_id), PRIMARY KEY (dog_id), UNIQUE (dog_name, dog_num)); CREATE TABLE monkeys ( dog_id INTEGER NOT NULL, monkey_num INTEGER, monkey_name VARCHAR(25), PRIMARY KEY (dog_id, monkey_num), FOREIGN_KEY (dog_id) REFERENCES dogs (dog_id)); CREATE INDEX mnk_dog_id_idx ON monkeys (dog_id); CREAIE INDEX mnk_mnk_num_idx ON monkeys (monkey_num); CREATE TABLE warthogs ( warthog_id INTEGER NOT NULL, warthog_num INTEGER, color VARCHAR(25) NOT NULL, dog_id INTEGER NOT NULL, monkey_num INTEGER NOT NULL, parrot_id INTEGER, beak CHAR(l), worm_id INTEGER, PRIMARY KEY (warthog_id), FOREIGN KEY (parrot_id, beak) REFERENCES parrots (parrot_id, beak) FOREIGN KEY (dog_id, monkey_num) REFERENCES monkeys (dog_id, monkey_nun) FOREIGN KEY (worm_id) REFERENCES worms (worm_id)); CREATE UNIQUE INDEX whg_whg_id_idx ON warthogs (warthog_id) CREATE INDEX whg_dog_id_idx ON warthogs (dog_id); CREATE INDEX whg_mnk_num_idx ON warthogs (monkey_num) CREATE INDEX whg_wrm_id_idx ON warthogs (worm_id); CREATE INDEX IDX_warthogs_1 ON warthogs (monkey_num, dog_id) CREATE INDEX lOX warthogs_2 ON warthogs (beak, parrot_id); CREATE TABLE worms ( worm_id INTEGER NOT NULL, abbreviation CHAR(l), PRIMARY KEY worm_id)); CREATE TABLE parrots ( parrot_id INTEGER NOT NULL, beak CHAR(1) NOT NULL, abbreviation CHAR(1), PRIMARY KEY (parrot_id, beak)); CREATE INDEX prt_prt_id_idx ON parrots (parrot_id) CREATE INDEX prt_beak_idx ON parrots (beak): CREATE TABLE deers ( deer_id INTEGER NOT NULL, deer_name VARCHAR(40), PRIMARY KEY (deer_id)); CREATE UNIQUE INDEX der_der_id_unq_idx ON deers (deer_id); CREATE TABLE rhinos ( rhino_id INTEGER NOT NULL, rhino_name VARCHAR(255), CONSTRAINT rhn_rhn_name_unique UNIQUE, CONSTRAINT PK_rhn PRIMARY KEY (rhino_id)); CREATE UNIQUE INDEX rhn_rhn_id_unq_idx ON rhinos (rhino_id); CREATE TABLE tigers ( tiger_id INTEGER NOT NULL, tiger_name VARCHAR(255), PRIMARY KEY (tiger_id)); CREATE UNIQUE INDEX tgr_tgr_id_unq_idx ON tigers (tiger_id); CREATE TABLE frogs ( frog_id INTEGER NOT NULL, warthog_id INTEGER NOT NULL, rhino_id INTEGER NOT NULL, deer_id INTEGER NOT NULL, sound_id INTEGER, tiger_id INTEGER, leg_length VARCHAR(255), jump_distance VARCHAR(lOO), PRIMARY KEY (frog_id)); ALTER TABLE frogs ADD FOREIGN KEY (warthog_id) REFERENCES warthogs (warthog_id), ALTER TABLE frogs ADD FOREIGN KEY (rhino_id) REFERENCES rhinos (rhino_id); ALTER TABLE frogs ADD FOREIGN KEY (deer id) REFERENCES deers (deer_id) ALTER TABLE frogs ADD FOREIGN KEY (sound_id) REFERENCES sounds (sound id); ALTER TABLE frogs ADD FOREIGN KEY (tiger_id) REFERENCES tigers (tiger_id); CREATE UNIQUE INDEX frg_frg_id_unq_idx ON frogs (frog_id); CREATE UNIQUE INDEX frg_w_r_d_t_unq_idx ON frogs (warthog_id, rhino_id, deer_id, tiger_id); CREATE INDEX frg_whg_id_idx ON frogs (warthog_id); CREATE INDEX frg rhn_id_idx ON frogs (rhino_id); CREATE INDEX frg_der_id_idx ON frogs (deer_id); CREATE INDEX frg_snd_id_idx ON frogs (sound_id); CREATE INDEX frg_tgr_id_idx ON frogs (tiger_id); CREATE TABLE lions ( lion_id INTEGER NOT NULL, deer_id INTEGER, PRIMARY KEY (lion_id)); CREATE UNIQUE INDEX lin_lin_id_unq_idx ON lions (lion_id); CREATE TABLE frogs_lions ( frog_id INTEGER NOT NULL, lion_id INTEGER NOT NULL, PRIMARY KEY (frog_id, lion_id)); ALTER TABLE frogs_lions ADD FOREIGN KEY (lion_id) REFERENCES lions (lion_id); ALTER TABLE frogs_lions ADD FOREIGN KEY (frog id) REFERENCES frogs (frog_id); CREATE UNIQUE INDEX frg_lin_frg_id_lin_id_unq_idx ON frogs_lions (frog_id, lion_id); CREATE INDEX frg_lin_lin_id_idx ON frogs_lions (lion_id); CREATE INDEX frg_lin_frg_id_idx ON frogs_lions (frog_id); From pgsql-performance-owner@postgresql.org Sun Dec 14 05:55:14 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B5DE9D1DB0B for ; Sun, 14 Dec 2003 09:55:10 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97024-05 for ; Sun, 14 Dec 2003 05:54:55 -0400 (AST) Received: from kiwi.iasi.rdsnet.ro (kiwi.iasi.rdsnet.ro [213.157.176.3]) by svr1.postgresql.org (Postfix) with ESMTP id 0E0AED1DAE9 for ; Sun, 14 Dec 2003 05:54:41 -0400 (AST) Received: from blackblue.iasi.rdsnet.ro (blackblue.iasi.rdsnet.ro [213.157.176.7]) by kiwi.iasi.rdsnet.ro (8.12.9/8.12.9) with ESMTP id hBE9sfrv013841 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sun, 14 Dec 2003 11:54:41 +0200 Received: from blackblue.iasi.rdsnet.ro (localhost.localdomain [127.0.0.1]) by blackblue.iasi.rdsnet.ro (8.12.9/8.12.9) with ESMTP id hBE9sfxd015236 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sun, 14 Dec 2003 11:54:41 +0200 Received: from localhost (Mituc@localhost) by blackblue.iasi.rdsnet.ro (8.12.9/8.12.9/Submit) with ESMTP id hBE9sfZL015232; Sun, 14 Dec 2003 11:54:41 +0200 X-Authentication-Warning: blackblue.iasi.rdsnet.ro: Mituc owned process doing -bs Date: Sun, 14 Dec 2003 11:54:41 +0200 (EET) From: Tarhon-Onu Victor To: Kari Lavikka Cc: pgsql-performance@postgresql.org Subject: Re: a lot of problems with pg 7.4 In-Reply-To: Message-ID: References: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/218 X-Sequence-Number: 5078 On Sat, 13 Dec 2003, Kari Lavikka wrote: > I evaluated pg 7.4 on our development server and it looked just fine > but performance with production loads seems to be quite poor. Most of > performance problems are caused by nonsensical query plans but there's > also some strange slowness that I can't locate. I had the same problem. I use Fedora Core 1 and after I updated from 7.4RC1/7.4RC2 (I build my own RPMs) to 7.4 using the binary RPMs from a mirror site and sometimes I had to restart postmaster to make something work. I rebuilt the src.rpm from current rawhide (7.4-5) and now everything is ok. The guys from redhat/fedora also add some patches (rpm-pgsql-7.4.patch seems to be the most important, the rest seem to be for a proper compile) but I didn't have the time to test if the loss of performance is because in the original binary RPMs from postgresql.org the patch(es) is(are) not present, because of the compiler and optflags used to build the RPMs are not chosed well or something else. I used gcc 3.3.2 (from FC1 distro) and the following optflags: - On a P4 machine: optflags: i686 -O2 -g -march=pentium4 -msse2 -mfpmath=sse -fomit-frame-pointer -fforce-addr -fforce-mem -maccumulate-outgoing-args -finline-limit=2048 - On a Celeron Tualatin: optflags: i686 -O2 -g -march=pentium3 -msse -mfpmath=sse -fomit-frame-pointer -fforce-addr -fforce-mem -maccumulate-outgoing-args -finline-limit=2048 So, if you use the original binaries from postgresql.org try to recompile from sources setting CFLAGS and CXXFLAGS to proper values (maybe -msse2 -mfpmath=sse are not a good choice, you can try removing them). If not then review your postgresql configuration (buffers, memory, page cost, etc), because 7.4 seems to be faster than 7.3 and there is no reason for it to run slower on your system. -- Any views or opinions presented within this e-mail are solely those of the author and do not necessarily represent those of any company, unless otherwise expressly stated. From pgsql-performance-owner@postgresql.org Sun Dec 14 13:14:35 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A0491D1DBBB for ; Sun, 14 Dec 2003 17:14:33 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28694-02 for ; Sun, 14 Dec 2003 13:14:04 -0400 (AST) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id B9B6AD1B47D for ; Sun, 14 Dec 2003 13:14:02 -0400 (AST) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id BE9BF8DD3; Sun, 14 Dec 2003 18:14:00 +0100 (CET) Date: Sun, 14 Dec 2003 18:14:00 +0100 (CET) From: Dennis Bjorklund To: Kari Lavikka Cc: pgsql-performance@postgresql.org Subject: Re: a lot of problems with pg 7.4 In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/219 X-Sequence-Number: 5079 On Sat, 13 Dec 2003, Kari Lavikka wrote: > I evaluated pg 7.4 on our development server and it looked just fine > but performance with production loads seems to be quite poor. Most of > performance problems are caused by nonsensical query plans Some of the estimates that pg made in the plans you showed was way off. I assume you have run VACUUM ANALYZE recently? If that does not help maybe you need to increaste the statistics gathering on some columns so that pg makes better estimates. With the wrong statistics it's not strange that pg chooses bad plans. -- /Dennis From pgsql-performance-owner@postgresql.org Sun Dec 14 13:56:31 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 894E6D1DC04 for ; Sun, 14 Dec 2003 17:56:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28518-06 for ; Sun, 14 Dec 2003 13:56:00 -0400 (AST) Received: from mail.libertyrms.com (unknown [209.167.124.227]) by svr1.postgresql.org (Postfix) with ESMTP id 01D27D1DB9D for ; Sun, 14 Dec 2003 13:56:00 -0400 (AST) Received: from [10.1.2.146] (helo=dba3.int.libertyrms.info) by mail.libertyrms.com with esmtp (Exim 3.22 #3 (Debian)) id 1AVaT6-0003PR-00; Sun, 14 Dec 2003 12:55:48 -0500 Received: by dba3.int.libertyrms.info (Postfix, from userid 1019) id 49F4C138CA; Sun, 14 Dec 2003 12:55:48 -0500 (EST) Date: Sun, 14 Dec 2003 12:55:48 -0500 From: Andrew Sullivan To: Bruce Momjian Cc: pgsql-performance@postgresql.org Subject: Re: Tuning for mid-size server Message-ID: <20031214175548.GA9375@libertyrms.info> Mail-Followup-To: Andrew Sullivan , Bruce Momjian , pgsql-performance@postgresql.org References: <20031021213408.GA12200@libertyrms.info> <200312140542.hBE5gLI21410@candle.pha.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200312140542.hBE5gLI21410@candle.pha.pa.us> User-Agent: Mutt/1.5.4i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/220 X-Sequence-Number: 5080 On Sun, Dec 14, 2003 at 12:42:21AM -0500, Bruce Momjian wrote: > > I know this is an old email, but have you tested larger shared buffers > in CVS HEAD with Jan's new cache replacement policy? Not yet. It's on our TODO list, for sure, because the consequences of relying too much on the filesystem buffers under certain perverse loads is lousy database performance _precisely_ when we need it. I expect some testing of this type some time in January. A -- ---- Andrew Sullivan 204-4141 Yonge Street Afilias Canada Toronto, Ontario Canada M2P 2A8 +1 416 646 3304 x110 From pgsql-admin-owner@postgresql.org Sun Dec 14 18:53:04 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6EF3CD1B524 for ; Sun, 14 Dec 2003 22:53:02 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53359-03 for ; Sun, 14 Dec 2003 18:52:35 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id E7932D1B45B for ; Sun, 14 Dec 2003 18:52:33 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBEMqZ19003271; Sun, 14 Dec 2003 17:52:36 -0500 (EST) To: Neil Conway Cc: aspire420@hotpop.com, pgsql-admin@postgresql.org, pgsql-performance@postgresql.org Subject: Re: [PERFORM] Tables Without OIDS and its effect In-reply-to: <87he05pr8i.fsf@mailbox.samurai.com> References: <3FD9E916.8050005@sancharnet.in> <87he05pr8i.fsf@mailbox.samurai.com> Comments: In-reply-to Neil Conway message dated "Fri, 12 Dec 2003 18:10:21 -0500" Date: Sun, 14 Dec 2003 17:52:35 -0500 Message-ID: <3270.1071442355@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/185 X-Sequence-Number: 11692 Neil Conway writes: > BTW, we intend to phase out the use of OIDs for user tables in the > long term. I don't believe anyone has proposed removing the facility altogether. There's a big difference between making the default behavior be not to have OIDs and removing the ability to have OIDs. regards, tom lane From pgsql-admin-owner@postgresql.org Mon Dec 15 14:42:33 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0A2F1D1B48A; Mon, 15 Dec 2003 02:10:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69240-09; Sun, 14 Dec 2003 22:10:16 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id EEDABD1B4C6; Sun, 14 Dec 2003 22:10:13 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 5BEB2218C; Sun, 14 Dec 2003 21:10:16 -0500 (EST) To: Tom Lane Cc: aspire420@hotpop.com, pgsql-admin@postgresql.org, pgsql-performance@postgresql.org Subject: Re: [PERFORM] Tables Without OIDS and its effect From: Neil Conway In-Reply-To: <3270.1071442355@sss.pgh.pa.us> (Tom Lane's message of "Sun, 14 Dec 2003 17:52:35 -0500") References: <3FD9E916.8050005@sancharnet.in> <87he05pr8i.fsf@mailbox.samurai.com> <3270.1071442355@sss.pgh.pa.us> Date: Sun, 14 Dec 2003 21:10:15 -0500 Message-ID: <87k74yg7aw.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/207 X-Sequence-Number: 11713 Tom Lane writes: > I don't believe anyone has proposed removing the facility > altogether. There's a big difference between making the default > behavior be not to have OIDs and removing the ability to have OIDs. Right, that's what I had meant to say. Sorry for the inaccuracy. -Neil From pgsql-performance-owner@postgresql.org Mon Dec 15 00:17:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 23FFAD1B475 for ; Mon, 15 Dec 2003 04:17:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 82579-09 for ; Mon, 15 Dec 2003 00:17:15 -0400 (AST) Received: from xyzzy.dhs.org (H98.C214.tor.velocet.net [216.138.214.98]) by svr1.postgresql.org (Postfix) with ESMTP id 53D77D1B44A for ; Mon, 15 Dec 2003 00:17:15 -0400 (AST) Received: from xyzzy.dhs.org (unknown [192.168.1.74]) by xyzzy.dhs.org (Postfix) with ESMTP id D27BD1B25E; Sun, 14 Dec 2003 23:17:13 -0500 (EST) Message-ID: <3FDD35CF.6080103@xyzzy.dhs.org> Date: Sun, 14 Dec 2003 23:17:19 -0500 From: "Andrew G. Hammond" User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031013 Thunderbird/0.3 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Jeff Bohmer Cc: pgsql-performance@postgresql.org Subject: Re: Hardware suggestions for Linux/PGSQL server 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 postgresql.org X-Archive-Number: 200312/223 X-Sequence-Number: 5083 I don't know what your budget is, but there are now 10k RPM SATA 150 drives on the market. Their price/performance is impressive. You may want to consider going with a bunch of these instead of SCSI disks (more spindles vs. faster spindles). 3ware makes a hardware raid card that can drive up to 12 SATA disks. I have been told by a few people who have used it that the linux driver is very solid. Drew Jeff Bohmer wrote: >> Just one more piece of advice, you might want to look into a good >> battery >> backed cache hardware RAID controller. They work quite well for heavily >> updated databases. The more drives you throw at the RAID array the >> faster >> it will be. > > > I've seen this list often recommended such a setup. We'll probably > get battery-backed write cache and start out with a 4 disk RAID 10 > array. Then add more disks and change RAID 5 if more read performance > is needed. > > Thanks, > - Jeff From pgsql-performance-owner@postgresql.org Mon Dec 15 01:30:17 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1992CD1B4C3 for ; Mon, 15 Dec 2003 05:30:15 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89743-09 for ; Mon, 15 Dec 2003 01:29:47 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 48F42D1B475 for ; Mon, 15 Dec 2003 01:29:46 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBF5TjU6033273 for ; Mon, 15 Dec 2003 05:29:45 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBF5Kcla022421 for pgsql-performance@postgresql.org; Mon, 15 Dec 2003 05:20:38 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Hardware suggestions for Linux/PGSQL server Date: Mon, 15 Dec 2003 00:14:29 -0500 Organization: cbbrowne Computing Inc Lines: 23 Message-ID: References: <3FDD35CF.6080103@xyzzy.dhs.org> 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/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:H/KAfwMrnYKY8ZSp0EHFd9u1trI= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/224 X-Sequence-Number: 5084 In the last exciting episode, drew@xyzzy.dhs.org ("Andrew G. Hammond") wrote: > I don't know what your budget is, but there are now 10k RPM SATA 150 > drives on the market. Their price/performance is impressive. You may > want to consider going with a bunch of these instead of SCSI disks > (more spindles vs. faster spindles). 3ware makes a hardware raid > card that can drive up to 12 SATA disks. I have been told by a few > people who have used it that the linux driver is very solid. We got a couple of those in for testing purposes; when opportunity presents itself, I'll have to check to see if they are any more honest about commits than traditional IDE drives. If they still "lie" the same way IDE drives do, it is entirely possible that they are NOT nearly as impressive as you presently imagine. It's not much good if they're "way fast" if you can't trust them to actually store data when they claim it is stored... -- (reverse (concatenate 'string "gro.gultn" "@" "enworbbc")) http://www.ntlug.org/~cbbrowne/lisp.html "Much of this software was user-friendly, meaning that it was intended for users who did not know anything about computers, and furthermore had absolutely no intention whatsoever of learning." -- A. S. Tanenbaum, "Modern Operating Systems, ch 1.2.4" From pgsql-performance-owner@postgresql.org Mon Dec 15 12:38:23 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 28E1CD1DCF5 for ; Mon, 15 Dec 2003 16:38:21 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71518-07 for ; Mon, 15 Dec 2003 12:37:58 -0400 (AST) Received: from visionlink.org (mail.visionlink.org [208.139.207.159]) by svr1.postgresql.org (Postfix) with SMTP id 7E35BD1B461 for ; Mon, 15 Dec 2003 12:37:51 -0400 (AST) Received: from [65.100.170.142] (HELO [192.168.1.201]) by visionlink.org (Stalker SMTP Server 1.8b8) with ESMTP id S.0002324650 for ; Mon, 15 Dec 2003 09:39:01 -0700 Mime-Version: 1.0 X-Sender: bohmer@mail.visionlink.org Message-Id: In-Reply-To: References: <3FDD35CF.6080103@xyzzy.dhs.org> Date: Mon, 15 Dec 2003 09:37:29 -0700 To: pgsql-performance@postgresql.org From: Jeff Bohmer Subject: Re: Hardware suggestions for Linux/PGSQL server Content-Type: text/plain; charset="us-ascii" ; format="flowed" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/225 X-Sequence-Number: 5085 >In the last exciting episode, drew@xyzzy.dhs.org ("Andrew G. Hammond") wrote: >> I don't know what your budget is, but there are now 10k RPM SATA 150 >> drives on the market. Their price/performance is impressive. You may >> want to consider going with a bunch of these instead of SCSI disks >> (more spindles vs. faster spindles). 3ware makes a hardware raid >> card that can drive up to 12 SATA disks. I have been told by a few >> people who have used it that the linux driver is very solid. > >We got a couple of those in for testing purposes; when opportunity >presents itself, I'll have to check to see if they are any more honest >about commits than traditional IDE drives. > >If they still "lie" the same way IDE drives do, it is entirely >possible that they are NOT nearly as impressive as you presently >imagine. It's not much good if they're "way fast" if you can't trust >them to actually store data when they claim it is stored... We lost data because of this very problem when a UPS didn't signal the shut down before it ran out of juice. Here's an excellent explanation of the problem: http://archives.postgresql.org/pgsql-general/2003-10/msg01343.php This post indicates that SATA drives still have problems, but a new ATA standard might fix things in the future: http://archives.postgresql.org/pgsql-general/2003-10/msg01395.php SATA RAID is a good option for a testing server, though. - Jeff -- Jeff Bohmer VisionLink, Inc. _________________________________ 303.402.0170 www.visionlink.org _________________________________ People. Tools. Change. Community. From pgsql-admin-owner@postgresql.org Mon Dec 15 13:01:56 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5A351D1DCFD; Mon, 15 Dec 2003 17:01:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73663-10; Mon, 15 Dec 2003 13:01:22 -0400 (AST) Received: from host2.hostseguro.com (66-98-192-97.hostseguro.com [66.98.192.97]) by svr1.postgresql.org (Postfix) with ESMTP id 6BC63D1B430; Mon, 15 Dec 2003 13:01:16 -0400 (AST) Received: from cpanel by host2.hostseguro.com with local (Exim 4.24) id 1AVw5r-0006Xf-HB; Mon, 15 Dec 2003 15:01:15 -0200 Received: from 200-180-185-142.paemt7003.dsl.brasiltelecom.net.br (200-180-185-142.paemt7003.dsl.brasiltelecom.net.br [200.180.185.142]) by sistemica.info (IMP) with HTTP for ; Mon, 15 Dec 2003 15:01:15 -0200 Message-ID: <1071507675.3fdde8db77a2e@sistemica.info> Date: Mon, 15 Dec 2003 15:01:15 -0200 From: Rhaoni Chiu Pereira To: PostgreSQL ADMIN , PostgreSQL Performance Subject: Unsubscribe MIME-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit User-Agent: Internet Messaging Program (IMP) 3.2.1 X-Originating-IP: 200.180.185.142 X-MailScanner-Information: Verificado pelo McAfee VirusScan / Scanned by McAfee VirusScan X-MailScanner: Nao infectado / Found to be clean X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - host2.hostseguro.com X-AntiAbuse: Original Domain - postgresql.org X-AntiAbuse: Originator/Caller UID/GID - [32001 32001] / [47 12] X-AntiAbuse: Sender Address Domain - sistemica.info X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/197 X-Sequence-Number: 11704 From pgsql-performance-owner@postgresql.org Mon Dec 15 14:06:20 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CFEADD1DCF5 for ; Mon, 15 Dec 2003 18:06:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89394-01 for ; Mon, 15 Dec 2003 14:05:56 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id A7BA0D1DDDB for ; Mon, 15 Dec 2003 14:05:49 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBFI5p19013829; Mon, 15 Dec 2003 13:05:52 -0500 (EST) To: LIANHE SHAO Cc: pgsql-performance Subject: Re: Index problem or function problem? In-reply-to: <4595eb458230.4582304595eb@jhmimail.jhmi.edu> References: <4595eb458230.4582304595eb@jhmimail.jhmi.edu> Comments: In-reply-to LIANHE SHAO message dated "Tue, 09 Dec 2003 23:24:19 +0000" Date: Mon, 15 Dec 2003 13:05:51 -0500 Message-ID: <13828.1071511551@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/227 X-Sequence-Number: 5087 LIANHE SHAO writes: > PGA=> explain select ei.expid, er.geneid, > er.sampleid, ei.annotation, si.samplename, > ei.title as exp_name, aaa.chip, > aaa.sequence_derived_from as accession_number, > aaa.gene_symbol, aaa.title as gene_function, > er.exprs, er.mas5exprs from expressiondata er, > experimentinfo ei, sampleinfo si, > affy_array_annotation aaa where exists (select > distinct ei.expid from experimentinfo) and > ei.annotation = aaa.chip and (lower (aaa.title) > like '%mif%' or lower(aaa.sequence_description) like > '%mif%') and exists (select distinct ei.annotation > from experimentinfo) and ei.expid = er.expid and > er.expid = si.expid and er.sampleid = si.sampleid > and er.geneid = aaa.probeset_id order by si.sampleid > limit 20; What is the purpose of the EXISTS() clauses? They are almost surely not doing what you intended, because AFAICS they are just an extremely expensive means of producing a constant-TRUE result. In exists (select distinct ei.expid from experimentinfo) "ei.expid" is an outer reference, which will necessarily be the same value over all rows of the sub-select. After computing this same value for every row of experimentinfo, the system performs a DISTINCT operation (sort + unique, not cheap) ... and then all it checks for is whether at least one row was produced, which means the DISTINCT operation was completely unnecessary. The only way the EXISTS could return false is if experimentinfo were empty, but if it were so then the outer FROM would've produced no rows and we'd not have got to WHERE anyway. I'm not sure why you get a worse plan for the simpler variant of the query; it would help to see EXPLAIN ANALYZE rather than EXPLAIN output. But it's not worth trying to improve the performance until you are calculating correct answers, and I suspect the above is not doing what you are after at all. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Dec 15 14:22:20 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F3E5ED1B4CE for ; Mon, 15 Dec 2003 18:22:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88066-06 for ; Mon, 15 Dec 2003 14:21:55 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id CD1D8D1DCDA for ; Mon, 15 Dec 2003 14:21:48 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBFILl19013899; Mon, 15 Dec 2003 13:21:48 -0500 (EST) To: Hartmut Raschick Cc: pgsql-performance@postgresql.org, Rod Taylor Subject: Re: TRUNCATE veeeery slow compared to DELETE in 7.4 In-reply-to: <3FD99CDA.8161D80A@ke-elektronik.de> References: <3FD484B6.EF39A92F@ke-elektronik.de> <200312100918.02501.josh@agliodbs.com> <8000.1071086054@sss.pgh.pa.us> <3FD99CDA.8161D80A@ke-elektronik.de> Comments: In-reply-to Hartmut Raschick message dated "Fri, 12 Dec 2003 11:47:54 +0100" Date: Mon, 15 Dec 2003 13:21:47 -0500 Message-ID: <13898.1071512507@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/228 X-Sequence-Number: 5088 Hartmut Raschick writes: > [ TRUNCATE is much slower in 7.4 than in 7.3 ] After looking into this, I think this is because when Rod Taylor reimplemented TRUNCATE to make it transaction-safe, he essentially turned it into a variant of CLUSTER. It is slow because it is creating and deleting dummy tables and indexes. I think this is not really necessary and it could be done better while still being transaction-safe. All we really need is to create a new empty table file, update the table's pg_class row with the new relfilenode, mark the old file for deletion, and then run REINDEX TABLE (which will perform similar shenanigans with the indexes). regards, tom lane From pgsql-admin-owner@postgresql.org Mon Dec 15 18:55:15 2003 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6E2A7D1CAFD for ; Mon, 15 Dec 2003 22:55:07 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 30267-07 for ; Mon, 15 Dec 2003 18:54:46 -0400 (AST) Received: from smtp.covadmail.net (mx05.covadmail.net [63.65.120.65]) by svr1.postgresql.org (Postfix) with SMTP id 44052D1C980 for ; Mon, 15 Dec 2003 18:54:38 -0400 (AST) Received: (covad.net 2977 invoked from network); 15 Dec 2003 22:54:41 -0000 Received: from unknown (HELO meer.net) (jhefner@67.100.190.154) by sun-qmail15 with SMTP; 15 Dec 2003 22:54:40 -0000 Message-ID: <3FDE3BAB.35563F@meer.net> Date: Mon, 15 Dec 2003 17:54:35 -0500 From: Jeremy Hefner X-Mailer: Mozilla 4.8 [en] (Windows NT 5.0; U) X-Accept-Language: en MIME-Version: 1.0 Cc: PostgreSQL ADMIN , PostgreSQL Performance Subject: unsubscribe References: <1071507675.3fdde8db77a2e@sistemica.info> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=EMAIL_ATTRIBUTION, QUOTED_EMAIL_TEXT, RCVD_IN_OSIRUSOFT_COM, REFERENCES, REPLY_WITH_QUOTES X-Spam-Level: X-Archive-Number: 200312/211 X-Sequence-Number: 11717 unsubscribe Rhaoni Chiu Pereira wrote: > > ---------------------------(end of broadcast)--------------------------- > TIP 7: don't forget to increase your free space map settings From pgsql-performance-owner@postgresql.org Tue Dec 16 11:42:01 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 543B8D1DBF5 for ; Tue, 16 Dec 2003 15:41:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55005-08 for ; Tue, 16 Dec 2003 11:41:32 -0400 (AST) Received: from ml370.omarnet.it (host163-247.pool62110.interbusiness.it [62.110.247.163]) by svr1.postgresql.org (Postfix) with ESMTP id D6EC2D1DBC4 for ; Tue, 16 Dec 2003 11:41:25 -0400 (AST) Received: by ml370.omarnet.it with Internet Mail Service (5.5.2657.72) id ; Tue, 16 Dec 2003 16:40:04 +0100 Message-ID: <7965BF282720D8118ED70050BA4E75760A9658@ml370.omarnet.it> From: Claudia D'amato To: pgsql-performance@postgresql.org Subject: Postgres respond after toomany times to a query view Date: Tue, 16 Dec 2003 16:40:03 +0100 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2657.72) Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C3C3EA.DF546700" X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.6 tagged_above=0.0 required=5.0 tests=BAYES_30, HTML_20_30 X-Spam-Level: X-Archive-Number: 200312/229 X-Sequence-Number: 5089 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. ------_=_NextPart_001_01C3C3EA.DF546700 Content-Type: text/plain; charset="iso-8859-1" Hi, I am developing a program using postgres and linux like operating system. My problem is this: I have a quite complicated view with roughly 10000 record. When I execute a simple query like this "select * from myview" postgres respond after 50 - 55 minutes roughly. I hope that someone can help me with some suggestion about reason of this behavior and some solution to reduce time ti have results. Thank you for your attentions and I hope to receive some feedback as soon as possible ------_=_NextPart_001_01C3C3EA.DF546700 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Postgres respond after toomany times to a query view

Hi, I am developing a program using postgr= es and linux like operating system. My problem is this:
I have a quite complicated view with roug= hly 10000 record. When I execute a simple query like this
        "select * from myview"
postgres respond after 50 - 55 minutes ro= ughly. I hope that someone can help me with some suggestion about reason of= this behavior and some solution to reduce time ti have results. Thank you = for your attentions and I hope to receive some feedback as soon as possible=

------_=_NextPart_001_01C3C3EA.DF546700-- From pgsql-performance-owner@postgresql.org Tue Dec 16 12:57:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F0E9FD1B449 for ; Tue, 16 Dec 2003 16:56:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69282-04 for ; Tue, 16 Dec 2003 12:56:28 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id D8431D1B46D for ; Tue, 16 Dec 2003 12:56:27 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hBGGsL7N021407; Tue, 16 Dec 2003 09:54:22 -0700 (MST) Date: Tue, 16 Dec 2003 09:37:28 -0700 (MST) From: "scott.marlowe" To: Kari Lavikka Cc: Subject: Re: a lot of problems with pg 7.4 In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/230 X-Sequence-Number: 5090 On Sat, 13 Dec 2003, Kari Lavikka wrote: > > Hi! > > We have been running a rather busy website using pg 7.3 as the database. > Peak hitrate is something like 120 request / second without images and > other static stuff. The site is a sort of image gallery for IRC users. > > I evaluated pg 7.4 on our development server and it looked just fine > but performance with production loads seems to be quite poor. Most of > performance problems are caused by nonsensical query plans but there's > also some strange slowness that I can't locate. Have you analyzed your database since putting the new data into it? Also, you might need to increase your statistics target before analyzing to get proper results as well. From pgsql-performance-owner@postgresql.org Tue Dec 16 17:15:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 62F36D1B567 for ; Tue, 16 Dec 2003 17:06:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71132-05 for ; Tue, 16 Dec 2003 13:06:26 -0400 (AST) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id A4260D1B579 for ; Tue, 16 Dec 2003 13:06:25 -0400 (AST) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id hBGH685f030864 for ; Tue, 16 Dec 2003 12:06:10 -0500 Reply-To: From: "Nick Fankhauser - Doxpop" To: "Pgsql-Performance@Postgresql. Org" Subject: Nested loop question Date: Tue, 16 Dec 2003 12:06:20 -0500 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 V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/237 X-Sequence-Number: 5097 Hi- I'm trying to optimize a query that I *think* should run very fast. Essentially, I'm joining two tables that have very selective indexes and constraining the query on an indexed field. (There's a third small lookup table in the mix, but it doesn't really affect the bottom line.) actor is a table containing roughly 3 million rows with an index on actor_full_name_uppercase and a unique index on actor_id. actor_summary also contains roughly 3 million rows. Its PK is a unique combined index on (actor_id, county_id, case_disp_global_code). The vast majority of the rows in actor correspond to a single row in actor_summary I'd estimate this at 95% or more. The remaining actors with multiple records generally have two corresponding rows in actor summary. Actor summary was created as a performance enhancer, where we can store some pre-calculated values such as the number of court cases an actor is involved in. The constraint is applied first, with reasonable speed. In the example below, it takes about 15 seconds to gather the matches in actor. I'm unsure what is happening next. I notice that an index scan is occurring on actor_summary_pk, with an "actual time" of 9.15, but then it looks like a nested loop occurs at the next level to join these tables. Does this mean that each probe of the actor_summary index will take 9.15 msec, but the nested loop is going to do this once for each actor_id? The nested loop appears to be where most of my time is going, so I'm focusing on this area, but don't know if there is a better approach to this join. Is there a more efficient means than a nested loop to handle such a join? Would a different method be chosen if there was exactly one row in actor_summary for every row in actor? -Nick The query & explain analyze: alpha=# alpha=# alpha=# explain analyze alpha-# select alpha-# min(actor.actor_id) as actor_id, alpha-# min(actor.actor_entity_type) as actor_entity_type, alpha-# min(actor.role_class_code) as role_class_code, alpha-# min(actor.actor_full_name) as actor_full_name, alpha-# min(actor.actor_person_date_of_birth) as actor_person_date_of_birth, alpha-# min(actor.actor_entity_acronym) as actor_entity_acronym, alpha-# min(actor.actor_person_last_name) as actor_person_last_name, alpha-# min(actor.actor_person_first_name) as actor_person_first_name, alpha-# min(actor.actor_person_middle_name) as actor_person_middle_name, alpha-# min(actor.actor_person_name_suffix) as actor_person_name_suffix, alpha-# min(actor.actor_person_place_of_birth) as actor_person_place_of_birth, alpha-# min(actor.actor_person_height) as actor_person_height, alpha-# min(actor.actor_person_height_unit) as actor_person_height_unit, alpha-# min(actor.actor_person_weight) as actor_person_weight, alpha-# min(actor.actor_person_weight_unit) as actor_person_weight_unit, alpha-# min(actor.actor_person_ethnicity) as actor_person_ethnicity, alpha-# min(actor.actor_person_citizenship_count) as actor_person_citizenship_count, alpha-# min(actor.actor_person_hair_color) as actor_person_hair_color, alpha-# min(actor.actor_person_scars_marks_tatto) as actor_person_scars_marks_tatto, alpha-# min(actor.actor_person_marital_status) as actor_person_marital_status, alpha-# min(actor.actor_alias_for_actor_id) as actor_alias_for_actor_id, alpha-# min(to_char(data_source.source_last_update, 'MM/DD/YYYY HH12:MI AM TZ')) as last_update, alpha-# min(actor_summary.single_case_public_id) as case_public_id, alpha-# min(actor_summary.single_case_id) as case_id, alpha-# sum(actor_summary.case_count)as case_count alpha-# from alpha-# actor, alpha-# actor_summary, alpha-# data_source alpha-# where alpha-# actor.actor_id = actor_summary.actor_id alpha-# and data_source.source_id = actor.source_id alpha-# and actor_full_name_uppercase like upper('sanders%') alpha-# group by alpha-# actor.actor_id alpha-# order by alpha-# min(actor.actor_full_name_uppercase), alpha-# case_count desc, alpha-# min(actor_summary.case_disp_global_code) alpha-# limit alpha-# 1000 alpha-# ; QUERY PLAN ---------------------------------------------------------------------------- ---------------------------------------------------------------------------- ------------------------------- Limit (cost=2555.58..2555.59 rows=1 width=547) (actual time=48841.76..48842.90 rows=1000 loops=1) -> Sort (cost=2555.58..2555.59 rows=1 width=547) (actual time=48841.76..48842.18 rows=1001 loops=1) Sort Key: min((actor.actor_full_name_uppercase)::text), sum(actor_summary.case_count), min((actor_summary.case_disp_global_code)::text) -> Aggregate (cost=2555.50..2555.57 rows=1 width=547) (actual time=48604.17..48755.28 rows=3590 loops=1) -> Group (cost=2555.50..2555.50 rows=1 width=547) (actual time=48604.04..48647.91 rows=3594 loops=1) -> Sort (cost=2555.50..2555.50 rows=1 width=547) (actual time=48604.01..48605.70 rows=3594 loops=1) Sort Key: actor.actor_id -> Nested Loop (cost=1.14..2555.49 rows=1 width=547) (actual time=69.09..48585.83 rows=3594 loops=1) -> Hash Join (cost=1.14..900.39 rows=204 width=475) (actual time=46.92..15259.02 rows=3639 loops=1) Hash Cond: ("outer".source_id = "inner".source_id) -> Index Scan using actor_full_name_uppercase on actor (cost=0.00..895.04 rows=222 width=463) (actual time=46.54..15220.77 rows=3639 loops=1) Index Cond: ((actor_full_name_uppercase >= 'SANDERS'::character varying) AND (actor_full_name_uppercase < 'SANDERT'::character varying)) Filter: (actor_full_name_uppercase ~~ 'SANDERS%'::text) -> Hash (cost=1.11..1.11 rows=11 width=12) (actual time=0.05..0.05 rows=0 loops=1) -> Seq Scan on data_source (cost=0.00..1.11 rows=11 width=12) (actual time=0.02..0.04 rows=11 loops=1) -> Index Scan using actor_summary_pk on actor_summary (cost=0.00..8.11 rows=1 width=72) (actual time=9.14..9.15 rows=1 loops=3639) Index Cond: ("outer".actor_id = actor_summary.actor_id) Total runtime: 48851.85 msec (18 rows) --------------------------------------------------------------------- Nick Fankhauser nickf@doxpop.com Phone 1.765.965.7363 Fax 1.765.962.9788 doxpop - Court records at your fingertips - http://www.doxpop.com/ From pgsql-performance-owner@postgresql.org Tue Dec 16 13:12:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 18554D1B535 for ; Tue, 16 Dec 2003 17:12:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 72240-07 for ; Tue, 16 Dec 2003 13:11:49 -0400 (AST) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id C17C8D1B475 for ; Tue, 16 Dec 2003 13:11:48 -0400 (AST) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id hBGHBk5f030965 for ; Tue, 16 Dec 2003 12:11:48 -0500 Reply-To: From: "Nick Fankhauser" To: "Pgsql-Performance@Postgresql. Org" Subject: Nested loop performance Date: Tue, 16 Dec 2003 12:11:59 -0500 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 V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/231 X-Sequence-Number: 5091 Hi- I'm trying to optimize a query that I *think* should run very fast. Essentially, I'm joining two tables that have very selective indexes and constraining the query on an indexed field. (There's a third small lookup table in the mix, but it doesn't really affect the bottom line.) actor is a table containing roughly 3 million rows with an index on actor_full_name_uppercase and a unique index on actor_id. actor_summary also contains roughly 3 million rows. Its PK is a unique combined index on (actor_id, county_id, case_disp_global_code). The vast majority of the rows in actor correspond to a single row in actor_summary I'd estimate this at 95% or more. The remaining actors with multiple records generally have two corresponding rows in actor summary. Actor summary was created as a performance enhancer, where we can store some pre-calculated values such as the number of court cases an actor is involved in. The constraint is applied first, with reasonable speed. In the example below, it takes about 15 seconds to gather the matches in actor. I'm unsure what is happening next. I notice that an index scan is occurring on actor_summary_pk, with an "actual time" of 9.15, but then it looks like a nested loop occurs at the next level to join these tables. Does this mean that each probe of the actor_summary index will take 9.15 msec, but the nested loop is going to do this once for each actor_id? The nested loop appears to be where most of my time is going, so I'm focusing on this area, but don't know if there is a better approach to this join. Is there a more efficient means than a nested loop to handle such a join? Would a different method be chosen if there was exactly one row in actor_summary for every row in actor? -Nick The query & explain analyze: alpha=# alpha=# alpha=# explain analyze alpha-# select alpha-# min(actor.actor_id) as actor_id, alpha-# min(actor.actor_entity_type) as actor_entity_type, alpha-# min(actor.role_class_code) as role_class_code, alpha-# min(actor.actor_full_name) as actor_full_name, alpha-# min(actor.actor_person_date_of_birth) as actor_person_date_of_birth, alpha-# min(actor.actor_entity_acronym) as actor_entity_acronym, alpha-# min(actor.actor_person_last_name) as actor_person_last_name, alpha-# min(actor.actor_person_first_name) as actor_person_first_name, alpha-# min(actor.actor_person_middle_name) as actor_person_middle_name, alpha-# min(actor.actor_person_name_suffix) as actor_person_name_suffix, alpha-# min(actor.actor_person_place_of_birth) as actor_person_place_of_birth, alpha-# min(actor.actor_person_height) as actor_person_height, alpha-# min(actor.actor_person_height_unit) as actor_person_height_unit, alpha-# min(actor.actor_person_weight) as actor_person_weight, alpha-# min(actor.actor_person_weight_unit) as actor_person_weight_unit, alpha-# min(actor.actor_person_ethnicity) as actor_person_ethnicity, alpha-# min(actor.actor_person_citizenship_count) as actor_person_citizenship_count, alpha-# min(actor.actor_person_hair_color) as actor_person_hair_color, alpha-# min(actor.actor_person_scars_marks_tatto) as actor_person_scars_marks_tatto, alpha-# min(actor.actor_person_marital_status) as actor_person_marital_status, alpha-# min(actor.actor_alias_for_actor_id) as actor_alias_for_actor_id, alpha-# min(to_char(data_source.source_last_update, 'MM/DD/YYYY HH12:MI AM TZ')) as last_update, alpha-# min(actor_summary.single_case_public_id) as case_public_id, alpha-# min(actor_summary.single_case_id) as case_id, alpha-# sum(actor_summary.case_count)as case_count alpha-# from alpha-# actor, alpha-# actor_summary, alpha-# data_source alpha-# where alpha-# actor.actor_id = actor_summary.actor_id alpha-# and data_source.source_id = actor.source_id alpha-# and actor_full_name_uppercase like upper('sanders%') alpha-# group by alpha-# actor.actor_id alpha-# order by alpha-# min(actor.actor_full_name_uppercase), alpha-# case_count desc, alpha-# min(actor_summary.case_disp_global_code) alpha-# limit alpha-# 1000 alpha-# ; QUERY PLAN ---------------------------------------------------------------------------- ---------------------------------------------------------------------------- ------------------------------- Limit (cost=2555.58..2555.59 rows=1 width=547) (actual time=48841.76..48842.90 rows=1000 loops=1) -> Sort (cost=2555.58..2555.59 rows=1 width=547) (actual time=48841.76..48842.18 rows=1001 loops=1) Sort Key: min((actor.actor_full_name_uppercase)::text), sum(actor_summary.case_count), min((actor_summary.case_disp_global_code)::text) -> Aggregate (cost=2555.50..2555.57 rows=1 width=547) (actual time=48604.17..48755.28 rows=3590 loops=1) -> Group (cost=2555.50..2555.50 rows=1 width=547) (actual time=48604.04..48647.91 rows=3594 loops=1) -> Sort (cost=2555.50..2555.50 rows=1 width=547) (actual time=48604.01..48605.70 rows=3594 loops=1) Sort Key: actor.actor_id -> Nested Loop (cost=1.14..2555.49 rows=1 width=547) (actual time=69.09..48585.83 rows=3594 loops=1) -> Hash Join (cost=1.14..900.39 rows=204 width=475) (actual time=46.92..15259.02 rows=3639 loops=1) Hash Cond: ("outer".source_id = "inner".source_id) -> Index Scan using actor_full_name_uppercase on actor (cost=0.00..895.04 rows=222 width=463) (actual time=46.54..15220.77 rows=3639 loops=1) Index Cond: ((actor_full_name_uppercase >= 'SANDERS'::character varying) AND (actor_full_name_uppercase < 'SANDERT'::character varying)) Filter: (actor_full_name_uppercase ~~ 'SANDERS%'::text) -> Hash (cost=1.11..1.11 rows=11 width=12) (actual time=0.05..0.05 rows=0 loops=1) -> Seq Scan on data_source (cost=0.00..1.11 rows=11 width=12) (actual time=0.02..0.04 rows=11 loops=1) -> Index Scan using actor_summary_pk on actor_summary (cost=0.00..8.11 rows=1 width=72) (actual time=9.14..9.15 rows=1 loops=3639) Index Cond: ("outer".actor_id = actor_summary.actor_id) Total runtime: 48851.85 msec (18 rows) --------------------------------------------------------------------- Nick Fankhauser nickf@doxpop.com Phone 1.765.965.7363 Fax 1.765.962.9788 doxpop - Court records at your fingertips - http://www.doxpop.com/ From pgsql-performance-owner@postgresql.org Tue Dec 16 17:29:09 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8E968D1D903 for ; Tue, 16 Dec 2003 21:29:06 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 13888-05 for ; Tue, 16 Dec 2003 17:28:36 -0400 (AST) Received: from fuji.krosing.net (217-159-136-226-dsl.kt.estpak.ee [217.159.136.226]) by svr1.postgresql.org (Postfix) with ESMTP id 05EB5D1D387 for ; Tue, 16 Dec 2003 17:24:53 -0400 (AST) Received: from fuji.krosing.net (localhost.localdomain [127.0.0.1]) by fuji.krosing.net (8.12.8/8.12.8) with ESMTP id hBGLOkLx005536; Tue, 16 Dec 2003 23:24:47 +0200 Received: (from hannu@localhost) by fuji.krosing.net (8.12.8/8.12.8/Submit) id hBGLOjMY005534; Tue, 16 Dec 2003 23:24:45 +0200 X-Authentication-Warning: fuji.krosing.net: hannu set sender to hannu@tm.ee using -f Subject: Re: Excessive rows/tuples seriously degrading query From: Hannu Krosing To: "Chadwick, Russell" Cc: pgsql-performance@postgresql.org In-Reply-To: References: Content-Type: text/plain Content-Transfer-Encoding: 7bit Message-Id: <1071609885.5397.8.camel@fuji.krosing.net> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Tue, 16 Dec 2003 23:24:45 +0200 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/239 X-Sequence-Number: 5099 Chadwick, Russell kirjutas L, 13.12.2003 kell 00:40: > > Hello everyone. > Can anyone explain why this table which has never had more than a > couple rows in it shows > 500k in the query planner even after running > vacuum full. Its terribly slow to return 2 rows of data. The 2 rows > in it are being updated a lot but I couldn't find any explanation for > this behavior. It can be that there is an idle transaction somewhere that has locked a lot of rows (i.e. all your updates have been running inside the same transaction for hour or days) try: $ ps ax| grep post on my linux box this gives 1683 ? S 0:00 /usr/bin/postmaster -p 5432 1704 ? S 0:00 postgres: stats buffer process 1705 ? S 0:00 postgres: stats collector process 5520 ? S 0:00 postgres: hu hannu [local] idle in transaction 5524 pts/2 S 0:00 grep post where backend 5520 seems to be the culprit. > Anything I could try besides droping db and recreating? make sure that no other backend is connected to db and do your > vacuum full; analyze; or if there seems to be something unidentifieable making your table unusable, then just recreate that table: begin; create table stock_log_positions_tmp as select * from stock_log_positions; drop table stock_log_positions; alter table stock_log_positions_tmp rename to stock_log_positions; -- if you have any constraints, indexes or foreign keys -- then recreate them here as well commit; > Thanks - Russ > --------------- hannu From pgsql-performance-owner@postgresql.org Tue Dec 16 18:48:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 25150D1B473 for ; Tue, 16 Dec 2003 22:48:37 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21457-08 for ; Tue, 16 Dec 2003 18:48:08 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 1D321D1CA77 for ; Tue, 16 Dec 2003 18:48:06 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 68DA522C4; Tue, 16 Dec 2003 17:48:07 -0500 (EST) To: "Sean P. Thomas" Cc: pgsql-performance@postgresql.org Subject: Re: Optimizing FK & PK performance... From: Neil Conway In-Reply-To: <3FD89D72.8080508@ulanji.com> (Sean P. Thomas's message of "Thu, 11 Dec 2003 11:38:10 -0500") References: <3FD89D72.8080508@ulanji.com> Date: Tue, 16 Dec 2003 17:47:59 -0500 Message-ID: <87fzfkl6qo.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/240 X-Sequence-Number: 5100 "Sean P. Thomas" writes: > 1. Is there any performance difference for declaring a primary or > foreign key a column or table contraint? From the documentation, > which way is faster and/or scales better: > > CREATE TABLE distributors ( > did integer, > name varchar(40), > PRIMARY KEY(did) > ); > > CREATE TABLE distributors ( > did integer PRIMARY KEY, > name varchar(40) > ); These are equivalent -- the performance should be the same. -Neil From pgsql-performance-owner@postgresql.org Tue Dec 16 18:51:50 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7FFABD1B4BD for ; Tue, 16 Dec 2003 22:51:49 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 24367-07 for ; Tue, 16 Dec 2003 18:51:20 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id B9369D1B4A8 for ; Tue, 16 Dec 2003 18:51:17 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id 177A72357; Tue, 16 Dec 2003 17:51:19 -0500 (EST) To: "David Shadovitz" Cc: pgsql-performance@postgresql.org, david@shadovitz.com Subject: Re: Why is VACUUM ANALYZE
so slow? From: Neil Conway In-Reply-To: <20031209211444.M99946@www.shadovitz.com> (David Shadovitz's message of "Tue, 9 Dec 2003 14:14:44 -0700") References: <20031209211444.M99946@www.shadovitz.com> Date: Tue, 16 Dec 2003 17:51:18 -0500 Message-ID: <87brq8l6l5.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/241 X-Sequence-Number: 5101 "David Shadovitz" writes: > I'm running PG 7.2.2 on RH Linux 8.0. Note that this version of PostgreSQL is quite old. > I'd like to know why "VACUUM ANALYZE
" is extemely slow (hours) for > certain tables. Is there another concurrent transaction that has modified the table but has not committed? VACUUM ANALYZE will need to block waiting for it. You might be able to get some insight into this by examining the pg_locks system view: http://www.postgresql.org/docs/current/static/monitoring-locks.html As well as the pg_stat_activity view. -Neil From pgsql-general-owner@postgresql.org Tue Dec 16 19:50:16 2003 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1CAC9D1B531; Tue, 16 Dec 2003 23:50:14 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31799-08; Tue, 16 Dec 2003 19:49:45 -0400 (AST) Received: from mail.osdl.org (fw.osdl.org [65.172.181.6]) by svr1.postgresql.org (Postfix) with ESMTP id A8A03D1BC5C; Tue, 16 Dec 2003 19:49:37 -0400 (AST) Received: from ibm-a.pdx.osdl.net (ibm-a.pdx.osdl.net [172.20.1.50]) by mail.osdl.org (8.11.6/8.11.6) with ESMTP id hBGNneZ21642; Tue, 16 Dec 2003 15:49:40 -0800 Subject: update slows down in pl/pgsql function From: Jenny Zhang To: perf-pgsql , pgsql-general@postgresql.org Content-Type: text/plain Organization: osdl Message-Id: <1071618760.10925.17.camel@ibm-a.pdx.osdl.net> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Tue, 16 Dec 2003 15:52:41 -0800 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/917 X-Sequence-Number: 54310 I have stored procedure written in pl/pgsql which takes about 13 seconds to finish. I was able to identify that the slowness is caused by one update SQL: UPDATE shopping_cart SET sc_sub_total=sc_subtotal, sc_date=now() WHERE sc_id=sc_id; If I comment this sql out, the stored procedure returns within 1 second. What puzzles me is that if I execute the same update SQL in psql interface, it returns very fast. The following is the explain analyze output for that SQL. #>explain analyze UPDATE shopping_cart SET sc_sub_total=1, sc_date=now() where sc_id=260706; QUERY PLAN ---------------------------------------------------------------------------------------------------------------------------------- Index Scan using shopping_cart_pkey on shopping_cart (cost=0.00..5.01 rows=1 width=144) (actual time=0.22..0.37 rows=1 loops=1) Index Cond: (sc_id = 260706::numeric) Total runtime: 1.87 msec (3 rows) Is it true that using pl/pgsql increases the overhead that much? TIA, Jenny -- Jenny Zhang Open Source Development Lab 12725 SW Millikan Way, Suite 400 Beaverton, OR 97005 (503)626-2455 ext 31 From pgsql-general-owner@postgresql.org Tue Dec 16 19:55:05 2003 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C0F17D1B494; Tue, 16 Dec 2003 23:55:03 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 32497-06; Tue, 16 Dec 2003 19:54:34 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 14236D1B447; Tue, 16 Dec 2003 19:54:32 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 3BD0C3548A; Tue, 16 Dec 2003 15:54:34 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 3A54235405; Tue, 16 Dec 2003 15:54:34 -0800 (PST) Date: Tue, 16 Dec 2003 15:54:34 -0800 (PST) From: Stephan Szabo To: Jenny Zhang Cc: perf-pgsql , pgsql-general@postgresql.org Subject: Re: update slows down in pl/pgsql function In-Reply-To: <1071618760.10925.17.camel@ibm-a.pdx.osdl.net> Message-ID: <20031216155342.J28271@megazone.bigpanda.com> References: <1071618760.10925.17.camel@ibm-a.pdx.osdl.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/918 X-Sequence-Number: 54311 On Tue, 16 Dec 2003, Jenny Zhang wrote: > I have stored procedure written in pl/pgsql which takes about 13 seconds > to finish. I was able to identify that the slowness is caused by one > update SQL: > > UPDATE shopping_cart SET sc_sub_total=sc_subtotal, sc_date=now() > WHERE sc_id=sc_id; Umm, is that exactly the condition you're using? Isn't that going to update the entire table? From pgsql-performance-owner@postgresql.org Tue Dec 16 19:56:32 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CE763D1BA6E for ; Tue, 16 Dec 2003 23:56:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34056-01 for ; Tue, 16 Dec 2003 19:56:00 -0400 (AST) Received: from jalapeno.crazydogs.org (jalapeno.jellybean.co.uk [212.78.70.100]) by svr1.postgresql.org (Postfix) with ESMTP id 1A8BED1B447 for ; Tue, 16 Dec 2003 19:55:57 -0400 (AST) Received: from rp by jalapeno.crazydogs.org with local (Exim 3.35 #1 (Debian)) id 1AWP2T-0001Mi-00 for ; Tue, 16 Dec 2003 23:55:41 +0000 Date: Tue, 16 Dec 2003 23:55:41 +0000 From: Richard Poole To: "Pgsql-Performance@Postgresql. Org" Subject: Re: Nested loop performance Message-ID: <20031216235541.GB8054@guests.deus.net> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.3.28i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/244 X-Sequence-Number: 5104 On Tue, Dec 16, 2003 at 12:11:59PM -0500, Nick Fankhauser wrote: > > I'm trying to optimize a query that I *think* should run very fast. > Essentially, I'm joining two tables that have very selective indexes and > constraining the query on an indexed field. (There's a third small lookup > table in the mix, but it doesn't really affect the bottom line.) > > actor is a table containing roughly 3 million rows with an index on > actor_full_name_uppercase and a unique index on actor_id. > > actor_summary also contains roughly 3 million rows. Its PK is a unique > combined index on (actor_id, county_id, case_disp_global_code). ... > I'm unsure what is happening next. I notice that an index scan is occurring > on actor_summary_pk, with an "actual time" of 9.15, but then it looks like a > nested loop occurs at the next level to join these tables. Does this mean > that each probe of the actor_summary index will take 9.15 msec, but the > nested loop is going to do this once for each actor_id? ... > Is there a more efficient means than a nested loop to handle such a join? > Would a different method be chosen if there was exactly one row in > actor_summary for every row in actor? It seems that your basic problem is that you're fetching lots of rows from two big ol' tables. The innermost estimation mistake being made by the planner is that the restriction on actor_full_name_uppercase will be much more selective than it is; it thinks there will be 222 matching actors and in fact there are 3639. But being right about this wouldn't make things a lot quicker, if it would make them quicker at all; the index scan for them is taking about 15 seconds and presumably a sequential scan of that table would be at least in the same ballpark. Once it's got those rows it needs to look up matches for them in actor_summary. Again, that's 3639 index scans of an index into a wide-ish table; your interpretation of the 9.15 is correct. (9 ms * 3639 rows =~ 30 seconds). It doesn't seem to me that there would be a substantially better plan for this query with your tables as they stand. If your data were more normalised, then your big scans might be quicker (because their rows would be smaller so they would hit fewer disk pages), and the extra lookups in your detail tables would only be done for the rows which actually ended up getting returned - but that would hardly be likely to make an order-of-magnitude difference to your overall speed. If it were my query and I really really needed it to be considerably faster, I'd think about hyper-normalising in the hope that my main tables would shrink so far I could keep them in RAM effectively all the time. The answers to your direct questions are (1) yes, (2) no, not really, and (3) no. Richard From pgsql-general-owner@postgresql.org Tue Dec 16 20:05:03 2003 X-Original-To: pgsql-general-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C95D7D1B442; Wed, 17 Dec 2003 00:04:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33703-06; Tue, 16 Dec 2003 20:04:31 -0400 (AST) Received: from mail.osdl.org (fw.osdl.org [65.172.181.6]) by svr1.postgresql.org (Postfix) with ESMTP id E3405D1BB71; Tue, 16 Dec 2003 20:04:26 -0400 (AST) Received: from ibm-a.pdx.osdl.net (ibm-a.pdx.osdl.net [172.20.1.50]) by mail.osdl.org (8.11.6/8.11.6) with ESMTP id hBH04QZ25571; Tue, 16 Dec 2003 16:04:26 -0800 Subject: Re: update slows down in pl/pgsql function From: Jenny Zhang To: Stephan Szabo Cc: perf-pgsql , general-pgsql In-Reply-To: <20031216155342.J28271@megazone.bigpanda.com> References: <1071618760.10925.17.camel@ibm-a.pdx.osdl.net> <20031216155342.J28271@megazone.bigpanda.com> Content-Type: text/plain Organization: osdl Message-Id: <1071619646.10925.19.camel@ibm-a.pdx.osdl.net> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Tue, 16 Dec 2003 16:07:26 -0800 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/919 X-Sequence-Number: 54312 Oops, I named the var name the same as the column name. Changing it to something else solved the problem. Thanks, Jenny On Tue, 2003-12-16 at 15:54, Stephan Szabo wrote: > On Tue, 16 Dec 2003, Jenny Zhang wrote: > > > I have stored procedure written in pl/pgsql which takes about 13 seconds > > to finish. I was able to identify that the slowness is caused by one > > update SQL: > > > > UPDATE shopping_cart SET sc_sub_total=sc_subtotal, sc_date=now() > > WHERE sc_id=sc_id; > > Umm, is that exactly the condition you're using? Isn't that going to > update the entire table? > > ---------------------------(end of broadcast)--------------------------- > TIP 9: 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 Dec 16 21:17:56 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 67D79D1CB3F for ; Wed, 17 Dec 2003 01:17:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 40437-05 for ; Tue, 16 Dec 2003 21:17:26 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id B8648D1CB21 for ; Tue, 16 Dec 2003 21:17:22 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 6C5DF3524B; Tue, 16 Dec 2003 17:17:26 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 6AEF734DBC; Tue, 16 Dec 2003 17:17:26 -0800 (PST) Date: Tue, 16 Dec 2003 17:17:26 -0800 (PST) From: Stephan Szabo To: Nick Fankhauser Cc: "Pgsql-Performance@Postgresql. Org" Subject: Re: Nested loop performance In-Reply-To: Message-ID: <20031216171625.A31260@megazone.bigpanda.com> References: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/247 X-Sequence-Number: 5107 On Tue, 16 Dec 2003, Nick Fankhauser wrote: > Is there a more efficient means than a nested loop to handle such a join? > Would a different method be chosen if there was exactly one row in > actor_summary for every row in actor? As a question, what does explain analyze give you if you set enable_nestloop=false; before trying the query? From pgsql-performance-owner@postgresql.org Tue Dec 16 21:14:43 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 33F1ED1B47D for ; Wed, 17 Dec 2003 01:14:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 40647-05 for ; Tue, 16 Dec 2003 21:14:14 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id DC553D1B4B1 for ; Tue, 16 Dec 2003 21:14:09 -0400 (AST) Received: from familyhealth.com.au (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (8.12.9p1/8.12.9) with ESMTP id hBH1DtoD085143; Wed, 17 Dec 2003 09:13:56 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <3FDFAEA8.1090607@familyhealth.com.au> Date: Wed, 17 Dec 2003 09:17:28 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Sean P. Thomas" Cc: pgsql-performance@postgresql.org Subject: Re: Optimizing FK & PK performance... References: <3FD89D72.8080508@ulanji.com> In-Reply-To: <3FD89D72.8080508@ulanji.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/246 X-Sequence-Number: 5106 > 1. Is there any performance difference for declaring a primary or > foreign key a column or table contraint? From the documentation, which > way is faster and/or scales better: > > > CREATE TABLE distributors ( > did integer, > name varchar(40), > PRIMARY KEY(did) > ); > > CREATE TABLE distributors ( > did integer PRIMARY KEY, > name varchar(40) > ); No difference - they're parsed to exactly the same thing (the first version). > 2. Is DEFERRABLE and INITIALLY IMMEDIATE or INITIALLY DEFERRABLE > perferred for performance? We generally have very small transactions > (web app) but we utilize a model of: No idea on this one :/ Chris From pgsql-performance-owner@postgresql.org Wed Dec 17 01:31:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 685F5D1B439 for ; Wed, 17 Dec 2003 05:30:59 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68158-08 for ; Wed, 17 Dec 2003 01:30:28 -0400 (AST) Received: from mallard.mail.pas.earthlink.net (mallard.mail.pas.earthlink.net [207.217.120.48]) by svr1.postgresql.org (Postfix) with ESMTP id A2B6DD1B43A for ; Wed, 17 Dec 2003 01:30:27 -0400 (AST) Received: from hsa158.pool003.at101.earthlink.net ([216.249.70.158] helo=shadovitzcmptr) by mallard.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 1AWUGM-0004PZ-00; Tue, 16 Dec 2003 21:30:22 -0800 Received: by localhost with Microsoft MAPI; Tue, 16 Dec 2003 20:45:44 -0800 Message-ID: <01C3C415.9359D080.david@shadovitz.com> From: David Shadovitz Reply-To: "david@shadovitz.com" To: 'Neil Conway' Cc: "pgsql-performance@postgresql.org" Subject: Re: Why is VACUUM ANALYZE
so slow? Date: Tue, 16 Dec 2003 20:37:02 -0800 X-Mailer: Microsoft Internet E-mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/249 X-Sequence-Number: 5109 Neil, Thanks for the good advice. I noticed that I had some sessions for which I could not account, and I think even a 2nd postmaster running. It looks like I've cleaned everything up, and now I can VACUUM and I can DROP an index which wouldn't drop. And I'm looking into upgrading PostgreSQL. -David On Tuesday, December 16, 2003 2:51 PM, Neil Conway [SMTP:neilc@samurai.com] wrote: > "David Shadovitz" writes: > > I'm running PG 7.2.2 on RH Linux 8.0. > > Note that this version of PostgreSQL is quite old. > > > I'd like to know why "VACUUM ANALYZE
" is extemely slow (hours) for > > certain tables. > > Is there another concurrent transaction that has modified the table > but has not committed? VACUUM ANALYZE will need to block waiting for > it. You might be able to get some insight into this by examining the > pg_locks system view: > > http://www.postgresql.org/docs/current/static/monitoring-locks.html > > As well as the pg_stat_activity view. > > -Neil From pgsql-performance-owner@postgresql.org Wed Dec 17 01:30:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B1998D1B42F for ; Wed, 17 Dec 2003 05:30:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70425-03 for ; Wed, 17 Dec 2003 01:30:24 -0400 (AST) Received: from mallard.mail.pas.earthlink.net (mallard.mail.pas.earthlink.net [207.217.120.48]) by svr1.postgresql.org (Postfix) with ESMTP id 8AF41D1B437 for ; Wed, 17 Dec 2003 01:30:23 -0400 (AST) Received: from hsa158.pool003.at101.earthlink.net ([216.249.70.158] helo=shadovitzcmptr) by mallard.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 1AWUGN-0004Pg-00 for pgsql-performance@postgresql.org; Tue, 16 Dec 2003 21:30:23 -0800 Received: by localhost with Microsoft MAPI; Tue, 16 Dec 2003 20:45:44 -0800 Message-ID: <01C3C415.93ADBCE0.david@shadovitz.com> From: David Shadovitz Reply-To: "david@shadovitz.com" To: "pgsql-performance@postgresql.org" Subject: Why is restored database faster? Date: Tue, 16 Dec 2003 20:42:58 -0800 X-Mailer: Microsoft Internet E-mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/248 X-Sequence-Number: 5108 I backed up my database using pg_dump, and then restored it onto a different server using psql. I see that the query "SELECT COUNT(*) FROM myTable" executes immediately on the new server but takes several seconds on the old one. (The servers are identical.) What could account for this difference? Clustering? How can I get the original server to perform as well as the new one? Thanks. -David From pgsql-performance-owner@postgresql.org Wed Dec 17 02:00:50 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C115DD1B437 for ; Wed, 17 Dec 2003 06:00:48 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73570-03 for ; Wed, 17 Dec 2003 02:00:17 -0400 (AST) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 35D35D1B43F for ; Wed, 17 Dec 2003 02:00:16 -0400 (AST) Received: from tokyo.samurai.com (d226-89-59.home.cgocable.net [24.226.89.59]) by bob.samurai.com (Postfix) with ESMTP id E179223BC; Wed, 17 Dec 2003 01:00:14 -0500 (EST) To: "david@shadovitz.com" Cc: "pgsql-performance@postgresql.org" Subject: Re: Why is restored database faster? From: Neil Conway In-Reply-To: <01C3C415.93ADBCE0.david@shadovitz.com> (David Shadovitz's message of "Tue, 16 Dec 2003 20:42:58 -0800") References: <01C3C415.93ADBCE0.david@shadovitz.com> Date: Wed, 17 Dec 2003 01:00:13 -0500 Message-ID: <87iskgugpe.fsf@mailbox.samurai.com> User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/250 X-Sequence-Number: 5110 David Shadovitz writes: > What could account for this difference? Lots of things -- disk fragmentation, expired tuples that aren't being cleaned up by VACUUM due to a long-lived transaction, the state of the kernel buffer cache, the configuration of the kernel, etc. > How can I get the original server to perform as well as the new one? Well, you can start by giving us some more information. For example, what is the output of VACUUM VERBOSE on the slow server? How much disk space does the database directory take up on both machines? (BTW, "SELECT count(*) FROM table" isn't a particularly good DBMS performance indication...) -Neil From pgsql-performance-owner@postgresql.org Wed Dec 17 02:31:49 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1CDA2D1C4ED for ; Wed, 17 Dec 2003 06:31:48 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71054-09 for ; Wed, 17 Dec 2003 02:31:17 -0400 (AST) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id DE46ED1C4D0 for ; Wed, 17 Dec 2003 02:31:13 -0400 (AST) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id hBH6VIbo014665 for ; Wed, 17 Dec 2003 12:01:18 +0530 Received: from persistent.co.in (daithan.intranet.pspl.co.in [192.168.7.161]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id hBH6VDQK014554; Wed, 17 Dec 2003 12:01:13 +0530 Message-ID: <3FDFF82A.2070201@persistent.co.in> Date: Wed, 17 Dec 2003 12:01:06 +0530 From: Shridhar Daithankar Organization: Persistent Systems Pvt. Ltd. User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Neil Conway Cc: "david@shadovitz.com" , "pgsql-performance@postgresql.org" Subject: Re: Why is restored database faster? References: <01C3C415.93ADBCE0.david@shadovitz.com> <87iskgugpe.fsf@mailbox.samurai.com> In-Reply-To: <87iskgugpe.fsf@mailbox.samurai.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/251 X-Sequence-Number: 5111 Neil Conway wrote: >>How can I get the original server to perform as well as the new one? Well, you have the answer. Dump the database, stop postmaster and restore it. That should be faster than original one. > > (BTW, "SELECT count(*) FROM table" isn't a particularly good DBMS > performance indication...) Particularly in case of postgresql..:-) Shridhar From pgsql-performance-owner@postgresql.org Wed Dec 17 02:43:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B9694D1B4C8 for ; Wed, 17 Dec 2003 06:43:19 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77260-06 for ; Wed, 17 Dec 2003 02:42:49 -0400 (AST) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 59368D1C4E2 for ; Wed, 17 Dec 2003 02:42:47 -0400 (AST) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id D54BB8DD3; Wed, 17 Dec 2003 07:42:46 +0100 (CET) Date: Wed, 17 Dec 2003 07:42:46 +0100 (CET) From: Dennis Bjorklund To: David Shadovitz Cc: "pgsql-performance@postgresql.org" Subject: Re: Why is restored database faster? In-Reply-To: <01C3C415.93ADBCE0.david@shadovitz.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/252 X-Sequence-Number: 5112 On Tue, 16 Dec 2003, David Shadovitz wrote: > I backed up my database using pg_dump, and then restored it onto a different > server using psql. I see that the query "SELECT COUNT(*) FROM myTable" > executes immediately on the new server but takes several seconds on the old > one. (The servers are identical.) > > What could account for this difference? Clustering? How can I get the > original server to perform as well as the new one? You probably need to run VACUUM FULL. It locks the tables during its execution so only do it when the database is not in full use. If this helps you probably need to do normal vacuums more often and maybe tune the max_fsm_pages to be bigger. -- /Dennis From pgsql-performance-owner@postgresql.org Wed Dec 17 05:52:52 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 32F42D1B565 for ; Wed, 17 Dec 2003 09:52:51 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 98641-06 for ; Wed, 17 Dec 2003 05:52:22 -0400 (AST) Received: from anchor-post-34.mail.demon.net (anchor-post-34.mail.demon.net [194.217.242.92]) by svr1.postgresql.org (Postfix) with ESMTP id 00DC5D1C990 for ; Wed, 17 Dec 2003 05:52:20 -0400 (AST) Received: from mwynhau.demon.co.uk ([193.237.186.96] helo=mainbox.archonet.com) by anchor-post-34.mail.demon.net with esmtp (Exim 3.35 #1) id 1AWYLs-000Fn1-0Y; Wed, 17 Dec 2003 09:52:20 +0000 Received: by mainbox.archonet.com (Postfix, from userid 529) id 3BCCA167DB; Wed, 17 Dec 2003 09:52:19 +0000 (GMT) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id E5F0615987; Wed, 17 Dec 2003 09:52:16 +0000 (GMT) From: Richard Huxton To: , "Pgsql-Performance@Postgresql. Org" Subject: Re: Nested loop question Date: Wed, 17 Dec 2003 09:52:15 +0000 User-Agent: KMail/1.5 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: <200312170952.15759.dev@archonet.com> X-Bogosity: No, tests=bogofilter, spamicity=0.000000, version=0.15.3 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/253 X-Sequence-Number: 5113 On Tuesday 16 December 2003 17:06, Nick Fankhauser - Doxpop wrote: > Hi- > > I'm trying to optimize a query that I *think* should run very fast. > Essentially, I'm joining two tables that have very selective indexes and > constraining the query on an indexed field. (There's a third small lookup > table in the mix, but it doesn't really affect the bottom line.) > I'm unsure what is happening next. I notice that an index scan is occurring > on actor_summary_pk, with an "actual time" of 9.15, but then it looks like > a nested loop occurs at the next level to join these tables. Does this mean > that each probe of the actor_summary index will take 9.15 msec, but the > nested loop is going to do this once for each actor_id? That's right - you need to multiply the actual time by the number of loops. In your case this would seem to be about 33 seconds. > -> Index Scan using actor_summary_pk on > actor_summary (cost=0.00..8.11 rows=1 width=72) (actual time=9.14..9.15 > rows=1 loops=3639) > Index Cond: ("outer".actor_id = > actor_summary.actor_id) > The nested loop appears to be where most of my time is going, so I'm > focusing on this area, but don't know if there is a better approach to this > join. > > Is there a more efficient means than a nested loop to handle such a join? > Would a different method be chosen if there was exactly one row in > actor_summary for every row in actor? Hmm - tricky to say in your case. PG has decided to filter on actor then look up the corresponding values in actor_summary. Given that you have 3 million rows in both tables that seems a reasonable approach. You could always try forcing different plans by switching the various ENABLE_HASHJOIN etc options (see the runtime configuration section of the manuals). I'm not sure that will help you here though. The fact that it's taking you 9ms to do each index lookup suggests to me that it's going to disk each time. Does that sound plausible, or do you think you have enough RAM to cache your large indexes? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Wed Dec 17 11:26:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id AACABD1B486 for ; Wed, 17 Dec 2003 15:26:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 45255-01 for ; Wed, 17 Dec 2003 11:26:16 -0400 (AST) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id 7D279D1B488 for ; Wed, 17 Dec 2003 11:26:12 -0400 (AST) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id hBHFQ65j020851; Wed, 17 Dec 2003 10:26:13 -0500 Reply-To: From: "Nick Fankhauser" To: "Richard Huxton" , "Pgsql-Performance@Postgresql. Org" Subject: Re: Nested loop question Date: Wed, 17 Dec 2003 10:26:19 -0500 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: <200312170952.15759.dev@archonet.com> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/255 X-Sequence-Number: 5115 > The fact that it's taking you 9ms to do each index lookup > suggests to me that > it's going to disk each time. Does that sound plausible, or do > you think you > have enough RAM to cache your large indexes? I'm sure we don't have enough RAM to cache all of our large indexes, so your supposition makes sense. We have 1GB on this machine. In responding to the performance problems we're having, one of the questions has been adding memory vs crafting "helper" tables to speed things up. The issue is that this database needs to be able to scale easily to about 10 times the size, so although we could easily triple the memory at reasonable expense, we'd eventually hit a wall. Is there any solid method to insure that a particular index always resides in memory? A hybrid approach that might scale reliably would be to bump up our memory and then make sure key indexes are cached. however, I'm concerned that if we didn't have a way to ensure that the indexes that we choose remain cached, we would have very inconsistent responses. Thanks for your ideas! -Nick From pgsql-performance-owner@postgresql.org Wed Dec 17 11:26:44 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5F0CBD1B47D for ; Wed, 17 Dec 2003 15:26:42 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 45227-01 for ; Wed, 17 Dec 2003 11:26:15 -0400 (AST) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id 31BECD1B47E for ; Wed, 17 Dec 2003 11:26:11 -0400 (AST) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id hBHFQ65k020851; Wed, 17 Dec 2003 10:26:14 -0500 Reply-To: From: "Nick Fankhauser" To: "Stephan Szabo" , "Nick Fankhauser" Cc: "Pgsql-Performance@Postgresql. Org" Subject: Re: Nested loop performance Date: Wed, 17 Dec 2003 10:26:20 -0500 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: <20031216171625.A31260@megazone.bigpanda.com> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/254 X-Sequence-Number: 5114 > As a question, what does explain analyze give you if you > set enable_nestloop=false; before trying the query? Here are the results- It looks quite a bit more painful than the other plan, although the wall time is in the same ballpark. alpha=# explain analyze alpha-# select alpha-# min(actor.actor_id) as actor_id, alpha-# min(actor.actor_entity_type) as actor_entity_type, alpha-# min(actor.role_class_code) as role_class_code, alpha-# min(actor.actor_full_name) as actor_full_name, alpha-# min(actor.actor_person_date_of_birth) as actor_person_date_of_birth, alpha-# min(actor.actor_entity_acronym) as actor_entity_acronym, alpha-# min(actor.actor_person_last_name) as actor_person_last_name, alpha-# min(actor.actor_person_first_name) as actor_person_first_name, alpha-# min(actor.actor_person_middle_name) as actor_person_middle_name, alpha-# min(actor.actor_person_name_suffix) as actor_person_name_suffix, alpha-# min(actor.actor_person_place_of_birth) as actor_person_place_of_birth, alpha-# min(actor.actor_person_height) as actor_person_height, alpha-# min(actor.actor_person_height_unit) as actor_person_height_unit, alpha-# min(actor.actor_person_weight) as actor_person_weight, alpha-# min(actor.actor_person_weight_unit) as actor_person_weight_unit, alpha-# min(actor.actor_person_ethnicity) as actor_person_ethnicity, alpha-# min(actor.actor_person_citizenship_count) as actor_person_citizenship_count, alpha-# min(actor.actor_person_hair_color) as actor_person_hair_color, alpha-# min(actor.actor_person_scars_marks_tatto) as actor_person_scars_marks_tatto, alpha-# min(actor.actor_person_marital_status) as actor_person_marital_status, alpha-# min(actor.actor_alias_for_actor_id) as actor_alias_for_actor_id, alpha-# min(to_char(data_source.source_last_update, 'MM/DD/YYYY HH12:MI AM TZ')) as last_update, alpha-# min(actor_summary.single_case_public_id) as case_public_id, alpha-# min(actor_summary.single_case_id) as case_id, alpha-# sum(actor_summary.case_count)as case_count alpha-# from alpha-# actor, alpha-# actor_summary, alpha-# data_source alpha-# where alpha-# actor.actor_id = actor_summary.actor_id alpha-# and data_source.source_id = actor.source_id alpha-# and actor.actor_full_name_uppercase like upper('sanders%') alpha-# group by alpha-# actor.actor_id alpha-# order by alpha-# min(actor.actor_full_name_uppercase), alpha-# case_count desc, alpha-# min(actor_summary.case_disp_global_code) alpha-# limit alpha-# 1000; QUERY PLAN ---------------------------------------------------------------------------- ---------------------------------------------------------------------------- ---------------- ---------------------- Limit (cost=168919.98..168920.03 rows=20 width=548) (actual time=91247.95..91249.05 rows=1000 loops=1) -> Sort (cost=168919.98..168920.03 rows=20 width=548) (actual time=91247.95..91248.35 rows=1001 loops=1) Sort Key: min((actor.actor_full_name_uppercase)::text), sum(actor_summary.case_count), min((actor_summary.case_disp_global_code)::text) -> Aggregate (cost=168904.95..168919.54 rows=20 width=548) (actual time=91015.00..91164.68 rows=3590 loops=1) -> Group (cost=168904.95..168905.95 rows=201 width=548) (actual time=90999.87..91043.25 rows=3594 loops=1) -> Sort (cost=168904.95..168905.45 rows=201 width=548) (actual time=90999.83..91001.57 rows=3594 loops=1) Sort Key: actor.actor_id -> Hash Join (cost=903.08..168897.24 rows=201 width=548) (actual time=25470.63..90983.45 rows=3594 loops=1) Hash Cond: ("outer".actor_id = "inner".actor_id) -> Seq Scan on actor_summary (cost=0.00..150715.43 rows=3455243 width=73) (actual time=8.03..52902.24 rows=3455243 loops=1) -> Hash (cost=902.57..902.57 rows=204 width=475) (actual time=25459.92..25459.92 rows=0 loops=1) -> Hash Join (cost=1.14..902.57 rows=204 width=475) (actual time=155.92..25451.25 rows=3639 loops=1) Hash Cond: ("outer".source_id = "inner".source_id) -> Index Scan using actor_full_name_uppercase on actor (cost=0.00..897.20 rows=223 width=463) (actual time=144.93..25404. 10 rows=3639 loops=1) Index Cond: ((actor_full_name_uppercase >= 'SANDERS'::character varying) AND (actor_full_name_uppercase < 'SANDERT':: character varying)) Filter: (actor_full_name_uppercase ~~ 'SANDERS%'::text) -> Hash (cost=1.11..1.11 rows=11 width=12) (actual time=10.66..10.66 rows=0 loops=1) -> Seq Scan on data_source (cost=0.00..1.11 rows=11 width=12) (actual time=10.63..10.64 rows=11 loops=1) Total runtime: 91275.18 msec (19 rows) alpha=# From pgsql-performance-owner@postgresql.org Wed Dec 17 11:27:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 06A58D1BB71 for ; Wed, 17 Dec 2003 15:26:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 39356-07 for ; Wed, 17 Dec 2003 11:26:30 -0400 (AST) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id E7190D1B4BB for ; Wed, 17 Dec 2003 11:26:25 -0400 (AST) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id hBHFQ65l020851; Wed, 17 Dec 2003 10:26:15 -0500 Reply-To: From: "Nick Fankhauser" To: "Richard Poole" , "Pgsql-Performance@Postgresql. Org" Subject: Re: Nested loop performance Date: Wed, 17 Dec 2003 10:26:25 -0500 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: <20031216235541.GB8054@guests.deus.net> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/256 X-Sequence-Number: 5116 > It seems that your basic problem is that you're fetching lots of rows > from two big ol' tables. > It doesn't seem to me that there would be a substantially better plan > for this query with your tables as they stand. That's more or less the conclusion I had come to. I was just hoping someone else could point out an approach I've been missing. (sigh!) > If your data were more > normalised, then your big scans might be quicker (because their rows > would be smaller so they would hit fewer disk pages), This started off as a 5-table join on well-normalized data. Unfortunately, the actor table doesn't get any smaller, and the work involved in calculating the "case_count" information on the fly was clearly becoming a problem- particularly with actors that had a heavy caseload. (Busy attorneys and judges.) The actor_summary approach makes these previous problem cases go away, but the payback is that (as you correctly pointed out) queries on average citizens who only have one case suffer from the de-normalized approach. We're currently considering the approach of just returning all of the rows to our application, and doing the aggregation and limit work in the app. The inconsistency of the data makes it very tough for the query planner to come up with an strategy that is always a winner. Thanks for your thoughts! -Nick From pgsql-performance-owner@postgresql.org Mon Dec 22 16:38:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5A462D1B437; Wed, 17 Dec 2003 16:58:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 58048-03; Wed, 17 Dec 2003 12:58:15 -0400 (AST) Received: from smxsat1.smxs.net (smxsat1.smxs.net [213.150.10.1]) by svr1.postgresql.org (Postfix) with ESMTP id B8164D1B441; Wed, 17 Dec 2003 12:58:12 -0400 (AST) Received: from m01x1.s-mxs.net [10.3.55.201] by smxsat1.smxs.net over TLS secured channel with XWall v3.28g ; Wed, 17 Dec 2003 17:57:54 +0100 Received: from m0102.s-mxs.net [10.3.55.2] by m01x1.s-mxs.net with XWall v3.28f ; Wed, 17 Dec 2003 17:57:53 +0100 Received: from m0114.s-mxs.net ([10.3.55.14]) by m0102.s-mxs.net with Microsoft SMTPSVC(5.0.2195.5329); Wed, 17 Dec 2003 17:57:53 +0100 X-MimeOLE: Produced By Microsoft Exchange V6.0.6503.0 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: [HACKERS] fsync method checking Date: Wed, 17 Dec 2003 17:57:52 +0100 Message-ID: <46C15C39FEB2C44BA555E356FBCD6FA496208C@m0114.s-mxs.net> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [HACKERS] fsync method checking Thread-Index: AcPA90yf7GDEkd2TQ9idLmOL/AmsmQDxrHRQ From: "Zeugswetter Andreas SB SD" To: "Tom Lane" , "Manfred Spraul" Cc: "Bruce Momjian" , "Mark Kirkwood" , , "PostgreSQL-development" X-OriginalArrivalTime: 17 Dec 2003 16:57:53.0114 (UTC) FILETIME=[E92963A0:01C3C4BE] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/284 X-Sequence-Number: 5144 > Ideally that path isn't taken very often. But I'm currently having a > discussion off-list with a CMU student who seems to be seeing a case > where it happens a lot. (She reports that both WALWriteLock and > WALInsertLock are causes of a lot of process blockages, which seems to > mean that a lot of the WAL I/O is being done with both held, which would > have to mean that AdvanceXLInsertBuffer is doing the I/O.=20=20 > More when we figure out what's going on exactly...) I would figure, that this is in a situation where a large transaction fills one XLInsertBuffer, and a lot of WAL buffers are not yet written. Andreas From pgsql-performance-owner@postgresql.org Wed Dec 17 14:35:02 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 850BAD1B8BC for ; Wed, 17 Dec 2003 18:35:00 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69990-10 for ; Wed, 17 Dec 2003 14:34:30 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 61257D1B534 for ; Wed, 17 Dec 2003 14:34:29 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBHIYO19008942; Wed, 17 Dec 2003 13:34:27 -0500 (EST) To: Hannu Krosing Cc: "Chadwick, Russell" , pgsql-performance@postgresql.org Subject: Re: Excessive rows/tuples seriously degrading query In-reply-to: <1071609885.5397.8.camel@fuji.krosing.net> References: <1071609885.5397.8.camel@fuji.krosing.net> Comments: In-reply-to Hannu Krosing message dated "Tue, 16 Dec 2003 23:24:45 +0200" Date: Wed, 17 Dec 2003 13:34:24 -0500 Message-ID: <8941.1071686064@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/257 X-Sequence-Number: 5117 Hannu Krosing writes: > Chadwick, Russell kirjutas L, 13.12.2003 kell 00:40: >> Can anyone explain why this table which has never had more than a >> couple rows in it shows > 500k in the query planner even after running >> vacuum full. > It can be that there is an idle transaction somewhere that has locked a > lot of rows (i.e. all your updates have been running inside the same > transaction for hour or days) In fact an old open transaction is surely the issue, given that the VACUUM report shows a huge number of "kept" tuples: >> INFO: Pages 4773: Changed 1, reaped 767, Empty 0, New 0; Tup 613737: Vac 57620, Keep/VTL 613735/613713, UnUsed 20652, MinLen 52, MaxLen 52; Re-using: Free/Avail. Space 4322596/4322596; EndEmpty/Avail. Pages 0/4773. >> CPU 9.11s/13.68u sec elapsed 22.94 sec. "Keep" is the number of tuples that are committed dead but can't be removed yet because there is some other open transaction that is old enough that it should be able to see them if it looks. Apparently the access pattern on this table is constant updates of the two logical rows, leaving lots and lots of dead versions. You need to vacuum it more often to keep down the amount of deadwood, and you need to avoid having very-long-running transactions open when you vacuum. regards, tom lane From pgsql-performance-owner@postgresql.org Wed Dec 17 15:57:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F21A8D1B452 for ; Wed, 17 Dec 2003 19:57:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81206-10 for ; Wed, 17 Dec 2003 15:56:51 -0400 (AST) Received: from shire.ontko.com (shire.ontko.com [199.164.165.1]) by svr1.postgresql.org (Postfix) with ESMTP id DD9A2D1B459 for ; Wed, 17 Dec 2003 15:56:49 -0400 (AST) Received: from nick (bilbo.ontko.com [199.164.165.101]) by shire.ontko.com (8.12.3/8.12.3/Debian-6.6) with SMTP id hBHJun5f027972 for ; Wed, 17 Dec 2003 14:56:50 -0500 Reply-To: From: "Nick Fankhauser" To: "Pgsql-Performance@Postgresql. Org" Subject: Adding RAM: seeking advice & warnings of hidden "gotchas" Date: Wed, 17 Dec 2003 14:57:02 -0500 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 V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/258 X-Sequence-Number: 5118 Hi- After having done my best to squeeze better performance out of our application by tuning within our existing resources, I'm falling back on adding memory as a short-term solution while we get creative for a long-term fix. I'm curious about what experiences others have had with the process of adding big chunks of RAM. In particular, if I'm trying to encourage the OS to cache more of my index information in RAM, what sort of configuration should I do at both the PostgreSQL and OS level? In a slightly off-topic vein, I'd also like to hear about it if anyone knows about any gotchas at the OS level that might become a problem. The server is a dual processor Athlon 1.2GHz box with hardware SCSI RAID. It currently has 1 GB RAM, and we're planning to add one GB more for a total of 2GB. The OS is Debian Linux Kernel 2.4.x, and we're on PostgreSQL v7.3.2 My current memory related settings are: SHMMAX and SHMALL set to 128MB (OS setting) shared buffers 8192 (64MB) sort_mem 16384 (16MB) effective_cache_size 65536 (512MB) We support up to 70 active users, sharing a connection pool of 16 connections. Most of the queries center around 3 tables that are about 1.5 GB each. Thanks. -Nick --------------------------------------------------------------------- Nick Fankhauser nickf@doxpop.com Phone 1.765.965.7363 Fax 1.765.962.9788 doxpop - Court records at your fingertips - http://www.doxpop.com/ From pgsql-performance-owner@postgresql.org Wed Dec 17 17:39:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 67B67D1BA62 for ; Wed, 17 Dec 2003 21:39:41 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06735-10 for ; Wed, 17 Dec 2003 17:39:12 -0400 (AST) Received: from main.wiredfool.com (dsl231-055-035.sea1.dsl.speakeasy.net [216.231.55.35]) by svr1.postgresql.org (Postfix) with ESMTP id DFC97D1B477 for ; Wed, 17 Dec 2003 17:38:47 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 403A22728F; Wed, 17 Dec 2003 13:33:38 -0800 (PST) Received: from main.wiredfool.com ([127.0.0.1]) by localhost (cabbage [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 20898-05; Wed, 17 Dec 2003 13:33:34 -0800 (PST) Received: from [IPv6:::1] (localhost [127.0.0.1]) by main.wiredfool.com (Postfix) with ESMTP id 9726A2728D; Wed, 17 Dec 2003 13:33:34 -0800 (PST) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v606) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <63587126-30D9-11D8-A85B-0003930F2A6C@soroos.net> Content-Transfer-Encoding: 7bit Cc: "Pgsql-Performance@Postgresql. Org" From: Eric Soroos Subject: Re: Adding RAM: seeking advice & warnings of hidden "gotchas" Date: Wed, 17 Dec 2003 13:38:44 -0800 To: X-Mailer: Apple Mail (2.606) X-Virus-Scanned: by amavisd-new-20030616-p3 (Debian) at main.wiredfool.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/259 X-Sequence-Number: 5119 On Dec 17, 2003, at 11:57 AM, Nick Fankhauser wrote: > Hi- > > After having done my best to squeeze better performance out of our > application by tuning within our existing resources, I'm falling back > on > adding memory as a short-term solution while we get creative for a > long-term > fix. I'm curious about what experiences others have had with the > process of > adding big chunks of RAM. In particular, if I'm trying to encourage > the OS > to cache more of my index information in RAM, what sort of > configuration > should I do at both the PostgreSQL and OS level? You need bigmem compiled in the kernel, which you should already have at the 1 gig level iirc. You should bump up your effective cache size, probably to around 1.75 gig. I wouldn't bump up the shared buffers beyond where you have them now. If you're swapping out sorts to disk, you may gain boosting sortmem some since you have the additional memory to use. > The server is a dual processor Athlon 1.2GHz box with hardware SCSI > RAID. It > currently has 1 GB RAM, and we're planning to add one GB more for a > total of > 2GB. The OS is Debian Linux Kernel 2.4.x, and we're on PostgreSQL > v7.3.2 I've got a machine running Debian Stable w/2.4.x, 1.3 ghz p3, 1.5 gig ram, pg 7.2.4 and it's rock solid. eric From pgsql-performance-owner@postgresql.org Wed Dec 17 18:19:35 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 022EDD1B495 for ; Wed, 17 Dec 2003 22:19:34 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15132-03 for ; Wed, 17 Dec 2003 18:19:04 -0400 (AST) Received: from bayswater1.ymogen.net (host-154-240-27-217.pobox.net.uk [217.27.240.154]) by svr1.postgresql.org (Postfix) with ESMTP id 08CC0D1B45D for ; Wed, 17 Dec 2003 18:19:00 -0400 (AST) Received: from finisterre (82-68-132-233.dsl.in-addr.zen.co.uk [82.68.132.233]) by bayswater1.ymogen.net (Postfix) with SMTP id A7A4B9AE38; Wed, 17 Dec 2003 22:18:45 +0000 (GMT) From: "Matt Clark" To: , "Pgsql-Performance@Postgresql. Org" Subject: Re: Adding RAM: seeking advice & warnings of hidden "gotchas" Date: Wed, 17 Dec 2003 22:18:45 -0000 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 (9.0.2910.0) Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 In-Reply-To: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/260 X-Sequence-Number: 5120 If you have 3 1.5GB tables then you might as well go for 4GB while you're at it. Make sure you've got a bigmem kernel either running or available, and boost effective_cache_size by whatever amount you increase the RAM by. We run a Quad Xeon/4GB server on Redhat 7.3 and it's solid as a rock. There is no way I know of to get indexes preferentially cached over data though. Matt > -----Original Message----- > From: pgsql-performance-owner@postgresql.org > [mailto:pgsql-performance-owner@postgresql.org]On Behalf Of Nick > Fankhauser > Sent: 17 December 2003 19:57 > To: Pgsql-Performance@Postgresql. Org > Subject: [PERFORM] Adding RAM: seeking advice & warnings of hidden > "gotchas" > > > Hi- > > After having done my best to squeeze better performance out of our > application by tuning within our existing resources, I'm falling back on > adding memory as a short-term solution while we get creative for > a long-term > fix. I'm curious about what experiences others have had with the > process of > adding big chunks of RAM. In particular, if I'm trying to encourage the OS > to cache more of my index information in RAM, what sort of configuration > should I do at both the PostgreSQL and OS level? > > In a slightly off-topic vein, I'd also like to hear about it if > anyone knows > about any gotchas at the OS level that might become a problem. > > The server is a dual processor Athlon 1.2GHz box with hardware > SCSI RAID. It > currently has 1 GB RAM, and we're planning to add one GB more for > a total of > 2GB. The OS is Debian Linux Kernel 2.4.x, and we're on PostgreSQL v7.3.2 > > My current memory related settings are: > > SHMMAX and SHMALL set to 128MB (OS setting) > shared buffers 8192 (64MB) > sort_mem 16384 (16MB) > effective_cache_size 65536 (512MB) > > > We support up to 70 active users, sharing a connection pool of 16 > connections. Most of the queries center around 3 tables that are about 1.5 > GB each. > > > Thanks. > -Nick > > --------------------------------------------------------------------- > Nick Fankhauser > > nickf@doxpop.com Phone 1.765.965.7363 Fax 1.765.962.9788 > doxpop - Court records at your fingertips - http://www.doxpop.com/ > > > > ---------------------------(end of broadcast)--------------------------- > TIP 7: don't forget to increase your free space map settings > From pgsql-performance-owner@postgresql.org Thu Dec 18 00:00:12 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C20E4D1B449 for ; Thu, 18 Dec 2003 03:59:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50374-09 for ; Wed, 17 Dec 2003 23:59:25 -0400 (AST) Received: from snipe.mail.pas.earthlink.net (snipe.mail.pas.earthlink.net [207.217.120.62]) by svr1.postgresql.org (Postfix) with ESMTP id 75BCDD1B44D for ; Wed, 17 Dec 2003 23:59:24 -0400 (AST) Received: from hsa218.pool032.at101.earthlink.net ([216.249.101.218] helo=shadovitzcmptr) by snipe.mail.pas.earthlink.net with smtp (Exim 3.33 #1) id 1AWpJW-00001w-00; Wed, 17 Dec 2003 19:59:02 -0800 Received: by localhost with Microsoft MAPI; Wed, 17 Dec 2003 19:55:01 -0800 Message-ID: <01C3C4D7.A7F88280.david@shadovitz.com> From: David Shadovitz Reply-To: "david@shadovitz.com" To: 'Dennis Bjorklund' , "'shridhar_daithankar@persistent.co.in'" , "'neilc@samurai.com'" Cc: "pgsql-performance@postgresql.org" Subject: Re: Why is restored database faster? Date: Wed, 17 Dec 2003 19:54:45 -0800 X-Mailer: Microsoft Internet E-mail/MAPI - 8.0.0.4211 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/261 X-Sequence-Number: 5121 Dennis, Shridhar, and Neil, Thanks for your input. Here are my responses: I ran VACUUM FULL on the table in question. Although that did reduce "Pages" and "UnUsed", the "SELECT *" query is still much slower on this installation than in the new, restored one. Old server: # VACUUM FULL abc; VACUUM # VACUUM VERBOSE abc; NOTICE: --Relation abc-- NOTICE: Pages 1526: Changed 0, Empty 0; Tup 91528; Vac 0, Keep 0, UnUsed 32. Total CPU 0.07s/0.52u sec elapsed 0.60 sec. VACUUM New server: # VACUUM VERBOSE abc; NOTICE: --Relation abc-- NOTICE: Pages 1526: Changed 0, Empty 0; Tup 91528; Vac 0, Keep 0, UnUsed 0. Total CPU 0.02s/0.00u sec elapsed 0.02 sec. VACUUM max_fsm_pages is at its default value, 10000. People don't have the practice of dumping and restoring just for the purpose of improving performance, do they? Neil asked how much disk space the database directory takes on each machine. What directory is of interest? The whole thing takes up about 875 MB on each machine. -David From pgsql-performance-owner@postgresql.org Thu Dec 18 02:48:21 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 74623D1B46D for ; Thu, 18 Dec 2003 06:48:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69309-03 for ; Thu, 18 Dec 2003 02:47:50 -0400 (AST) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id 556EED1B46E for ; Thu, 18 Dec 2003 02:47:47 -0400 (AST) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id hBI6lqG1018277 for ; Thu, 18 Dec 2003 12:17:52 +0530 Received: from daithan.intranet.pspl.co.in (daithan.intranet.pspl.co.in [192.168.7.161]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id hBI6lpQK018257 for ; Thu, 18 Dec 2003 12:17:51 +0530 From: Shridhar Daithankar To: "pgsql-performance@postgresql.org" Subject: Re: Why is restored database faster? Date: Thu, 18 Dec 2003 12:17:12 +0530 User-Agent: KMail/1.5.4 References: <01C3C4D7.A7F88280.david@shadovitz.com> In-Reply-To: <01C3C4D7.A7F88280.david@shadovitz.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312181217.12520.shridhar_daithankar@persistent.co.in> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/262 X-Sequence-Number: 5122 On Thursday 18 December 2003 09:24, David Shadovitz wrote: > Old server: > # VACUUM FULL abc; > VACUUM > # VACUUM VERBOSE abc; > NOTICE: --Relation abc-- > NOTICE: Pages 1526: Changed 0, Empty 0; Tup 91528; Vac 0, Keep 0, UnUsed > 32. Total CPU 0.07s/0.52u sec elapsed 0.60 sec. > VACUUM > > New server: > # VACUUM VERBOSE abc; > NOTICE: --Relation abc-- > NOTICE: Pages 1526: Changed 0, Empty 0; Tup 91528; Vac 0, Keep 0, UnUsed > 0. Total CPU 0.02s/0.00u sec elapsed 0.02 sec. > VACUUM > > max_fsm_pages is at its default value, 10000. Well, then the only issue left is file sytem defragmentation. Which file system is this anyway > People don't have the practice of dumping and restoring just for the > purpose of improving performance, do they? Well, at times it is required. Especially if it is update intensive environment. An no database is immune to that > Neil asked how much disk space the database directory takes on each > machine. What directory is of interest? The whole thing takes up about 875 > MB on each machine. That is fairly small.. Should not take much time..in my guess, the time it takes to vacuum is more than time to dump and reload. Another quick way to defragment a file system is to copy entire data directory to another partition(Shutdown postmaster first), delete it from original partition and move back. Contegous wriing to a partition results in defragmentation effectively. Try it and see if it helps. It could be much less trouble than dump/restore.. HTH Shridhar From pgsql-performance-owner@postgresql.org Thu Dec 18 11:13:07 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 68832D1B491 for ; Thu, 18 Dec 2003 15:13:03 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 30772-08 for ; Thu, 18 Dec 2003 11:12:18 -0400 (AST) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 05E88D1B4CF for ; Thu, 18 Dec 2003 11:12:14 -0400 (AST) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 9A14E8DD4; Thu, 18 Dec 2003 16:12:16 +0100 (CET) Date: Thu, 18 Dec 2003 16:12:16 +0100 (CET) From: Dennis Bjorklund To: Shridhar Daithankar Cc: "pgsql-performance@postgresql.org" Subject: Re: Why is restored database faster? In-Reply-To: <200312181217.12520.shridhar_daithankar@persistent.co.in> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/263 X-Sequence-Number: 5123 On Thu, 18 Dec 2003, Shridhar Daithankar wrote: > Well, then the only issue left is file sytem defragmentation. And the internal fragmentation that can be "fixed" with the CLUSTER command. -- /Dennis From pgsql-performance-owner@postgresql.org Mon Dec 22 16:38:16 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2959DD1B45D for ; Thu, 18 Dec 2003 17:05:51 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52007-04 for ; Thu, 18 Dec 2003 13:05:01 -0400 (AST) Received: from smtp-slave1.fsdata.se (smtp-gw.fsdata.se [195.35.82.150]) by svr1.postgresql.org (Postfix) with ESMTP id 3CF6FD1B45E for ; Thu, 18 Dec 2003 13:05:00 -0400 (AST) Received: from www3.aname.net (www3.aname.net [62.119.28.103]) by smtp-slave1.fsdata.se (8.12.10/8.12.0) with ESMTP id hBIH4qIY023254 for ; Thu, 18 Dec 2003 18:04:52 +0100 Received: from www3.aname.net (localhost [127.0.0.1]) by www3.aname.net (8.12.10/8.12.10) with ESMTP id hBIH4qgX010135 for ; Thu, 18 Dec 2003 18:04:52 +0100 Received: (from servi008@localhost) by www3.aname.net (8.12.10/8.12.0/Submit) id hBIH4qJj010134; Thu, 18 Dec 2003 18:04:52 +0100 Date: Thu, 18 Dec 2003 18:04:52 +0100 From: Conny Thimren Reply-To: conny.thimren@servit.se Subject: general peformance question To: pgsql-performance@postgresql.org Message-ID: MIME-Version: 1.0 Content-Type: TEXT/plain; CHARSET=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Importance: Normal X-Mailer: FS Webmail v2.0 X-FSDATA-AntiSpamScore: 1.8 X-FSDATA-AntiSpamRules: MIME_QP_DEFICIENT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/285 X-Sequence-Number: 5145 Hi, This a kind of newbie-question. I've been using Postgres for a long time in= a low transction environment - and it is great. Now I've got an inquiry for using Postgresql in a heavy-load on-line system= . This system must handle something like 20 questions per sec with a respon= se time at 1/10 sec. Each question will result in approx 5-6 reads and a co= uple of updates. Anybody have a feeling if this is realistic on a Intelbased Linux server wi= th Postgresql. Ofcourse I know that this is too little info for an exact an= swer but - as I said - maybe someone can give a hint if it's possible. Mayb= e someone with heavy-load can give an example of what is possible... Regards Conny Thimr=E9n From pgsql-performance-owner@postgresql.org Thu Dec 18 14:44:31 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 555DDD1B906 for ; Thu, 18 Dec 2003 18:44:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 64935-03 for ; Thu, 18 Dec 2003 14:43:41 -0400 (AST) Received: from morework.geizhals.at (home.geizhals.at [213.229.14.34]) by svr1.postgresql.org (Postfix) with ESMTP id DC588D1B8C0 for ; Thu, 18 Dec 2003 14:43:38 -0400 (AST) Received: from localhost (morework [127.0.0.1]) by morework.geizhals.at (Postfix) with ESMTP id 84B21524176 for ; Thu, 18 Dec 2003 19:43:25 +0100 (CET) Received: from geizhals.at (chello080110242194.117.11.tuwien.teleweb.at [80.110.242.194]) by morework.geizhals.at (Postfix) with ESMTP id 4623D52410E for ; Thu, 18 Dec 2003 19:43:25 +0100 (CET) Message-ID: <3FE1F5A1.6080908@geizhals.at> Date: Thu, 18 Dec 2003 19:44:49 +0100 From: "Marinos J. Yannikos" User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: why do optimizer parameters have to be set manually? Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by AMaViS snapshot-20020222 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/264 X-Sequence-Number: 5124 Hi, it seems to me that the optimizer parameters (like random_page_cost etc.) could easily be calculated and adjusted dynamically be the DB backend based on the planner's cost estimates and actual run times for different queries. Perhaps the developers could comment on that? I'm not sure how the parameters are used internally (apart from whatever "EXPLAIN" shows), but if cpu_operator_cost is the same for all operators, this should probably also be adjusted for individual operators (I suppose that ">" is not as costly as "~*"). As far as the static configuration is concerned, I'd be interested in other users' parameters and hardware configurations. Here's ours (for a write-intensive db that also performs many queries with regular expression matching): effective_cache_size = 1000000 # typically 8KB each #random_page_cost = 0.2 # units are one sequential page fetch cost random_page_cost = 3 # units are one sequential page fetch cost #cpu_tuple_cost = 0.01 # (same) cpu_index_tuple_cost = 0.01 # (same) 0.1 #cpu_operator_cost = 0.0025 # (same) cpu_operator_cost = 0.025 # (same) other options: shared_buffers = 240000 # 2*max_connections, min 16, typically 8KB each max_fsm_relations = 10000 # min 10, fsm is free space map, ~40 bytes max_fsm_pages = 10000000 # min 1000, fsm is free space map, ~6 bytes #max_locks_per_transaction = 20 # min 10 wal_buffers = 128 # min 4, typically 8KB each sort_mem = 800000 # min 64, size in KB vacuum_mem = 100000 # min 1024, size in KB checkpoint_segments = 80 # in logfile segments, min 1, 16MB each checkpoint_timeout = 300 # range 30-3600, in seconds commit_delay = 100000 # range 0-100000, in microseconds commit_siblings = 5 # range 1-1000 12GB RAM, dual 2,80GHz Xeon, 6x 10K rpm disks in a RAID-5, Linux 2.4.23 with HT enabled. Regards, Marinos From pgsql-performance-owner@postgresql.org Thu Dec 18 16:31:22 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C89D7D1B4C8 for ; Thu, 18 Dec 2003 20:31:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77069-06 for ; Thu, 18 Dec 2003 16:30:33 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id EEDF2D1B53E for ; Thu, 18 Dec 2003 16:30:31 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBIKUVU6044642 for ; Thu, 18 Dec 2003 20:30:31 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBIKA0IS019960 for pgsql-performance@postgresql.org; Thu, 18 Dec 2003 20:10:00 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: why do optimizer parameters have to be set manually? Date: Thu, 18 Dec 2003 14:17:43 -0500 Organization: Hub.Org Networking Services Lines: 23 Message-ID: <60n09q9bqg.fsf@dev6.int.libertyrms.info> References: <3FE1F5A1.6080908@geizhals.at> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@news.hub.org User-Agent: Gnus/5.1003 (Gnus v5.10.3) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:986zF5HVCKX8G4/7FrcwSG24Oek= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/266 X-Sequence-Number: 5126 mjy@geizhals.at ("Marinos J. Yannikos") writes: > it seems to me that the optimizer parameters (like random_page_cost > etc.) could easily be calculated and adjusted dynamically be the DB > backend based on the planner's cost estimates and actual run times for > different queries. Perhaps the developers could comment on that? Yes, it seems like a Small Matter Of Programming. http://wombat.doc.ic.ac.uk/foldoc/foldoc.cgi?SMOP In seriousness, yes, it would seem a reasonable idea to calculate some of these values a bit more dynamically. I would be inclined to start with something that ran a workload, and provided static values based on how that workload went. That would require NO intervention inside the DB server; it could be accomplished simply by writing a database script. Feel free to contribute either a script or a backend "hack"... -- let name="cbbrowne" and tld="libertyrms.info" in String.concat "@" [name;tld];; Christopher Browne (416) 646 3304 x124 (land) From pgsql-performance-owner@postgresql.org Mon Dec 22 16:39:10 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 697E0D1B442 for ; Thu, 18 Dec 2003 19:42:09 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68591-08 for ; Thu, 18 Dec 2003 15:41:20 -0400 (AST) Received: from smtp4.hy.skanova.net (smtp4.hy.skanova.net [195.67.199.133]) by svr1.postgresql.org (Postfix) with ESMTP id 62E3FD1B444 for ; Thu, 18 Dec 2003 15:41:02 -0400 (AST) Received: from Foo (h224n2fls35o962.telia.com [217.211.189.224]) by smtp4.hy.skanova.net (8.12.10/8.12.10) with SMTP id hBIJegr5012844 for ; Thu, 18 Dec 2003 20:40:42 +0100 (CET) Message-ID: <004901c3c59f$365de160$6400a8c0@telia.com> From: "Conny Thimren" To: Subject: general performance question Date: Thu, 18 Dec 2003 20:43:29 +0100 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0046_01C3C5A7.97E6C700" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4920.2300 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4920.2300 X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=1.2 tagged_above=0.0 required=5.0 tests=HTML_10_20 X-Spam-Level: * X-Archive-Number: 200312/287 X-Sequence-Number: 5147 This is a multi-part message in MIME format. ------=_NextPart_000_0046_01C3C5A7.97E6C700 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi,=20 This a kind of newbie-question. I've been using Postgres for a long time in= a low transction environment - and it is great.=20 Now I've got an inquiry for using Postgresql in a heavy-load on-line system= . This system must handle something like 20 questions per sec with a respon= se time at 1/10 sec. Each question will result in approx 5-6 reads and a co= uple of updates.=20 Anybody have a feeling if this is realistic on a Intelbased Linux server wi= th Postgresql. Ofcourse I know that this is too little info for an exact an= swer but - as I said - maybe someone can give a hint if it's possible. Mayb= e someone with heavy-load can give an example of what is possible...=20 Regards=20 Conny Thimr=E9n=20 ------=_NextPart_000_0046_01C3C5A7.97E6C700 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi= ,
This a=20 kind of newbie-question. I've been using Postgres for a long time in a low= =20 transction environment - and it is great.

Now I've got an inquiry f= or=20 using Postgresql in a heavy-load on-line system. This system must handle=20 something like 20 questions per sec with a response time at 1/10 sec. Each= =20 question will result in approx 5-6 reads and a couple of updates.
Anybo= dy=20 have a feeling if this is realistic on a Intelbased Linux server with=20 Postgresql. Ofcourse I know that this is too little info for an exact answe= r but=20 - as I said - maybe someone can give a hint if it's possible. Maybe someone= with=20 heavy-load can give an example of what is possible...

Regards
C= onny=20 Thimr=E9n

------=_NextPart_000_0046_01C3C5A7.97E6C700-- From pgsql-performance-owner@postgresql.org Thu Dec 18 15:57:44 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 28CFDD1B437 for ; Thu, 18 Dec 2003 19:57:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73381-03 for ; Thu, 18 Dec 2003 15:56:55 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 111F3D1B48D for ; Thu, 18 Dec 2003 15:56:37 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBIJuL19022687; Thu, 18 Dec 2003 14:56:21 -0500 (EST) To: "Marinos J. Yannikos" Cc: pgsql-performance@postgresql.org Subject: Re: why do optimizer parameters have to be set manually? In-reply-to: <3FE1F5A1.6080908@geizhals.at> References: <3FE1F5A1.6080908@geizhals.at> Comments: In-reply-to "Marinos J. Yannikos" message dated "Thu, 18 Dec 2003 19:44:49 +0100" Date: Thu, 18 Dec 2003 14:56:20 -0500 Message-ID: <22686.1071777380@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/265 X-Sequence-Number: 5125 "Marinos J. Yannikos" writes: > it seems to me that the optimizer parameters (like random_page_cost > etc.) could easily be calculated and adjusted dynamically be the DB > backend based on the planner's cost estimates and actual run times for > different queries. Perhaps the developers could comment on that? No, they are not that easy to determine. In particular I think the idea of automatically feeding back error measurements is hopeless, because you cannot tell which parameters are wrong. > I'm not sure how the parameters are used internally (apart from whatever > "EXPLAIN" shows), but if cpu_operator_cost is the same for all > operators, this should probably also be adjusted for individual > operators (I suppose that ">" is not as costly as "~*"). In theory perhaps, but in practice this is far down in the noise in most situations. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Dec 18 19:19:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A20DAD1B43E for ; Thu, 18 Dec 2003 23:19:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01372-09 for ; Thu, 18 Dec 2003 19:19:10 -0400 (AST) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id F3D73D1B459 for ; Thu, 18 Dec 2003 19:19:06 -0400 (AST) Received: (qmail 13026 invoked from network); 18 Dec 2003 23:18:49 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 18 Dec 2003 23:18:49 -0000 Subject: is it possible to get the optimizer to use indexes with a like clause From: Dave Cramer Reply-To: pg@fastcrypt.com To: pgsql-performance@postgresql.org In-Reply-To: <60n09q9bqg.fsf@dev6.int.libertyrms.info> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1071789528.1629.10.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 18 Dec 2003 18:18:48 -0500 Content-Transfer-Encoding: 7bit X-AntiVirus: scanned for viruses by AMaViS 0.2.1 (http://amavis.org/) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/267 X-Sequence-Number: 5127 It appears that the optimizer only uses indexes for = clause? Dave From pgsql-performance-owner@postgresql.org Mon Dec 22 16:39:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 626E7D1B8B6 for ; Fri, 19 Dec 2003 01:41:20 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14803-03 for ; Thu, 18 Dec 2003 21:40:33 -0400 (AST) Received: from main.gmane.org (main.gmane.org [80.91.224.249]) by svr1.postgresql.org (Postfix) with ESMTP id 24876D1B439 for ; Thu, 18 Dec 2003 21:40:29 -0400 (AST) Received: from root by main.gmane.org with local (Exim 3.35 #1 (Debian)) id 1AX9cl-0005Rv-00 for ; Fri, 19 Dec 2003 02:40:15 +0100 X-Injected-Via-Gmane: http://gmane.org/ To: pgsql-performance@postgresql.org Received: from sea.gmane.org ([80.91.224.252]) by main.gmane.org with esmtp (Exim 3.35 #1 (Debian)) id 1AX9an-0005QT-00 for ; Fri, 19 Dec 2003 02:38:13 +0100 Received: from news by sea.gmane.org with local (Exim 3.35 #1 (Debian)) id 1AX9an-0001T2-00 for ; Fri, 19 Dec 2003 02:38:13 +0100 From: Doug McNaught Subject: Re: is it possible to get the optimizer to use indexes Date: Thu, 18 Dec 2003 20:38:13 -0500 Lines: 12 Message-ID: <87d6alsi2i.fsf@asmodeus.mcnaught.org> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@sea.gmane.org User-Agent: Gnus/5.1002 (Gnus v5.10.2) Emacs/20.7 (gnu/linux) Cancel-Lock: sha1:3UKEZsz1UOTkQlhldXduaTM4I+o= X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/286 X-Sequence-Number: 5146 Christopher Kings-Lynne writes: >> It appears that the optimizer only uses indexes for = clause? > > The optimizer will used indexes for LIKE clauses, so long as the > clause is a prefix search, eg: > > SELECT * FROM test WHERE a LIKE 'prf%'; Doesn't this still depend on your locale? -Doug From pgsql-performance-owner@postgresql.org Thu Dec 18 21:34:56 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0C474D1B439 for ; Fri, 19 Dec 2003 01:34:55 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15855-03 for ; Thu, 18 Dec 2003 21:34:08 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id B3995D1B48E for ; Thu, 18 Dec 2003 21:34:03 -0400 (AST) Received: from familyhealth.com.au (work-40.internal [192.168.0.40]) by houston.familyhealth.com.au (8.12.9p1/8.12.9) with ESMTP id hBJ1XkoD062763; Fri, 19 Dec 2003 09:33:46 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <3FE256AA.8000707@familyhealth.com.au> Date: Fri, 19 Dec 2003 09:38:50 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pg@fastcrypt.com Cc: pgsql-performance@postgresql.org Subject: Re: is it possible to get the optimizer to use indexes References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> In-Reply-To: <1071789528.1629.10.camel@localhost.localdomain> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/268 X-Sequence-Number: 5128 > It appears that the optimizer only uses indexes for = clause? The optimizer will used indexes for LIKE clauses, so long as the clause is a prefix search, eg: SELECT * FROM test WHERE a LIKE 'prf%'; Chris From pgsql-performance-owner@postgresql.org Thu Dec 18 23:09:31 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DA271D1B478 for ; Fri, 19 Dec 2003 03:09:29 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21276-05 for ; Thu, 18 Dec 2003 23:08:43 -0400 (AST) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 2785DD1B47D for ; Thu, 18 Dec 2003 23:08:39 -0400 (AST) Received: (qmail 32257 invoked from network); 19 Dec 2003 03:08:37 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 19 Dec 2003 03:08:37 -0000 Subject: Re: is it possible to get the optimizer to use indexes From: Dave Cramer Reply-To: pg@fastcrypt.com To: Christopher Kings-Lynne Cc: pgsql-performance@postgresql.org In-Reply-To: <3FE256AA.8000707@familyhealth.com.au> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1071803317.1629.16.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 18 Dec 2003 22:08:37 -0500 Content-Transfer-Encoding: 7bit X-AntiVirus: scanned for viruses by AMaViS 0.2.1 (http://amavis.org/) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/269 X-Sequence-Number: 5129 after vacuum verbose analyze, I still get explain select * from isppm where item_upc_cd like '06038301234'; QUERY PLAN ----------------------------------------------------------------------- Seq Scan on isppm (cost=100000000.00..100009684.89 rows=2 width=791) Filter: (item_upc_cd ~~ '06038301234'::text) (2 rows) isp=# explain select * from isppm where item_upc_cd = '06038301234'; QUERY PLAN ------------------------------------------------------------------------ Index Scan using isppm_x0 on isppm (cost=0.00..5.86 rows=2 width=791) Index Cond: (item_upc_cd = '06038301234'::bpchar) (2 rows) Dave On Thu, 2003-12-18 at 20:38, Christopher Kings-Lynne wrote: > > It appears that the optimizer only uses indexes for = clause? > > The optimizer will used indexes for LIKE clauses, so long as the clause > is a prefix search, eg: > > SELECT * FROM test WHERE a LIKE 'prf%'; > > Chris > > From pgsql-performance-owner@postgresql.org Thu Dec 18 23:31:25 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 88CD3D1B4B7 for ; Fri, 19 Dec 2003 03:31:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20768-06 for ; Thu, 18 Dec 2003 23:30:36 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 7B31AD1B45A for ; Thu, 18 Dec 2003 23:30:32 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBJ3UVU6081650 for ; Fri, 19 Dec 2003 03:30:32 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBJ3SVB1078996 for pgsql-performance@postgresql.org; Fri, 19 Dec 2003 03:28:31 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: is it possible to get the optimizer to use indexes with a like clause Date: Thu, 18 Dec 2003 22:22:38 -0500 Organization: cbbrowne Computing Inc Lines: 21 Message-ID: References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> 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/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, linux) Cancel-Lock: sha1:Q3cISRgx/BUPdg4/ElO3oOiwRiA= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/270 X-Sequence-Number: 5130 pg@fastcrypt.com (Dave Cramer) wrote: > It appears that the optimizer only uses indexes for = clause? It can use indices only if there is a given prefix. Thus: where text_field like 'A%' can use the index, essentially transforming this into the clauses where text_field >= 'A' and text_field < 'B'. You can't get much out of an index for where text_field like '%SOMETHING' -- (reverse (concatenate 'string "moc.enworbbc" "@" "enworbbc")) http://www3.sympatico.ca/cbbrowne/wp.html "When the grammar checker identifies an error, it suggests a correction and can even makes some changes for you." -- Microsoft Word for Windows 2.0 User's Guide, p.35: From pgsql-performance-owner@postgresql.org Thu Dec 18 23:37:04 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2B053D1B45C for ; Fri, 19 Dec 2003 03:37:02 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19710-08 for ; Thu, 18 Dec 2003 23:36:17 -0400 (AST) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 328A7D1B441 for ; Thu, 18 Dec 2003 23:36:13 -0400 (AST) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 5D686355CF; Thu, 18 Dec 2003 19:36:02 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 5BEFB3556C; Thu, 18 Dec 2003 19:36:02 -0800 (PST) Date: Thu, 18 Dec 2003 19:36:02 -0800 (PST) From: Stephan Szabo To: Dave Cramer Cc: Christopher Kings-Lynne , pgsql-performance@postgresql.org Subject: Re: is it possible to get the optimizer to use indexes In-Reply-To: <1071803317.1629.16.camel@localhost.localdomain> Message-ID: <20031218193514.Q20465@megazone.bigpanda.com> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> <1071803317.1629.16.camel@localhost.localdomain> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/271 X-Sequence-Number: 5131 On Thu, 18 Dec 2003, Dave Cramer wrote: > after vacuum verbose analyze, I still get > > explain select * from isppm where item_upc_cd like '06038301234'; > QUERY PLAN > ----------------------------------------------------------------------- > Seq Scan on isppm (cost=100000000.00..100009684.89 rows=2 width=791) > Filter: (item_upc_cd ~~ '06038301234'::text) > (2 rows) IIRC, the other limitation is that it only does so in "C" locale due to wierdnesses in other locales. From pgsql-performance-owner@postgresql.org Thu Dec 18 23:44:54 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 48C79D1B46E for ; Fri, 19 Dec 2003 03:44:51 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 29658-03 for ; Thu, 18 Dec 2003 23:44:05 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id E2BEFD1B45A for ; Thu, 18 Dec 2003 23:44:00 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBJ3i019006122; Thu, 18 Dec 2003 22:44:00 -0500 (EST) To: pg@fastcrypt.com Cc: Christopher Kings-Lynne , pgsql-performance@postgresql.org Subject: Re: is it possible to get the optimizer to use indexes In-reply-to: <1071803317.1629.16.camel@localhost.localdomain> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> <1071803317.1629.16.camel@localhost.localdomain> Comments: In-reply-to Dave Cramer message dated "18 Dec 2003 22:08:37 -0500" Date: Thu, 18 Dec 2003 22:44:00 -0500 Message-ID: <6121.1071805440@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/272 X-Sequence-Number: 5132 Dave Cramer writes: > after vacuum verbose analyze, I still get [a seqscan] The other gating factor is that you have to have initdb'd in C locale. Non-C locales tend to use wild and wooly sort orders that are not compatible with what LIKE needs. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 19 05:51:10 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A491AD1B486 for ; Fri, 19 Dec 2003 09:51:08 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71215-09 for ; Fri, 19 Dec 2003 05:50:20 -0400 (AST) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id A76B5D1B48F for ; Fri, 19 Dec 2003 05:50:17 -0400 (AST) Received: (qmail 7249 invoked from network); 19 Dec 2003 09:50:13 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 19 Dec 2003 09:50:13 -0000 Subject: Re: is it possible to get the optimizer to use indexes From: Dave Cramer Reply-To: pg@fastcrypt.com To: Tom Lane Cc: Christopher Kings-Lynne , pgsql-performance@postgresql.org In-Reply-To: <6121.1071805440@sss.pgh.pa.us> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> <1071803317.1629.16.camel@localhost.localdomain> <6121.1071805440@sss.pgh.pa.us> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1071827412.1627.25.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 19 Dec 2003 04:50:12 -0500 Content-Transfer-Encoding: 7bit X-AntiVirus: scanned for viruses by AMaViS 0.2.1 (http://amavis.org/) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/273 X-Sequence-Number: 5133 So even in a north-american locale, such as en_CA this will be a problem? Dave On Thu, 2003-12-18 at 22:44, Tom Lane wrote: > Dave Cramer writes: > > after vacuum verbose analyze, I still get [a seqscan] > > The other gating factor is that you have to have initdb'd in C locale. > Non-C locales tend to use wild and wooly sort orders that are not > compatible with what LIKE needs. > > regards, tom lane > > From pgsql-performance-owner@postgresql.org Fri Dec 19 09:52:59 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 28034D1B449 for ; Fri, 19 Dec 2003 13:52:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03274-04 for ; Fri, 19 Dec 2003 09:52:12 -0400 (AST) Received: from morework.geizhals.at (home.geizhals.at [213.229.14.34]) by svr1.postgresql.org (Postfix) with ESMTP id 88178D1B43A for ; Fri, 19 Dec 2003 09:52:08 -0400 (AST) Received: from localhost (morework [127.0.0.1]) by morework.geizhals.at (Postfix) with ESMTP id 4A808524177; Fri, 19 Dec 2003 14:51:58 +0100 (CET) Received: from geizhals.at (chello080110242194.117.11.tuwien.teleweb.at [80.110.242.194]) by morework.geizhals.at (Postfix) with ESMTP id 0046C52416C; Fri, 19 Dec 2003 14:51:57 +0100 (CET) Message-ID: <3FE302D2.7080901@geizhals.at> Date: Fri, 19 Dec 2003 14:53:22 +0100 From: "Marinos J. Yannikos" User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: why do optimizer parameters have to be set manually? References: <3FE1F5A1.6080908@geizhals.at> <22686.1071777380@sss.pgh.pa.us> In-Reply-To: <22686.1071777380@sss.pgh.pa.us> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by AMaViS snapshot-20020222 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/274 X-Sequence-Number: 5134 Tom Lane wrote: > No, they are not that easy to determine. In particular I think the idea > of automatically feeding back error measurements is hopeless, because > you cannot tell which parameters are wrong. Isn't it just a matter of solving an equation system with n variables (n being the number of parameters), where each equation stands for the calculation of the run time of a particular query? I.e. something like this for a sequential scan over 1000 rows with e.g. 2 operators used per iteration that took 2 seconds (simplified so that the costs are actual timings and not relative costs to a base value): 1000 * sequential_scan_cost + 1000 * 2 * cpu_operator_cost = 2.0 seconds With a sufficient number of equations (not just n, since not all query plans use all the parameters) this system can be solved for the particular query mix that was used. E.g. with a second sequential scan over 2000 rows with 1 operator per iteration that took 3 seconds you can derive: sequential_scan_cost = 1ms cpu_operator_cost = 0.5ms This could probably be implemented with very little overhead compared to the actual run times of the queries. Regard, Marinos From pgsql-performance-owner@postgresql.org Fri Dec 19 10:39:30 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A463BD1B44B for ; Fri, 19 Dec 2003 14:39:23 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15134-01 for ; Fri, 19 Dec 2003 10:38:38 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 0E037D1B43A for ; Fri, 19 Dec 2003 10:38:29 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBJEcE19009225; Fri, 19 Dec 2003 09:38:15 -0500 (EST) To: pg@fastcrypt.com Cc: Christopher Kings-Lynne , pgsql-performance@postgresql.org Subject: Re: is it possible to get the optimizer to use indexes In-reply-to: <1071827412.1627.25.camel@localhost.localdomain> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> <1071803317.1629.16.camel@localhost.localdomain> <6121.1071805440@sss.pgh.pa.us> <1071827412.1627.25.camel@localhost.localdomain> Comments: In-reply-to Dave Cramer message dated "19 Dec 2003 04:50:12 -0500" Date: Fri, 19 Dec 2003 09:38:14 -0500 Message-ID: <9224.1071844694@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/275 X-Sequence-Number: 5135 Dave Cramer writes: > So even in a north-american locale, such as en_CA this will be a > problem? If it's not "C" we won't try to optimize LIKE. I know en_US does not work (case-insensitive, funny rules about spaces, etc) and I would expect en_CA has the same issues. If you're using 7.4 you have the option to create a special index instead of re-initdb'ing your whole database. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 19 11:08:06 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 313E7D1B43A for ; Fri, 19 Dec 2003 15:08:05 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 16914-07 for ; Fri, 19 Dec 2003 11:07:19 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 76D21D1B42F for ; Fri, 19 Dec 2003 11:07:13 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBJF7F19009383; Fri, 19 Dec 2003 10:07:15 -0500 (EST) To: "Marinos J. Yannikos" Cc: pgsql-performance@postgresql.org Subject: Re: why do optimizer parameters have to be set manually? In-reply-to: <3FE302D2.7080901@geizhals.at> References: <3FE1F5A1.6080908@geizhals.at> <22686.1071777380@sss.pgh.pa.us> <3FE302D2.7080901@geizhals.at> Comments: In-reply-to "Marinos J. Yannikos" message dated "Fri, 19 Dec 2003 14:53:22 +0100" Date: Fri, 19 Dec 2003 10:07:15 -0500 Message-ID: <9382.1071846435@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/276 X-Sequence-Number: 5136 "Marinos J. Yannikos" writes: > Tom Lane wrote: >> No, they are not that easy to determine. In particular I think the idea >> of automatically feeding back error measurements is hopeless, because >> you cannot tell which parameters are wrong. > Isn't it just a matter of solving an equation system with n variables (n > being the number of parameters), where each equation stands for the > calculation of the run time of a particular query? If we knew all the variables involved, it might be (though since the equations would be nonlinear, the solution would be more difficult than you suppose). The real problems are: 1. There is lots of noise in any real-world measurement, mostly due to competition from other processes. 2. There are effects we don't even try to model, such as the current contents of kernel cache. Everybody who's done any work with Postgres knows that for small-to-middling tables, running the same query twice in a row will yield considerably different runtimes, because the second time through all the data will be in kernel cache. But we don't have any useful way to model that in the optimizer, since we can't see what the kernel has in its buffers. 3. Even for the effects we do try to model, some of the equations are pretty ad-hoc and might not fit real data very well. (I have little confidence in the current correction for index order correlation, for example.) In short, if you just try to fit the present cost equations to real data, what you'll get will inevitably be "garbage in, garbage out". You could easily end up with parameter values that are much less realistic than the defaults. Over time we'll doubtless improve the optimizer's cost models, and someday we might get to a point where this wouldn't be a fool's errand, but I don't see it happening in the foreseeable future. I think a more profitable approach is to set up special test code to try to approximate the value of individual parameters measured in isolation. For instance, the current default of 4.0 for random_page_cost was developed through rather extensive testing a few years ago, and I think it's still a decent average value (for the case where you are actually doing I/O, mind you). But if your disks have particularly fast or slow seek times, maybe it's not good for you. It might be useful to package up a test program that repeats those measurements on particular systems --- though the problem of noisy measurements still applies. It is not easy or cheap to get a measurement that isn't skewed by kernel caching behavior. (You need a test file significantly larger than RAM, and even then you'd better repeat the measurement quite a few times to see how much noise there is in it.) regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 19 12:01:35 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 515FED1B48D for ; Fri, 19 Dec 2003 16:01:33 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 25138-10 for ; Fri, 19 Dec 2003 12:00:44 -0400 (AST) Received: from webware.ee (webware.ee [212.7.7.3]) by svr1.postgresql.org (Postfix) with ESMTP id 3AC97D1B471 for ; Fri, 19 Dec 2003 12:00:30 -0400 (AST) Received: from webware.ee (pc234.host4.starman.ee [62.65.196.234]) by webware.ee (Postfix) with ESMTP id D7DF1378100; Fri, 19 Dec 2003 18:01:34 +0200 (EET) Message-ID: <3FE32099.40901@webware.ee> Date: Fri, 19 Dec 2003 18:00:25 +0200 From: =?ISO-8859-1?Q?Erki_Kaldj=E4rv?= Organization: Webware =?ISO-8859-1?Q?O=DC?= User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-GB; rv:1.0.1) Gecko/20020823 Netscape/7.0 X-Accept-Language: et,en MIME-Version: 1.0 To: Tom Lane Cc: pg@fastcrypt.com, Christopher Kings-Lynne , pgsql-performance@postgresql.org Subject: Re: is it possible to get the optimizer to use indexes References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> <1071803317.1629.16.camel@localhost.localdomain> <6121.1071805440@sss.pgh.pa.us> <1071827412.1627.25.camel@localhost.localdomain> <9224.1071844694@sss.pgh.pa.us> Content-Type: multipart/alternative; boundary="------------040301000406030407010001" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/277 X-Sequence-Number: 5137 --------------040301000406030407010001 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit Hello, i got indexes to work with "text_pattern_ops" for locale et_EE. So instead of: create index some_index_name on some_table(some_text_field); nor create index some_index_name on some_table(some_text_field text_ops); try to create index as follows: create index some_index_name on some_table(some_text_field text_pattern_ops); Note that text_pattern_ops is available pg >= 7.4. Regards, Erki Kaldj�rv Webware O� www.webware.ee Tom Lane wrote: >Dave Cramer writes: > > >>So even in a north-american locale, such as en_CA this will be a >>problem? >> >> > >If it's not "C" we won't try to optimize LIKE. I know en_US does not >work (case-insensitive, funny rules about spaces, etc) and I would >expect en_CA has the same issues. > >If you're using 7.4 you have the option to create a special index >instead of re-initdb'ing your whole database. > > regards, tom lane > >---------------------------(end of broadcast)--------------------------- >TIP 4: Don't 'kill -9' the postmaster > > --------------040301000406030407010001 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hello,

i got indexes to work with "text_pattern_ops" for locale et_EE.

So instead of:
create index some_index_name on some_table(some_text_field);

nor

create index some_index_name on some_table(some_text_field text_ops);

try to create index as follows:
create index some_index_name on some_table(some_text_field text_pattern_ops);

Note that text_pattern_ops is available pg >= 7.4.

Regards,

Erki Kaldjärv
Webware OÜ
www.webware.ee

Tom Lane wrote:
Dave Cramer <pg@fastcrypt.com> writes:
  
So even in a north-american locale, such as en_CA this will be a
problem?
    

If it's not "C" we won't try to optimize LIKE.  I know en_US does not
work (case-insensitive, funny rules about spaces, etc) and I would
expect en_CA has the same issues.

If you're using 7.4 you have the option to create a special index
instead of re-initdb'ing your whole database.

			regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

--------------040301000406030407010001-- From pgsql-performance-owner@postgresql.org Fri Dec 19 14:06:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 32D1AD1BA6E for ; Fri, 19 Dec 2003 18:05:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 46488-05 for ; Fri, 19 Dec 2003 14:04:57 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 3F6D8D1B8EA for ; Fri, 19 Dec 2003 14:00:45 -0400 (AST) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id hBJI0cU6001986 for ; Fri, 19 Dec 2003 18:00:38 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id hBJHiDIq063765 for pgsql-performance@postgresql.org; Fri, 19 Dec 2003 17:44:13 GMT From: William Yu X-Newsgroups: comp.databases.postgresql.performance Subject: Re: why do optimizer parameters have to be set manually? Date: Fri, 19 Dec 2003 09:44:17 -0800 Organization: Hub.Org Networking Services Lines: 10 Message-ID: References: <3FE1F5A1.6080908@geizhals.at> <22686.1071777380@sss.pgh.pa.us> <3FE302D2.7080901@geizhals.at> <9382.1071846435@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@news.hub.org User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en In-Reply-To: <9382.1071846435@sss.pgh.pa.us> To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/278 X-Sequence-Number: 5138 Tom Lane wrote: > easy or cheap to get a measurement that isn't skewed by kernel caching > behavior. (You need a test file significantly larger than RAM, and > even then you'd better repeat the measurement quite a few times to see > how much noise there is in it.) I found a really fast way in Linux to flush the kernel cache and that is to unmount the drive and then remount. Beats having to read though a file > RAM everytime. From pgsql-performance-owner@postgresql.org Mon Dec 22 06:52:49 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 585D9D1B4AE for ; Mon, 22 Dec 2003 10:52:48 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97397-01 for ; Mon, 22 Dec 2003 06:52:00 -0400 (AST) Received: from nitrogen.id.pl (nitrogen.id.pl [193.178.214.5]) by svr1.postgresql.org (Postfix) with SMTP id D039AD1B518 for ; Mon, 22 Dec 2003 06:51:26 -0400 (AST) Received: (qmail 5148 invoked by uid 0); 22 Dec 2003 10:51:13 -0000 Received: from unknown (HELO siaco.id.pl) (213.25.114.8) by smtp.id.pl with SMTP; 22 Dec 2003 10:51:12 -0000 Received: (qmail 24417 invoked by uid 1000); 22 Dec 2003 10:51:08 -0000 Resent-From: Ryszard Lach Resent-Date: Mon, 22 Dec 2003 11:51:08 +0100 Resent-Message-ID: <20031222105108.GB23673@siaco.id.pl> Resent-To: pgsql-performance@postgresql.org Date: Mon, 22 Dec 2003 11:39:18 +0100 From: Ryszard Lach To: pgsql-performance@postgresql.org Subject: "select max/count(id)" not using index Message-ID: <20031222103918.GA23673@siaco.id.pl> Reply-To: Ryszard Lach Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-2 Content-Disposition: inline X-My-GPG-Key: echo | mail -s "send key pub" ryszard@lach.name User-Agent: Mutt/1.5.4i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/279 X-Sequence-Number: 5139 Hi. I have a table with 24k records and btree index on column 'id'. Is this normal, that 'select max(id)' or 'select count(id)' causes a sequential scan? It takes over 24 seconds (on a pretty fast machine): => explain ANALYZE select max(id) from ogloszenia; QUERY PLAN ---------------------------------------------------------------------- Aggregate (cost=3511.05..3511.05 rows=1 width=4) (actual time=24834.629..24834.629 rows=1 loops=1) -> Seq Scan on ogloszenia (cost=0.00..3473.04 rows=15204 width=4) (actual time=0.013..24808.377 rows=16873 loops=1) Total runtime: 24897.897 ms Maybe it's caused by a number of varchar fields in this table? However, 'id' column is 'integer' and is primary key. Clustering table on index created on 'id' makes such a queries many faster, but they still use a sequential scan. Richard. -- "First they ignore you. Then they laugh at you. Then they fight you. Then you win." - Mohandas Gandhi. From pgsql-performance-owner@postgresql.org Mon Dec 22 06:57:46 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ECA32D1B456 for ; Mon, 22 Dec 2003 10:57:44 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92280-08 for ; Mon, 22 Dec 2003 06:56:58 -0400 (AST) Received: from relay01.kbs.net.au (relay01.kbs.net.au [203.220.32.149]) by svr1.postgresql.org (Postfix) with ESMTP id 43F2AD1B496 for ; Mon, 22 Dec 2003 06:56:55 -0400 (AST) Received: from [203.221.246.88] (helo=familyhealth.com.au) by relay01.kbs.net.au with esmtp (Exim 3.36 #2) id 1AYNk6-0005RM-00; Mon, 22 Dec 2003 21:56:55 +1100 Message-ID: <3FE6CDF2.10608@familyhealth.com.au> Date: Mon, 22 Dec 2003 18:56:50 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Ryszard Lach Cc: pgsql-performance@postgresql.org Subject: Re: "select max/count(id)" not using index References: <20031222103918.GA23673@siaco.id.pl> In-Reply-To: <20031222103918.GA23673@siaco.id.pl> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/280 X-Sequence-Number: 5140 > I have a table with 24k records and btree index on column 'id'. Is this > normal, that 'select max(id)' or 'select count(id)' causes a sequential > scan? It takes over 24 seconds (on a pretty fast machine): > > => explain ANALYZE select max(id) from ogloszenia; Yes, it is. It is a known issue with Postgres's extensible operator architecture. The work around is to have an index on the id column and do this instead: SELECT id FROM ogloszenia ORDER BY id DESC LIMIT 1; Which will be really fast. Chris From pgsql-performance-owner@postgresql.org Mon Dec 22 11:29:48 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CF1A7D1B42F for ; Mon, 22 Dec 2003 15:29:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28934-10 for ; Mon, 22 Dec 2003 11:29:00 -0400 (AST) Received: from mailout04.sul.t-online.com (mailout04.sul.t-online.com [194.25.134.18]) by svr1.postgresql.org (Postfix) with ESMTP id E61CED1B44D for ; Mon, 22 Dec 2003 11:28:31 -0400 (AST) Received: from fwd04.aul.t-online.de by mailout04.sul.t-online.com with smtp id 1AYNn6-0004cf-03; Mon, 22 Dec 2003 12:00:00 +0100 Received: from router.azrael.de (XHtrgcZ1Qe5BoippkwgC4OGlA9YmT16ZUYhl4QH2-TxgYlIvjV4u4O@[80.141.232.167]) by fmrl04.sul.t-online.com with esmtp id 1AYNmu-1y12US0; Mon, 22 Dec 2003 11:59:48 +0100 Received: from azrael.azrael.de (azrael.azrael.de [192.168.202.18]) by router.azrael.de (8.11.4/8.11.4) with ESMTP id hBMAxkt15644; Mon, 22 Dec 2003 11:59:47 +0100 Date: Mon, 22 Dec 2003 11:59:58 +0100 From: Evil Azrael X-Priority: 3 (Normal) Message-ID: <110164283326.20031222115958@evilazrael.de> To: Ryszard Lach Cc: pgsql-performance@postgresql.org Subject: Re: "select max/count(id)" not using index In-Reply-To: <20031222103918.GA23673@siaco.id.pl> References: <20031222103918.GA23673@siaco.id.pl> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-2 Content-Transfer-Encoding: 8bit X-Seen: false X-ID: XHtrgcZ1Qe5BoippkwgC4OGlA9YmT16ZUYhl4QH2-TxgYlIvjV4u4O@t-dialin.net X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/283 X-Sequence-Number: 5143 Guten Tag Ryszard Lach, Am Montag, 22. Dezember 2003 um 11:39 schrieben Sie: RL> Hi. RL> I have a table with 24k records and btree index on column 'id'. Is this RL> normal, that 'select max(id)' or 'select count(id)' causes a sequential RL> scan? It takes over 24 seconds (on a pretty fast machine): Yes, that was occasionally discussed on the mailinglists. For the max(id) you can use instead "SELECT id FROM table ORDER BY id DESC LIMIT 1" Christoph Nelles =>> explain ANALYZE select max(id) from ogloszenia; RL> QUERY PLAN RL> ---------------------------------------------------------------------- RL> Aggregate (cost=3511.05..3511.05 rows=1 width=4) (actual RL> time=24834.629..24834.629 rows=1 loops=1) RL> -> Seq Scan on ogloszenia (cost=0.00..3473.04 rows=15204 width=4) RL> (actual time=0.013..24808.377 rows=16873 loops=1) RL> Total runtime: 24897.897 ms RL> Maybe it's caused by a number of varchar fields in this table? However, RL> 'id' column is 'integer' and is primary key. RL> Clustering table on index created on 'id' makes such a queries RL> many faster, but they still use a sequential scan. RL> Richard. -- Mit freundlichen Gr�ssen Evil Azrael mailto:evilazrael@evilazrael.de From pgsql-performance-owner@postgresql.org Mon Dec 22 07:03:58 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 31E56D1B4B7 for ; Mon, 22 Dec 2003 11:03:56 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97406-07 for ; Mon, 22 Dec 2003 07:03:08 -0400 (AST) Received: from kix.fsv.cvut.cz (Kix.FSV.CVUT.CZ [147.32.129.84]) by svr1.postgresql.org (Postfix) with ESMTP id B2A51D1B487 for ; Mon, 22 Dec 2003 07:03:04 -0400 (AST) Received: from localhost (stehule@localhost) by kix.fsv.cvut.cz (8.11.6/8.11.6) with ESMTP id hBMB35b27958; Mon, 22 Dec 2003 12:03:05 +0100 Date: Mon, 22 Dec 2003 12:03:05 +0100 (CET) From: Pavel Stehule To: Ryszard Lach Cc: pgsql-performance@postgresql.org, Ryszard Lach Subject: Re: "select max/count(id)" not using index In-Reply-To: <20031222103918.GA23673@siaco.id.pl> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/281 X-Sequence-Number: 5141 Hello It is normal behavior PostgreSQL. Use SELECT id FROM tabulka ORDER BY id DESC LIMIT 1; regards Pavel On Mon, 22 Dec 2003, Ryszard Lach wrote: > Hi. > > I have a table with 24k records and btree index on column 'id'. Is this > normal, that 'select max(id)' or 'select count(id)' causes a sequential > scan? It takes over 24 seconds (on a pretty fast machine): > > => explain ANALYZE select max(id) from ogloszenia; > QUERY PLAN > ---------------------------------------------------------------------- > Aggregate (cost=3511.05..3511.05 rows=1 width=4) (actual > time=24834.629..24834.629 rows=1 loops=1) > -> Seq Scan on ogloszenia (cost=0.00..3473.04 rows=15204 width=4) > (actual time=0.013..24808.377 rows=16873 loops=1) > Total runtime: 24897.897 ms > > Maybe it's caused by a number of varchar fields in this table? However, > 'id' column is 'integer' and is primary key. > > Clustering table on index created on 'id' makes such a queries > many faster, but they still use a sequential scan. > > Richard. > > From pgsql-performance-owner@postgresql.org Mon Dec 22 07:04:57 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 21365D1B487 for ; Mon, 22 Dec 2003 11:04:55 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 94268-09 for ; Mon, 22 Dec 2003 07:04:07 -0400 (AST) Received: from serwer.skawsoft.com.pl (skawina.eu.org [80.48.213.66]) by svr1.postgresql.org (Postfix) with ESMTP id 236E4D1B4CF for ; Mon, 22 Dec 2003 07:04:04 -0400 (AST) Received: from klaster.net (core-1.citynet.pl [80.48.135.69]) by serwer.skawsoft.com.pl (Postfix) with ESMTP id A53DF2B3CB; Mon, 22 Dec 2003 12:03:32 +0100 (CET) Message-ID: <3FE6CF91.6070300@klaster.net> Date: Mon, 22 Dec 2003 12:03:45 +0100 From: Tomasz Myrta User-Agent: Mozilla/5.0 (Windows; U; Win 9x 4.90; pl-PL; rv:1.5) Gecko/20031007 X-Accept-Language: pl, en-us, en MIME-Version: 1.0 To: Ryszard Lach Cc: pgsql-performance@postgresql.org Subject: Re: "select max/count(id)" not using index References: <20031222103918.GA23673@siaco.id.pl> In-Reply-To: <20031222103918.GA23673@siaco.id.pl> Content-Type: text/plain; charset=ISO-8859-2; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/282 X-Sequence-Number: 5142 Dnia 2003-12-22 11:39, U�ytkownik Ryszard Lach napisa�: > Hi. > > I have a table with 24k records and btree index on column 'id'. Is this > normal, that 'select max(id)' or 'select count(id)' causes a sequential > scan? It takes over 24 seconds (on a pretty fast machine): 'select count(id)' Yes, this is normal. Because of MVCC all rows must be checked and Postgres doesn't cache count(*) like Mysql. 'select max(id)' This is also normal, but try to change this query into: select id from some_table order by id desc limit 1; What is your Postgresql version? Regards, Tomasz Myrta From pgsql-performance-owner@postgresql.org Mon Dec 22 16:39:11 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4C441D1B44D for ; Mon, 22 Dec 2003 19:13:02 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69235-03 for ; Mon, 22 Dec 2003 15:12:13 -0400 (AST) Received: from wren.rentec.com (unknown [65.213.84.9]) by svr1.postgresql.org (Postfix) with ESMTP id 08706D1B4B1 for ; Mon, 22 Dec 2003 15:12:10 -0400 (AST) Received: from rentec.com (IDENT:618@murre.rentec.com [172.26.132.91]) by wren.rentec.com (8.12.1/8.12.1) with ESMTP id hBMJBt12012828 for ; Mon, 22 Dec 2003 14:11:55 -0500 (EST) Message-ID: <3FE741FA.4000002@rentec.com> Date: Mon, 22 Dec 2003 14:11:54 -0500 From: Michael Guerin User-Agent: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:1.4) Gecko/20030703 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: postgresql performance on linux port Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/288 X-Sequence-Number: 5148 I just restored a database running on a solaris box to a linux box and queries take forever to execute. The linux box is faster and has twice the memory allocated to postgresql, is there anything obvious that I should look at? It is using a journal file system. From pgsql-performance-owner@postgresql.org Mon Dec 22 16:47:24 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 05A7BD1B440 for ; Mon, 22 Dec 2003 20:47:22 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 83012-05 for ; Mon, 22 Dec 2003 16:46:33 -0400 (AST) Received: from mail.hive.nj2.inquent.com (mc.carriermail.com [205.178.180.9]) by svr1.postgresql.org (Postfix) with SMTP id AF045D1B518 for ; Mon, 22 Dec 2003 16:45:40 -0400 (AST) Received: (qmail 9261 invoked from network); 22 Dec 2003 20:45:26 -0000 Received: from unknown (HELO ?10.0.2.2?) (216.208.117.7) by 205.178.180.9 with SMTP; 22 Dec 2003 20:45:26 -0000 Subject: Re: general peformance question From: Rod Taylor To: conny.thimren@servit.se Cc: Postgresql Performance In-Reply-To: References: Content-Type: text/plain Message-Id: <1072125923.17267.55.camel@jester> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Mon, 22 Dec 2003 15:45:24 -0500 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/289 X-Sequence-Number: 5149 On Thu, 2003-12-18 at 12:04, Conny Thimren wrote: > Hi, > This a kind of newbie-question. I've been using Postgres for a long time in a low transction environment - and it is great. > > Now I've got an inquiry for using Postgresql in a heavy-load on-line system. This system must handle something like 20 questions per sec with a response time at 1/10 sec. Each question will result in approx 5-6 reads and a couple of updates. > Anybody have a feeling if this is realistic on a Intelbased Linux server with Postgresql. Ofcourse I know that this is too little info for an exact answer but - as I said - maybe someone can give a hint if it's possible. Maybe someone with heavy-load can give an example of what is possible... Ok, is that 20 questions per second (20 in parallel taking 1 second each) or serialized taking 50ms each. Are they simple selects / updates (less than 10 rows in result set, very simple joins) or are they monster 30 table join queries? From pgsql-performance-owner@postgresql.org Wed Dec 24 20:53:14 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A1333D1B44B for ; Mon, 22 Dec 2003 21:15:35 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 84583-06 for ; Mon, 22 Dec 2003 17:14:47 -0400 (AST) Received: from ganymede.hub.org (u46n208.hfx.eastlink.ca [24.222.46.208]) by svr1.postgresql.org (Postfix) with ESMTP id A26E0D1B4E0 for ; Mon, 22 Dec 2003 17:14:45 -0400 (AST) Received: by ganymede.hub.org (Postfix, from userid 1000) id 21E7638B92; Mon, 22 Dec 2003 17:11:25 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id 1835A386C7 for ; Mon, 22 Dec 2003 17:11:25 -0400 (AST) Date: Mon, 22 Dec 2003 17:11:25 -0400 (AST) From: "Marc G. Fournier" To: pgsql-performance@postgresql.org Subject: mnogosearch under 7.4 ... Message-ID: <20031222165147.S916@ganymede.hub.org> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/293 X-Sequence-Number: 5153 G'day all ... Dave asked me today about 'slow downs' on the search engines, so am looking at the various queries generated by enabling log_statement/log_duration, to get a feel for is something is "off" ... and the following seems a bit weird ... QueryA and QueryB are the same query, but against two different tables in the databases ... QueryA takes ~4x longer to run then QueryB, but both EXPLAINs look similar ... in fact, looking at the EXPLAIN ANALYZE output, I would expect that QueryB would be the slower of the two ... but, the actual vs estimated times for ndict5/ndict4 seem off (ndict4 is estimated high, ndict5 is estimated low) ... QueryA: 186_archives=# explain analyze SELECT ndict5.url_id,ndict5.intag FROM ndict5, url WHERE ndict5.word_id=1343124681 AND url.rec_id=ndict5.url_id AND ((url.url || '') LIKE 'http://archives.postgresql.org/%%'); QUERY PLAN -------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=0.00..69799.69 rows=44 width=8) (actual time=113.067..26477.672 rows=14112 loops=1) -> Index Scan using n5_word on ndict5 (cost=0.00..34321.89 rows=8708 width=8) (actual time=27.349..25031.666 rows=15501 loops=1) Index Cond: (word_id = 1343124681) -> Index Scan using url_rec_id on url (cost=0.00..4.06 rows=1 width=4) (actual time=0.061..0.068 rows=1 loops=15501) Index Cond: (url.rec_id = "outer".url_id) Filter: ((url || ''::text) ~~ 'http://archives.postgresql.org/%%'::text) Total runtime: 26550.566 ms (7 rows) QueryB: 186_archives=# explain analyze SELECT ndict4.url_id,ndict4.intag FROM ndict4, url WHERE ndict4.word_id=-2038735111 AND url.rec_id=ndict4.url_id AND ((url.url || '') LIKE 'http://archives.postgresql.org/%%'); QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------ Nested Loop (cost=0.00..99120.97 rows=62 width=8) (actual time=26.330..6630.581 rows=2694 loops=1) -> Index Scan using n4_word on ndict4 (cost=0.00..48829.52 rows=12344 width=8) (actual time=7.954..6373.098 rows=2900 loops=1) Index Cond: (word_id = -2038735111) -> Index Scan using url_rec_id on url (cost=0.00..4.06 rows=1 width=4) (actual time=0.059..0.066 rows=1 loops=2900) Index Cond: (url.rec_id = "outer".url_id) Filter: ((url || ''::text) ~~ 'http://archives.postgresql.org/%%'::text) Total runtime: 6643.462 ms (7 rows) ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664 From pgsql-performance-owner@postgresql.org Mon Dec 22 18:50:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B23BDD1B448 for ; Mon, 22 Dec 2003 22:50:39 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97133-07 for ; Mon, 22 Dec 2003 18:49:53 -0400 (AST) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 7065BD1B450 for ; Mon, 22 Dec 2003 18:49:50 -0400 (AST) Received: from css120.ihs.com (css120.ihs.com [170.207.105.120]) by mail1.ihs.com (8.12.10/8.12.10) with ESMTP id hBMMmZeD000851; Mon, 22 Dec 2003 15:48:35 -0700 (MST) Date: Mon, 22 Dec 2003 15:30:56 -0700 (MST) From: "scott.marlowe" To: Conny Thimren Cc: Subject: Re: general peformance question In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-IHS-MailScanner: Found to be clean X-IHS-MailScanner-SpamCheck: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/291 X-Sequence-Number: 5151 On Thu, 18 Dec 2003, Conny Thimren wrote: > Hi, > This a kind of newbie-question. I've been using Postgres for a long time in a low transction environment - and it is great. > > Now I've got an inquiry for using Postgresql in a heavy-load on-line system. This system must handle something like 20 questions per sec with a response time at 1/10 sec. Each question will result in approx 5-6 reads and a couple of updates. > Anybody have a feeling if this is realistic on a Intelbased Linux server with Postgresql. Ofcourse I know that this is too little info for an exact answer but - as I said - maybe someone can give a hint if it's possible. Maybe someone with heavy-load can give an example of what is possible... That really depends on how heavy each query is, so it's hard to say from what little you've given us. If you are doing simple banking style transactions, then you can easily handle this load, if you are talking a simple shopping cart, ditto, if, however, you are talking about queries that run 4 or 5 tables with millions of rows againts each other, you're gonna have to test it yourself. With the autovacuum daemon running, I ran a test overnight of pgbench (more for general purpose burn in than anything else) pgbench -i -s 100 pgbench -c 50 -t 250000 that's 10 million transactions, and it took just over twelve hours to complete at 220+ transactions per second. so, for financials, you're likely to find it easy to meet your target. But as the tables get bigger / more complex / more interconnected you'll see a drop in performance. From pgsql-performance-owner@postgresql.org Mon Dec 22 18:33:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2B723D1B446 for ; Mon, 22 Dec 2003 22:33:00 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96705-06 for ; Mon, 22 Dec 2003 18:32:12 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 5DBDCD1B450 for ; Mon, 22 Dec 2003 18:32:09 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBMMWB19018709; Mon, 22 Dec 2003 17:32:11 -0500 (EST) To: Michael Guerin Cc: pgsql-performance@postgresql.org Subject: Re: postgresql performance on linux port In-reply-to: <3FE741FA.4000002@rentec.com> References: <3FE741FA.4000002@rentec.com> Comments: In-reply-to Michael Guerin message dated "Mon, 22 Dec 2003 14:11:54 -0500" Date: Mon, 22 Dec 2003 17:32:11 -0500 Message-ID: <18708.1072132331@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/290 X-Sequence-Number: 5150 Michael Guerin writes: > I just restored a database running on a solaris box to a linux box > and queries take forever to execute. Did you remember to run ANALYZE? Have you applied the same configuration settings that you were using before? regards, tom lane From pgsql-performance-owner@postgresql.org Wed Dec 24 21:06:38 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 89BC2D1B480 for ; Tue, 23 Dec 2003 00:34:31 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08231-07 for ; Mon, 22 Dec 2003 20:33:44 -0400 (AST) Received: from wren.rentec.com (unknown [65.213.84.9]) by svr1.postgresql.org (Postfix) with ESMTP id F35FDD1B4A3 for ; Mon, 22 Dec 2003 20:33:40 -0400 (AST) Received: from rentec.com (guerinpc.rentec.com [172.16.161.26]) by wren.rentec.com (8.12.1/8.12.1) with ESMTP id hBN0XZ12013021; Mon, 22 Dec 2003 19:33:35 -0500 (EST) Message-ID: <3FE78D61.2F0AECAD@rentec.com> Date: Mon, 22 Dec 2003 19:33:37 -0500 From: Michael Guerin X-Mailer: Mozilla 4.79 [en] (Windows NT 5.0; U) X-Accept-Language: en MIME-Version: 1.0 To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: postgresql performance on linux port References: <3FE741FA.4000002@rentec.com> <18708.1072132331@sss.pgh.pa.us> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/294 X-Sequence-Number: 5154 Hi Tom, I don't believe I did run Analyze, I was under the assumption that the statistics would have been up to date when the indexes were created. Thanks for the quick response. -mike Tom Lane wrote: > Michael Guerin writes: > > I just restored a database running on a solaris box to a linux box > > and queries take forever to execute. > > Did you remember to run ANALYZE? Have you applied the same > configuration settings that you were using before? > > regards, tom lane From pgsql-performance-owner@postgresql.org Wed Dec 24 09:59:36 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D2246D1B466 for ; Wed, 24 Dec 2003 13:59:30 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 51717-02 for ; Wed, 24 Dec 2003 09:58:45 -0400 (AST) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 4D199D1B446 for ; Wed, 24 Dec 2003 09:58:41 -0400 (AST) Received: (qmail 19690 invoked from network); 24 Dec 2003 13:58:34 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 24 Dec 2003 13:58:34 -0000 Subject: Re: is it possible to get the optimizer to use indexes From: Dave Cramer Reply-To: pg@fastcrypt.com To: Doug McNaught Cc: pgsql-performance@postgresql.org In-Reply-To: <87d6alsi2i.fsf@asmodeus.mcnaught.org> References: <3FE1F5A1.6080908@geizhals.at> <60n09q9bqg.fsf@dev6.int.libertyrms.info> <1071789528.1629.10.camel@localhost.localdomain> <3FE256AA.8000707@familyhealth.com.au> <87d6alsi2i.fsf@asmodeus.mcnaught.org> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1072274314.1687.12.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 24 Dec 2003 08:58:34 -0500 Content-Transfer-Encoding: 7bit X-AntiVirus: scanned for viruses by AMaViS 0.2.1 (http://amavis.org/) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/292 X-Sequence-Number: 5152 Doug, Yes, it does depend on the locale, you can get around this in 7.4 by building the index with smart operators Dave On Thu, 2003-12-18 at 20:38, Doug McNaught wrote: > Christopher Kings-Lynne writes: > > >> It appears that the optimizer only uses indexes for = clause? > > > > The optimizer will used indexes for LIKE clauses, so long as the > > clause is a prefix search, eg: > > > > SELECT * FROM test WHERE a LIKE 'prf%'; > > Doesn't this still depend on your locale? > > -Doug > > > ---------------------------(end of broadcast)--------------------------- > TIP 3: 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 Wed Dec 24 21:06:56 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 375B3D1B94F for ; Wed, 24 Dec 2003 16:27:52 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66842-02 for ; Wed, 24 Dec 2003 12:27:03 -0400 (AST) Received: from HQEX1.sapiens.int (unknown [209.88.187.126]) by svr1.postgresql.org (Postfix) with ESMTP id 61FE0D1B481 for ; Wed, 24 Dec 2003 12:26:58 -0400 (AST) Received: from corigin.co.il ([192.168.168.3]) by HQEX1.sapiens.int with Microsoft SMTPSVC(5.0.2195.5329); Wed, 24 Dec 2003 18:26:51 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.0.6375.0 Content-Class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C3CA3A.93F9A512" Subject: Date: Wed, 24 Dec 2003 18:25:43 +0200 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: thread-index: AcPKOpP/Yr548916RgyY/vZTtpBlJA== From: "Michael Rothschild" To: X-OriginalArrivalTime: 24 Dec 2003 16:26:51.0370 (UTC) FILETIME=[BC5E0CA0:01C3CA3A] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/295 X-Sequence-Number: 5155 This is a multi-part message in MIME format. ------_=_NextPart_001_01C3CA3A.93F9A512 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable I consider using PostgreSQL for a project we have in our company and, to get a better picture of the product, I started scanning its source code and internal documentation. Based on what I saw (and maybe I didn't see enough) it seems that the optimizer will always decide to repeatedly scan the whole row set returned by sub selects in the context of an IN clause sequentially, as opposed to what I would expect it to do (which is to create some index or hash structure to improve performance). For example, if I have the following query: Select * from a where x in (select y from b where z=3D7)=20 Then I would expect an index or hash structure to be created for b.y when it is first scanned and brought into the cache but I couldn't see it happening in the source. As I said, I only inferred it from reading the source - not from actual experiments - so I may be wrong. 1. Am I wrong? 2. If I'm right, is there any plan to change it (after all, in the context of an IN clause, an index on the returned row set is all that is needed - the row set itself does not seem to matter). =20 Thank you, =20 Michael Rothschild ------_=_NextPart_001_01C3CA3A.93F9A512 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Message
I consider using PostgreSQL for a project we hav= e in=20 our company and, to get a better picture of the product, I s= tarted=20 scanning its source code and internal documentation.
Based on what I saw (and maybe I didn't see enou= gh)=20 it seems that the optimizer will always decide to repeatedly scan th= e=20 whole row set returned by sub selects in the context of an IN=20 clause sequentially, as opposed to what I would expect it to do (which= is=20 to create some index or hash structure to improve=20 performance).
For example, if I have the follow= ing=20 query:
Select * from a where x in (selec= t y from=20 b where z=3D7)
Then I would expect an index or hash structure to be = created=20 for b.y when it is first scanned and brought into the cache but I couldn't = see=20 it happening in the source.
As I said, I only inferred it from readin= g the=20 source - not from actual experiments - so I may be=20 wrong.
1. Am I= =20 wrong?
2. If I'm= right, is=20 there any plan to change it (after all, in the context of an IN clause, an = index=20 on the returned row set is all that is needed - the row set itself does not= seem=20 to matter).
 
Thank=20 you,
 
Michael=20 Rothschild
=00 ------_=_NextPart_001_01C3CA3A.93F9A512-- From pgsql-performance-owner@postgresql.org Wed Dec 24 22:48:33 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 30934D1B455 for ; Thu, 25 Dec 2003 02:48:31 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31885-01 for ; Wed, 24 Dec 2003 22:47:45 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 4F4CBD1B471 for ; Wed, 24 Dec 2003 22:47:40 -0400 (AST) Received: from houston.familyhealth.com.au (chriskl@localhost [127.0.0.1]) by houston.familyhealth.com.au (8.12.9p1/8.12.9) with ESMTP id hBP2lToD058752; Thu, 25 Dec 2003 10:47:29 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Received: from localhost (chriskl@localhost) by houston.familyhealth.com.au (8.12.9p1/8.12.9/Submit) with ESMTP id hBP2lTrK058749; Thu, 25 Dec 2003 10:47:29 +0800 (WST) X-Authentication-Warning: houston.familyhealth.com.au: chriskl owned process doing -bs Date: Thu, 25 Dec 2003 10:47:28 +0800 (WST) From: Christopher Kings-Lynne To: Michael Rothschild Cc: pgsql-performance@postgresql.org Subject: Re: In-Reply-To: Message-ID: <20031225104501.V58695-100000@houston.familyhealth.com.au> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/296 X-Sequence-Number: 5156 > For example, if I have the following query: > Select * from a where x in (select y from b where z=7) > Then I would expect an index or hash structure to be created for b.y > when it is first scanned and brought into the cache but I couldn't see > it happening in the source. > As I said, I only inferred it from reading the source - not from actual > experiments - so I may be wrong. > 1. Am I wrong? You are wrong - this is old behaviour and one of the major speed improvements of PostgreSQL 7.4 is that IN subqueries now use a hash index and hence they are much faster. Chris From pgsql-performance-owner@postgresql.org Thu Dec 25 12:29:53 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 88F91D1B450 for ; Thu, 25 Dec 2003 16:29:50 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96122-06 for ; Thu, 25 Dec 2003 12:29:01 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 4A210D1B455 for ; Thu, 25 Dec 2003 12:29:00 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBPGSs19020566; Thu, 25 Dec 2003 11:28:54 -0500 (EST) To: "Michael Rothschild" Cc: pgsql-performance@postgresql.org Subject: Re: In-reply-to: References: Comments: In-reply-to "Michael Rothschild" message dated "Wed, 24 Dec 2003 18:25:43 +0200" Date: Thu, 25 Dec 2003 11:28:54 -0500 Message-ID: <20565.1072369734@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/297 X-Sequence-Number: 5157 "Michael Rothschild" writes: > Based on what I saw (and maybe I didn't see enough) it seems that the > optimizer will always decide to repeatedly scan the whole row set > returned by sub selects in the context of an IN clause sequentially, What version were you looking at? regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 26 20:12:09 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E9133D1B479 for ; Sat, 27 Dec 2003 00:12:07 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 59501-07 for ; Fri, 26 Dec 2003 20:11:20 -0400 (AST) Received: from sccrmhc13.comcast.net (sccrmhc13.comcast.net [204.127.202.64]) by svr1.postgresql.org (Postfix) with ESMTP id 856E1D1B43C for ; Fri, 26 Dec 2003 20:11:17 -0400 (AST) Received: from juxtapose (c-24-1-81-113.client.comcast.net[24.1.81.113]) by comcast.net (sccrmhc13) with SMTP id <2003122700111901600ro8oie>; Sat, 27 Dec 2003 00:11:19 +0000 From: "Keith Bottner" To: Subject: What's faster? Date: Fri, 26 Dec 2003 18:11:19 -0600 Message-ID: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0084_01C3CBDB.A958C7B0" X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.4510 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.6 tagged_above=0.0 required=5.0 tests=HTML_30_40 X-Spam-Level: X-Archive-Number: 200312/298 X-Sequence-Number: 5158 This is a multi-part message in MIME format. ------=_NextPart_000_0084_01C3CBDB.A958C7B0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable I have a database where the vast majority of information that is related to a customer never changes. However, there is a single field (i.e. balance) that changes potentially tens to hundreds of times per day per customer (customers ranging in the 1000s to 10000s). This information is not indexed. Because Postgres requires VACUUM ANALYZE more frequently on updated tables, should I break this single field out into its own table, and if so what kind of a speed up can I expect to achieve. I would be appreciative of any guidance offered. =20 BTW, currently using Postgres 7.3.4 =20 =20 Keith =20 =20 ------=_NextPart_000_0084_01C3CBDB.A958C7B0 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Message
I have a = database=20 where the vast majority of information that is related to a customer never= =20 changes. However, there is a single field (i.e. balance) that changes= =20 potentially tens to hundreds of times per day per customer (customers rangi= ng in=20 the 1000s to 10000s). This information is not indexed. Because Postgres req= uires=20 VACUUM ANALYZE more frequently on updated tables, should I break this singl= e=20 field out into its own table, and if so what kind of a speed up can I expec= t to=20 achieve. I would be appreciative of any guidance offered.
 
BTW, curr= ently using=20 Postgres 7.3.4
 
 
Keith
 
 
------=_NextPart_000_0084_01C3CBDB.A958C7B0-- From pgsql-performance-owner@postgresql.org Fri Dec 26 20:49:55 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 58E82D1B432 for ; Sat, 27 Dec 2003 00:49:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 60019-09 for ; Fri, 26 Dec 2003 20:49:08 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id E5A76D1B449 for ; Fri, 26 Dec 2003 20:49:04 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBR0n719008603; Fri, 26 Dec 2003 19:49:07 -0500 (EST) To: "Keith Bottner" Cc: pgsql-performance@postgresql.org Subject: Re: What's faster? In-reply-to: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> References: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> Comments: In-reply-to "Keith Bottner" message dated "Fri, 26 Dec 2003 18:11:19 -0600" Date: Fri, 26 Dec 2003 19:49:07 -0500 Message-ID: <8602.1072486147@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/299 X-Sequence-Number: 5159 "Keith Bottner" writes: > I have a database where the vast majority of information that is related to > a customer never changes. However, there is a single field (i.e. balance) > that changes potentially tens to hundreds of times per day per customer > (customers ranging in the 1000s to 10000s). This information is not indexed. > Because Postgres requires VACUUM ANALYZE more frequently on updated tables, > should I break this single field out into its own table, Very likely a good idea, if the primary key that you'd need to add to identify the balance is narrow. Hard to say exactly how large the benefit would be, but I'd think the update costs would be reduced considerably. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Dec 26 21:07:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3C8ECD1B450 for ; Sat, 27 Dec 2003 01:07:11 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 65119-08 for ; Fri, 26 Dec 2003 21:06:25 -0400 (AST) Received: from gw.tssi.com (gw.tssi.com [198.147.197.1]) by svr1.postgresql.org (Postfix) with ESMTP id B77C3D1B432 for ; Fri, 26 Dec 2003 21:06:21 -0400 (AST) Received: from gw.tssi.com (nolan@gw.tssi.com [127.0.0.1] (may be forged)) by gw.tssi.com (8.12.6/8.12.6) with ESMTP id hBR16MjC002312; Fri, 26 Dec 2003 19:06:23 -0600 Received: (from nolan@localhost) by gw.tssi.com (8.12.6/8.12.6/Submit) id hBR16MLg002310; Fri, 26 Dec 2003 19:06:22 -0600 From: Mike Nolan Message-Id: <200312270106.hBR16MLg002310@gw.tssi.com> Subject: Re: What's faster? To: kbottner@comcast.net (Keith Bottner) Date: Fri, 26 Dec 2003 19:06:21 -0600 (CST) Cc: pgsql-performance@postgresql.org In-Reply-To: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> from "Keith Bottner" at Dec 26, 2003 06:11:19 PM X-Mailer: ELM [version 2.5 PL3] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/300 X-Sequence-Number: 5160 > Because Postgres requires VACUUM ANALYZE more frequently on updated tables, > should I break this single field out into its own table, and if so what kind > of a speed up can I expect to achieve. I would be appreciative of any > guidance offered. Unless that field is part of the key, I wouldn't think that a vacuum analyze would be needed, as the key distribution isn't changing. I don't know if that is still true if that field is indexed. Tom? Even then, as I understand things vacuum analyze doesn't rebuild indexes, so I could see a need to drop and rebuild indexes on a regular basis, even if you move that field into a separate table. -- Mike Nolan From pgsql-performance-owner@postgresql.org Sat Dec 27 00:01:00 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D7AEDD1B436 for ; Sat, 27 Dec 2003 04:00:58 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80619-03 for ; Sat, 27 Dec 2003 00:00:08 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 67D6CD1B44B for ; Sat, 27 Dec 2003 00:00:08 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBR40419009345; Fri, 26 Dec 2003 23:00:04 -0500 (EST) To: Mike Nolan Cc: kbottner@comcast.net (Keith Bottner), pgsql-performance@postgresql.org Subject: Re: What's faster? In-reply-to: <200312270106.hBR16MLg002310@gw.tssi.com> References: <200312270106.hBR16MLg002310@gw.tssi.com> Comments: In-reply-to Mike Nolan message dated "Fri, 26 Dec 2003 19:06:21 -0600" Date: Fri, 26 Dec 2003 23:00:03 -0500 Message-ID: <9344.1072497603@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/301 X-Sequence-Number: 5161 Mike Nolan writes: >> Because Postgres requires VACUUM ANALYZE more frequently on updated tables, >> should I break this single field out into its own table, and if so what kind >> of a speed up can I expect to achieve. I would be appreciative of any >> guidance offered. > Unless that field is part of the key, I wouldn't think that a vacuum > analyze would be needed, as the key distribution isn't changing. The "analyze" wouldn't matter ... but the "vacuum" would. He needs to get rid of the dead rows in a timely fashion. The wider the rows, the more disk space is at stake. Also, if he has more than just a primary index on the main table, the cost of updating the secondary indexes must be considered. A balance-only table would presumably have just one index to update. Against all this you have to weigh the cost of doing a join to get the balance, so it's certainly not a no-brainer choice. But I think it's surely worth considering such a design. regards, tom lane From pgsql-performance-owner@postgresql.org Sat Dec 27 06:52:56 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0A41FD1B4AC for ; Sat, 27 Dec 2003 10:52:54 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31612-02 for ; Sat, 27 Dec 2003 06:52:07 -0400 (AST) Received: from druid.net (druid.net [216.126.72.98]) by svr1.postgresql.org (Postfix) with ESMTP id 6A77ED1B47A for ; Sat, 27 Dec 2003 06:52:04 -0400 (AST) Received: by druid.net (Postfix, from userid 1000) id 8784F1B52; Sat, 27 Dec 2003 05:52:08 -0500 (EST) From: "D'Arcy J.M. Cain" To: "Keith Bottner" , Subject: Re: What's faster? Date: Sat, 27 Dec 2003 05:52:07 -0500 User-Agent: KMail/1.5.2 References: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> In-Reply-To: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200312270552.07823.darcy@druid.net> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/302 X-Sequence-Number: 5162 On December 26, 2003 07:11 pm, Keith Bottner wrote: > I have a database where the vast majority of information that is related to > a customer never changes. However, there is a single field (i.e. balance) > that changes potentially tens to hundreds of times per day per customer > (customers ranging in the 1000s to 10000s). This information is not > indexed. Because Postgres requires VACUUM ANALYZE more frequently on > updated tables, should I break this single field out into its own table, > and if so what kind of a speed up can I expect to achieve. I would be > appreciative of any guidance offered. We went through this recently. One thing we found that may apply to you is how many fields in the client record have a foreign key constraint. We find that tables with lots of FKeys are a lot more intensive on updates. In our case it was another table, think of it as an order or header table with a balance, that has over 10 million records. Sometimes we have 200,000 transactions a day where we have to check the balance. We eventually moved every field that could possibly be updated on a regular basis out to separate tables. The improvement was dramatic. -- D'Arcy J.M. Cain | Democracy is three wolves http://www.druid.net/darcy/ | and a sheep voting on +1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner. From pgsql-performance-owner@postgresql.org Sat Dec 27 14:31:33 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E3E4ED1B4B7 for ; Sat, 27 Dec 2003 18:31:31 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71990-02 for ; Sat, 27 Dec 2003 14:30:42 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 3AF4BD1B4E1 for ; Sat, 27 Dec 2003 14:30:41 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBRIUd19012772; Sat, 27 Dec 2003 13:30:39 -0500 (EST) To: "D'Arcy J.M. Cain" Cc: "Keith Bottner" , pgsql-performance@postgresql.org Subject: Re: What's faster? In-reply-to: <200312270552.07823.darcy@druid.net> References: <008301c3cc0d$f3f337b0$7d00a8c0@juxtapose> <200312270552.07823.darcy@druid.net> Comments: In-reply-to "D'Arcy J.M. Cain" message dated "Sat, 27 Dec 2003 05:52:07 -0500" Date: Sat, 27 Dec 2003 13:30:38 -0500 Message-ID: <12771.1072549838@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/303 X-Sequence-Number: 5163 "D'Arcy J.M. Cain" writes: > We went through this recently. One thing we found that may apply to you is > how many fields in the client record have a foreign key constraint. We find > that tables with lots of FKeys are a lot more intensive on updates. BTW, this should have gotten better in 7.3.4 and later --- there is logic to skip checking an FKey reference if the referencing columns didn't change during the update. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Dec 29 12:36:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 148C7D1B472 for ; Mon, 29 Dec 2003 16:36:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 38588-05 for ; Mon, 29 Dec 2003 12:35:57 -0400 (AST) Received: from bittern.mail.pas.earthlink.net (bittern.mail.pas.earthlink.net [207.217.120.119]) by svr1.postgresql.org (Postfix) with ESMTP id C82C6D1B443 for ; Mon, 29 Dec 2003 12:35:56 -0400 (AST) Received: from dpc6682160001.direcpc.com ([66.82.160.1] helo=earthlink.net) by bittern.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1Ab0NA-0006qe-00 for pgsql-performance@postgresql.org; Mon, 29 Dec 2003 08:36:07 -0800 Message-ID: <3FF057DF.7000605@earthlink.net> Date: Mon, 29 Dec 2003 11:35:43 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Question about difference in performance of 2 queries on large table Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/304 X-Sequence-Number: 5164 To all, The facts: PostgreSQL 7.4.0 running on BSD 5.1 on Dell 2650 with 4GB RAM, 5 SCSI drives in hardware RAID 0 configuration. Database size with indexes is currently 122GB. Schema for the table in question is at the end of this email. The DB has been vacuumed full and analyzed. Between 2 and 3 million records are added to the table in question each night. An analyze is run on the entire DB after the data has been loaded each night. There are no updates or deletes of records during the nightly load, only insertions. I am trying to understand why the performance between the two queries below is so different. I am trying to find the count of all pages that have a 'valid' content_key. -1 is our 'we don't have any content' key. The first plan below has horrendous performance. we only get about 2% CPU usage and iostat shows 3-5 MB/sec IO. The second plan runs at 30% cpu and 15-30MB.sec IO. Could someone shed some light on why the huge difference in performance? Both are doing index scans plus a filter. We have no content_keys below -1 at this time so the queries return the same results. Thanks. --sean explain select count (distinct (persistent_cookie_key) ) from f_pageviews where date_key between 305 and 334 and content_key > -1; QUERY PLAN ------------------------------------------------------------------------------------------------------------ Aggregate (cost=688770.29..688770.29 rows=1 width=4) -> Index Scan using idx_pageviews_content on f_pageviews (cost=0.00..645971.34 rows=17119580 width=4) Index Cond: (content_key > -1) Filter: ((date_key >= 305) AND (date_key <= 334)) (4 rows) explain select count (distinct (persistent_cookie_key) ) from f_pageviews where date_key between 305 and 334 and content_key <> -1; QUERY PLAN ------------------------------------------------------------------------------------------------------------------ Aggregate (cost=1365419.12..1365419.12 rows=1 width=4) -> Index Scan using idx_pageviews_date_nov_2003 on f_pageviews (cost=0.00..1322615.91 rows=17121284 width=4) Index Cond: ((date_key >= 305) AND (date_key <= 334)) Filter: (content_key <> -1) (4 rows) \d f_pageviews Table "public.f_pageviews" Column | Type | Modifiers ------------------------+---------+------------------------------------------------------------- id | integer | not null default nextval('public.f_pageviews_id_seq'::text) date_key | integer | not null time_key | integer | not null content_key | integer | not null location_key | integer | not null session_key | integer | not null subscriber_key | text | not null persistent_cookie_key | integer | not null ip_key | integer | not null referral_key | integer | not null servlet_key | integer | not null tracking_key | integer | not null provider_key | text | not null marketing_campaign_key | integer | not null orig_airport | text | not null dest_airport | text | not null commerce_page | boolean | not null default false job_control_number | integer | not null sequenceid | integer | not null default 0 url_key | integer | not null useragent_key | integer | not null web_server_name | text | not null default 'Not Available'::text cpc | integer | not null default 0 referring_servlet_key | integer | default 1 first_page_key | integer | default 1 newsletterid_key | text | not null default 'Not Available'::text Indexes: "f_pageviews_pkey" primary key, btree (id) "idx_pageviews_content" btree (content_key) "idx_pageviews_date_dec_2003" btree (date_key) WHERE ((date_key >= 335) AND (date_key <= 365)) "idx_pageviews_date_nov_2003" btree (date_key) WHERE ((date_key >= 304) AND (date_key <= 334)) "idx_pageviews_referring_servlet" btree (referring_servlet_key) "idx_pageviews_servlet" btree (servlet_key) "idx_pageviews_session" btree (session_key) From pgsql-performance-owner@postgresql.org Mon Dec 29 12:49:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1536CD1B443 for ; Mon, 29 Dec 2003 16:49:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 41877-02 for ; Mon, 29 Dec 2003 12:48:57 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 59E79D1B4A0 for ; Mon, 29 Dec 2003 12:48:10 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBTGm819015477; Mon, 29 Dec 2003 11:48:08 -0500 (EST) To: Sean Shanny Cc: pgsql-performance@postgresql.org Subject: Re: Question about difference in performance of 2 queries on large table In-reply-to: <3FF057DF.7000605@earthlink.net> References: <3FF057DF.7000605@earthlink.net> Comments: In-reply-to Sean Shanny message dated "Mon, 29 Dec 2003 11:35:43 -0500" Date: Mon, 29 Dec 2003 11:48:07 -0500 Message-ID: <15476.1072716487@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/305 X-Sequence-Number: 5165 Please show EXPLAIN ANALYZE output for your queries, not just EXPLAIN. Also it would be useful to see the pg_stats rows for the date_key and content_key columns. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Dec 29 12:50:10 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CBBBDD1B513 for ; Mon, 29 Dec 2003 16:50:08 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 38936-08 for ; Mon, 29 Dec 2003 12:49:20 -0400 (AST) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 7EBF4D1B436 for ; Mon, 29 Dec 2003 12:48:51 -0400 (AST) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 4126B8DD6; Mon, 29 Dec 2003 17:48:49 +0100 (CET) Date: Mon, 29 Dec 2003 17:48:49 +0100 (CET) From: Dennis Bjorklund To: Sean Shanny Cc: pgsql-performance@postgresql.org Subject: Re: Question about difference in performance of 2 queries In-Reply-To: <3FF057DF.7000605@earthlink.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/306 X-Sequence-Number: 5166 On Mon, 29 Dec 2003, Sean Shanny wrote: > The first plan below has horrendous performance. we only get about 2% > CPU usage and iostat shows 3-5 MB/sec IO. The second plan runs at 30% > cpu and 15-30MB.sec IO. > > Could someone shed some light on why the huge difference in > performance? Both are doing index scans plus a filter. We have no > content_keys below -1 at this time so the queries return the same results. EXPLAIN ANALYZE gives more information then EXPLAIN, and is prefered. It uses different indexes in the two queries, and one seems to be faster then the other. Why, I can't tell yet. I would assume that you would get the fastet result if you had an index (content_key, date_key) I don't know if pg will even use an index to speed up a <> operation. When you had > then it could use the idx_pageviews_content index. Why it choose that when the other would be faster I don't know. Maybe explain analyze will give some hint. -- /Dennis From pgsql-performance-owner@postgresql.org Mon Dec 29 13:45:15 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0C32DD1B440 for ; Mon, 29 Dec 2003 17:45:13 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 47911-09 for ; Mon, 29 Dec 2003 13:44:23 -0400 (AST) Received: from bittern.mail.pas.earthlink.net (bittern.mail.pas.earthlink.net [207.217.120.119]) by svr1.postgresql.org (Postfix) with ESMTP id 0AB97D1B4A0 for ; Mon, 29 Dec 2003 13:44:23 -0400 (AST) Received: from dpc6682160001.direcpc.com ([66.82.160.1] helo=earthlink.net) by bittern.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1Ab1RQ-0007EE-00; Mon, 29 Dec 2003 09:44:34 -0800 Message-ID: <3FF067EB.4050405@earthlink.net> Date: Mon, 29 Dec 2003 12:44:11 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Cc: Tom Lane Subject: Re: Question about difference in performance of 2 queries References: <3FF057DF.7000605@earthlink.net> <15476.1072716487@sss.pgh.pa.us> In-Reply-To: <15476.1072716487@sss.pgh.pa.us> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/307 X-Sequence-Number: 5167 I am running explain analyze now and will post results as they finish. Thanks. --sean Tom Lane wrote: >Please show EXPLAIN ANALYZE output for your queries, not just EXPLAIN. >Also it would be useful to see the pg_stats rows for the date_key and >content_key columns. > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Mon Dec 29 14:47:13 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B2DAFD1B436 for ; Mon, 29 Dec 2003 18:47:11 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 56288-05 for ; Mon, 29 Dec 2003 14:46:22 -0400 (AST) Received: from bittern.mail.pas.earthlink.net (bittern.mail.pas.earthlink.net [207.217.120.119]) by svr1.postgresql.org (Postfix) with ESMTP id 5CE99D1B437 for ; Mon, 29 Dec 2003 14:46:21 -0400 (AST) Received: from dpc6682160001.direcpc.com ([66.82.160.1] helo=earthlink.net) by bittern.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1Ab2PP-0007Sj-00; Mon, 29 Dec 2003 10:46:33 -0800 Message-ID: <3FF07671.7050706@earthlink.net> Date: Mon, 29 Dec 2003 13:46:09 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Cc: Tom Lane Subject: Re: Question about difference in performance of 2 queries References: <3FF057DF.7000605@earthlink.net> <15476.1072716487@sss.pgh.pa.us> In-Reply-To: <15476.1072716487@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 postgresql.org X-Archive-Number: 200312/308 X-Sequence-Number: 5168 Here is the pg_stats data. The explain analyze queries are still running. select * from pg_stats where tablename = 'f_pageviews' and attname = 'date_key'; schemaname | tablename | attname | null_frac | avg_width | n_distinct | most_common_vals | most_common_freqs | histogram_bounds | correlation ------------+-------------+----------+-----------+-----------+------------+-------------------------------------------+---------------------------------------------------------------------------------------------------+-----------------------------------------------+------------- public | f_pageviews | date_key | 0 | 4 | 60 | {335,307,309,336,308,321,314,342,322,316} | {0.0283333,0.0243333,0.0243333,0.0243333,0.024,0.0233333,0.0226667,0.0226667,0.0223333,0.0216667} | {304,311,318,325,329,334,341,346,351,356,363} | 0.345026 (1 row) select * from pg_stats where tablename = 'f_pageviews' and attname = 'content_key'; schemaname | tablename | attname | null_frac | avg_width | n_distinct | most_common_vals | most_common_freqs | histogram_bounds | correlation ------------+-------------+-------------+-----------+-----------+------------+------------------+-----------------------+-------------------------------------------------------------------------------------+------------- public | f_pageviews | content_key | 0 | 4 | 983 | {-1,1528483} | {0.749333,0.00166667} | {38966,323835,590676,717061,919148,1091875,1208244,1299702,1375366,1434079,1528910} | 0.103399 (1 row) Thanks. --sean Tom Lane wrote: >Please show EXPLAIN ANALYZE output for your queries, not just EXPLAIN. >Also it would be useful to see the pg_stats rows for the date_key and >content_key columns. > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Mon Dec 29 15:26:42 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 28712D1B498 for ; Mon, 29 Dec 2003 19:26:39 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 62922-09 for ; Mon, 29 Dec 2003 15:25:51 -0400 (AST) Received: from bittern.mail.pas.earthlink.net (bittern.mail.pas.earthlink.net [207.217.120.119]) by svr1.postgresql.org (Postfix) with ESMTP id CD654D1B450 for ; Mon, 29 Dec 2003 15:25:49 -0400 (AST) Received: from dpc6682160001.direcpc.com ([66.82.160.1] helo=earthlink.net) by bittern.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1Ab31b-0000in-00 for pgsql-performance@postgresql.org; Mon, 29 Dec 2003 11:26:02 -0800 Message-ID: <3FF07FB2.5060608@earthlink.net> Date: Mon, 29 Dec 2003 14:25:38 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Question about difference in performance of 2 queries References: <3FF057DF.7000605@earthlink.net> <15476.1072716487@sss.pgh.pa.us> In-Reply-To: <15476.1072716487@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 postgresql.org X-Archive-Number: 200312/309 X-Sequence-Number: 5169 Here is one of the explain analyzes. This is the from the faster query. Ignore the total runtime as we are currently doing other queries on this machine so it is slightly loaded. Thanks. --sean explain analyze select count (distinct (persistent_cookie_key) ) from f_pageviews where date_key between 305 and 334 and content_key <> -1; QUERY PLAN -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Aggregate (cost=1384925.95..1384925.95 rows=1 width=4) (actual time=4541462.030..4541462.034 rows=1 loops=1) -> Index Scan using idx_pageviews_date_nov_2003 on f_pageviews (cost=0.00..1343566.52 rows=16543772 width=4) (actual time=83.267..4286664.678 rows=15710722 loops=1) Index Cond: ((date_key >= 305) AND (date_key <= 334)) Filter: (content_key <> -1) Total runtime: 4541550.832 ms (5 rows) Tom Lane wrote: >Please show EXPLAIN ANALYZE output for your queries, not just EXPLAIN. >Also it would be useful to see the pg_stats rows for the date_key and >content_key columns. > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Mon Dec 29 15:40:08 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D5CEAD1B4AC for ; Mon, 29 Dec 2003 19:40:07 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68108-02 for ; Mon, 29 Dec 2003 15:39:20 -0400 (AST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id A10E0D1B49C for ; Mon, 29 Dec 2003 15:39:14 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.10/8.12.10) with ESMTP id hBTJdC19016346; Mon, 29 Dec 2003 14:39:12 -0500 (EST) To: Sean Shanny Cc: pgsql-performance@postgresql.org Subject: Re: Question about difference in performance of 2 queries In-reply-to: <3FF07671.7050706@earthlink.net> References: <3FF057DF.7000605@earthlink.net> <15476.1072716487@sss.pgh.pa.us> <3FF07671.7050706@earthlink.net> Comments: In-reply-to Sean Shanny message dated "Mon, 29 Dec 2003 13:46:09 -0500" Date: Mon, 29 Dec 2003 14:39:11 -0500 Message-ID: <16345.1072726751@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/310 X-Sequence-Number: 5170 Sean Shanny writes: > Here is the pg_stats data. The explain analyze queries are still running. > select * from pg_stats where tablename = 'f_pageviews' and attname = > 'content_key'; > schemaname | tablename | attname | null_frac | avg_width | > n_distinct | most_common_vals | most_common_freqs > | > histogram_bounds | correlation > ------------+-------------+-------------+-----------+-----------+------------+------------------+-----------------------+-------------------------------------------------------------------------------------+------------- > public | f_pageviews | content_key | 0 | 4 | > 983 | {-1,1528483} | {0.749333,0.00166667} | Oh-ho, I see the problem: about 75% of your table has content_key = -1. Why is that a problem, you ask? Well, the planner realizes that "content_key > -1" is a pretty good restriction condition (better than the date condition, apparently) and so it tries to use that as the index scan condition. The problem is that in 7.4 and before, the btree index code implements a "> -1" scan starting boundary by finding the first -1 and then advancing to the first key that's not -1. So you end up scanning through 75% of the index before anything useful happens :-( I just fixed this poor behavior in CVS tip a couple weeks ago: http://archives.postgresql.org/pgsql-committers/2003-12/msg00220.php but the patch seems too large and unproven to risk back-patching into 7.4.*. If you expect that a pretty large fraction of your data will always have dummy content_key, it'd probably be worth changing the index to not index -1's at all --- that is, make it a partial index with the condition "WHERE content_key > -1". Another workaround is to leave the index as-is but phrase the query WHERE condition as "content_key >= 0" instead of "> -1". regards, tom lane From pgsql-performance-owner@postgresql.org Mon Dec 29 15:53:41 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6308ED1B48E for ; Mon, 29 Dec 2003 19:53:40 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70004-01 for ; Mon, 29 Dec 2003 15:52:51 -0400 (AST) Received: from bittern.mail.pas.earthlink.net (bittern.mail.pas.earthlink.net [207.217.120.119]) by svr1.postgresql.org (Postfix) with ESMTP id 29209D1B436 for ; Mon, 29 Dec 2003 15:52:50 -0400 (AST) Received: from dpc6682160001.direcpc.com ([66.82.160.1] helo=earthlink.net) by bittern.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 1Ab3Ri-0006nC-00 for pgsql-performance@postgresql.org; Mon, 29 Dec 2003 11:53:01 -0800 Message-ID: <3FF08606.2080204@earthlink.net> Date: Mon, 29 Dec 2003 14:52:38 -0500 From: Sean Shanny User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Question about difference in performance of 2 queries References: <3FF057DF.7000605@earthlink.net> <15476.1072716487@sss.pgh.pa.us> <3FF07671.7050706@earthlink.net> <16345.1072726751@sss.pgh.pa.us> In-Reply-To: <16345.1072726751@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 postgresql.org X-Archive-Number: 200312/311 X-Sequence-Number: 5171 Tom, Thanks. I will make the changes you suggest concerning the indexes. I am finding partial indexes to be very handy. :-) I canceled the explain analyze on the other query as we have found the problem and who knows how long it would take to complete. Thanks again. --sean Tom Lane wrote: >Sean Shanny writes: > > >>Here is the pg_stats data. The explain analyze queries are still running. >> >> > > > >>select * from pg_stats where tablename = 'f_pageviews' and attname = >>'content_key'; >> schemaname | tablename | attname | null_frac | avg_width | >>n_distinct | most_common_vals | most_common_freqs >>| >>histogram_bounds | correlation >>------------+-------------+-------------+-----------+-----------+------------+------------------+-----------------------+-------------------------------------------------------------------------------------+------------- >> public | f_pageviews | content_key | 0 | 4 | >>983 | {-1,1528483} | {0.749333,0.00166667} | >> >> > >Oh-ho, I see the problem: about 75% of your table has content_key = -1. > >Why is that a problem, you ask? Well, the planner realizes that >"content_key > -1" is a pretty good restriction condition (better than >the date condition, apparently) and so it tries to use that as the index >scan condition. The problem is that in 7.4 and before, the btree index >code implements a "> -1" scan starting boundary by finding the first -1 >and then advancing to the first key that's not -1. So you end up >scanning through 75% of the index before anything useful happens :-( > >I just fixed this poor behavior in CVS tip a couple weeks ago: >http://archives.postgresql.org/pgsql-committers/2003-12/msg00220.php >but the patch seems too large and unproven to risk back-patching into >7.4.*. > >If you expect that a pretty large fraction of your data will always have >dummy content_key, it'd probably be worth changing the index to not >index -1's at all --- that is, make it a partial index with the >condition "WHERE content_key > -1". Another workaround is to leave the >index as-is but phrase the query WHERE condition as "content_key >= 0" >instead of "> -1". > > regards, tom lane > >---------------------------(end of broadcast)--------------------------- >TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > > > From pgsql-performance-owner@postgresql.org Mon Dec 29 16:34:47 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B67F8D1B436 for ; Mon, 29 Dec 2003 20:34:46 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75028-10 for ; Mon, 29 Dec 2003 16:33:59 -0400 (AST) Received: from lorax.kcilink.com (lorax.kciLink.com [206.112.95.1]) by svr1.postgresql.org (Postfix) with ESMTP id B42E4D1B43C for ; Mon, 29 Dec 2003 16:33:57 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id 39A9F3E42 for ; Mon, 29 Dec 2003 15:33:58 -0500 (EST) Received: from lorax.kcilink.com ([127.0.0.1]) by localhost (lorax.kcilink.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 02693-01-2 for ; Mon, 29 Dec 2003 15:33:57 -0500 (EST) Received: from lorax.kcilink.com (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id 9606F3E33 for ; Mon, 29 Dec 2003 15:33:57 -0500 (EST) Received: (from news@localhost) by lorax.kcilink.com (8.12.9p2/8.12.9/Submit) id hBTKXvGA024420 for pgsql-performance@postgresql.org; Mon, 29 Dec 2003 15:33:57 -0500 (EST) (envelope-from news) To: pgsql-performance@postgresql.org Path: not-for-mail From: Vivek Khera Newsgroups: ml.postgres.performance Subject: deferred foreign keys Date: Mon, 29 Dec 2003 15:33:57 -0500 Organization: Khera Communications, Inc., Rockville, MD Lines: 27 Message-ID: NNTP-Posting-Host: yertle.kcilink.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: lorax.kcilink.com 1072730037 59545 65.205.34.180 (29 Dec 2003 20:33:57 GMT) X-Complaints-To: daemon@kciLink.com NNTP-Posting-Date: Mon, 29 Dec 2003 20:33:57 +0000 (UTC) User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, berkeley-unix) Cancel-Lock: sha1:9TYKO5cEojn18uHttgg36sHTv1E= X-Virus-Scanned: by amavisd-new at kciLink.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/312 X-Sequence-Number: 5172 I'm observing that when I have many processes doing some work on my system that the transactions run along almost in lockstep. It appears from messages posted here that the foreign keys are acquiring and holding locks during the transactions, which seems like it would cause this behavior. I'd like to experiment with deferred foreign key checks so that the lock is only held during the commit when the checks are done. My questions are: 1) can I, and if so, how do I convert my existing FK's to deferrable without drop/create of the keys. Some of the keys take a long time to create and I'd like to avoid the hit. 2) do I increase the liklihood of deadlocks when the FK locks are being acquired or is it just as likeley as with the current non-deferred checking? I'm running 7.4 (soon to be 7.4.1) -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Vivek Khera, Ph.D. Khera Communications, Inc. Internet: khera@kciLink.com Rockville, MD +1-301-869-4449 x806 AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/ From pgsql-performance-owner@postgresql.org Tue Dec 30 11:03:15 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 76658D1B4A2 for ; Tue, 30 Dec 2003 15:03:13 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 09343-06 for ; Tue, 30 Dec 2003 11:02:28 -0400 (AST) Received: from mx.tripadvisor.com (unknown [151.203.96.35]) by svr1.postgresql.org (Postfix) with ESMTP id 34B2CD1B554 for ; Tue, 30 Dec 2003 11:02:17 -0400 (AST) Received: from [192.168.35.16] (shanny.tripadvisor.com [192.168.35.16]) by mx.tripadvisor.com (8.12.10/8.12.10) with ESMTP id hBUF2I9p088142; Tue, 30 Dec 2003 10:02:18 -0500 (EST) (envelope-from nshanny@tripadvisor.com) In-Reply-To: <16345.1072726751@sss.pgh.pa.us> References: <3FF057DF.7000605@earthlink.net> <15476.1072716487@sss.pgh.pa.us> <3FF07671.7050706@earthlink.net> <16345.1072726751@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v609) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit Cc: Sean Shanny , pgsql-performance@postgresql.org From: Nicholas Shanny Subject: Re: Question about difference in performance of 2 queries Date: Tue, 30 Dec 2003 10:00:30 -0500 To: Tom Lane X-Mailer: Apple Mail (2.609) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/313 X-Sequence-Number: 5173 Tom, I understand the problem and your solution makes sense although I am still puzzled by the machine under-utilization. If you run the original query and monitor the IO/CPU usage you find that it is minimal. here is the output from iostat 1 for a brief portion of the query. I am very curious to understand why when scanning the index the IO/CPU utilization is seemingly low. Cheers Nick Shanny TripAdvisor, Inc. 0 77 32.00 106 3.31 0.00 0 0.00 0.00 0 0.00 0 0 2 0 98 0 76 32.00 125 3.90 0.00 0 0.00 0.00 0 0.00 0 0 2 0 97 0 76 32.00 125 3.90 0.00 0 0.00 0.00 0 0.00 0 0 1 1 98 0 76 32.75 127 4.05 0.00 0 0.00 0.00 0 0.00 0 0 1 0 99 tty aacd0 acd0 fd0 cpu tin tout KB/t tps MB/s KB/t tps MB/s KB/t tps MB/s us ni sy in id 0 76 32.00 127 3.96 0.00 0 0.00 0.00 0 0.00 0 0 3 0 97 0 229 32.24 135 4.24 0.00 0 0.00 0.00 0 0.00 0 0 4 0 95 0 76 32.00 129 4.02 0.00 0 0.00 0.00 0 0.00 0 0 2 0 97 0 76 32.00 123 3.84 0.00 0 0.00 0.00 0 0.00 0 0 2 0 98 0 76 31.72 115 3.56 0.00 0 0.00 0.00 0 0.00 0 0 2 0 98 0 76 32.50 126 3.99 0.00 0 0.00 0.00 0 0.00 0 0 3 1 96 0 76 32.00 123 3.84 0.00 0 0.00 0.00 0 0.00 0 0 3 0 97 0 76 32.00 122 3.81 0.00 0 0.00 0.00 0 0.00 1 0 2 0 97 0 76 32.00 135 4.21 0.00 0 0.00 0.00 0 0.00 0 0 2 1 97 0 76 32.00 97 3.03 0.00 0 0.00 0.00 0 0.00 0 0 3 0 97 On Dec 29, 2003, at 2:39 PM, Tom Lane wrote: > Sean Shanny writes: >> Here is the pg_stats data. The explain analyze queries are still >> running. > >> select * from pg_stats where tablename = 'f_pageviews' and attname = >> 'content_key'; >> schemaname | tablename | attname | null_frac | avg_width | >> n_distinct | most_common_vals | most_common_freqs >> | >> histogram_bounds | correlation >> ------------+-------------+-------------+-----------+----------- >> +------------+------------------+----------------------- >> +--------------------------------------------------------------------- >> ----------------+------------- >> public | f_pageviews | content_key | 0 | 4 | >> 983 | {-1,1528483} | {0.749333,0.00166667} | > > Oh-ho, I see the problem: about 75% of your table has content_key = -1. > > Why is that a problem, you ask? Well, the planner realizes that > "content_key > -1" is a pretty good restriction condition (better than > the date condition, apparently) and so it tries to use that as the > index > scan condition. The problem is that in 7.4 and before, the btree index > code implements a "> -1" scan starting boundary by finding the first -1 > and then advancing to the first key that's not -1. So you end up > scanning through 75% of the index before anything useful happens :-( > > I just fixed this poor behavior in CVS tip a couple weeks ago: > http://archives.postgresql.org/pgsql-committers/2003-12/msg00220.php > but the patch seems too large and unproven to risk back-patching into > 7.4.*. > > If you expect that a pretty large fraction of your data will always > have > dummy content_key, it'd probably be worth changing the index to not > index -1's at all --- that is, make it a partial index with the > condition "WHERE content_key > -1". Another workaround is to leave the > index as-is but phrase the query WHERE condition as "content_key >= 0" > instead of "> -1". > > regards, tom lane > > ---------------------------(end of > broadcast)--------------------------- > TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > From pgsql-performance-owner@postgresql.org Tue Dec 30 12:33:40 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5E617D1B8C8 for ; Tue, 30 Dec 2003 16:33:38 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21075-05 for ; Tue, 30 Dec 2003 12:32:48 -0400 (AST) Received: from mail4.speakeasy.net (mail4.speakeasy.net [216.254.0.204]) by svr1.postgresql.org (Postfix) with ESMTP id 7AA26D1B468 for ; Tue, 30 Dec 2003 12:32:47 -0400 (AST) Received: (qmail 25116 invoked from network); 30 Dec 2003 16:32:46 -0000 Received: from atlas.jonathangardner.net (HELO jonathangardner.net) (jgardn@[66.92.192.166]) (envelope-sender ) by mail4.speakeasy.net (qmail-ldap-1.03) with RC4-MD5 encrypted SMTP for ; 30 Dec 2003 16:32:46 -0000 From: Jonathan Gardner To: pgsql-performance@postgresql.org Subject: DELETE ... WHERE ctid IN (...) vs. Iteration Date: Tue, 30 Dec 2003 08:32:38 -0800 User-Agent: KMail/1.5.94 MIME-Version: 1.0 Content-Disposition: inline Content-Type: Text/Plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Message-Id: <200312300832.44945.jgardner@jonathangardner.net> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/314 X-Sequence-Number: 5174 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I guess this may have come up before, but now that 7.4 has the IN with=20 improved performance, it may be time to revisit this topic. Compare these two algorithms (in plpgsql): (a) DELETE FROM foo WHERE ctid IN ( SELECT foo.ctid FROM ... WHERE ... ); (b) FOR result IN SELECT foo.ctid FROM ... WHERE ... LOOP DELETE FROM foo WHERE ctid =3D result; END LOOP; My poor understanding of how the IN operator works leaves me to believe=20 that for a large set of data in the IN group, a hash is used and a=20 tablescan done on foo. However, for a small set of data in the IN group,= =20 no tablescan is performed. I assume that (a) works at O(ln(N)) for large N, and O(N) for small N,=20 while (b) works at O(N) universally. Therefore, (a) is the superior=20 algorithm. I welcome criticism and correction. - --=20 Jonathan Gardner jgardner@jonathangardner.net Live Free, Use Linux! -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQE/8aipWgwF3QvpWNwRAk8GAJoDWISjxG7LMB1FdCFmwlOafsmZTwCePx18 lyHLNBJ8nP0RHzv6WfRzQ+M=3D =3DFPdW -----END PGP SIGNATURE----- From pgsql-performance-owner@postgresql.org Wed Dec 31 13:18:45 2003 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 73686D1B440 for ; Wed, 31 Dec 2003 17:18:43 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11455-03 for ; Wed, 31 Dec 2003 13:17:54 -0400 (AST) Received: from lorax.kcilink.com (lorax.kciLink.com [206.112.95.1]) by svr1.postgresql.org (Postfix) with ESMTP id 5EF6CD1C981 for ; Wed, 31 Dec 2003 13:17:53 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id 9F0FB3E65 for ; Wed, 31 Dec 2003 12:17:51 -0500 (EST) Received: from lorax.kcilink.com ([127.0.0.1]) by localhost (lorax.kcilink.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 84374-05-4 for ; Wed, 31 Dec 2003 12:17:51 -0500 (EST) Received: from lorax.kcilink.com (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id E37CA3E4D for ; Wed, 31 Dec 2003 12:17:50 -0500 (EST) Received: (from news@localhost) by lorax.kcilink.com (8.12.9p2/8.12.9/Submit) id hBVHHogZ033782 for pgsql-performance@postgresql.org; Wed, 31 Dec 2003 12:17:50 -0500 (EST) (envelope-from news) To: pgsql-performance@postgresql.org Path: not-for-mail From: Vivek Khera Newsgroups: ml.postgres.performance Subject: Re: deferred foreign keys Date: Wed, 31 Dec 2003 12:17:50 -0500 Organization: Khera Communications, Inc., Rockville, MD Lines: 16 Message-ID: References: NNTP-Posting-Host: yertle.kcilink.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: lorax.kcilink.com 1072891070 90029 65.205.34.180 (31 Dec 2003 17:17:50 GMT) X-Complaints-To: daemon@kciLink.com NNTP-Posting-Date: Wed, 31 Dec 2003 17:17:50 +0000 (UTC) User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, berkeley-unix) Cancel-Lock: sha1:0SjR8PMhIyH041mh3DSKaJ7YFMA= X-Virus-Scanned: by amavisd-new at kciLink.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200312/315 X-Sequence-Number: 5175 One more question: does the FK checker know to skip checking a constraint if the column in question did not change during an update? That is, if I have a user table that references an owner_id in an owners table as a foreign key, but I update fields other than owner_id in the user table, will it still try to verify that owner_id is a valid value even though it did not change? I'm using PG 7.4. Thanks. -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Vivek Khera, Ph.D. Khera Communications, Inc. Internet: khera@kciLink.com Rockville, MD +1-301-869-4449 x806 AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/