From pgsql-performance-owner@postgresql.org Wed Mar 31 21:14:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 25BEAD1E8CE for ; Thu, 1 Apr 2004 01:14:46 +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 80332-01 for ; Wed, 31 Mar 2004 21:14:45 -0400 (AST) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 15D8BD1E8AD for ; Wed, 31 Mar 2004 21:14:42 -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 i311EYWL071036; Thu, 1 Apr 2004 09:14:34 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <406B6E0B.9050200@familyhealth.com.au> Date: Thu, 01 Apr 2004 09:19:07 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Priem, Alexander" Cc: pgsql-performance@postgresql.org Subject: Re: What index for 'like (%keyword%)' ??? References: <2A07EC2D0BC2774AAD6F74769F60D52A08329A@ahmose.cict_ad.nl> In-Reply-To: <2A07EC2D0BC2774AAD6F74769F60D52A08329A@ahmose.cict_ad.nl> 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: 200403/444 X-Sequence-Number: 6300 > I am building a query which uses a clause like "Where doc_description like > '%keyword%'". I know a normal index won't be of any use here, but since the > table in question will get fairly big, I do want to use an index. > > Can anyone give me some advise on what kind of index I can use here? Or > shouldn't I use one in this case? You have to use a proper full text indexing scheme. Investigate contrib/tsearch2 module in the postgres distribution. Chirs From pgsql-performance-owner@postgresql.org Thu Apr 1 15:21:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E2261D1EB0F for ; Thu, 1 Apr 2004 19:16:48 +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 52634-08 for ; Thu, 1 Apr 2004 15:16:31 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 1D212D1EA3B for ; Thu, 1 Apr 2004 15:16:17 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001CFC; 1 Apr 2004 20:20:15 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 1 Apr 2004 20:19:44 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001CFB; 1 Apr 2004 20:19:35 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Thu, 01 Apr 2004 20:19:34 +0100 MIME-Version: 1.0 Subject: PostgreSQL and Linux 2.6 kernel. Message-ID: <406C7956.17885.4B2DCD2@localhost> In-reply-to: <5090.1080747621@sss.pgh.pa.us> References: X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/1 X-Sequence-Number: 6301 As part of my ongoing evaluation of PostgreSQL I have been doing a little stress testing. I though I would share an interesting result here.. Machine spec: 500 MHz PIII 256MB RAM "old-ish" IDE HD (5400RPM) Linux 2.4.22 kernel (Madrake 9.2) I have PostgreSQL 7.4.1 installed and have managed to load up a 1.4 GB database from MS SQLServer. Vaccum analyzed it. As a test in PosgreSQL I issued a statement to update a single column of a table containing 2.8 million rows with the values of a column in a table with similar rowcount. Using the above spec I had to stop the server after 17 hours. The poor thing was thrashing the hard disk and doing more swapping than useful work. Having obtained a copy of Mandrake 10.0 with the 2.6 kernal I though I would give it a go. Same hardware. Same setup. Same database loaded up. Same postgresql.conf file to make sure all the settings were the same. Vaccum analyzed it. same update statement COMPLETED in 2 hours 50 minutes. I'm impressed. I could see from vmstat that the system was achieving much greater IO thoughput than the 2.4 kernel. Although the system was still swapping there seems to be a completely different memory management pattern that suits PostgreSQL very well. Just to see that this wasn't a coincidence I am repeating the test. It is now into the 14th hour using the old 2.4 kernel. I'm going to give up..... Has anyone else done any comparative testing with the 2.6 kernel? Cheers, Gary. From pgsql-performance-owner@postgresql.org Mon Apr 5 15:59:37 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 40187D1E928 for ; Thu, 1 Apr 2004 20:19:31 +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 80056-10 for ; Thu, 1 Apr 2004 16:19:25 -0400 (AST) Received: from mail2.fbab.net (spectre.fbab.net [212.214.165.139]) by svr1.postgresql.org (Postfix) with SMTP id 69A9FD1E8FB for ; Thu, 1 Apr 2004 16:19:22 -0400 (AST) Received: (qmail 17709 invoked by uid 136); 1 Apr 2004 20:19:21 -0000 Received: from mag@fbab.net by mail2.fbab.net by uid 133 with qmail-scanner-1.20rc1 (avp: 4.0.3.0. Clear:RC:0:. Processed in 0.053725 secs); 01 Apr 2004 20:19:21 -0000 Received: from unknown (HELO fbab.net) (magpool1@212.214.165.129) by mail2.fbab.net with SMTP; 1 Apr 2004 20:19:20 -0000 Message-ID: <406C7942.40900@fbab.net> Date: Thu, 01 Apr 2004 22:19:14 +0200 From: "Magnus Naeslund(t)" User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Gary Doades Cc: pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. References: <406C7956.17885.4B2DCD2@localhost> In-Reply-To: <406C7956.17885.4B2DCD2@localhost> 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: 200404/46 X-Sequence-Number: 6346 Gary Doades wrote: > > Has anyone else done any comparative testing with the 2.6 kernel? > I know for a fact that certain stuff is recognized differently between 2.2, 2.4 and 2.6 kernels. For example i have one box that i installed debian stable on that used a 2.2 kernel which automatically tuned on DMA on the harddrive, didn't do it on a 2.4 kernel, but on 2.6 one it saw it as DMA able. Such things can dramatically affect performance, so make sure to compare what capabilities the kernel thinks your hardware has between the kernels first... But i'll grant that the 2.6 kernel is a great deal faster on some of our test servers. Regards Magnus From pgsql-performance-owner@postgresql.org Thu Apr 1 19:00:51 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6C79BD1E159 for ; Thu, 1 Apr 2004 23:00:48 +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 45099-06 for ; Thu, 1 Apr 2004 19:00:48 -0400 (AST) Received: from palle.girgensohn.se (1-2-8-5a.asp.sth.bostream.se [82.182.157.66]) by svr1.postgresql.org (Postfix) with ESMTP id 1C1A2D1E0F6 for ; Thu, 1 Apr 2004 19:00:45 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by palle.girgensohn.se (8.12.11/8.12.11) with ESMTP id i31N0jsA036201; Fri, 2 Apr 2004 01:00:45 +0200 (CEST) (envelope-from girgen@pingpong.net) Date: Fri, 02 Apr 2004 01:00:45 +0200 From: Palle Girgensohn To: pgsql-performance@postgresql.org Subject: single index on more than two coulumns a bad thing? Message-ID: <24070000.1080860445@palle.girgensohn.se> X-Mailer: Mulberry/3.1.0 (Linux/x86) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/2 X-Sequence-Number: 6302 Hi, I remember reading a post ages ago, maybe from Vadim, about the fact that people creating indices on more than two columns will be the first to be put againts the wall when the revolution comes... sort of... Is it always bad to create index xx on yy (field1, field2, field3); I guess the problem is that the index might often grow bigger than the table, or at least big enough not to speed up the queries? /Palle From pgsql-performance-owner@postgresql.org Thu Apr 1 20:36:23 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1A26FD1D44F for ; Fri, 2 Apr 2004 00:36:17 +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 71483-09 for ; Thu, 1 Apr 2004 20:36:17 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 2369FD1E9D5 for ; Thu, 1 Apr 2004 20:36:14 -0400 (AST) Received: from [63.195.55.98] (HELO 192.168.1.29) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4750237; Thu, 01 Apr 2004 16:37:35 -0800 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Palle Girgensohn , pgsql-performance@postgresql.org Subject: Re: single index on more than two coulumns a bad thing? Date: Thu, 1 Apr 2004 16:35:45 -0800 User-Agent: KMail/1.5.4 References: <24070000.1080860445@palle.girgensohn.se> In-Reply-To: <24070000.1080860445@palle.girgensohn.se> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404011635.45440.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/3 X-Sequence-Number: 6303 Palle, > Is it always bad to create index xx on yy (field1, field2, field3); No, it seldom bad, in fact. I have some indexes that run up to seven columns, becuase they are required for unique keys. Indexes of 3-4 columns are often *required* for many-to-many join tables. I'm afraid that you've been given some misleading advice. > I guess the problem is that the index might often grow bigger than the > table, or at least big enough not to speed up the queries? Well, yes ... a 4-column index on a 5-column table could be bigger than the table if allowed to bloat and not re-indexed. But that's just a reason for better maintainence. -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Apr 2 00:38:15 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E6459D1EA71 for ; Fri, 2 Apr 2004 04:36:59 +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 42904-01 for ; Fri, 2 Apr 2004 00:36:48 -0400 (AST) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6835BD1D5A3 for ; Fri, 2 Apr 2004 00:36:47 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i324ac56018755; Thu, 1 Apr 2004 23:36:38 -0500 (EST) To: josh@agliodbs.com Cc: Palle Girgensohn , pgsql-performance@postgresql.org Subject: Re: single index on more than two coulumns a bad thing? In-reply-to: <200404011635.45440.josh@agliodbs.com> References: <24070000.1080860445@palle.girgensohn.se> <200404011635.45440.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Thu, 01 Apr 2004 16:35:45 -0800" Date: Thu, 01 Apr 2004 23:36:38 -0500 Message-ID: <18754.1080880598@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/4 X-Sequence-Number: 6304 Josh Berkus writes: >> Is it always bad to create index xx on yy (field1, field2, field3); > I'm afraid that you've been given some misleading advice. I'd say it's a matter of getting your optimizations straight. If you have a query that can make use of that index, and the query is executed often enough to make it worth maintaining the index during table updates, then by all means make the index. The standard advice is meant to warn you against creating a zillion indexes without any thought to what you'll be paying in update costs. Indexes with more than a couple of columns are usually of only narrow applicability, and so you have to be sure that they'll really pay for themselves... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 2 00:44:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6A6DDD1D5A3 for ; Fri, 2 Apr 2004 04:44:32 +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 41032-07 for ; Fri, 2 Apr 2004 00:44:31 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 9D003D1D2B5 for ; Fri, 2 Apr 2004 00:44:30 -0400 (AST) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 094E617C4A8; Thu, 1 Apr 2004 23:44:30 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1B9GXd-0005Qq-00; Thu, 01 Apr 2004 23:44:29 -0500 To: "Damien Dougan" Cc: Subject: Re: Index Performance Help References: <004b01c3ebe1$7e0c9a70$7701a8c0@pestilence> In-Reply-To: <004b01c3ebe1$7e0c9a70$7701a8c0@pestilence> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 01 Apr 2004 23:44:29 -0500 Message-ID: <87d66rouoi.fsf@stark.xeocode.com> Lines: 25 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: 200404/5 X-Sequence-Number: 6305 "Damien Dougan" writes: > Sample analyze output for an initial query: > > hydradb=# explain analyze select * from pvsubscriber where actorid = > 'b3432-asdas-232-Subscriber793500'; I take it pvsubscriber is a view? What's the definition of your view? > -> Index Scan using mc_actor_key on mc_actor > (cost=0.00..4.08 rows=1 width=69) > (actual time=39.497..39.499 rows=1 loops=1) Is this table regularly vacuumed? Is it possible it has lots of dead records with this value for actorid? Try running vacuum full, or better "vacuum full verbose" and keep the output, it might explain. What version of postgres is this? You might try reindexing all your indexes (but particularly this one). Older versions of postgres were prone to index bloat problems. -- greg From pgsql-performance-owner@postgresql.org Fri Apr 2 00:59:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7FBC1D1D606 for ; Fri, 2 Apr 2004 04:59:49 +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 49526-01 for ; Fri, 2 Apr 2004 00:59:48 -0400 (AST) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 77B85D1D2B5 for ; Fri, 2 Apr 2004 00:59:47 -0400 (AST) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id EB42217C464; Thu, 1 Apr 2004 23:59:46 -0500 (EST) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1B9GmQ-0005Ss-00; Thu, 01 Apr 2004 23:59:46 -0500 To: Palle Girgensohn Cc: pgsql-performance@postgresql.org Subject: Re: single index on more than two coulumns a bad thing? References: <24070000.1080860445@palle.girgensohn.se> In-Reply-To: <24070000.1080860445@palle.girgensohn.se> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 01 Apr 2004 23:59:46 -0500 Message-ID: <871xn7otz1.fsf@stark.xeocode.com> 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: 200404/6 X-Sequence-Number: 6306 Palle Girgensohn writes: > Is it always bad to create index xx on yy (field1, field2, field3); All generalisations are false... Seriously, it's true that as the length of your index key gets longer the harder and harder it is to justify it. That doesn't mean they're always wrong, but you should consider whether a shorter key would perform just as well. The other problem with long index keys is that they often show up in the same place as having dozens of indexes on the same table. Usually in shops where the indexes were created after the fact looking at specific queries. -- greg From pgsql-performance-owner@postgresql.org Fri Apr 2 02:32:23 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E4B59D1EC4E for ; Fri, 2 Apr 2004 06:32:18 +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 71138-05 for ; Fri, 2 Apr 2004 02:32:17 -0400 (AST) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id AE36AD1EC43 for ; Fri, 2 Apr 2004 02:32:13 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i326WBJi023110; Fri, 2 Apr 2004 01:32:11 -0500 (EST) To: "Gary Doades" Cc: pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. In-reply-to: <406C7956.17885.4B2DCD2@localhost> References: <406C7956.17885.4B2DCD2@localhost> Comments: In-reply-to "Gary Doades" message dated "Thu, 01 Apr 2004 20:19:34 +0100" Date: Fri, 02 Apr 2004 01:32:10 -0500 Message-ID: <23109.1080887530@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/7 X-Sequence-Number: 6307 "Gary Doades" writes: > As a test in PosgreSQL I issued a statement to update a single column > of a table containing 2.8 million rows with the values of a column in > a table with similar rowcount. Using the above spec I had to stop the > server after 17 hours. The poor thing was thrashing the hard disk and > doing more swapping than useful work. This statement is pretty much content-free, since you did not show us the table schemas, the query, or the EXPLAIN output for the query. (I'll forgive you the lack of EXPLAIN ANALYZE, but you could easily have provided all the other hard facts.) There's really no way to tell where the bottleneck is. Maybe it's a kernel-level issue, but I would not bet on that without more evidence. I'd definitely not bet on it without direct confirmation that the same query plan was used in both setups. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 2 03:04:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3C8B8D1EA9D for ; Fri, 2 Apr 2004 07:04:14 +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 77782-07 for ; Fri, 2 Apr 2004 03:04:13 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 623A6D1EA89 for ; Fri, 2 Apr 2004 03:04:11 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001D53; 2 Apr 2004 08:07:59 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 2 Apr 2004 08:07:44 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001D52; 2 Apr 2004 08:07:39 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Fri, 02 Apr 2004 08:07:38 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <406D1F4A.27973.73B2299@localhost> In-reply-to: <23109.1080887530@sss.pgh.pa.us> References: <406C7956.17885.4B2DCD2@localhost> X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/8 X-Sequence-Number: 6308 The post was not intended to be content-rich, just my initial feedback after only just switching to 2.6. Since I had largely given up on this particular line of attack using 2.4 I didn't think to do a detailed analysis at this time. I was also hoping that others would add to the discussion. As this could become important I will be doing more analysis, but due to the nature of the issue and trying to keep as many factors constant as possible, this may take some time. Cheers, Gary. On 2 Apr 2004 at 1:32, Tom Lane wrote: > "Gary Doades" writes: > > As a test in PosgreSQL I issued a statement to update a single column > > of a table containing 2.8 million rows with the values of a column in > > a table with similar rowcount. Using the above spec I had to stop the > > server after 17 hours. The poor thing was thrashing the hard disk and > > doing more swapping than useful work. > > This statement is pretty much content-free, since you did not show us > the table schemas, the query, or the EXPLAIN output for the query. > (I'll forgive you the lack of EXPLAIN ANALYZE, but you could easily > have provided all the other hard facts.) There's really no way to tell > where the bottleneck is. Maybe it's a kernel-level issue, but I would > not bet on that without more evidence. I'd definitely not bet on it > without direct confirmation that the same query plan was used in both > setups. > > regards, tom lane > > > -- > Incoming mail is certified Virus Free. > Checked by AVG Anti-Virus (http://www.grisoft.com). > Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 > From pgsql-performance-owner@postgresql.org Fri Apr 2 07:56:31 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9FC6DD1E930 for ; Fri, 2 Apr 2004 11:56:29 +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 88171-02 for ; Fri, 2 Apr 2004 07:56:30 -0400 (AST) Received: from hotmail.com (law10-oe58.law10.hotmail.com [64.4.14.193]) by svr1.postgresql.org (Postfix) with ESMTP id 7406ED1E8C6 for ; Fri, 2 Apr 2004 07:56:27 -0400 (AST) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Fri, 2 Apr 2004 03:56:30 -0800 Received: from 67.81.102.201 by law10-oe58.law10.hotmail.com with DAV; Fri, 02 Apr 2004 11:56:30 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman2@hotmail.com] X-Sender: awerman2@hotmail.com From: "Aaron Werman" To: , "Palle Girgensohn" , References: <24070000.1080860445@palle.girgensohn.se> <200404011635.45440.josh@agliodbs.com> Subject: Re: single index on more than two coulumns a bad thing? Date: Fri, 2 Apr 2004 06:56:42 -0500 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 6.00.2800.1158 X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Message-ID: X-OriginalArrivalTime: 02 Apr 2004 11:56:30.0273 (UTC) FILETIME=[89261310:01C418A9] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/9 X-Sequence-Number: 6309 another thing that I have all over the place is a hierarchy: index on grandfather_table(grandfather) index on father_table(grandfather, father) index on son_table(grandfather, father, son) almost all of my indices are composite. Are you thinking about composite indices with low cardinality leading columns? /Aaron ----- Original Message ----- From: "Josh Berkus" To: "Palle Girgensohn" ; Sent: Thursday, April 01, 2004 7:35 PM Subject: Re: [PERFORM] single index on more than two coulumns a bad thing? > Palle, > > > Is it always bad to create index xx on yy (field1, field2, field3); > > No, it seldom bad, in fact. I have some indexes that run up to seven > columns, becuase they are required for unique keys. > > Indexes of 3-4 columns are often *required* for many-to-many join tables. > > I'm afraid that you've been given some misleading advice. > > > I guess the problem is that the index might often grow bigger than the > > table, or at least big enough not to speed up the queries? > > Well, yes ... a 4-column index on a 5-column table could be bigger than the > table if allowed to bloat and not re-indexed. But that's just a reason for > better maintainence. > > -- > -Josh Berkus > Aglio Database Solutions > San Francisco > > > ---------------------------(end of broadcast)--------------------------- > TIP 7: don't forget to increase your free space map settings > From pgsql-performance-owner@postgresql.org Fri Apr 2 10:56:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 330AAD1ECD5 for ; Fri, 2 Apr 2004 14:55:28 +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 43598-09 for ; Fri, 2 Apr 2004 10:55:30 -0400 (AST) Received: from artemis.le.ac.uk (artemis.le.ac.uk [143.210.16.126]) by svr1.postgresql.org (Postfix) with ESMTP id 436A7D1EC47 for ; Fri, 2 Apr 2004 10:55:26 -0400 (AST) Received: from [143.210.36.58] (helo=mail.star.le.ac.uk) by artemis.le.ac.uk with smtp (Exim 4.30) id 1B9Q4w-0004g1-51 for pgsql-performance@postgresql.org; Fri, 02 Apr 2004 15:55:30 +0100 Received: (qmail 15397 invoked from network); 2 Apr 2004 14:55:48 -0000 Received: from peneca.star.le.ac.uk (143.210.36.224) by mail.star.le.ac.uk with SMTP; 2 Apr 2004 14:55:48 -0000 Date: Fri, 2 Apr 2004 15:55:26 +0100 (BST) From: Clive Page To: pgsql-performance@postgresql.org Subject: Spatial join insists on sequential scan of larger table 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: 200404/10 X-Sequence-Number: 6310 I am trying to do a spatial join between two tables each of which has a column of type BOX called ERRBOX, with R-TREE indices created on both. The smaller table, xmm1, has 56,711 rows, the larger one, twomass, has 177,757,299 rows. The most efficient way to join these is to do a sequential scan of the smaller table, and an R-tree lookup on the larger. However for a simple inner join the optimiser seems to want to do the reverse, for example: EXPLAIN SELECT x.ra AS xra, x.decl AS xdecl, t.ra AS tra, t.decl AS tdecl FROM xmm1 AS x INNER JOIN twomass AS t ON x.errbox && t.errbox; QUERY PLAN ---------------------------------------------------------------------------------- Nested Loop (cost=0.00..196642756520.34 rows=49506496044 width=32) -> Seq Scan on twomass t (cost=0.00..9560002.72 rows=177023872 width=48) -> Index Scan using xmm1box on xmm1 x (cost=0.00..1107.28 rows=280 width=48) Index Cond: (x.errbox && "outer".errbox) Reversing the join condition (i.e. t.errbox && x.errbox) and similar make no difference, nor does using the old implicit join syntax. If, however, I specify an outer join such as: EXPLAIN SELECT x.ra AS xra, x.decl AS xdecl, t.ra AS tra, t.decl AS tdecl FROM xmm1 AS x LEFT OUTER JOIN twomass AS t ON x.errbox && t.errbox; QUERY PLAN ---------------------------------------------------------------------------------------- Nested Loop Left Join (cost=0.00..198945259325.90 rows=49506496044 width=32) -> Seq Scan on xmm1 x (cost=0.00..8592.32 rows=55932 width=48) -> Index Scan using tbox on twomass t (cost=0.00..3545848.88 rows=885119 width=48) Index Cond: ("outer".errbox && t.errbox) This executes, it need hardly be said, a whole lot faster. I found that I can also force a sequential scan of the smaller table by dropping its R-tree index, but I may need this in other operations, so this isn't a very satisfactory solution. It's odd that an outer join should be faster than an inner one, or to put it another way, after dropping an index there is more than an order of magnitude speed increase. I'm using Postgres 7.4.1 on Red Hat Linux. Has anyone had similar problems with spatial joins? -- Clive Page Dept of Physics & Astronomy, University of Leicester, Leicester, LE1 7RH, U.K. From pgsql-performance-owner@postgresql.org Fri Apr 2 11:47:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 66E0FD1EB5F for ; Fri, 2 Apr 2004 15:46:25 +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 69256-01 for ; Fri, 2 Apr 2004 11:46:27 -0400 (AST) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 99D67D1EB47 for ; Fri, 2 Apr 2004 11:46:22 -0400 (AST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i32FkRkc027944; Fri, 2 Apr 2004 10:46:27 -0500 (EST) To: Clive Page Cc: pgsql-performance@postgresql.org Subject: Re: Spatial join insists on sequential scan of larger table In-reply-to: References: Comments: In-reply-to Clive Page message dated "Fri, 02 Apr 2004 15:55:26 +0100" Date: Fri, 02 Apr 2004 10:46:26 -0500 Message-ID: <27943.1080920786@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/11 X-Sequence-Number: 6311 Clive Page writes: > This executes, it need hardly be said, a whole lot faster. Could we see EXPLAIN ANALYZE output? The estimated costs for the two cases are nearly the same, which says to me that there's something wrong with the cost model for r-tree lookups, but I don't know what it is. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 2 11:53:10 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A3159D1EB59 for ; Fri, 2 Apr 2004 15:52:39 +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 67475-06 for ; Fri, 2 Apr 2004 11:52:42 -0400 (AST) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id 33C36D1EB5F for ; Fri, 2 Apr 2004 11:52:37 -0400 (AST) Received: (qmail 5373 invoked by uid 500); 2 Apr 2004 15:56:04 -0000 Date: Fri, 2 Apr 2004 09:56:04 -0600 From: Bruno Wolff III To: Palle Girgensohn Cc: pgsql-performance@postgresql.org Subject: Re: single index on more than two coulumns a bad thing? Message-ID: <20040402155604.GC5151@wolff.to> Mail-Followup-To: Palle Girgensohn , pgsql-performance@postgresql.org References: <24070000.1080860445@palle.girgensohn.se> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <24070000.1080860445@palle.girgensohn.se> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/12 X-Sequence-Number: 6312 On Fri, Apr 02, 2004 at 01:00:45 +0200, Palle Girgensohn wrote: > > Is it always bad to create index xx on yy (field1, field2, field3); > > I guess the problem is that the index might often grow bigger than the > table, or at least big enough not to speed up the queries? One place where you need them in postgres is enforcing unique multicolumn keys. These will get created implicitly from the unique (or primary key) constraint. It isn't all that unusual to have a table that describes a many to many (to many ...) relationship where the primary key is all of the columns. From pgsql-performance-owner@postgresql.org Fri Apr 2 12:06:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 05A85D1EB55 for ; Fri, 2 Apr 2004 16:05:00 +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 75182-05 for ; Fri, 2 Apr 2004 12:05:03 -0400 (AST) Received: from artemis.le.ac.uk (ntp2c.le.ac.uk [143.210.16.126]) by svr1.postgresql.org (Postfix) with ESMTP id CE6EED1ECBC for ; Fri, 2 Apr 2004 12:04:57 -0400 (AST) Received: from [143.210.36.58] (helo=mail.star.le.ac.uk) by artemis.le.ac.uk with smtp (Exim 4.30) id 1B9RAE-00052C-1D for pgsql-performance@postgresql.org; Fri, 02 Apr 2004 17:05:02 +0100 Received: (qmail 3565 invoked from network); 2 Apr 2004 16:05:23 -0000 Received: from peneca.star.le.ac.uk (143.210.36.224) by mail.star.le.ac.uk with SMTP; 2 Apr 2004 16:05:23 -0000 Date: Fri, 2 Apr 2004 17:05:01 +0100 (BST) From: Clive Page To: pgsql-performance@postgresql.org Subject: Re: Spatial join insists on sequential scan of larger In-Reply-To: <27943.1080920786@sss.pgh.pa.us> 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: 200404/13 X-Sequence-Number: 6313 On Fri, 2 Apr 2004, Tom Lane wrote: > Could we see EXPLAIN ANALYZE output? Certainly, but that's going to take a little time (as the ANALYZE causes it to run the actual query, which I only just discovered), so may have to wait until Monday if I don't get time to finish it this afternoon. -- Clive Page Dept of Physics & Astronomy, University of Leicester, Leicester, LE1 7RH, U.K. From pgsql-performance-owner@postgresql.org Sat Apr 3 06:47:07 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 78170D1ECA7 for ; Sat, 3 Apr 2004 10:47:05 +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 98402-03 for ; Sat, 3 Apr 2004 06:47:15 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 73DCDD1ECFC for ; Sat, 3 Apr 2004 06:47:02 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001DF5; 3 Apr 2004 11:51:09 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 3 Apr 2004 11:51:02 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001DF4; 3 Apr 2004 11:50:52 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Sat, 03 Apr 2004 11:50:51 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <406EA51B.21218.D2DE061@localhost> X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/14 X-Sequence-Number: 6314 On 2 Apr 2004 at 22:36, pgsql-performance@postgresql. wrote: OK, some more detail: Before wiping 2.4 off my test box for the second time: SQL Statement for update: update staff_booking set time_from = r.time_from from order_reqt r where r.reqt_id = staff_booking.reqt_id; Explain: (on 2.4) QUERY PLAN Merge Join (cost=0.00..185731.30 rows=2845920 width=92) Merge Cond: ("outer".reqt_id = "inner".reqt_id) -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..53068.20 rows=2206291 width=6) -> Index Scan using staff_book_idx2 on staff_booking (cost=0.00..99579.21 rows=2845920 width=90) Total execution time: 18 hours 12 minutes vacuum full analyze: total time 3 hours 22 minutes Wait 2 hours for re-install 2.6, set params etc. restore database. Same SQL Statement Explain: (on 2.6) QUERY PLAN Merge Join (cost=0.00..209740.24 rows=2845920 width=92) Merge Cond: ("outer".reqt_id = "inner".reqt_id) -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..50734.20 rows=2206291 width=6) -> Index Scan using staff_book_idx2 on staff_booking (cost=0.00..117921.92 rows=2845920 width=90) Total execution time: 2 hours 53 minutes vacuum full analyze: total time 1 hours 6 minutes Table definitions for the two tables involved: CREATE TABLE ORDER_REQT ( REQT_ID SERIAL, ORDER_ID integer NOT NULL, DAYOFWEEK smallint NOT NULL CHECK (DAYOFWEEK BETWEEN 0 AND 6), TIME_FROM smallint NOT NULL CHECK (TIME_FROM BETWEEN 0 AND 1439), DURATION smallint NOT NULL CHECK (DURATION BETWEEN 0 AND 1439), PRODUCT_ID integer NOT NULL, NUMBER_REQT smallint NOT NULL DEFAULT (1), WROPTIONS integer NOT NULL DEFAULT 0, UID_REF integer NOT NULL, DT_STAMP timestamp NOT NULL DEFAULT current_timestamp, Sentinel_Priority integer NOT NULL DEFAULT 0, PERIOD smallint NOT NULL DEFAULT 1 CHECK (PERIOD BETWEEN -2 AND 4), FREQUENCY smallint NOT NULL DEFAULT 1, PRIMARY KEY (REQT_ID) ); CREATE TABLE STAFF_BOOKING ( BOOKING_ID SERIAL, REQT_ID integer NOT NULL, ENTITY_TYPE smallint NOT NULL DEFAULT 3 check(ENTITY_TYPE in(3,4)), STAFF_ID integer NOT NULL, CONTRACT_ID integer NOT NULL, TIME_FROM smallint NOT NULL CHECK (TIME_FROM BETWEEN 0 AND 1439), DURATION smallint NOT NULL CHECK (DURATION BETWEEN 0 AND 1439), PERIOD smallint NOT NULL DEFAULT 1 CHECK (PERIOD BETWEEN -2 AND 4), FREQUENCY smallint NOT NULL DEFAULT 1, TRAVEL_TO smallint NOT NULL DEFAULT 0, UID_REF integer NOT NULL, DT_STAMP timestamp NOT NULL DEFAULT current_timestamp, SELL_PRICE numeric(10,4) NOT NULL DEFAULT 0, COST_PRICE numeric(10,4) NOT NULL DEFAULT 0, MIN_SELL_PRICE numeric(10,4) NOT NULL DEFAULT 0, MIN_COST_PRICE numeric(10,4) NOT NULL DEFAULT 0, Sentinel_Priority integer NOT NULL DEFAULT 0, CHECK_INTERVAL smallint NOT NULL DEFAULT 0, STATUS smallint NOT NULL DEFAULT 0, WROPTIONS integer NOT NULL DEFAULT 0, PRIMARY KEY (BOOKING_ID) ); Foreign keys: ALTER TABLE ORDER_REQT ADD FOREIGN KEY ( ORDER_ID ) REFERENCES MAIN_ORDER ( ORDER_ID ) ON DELETE CASCADE; ALTER TABLE ORDER_REQT ADD FOREIGN KEY ( PRODUCT_ID ) REFERENCES PRODUCT ( PRODUCT_ID ); ALTER TABLE STAFF_BOOKING ADD FOREIGN KEY ( CONTRACT_ID ) REFERENCES STAFF_CONTRACT ( CONTRACT_ID ); ALTER TABLE STAFF_BOOKING ADD FOREIGN KEY ( STAFF_ID ) REFERENCES STAFF ( STAFF_ID ); Indexes: CREATE INDEX FK_IDX_ORDER_REQT ON ORDER_REQT ( ORDER_ID ); CREATE INDEX FK_IDX_ORDER_REQT_2 ON ORDER_REQT ( PRODUCT_ID ); CREATE INDEX ORDER_REQT_IDX ON ORDER_REQT ( ORDER_ID, PRODUCT_ID ); CREATE INDEX ORDER_REQT_IDX4 ON ORDER_REQT ( REQT_ID, TIME_FROM, DURATION ); CREATE INDEX FK_IDX_STAFF_BOOKING ON STAFF_BOOKING ( CONTRACT_ID ); CREATE INDEX FK_IDX_STAFF_BOOKING_2 ON STAFF_BOOKING ( STAFF_ID ); CREATE INDEX STAFF_BOOK_IDX1 ON STAFF_BOOKING ( STAFF_ID, REQT_ID ); CREATE INDEX STAFF_BOOK_IDX2 ON STAFF_BOOKING ( REQT_ID ); CREATE INDEX STAFF_BOOK_IDX3 ON STAFF_BOOKING ( BOOKING_ID, REQT_ID ); CREATE INDEX STAFF_BOOK_IDX4 ON STAFF_BOOKING ( BOOKING_ID, CONTRACT_ID ); There are no indexes on the columns involved in the update, they are not required for my usual select statements. This is an attempt to slightly denormalise the design to get the performance up comparable to SQL Server 2000. We hope to move some of our databases over to PostgreSQL later in the year and this is part of the ongoing testing. SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) so I am hand optimising some of the more frequently used SQL and/or tweaking the database design slightly. Later, after deciphering SQLServers graphical plans I will attempt to post comparitive performance/access plans, using the same data of course, if anyone would be interested.... Cheers, Gary. On 2 Apr 2004 at 1:32, Tom Lane wrote: > "Gary Doades" writes: > > As a test in PosgreSQL I issued a statement to update a single column > > of a table containing 2.8 million rows with the values of a column in > > a table with similar rowcount. Using the above spec I had to stop the > > server after 17 hours. The poor thing was thrashing the hard disk and > > doing more swapping than useful work. > > This statement is pretty much content-free, since you did not show us > the table schemas, the query, or the EXPLAIN output for the query. > (I'll forgive you the lack of EXPLAIN ANALYZE, but you could easily > have provided all the other hard facts.) There's really no way to tell > where the bottleneck is. Maybe it's a kernel-level issue, but I would > not bet on that without more evidence. I'd definitely not bet on it > without direct confirmation that the same query plan was used in both > setups. > > regards, tom lane > > ---------------------------(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 > > > -- > Incoming mail is certified Virus Free. > Checked by AVG Anti-Virus (http://www.grisoft.com). > Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 > From pgsql-performance-owner@postgresql.org Sat Apr 3 09:23:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 05C7DD1DB5B for ; Sat, 3 Apr 2004 13:23:58 +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 31923-10 for ; Sat, 3 Apr 2004 09:24:09 -0400 (AST) Received: from palle.girgensohn.se (1-2-8-5a.asp.sth.bostream.se [82.182.157.66]) by svr1.postgresql.org (Postfix) with ESMTP id 1FC6CD1D55F for ; Sat, 3 Apr 2004 09:23:55 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by palle.girgensohn.se (8.12.11/8.12.11) with ESMTP id i33DO7LF019918; Sat, 3 Apr 2004 15:24:07 +0200 (CEST) (envelope-from girgen@pingpong.net) Date: Sat, 03 Apr 2004 15:24:07 +0200 From: Palle Girgensohn To: Bruno Wolff III Cc: pgsql-performance@postgresql.org Subject: Re: single index on more than two coulumns a bad thing? Message-ID: <1730000.1080998647@palle.girgensohn.se> In-Reply-To: <20040402155604.GC5151@wolff.to> References: <24070000.1080860445@palle.girgensohn.se> <20040402155604.GC5151@wolff.to> X-Mailer: Mulberry/3.1.0 (Linux/x86) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/15 X-Sequence-Number: 6315 --On fredag, april 02, 2004 09.56.04 -0600 Bruno Wolff III wrote: > On Fri, Apr 02, 2004 at 01:00:45 +0200, > Palle Girgensohn wrote: >> >> Is it always bad to create index xx on yy (field1, field2, field3); >> >> I guess the problem is that the index might often grow bigger than the >> table, or at least big enough not to speed up the queries? > > One place where you need them in postgres is enforcing unique multicolumn > keys. These will get created implicitly from the unique (or primary key) > constraint. It isn't all that unusual to have a table that describes > a many to many (to many ...) relationship where the primary key is all > of the columns. True, of course! /Palle From pgsql-performance-owner@postgresql.org Sat Apr 3 15:00:10 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A1EFED1ECBD for ; Sat, 3 Apr 2004 19:00:01 +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 22193-06 for ; Sat, 3 Apr 2004 15:00:14 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 55A5CD1ECA7 for ; Sat, 3 Apr 2004 14:59:54 -0400 (AST) Received: from [63.195.55.98] (HELO 192.168.1.29) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4763818; Sat, 03 Apr 2004 11:01:30 -0800 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: "Gary Doades" , pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Sat, 3 Apr 2004 10:59:39 -0800 User-Agent: KMail/1.5.4 References: <406EA51B.21218.D2DE061@localhost> In-Reply-To: <406EA51B.21218.D2DE061@localhost> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404031059.39268.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/16 X-Sequence-Number: 6316 Gary, > There are no indexes on the columns involved in the update, they are > not required for my usual select statements. This is an attempt to > slightly denormalise the design to get the performance up comparable > to SQL Server 2000. We hope to move some of our databases over to > PostgreSQL later in the year and this is part of the ongoing testing. > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > so I am hand optimising some of the more frequently used > SQL and/or tweaking the database design slightly. Hmmm ... that hasn't been my general experience on complex queries. However, it may be due to a difference in ANALYZE statistics. I'd love to see you increase your default_stats_target, re-analyze, and see if PostgreSQL gets "smarter". -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Sat Apr 3 17:30:14 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 041E4D1ED8D for ; Sat, 3 Apr 2004 21:30:13 +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 59107-07 for ; Sat, 3 Apr 2004 17:30:27 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 77832D1ED91 for ; Sat, 3 Apr 2004 17:30:10 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001E59; 3 Apr 2004 22:34:24 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 3 Apr 2004 22:34:06 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001E58; 3 Apr 2004 22:33:56 +0100 Resent-From: "Gary Doades" Resent-To: pgsql-performance@postgresql.org Resent-date: Sat, 03 Apr 2004 22:33:56 +0100 X-cs: R From: Gary Doades X-RS-ID: X-RS-Flags: 0,0,1,1,0,0,0 X-RS-Header: In-reply-to: <200404031059.39268.josh@agliodbs.com> X-RS-Header: References: <406EA51B.21218.D2DE061@localhost> X-RS-Sigset: 0 To: pgsql-performance-owner@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: 8BIT Date: Sat, 03 Apr 2004 20:32:35 +0100 Message-ID: Resent-Message-Id: <20040403213010.77832D1ED91@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/17 X-Sequence-Number: 6317 Actually it hasn't been my experience either. Most of my queries against the database, large and small are either a little quicker or no real difference. I have only really noticed big differences under stress when memory (RAM) is being squeezed. The main winner on 2.6 seems to be write performance and memory management. Unfortunately I only have one test machine and I can't really keep switching between 2.4 and 2.6 to do the comparisons. I had written down 27 timings from a set of SQL of varying complexity using the 2.4 kernel. Each SQL statement was executed 10 times and the average of the last 5 was used. I can only really compare those timings against the new installation on 2.6. I know that this is not ideal "real world" testing, but it is good enough for me at the moment. Unless anyone has contradictory indications then I will proceed with 2.6. I did increase the default stats target from 10 to 50 and re-analysed. The explain numbers are slightly different, but the time to run was almost the same. Not surprising since the plan was the same. QUERY PLAN Merge Join (cost=0.00..192636.20 rows=2845920 width=92) Merge Cond: ("outer".reqt_id = "inner".reqt_id) -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..52662.40 rows=2206291 width=6) -> Index Scan using staff_book_idx2 on staff_booking (cost=0.00..102529.28 rows=2845920 width=90) On 3 Apr 2004 at 10:59, Josh Berkus wrote: Gary, > There are no indexes on the columns involved in the update, they are > not required for my usual select statements. This is an attempt to > slightly denormalise the design to get the performance up comparable > to SQL Server 2000. We hope to move some of our databases over to > PostgreSQL later in the year and this is part of the ongoing testing. > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > so I am hand optimising some of the more frequently used > SQL and/or tweaking the database design slightly. Hmmm ... that hasn't been my general experience on complex queries. However, it may be due to a difference in ANALYZE statistics. I'd love to see you increase your default_stats_target, re-analyze, and see if PostgreSQL gets "smarter". -- -Josh Berkus Aglio Database Solutions San Francisco ---------------------------(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 -- Incoming mail is certified Virus Free. Checked by AVG Anti-Virus (http://www.grisoft.com). Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 From pgsql-performance-owner@postgresql.org Sat Apr 3 17:31:02 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 45624D1E2D7 for ; Sat, 3 Apr 2004 21:31:01 +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 62435-02 for ; Sat, 3 Apr 2004 17:31:15 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id B7E31D1CAD0 for ; Sat, 3 Apr 2004 17:30:58 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001E5B; 3 Apr 2004 22:35:12 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 3 Apr 2004 22:35:01 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001E5A; 3 Apr 2004 22:34:54 +0100 Resent-From: "Gary Doades" Resent-To: pgsql-performance@postgresql.org Resent-date: Sat, 03 Apr 2004 22:34:54 +0100 X-cs: R From: Gary Doades X-RS-ID: X-RS-Flags: 0,0,1,1,0,0,0 X-RS-Header: In-reply-to: <200404031059.39268.josh@agliodbs.com> X-RS-Header: References: <406EA51B.21218.D2DE061@localhost> X-RS-Sigset: 0 To: pgsql-performance-owner@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: 8BIT Date: Sat, 03 Apr 2004 21:16:10 +0100 Message-ID: Resent-Message-Id: <20040403213058.B7E31D1CAD0@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/18 X-Sequence-Number: 6318 Sorry, I think I misread your post in my last reply. I thought you were still talking about the big update.... The main thing I have noticed about SQLServer is it seems more willing to do hash or merge joins than PostgreSQL. I have experimented with various postgresql.conf parameters and even turned off nested loops to see the difference. When actually getting a merge join out of PostgreSQL when it wanted to do a nested loop it, not surprisingly, took longer to execute. Looking at the SQLServer plan it seemed to be spending MUCH less time in the sort operations than PostgreSQL. This is probably what leads SQLServer to go for hash/merge joins more often. The other problem is that the SQLServer timings are skewed by its query plan caching. For one query SQLserver plan said it spent 2% of its time in a big sort, the same query in PostgreSQL when hash join was forced spent 23% of its time on the sort (from explain analyse actual stats). I have played about with the sort_mem, but it doesn't make much diffrence. I have also noticed that SQLServer tends to fold more complex IN subselects into the main query using merge joins, maybe for the same reason as above. SQLServer seems to have some more "exotic" joins ("nested loop/left semi join","nested loop/left anti semi join"). These are probably just variants of nested loops, but I don't know enough about it to say if they make a difference. Clustered indexes and clustered index seeks also seem to be a big player in the more complex queries. I still have quite a lot comparitive testing and tuning to do before I can nail it down further, but I will let you know when I have some hard stats to go on. On 3 Apr 2004 at 10:59, Josh Berkus wrote: Gary, > There are no indexes on the columns involved in the update, they are > not required for my usual select statements. This is an attempt to > slightly denormalise the design to get the performance up comparable > to SQL Server 2000. We hope to move some of our databases over to > PostgreSQL later in the year and this is part of the ongoing testing. > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > so I am hand optimising some of the more frequently used > SQL and/or tweaking the database design slightly. Hmmm ... that hasn't been my general experience on complex queries. However, it may be due to a difference in ANALYZE statistics. I'd love to see you increase your default_stats_target, re-analyze, and see if PostgreSQL gets "smarter". -- -Josh Berkus Aglio Database Solutions San Francisco ---------------------------(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 -- Incoming mail is certified Virus Free. Checked by AVG Anti-Virus (http://www.grisoft.com). Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 From pgsql-performance-owner@postgresql.org Sat Apr 3 17:31:21 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A4F4ED1E9C8 for ; Sat, 3 Apr 2004 21:31:17 +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 63360-01 for ; Sat, 3 Apr 2004 17:31:31 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 7ACA1D1E9BB for ; Sat, 3 Apr 2004 17:31:14 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001E5D; 3 Apr 2004 22:35:28 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 3 Apr 2004 22:35:13 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001E5C; 3 Apr 2004 22:35:08 +0100 Resent-From: "Gary Doades" Resent-To: pgsql-performance@postgresql.org Resent-date: Sat, 03 Apr 2004 22:35:07 +0100 X-cs: R From: Gary Doades X-RS-ID: X-RS-Flags: 0,0,1,1,0,0,0 X-RS-Header: In-reply-to: <200404031338.i33DccbP015107@smtp.gvtc.com> X-RS-Header: References: <406EA51B.21218.D2DE061@localhost> X-RS-Sigset: 0 To: pgsql-performance-owner@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: 8BIT Date: Sat, 03 Apr 2004 21:20:49 +0100 Message-ID: Resent-Message-Id: <20040403213114.7ACA1D1E9BB@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/20 X-Sequence-Number: 6320 Thanks, I know about set showplan_text, but it is only the equivalent of explain, not explain analyze. The graphical plan gives full statistics, runtime, percentage cost, loop execution counts etc. which is much more useful. I don't know of a way of getting the graphical plan content in text form. Cheers, Gary. On 3 Apr 2004 at 6:50, @g v t c wrote: Use "Set Show_Plan" or something of the sort in Query Analyzer. Then run your SQL. This will change the graphical plan to a text plan similar to Postgresql or at least something close to readable. Gary Doades wrote: >On 2 Apr 2004 at 22:36, pgsql-performance@postgresql. wrote: > >OK, some more detail: > >Before wiping 2.4 off my test box for the second time: > >SQL Statement for update: >update staff_booking set time_from = r.time_from from order_reqt r where r.reqt_id = >staff_booking.reqt_id; > >Explain: (on 2.4) >QUERY PLAN >Merge Join (cost=0.00..185731.30 rows=2845920 width=92) > Merge Cond: ("outer".reqt_id = "inner".reqt_id) > -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..53068.20 >rows=2206291 width=6) > -> Index Scan using staff_book_idx2 on staff_booking (cost=0.00..99579.21 >rows=2845920 width=90) > >Total execution time: 18 hours 12 minutes > >vacuum full analyze: total time 3 hours 22 minutes > >Wait 2 hours for re-install 2.6, set params etc. >restore database. > >Same SQL Statement >Explain: (on 2.6) >QUERY PLAN >Merge Join (cost=0.00..209740.24 rows=2845920 width=92) > Merge Cond: ("outer".reqt_id = "inner".reqt_id) > -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..50734.20 >rows=2206291 width=6) > -> Index Scan using staff_book_idx2 on staff_booking (cost=0.00..117921.92 >rows=2845920 width=90) > >Total execution time: 2 hours 53 minutes > >vacuum full analyze: total time 1 hours 6 minutes > >Table definitions for the two tables involved: >CREATE TABLE ORDER_REQT >( > REQT_ID SERIAL, > ORDER_ID integer NOT NULL, > DAYOFWEEK smallint NOT NULL CHECK (DAYOFWEEK >BETWEEN 0 AND 6), > TIME_FROM smallint NOT NULL CHECK (TIME_FROM >BETWEEN 0 AND 1439), > DURATION smallint NOT NULL CHECK (DURATION >BETWEEN 0 AND 1439), > PRODUCT_ID integer NOT NULL, > NUMBER_REQT smallint NOT NULL DEFAULT (1), > WROPTIONS integer NOT NULL DEFAULT 0, > UID_REF integer NOT NULL, > DT_STAMP timestamp NOT NULL DEFAULT >current_timestamp, > Sentinel_Priority integer NOT NULL DEFAULT 0, > PERIOD smallint NOT NULL DEFAULT 1 CHECK >(PERIOD BETWEEN -2 AND 4), > FREQUENCY smallint NOT NULL DEFAULT 1, > PRIMARY KEY (REQT_ID) >); > >CREATE TABLE STAFF_BOOKING >( > BOOKING_ID SERIAL, > REQT_ID integer NOT NULL, > ENTITY_TYPE smallint NOT NULL DEFAULT 3 >check(ENTITY_TYPE in(3,4)), > STAFF_ID integer NOT NULL, > CONTRACT_ID integer NOT NULL, > TIME_FROM smallint NOT NULL CHECK (TIME_FROM >BETWEEN 0 AND 1439), > DURATION smallint NOT NULL CHECK (DURATION >BETWEEN 0 AND 1439), > PERIOD smallint NOT NULL DEFAULT 1 CHECK >(PERIOD BETWEEN -2 AND 4), > FREQUENCY smallint NOT NULL DEFAULT 1, > TRAVEL_TO smallint NOT NULL DEFAULT 0, > UID_REF integer NOT NULL, > DT_STAMP timestamp NOT NULL DEFAULT >current_timestamp, > SELL_PRICE numeric(10,4) NOT NULL DEFAULT 0, > COST_PRICE numeric(10,4) NOT NULL DEFAULT 0, > MIN_SELL_PRICE numeric(10,4) NOT NULL DEFAULT 0, > MIN_COST_PRICE numeric(10,4) NOT NULL DEFAULT 0, > Sentinel_Priority integer NOT NULL DEFAULT 0, > CHECK_INTERVAL smallint NOT NULL DEFAULT 0, > STATUS smallint NOT NULL DEFAULT 0, > WROPTIONS integer NOT NULL DEFAULT 0, > PRIMARY KEY (BOOKING_ID) >); > >Foreign keys: > >ALTER TABLE ORDER_REQT ADD > FOREIGN KEY > ( > ORDER_ID > ) REFERENCES MAIN_ORDER ( > ORDER_ID > ) ON DELETE CASCADE; > >ALTER TABLE ORDER_REQT ADD > FOREIGN KEY > ( > PRODUCT_ID > ) REFERENCES PRODUCT ( > PRODUCT_ID > ); > >ALTER TABLE STAFF_BOOKING ADD > FOREIGN KEY > ( > CONTRACT_ID > ) REFERENCES STAFF_CONTRACT ( > CONTRACT_ID > ); > >ALTER TABLE STAFF_BOOKING ADD > FOREIGN KEY > ( > STAFF_ID > ) REFERENCES STAFF ( > STAFF_ID > ); > > >Indexes: > >CREATE INDEX FK_IDX_ORDER_REQT > ON ORDER_REQT > ( > ORDER_ID > ); > >CREATE INDEX FK_IDX_ORDER_REQT_2 > ON ORDER_REQT > ( > PRODUCT_ID > ); > >CREATE INDEX ORDER_REQT_IDX ON ORDER_REQT >( > ORDER_ID, > PRODUCT_ID >); > >CREATE INDEX ORDER_REQT_IDX4 ON ORDER_REQT >( > REQT_ID, > TIME_FROM, > DURATION >); > >CREATE INDEX FK_IDX_STAFF_BOOKING > ON STAFF_BOOKING > ( > CONTRACT_ID > ); > >CREATE INDEX FK_IDX_STAFF_BOOKING_2 > ON STAFF_BOOKING > ( > STAFF_ID > ); > >CREATE INDEX STAFF_BOOK_IDX1 ON STAFF_BOOKING >( > STAFF_ID, > REQT_ID >); > >CREATE INDEX STAFF_BOOK_IDX2 ON STAFF_BOOKING >( > REQT_ID >); > >CREATE INDEX STAFF_BOOK_IDX3 ON STAFF_BOOKING >( > BOOKING_ID, > REQT_ID >); > > >CREATE INDEX STAFF_BOOK_IDX4 ON STAFF_BOOKING >( > BOOKING_ID, > CONTRACT_ID >); > >There are no indexes on the columns involved in the update, they are >not required for my usual select statements. This is an attempt to >slightly denormalise the design to get the performance up comparable >to SQL Server 2000. We hope to move some of our databases over to >PostgreSQL later in the year and this is part of the ongoing testing. >SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) >so I am hand optimising some of the more frequently used >SQL and/or tweaking the database design slightly. > >Later, after deciphering SQLServers graphical plans I will attempt to >post comparitive performance/access plans, using the same data of >course, if anyone would be interested.... > >Cheers, >Gary. > > > >On 2 Apr 2004 at 1:32, Tom Lane wrote: > > > >>"Gary Doades" writes: >> >> >>>As a test in PosgreSQL I issued a statement to update a single column >>>of a table containing 2.8 million rows with the values of a column in >>>a table with similar rowcount. Using the above spec I had to stop the >>>server after 17 hours. The poor thing was thrashing the hard disk and >>>doing more swapping than useful work. >>> >>> >> >>This statement is pretty much content-free, since you did not show us >>the table schemas, the query, or the EXPLAIN output for the query. >>(I'll forgive you the lack of EXPLAIN ANALYZE, but you could easily >>have provided all the other hard facts.) There's really no way to tell >>where the bottleneck is. Maybe it's a kernel-level issue, but I would >>not bet on that without more evidence. I'd definitely not bet on it >>without direct confirmation that the same query plan was used in both >>setups. >> >> regards, tom lane >> >>---------------------------(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 >> >> >>-- >>Incoming mail is certified Virus Free. >>Checked by AVG Anti-Virus (http://www.grisoft.com). >>Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 >> >> >> > > > >---------------------------(end of broadcast)--------------------------- >TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) > > > > -- Incoming mail is certified Virus Free. Checked by AVG Anti-Virus (http://www.grisoft.com). Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 From pgsql-performance-owner@postgresql.org Sat Apr 3 17:31:19 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 67F53D1E9D3 for ; Sat, 3 Apr 2004 21:31:18 +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 59322-06 for ; Sat, 3 Apr 2004 17:31:31 -0400 (AST) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 7D9C5D1E9BC for ; Sat, 3 Apr 2004 17:31:14 -0400 (AST) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001E5F; 3 Apr 2004 22:35:28 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 3 Apr 2004 22:35:24 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001E5E; 3 Apr 2004 22:35:18 +0100 Resent-From: "Gary Doades" Resent-To: pgsql-performance@postgresql.org Resent-date: Sat, 03 Apr 2004 22:35:18 +0100 X-cs: R From: Gary Doades X-RS-ID: X-RS-Flags: 0,0,1,1,0,0,0 X-RS-Header: In-reply-to: <200404031059.39268.josh@agliodbs.com> X-RS-Header: References: <406EA51B.21218.D2DE061@localhost> X-RS-Sigset: 0 To: pgsql-performance-owner@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: 8BIT Date: Sat, 03 Apr 2004 22:29:01 +0100 Message-ID: Resent-Message-Id: <20040403213114.7D9C5D1E9BC@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/19 X-Sequence-Number: 6319 Following on from Josh's response and my previous reply on SQLServer planning. The main problem query is this one: SELECT VS.*,VL.TEL1,SC.CONTRACT_ID,SC.CONTRACT_REF, SC.MAX_HOURS, SC.MIN_HOURS, (SELECT COUNT(*) FROM TIMESHEET_DETAIL JOIN MAIN_ORDER ON (MAIN_ORDER.ORDER_ID = TIMESHEET_DETAIL.ORDER_ID AND MAIN_ORDER.CLIENT_ID = 6) WHERE TIMESHEET_DETAIL.CONTRACT_ID = SC.CONTRACT_ID) AS VISITS, (SELECT (SUM(R.DURATION+1))/60.0 FROM ORDER_REQT R JOIN STAFF_BOOKING B ON (B.REQT_ID = R.REQT_ID) JOIN BOOKING_PLAN BP ON (BP.BOOKING_ID = B.BOOKING_ID) WHERE B.CONTRACT_ID = SC.CONTRACT_ID AND BP.BOOKING_DATE BETWEEN '2004-06-12' AND '2004-06-18') AS RHOURS FROM VSTAFF VS JOIN STAFF_CONTRACT SC ON (SC.STAFF_ID = VS.STAFF_ID) JOIN VLOCATION VL ON (VL.LOCATION_ID = VS.LOCATION_ID) JOIN SEARCH_REQT_RESULT SR ON (SR.STAFF_ID = VS.STAFF_ID) WHERE SR.SEARCH_ID = 1 AND SC.CONTRACT_ID IN (SELECT C.CONTRACT_ID FROM STAFF_PRODUCT P,STAFF_CONTRACT C WHERE P.CONTRACT_ID=C.CONTRACT_ID AND C.STAFF_ID = VS.STAFF_ID AND P.PRODUCT_ID IN (SELECT PRODUCT_ID FROM SEARCH_ORDER_REQT WHERE SEARCH_ID = 1) AND C.AVAIL_DATE_FROM <= '2004-06-12' AND C.AVAIL_DATE_TO >= '2004-06-18' GROUP BY C.CONTRACT_ID HAVING (COUNT(C.CONTRACT_ID) = (SELECT COUNT(DISTINCT PRODUCT_ID) FROM SEARCH_ORDER_REQT WHERE SEARCH_ID = 1))) The explain analyze is: QUERY PLAN Nested Loop (cost=101.54..1572059.57 rows=135 width=152) (actual time=13749.100..1304586.501 rows=429 loops=1) InitPlan -> Index Scan using fk_idx_wruserarea on wruserarea (cost=3.26..6.52 rows=1 width=4) (actual time=0.944..0.944 rows=1 loops=1) Index Cond: (area_id = 1) Filter: (uid = $4) InitPlan -> Seq Scan on wruser (cost=0.00..3.26 rows=1 width=4) (actual time=0.686..0.691 rows=1 loops=1) Filter: ((username)::name = "current_user"()) -> Hash Join (cost=95.02..3701.21 rows=215 width=138) (actual time=100.476..1337.392 rows=429 loops=1) Hash Cond: ("outer".staff_id = "inner".staff_id) Join Filter: (subplan) -> Seq Scan on staff_contract sc (cost=0.00..33.24 rows=1024 width=37) (actual time=0.114..245.366 rows=1024 loops=1) -> Hash (cost=93.95..93.95 rows=430 width=109) (actual time=38.563..38.563 rows=0 loops=1) -> Hash Join (cost=47.47..93.95 rows=430 width=109) (actual time=15.502..36.627 rows=429 loops=1) Hash Cond: ("outer".staff_id = "inner".staff_id) -> Seq Scan on staff (cost=34.61..66.48 rows=1030 width=105) (actual time=9.655..15.264 rows=1030 loops=1) Filter: ((hashed subplan) OR $5) SubPlan -> Seq Scan on staff_area (cost=10.73..33.38 rows=493 width=4) (actual time=8.452..8.452 rows=0 loops=1) Filter: ((hashed subplan) OR (area_id = 1)) SubPlan -> Seq Scan on wruserarea (cost=3.26..10.72 rows=5 width=4) (actual time=0.977..1.952 rows=1 loops=1) Filter: (uid = $1) InitPlan -> Seq Scan on wruser (cost=0.00..3.26 rows=1 width=4) (actual time=0.921..0.926 rows=1 loops=1) Filter: ((username)::name = "current_user"()) -> Hash (cost=11.79..11.79 rows=430 width=4) (actual time=5.705..5.705 rows=0 loops=1) -> Index Scan using fk_idx_search_reqt_result on search_reqt_result sr (cost=0.00..11.79 rows=430 width=4) (actual time=0.470..4.482 rows=429 loops=1) Index Cond: (search_id = 1) SubPlan -> HashAggregate (cost=8.32..8.32 rows=1 width=4) (actual time=2.157..2.157 rows=1 loops=429) Filter: (count(contract_id) = $9) InitPlan -> Aggregate (cost=1.04..1.04 rows=1 width=4) (actual time=0.172..0.173 rows=1 loops=1) -> Seq Scan on search_order_reqt (cost=0.00..1.04 rows=1 width=4) (actual time=0.022..0.038 rows=1 loops=1) Filter: (search_id = 1) -> Hash IN Join (cost=1.04..7.27 rows=1 width=4) (actual time=2.064..2.117 rows=1 loops=429) Hash Cond: ("outer".product_id = "inner".product_id) -> Nested Loop (cost=0.00..6.19 rows=7 width=8) (actual time=1.112..2.081 rows=8 loops=429) -> Index Scan using fk_idx_staff_contract_2 on staff_contract c (cost=0.00..3.03 rows=1 width=4) (actual time=0.206..0.245 rows=1 loops=429) Index Cond: (staff_id = $8) Filter: ((avail_date_from <= '2004-06-12'::date) AND (avail_date_to >= '2004-06-18'::date)) -> Index Scan using fk_idx_staff_product on staff_product p (cost=0.00..3.08 rows=6 width=8) (actual time=0.873..1.764 rows=8 loops=429) Index Cond: (p.contract_id = "outer".contract_id) -> Hash (cost=1.04..1.04 rows=1 width=4) (actual time=0.086..0.086 rows=0 loops=1) -> Seq Scan on search_order_reqt (cost=0.00..1.04 rows=1 width=4) (actual time=0.037..0.050 rows=1 loops=1) Filter: (search_id = 1) -> Index Scan using location_pkey on "location" (cost=0.00..12.66 rows=1 width=18) (actual time=0.876..0.887 rows=1 loops=429) Index Cond: ("location".location_id = "outer".location_id) Filter: ((area_id = 1) OR (subplan)) SubPlan -> Index Scan using fk_idx_wruserarea, fk_idx_wruserarea on wruserarea (cost=3.26..9.64 rows=1 width=4) (never executed) Index Cond: ((area_id = 1) OR (area_id = $7)) Filter: (uid = $6) InitPlan -> Seq Scan on wruser (cost=0.00..3.26 rows=1 width=4) (never executed) Filter: ((username)::name = "current_user"()) SubPlan -> Aggregate (cost=11233.28..11233.29 rows=1 width=2) (actual time=3036.814..3036.815 rows=1 loops=429) -> Nested Loop (cost=10391.71..11233.21 rows=30 width=2) (actual time=2817.923..3036.516 rows=34 loops=429) -> Hash Join (cost=10391.71..11142.43 rows=30 width=4) (actual time=2813.349..3007.936 rows=34 loops=429) Hash Cond: ("outer".booking_id = "inner".booking_id) -> Index Scan using booking_plan_idx2 on booking_plan bp (cost=0.00..572.52 rows=23720 width=4) (actual time=0.070..157.028 rows=24613 loops=429) Index Cond: ((booking_date >= '2004-06-12'::date) AND (booking_date <= '2004-06-18'::date)) -> Hash (cost=10382.78..10382.78 rows=3571 width=8) (actual time=2746.122..2746.122 rows=0 loops=429) -> Index Scan using fk_idx_staff_booking on staff_booking b (cost=0.00..10382.78 rows=3571 width=8) (actual time=14.168..2733.315 rows=3815 loops=429) Index Cond: (contract_id = $0) -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..3.01 rows=1 width=6) (actual time=0.826..0.832 rows=1 loops=14401) Index Cond: ("outer".reqt_id = r.reqt_id) -> Aggregate (cost=363.94..363.94 rows=1 width=0) (actual time=0.057..0.058 rows=1 loops=429) -> Nested Loop (cost=0.00..363.94 rows=1 width=0) (actual time=0.034..0.034 rows=0 loops=429) -> Index Scan using fk_idx_main_order on main_order (cost=0.00..4.99 rows=1 width=4) (actual time=0.031..0.031 rows=0 loops=429) Index Cond: (client_id = 6) -> Index Scan using fk_idx_timesheet_detail_3 on timesheet_detail (cost=0.00..358.93 rows=1 width=4) (never executed) Index Cond: ("outer".order_id = timesheet_detail.order_id) Filter: (contract_id = $0) Total runtime: 1304591.861 ms Long Time! The main issue here is that the RHOURS subselect is executed as a nested join 429 times. unfortunately this is an expensive subquery. SQLServer executed this in just over 1 second on comparable hardware. Looking at its execution plan it flattens out the two subselects with a merge join. So I manually rewrote the query using derived tables and joins as: SELECT VS.*,VL.TEL1,SC.CONTRACT_ID,SC.CONTRACT_REF, SC.MAX_HOURS, SC.MIN_HOURS, TBOOK.RHOURS, TVIS.VISITS FROM SEARCH_REQT_RESULT SR JOIN STAFF_CONTRACT SC ON (SR.STAFF_ID = SC.STAFF_ID) AND SC.AVAIL_DATE_FROM <= '2004-06-12' AND SC.AVAIL_DATE_TO >= '2004-06-18' JOIN VSTAFF VS ON (VS.STAFF_ID = SC.STAFF_ID) JOIN VLOCATION VL ON (VL.LOCATION_ID = VS.LOCATION_ID) LEFT OUTER JOIN (SELECT B.CONTRACT_ID, SUM(R.DURATION+1)/60.0 AS RHOURS FROM STAFF_BOOKING B JOIN BOOKING_PLAN BP ON (BP.BOOKING_ID = B.BOOKING_ID) AND BP.BOOKING_DATE BETWEEN '2004-06-12' AND '2004-06-18' JOIN ORDER_REQT R ON (R.REQT_ID = B.REQT_ID) GROUP BY B.CONTRACT_ID) AS TBOOK ON (SC.CONTRACT_ID = TBOOK.CONTRACT_ID) LEFT OUTER JOIN (SELECT CONTRACT_ID,COUNT(*) AS VISITS FROM TIMESHEET_DETAIL JOIN MAIN_ORDER ON (MAIN_ORDER.ORDER_ID = TIMESHEET_DETAIL.ORDER_ID) WHERE MAIN_ORDER.CLIENT_ID = 6 GROUP BY CONTRACT_ID) AS TVIS ON (TVIS.CONTRACT_ID = SC.CONTRACT_ID) JOIN (SELECT P.CONTRACT_ID FROM STAFF_PRODUCT P, SEARCH_ORDER_REQT SR WHERE P.PRODUCT_ID = SR.PRODUCT_ID AND SR.SEARCH_ID = 1 GROUP BY P.CONTRACT_ID HAVING (COUNT(P.CONTRACT_ID) = (SELECT COUNT(DISTINCT PRODUCT_ID) FROM SEARCH_ORDER_REQT WHERE SEARCH_ID = 1))) AS TCONT ON (TCONT.CONTRACT_ID = SC.CONTRACT_ID) WHERE SR.SEARCH_ID = 1 With the explain analyze as: QUERY PLAN Hash Join (cost=137054.42..137079.74 rows=159 width=192) (actual time=6228.354..6255.058 rows=429 loops=1) Hash Cond: ("outer".contract_id = "inner".contract_id) InitPlan -> Index Scan using fk_idx_wruserarea on wruserarea (cost=3.26..6.52 rows=1 width=4) (actual time=0.850..0.850 rows=1 loops=1) Index Cond: (area_id = 1) Filter: (uid = $3) InitPlan -> Seq Scan on wruser (cost=0.00..3.26 rows=1 width=4) (actual time=0.670..0.675 rows=1 loops=1) Filter: ((username)::name = "current_user"()) -> Subquery Scan tcont (cost=152.63..161.81 rows=612 width=4) (actual time=36.312..42.268 rows=612 loops=1) -> HashAggregate (cost=152.63..155.69 rows=612 width=4) (actual time=36.301..40.040 rows=612 loops=1) Filter: (count(contract_id) = $7) InitPlan -> Aggregate (cost=1.04..1.04 rows=1 width=4) (actual time=0.107..0.108 rows=1 loops=1) -> Seq Scan on search_order_reqt (cost=0.00..1.04 rows=1 width=4) (actual time=0.025..0.037 rows=1 loops=1) Filter: (search_id = 1) -> Hash Join (cost=1.04..148.53 rows=612 width=4) (actual time=0.419..32.284 rows=612 loops=1) Hash Cond: ("outer".product_id = "inner".product_id) -> Seq Scan on staff_product p (cost=0.00..109.91 rows=6291 width=8) (actual time=0.117..17.943 rows=6291 loops=1) -> Hash (cost=1.04..1.04 rows=1 width=4) (actual time=0.190..0.190 rows=0 loops=1) -> Seq Scan on search_order_reqt sr (cost=0.00..1.04 rows=1 width=4) (actual time=0.165..0.177 rows=1 loops=1) Filter: (search_id = 1) -> Hash (cost=136894.61..136894.61 rows=266 width=192) (actual time=6191.923..6191.923 rows=0 loops=1) -> Merge Left Join (cost=136886.03..136894.61 rows=266 width=192) (actual time=6143.315..6189.685 rows=429 loops=1) Merge Cond: ("outer".contract_id = "inner".contract_id) -> Merge Left Join (cost=136517.64..136525.04 rows=266 width=184) (actual time=6142.896..6171.676 rows=429 loops=1) Merge Cond: ("outer".contract_id = "inner".contract_id) -> Sort (cost=5529.68..5530.34 rows=266 width=152) (actual time=129.548..130.027 rows=429 loops=1) Sort Key: sc.contract_id -> Nested Loop (cost=88.35..5518.96 rows=266 width=152) (actual time=33.213..121.666 rows=429 loops=1) -> Hash Join (cost=88.35..143.88 rows=424 width=138) (actual time=32.739..76.357 rows=429 loops=1) Hash Cond: ("outer".staff_id = "inner".staff_id) -> Hash Join (cost=47.47..93.95 rows=430 width=109) (actual time=15.232..40.040 rows=429 loops=1) Hash Cond: ("outer".staff_id = "inner".staff_id) -> Seq Scan on staff (cost=34.61..66.48 rows=1030 width=105) (actual time=9.412..16.105 rows=1030 loops=1) Filter: ((hashed subplan) OR $4) SubPlan -> Seq Scan on staff_area (cost=10.73..33.38 rows=493 width=4) (actual time=8.380..8.380 rows=0 loops=1) Filter: ((hashed subplan) OR (area_id = 1)) SubPlan -> Seq Scan on wruserarea (cost=3.26..10.72 rows=5 width=4) (actual time=0.953..1.941 rows=1 loops=1) Filter: (uid = $0) InitPlan -> Seq Scan on wruser (cost=0.00..3.26 rows=1 width=4) (actual time=0.902..0.908 rows=1 loops=1) Filter: ((username)::name = "current_user"()) -> Hash (cost=11.79..11.79 rows=430 width=4) (actual time=5.670..5.670 rows=0 loops=1) -> Index Scan using fk_idx_search_reqt_result on search_reqt_result sr (cost=0.00..11.79 rows=430 width=4) (actual time=0.448..4.516 rows=429 loops=1) Index Cond: (search_id = 1) -> Hash (cost=38.36..38.36 rows=1008 width=37) (actual time=17.386..17.386 rows=0 loops=1) -> Seq Scan on staff_contract sc (cost=0.00..38.36 rows=1008 width=37) (actual time=0.222..14.063 rows=1008 loops=1) Filter: ((avail_date_from <= '2004-06-12'::date) AND (avail_date_to >= '2004-06-18'::date)) -> Index Scan using location_pkey on "location" (cost=0.00..12.66 rows=1 width=18) (actual time=0.043..0.050 rows=1 loops=429) Index Cond: ("location".location_id = "outer".location_id) Filter: ((area_id = 1) OR (subplan)) SubPlan -> Index Scan using fk_idx_wruserarea, fk_idx_wruserarea on wruserarea (cost=3.26..9.64 rows=1 width=4) (never executed) Index Cond: ((area_id = 1) OR (area_id = $6)) Filter: (uid = $5) InitPlan -> Seq Scan on wruser (cost=0.00..3.26 rows=1 width=4) (never executed) Filter: ((username)::name = "current_user"()) -> Sort (cost=130987.97..130989.96 rows=797 width=36) (actual time=6013.254..6014.112 rows=746 loops=1) Sort Key: tbook.contract_id -> Subquery Scan tbook (cost=130933.62..130949.56 rows=797 width=36) (actual time=5993.070..6007.677 rows=746 loops=1) -> HashAggregate (cost=130933.62..130941.59 rows=797 width=6) (actual time=5993.055..6004.099 rows=746 loops=1) -> Merge Join (cost=74214.90..130815.02 rows=23720 width=6) (actual time=4950.951..5807.985 rows=24613 loops=1) Merge Cond: ("outer".reqt_id = "inner".reqt_id) -> Index Scan using order_reqt_pkey on order_reqt r (cost=0.00..50734.20 rows=2206291 width=6) (actual time=0.444..2753.374 rows=447439 loops=1) -> Sort (cost=74214.90..74274.20 rows=23720 width=8) (actual time=1822.405..1856.081 rows=24613 loops=1) Sort Key: b.reqt_id -> Nested Loop (cost=0.00..72491.19 rows=23720 width=8) (actual time=1.955..1633.124 rows=24613 loops=1) -> Index Scan using booking_plan_idx2 on booking_plan bp (cost=0.00..572.52 rows=23720 width=4) (actual time=1.468..243.827 rows=24613 loops=1) Index Cond: ((booking_date >= '2004-06- 12'::date) AND (booking_date <= '2004-06-18'::date)) -> Index Scan using staff_booking_pkey on staff_booking b (cost=0.00..3.02 rows=1 width=12) (actual time=0.037..0.042 rows=1 loops=24613) Index Cond: ("outer".booking_id = b.booking_id) -> Sort (cost=368.38..368.55 rows=68 width=12) (actual time=0.338..0.338 rows=0 loops=1) Sort Key: tvis.contract_id -> Subquery Scan tvis (cost=365.46..366.31 rows=68 width=12) (actual time=0.307..0.307 rows=0 loops=1) -> HashAggregate (cost=365.46..365.63 rows=68 width=4) (actual time=0.302..0.302 rows=0 loops=1) -> Nested Loop (cost=0.00..365.12 rows=68 width=4) (actual time=0.290..0.290 rows=0 loops=1) -> Index Scan using fk_idx_main_order on main_order (cost=0.00..4.99 rows=1 width=4) (actual time=0.286..0.286 rows=0 loops=1) Index Cond: (client_id = 6) -> Index Scan using fk_idx_timesheet_detail_3 on timesheet_detail (cost=0.00..358.63 rows=120 width=8) (never executed) Index Cond: ("outer".order_id = timesheet_detail.order_id) Total runtime: 6266.205 ms This now gives me the same results, but with orders of magnitude better execution times! Oddly enough, SQLServer really struggles with the second query, taking longer then PostgreSQL!!!! Regards, Gary. On 3 Apr 2004 at 10:59, Josh Berkus wrote: Gary, > There are no indexes on the columns involved in the update, they are > not required for my usual select statements. This is an attempt to > slightly denormalise the design to get the performance up comparable > to SQL Server 2000. We hope to move some of our databases over to > PostgreSQL later in the year and this is part of the ongoing testing. > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > so I am hand optimising some of the more frequently used > SQL and/or tweaking the database design slightly. Hmmm ... that hasn't been my general experience on complex queries. However, it may be due to a difference in ANALYZE statistics. I'd love to see you increase your default_stats_target, re-analyze, and see if PostgreSQL gets "smarter". -- -Josh Berkus Aglio Database Solutions San Francisco ---------------------------(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 -- Incoming mail is certified Virus Free. Checked by AVG Anti-Virus (http://www.grisoft.com). Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 From pgsql-performance-owner@postgresql.org Sat Apr 3 18:34:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 56A8AD1ED6D for ; Sat, 3 Apr 2004 22:34:34 +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 76300-02 for ; Sat, 3 Apr 2004 18:34:48 -0400 (AST) Received: from artemis.le.ac.uk (mailsend.le.ac.uk [143.210.16.126]) by svr1.postgresql.org (Postfix) with ESMTP id E04DED1ED69 for ; Sat, 3 Apr 2004 18:34:31 -0400 (AST) Received: from [143.210.36.58] (helo=mail.star.le.ac.uk) by artemis.le.ac.uk with smtp (Exim 4.30) id 1B9tiz-0004iN-Sn for pgsql-performance@postgresql.org; Sat, 03 Apr 2004 23:34:49 +0100 Received: (qmail 11789 invoked from network); 3 Apr 2004 22:35:11 -0000 Received: from sparky.star.le.ac.uk (143.210.36.10) by mail.star.le.ac.uk with SMTP; 3 Apr 2004 22:35:11 -0000 Date: Sat, 3 Apr 2004 23:35:11 +0100 (BST) From: Clive Page To: pgsql-performance@postgresql.org Subject: Re: Spatial join insists on sequential scan of larger In-Reply-To: <27943.1080920786@sss.pgh.pa.us> 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: 200404/21 X-Sequence-Number: 6321 On Fri, 2 Apr 2004, Tom Lane wrote: > Could we see EXPLAIN ANALYZE output? The original EXPLAIN output was: QUERY PLAN ---------------------------------------------------------------------------------- Nested Loop (cost=0.00..196642756520.34 rows=49506496044 width=32) -> Seq Scan on twomass t (cost=0.00..9560002.72 rows=177023872 width=48) -> Index Scan using xmm1box on xmm1 x (cost=0.00..1107.28 rows=280 width=48) Index Cond: (x.errbox && "outer".errbox) The EXPLAIN ANALYZE query was: explain analyze SELECT x.ra AS xra, x.decl AS xdecl, t.ra AS tra, t.decl AS tdecl INTO tempjoin FROM xmm1 AS x INNER JOIN twomass AS t ON x.errbox && t.errbox; And this produced: \timing Timing is on. dw=# \i join1.sql QUERY PLAN -------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=0.00..196642756520.34 rows=49506496044 width=32) (actual time=701.919..7796111.624 rows=1513 loops=1) -> Seq Scan on twomass t (cost=0.00..9560002.72 rows=177023872 width=48) (actual time=22.064..617462.486 rows=177757299 loops=1) -> Index Scan using xmmbox on xmm1 x (cost=0.00..1107.28 rows=280 width=48) (actual time=0.036..0.036 rows=0 loops=177757299) Index Cond: (x.errbox && "outer".errbox) Total runtime: 7796410.533 ms (5 rows) Time: 7796996.093 ms -- Clive Page Dept of Physics & Astronomy, University of Leicester, Leicester, LE1 7RH, U.K. From pgsql-performance-owner@postgresql.org Sat Apr 3 18:43:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B022ED1CA77 for ; Sat, 3 Apr 2004 22:43:22 +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 76300-06 for ; Sat, 3 Apr 2004 18:43:37 -0400 (AST) Received: from hotmail.com (law10-oe20.law10.hotmail.com [64.4.14.124]) by svr1.postgresql.org (Postfix) with ESMTP id BB8B4D1B4BB for ; Sat, 3 Apr 2004 18:43:20 -0400 (AST) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Sat, 3 Apr 2004 14:43:39 -0800 Received: from 67.81.102.201 by law10-oe20.law10.hotmail.com with DAV; Sat, 03 Apr 2004 22:43:39 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman2@hotmail.com] X-Sender: awerman2@hotmail.com From: "Aaron Werman" To: , "Gary Doades" , References: <406EA51B.21218.D2DE061@localhost> <200404031059.39268.josh@agliodbs.com> Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Sat, 3 Apr 2004 17:43:52 -0500 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 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Message-ID: X-OriginalArrivalTime: 03 Apr 2004 22:43:39.0389 (UTC) FILETIME=[1B84BAD0:01C419CD] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/22 X-Sequence-Number: 6322 Almost any cross dbms migration shows a drop in performance. The engine effectively trains developers and administrators in what works and what doesn't. The initial migration thus compares a tuned to an untuned version. /Aaron ----- Original Message ----- From: "Josh Berkus" To: "Gary Doades" ; Sent: Saturday, April 03, 2004 1:59 PM Subject: Re: [PERFORM] PostgreSQL and Linux 2.6 kernel. > Gary, > > > There are no indexes on the columns involved in the update, they are > > not required for my usual select statements. This is an attempt to > > slightly denormalise the design to get the performance up comparable > > to SQL Server 2000. We hope to move some of our databases over to > > PostgreSQL later in the year and this is part of the ongoing testing. > > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > > so I am hand optimising some of the more frequently used > > SQL and/or tweaking the database design slightly. > > Hmmm ... that hasn't been my general experience on complex queries. However, > it may be due to a difference in ANALYZE statistics. I'd love to see you > increase your default_stats_target, re-analyze, and see if PostgreSQL gets > "smarter". > > -- > -Josh Berkus > Aglio Database Solutions > San Francisco > > > ---------------------------(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 Sat Apr 3 19:52:37 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DC9CFD1EA01 for ; Sat, 3 Apr 2004 23:52:34 +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 90994-10 for ; Sat, 3 Apr 2004 19:52:49 -0400 (AST) Received: from jinx.internetstaff.com (unknown [63.214.174.240]) by svr1.postgresql.org (Postfix) with ESMTP id 276E7D1E9B0 for ; Sat, 3 Apr 2004 19:52:32 -0400 (AST) Received: from localhost (localhost [127.0.0.1]) by jinx.internetstaff.com (Postfix) with ESMTP id 621034BC0BC for ; Sat, 3 Apr 2004 16:52:50 -0700 (MST) Received: from jinx.internetstaff.com ([127.0.0.1]) by localhost (jinx [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 28839-03 for ; Sat, 3 Apr 2004 16:52:50 -0700 (MST) Subject: Re: PostgreSQL and Linux 2.6 kernel. From: Cott Lang To: pgsql-performance@postgresql.org In-Reply-To: <406EA51B.21218.D2DE061@localhost> References: <406EA51B.21218.D2DE061@localhost> Content-Type: text/plain Message-Id: <1081036369.11561.1.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 (1.4.6-1) Date: Sat, 03 Apr 2004 16:52:49 -0700 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at internetstaff.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/23 X-Sequence-Number: 6323 On Sat, 2004-04-03 at 03:50, Gary Doades wrote: > On 2 Apr 2004 at 22:36, pgsql-performance@postgresql. wrote: > > OK, some more detail: > > Before wiping 2.4 off my test box for the second time: Perhaps I missed it, but which io scheduler are you using under 2.6? From pgsql-performance-owner@postgresql.org Sat Apr 3 23:27:02 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D6FCBD1CA77 for ; Sun, 4 Apr 2004 03:27:00 +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 43753-03 for ; Sat, 3 Apr 2004 23:27:16 -0400 (AST) Received: from gw.tssi.com (gw.tssi.com [198.147.197.1]) by svr1.postgresql.org (Postfix) with ESMTP id 64292D1ED2F for ; Sat, 3 Apr 2004 23:26:57 -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 i343Nx8S002749; Sat, 3 Apr 2004 21:23:59 -0600 Received: (from nolan@localhost) by gw.tssi.com (8.12.6/8.12.6/Submit) id i343Nw3Z002747; Sat, 3 Apr 2004 21:23:58 -0600 From: Mike Nolan Message-Id: <200404040323.i343Nw3Z002747@gw.tssi.com> Subject: Re: PostgreSQL and Linux 2.6 kernel. To: awerman2@hotmail.com (Aaron Werman) Date: Sat, 3 Apr 2004 21:23:57 -0600 (CST) Cc: josh@agliodbs.com, gpd@gpdnet.co.uk (Gary Doades), pgsql-performance@postgresql.org In-Reply-To: from "Aaron Werman" at Apr 03, 2004 05:43:52 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: 200404/24 X-Sequence-Number: 6324 > Almost any cross dbms migration shows a drop in performance. The engine > effectively trains developers and administrators in what works and what > doesn't. The initial migration thus compares a tuned to an untuned version. I think it is also possible that Microsoft has more programmers working on tuning issues for SQL Server than PostgreSQL has working on the whole project. -- Mike Nolan From pgsql-performance-owner@postgresql.org Sun Apr 4 05:13:48 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 85F43D1B8B6 for ; Sun, 4 Apr 2004 08:13:46 +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 08505-02 for ; Sun, 4 Apr 2004 05:14:03 -0300 (ADT) Received: from mail.networkprograms.com (unknown [203.190.139.254]) by svr1.postgresql.org (Postfix) with ESMTP id 7F11AD1E9B8 for ; Sun, 4 Apr 2004 05:13:40 -0300 (ADT) Received: from KAMALR ([192.9.203.78]) by mail.networkprograms.com (8.12.9/8.12.9) with SMTP id i348H1bB028241 for ; Sun, 4 Apr 2004 13:47:01 +0530 Message-ID: <02ad01c41a1d$4255a7a0$4ecb09c0@KAMALR> Reply-To: "Kamalraj Singh Madhan" From: "Kamalraj Singh Madhan" To: References: <200404040323.i343Nw3Z002747@gw.tssi.com> Subject: Substitute for this oracle query in postGre Date: Sun, 4 Apr 2004 13:47:24 +0530 MIME-Version: 1.0 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 2.40 X-Virus-Scanned: by amavisd-new at postgresql.org Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="iso-8859-1" Content-Disposition: inline X-Archive-Number: 200404/25 X-Sequence-Number: 6325 Hi Friends, Does anybody know the substitute of the oracle function 'connect by prior' in postgre sql. The query is basically being used to get a tree structure of records. The query in oracle is :- select pkmsgid from mstmessage connect by prior pkmsgid = msgparentid start with msgparentid = 1 Kindly suggest. regards Kamal ********************************************************************* Network Programs is a SEI CMM Level 5 Certified Company ******************************************************************** The information contained in this communication (including any attachments) is intended solely for the use of the individual or entity to whom it is addressed and others authorized to receive it. It may contain confidential or legally privileged information. If you are not the intended recipient you are hereby notified that any disclosure, copying, distribution or taking any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you have received this communication in error, please notify us immediately by responding to this email and then delete it from your system. Network Programs (India) Limited is neither liable for the proper and complete transmission of the information contained in this communication nor for any delay in its receipt. ********************************************************************* From pgsql-performance-owner@postgresql.org Sun Apr 4 05:45:21 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 012D9D1EA65 for ; Sun, 4 Apr 2004 08:45:21 +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 12296-07 for ; Sun, 4 Apr 2004 05:45:37 -0300 (ADT) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 83F04D1EA01 for ; Sun, 4 Apr 2004 05:45:17 -0300 (ADT) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001EC1; 4 Apr 2004 09:49:38 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 4 Apr 2004 09:49:30 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001EC0; 4 Apr 2004 09:49:24 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Sun, 04 Apr 2004 09:49:23 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <406FDA23.19934.24163F4@localhost> In-reply-to: X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/26 X-Sequence-Number: 6326 Possibly. A lot of my queries show comparable performance, some a little slower and a few a little faster. There are a few, however, that really grind on PostgreSQL. I am leaning patterns from these to try and and target the most likely performance problems to come and hand tune these types of SQL. I'm not complaining about PostgreSQL or saying that SQLServer is better, in most cases it is not. SQLServer seems to be more predictable and forgiving in performance which tends to make for lazy SQL programming. It also has implications when the SQL is dynamically created based on user input, there are more chances of PostgreSQL hitting a performance problem than SQLServer. Overall I'm still very impressed with PostgreSQL. Given the $7000 per processor licence for SQLServer makes the case for PostgreSQL even stronger! Cheers, Gary. On 3 Apr 2004 at 17:43, Aaron Werman wrote: Almost any cross dbms migration shows a drop in performance. The engine effectively trains developers and administrators in what works and what doesn't. The initial migration thus compares a tuned to an untuned version. /Aaron ----- Original Message ----- From: "Josh Berkus" To: "Gary Doades" ; Sent: Saturday, April 03, 2004 1:59 PM Subject: Re: [PERFORM] PostgreSQL and Linux 2.6 kernel. > Gary, > > > There are no indexes on the columns involved in the update, they are > > not required for my usual select statements. This is an attempt to > > slightly denormalise the design to get the performance up comparable > > to SQL Server 2000. We hope to move some of our databases over to > > PostgreSQL later in the year and this is part of the ongoing testing. > > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > > so I am hand optimising some of the more frequently used > > SQL and/or tweaking the database design slightly. > > Hmmm ... that hasn't been my general experience on complex queries. However, > it may be due to a difference in ANALYZE statistics. I'd love to see you > increase your default_stats_target, re-analyze, and see if PostgreSQL gets > "smarter". > > -- > -Josh Berkus > Aglio Database Solutions > San Francisco > > > ---------------------------(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 6: Have you searched our list archives? http://archives.postgresql.org -- Incoming mail is certified Virus Free. Checked by AVG Anti-Virus (http://www.grisoft.com). Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 From pgsql-performance-owner@postgresql.org Sun Apr 4 06:06:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F212ED1D2CE for ; Sun, 4 Apr 2004 09:06:57 +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 18031-06 for ; Sun, 4 Apr 2004 06:07:13 -0300 (ADT) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 59A5DD1BB89 for ; Sun, 4 Apr 2004 06:06:53 -0300 (ADT) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001ECC; 4 Apr 2004 10:11:14 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 4 Apr 2004 10:11:00 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001ECB; 4 Apr 2004 10:11:00 +0100 Resent-From: "Gary Doades" Resent-To: pgsql-performance@postgresql.org Resent-date: Sun, 04 Apr 2004 10:10:59 +0100 X-cs: R From: Gary Doades X-RS-ID: X-RS-Flags: 0,0,1,1,0,0,0 X-RS-Header: In-reply-to: <200404040323.i343Nw3Z002747@gw.tssi.com> X-RS-Header: References: from "Aaron Werman" at Apr 03, 2004 05:43:52 PM X-RS-Sigset: 0 To: Mike Nolan Subject: Re: PostgreSQL and Linux 2.6 kernel. MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: 8BIT Date: Sun, 04 Apr 2004 09:52:40 +0100 Message-ID: Resent-Message-Id: <20040404090653.59A5DD1BB89@svr1.postgresql.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/29 X-Sequence-Number: 6329 On 3 Apr 2004 at 21:23, Mike Nolan wrote: > > Almost any cross dbms migration shows a drop in performance. The engine > > effectively trains developers and administrators in what works and what > > doesn't. The initial migration thus compares a tuned to an untuned version. > > I think it is also possible that Microsoft has more programmers working > on tuning issues for SQL Server than PostgreSQL has working on the > whole project. > -- > Mike Nolan > Agreed. Also considering the high price of SQLServer it is in their interests to spend a lot of resources on tuning/performance to give it a commercial edge over it rivals and in silly benchmark scores. Cheers, Gary. > -- > Incoming mail is certified Virus Free. > Checked by AVG Anti-Virus (http://www.grisoft.com). > Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 > From pgsql-performance-owner@postgresql.org Sun Apr 4 05:52:33 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2646FD1BB89 for ; Sun, 4 Apr 2004 08:52:32 +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 17233-01 for ; Sun, 4 Apr 2004 05:52:49 -0300 (ADT) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 4DC0BD1EAC4 for ; Sun, 4 Apr 2004 05:52:29 -0300 (ADT) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001EC6; 4 Apr 2004 09:56:50 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 4 Apr 2004 09:56:40 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001EC5; 4 Apr 2004 09:56:34 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Sun, 04 Apr 2004 09:56:34 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <406FDBD2.9910.247F995@localhost> In-reply-to: <1081036369.11561.1.camel@localhost.localdomain> References: <406EA51B.21218.D2DE061@localhost> X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/27 X-Sequence-Number: 6327 Unfortunately I don't understand the question! My background is the primarily Win32. The last time I used a *nix OS was about 20 years ago apart from occasional dips into the linux OS over the past few years. If you can tell be how to find out what you want I will gladly give you the information. Regards, Gary. On 3 Apr 2004 at 16:52, Cott Lang wrote: > On Sat, 2004-04-03 at 03:50, Gary Doades wrote: > > On 2 Apr 2004 at 22:36, pgsql-performance@postgresql. wrote: > > > > OK, some more detail: > > > > Before wiping 2.4 off my test box for the second time: > > Perhaps I missed it, but which io scheduler are you using under 2.6? > > > > ---------------------------(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 > > > -- > Incoming mail is certified Virus Free. > Checked by AVG Anti-Virus (http://www.grisoft.com). > Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 > From pgsql-performance-owner@postgresql.org Sun Apr 4 05:56:17 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2114AD1EB92 for ; Sun, 4 Apr 2004 08:56:16 +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 12736-10 for ; Sun, 4 Apr 2004 05:56:33 -0300 (ADT) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 588F5D1EA91 for ; Sun, 4 Apr 2004 05:56:13 -0300 (ADT) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001EC9; 4 Apr 2004 10:00:34 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 4 Apr 2004 10:00:21 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001EC8; 4 Apr 2004 10:00:10 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Sun, 04 Apr 2004 10:00:10 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <406FDCAA.30459.24B450F@localhost> In-reply-to: X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/28 X-Sequence-Number: 6328 Unfortunately I have to try and keep both SQLServer and PostgreSQL compatibilty. Our main web application is currently SQLServer, but we want to migrate customers who don't care what the DB server is over to PostgreSQL. Some of our larger customers demand SQLServer, you know how it is! I don't want to maintain two sets of code or SQL, so I am trying to find common ground. The code is not a problem, but the SQL sometimes is. Cheers, Gary. On 3 Apr 2004 at 17:43, Aaron Werman wrote: > Almost any cross dbms migration shows a drop in performance. The engine > effectively trains developers and administrators in what works and what > doesn't. The initial migration thus compares a tuned to an untuned version. > > /Aaron > > ----- Original Message ----- > From: "Josh Berkus" > To: "Gary Doades" ; > Sent: Saturday, April 03, 2004 1:59 PM > Subject: Re: [PERFORM] PostgreSQL and Linux 2.6 kernel. > > > > Gary, > > > > > There are no indexes on the columns involved in the update, they are > > > not required for my usual select statements. This is an attempt to > > > slightly denormalise the design to get the performance up comparable > > > to SQL Server 2000. We hope to move some of our databases over to > > > PostgreSQL later in the year and this is part of the ongoing testing. > > > SQLServer's query optimiser is a bit smarter that PostgreSQL's (yet) > > > so I am hand optimising some of the more frequently used > > > SQL and/or tweaking the database design slightly. > > > > Hmmm ... that hasn't been my general experience on complex queries. > However, > > it may be due to a difference in ANALYZE statistics. I'd love to see you > > increase your default_stats_target, re-analyze, and see if PostgreSQL gets > > "smarter". > > > > -- > > -Josh Berkus > > Aglio Database Solutions > > San Francisco > > > > > > ---------------------------(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 > > > > > -- > Incoming mail is certified Virus Free. > Checked by AVG Anti-Virus (http://www.grisoft.com). > Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 > From pgsql-performance-owner@postgresql.org Sun Apr 4 06:44:23 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D5368D1ED14 for ; Sun, 4 Apr 2004 09:44:22 +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 25430-09 for ; Sun, 4 Apr 2004 06:44:41 -0300 (ADT) 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 EA32ED1ED0B for ; Sun, 4 Apr 2004 06:44:20 -0300 (ADT) Received: from tmsl-adsl.demon.co.uk ([80.177.114.181] helo=bacon.tmsl.demon.co.uk) by anchor-post-34.mail.demon.net with esmtp (Exim 3.35 #1) id 1BA4BI-0007bQ-0Y; Sun, 04 Apr 2004 10:44:44 +0100 Date: Sun, 4 Apr 2004 10:41:33 +0100 From: Paul Thomas To: Gary Doades Cc: "pgsql-performance @ postgresql . org" Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <20040404104133.A1704@bacon> References: <406EA51B.21218.D2DE061@localhost> <1081036369.11561.1.camel@localhost.localdomain> <406FDBD2.9910.247F995@localhost> Mime-Version: 1.0 Content-Type: text/plain; format=flowed; charset=ISO-8859-1 In-Reply-To: <406FDBD2.9910.247F995@localhost>; from gpd@gpdnet.co.uk on Sun, Apr 04, 2004 at 09:56:34 +0100 X-Mailer: Balsa 1.2.3 Lines: 30 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/30 X-Sequence-Number: 6330 On 04/04/2004 09:56 Gary Doades wrote: > Unfortunately I don't understand the question! > > My background is the primarily Win32. The last time I used a *nix OS > was about 20 years ago apart from occasional dips into the linux OS > over the past few years. If you can tell be how to find out what you want > > I will gladly give you the information. Googling threw up http://spider.tm/apr2004/cstory2.html Interesting and possibly relevant quote: "Benchmarks have shown that in certain conditions the anticipatory algorithm is almost 10 times faster than what 2.4 kernel supports". HTH -- Paul Thomas +------------------------------+---------------------------------------------+ | Thomas Micro Systems Limited | Software Solutions for Business | | Computer Consultants | http://www.thomas-micro-systems-ltd.co.uk | +------------------------------+---------------------------------------------+ From pgsql-performance-owner@postgresql.org Sun Apr 4 07:14:37 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7065FD1ED27 for ; Sun, 4 Apr 2004 10:14:36 +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 31706-06 for ; Sun, 4 Apr 2004 07:14:54 -0300 (ADT) Received: from relay01.kbs.net.au (relay01.kbs.net.au [203.220.32.149]) by svr1.postgresql.org (Postfix) with ESMTP id 337A2D1ED20 for ; Sun, 4 Apr 2004 07:14:33 -0300 (ADT) Received: from [203.221.247.226] (helo=familyhealth.com.au) by relay01.kbs.net.au with esmtp (Exim 3.36 #2) id 1BA4eR-0002IX-00; Sun, 04 Apr 2004 20:14:52 +1000 Message-ID: <406FE017.1010602@familyhealth.com.au> Date: Sun, 04 Apr 2004 18:14:47 +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: Kamalraj Singh Madhan Cc: pgsql-performance@postgresql.org Subject: Re: Substitute for this oracle query in postGre References: <200404040323.i343Nw3Z002747@gw.tssi.com> <02ad01c41a1d$4255a7a0$4ecb09c0@KAMALR> In-Reply-To: <02ad01c41a1d$4255a7a0$4ecb09c0@KAMALR> 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: 200404/31 X-Sequence-Number: 6331 Hi, Try looking at the contrib/tablefunc add-in module. Chris Kamalraj Singh Madhan wrote: > Hi Friends, > Does anybody know the substitute of the oracle function 'connect by > prior' in postgre sql. > The query is basically being used to get a tree structure of records. The > query in oracle is :- > > select pkmsgid > from mstmessage > connect by prior pkmsgid = msgparentid > start with msgparentid = 1 > > Kindly suggest. > > regards > Kamal > > > > ********************************************************************* > Network Programs is a SEI CMM Level 5 Certified Company > ******************************************************************** > The information contained in this communication (including any attachments) is > intended solely for the use of the individual or entity to whom it is addressed > and others authorized to receive it. It may contain confidential or legally > privileged information. If you are not the intended recipient you are hereby > notified that any disclosure, copying, distribution or taking any action in > reliance on the contents of this information is strictly prohibited and may be > unlawful. If you have received this communication in error, please notify us > immediately by responding to this email and then delete it from your system. > Network Programs (India) Limited is neither liable for the proper and complete > transmission of the information contained in this communication nor for any > delay in its receipt. > ********************************************************************* > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org From pgsql-performance-owner@postgresql.org Sun Apr 4 10:04:14 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8BE58D1EBBD for ; Sun, 4 Apr 2004 13:04:13 +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 67843-02 for ; Sun, 4 Apr 2004 10:04:32 -0300 (ADT) Received: from jinx.internetstaff.com (unknown [63.214.174.240]) by svr1.postgresql.org (Postfix) with ESMTP id 7E108D1EA61 for ; Sun, 4 Apr 2004 10:04:11 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by jinx.internetstaff.com (Postfix) with ESMTP id 0A04D4BC0C6 for ; Sun, 4 Apr 2004 06:04:36 -0700 (MST) Received: from jinx.internetstaff.com ([127.0.0.1]) by localhost (jinx [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 04831-03-10 for ; Sun, 4 Apr 2004 06:04:35 -0700 (MST) Subject: Re: PostgreSQL and Linux 2.6 kernel. From: Cott Lang To: pgsql-performance@postgresql.org In-Reply-To: <406FDBD2.9910.247F995@localhost> References: <406EA51B.21218.D2DE061@localhost> <406FDBD2.9910.247F995@localhost> Content-Type: text/plain Message-Id: <1081083874.2172.3.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 (1.4.6-1) Date: Sun, 04 Apr 2004 06:04:35 -0700 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at internetstaff.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/32 X-Sequence-Number: 6332 On Sun, 2004-04-04 at 01:56, Gary Doades wrote: > Unfortunately I don't understand the question! > > My background is the primarily Win32. The last time I used a *nix OS > was about 20 years ago apart from occasional dips into the linux OS > over the past few years. If you can tell be how to find out what you want > I will gladly give you the information. There are two available io schedulers in 2.6 (new feature), deadline and anticipatory. It should show be listed in the boot messages: dmesg | grep scheduler I've seen people arguing for each of the two schedulers, saying one is better than the other for databases. I'm curious which one you're using. :) From pgsql-performance-owner@postgresql.org Sun Apr 4 11:46:21 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A053FD1ECF6 for ; Sun, 4 Apr 2004 14:46:20 +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 90328-04 for ; Sun, 4 Apr 2004 11:46:40 -0300 (ADT) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id 51DA3D1ECF2 for ; Sun, 4 Apr 2004 11:46:18 -0300 (ADT) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001EED; 4 Apr 2004 15:50:43 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 4 Apr 2004 15:50:33 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001EEC; 4 Apr 2004 15:50:23 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Sun, 04 Apr 2004 15:50:22 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <40702EBE.5594.38BE2E3@localhost> In-reply-to: <1081083874.2172.3.camel@localhost.localdomain> References: <406FDBD2.9910.247F995@localhost> X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/33 X-Sequence-Number: 6333 It says: Using anticipatory io scheduler. This then fits with the earlier post on other observations of up to 10 times better performance, which I what I was seeing in in certain circumstances. Cheers, Gary. On 4 Apr 2004 at 6:04, Cott Lang wrote: > On Sun, 2004-04-04 at 01:56, Gary Doades wrote: > > Unfortunately I don't understand the question! > > > > My background is the primarily Win32. The last time I used a *nix OS > > was about 20 years ago apart from occasional dips into the linux OS > > over the past few years. If you can tell be how to find out what you want > > I will gladly give you the information. > > There are two available io schedulers in 2.6 (new feature), deadline and > anticipatory. It should show be listed in the boot messages: > > dmesg | grep scheduler > > I've seen people arguing for each of the two schedulers, saying one is > better than the other for databases. I'm curious which one you're > using. :) > > > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org > > > -- > Incoming mail is certified Virus Free. > Checked by AVG Anti-Virus (http://www.grisoft.com). > Version: 7.0.230 / Virus Database: 262.6.5 - Release Date: 31/03/2004 > From pgsql-performance-owner@postgresql.org Sun Apr 4 18:05:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8E104D1CCAD for ; Sun, 4 Apr 2004 21:05:27 +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 75512-08 for ; Sun, 4 Apr 2004 18:05:49 -0300 (ADT) Received: from usea-naimss1.unisys.com (usea-naimss1.unisys.com [192.61.61.103]) by svr1.postgresql.org (Postfix) with ESMTP id 19FD5D1CA77 for ; Sun, 4 Apr 2004 18:05:25 -0300 (ADT) Received: from usea-nagw2.na.uis.unisys.com ([129.224.72.18]unverified) by usea-naimss1 with InterScan Messaging Security Suite; Sun, 04 Apr 2004 16:09:52 -0500 Received: from usea-nagw2.na.uis.unisys.com ([129.224.72.53]) by usea-nagw2.na.uis.unisys.com with Microsoft SMTPSVC(5.0.2195.6713); Sun, 4 Apr 2004 16:05:53 -0500 Received: from gbmk-eugw1.eu.uis.unisys.com ([129.221.133.28]) by usea-nagw2.na.uis.unisys.com with Microsoft SMTPSVC(5.0.2195.6713); Sun, 4 Apr 2004 16:05:52 -0500 content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-MimeOLE: Produced By Microsoft Exchange V6.0.6487.1 Subject: Re: single index on more than two coulumns a bad thing? Date: Sun, 4 Apr 2004 22:06:11 +0100 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] single index on more than two coulumns a bad thing? Thread-Index: AcQYq7LTjW8JZm4tREecA3JXJST1fAB3B71g From: "Leeuw van der, Tim" To: X-OriginalArrivalTime: 04 Apr 2004 21:05:52.0958 (UTC) FILETIME=[9D4425E0:01C41A88] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/34 X-Sequence-Number: 6334 Hi Aaron, > -----Original Message----- > From: pgsql-performance-owner@postgresql.org > [mailto:pgsql-performance-owner@postgresql.org]On Behalf Of=20 > Aaron Werman > Sent: vrijdag 2 april 2004 13:57 >=20 >=20 > another thing that I have all over the place is a hierarchy: > index on grandfather_table(grandfather) > index on father_table(grandfather, father) > index on son_table(grandfather, father, son) >=20 It depends on your data-distribution, but I find that in almost all cases i= t's beneficial to have your indexes the other way round in such cases: index on grandfather_table(grandfather) index on father_table(father, grandfather) index on son_table(son, father, grandfather) That usually gives a less common, more selective value at the start of the = index, making the initial selection in the index smaller. And AFAIK I don't have to rewrite my queries for that; the planner doesn't = care about the order of expressions in the query that are on the same level. That said, I tend to use 'surrogate keys'; keys generated from sequences or= auto-number columns for my tables. It makes the tables less readable, but = the indexes remain smaller. Greetings, --Tim From pgsql-performance-owner@postgresql.org Mon Apr 5 16:19:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E1F3FD1EB5E for ; Mon, 5 Apr 2004 02:08:33 +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 47416-06 for ; Sun, 4 Apr 2004 23:08:56 -0300 (ADT) Received: from hotmail.com (bay9-dav30.bay9.hotmail.com [64.4.46.87]) by svr1.postgresql.org (Postfix) with ESMTP id A7883D1E9AE for ; Sun, 4 Apr 2004 23:08:30 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Sun, 4 Apr 2004 19:09:02 -0700 Received: from 67.81.102.201 by bay9-dav30.bay9.hotmail.com with DAV; Mon, 05 Apr 2004 02:09:01 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman@hotmail.com] X-Sender: awerman@hotmail.com From: "Aaron Werman" To: "Leeuw van der, Tim" , References: Subject: Re: single index on more than two coulumns a bad thing? Date: Sun, 4 Apr 2004 22:09:21 -0400 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 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Message-ID: X-OriginalArrivalTime: 05 Apr 2004 02:09:02.0223 (UTC) FILETIME=[F6E9B9F0:01C41AB2] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/48 X-Sequence-Number: 6348 You're absolutely correct that the general rule is to lead a composite index with the highest cardinality index columns for fastest selectivity. Indices and all physical design are based on usage. In this case of unique indices supporting primary keys in a hierarchy, it depends. For selection of small sets of arbitrary rows, your arrangement is best. For hierarchy based queries, such as "for grandparent of foo, and parent of bar, give average age of sons" - the hierarchy based index is often more efficient. Surrogate keys have a role, and can improve performance, but also carry an enormous penalty of intentionally obfuscating logical keys and data semantics, and almost always lead to data errors not being caught because they obscure irrational relationships. I hate them, but use them frequently in high transaction rate operational systems where there is much functional validation outside the dbms (and the apps behave therefore like object databases and surrogate keys are network database pointers) and in data warehousing (where downstream data cannot be corrected anyway). /Aaron ----- Original Message ----- From: "Leeuw van der, Tim" To: Sent: Sunday, April 04, 2004 5:06 PM Subject: Re: [PERFORM] single index on more than two coulumns a bad thing? Hi Aaron, > -----Original Message----- > From: pgsql-performance-owner@postgresql.org > [mailto:pgsql-performance-owner@postgresql.org]On Behalf Of > Aaron Werman > Sent: vrijdag 2 april 2004 13:57 > > > another thing that I have all over the place is a hierarchy: > index on grandfather_table(grandfather) > index on father_table(grandfather, father) > index on son_table(grandfather, father, son) > It depends on your data-distribution, but I find that in almost all cases it's beneficial to have your indexes the other way round in such cases: index on grandfather_table(grandfather) index on father_table(father, grandfather) index on son_table(son, father, grandfather) That usually gives a less common, more selective value at the start of the index, making the initial selection in the index smaller. And AFAIK I don't have to rewrite my queries for that; the planner doesn't care about the order of expressions in the query that are on the same level. That said, I tend to use 'surrogate keys'; keys generated from sequences or auto-number columns for my tables. It makes the tables less readable, but the indexes remain smaller. Greetings, --Tim ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Mon Apr 5 12:42:02 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B5AAFD1B97B for ; Mon, 5 Apr 2004 15:39:36 +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 24326-01 for ; Mon, 5 Apr 2004 12:39:57 -0300 (ADT) Received: from ol-mailrelay-01.fh-oldenburg.de (ol-mailrelay-01.fh-oldenburg.de [139.13.186.99]) by svr1.postgresql.org (Postfix) with ESMTP id 6C479D1B535 for ; Mon, 5 Apr 2004 12:39:26 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by ol-mailrelay-01.fh-oldenburg.de (Postfix) with ESMTP id 7EDC438C3F6 for ; Mon, 5 Apr 2004 17:31:39 +0200 (CEST) Received: from ol-mailrelay-01.fh-oldenburg.de ([127.0.0.1]) by localhost (ol-mailrelay-01 [127.0.0.1]) (amavisd-new, port 10023) with ESMTP id 26538-03 for ; Mon, 5 Apr 2004 17:31:39 +0200 (CEST) Received: from webmail.fh-oldenburg.de (ol-mailrelay-01.fh-oldenburg.de [139.13.186.99]) by ol-mailrelay-01.fh-oldenburg.de (Postfix) with SMTP id 4DB9C38C3BB for ; Mon, 5 Apr 2004 17:31:39 +0200 (CEST) Received: from 195.243.253.146 (SquirrelMail authenticated user kehlenbh) by webmail.fh-oldenburg.de with HTTP; Mon, 5 Apr 2004 17:31:39 +0200 (CEST) Message-ID: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> Date: Mon, 5 Apr 2004 17:31:39 +0200 (CEST) Subject: performance comparission postgresql/ms-sql server From: "Heiko Kehlenbrink" To: pgsql-performance@postgresql.org User-Agent: SquirrelMail/1.4.2 MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 Importance: Normal X-Virus-Scanned: by amavisd-new X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/35 X-Sequence-Number: 6335 hi list, i want to convince people to use postgresql instead of ms-sql server, so i set up a kind of comparission insert data / select data from postgresql / ms-sql server the table i use was pretty basic, id bigserial dist float8 x float8 y float8 z float8 i filled the table with a function which filled x,y,z with incremental increasing values (1,2,3,4,5,6...) and computing from that the dist value for every tupel (sqrt((x*x)+(y*y)+(z*z))). this works fine for both dbms postgresql needs 13:37 min for 10.000.000 tupel, ms-sql needs 1:01:27 h for 10.000.000 tupel. so far so good. i attached an index on the dist row and started to query the dbs with scripts which select a serial row of 100.000,200.000,500.000 tupels based on the dist row. i randomizly compute the start and the end distance and made a "select avg(dist) from table where dist > startdist and dist < enddist" Did the same with a table with 50.000.000 tupel in ms-sql and postgres. the outcome so far: 100.000 from 50.000.000: postgres: 0.88 sec ms-sql: 0.38 sec 200.000 from 50.000.000: postgres: 1.57 sec ms-sql: 0.54 sec 500.000 from 50.000.000: postgres: 3.66 sec ms-sql: 1.18 sec i try a lot of changes to the postgresql.conf regarding "Tuning PostgreSQL for performance" by Shridhar Daithankar, Josh Berkus which did not make a big diffrence to the answering times from postgresql. i'm pretty fine with the insert time... do you have any hints like compiler-flags and so on to get the answering time from postgresql equal to ms-sql? (btw both dbms were running on exactly the same hardware) i use suse 8.1 postgresql 7.2 compiled from the rpms for using postgis, but that is anothe story... 1.5 gig ram 1.8 mhz intel cpu every help welcome best regards heiko From pgsql-performance-owner@postgresql.org Mon Apr 5 12:46:41 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 73A21D1BA93 for ; Mon, 5 Apr 2004 15:37:46 +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 17791-07 for ; Mon, 5 Apr 2004 12:38:14 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 1AC96D1BA90 for ; Mon, 5 Apr 2004 12:37:44 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4774963; Mon, 05 Apr 2004 08:39:40 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Mike Nolan , awerman2@hotmail.com (Aaron Werman) Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Mon, 5 Apr 2004 08:36:52 -0700 User-Agent: KMail/1.4.3 Cc: gpd@gpdnet.co.uk (Gary Doades), pgsql-performance@postgresql.org References: <200404040323.i343Nw3Z002747@gw.tssi.com> In-Reply-To: <200404040323.i343Nw3Z002747@gw.tssi.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404050836.52115.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/36 X-Sequence-Number: 6336 Mike, > I think it is also possible that Microsoft has more programmers working > on tuning issues for SQL Server than PostgreSQL has working on the > whole project. Ah, but quantity != quality. Or they wouldn't be trolling our mailing lists trying to hire PostgreSQL programmers for the SQL Server project (really!). And we had nearly 200 contributors between 7.3 and 7.4 ... a respectable development staff for even a large corporation. Point taken, though, SQL Server has done a better job in opitimizing for "dumb" queries. This is something that PostgreSQL needs to work on, as is self-referential updates for large tables, which also tend to be really slow. Mind you, in SQL Server 7 I used to be able to crash the server with a big self-referential update, so this is a common database problem. Unfortunately, these days only Tom and Neil seem to be seriously working on the query planner (beg pardon in advance if I've missed someone) so I think the real answer is that we need another person interested in this kind of optimization before it's going to get much better. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Mon Apr 5 12:53:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CA4F9D1D12E for ; Mon, 5 Apr 2004 15:53:44 +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 29651-02 for ; Mon, 5 Apr 2004 12:54:13 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 952FCD1CCCA for ; Mon, 5 Apr 2004 12:53:42 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4775049; Mon, 05 Apr 2004 08:55:40 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: "Heiko Kehlenbrink" , pgsql-performance@postgresql.org Subject: Re: performance comparission postgresql/ms-sql server Date: Mon, 5 Apr 2004 08:52:51 -0700 User-Agent: KMail/1.4.3 References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> In-Reply-To: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404050852.51709.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/37 X-Sequence-Number: 6337 Heiko, > 100.000 from 50.000.000: > > postgres: 0.88 sec > ms-sql: 0.38 sec > > 200.000 from 50.000.000: > > postgres: 1.57 sec > ms-sql: 0.54 sec > > 500.000 from 50.000.000: > > postgres: 3.66 sec > ms-sql: 1.18 sec Questions: 1. Is this the time to return *all rows* or just the first row? Given the different way that PostgreSQL fetches rows to the client from MSSQL, it makes a difference. 2. What are your sort-mem and shared-mem settings? 3. Have you tried clustering the table? 4. Have you done a comparison of selecting random or scattered, instead of serial rows? MSSQL has a tendency to physically store rows in "order" which gives it a certain advantage in this kind of query. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Mon Apr 5 13:08:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D5302D1CCCA for ; Mon, 5 Apr 2004 15:54:30 +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 26581-06 for ; Mon, 5 Apr 2004 12:54:45 -0300 (ADT) Received: from frodo.hserus.net (frodo.hserus.net [204.74.68.40]) by svr1.postgresql.org (Postfix) with ESMTP id C09C1D1C516 for ; Mon, 5 Apr 2004 12:54:14 -0300 (ADT) Received: from concord.pspl.co.in ([202.54.11.72]:64379 helo=frodo.hserus.net) by frodo.hserus.net with asmtp (Cipher TLSv1:DHE-RSA-AES256-SHA:256) (Exim 4.31 #0) id 1BAWR0-000Nty-GP by authid with plain; Mon, 05 Apr 2004 21:24:51 +0530 Message-ID: <40718146.1010506@frodo.hserus.net> Date: Mon, 05 Apr 2004 21:24:46 +0530 From: Shridhar Daithankar 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: Heiko Kehlenbrink Cc: pgsql-performance@postgresql.org Subject: Re: performance comparission postgresql/ms-sql server References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> In-Reply-To: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> 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: 200404/38 X-Sequence-Number: 6338 Heiko Kehlenbrink wrote: > hi list, > > i want to convince people to use postgresql instead of ms-sql server, so i > set up a kind of comparission insert data / select data from postgresql / > ms-sql server > > the table i use was pretty basic, > > id bigserial > dist float8 > x float8 > y float8 > z float8 > > i filled the table with a function which filled x,y,z with incremental > increasing values (1,2,3,4,5,6...) and computing from that the dist value > for every tupel (sqrt((x*x)+(y*y)+(z*z))). > > this works fine for both dbms > > postgresql needs 13:37 min for 10.000.000 tupel, > ms-sql needs 1:01:27 h for 10.000.000 tupel. > > so far so good. > > i attached an index on the dist row and started to query the dbs with > scripts which select a serial row of 100.000,200.000,500.000 tupels based > on the dist row. > i randomizly compute the start and the end distance and made a "select > avg(dist) from table where dist > startdist and dist < enddist" Some basics to check quickly. 1. vacuum analyze the table before you start selecting. 2. for slow running queries, check explain analyze output and find out who takes maximum time. 3. Check for typecasting. You need to typecast the query correctly e.g. select avg(dist) from table where dist >startdist::float8 and dist; Mon, 5 Apr 2004 16:11:01 +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 31183-07 for ; Mon, 5 Apr 2004 13:11:28 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id EB892D1BCB9 for ; Mon, 5 Apr 2004 13:10:57 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i35GBafl012486; Mon, 5 Apr 2004 12:11:36 -0400 (EDT) To: "Heiko Kehlenbrink" Cc: pgsql-performance@postgresql.org Subject: Re: performance comparission postgresql/ms-sql server In-reply-to: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> Comments: In-reply-to "Heiko Kehlenbrink" message dated "Mon, 05 Apr 2004 17:31:39 +0200" Date: Mon, 05 Apr 2004 12:11:35 -0400 Message-ID: <12485.1081181495@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/39 X-Sequence-Number: 6339 "Heiko Kehlenbrink" writes: > i use suse 8.1 > postgresql 7.2 compiled from the rpms for using postgis, but that is > anothe story... 7.4 might be a little quicker; but in any case you should be doing this sort of comparison using the current release, no? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 5 13:27:00 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0865AD1BBA2 for ; Mon, 5 Apr 2004 16:27:00 +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 38980-10 for ; Mon, 5 Apr 2004 13:27:27 -0300 (ADT) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 46098D1B560 for ; Mon, 5 Apr 2004 13:26:56 -0300 (ADT) Received: (qmail 37033 invoked from network); 5 Apr 2004 16:27:38 -0000 Received: from unknown (HELO ?192.168.0.4?) (10.10.10.177) by 10.10.10.10 with SMTP; 5 Apr 2004 16:27:38 -0000 Mime-Version: 1.0 (Apple Message framework v609) Content-Transfer-Encoding: 7bit Message-Id: <49B1B198-871E-11D8-B8C1-000393D1F76E@torgo.978.org> Content-Type: text/plain; charset=US-ASCII; format=flowed To: pgsql-performance@postgresql.org From: Jeff Subject: SETOF performance Date: Mon, 5 Apr 2004 12:28:36 -0400 X-Mailer: Apple Mail (2.609) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/40 X-Sequence-Number: 6340 I think it was on this list - someone posted a message about SETOF being slower. Tom replied saying it was because it needed to create an on-disk tuplestore. I was just looking for some clarification - a SETOF function will always write the reslting tuples to disk (Not buffering in say a sort_mem sized buffer)? I think if that is the case I may need to go back and change some stuff around. I have a procedure that I broke out a bit to make life easier. Basically it goes for v_row in select blah from function_that_gets_data_from_some_cache(....) rowcount := rowcount + 1; return next v_row; end for; if rowcount = 0 then [same thing, but we call some_function_that_creates_data_for_cache] end if; Doing it this way means I avoid having to deal with it in the client and I also avoid having a giant stored procedure. (I like short & sweet things) What I've found for timings is this: select * from function_that_gets_data_from_some_cache() runs around 1.8 ms but select * from the_top_level_function() runs around 4.2ms (Yes, I know 4.2 ms is fast, but that is not the point). could this overhead be related to the SETOF tuplestores? Might it be better to use refcursor or something or bite the bullet and live with a giant procedure? -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Mon Apr 5 13:56:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1FEB3D1E9CF for ; Mon, 5 Apr 2004 16:43:13 +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 50730-04 for ; Mon, 5 Apr 2004 13:43:41 -0300 (ADT) Received: from mail.osdl.org (fw.osdl.org [65.172.181.6]) by svr1.postgresql.org (Postfix) with ESMTP id B49FBD1E9C5 for ; Mon, 5 Apr 2004 13:43:09 -0300 (ADT) Received: from osdl.org (markw@ibm-b.pdx.osdl.net [172.20.1.51]) by mail.osdl.org (8.11.6/8.11.6) with ESMTP id i35Ghd200769; Mon, 5 Apr 2004 09:43:40 -0700 Message-Id: <200404051643.i35Ghd200769@mail.osdl.org> Date: Mon, 5 Apr 2004 09:43:36 -0700 (PDT) From: markw@osdl.org Subject: Re: PostgreSQL and Linux 2.6 kernel. To: cott@internetstaff.com Cc: pgsql-performance@postgresql.org In-Reply-To: <1081083874.2172.3.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: 200404/41 X-Sequence-Number: 6341 On 4 Apr, Cott Lang wrote: > On Sun, 2004-04-04 at 01:56, Gary Doades wrote: >> Unfortunately I don't understand the question! >> >> My background is the primarily Win32. The last time I used a *nix OS >> was about 20 years ago apart from occasional dips into the linux OS >> over the past few years. If you can tell be how to find out what you want >> I will gladly give you the information. > > There are two available io schedulers in 2.6 (new feature), deadline and > anticipatory. It should show be listed in the boot messages: > > dmesg | grep scheduler > > I've seen people arguing for each of the two schedulers, saying one is > better than the other for databases. I'm curious which one you're > using. :) Our database tests (TPC fair use implementations) show that the deadline scheduler has an edge on the anticipatory scheduler. Depending on the current state of the AS scheduler, it can be within a few percent to 10% or so. I have some data with one of our tests here: http://developer.osdl.org/markw/fs/dbt2_project_results.html Mark From pgsql-performance-owner@postgresql.org Mon Apr 5 14:05:53 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C2A44D1BABB for ; Mon, 5 Apr 2004 17:05:04 +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 62338-01 for ; Mon, 5 Apr 2004 14:05:31 -0300 (ADT) Received: from www.totalcardinc.com (unknown [64.33.232.186]) by svr1.postgresql.org (Postfix) with ESMTP id BAC39D1B441 for ; Mon, 5 Apr 2004 14:04:59 -0300 (ADT) Received: from RSCHWARZW2K ([10.250.0.37]) (authenticated bits=0) by www.totalcardinc.com (8.12.8/8.12.8) with ESMTP id i35H5bTN019089 for ; Mon, 5 Apr 2004 12:05:37 -0500 From: "Rosser Schwarz" To: Subject: Re: atrocious update performance Date: Mon, 5 Apr 2004 12:05:37 -0500 Message-ID: <009801c41b30$37872380$2500fa0a@CardServices.TCI.com> 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 10.0.4510 In-Reply-To: <004301c411be$74004650$2500fa0a@CardServices.TCI.com> Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/42 X-Sequence-Number: 6342 After deinstalling and scrubbing PostgreSQL from my server and doing a clean build using a vanilla 7.4.2 tree, I'm rather more confident that foreign key validation is at cause in my performance problems. I recreated my schemas and ran the original update, with foreign keys referring to the identity column of the target table. The update took roughly two days, as I'd predicted based on my analysis of the previous installation. (I can't say how long with certainty, beyond that it finished some time between when I left work one night and came in the next morning, the second day after starting the query.) I'm not sure what was wrong with the previous install, such that the update took several days; two-ish days is long enough. Just this morning, however, I created a copy of the target table (all 4.7M rows), with absolutely no foreign keys referring to it, and ran the update against the copy. That update took 2300 seconds. The join columns were indexed in both cases. I'm in the process of migrating the machine to run kernel 2.6.4, following the thread started by Gary, though I suspect that the kernel revision is moot with respect to whether or not foreign keys are being incorrectly validated. I can keep the 2.4 kernel and modules around to run using the current versions for testing purposes, though any such work would necessarily be off-hours. Please advise of anything I can do to help narrow down the specific cause of the issue; I know just enough C to be mildly dangerous. /rls -- Rosser Schwarz Total Card, Inc. From pgsql-performance-owner@postgresql.org Mon Apr 5 14:51:22 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 470C9D1BAA5 for ; Mon, 5 Apr 2004 17:49:59 +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 80197-07 for ; Mon, 5 Apr 2004 14:50:26 -0300 (ADT) Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82]) by svr1.postgresql.org (Postfix) with ESMTP id 1B5BFD1BABD for ; Mon, 5 Apr 2004 14:49:55 -0300 (ADT) Received: from [206.19.64.3] (account jconway HELO joeconway.com) by joeconway.com (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 1259946; Mon, 05 Apr 2004 10:45:08 -0700 Message-ID: <40719C31.5030505@joeconway.com> Date: Mon, 05 Apr 2004 10:49:37 -0700 From: Joe Conway User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Jeff Cc: pgsql-performance@postgresql.org Subject: Re: SETOF performance References: <49B1B198-871E-11D8-B8C1-000393D1F76E@torgo.978.org> In-Reply-To: <49B1B198-871E-11D8-B8C1-000393D1F76E@torgo.978.org> 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: 200404/43 X-Sequence-Number: 6343 Jeff wrote: > I think it was on this list - someone posted a message about SETOF > being slower. Tom replied saying it was because it needed to create an > on-disk tuplestore. > > I was just looking for some clarification - a SETOF function will always > write the reslting tuples to disk (Not buffering in say a sort_mem sized > buffer)? I think at least part of what you're seeing is normal function call overhead. As far as tuplestores writing to disk, here's what the source says: In src/backend/utils/sort/tuplestore.c 8<--------------------------------------- * maxKBytes: how much data to store in memory (any data beyond this * amount is paged to disk). When in doubt, use work_mem. */ Tuplestorestate * tuplestore_begin_heap(bool randomAccess, bool interXact, int maxKBytes) 8<--------------------------------------- In src/backend/executor/execQual.c:ExecMakeTableFunctionResult(): 8<--------------------------------------- tupstore = tuplestore_begin_heap(true, false, work_mem); 8<--------------------------------------- So up to work_mem (sort_mem in 7.4 and earlier) should be stored in memory. Joe From pgsql-performance-owner@postgresql.org Mon Apr 5 16:35:10 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 66118D1B441 for ; Mon, 5 Apr 2004 18:10:25 +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 83270-10 for ; Mon, 5 Apr 2004 15:10:54 -0300 (ADT) Received: from mail.ez3pl.com (66-136-75-131.ded.swbell.net [66.136.75.131]) by svr1.postgresql.org (Postfix) with ESMTP id DB169D1B54E for ; Mon, 5 Apr 2004 15:10:21 -0300 (ADT) Received: from kbarnard ([172.25.96.149]) by mail.ez3pl.com (8.11.2/8.11.2) with ESMTP id i35I9sL20799 for ; Mon, 5 Apr 2004 13:09:54 -0500 From: "Kevin Barnard" To: Date: Mon, 05 Apr 2004 13:04:52 -0500 MIME-Version: 1.0 Subject: Re: atrocious update performance Message-ID: <40715974.10626.ED9A003@localhost> In-reply-to: <009801c41b30$37872380$2500fa0a@CardServices.TCI.com> References: <004301c411be$74004650$2500fa0a@CardServices.TCI.com> X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/49 X-Sequence-Number: 6349 On 5 Apr 2004 at 12:05, Rosser Schwarz wrote: > Just this morning, however, I created a copy of the target table (all > 4.7M rows), with absolutely no foreign keys referring to it, and ran > the update against the copy. That update took 2300 seconds. The > join columns were indexed in both cases. Have you added indexes for the custid column for tables account.acct accunt.orgacct and note? I haven't followed the entire thread but it you have cascading FK on those tables without an index on the column that could cause your delay. Kevin Barnard SpeedFC From pgsql-performance-owner@postgresql.org Mon Apr 5 16:14:52 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id AB300D1BAA5 for ; Mon, 5 Apr 2004 18:11:27 +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 90570-02 for ; Mon, 5 Apr 2004 15:11:56 -0300 (ADT) Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243]) by svr1.postgresql.org (Postfix) with ESMTP id EC736D1BAF8 for ; Mon, 5 Apr 2004 15:11:24 -0300 (ADT) Received: from Spooler by gpdnet.co.uk (Mercury/32 v4.01a) ID MO001FDC; 5 Apr 2004 19:16:08 +0100 Received: from spooler by gpdnet.co.uk (Mercury/32 v4.01a); 5 Apr 2004 19:16:02 +0100 Received: from gary (192.168.1.2) by gpdnet.co.uk (Mercury/32 v4.01a) with ESMTP ID MG001FDB; 5 Apr 2004 19:15:53 +0100 From: "Gary Doades" To: pgsql-performance@postgresql.org Date: Mon, 05 Apr 2004 19:11:56 +0100 MIME-Version: 1.0 Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <4071AF7C.15265.52B583C@localhost> In-reply-to: <200404050836.52115.josh@agliodbs.com> References: <200404040323.i343Nw3Z002747@gw.tssi.com> X-mailer: Pegasus Mail for Windows (v4.12a) Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Content-description: Mail message body X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/47 X-Sequence-Number: 6347 On 5 Apr 2004 at 8:36, Josh Berkus wrote: > > Point taken, though, SQL Server has done a better job in opitimizing for > "dumb" queries. This is something that PostgreSQL needs to work on, as is > self-referential updates for large tables, which also tend to be really slow. > Mind you, in SQL Server 7 I used to be able to crash the server with a big > self-referential update, so this is a common database problem. > I agree about the "dumb" queries (I'm not mine are *that* dumb :) ) When you can write SQL that looks right, feels right, gives the right answers during testing and SQLServer runs them really fast, you stop there and tend not to tinker with the SQL further. You *can* (I certainly do) achieve comparable performance with PostgreSQL, but you just have to work harder for it. Now that I have learned the characteristics of both servers I can write SQL that is pretty good on both. I suspect that there are people who evaluate PostgreSQL by executing their favorite SQLSever queries against it, see that it is slower and never bother to go further. Cheers, Gary. From pgsql-admin-owner@postgresql.org Mon Apr 5 20:32:58 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EC4C0D1B88F for ; Mon, 5 Apr 2004 19:42:50 +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 31757-09 for ; Mon, 5 Apr 2004 16:43:19 -0300 (ADT) Received: from loki.globexplorer.com (unknown [208.35.14.101]) by svr1.postgresql.org (Postfix) with ESMTP id AD36BD1B837 for ; Mon, 5 Apr 2004 16:42:43 -0300 (ADT) X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0 content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="Windows-1252" Content-Transfer-Encoding: quoted-printable Subject: Re: Raw devices vs. Filesystems Date: Mon, 5 Apr 2004 12:43:21 -0700 Message-ID: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [ADMIN] Raw devices vs. Filesystems Thread-Index: AcQbPaTHRcbj4ePWROSnezctjwhSmwAB89Ar From: "Gregory S. Williamson" To: "Christopher Browne" , X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/75 X-Sequence-Number: 13048 No point to beating a dead horse (other than the sheer joy of the thing) si= nce postgres does not have raw device support, but ... raw devices, at least on solaris, are about 10 times as fast as cooked file= systems for Informix. This might still be a gain for postgres' performance= , but the portability issues remain. raw device use in Informix is safer in terms of data because Informix does = not ever have to use the regular file system and so issues of buffering and= so on go away. My understanding -- fortunately not ever tried in the real = world -- is that postgres' WAL log system is as reliable as Informix writin= g to raw devices. raw devices can't be copied or tampered with with regular file tools (mv, c= p etc.); this changes how backups get done but also adds a layer of insulat= ion between valuable data and users. Greg Williamson DBA GlobeXplorer LLC -----Original Message----- From: Christopher Browne [mailto:cbbrowne@acm.org] Sent: Mon 3/29/2004 10:28 AM To: pgsql-admin@postgresql.org Cc:=09 Subject: Re: [ADMIN] Raw devices vs. Filesystems After takin a swig o' Arrakan spice grog, el_vigia_ec@hotmail.com ("Jaime C= asanova") belched out: > Can you tell me (or at least guide me to a palce where i can find the > answer) what are the benefits of filesystems over raw devices? For PostgreSQL, filesystems have the merit that you can actually use them. PostgreSQL doesn't support use of "raw devices." Two major benefits of using filesystems as opposed to raw devices are that: a) The use of raw devices is dramatically non-portable; you have to reimplement data access on every platform you are trying to support;=20 b) The use of raw devices essentially mandates that you implement some form of generic filesystem on top of them, which adds considerable complexity to your code. Two benefits to raw devices are claimed... c) It's faster. But that assumes that the "cooked" filesystems are implemented fairly badly. That was typically true, a dozen years ago, but it isn't so typical now, particularly with a fancy cacheing controller. d) It guarantees application control of update ordering. Of course, with a cacheing controller, or disk drives that lie to one degree or another, those guarantees might be gone anyways. There are other filesystem advantages, such as e) Shifting "cooked" data around may be as simple as a "mv," whereas reorganizing on raw disk requires DB-specific tools... > And what filesystem is the best for postgresql performance? That would depend, assortedly, on what OS you are using, what kind of hardware you are running on, what kind of usage patterns you have, as well as on how you define the notion of "best." Absent of any indication of any of those things, the best that can be said is "that depends..." --=20 (format nil "~S@~S" "cbbrowne" "acm.org") http://cbbrowne.com/info/languages.html TTY Message from The-XGP at MIT-AI: The-XGP@AI 02/59/69 02:59:69 Your XGP output is startling. ---------------------------(end of broadcast)--------------------------- TIP 8: explain analyze is your friend From pgsql-performance-owner@postgresql.org Mon Apr 5 20:22:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 129C4D1BB4C for ; Mon, 5 Apr 2004 19:58:59 +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 44582-03 for ; Mon, 5 Apr 2004 16:59:27 -0300 (ADT) Received: from www.totalcardinc.com (unknown [64.33.232.186]) by svr1.postgresql.org (Postfix) with ESMTP id 5E2FDD1BAFD for ; Mon, 5 Apr 2004 16:58:55 -0300 (ADT) Received: from RSCHWARZW2K ([10.250.0.37]) (authenticated bits=0) by www.totalcardinc.com (8.12.8/8.12.8) with ESMTP id i35JxWTN025873 for ; Mon, 5 Apr 2004 14:59:32 -0500 From: "Rosser Schwarz" To: Subject: Re: atrocious update performance Date: Mon, 5 Apr 2004 14:59:32 -0500 Message-ID: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> 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 10.0.4510 In-Reply-To: <40715974.10626.ED9A003@localhost> Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/51 X-Sequence-Number: 6351 while you weren't looking, Kevin Barnard wrote: > Have you added indexes for the custid column for tables > account.acct accunt.orgacct and note? They were indexed in the original case, yes. There was no need to index them in today's test case, as that was done purely in attempt to rule in or out foreign key validation as the cause of the performance hit. No foreign keys that might be validated, no need to index the foreign key columns. > I haven't followed the entire thread but it you have > cascading FK on those tables without an index on the > column that could cause your delay. The issue is that the foreign keys are being validated at all, when the column being referenced by those foreign keys (account.cust.custid) is never touched. Regardless of whether or not the referencing columns are indexed, validating them at all--in this specific case--is broken. The column they refer to is never touched; they should remain utterly ignorant of whatever happens to other columns in the same row. /rls -- Rosser Schwarz Total Card, Inc. From pgsql-performance-owner@postgresql.org Mon Apr 5 18:32:06 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A3311D1D25F for ; Mon, 5 Apr 2004 21:27:57 +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 85771-10 for ; Mon, 5 Apr 2004 18:28:24 -0300 (ADT) Received: from outbound.mailhop.org (outbound.mailhop.org [63.208.196.171]) by svr1.postgresql.org (Postfix) with ESMTP id 7A4AFD1CCC2 for ; Mon, 5 Apr 2004 18:27:52 -0300 (ADT) Received: from ool-4353b51a.dyn.optonline.net ([67.83.181.26] helo=zeut.net) by outbound.mailhop.org with asmtp (TLSv1:AES256-SHA:256) (Exim 4.20) id 1BAbdu-00083K-77; Mon, 05 Apr 2004 17:28:30 -0400 Message-ID: <4071CF81.2000204@zeut.net> Date: Mon, 05 Apr 2004 17:28:33 -0400 From: "Matthew T. O'Connor" User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Heiko Kehlenbrink Cc: pgsql-performance@postgresql.org Subject: Re: performance comparission postgresql/ms-sql server References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> In-Reply-To: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Mail-Handler: MailHop Outbound by DynDNS.org X-Report-Abuse-To: abuse@dyndns.org X-MHO-User: zeut X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/50 X-Sequence-Number: 6350 Heiko Kehlenbrink wrote: >i use suse 8.1 > postgresql 7.2 compiled from the rpms for using postgis, but that is > > > Try v7.4, there are many performance improvements. It may not make up all the differences but it should help. From pgsql-performance-owner@postgresql.org Mon Apr 5 21:48:53 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 109D8D1D206 for ; Mon, 5 Apr 2004 22:49:40 +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 16002-08 for ; Mon, 5 Apr 2004 19:50:09 -0300 (ADT) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 67EF3D1CCC1 for ; Mon, 5 Apr 2004 19:49:34 -0300 (ADT) 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 i35MmcBK015420; Mon, 5 Apr 2004 16:48:39 -0600 (MDT) Date: Mon, 5 Apr 2004 16:48:49 -0600 (MDT) From: "scott.marlowe" To: Kevin Barnard Cc: Subject: Re: atrocious update performance In-Reply-To: <40715974.10626.ED9A003@localhost> 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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/52 X-Sequence-Number: 6352 On Mon, 5 Apr 2004, Kevin Barnard wrote: > On 5 Apr 2004 at 12:05, Rosser Schwarz wrote: > > > Just this morning, however, I created a copy of the target table (all > > 4.7M rows), with absolutely no foreign keys referring to it, and ran > > the update against the copy. That update took 2300 seconds. The > > join columns were indexed in both cases. > > Have you added indexes for the custid column for tables account.acct accunt.orgacct > and note? > > I haven't followed the entire thread but it you have cascading FK on those tables > without an index on the column that could cause your delay. also make sure the fk/pk types match, or the index likely won't get used anyway. From pgsql-performance-owner@postgresql.org Mon Apr 5 22:37:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EB30FD1C9ED for ; Tue, 6 Apr 2004 00:19:39 +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 57188-01 for ; Mon, 5 Apr 2004 21:20:11 -0300 (ADT) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id C8E7DD1C995 for ; Mon, 5 Apr 2004 21:19:37 -0300 (ADT) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 13F8A35517; Mon, 5 Apr 2004 17:20:20 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 127A4350A3; Mon, 5 Apr 2004 17:20:20 -0700 (PDT) Date: Mon, 5 Apr 2004 17:20:20 -0700 (PDT) From: Stephan Szabo To: Rosser Schwarz Cc: pgsql-performance@postgresql.org Subject: Re: atrocious update performance In-Reply-To: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> Message-ID: <20040405170631.M19491@megazone.bigpanda.com> References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/54 X-Sequence-Number: 6354 On Mon, 5 Apr 2004, Rosser Schwarz wrote: > while you weren't looking, Kevin Barnard wrote: > > > Have you added indexes for the custid column for tables > > account.acct accunt.orgacct and note? > > They were indexed in the original case, yes. There was no > need to index them in today's test case, as that was done > purely in attempt to rule in or out foreign key validation > as the cause of the performance hit. No foreign keys that > might be validated, no need to index the foreign key columns. > > > I haven't followed the entire thread but it you have > > cascading FK on those tables without an index on the > > column that could cause your delay. > > The issue is that the foreign keys are being validated at > all, when the column being referenced by those foreign keys > (account.cust.custid) is never touched. > > Regardless of whether or not the referencing columns are > indexed, validating them at all--in this specific case--is > broken. The column they refer to is never touched; they > should remain utterly ignorant of whatever happens to other > columns in the same row. It shouldn't be checking the other table if the values of the key column hadn't changed. The ri_KeysEqual check should be causing it to return just before actually doing the check on the other table (it still does a few things before then but nothing that should be particularly expensive). In some simple tests on my 7.4.2 machine, this appears to work for me on pk cascade updates. It would be interesting to know if it's actually doing any checks for you, you might be able to poke around the triggers (backend/utils/adt/ri_triggers.c). From pgsql-performance-owner@postgresql.org Mon Apr 5 22:35:14 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 271DAD1B441 for ; Tue, 6 Apr 2004 00:53:30 +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 68383-02 for ; Mon, 5 Apr 2004 21:54:00 -0300 (ADT) Received: from leo.supplyfx.com (h-67-100-199-42.lsanca54.covad.net [67.100.199.42]) by svr1.postgresql.org (Postfix) with ESMTP id ACC14D1ED4E for ; Mon, 5 Apr 2004 21:53:26 -0300 (ADT) Received: from [10.0.100.21] (unknown [10.0.100.21]) by leo.supplyfx.com (Postfix) with ESMTP id 16108EC26D for ; Mon, 5 Apr 2004 17:54:08 -0700 (PDT) Mime-Version: 1.0 (Apple Message framework v612) In-Reply-To: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Qing Zhao Subject: possible improvement between G4 and G5 Date: Mon, 5 Apr 2004 17:53:49 -0700 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.612) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/53 X-Sequence-Number: 6353 Hi, We have got a G5 64-bit processor to replace an old G4 32-bit processor. Given everything else equal, should we see a big improvement on PG's performance? The other question I have is that, when I tried different size for shared_buffer ( i used 10,000, 1,000, 528, 256) and Max connections=32, it gives me error when I tried to start PG using pg_ctl start as postgres. It kept saying this is bigger than the system Shared Memory. So finally I started PG using SystemStarter start PostgreSQL and it seems starting OK. Any idea? Thanks a lot! Qing Zhao From pgsql-performance-owner@postgresql.org Tue Apr 6 12:01:06 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 910BFD1B91B for ; Tue, 6 Apr 2004 01:41:23 +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 84992-04 for ; Mon, 5 Apr 2004 22:41:54 -0300 (ADT) Received: from exchange.rampageusa.net (exchange.dslextreme.com [66.51.198.16]) by svr1.postgresql.org (Postfix) with ESMTP id 57B0DD1ED8E for ; Mon, 5 Apr 2004 22:41:19 -0300 (ADT) 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_01C41B78.3BF37693" Subject: Wierd issues Date: Mon, 5 Apr 2004 18:41:08 -0700 Message-ID: <79AEF92046759442A4AA04C2AB08C0B80834E2@exchange.corp.dslextreme.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Wierd issues Thread-Index: AcQbeDuW3/erCMLlTduaiu8lxIPCEQ== From: "Andrew Matthews" To: X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/63 X-Sequence-Number: 6363 This is a multi-part message in MIME format. ------_=_NextPart_001_01C41B78.3BF37693 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable This is what I got... =20 Two servers, one debian, one fedora =20 Debain dual 3ghz, 1 gig ram, ide, PostgreSQL 7.2.1 on i686-pc-linux-gnu, compiled by GCC 2.95.4 =20 =20 Fedora: Dual 3ghz, 1 gig ram, scsi, PostgreSQL 7.3.4-RH on i386-redhat-linux-gnu, compiled by GCC i386-redhat-linux-gcc (GCC) 3.3.2 20031022 (Red Hat Linux 3.3.2-1) =20 =20 Both have same databases, Both have had vacume full ran on them. Both doing the same query =20 Select * from vpopmail; The vpopmail is a view, this is the view =20 =20 View "vpopmail" Column | Type | Modifiers=20 -----------+------------------------+----------- pw_name | character varying(32) |=20 pw_domain | character varying(64) |=20 pw_passwd | character varying |=20 pw_uid | integer |=20 pw_gid | integer |=20 pw_gecos | character varying |=20 pw_dir | character varying(160) |=20 pw_shell | character varying(20) |=20 View definition: SELECT ea.email_name AS pw_name, ea.domain AS pw_domain, get_pwd(u.username, '127.0.0.1'::"varchar", '101'::"varchar", 'MD5'::"varchar") AS pw_passwd, 0 AS pw_uid, 0 AS pw_gid, ''::"varchar" AS pw_gecos, ei.directory AS pw_dir, ei.quota AS pw_shell FROM email_addresses ea, email_info ei, users u, user_resources ur WHERE (((((ea.user_resource_id =3D ei.user_resource_id) AND (get_pwd(u.username, '127.0.0.1'::"varchar", '101'::"varchar", 'MD5'::"varchar") IS NOT NULL)) AND (ur.id =3D ei.user_resource_id)) AND (u.id =3D ur.user_id)) AND (NOT (EXISTS (SELECT forwarding.email_id FROM forwarding WHERE (forwarding.email_id =3D ea.id))))); =20 =20 =20 Both are set to the same buffers and everything... this is the execution time: =20 Debian: Total runtime: 35594.81 msec =20 Fedora: Total runtime: 2279869.08 msec =20 Huge difference as you can see... here are the pastes of the stuff =20 Debain: =20 user_acl=3D# explain analyze SELECT count(*) from vpopmail; NOTICE: QUERY PLAN: =20 Aggregate (cost=3D438231.94..438231.94 rows=3D1 width=3D20) (actual time=3D35594.67..35594.67 rows=3D1 loops=3D1) -> Hash Join (cost=3D434592.51..438142.51 rows=3D35774 width=3D20) (act= ual time=3D34319.24..35537.11 rows=3D70613 loops=3D1) -> Seq Scan on email_info ei (cost=3D0.00..1721.40 rows=3D71640 width=3D4) (actual time=3D0.04..95.13 rows=3D71689 loops=3D1) -> Hash (cost=3D434328.07..434328.07 rows=3D35776 width=3D16) (actual time=3D34319.00..34319.00 rows=3D0 loops=3D1) -> Hash Join (cost=3D430582.53..434328.07 rows=3D35776 width=3D16) (actual time=3D2372.45..34207.21 rows=3D70613 loops=3D1) -> Seq Scan on users u (cost=3D0.00..1938.51 rows=3D71283 width=3D4) (actual time=3D0.81..30119.58 rows=3D70809 loops=3D= 1) -> Hash (cost=3D430333.64..430333.64 rows=3D35956 width=3D12) (actual time=3D2371.51..2371.51 rows=3D0 loops=3D1) -> Hash Join (cost=3D2425.62..430333.64 rows=3D35956 width=3D12) (actual time=3D176.73..2271.14 rows=3D71470 loops= =3D1) -> Seq Scan on email_addresses ea (cost=3D0.00..426393.25 rows=3D35956 width=3D4) (actual time=3D0.06..627.49 rows=3D71473 loops=3D1) SubPlan -> Index Scan using forwarding_idx on forwarding (cost=3D0.00..5.88 rows=3D1 width=3D4) (actual time=3D0.00..0.00 rows=3D0 loops=3D71960) -> Hash (cost=3D1148.37..1148.37 rows=3D71637 width=3D8) (actual time=3D176.38..176.38 rows=3D0 loops=3D1) -> Seq Scan on user_resources ur (cost=3D0.00..1148.37 rows=3D71637 width=3D8) (actual time=3D0.03..82.21 rows=3D71686 loops=3D1) Total runtime: 35594.81 msec =20 EXPLAIN =20 =20 =20 And for fedora it's =20 =20 Aggregate (cost=3D416775.52..416775.52 rows=3D1 width=3D20) (actual time=3D2279868.57..2279868.58 rows=3D1 loops=3D1) -> Hash Join (cost=3D413853.79..416686.09 rows=3D35772 width=3D20) (actual time=3D2279271.26..2279803.91 rows=3D70841 loops=3D1) Hash Cond: ("outer".user_resource_id =3D "inner".id) -> Seq Scan on email_info ei (cost=3D0.00..1666.07 rows=3D71907 width=3D4) (actual time=3D8.12..171.10 rows=3D71907 loops=3D1) -> Hash (cost=3D413764.36..413764.36 rows=3D35772 width=3D16) (actual time=3D2279263.03..2279263.03 rows=3D0 loops=3D1) -> Hash Join (cost=3D410712.87..413764.36 rows=3D35772 width=3D16) (actual time=3D993.90..2279008.72 rows=3D70841 loops=3D1) Hash Cond: ("outer".id =3D "inner".user_id) -> Seq Scan on users u (cost=3D0.00..1888.85 rows=3D71548 width=3D4) (actual time=3D18.38..2277152.51 rows=3D71028 loops= =3D1) Filter: (get_pwd(username, '127.0.0.1'::character varying, '101'::character varying, 'MD5'::character varying) IS NOT NULL) -> Hash (cost=3D410622.99..410622.99 rows=3D35952 width=3D12) (actual time=3D975.40..975.40 rows=3D0 loops=3D1) -> Hash Join (cost=3D408346.51..410622.99 rows=3D35952 width=3D12) (actual time=3D507.52..905.91 rows=3D71697 loops= =3D1) Hash Cond: ("outer".id =3D "inner".user_resource_id) -> Seq Scan on user_resources ur (cost=3D0.00..1108.04 rows=3D71904 width=3D8) (actual time=3D0.05..95.65 rows=3D71904 loops=3D1) -> Hash (cost=3D408256.29..408256.29 rows=3D36091 width=3D4) (actual time=3D507.33..507.33 rows=3D0 loops=3D1) -> Seq Scan on email_addresses ea (cost=3D0.00..408256.29 rows=3D36091 width=3D4) (actual time=3D0.15..43= 2.83 rows=3D71700 loops=3D1) Filter: (NOT (subplan)) SubPlan -> Index Scan using forwarding_idx on forwarding (cost=3D0.00..5.63 rows=3D1 width=3D4) (actual time=3D0.00..0.00 rows=3D0 loops=3D72182) Index Cond: (email_id =3D $0) Total runtime: 2279869.08 msec (20 rows) =20 =20 =20 Any suggestions? =20 I can't figure this out. There is no reason it should be that much of a difference, It's all the same value's, Thanks in advanced. =20 Andrew ------_=_NextPart_001_01C41B78.3BF37693 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

This is what I got…

 

Two servers, one debian, one fedora

 

Debain dual 3ghz, 1 gig ram, ide, PostgreSQL 7.2.1 on i686-pc-linux-gnu, compiled by GCC 2.95.4

 

 

Fedora: Dual 3ghz, 1 gig ram, scsi, PostgreSQL 7.3.4-RH = on i386-redhat-linux-gnu, compiled by GCC i386-redhat-linux-gcc (GCC) 3.3.2 20031022 (Red Hat Linux 3.3.2-1)

 

 

Both have same databases, Both have had vacume full ran = on them. Both doing the same query

 

Select * from vpopmail; The vpopmail is a view, this is = the view

 

 

         &n= bsp;      View "vpopmail"

  Column   |     = ;     Type       &nb= sp;  | Modifiers

-----------+------------------------+-----------

 pw_name   | character varying(32)  = |

 pw_domain | character varying(64)  |

 pw_passwd | character varying   &nb= sp;  |

 pw_uid    | integer  &nbs= p;             = |

 pw_gid    | integer  &nbs= p;             = |

 pw_gecos  | character varying  &nbs= p;   |

 pw_dir    | character varying(160) = |

 pw_shell  | character varying(20)  |

View definition: SELECT ea.email_name AS pw_name, ea.dom= ain AS pw_domain, get_pwd(u.username, '127.0.0.1'::"varchar", '101'::"varchar", 'MD5'::"varchar") AS pw_passwd, 0 AS pw_uid, 0 AS pw_gid, ''::"varchar" AS pw_gecos, ei.directory AS pw_dir, ei.quota AS pw_shell FROM email_addresses ea, email_info ei, users = u, user_resources ur WHERE (((((ea.user_resource_id =3D ei.user_resource_id) A= ND (get_pwd(u.username, '127.0.0.1'::"varchar", '101'::"varchar", 'MD5'::"varchar") IS NOT NULL)) AND (ur.id =3D ei.user_resource_id)) AND (u.id =3D ur.user_id)) AND (NOT (EXISTS (SELECT forwarding.email_id FROM forwarding WHERE (forwarding.email_id =3D ea.id)))));

 

 

 

Both are set to the same buffers and everything… t= his is the execution time:

 

Debian: Total runtime: 35594.81 msec

 

Fedora: Total runtime: 2279869.08 msec=

 

Huge difference as you can see… here are the paste= s of the stuff

 

Debain:

 

user_acl=3D# explain analyze SELECT count(*) from vpopma= il;

NOTICE:  QUERY PLAN:

 

Aggregate  (cost=3D438231.94..438231.94 rows=3D1 wi= dth=3D20) (actual time=3D35594.67..35594.67 rows=3D1 loops=3D1)

  ->  Hash Join  (cost=3D434592.51..43= 8142.51 rows=3D35774 width=3D20) (actual time=3D34319.24..35537.11 rows=3D70613 loops=3D1)<= /o:p>

        ->  S= eq Scan on email_info ei  (cost=3D0.00..1721.40 rows=3D71640 width=3D4) (actual time=3D0.04..95.13 ro= ws=3D71689 loops=3D1)

        ->  H= ash  (cost=3D434328.07..434328.07 rows=3D35776 width=3D16) (actual time=3D34319.00..34319.00 rows=3D0 loops=3D1)

         &n= bsp;    ->  Hash Join  (cost=3D430582.53..43432= 8.07 rows=3D35776 width=3D16) (actual time=3D2372.45..34207.21 rows=3D70613 loop= s=3D1)

         &n= bsp;          ->  Seq = Scan on users u  (cost=3D0.00..1938.51 rows=3D71283 width=3D4) (actual time=3D0.81..30119.58= rows=3D70809 loops=3D1)

         &n= bsp;          ->  Hash=   (cost=3D430333.64..430333.64 rows=3D35956 width=3D12) (actual time=3D2371.51..2371.51 rows=3D0 loops=3D1= )

         &n= bsp;            = ;    ->  Hash Join  (cost=3D2425.62..430333.64 rows=3D35956 width=3D12) (actual time=3D176.73..= 2271.14 rows=3D71470 loops=3D1)

         &n= bsp;            = ;          ->  Seq Sca= n on email_addresses ea  (cost=3D0.00..426393.25 rows=3D35956 width=3D4) (a= ctual time=3D0.06..627.49 rows=3D71473 loops=3D1)

         &n= bsp;            = ;            &n= bsp;   SubPlan

         &n= bsp;            = ;            &n= bsp;     ->  Index Scan using forwarding_idx on forwarding  (cost=3D0.00..5.88 rows=3D1 width= =3D4) (actual time=3D0.00..0.00 rows=3D0 loops=3D71960)

          &= nbsp;           &nbs= p;         ->  Hash&nb= sp; (cost=3D1148.37..1148.37 rows=3D71637 width=3D8) (actual time=3D176.38..176= .38 rows=3D0 loops=3D1)

         &n= bsp;            = ;            &n= bsp;   ->  Seq Scan on user_resources ur<= /st1:City>  (cost=3D0.00..1148.37 rows=3D71637 width=3D8) (actual time=3D0.03..82.21 ro= ws=3D71686 loops=3D1)

Total runtime: 35594.81 msec

 

EXPLAIN

 

 

 

And for fedora it’s

 

 

Aggregate  (cost=3D416775.52..416775.52 row=
s=3D1 width=3D20) (actual time=3D2279868.57..2279868.58 rows=3D1 loops=3D1)=

   ->&nb=
sp; Hash Join  (cost=3D413853.79..416686.09 rows=3D35772 width=3D20) (=
actual time=3D2279271.26..2279803.91 rows=3D70841 loops=3D1)
   &nb=
sp;     Hash Cond: ("outer".user_resource_id =
=3D "inner".id)
   &nb=
sp;     ->  Seq Scan on email_info ei  (co=
st=3D0.00..1666.07 rows=3D71907 width=3D4) (actual time=3D8.12..171.10 rows=
=3D71907 loops=3D1)
   &nb=
sp;     ->  Hash  (cost=3D413764.36..41376=
4.36 rows=3D35772 width=3D16) (actual time=3D2279263.03..2279263.03 rows=3D=
0 loops=3D1)
   &nb=
sp;           -> =
 Hash Join  (cost=3D410712.87..413764.36 rows=3D35772 width=3D16) (act=
ual time=3D993.90..2279008.72 rows=3D70841 loops=3D1)
   &nb=
sp;            =
     Hash Cond: ("outer".id =3D "inner&q=
uot;.user_id)
   &nb=
sp;            =
     ->  Seq Scan on users u  (cost=3D0.00=
..1888.85 rows=3D71548 width=3D4) (actual time=3D18.38..2277152.51 rows=3D7=
1028 loops=3D1)
   &nb=
sp;            =
           Filter: (get_p=
wd(username, '127.0.0.1'::character varying, '101'::character varying, 'MD5=
'::character varying) IS NOT NULL)
   &nb=
sp;            =
     ->  Hash  (cost=3D410622.99..410622.9=
9 rows=3D35952 width=3D12) (actual time=3D975.40..975.40 rows=3D0 loops=3D1=
)
   &nb=
sp;             =
;          ->  Ha=
sh Join  (cost=3D408346.51..410622.99 rows=3D35952 width=3D12) (actual=
 time=3D507.52..905.91 rows=3D71697 loops=3D1)
   &nb=
sp;            =
            &nb=
sp;    Hash Cond: ("outer".id =3D "inner"=
;.user_resource_id)
   &nb=
sp;            =
            &nb=
sp;    ->  Seq Scan on user_resources ur  (cost=3D=
0.00..1108.04 rows=3D71904 width=3D8) (actual time=3D0.05..95.65 rows=3D719=
04 loops=3D1)
   &nb=
sp;            =
            &nb=
sp;    ->  Hash  (cost=3D408256.29..408256.29 r=
ows=3D36091 width=3D4) (actual time=3D507.33..507.33 rows=3D0 loops=3D1)
   &nb=
sp;             =
;            &n=
bsp;         ->  Seq S=
can on email_addresses ea  (cost=3D0.00..408256.29 rows=3D36091 width=
=3D4) (actual time=3D0.15..432.83 rows=3D71700 loops=3D1)=

   &nb=
sp;            =
            &nb=
sp;            =
    Filter: (NOT (subplan))
   &nb=
sp;            =
            &nb=
sp;             =
;   SubPlan
   &nb=
sp;            =
            &nb=
sp;            =
      ->  Index Scan using forwarding_idx =
on forwarding  (cost=3D0.00..5.63 rows=3D1 width=3D4) (actual time=3D0=
.00..0.00 rows=3D0 loops=3D72182)
   &nb=
sp;            =
            &nb=
sp;            =
            Index Co=
nd: (email_id =3D $0)
 Total runtime: 22798=
69.08 msec

(20 rows)

 

 

 

Any suggestions?

 

I can’t figure this out. There is no reason it should be that much of a difference, It’s all the same value’s, Thanks in advanced.

 

Andrew

------_=_NextPart_001_01C41B78.3BF37693-- From pgsql-performance-owner@postgresql.org Tue Apr 6 02:46:47 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9F64FD1B4BE for ; Tue, 6 Apr 2004 02:46:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53968-08 for ; Tue, 6 Apr 2004 02:47:24 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E9B03D1B4B6 for ; Tue, 6 Apr 2004 02:46:38 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i365lNvI019782; Tue, 6 Apr 2004 01:47:23 -0400 (EDT) To: Qing Zhao Cc: pgsql-performance@postgresql.org Subject: Re: possible improvement between G4 and G5 In-reply-to: References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> Comments: In-reply-to Qing Zhao message dated "Mon, 05 Apr 2004 17:53:49 -0700" Date: Tue, 06 Apr 2004 01:47:22 -0400 Message-ID: <19781.1081230442@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/55 X-Sequence-Number: 6355 Qing Zhao writes: > We have got a G5 64-bit processor to replace an old G4 32-bit > processor. Given everything else equal, should we see a big > improvement on PG's performance? Nope. Database performance typically depends on disk performance first, and RAM size second. A 64-bit processor might help by allowing you to install more RAM, but you didn't say that you had. > The other question I have is that, when I tried different size for > shared_buffer ( i used 10,000, 1,000, 528, 256) and Max > connections=32, it gives me error when I tried to start PG using > pg_ctl start as postgres. It kept saying this is bigger than the > system Shared Memory. Out-of-the-box, Mac OS X has a very low SHMMAX limit. See the PG admin docs or the mail list archives about how to increase it. You should do this --- most people find that you want to set shared_buffers to 1000 or 10000 or so for best performance. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 6 04:26:42 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7A0E2D1B8DF for ; Tue, 6 Apr 2004 04:26:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 95895-03 for ; Tue, 6 Apr 2004 04:27:21 -0300 (ADT) Received: from frodo.hserus.net (frodo.hserus.net [204.74.68.40]) by svr1.postgresql.org (Postfix) with ESMTP id ECBFBD1B54E for ; Tue, 6 Apr 2004 04:26:35 -0300 (ADT) Received: from concord.pspl.co.in ([202.54.11.72]:61708 helo=frodo.hserus.net) by frodo.hserus.net with asmtp (Cipher TLSv1:DHE-RSA-AES256-SHA:256) (Exim 4.31 #0) id 1BAkzQ-0006du-E8 by authid with plain; Tue, 06 Apr 2004 12:57:21 +0530 Message-ID: <40725BD1.3050008@frodo.hserus.net> Date: Tue, 06 Apr 2004 12:57:13 +0530 From: Shridhar Daithankar 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: Heiko Kehlenbrink , Performance Subject: Re: performance comparission postgresql/ms-sql server References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> <40718146.1010506@frodo.hserus.net> <56755.195.243.253.146.1081233367.squirrel@webmail.fh-oldenburg.de> In-Reply-To: <56755.195.243.253.146.1081233367.squirrel@webmail.fh-oldenburg.de> 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: 200404/56 X-Sequence-Number: 6356 Heiko Kehlenbrink wrote: > hkehlenbrink@lin0493l:~> psql -d test -c 'explain analyse select avg(dist) > from massive2 where dist > (1000000*sqrt(3.0))::float8 and dist < > (1500000*sqrt(3.0))::float8;' > NOTICE: QUERY PLAN: > > Aggregate (cost=14884.61..14884.61 rows=1 width=8) (actual > time=3133.24..3133.24 rows=1 loops=1) > -> Index Scan using massive2_dist on massive2 (cost=0.00..13648.17 > rows=494573 width=8) (actual time=0.11..2061.38 rows=499999 loops=1) > Total runtime: 3133.79 msec > > EXPLAIN > > seems to me that most time was needed for the index scanning... Hmm... I would suggest if you are testing, you should try 7.4.2. 7.4 has some good optimisation for hash agregates though I am not sure if it apply to averaging. Also try forcing a seq. scan by turning off index scan. I guess index scan for so many rows is not exactly good thing even if tuple size if pretty small. Shridhar From pgsql-performance-owner@postgresql.org Tue Apr 6 05:04:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6F58AD1E172 for ; Tue, 6 Apr 2004 05:04:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01070-06 for ; Tue, 6 Apr 2004 05:05:09 -0300 (ADT) Received: from mail (unknown [202.108.142.212]) by svr1.postgresql.org (Postfix) with ESMTP id 80DB0D1B98D for ; Tue, 6 Apr 2004 05:04:22 -0300 (ADT) Received: from HUANGYQ ([61.149.153.189]) by mail.gthome.com (iPlanet Messaging Server 5.1 (built May 7 2001)) with ESMTPA id <0HVQ00E32PI95I@mail.gthome.com> for pgsql-performance@postgresql.org; Tue, 06 Apr 2004 16:13:22 +0800 (CST) Date: Tue, 06 Apr 2004 16:01:34 +0800 From: huang yaqin Subject: good pc but bad performance,why? To: "pgsql-performance@postgresql.org" Message-id: <0HVQ00E33PI95I@mail.gthome.com> MIME-version: 1.0 X-Mailer: Foxmail 4.2 [cn] Content-type: text/plain; charset=GB2312 Content-transfer-encoding: quoted-printable X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.9 tagged_above=0.0 required=5.0 tests=RCVD_IN_RFCI, TO_ADDRESS_EQ_REAL X-Spam-Level: X-Archive-Number: 200404/57 X-Sequence-Number: 6357 hello, I have some question when I use postgresql 7.4.1 on redhat adv server 2.1 . I use IBM335 as server, it has 4 cpus, 1G RAM. but I got very bad performan= ce. I can only do about 50 inserts per sencond. Event worse than my pc(PIII 800= ,256M RAM), can anyone give me some advice?=09 =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1huang yaqin =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1hyq@gthome.com =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A12004-04-06 From pgsql-performance-owner@postgresql.org Tue Apr 6 08:15:48 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1329ED1E9E1 for ; Tue, 6 Apr 2004 08:15:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 78119-09 for ; Tue, 6 Apr 2004 08:16:28 -0300 (ADT) 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 9EA7FD1E9C5 for ; Tue, 6 Apr 2004 08:15:40 -0300 (ADT) 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 1BAoZ8-0000MC-0X; Tue, 06 Apr 2004 12:16:27 +0100 Received: by mainbox.archonet.com (Postfix, from userid 529) id 900FF1666F; Tue, 6 Apr 2004 12:16:24 +0100 (BST) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 9934316486; Tue, 6 Apr 2004 12:16:22 +0100 (BST) From: Richard Huxton To: huang yaqin , "pgsql-performance@postgresql.org" Subject: Re: good pc but bad performance,why? Date: Tue, 6 Apr 2004 12:16:22 +0100 User-Agent: KMail/1.5 References: <0HVQ00E33PI95I@mail.gthome.com> In-Reply-To: <0HVQ00E33PI95I@mail.gthome.com> MIME-Version: 1.0 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404061216.22856.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: 200404/58 X-Sequence-Number: 6358 On Tuesday 06 April 2004 09:01, huang yaqin wrote: > hello, > > I have some question when I use postgresql 7.4.1 on redhat adv server 2.1 > . I use IBM335 as server, it has 4 cpus, 1G RAM. but I got very bad > performance. I can only do about 50 inserts per sencond. Event worse than > my pc(PIII 800,256M RAM), can anyone give me some advice? How have you tuned your postgresql.conf file? What disk systems do you have? What does vmstat/iostat show as the bottleneck in the system? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Tue Apr 6 09:46:08 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 07E8CD1D4DA for ; Tue, 6 Apr 2004 09:46:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15832-08 for ; Tue, 6 Apr 2004 09:46:51 -0300 (ADT) Received: from internet.potentialtech.com (h-66-167-251-6.phlapafg.covad.net [66.167.251.6]) by svr1.postgresql.org (Postfix) with ESMTP id CEBC2D1D391 for ; Tue, 6 Apr 2004 09:46:03 -0300 (ADT) Received: from potentialtech.com (pa-plum1c-102.pit.adelphia.net [24.53.179.102]) by internet.potentialtech.com (Postfix) with ESMTP id 48BBA69A71; Tue, 6 Apr 2004 08:46:51 -0400 (EDT) Message-ID: <4072A6B9.5050502@potentialtech.com> Date: Tue, 06 Apr 2004 08:46:49 -0400 From: Bill Moran User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031005 X-Accept-Language: en-us, en MIME-Version: 1.0 To: huang yaqin Cc: "pgsql-performance@postgresql.org" Subject: Re: good pc but bad performance,why? References: <0HVQ00E33PI95I@mail.gthome.com> In-Reply-To: <0HVQ00E33PI95I@mail.gthome.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/59 X-Sequence-Number: 6359 huang yaqin wrote: > hello, > > I have some question when I use postgresql 7.4.1 on redhat adv server 2.1 . > I use IBM335 as server, it has 4 cpus, 1G RAM. but I got very bad performance. > I can only do about 50 inserts per sencond. Event worse than my pc(PIII 800,256M RAM), can anyone give me some advice? Have you referenced this document?: http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Tue Apr 6 10:00:45 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 892A4D1D370 for ; Tue, 6 Apr 2004 10:00:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21737-06 for ; Tue, 6 Apr 2004 10:01:27 -0300 (ADT) 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 0935AD1D282 for ; Tue, 6 Apr 2004 10:00:39 -0300 (ADT) 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 1BAqCk-000IEF-0W for pgsql-performance@postgresql.org; Tue, 06 Apr 2004 14:01:26 +0100 Received: by mainbox.archonet.com (Postfix, from userid 529) id 77BDE1799E; Tue, 6 Apr 2004 14:01:25 +0100 (BST) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 23E1216486 for ; Tue, 6 Apr 2004 14:01:23 +0100 (BST) From: Richard Huxton To: pgsql-performance@postgresql.org Subject: Back to Linux 2.6 kernel thoughts... Date: Tue, 6 Apr 2004 14:01:23 +0100 User-Agent: KMail/1.5 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404061401.23189.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: 200404/60 X-Sequence-Number: 6360 I seem to remember discussion of anticipatory vs deadline scheduler in 2.6. Here is what Andrew Morton (I think) says: "The deadline scheduler has two additional scheduling queues that were not available to the 2.4 IO scheduler. The two new queues are a FIFO read queue and a FIFO write queue. This new multi-queue method allows for greater interactivity by giving the read requests a better deadline than write requests, thus ensuring that applications rarely will be delayed by read requests. Deadline scheduling is best suited for database servers and high disk performance systems. Morton has experienced up to 15 percent increases on database loads while using deadline scheduling." http://story.news.yahoo.com/news?tmpl=story&cid=75&e=2&u=/nf/20040405/tc_nf/23603 Nothing very in-depth in the story. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Tue Apr 6 11:40:41 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 15946D1B8DF for ; Tue, 6 Apr 2004 11:40:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71071-05 for ; Tue, 6 Apr 2004 11:41:21 -0300 (ADT) Received: from frodo.hserus.net (frodo.hserus.net [204.74.68.40]) by svr1.postgresql.org (Postfix) with ESMTP id C1EABD1B88D for ; Tue, 6 Apr 2004 11:40:32 -0300 (ADT) Received: from concord.pspl.co.in ([202.54.11.72]:65087 helo=frodo.hserus.net) by frodo.hserus.net with asmtp (Cipher TLSv1:DHE-RSA-AES256-SHA:256) (Exim 4.31 #0) id 1BArlQ-00099m-4v by authid with plain; Tue, 06 Apr 2004 20:11:20 +0530 Message-ID: <4072C18B.1030600@frodo.hserus.net> Date: Tue, 06 Apr 2004 20:11:15 +0530 From: Shridhar Daithankar 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: Heiko Kehlenbrink , Performance Subject: Re: performance comparission postgresql/ms-sql server References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> <40718146.1010506@frodo.hserus.net> <56755.195.243.253.146.1081233367.squirrel@webmail.fh-oldenburg.de> <40725BD1.3050008@frodo.hserus.net> <58199.195.243.253.146.1081238266.squirrel@webmail.fh-oldenburg.de> In-Reply-To: <58199.195.243.253.146.1081238266.squirrel@webmail.fh-oldenburg.de> 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: 200404/61 X-Sequence-Number: 6361 Heiko Kehlenbrink wrote: >>Hmm... I would suggest if you are testing, you should try 7.4.2. 7.4 has >>some >>good optimisation for hash agregates though I am not sure if it apply to >>averaging. > would be the last option till we are runing other applications on that 7.2 > system I can understand.. >>Also try forcing a seq. scan by turning off index scan. I guess index scan >>for >>so many rows is not exactly good thing even if tuple size if pretty small. > a sequential scann gives me the following result: > > HKehlenbrink@lin0493l:~> time psql -d test -c 'explain analyse select > avg(dist) from massive2 where dist > 1000000*sqrt(3.0)::float8 and dist < > 1500000*sqrt(3.0)::float8 ;' > NOTICE: QUERY PLAN: > > Aggregate (cost=1193714.43..1193714.43 rows=1 width=8) (actual > time=166718.54..166718.54 rows=1 loops=1) > -> Seq Scan on massive2 (cost=0.00..1192478.00 rows=494573 width=8) > (actual time=3233.22..165576.40 rows=499999 loops=1) > Total runtime: 166733.73 msec Certainly bad and not an option.. I can't think of anything offhand to speed this up.. Shridhar From pgsql-performance-owner@postgresql.org Tue Apr 6 11:42:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DFD42D1DC17 for ; Tue, 6 Apr 2004 11:42:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66886-08 for ; Tue, 6 Apr 2004 11:43:04 -0300 (ADT) Received: from MXR-8.estpak.ee (ld1.estpak.ee [194.126.101.98]) by svr1.postgresql.org (Postfix) with ESMTP id A9604D1D9E6 for ; Tue, 6 Apr 2004 11:42:15 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by MXR-8.estpak.ee (Postfix) with ESMTP id 651302993B; Tue, 6 Apr 2004 17:43:03 +0300 (EEST) Received: from MXR-8.estpak.ee ([127.0.0.1]) by localhost (Relay8 [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 32755-03; Tue, 6 Apr 2004 17:43:03 +0300 (EEST) Received: from future.ee (80-235-33-83-dsl.mus.estpak.ee [80.235.33.83]) by MXR-8.estpak.ee (Postfix) with ESMTP id 3F6242992C; Tue, 6 Apr 2004 17:43:02 +0300 (EEST) Message-ID: <4072C209.1060902@future.ee> Date: Tue, 06 Apr 2004 17:43:21 +0300 From: Kaarel User-Agent: Mozilla Thunderbird 0.5+ (X11/20040220) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Heiko Kehlenbrink Cc: pgsql-performance@postgresql.org Subject: Re: performance comparission postgresql/ms-sql server References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> In-Reply-To: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new-20030616-p7 (Debian) at neti.ee X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/62 X-Sequence-Number: 6362 Heiko Kehlenbrink wrote: >i want to convince people to use postgresql instead of ms-sql server, so i >set up a kind of comparission insert data / select data from postgresql / >ms-sql server > > [...] >do you have any hints like compiler-flags and so on to get the answering >time from postgresql equal to ms-sql? > >(btw both dbms were running on exactly the same hardware) > >i use suse 8.1 > postgresql 7.2 compiled from the rpms for using postgis, but that is >anothe story... > 1.5 gig ram > 1.8 mhz intel cpu > > >every help welcome > > Suse 8.1 comes with 2.4 series kernel I suppose. Many have witnessed a speed increase when using 2.6 series kernel. Might consider this too besides the newer PostgreSQL version already suggested. 2.6 has some scheduling options that are not enabled by default but may enhance database performance (http://story.news.yahoo.com/news?tmpl=story&cid=75&e=2&u=/nf/20040405/tc_nf/23603). Kaarel From pgsql-performance-owner@postgresql.org Tue Apr 6 12:42:52 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B1A65D1EB28 for ; Tue, 6 Apr 2004 12:42:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01203-06 for ; Tue, 6 Apr 2004 12:42:14 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 57AE4D1E146 for ; Tue, 6 Apr 2004 12:42:13 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i36Fg95R025242; Tue, 6 Apr 2004 11:42:09 -0400 (EDT) To: huang yaqin Cc: "pgsql-performance@postgresql.org" Subject: Re: good pc but bad performance,why? In-reply-to: <0HVQ00E33PI95I@mail.gthome.com> References: <0HVQ00E33PI95I@mail.gthome.com> Comments: In-reply-to huang yaqin message dated "Tue, 06 Apr 2004 16:01:34 +0800" Date: Tue, 06 Apr 2004 11:42:09 -0400 Message-ID: <25241.1081266129@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/64 X-Sequence-Number: 6364 huang yaqin writes: > I have some question when I use postgresql 7.4.1 on redhat adv server 2.1 . > I use IBM335 as server, it has 4 cpus, 1G RAM. but I got very bad performance. > I can only do about 50 inserts per sencond. Event worse than my pc(PIII 800,256M RAM), can anyone give me some advice? If the cheap machine appears to be able to commit more transactions per second than the better one, it's very likely because the cheap machine has a disk that lies about write completion. Is the server using SCSI disks by any chance? To a first approximation, IDE drives lie by default, SCSI drives don't. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 6 13:22:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 893CFD1ED0B for ; Tue, 6 Apr 2004 12:45:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01753-08 for ; Tue, 6 Apr 2004 12:45:32 -0300 (ADT) Received: from hotmail.com (law10-oe30.law10.hotmail.com [64.4.14.87]) by svr1.postgresql.org (Postfix) with ESMTP id A8F7AD1EB25 for ; Tue, 6 Apr 2004 12:45:31 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Tue, 6 Apr 2004 08:45:30 -0700 Received: from 67.81.102.201 by law10-oe30.law10.hotmail.com with DAV; Tue, 06 Apr 2004 15:45:30 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman2@hotmail.com] X-Sender: awerman2@hotmail.com From: "Aaron Werman" To: "Qing Zhao" , "Tom Lane" Cc: References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> <19781.1081230442@sss.pgh.pa.us> Subject: Re: possible improvement between G4 and G5 Date: Tue, 6 Apr 2004 11:45:29 -0400 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 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Message-ID: X-OriginalArrivalTime: 06 Apr 2004 15:45:30.0742 (UTC) FILETIME=[30C24160:01C41BEE] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/66 X-Sequence-Number: 6366 ----- Original Message ----- From: "Tom Lane" To: "Qing Zhao" Cc: Sent: Tuesday, April 06, 2004 1:47 AM Subject: Re: [PERFORM] possible improvement between G4 and G5 > Qing Zhao writes: > > We have got a G5 64-bit processor to replace an old G4 32-bit > > processor. Given everything else equal, should we see a big > > improvement on PG's performance? > > Nope. Database performance typically depends on disk performance first, > and RAM size second. I'm surprised by this thought. I tend to hit CPU bottlenecks more often than I/O ones. In most applications, db I/O is a combination of buffer misses and logging, which are both reasonably constrained. RAM size seems to me to be the best way to improve performance, and then CPU which is needed to perform the in-memory searching, locking, versioning, and processing, and finally I/O (this is not the case in small I/O subsystems - if you have less than a dozen drives, you're easily I/O bound). I/O is often the thing I tune first, because I can do it in place without buying hardware. Conceptually, an RDBMS converts slow random I/O into in memory processing and sequential logging writes. If successful, it should reduce the I/O overhead. /Aaron From pgsql-performance-owner@postgresql.org Tue Apr 6 12:53:11 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3489FD1E149 for ; Tue, 6 Apr 2004 12:52:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 07714-07 for ; Tue, 6 Apr 2004 12:52:43 -0300 (ADT) Received: from flake.decibel.org (flake.decibel.org [66.143.173.58]) by svr1.postgresql.org (Postfix) with SMTP id 5BA0CD1DC17 for ; Tue, 6 Apr 2004 12:52:42 -0300 (ADT) Received: (qmail 737 invoked by uid 1001); 6 Apr 2004 15:52:35 -0000 Date: Tue, 6 Apr 2004 10:52:35 -0500 From: "Jim C. Nasby" To: Tom Lane Cc: Qing Zhao , pgsql-performance@postgresql.org Subject: Re: possible improvement between G4 and G5 Message-ID: <20040406155235.GZ74840@nasby.net> References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> <19781.1081230442@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <19781.1081230442@sss.pgh.pa.us> User-Agent: Mutt/1.4.2.1i X-Operating-System: FreeBSD 4.9-RELEASE-p3 i386 X-Distributed: Join the Effort! http://www.distributed.net X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/65 X-Sequence-Number: 6365 On Tue, Apr 06, 2004 at 01:47:22AM -0400, Tom Lane wrote: > Qing Zhao writes: > > We have got a G5 64-bit processor to replace an old G4 32-bit > > processor. Given everything else equal, should we see a big > > improvement on PG's performance? > > Nope. Database performance typically depends on disk performance first, > and RAM size second. A 64-bit processor might help by allowing you to > install more RAM, but you didn't say that you had. Memory bandwidth is a consideration too, so you might see some performance improvements on a G5. We recently debated between Xeons and Opterons in a new PGSQL server and a little poking around on the lists indicated that the Opterons did perform better, presumably due to the increased memory bandwidth. Incidentally, this is why you need about 2x the CPUs on Sun hardware vs RS6000 hardware for database stuff (and that gets expensive if you're paying per CPU!). -- Jim C. Nasby, Database Consultant jim@nasby.net Member: Triangle Fraternity, Sports Car Club of America Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Tue Apr 6 16:16:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 258A6D1C951 for ; Tue, 6 Apr 2004 13:23:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20467-09 for ; Tue, 6 Apr 2004 13:23:05 -0300 (ADT) Received: from cronos.madness.at (madness.at [213.153.61.104]) by svr1.postgresql.org (Postfix) with ESMTP id 9A98CD1BAFC for ; Tue, 6 Apr 2004 13:23:04 -0300 (ADT) Received: from masterbrain.kaltenbrunner.cc ([195.70.118.74] helo=madness.at) by cronos.madness.at with asmtp (TLSv1:AES256-SHA:256) (Exim 4.30) id 1BAtLn-0006hY-0I; Tue, 06 Apr 2004 18:22:59 +0200 Message-ID: <4072D959.5090905@madness.at> Date: Tue, 06 Apr 2004 18:22:49 +0200 From: Stefan Kaltenbrunner User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7b) Gecko/20040326 X-Accept-Language: en-us, en MIME-Version: 1.0 To: huang yaqin Cc: pgsql-performance@postgresql.org Subject: Re: good pc but bad performance,why? References: <0HVQ00E33PI95I@mail.gthome.com> In-Reply-To: <0HVQ00E33PI95I@mail.gthome.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: 200404/68 X-Sequence-Number: 6368 huang yaqin wrote: > hello, > > I have some question when I use postgresql 7.4.1 on redhat adv server 2.1 . > I use IBM335 as server, it has 4 cpus, 1G RAM. but I got very bad performance. This is most likely a dual processor Xeon machine with HT, because the x335 is limited to two physical cpus. > I can only do about 50 inserts per sencond. Event worse than my pc(PIII 800,256M RAM), can anyone give me some advice? any chance you are using the onboard MPT-Fusion "Raid"controller with a RAID1 - we have seen absolutely horrible performance from these controllers here. Using them as a normal SCSI-Controller with Softwareraid on top fixed this for us ... stefan From pgsql-performance-owner@postgresql.org Tue Apr 6 15:51:54 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3E0C0D1E171 for ; Tue, 6 Apr 2004 15:51:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 94439-01 for ; Tue, 6 Apr 2004 15:51:52 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 21692D1E158 for ; Tue, 6 Apr 2004 15:51:50 -0300 (ADT) Received: from [63.195.55.98] (HELO 192.168.1.29) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4784265; Tue, 06 Apr 2004 11:53:11 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: "Aaron Werman" , "Qing Zhao" , "Tom Lane" Subject: Re: possible improvement between G4 and G5 Date: Tue, 6 Apr 2004 11:52:17 -0700 User-Agent: KMail/1.5.4 Cc: References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> <19781.1081230442@sss.pgh.pa.us> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404061152.17997.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/67 X-Sequence-Number: 6367 Aaron, > I'm surprised by this thought. I tend to hit CPU bottlenecks more often than > I/O ones. In most applications, db I/O is a combination of buffer misses and > logging, which are both reasonably constrained. Not my experience at all. In fact, the only times I've seen modern platforms max out the CPU was when: a) I had bad queries with bad plans, or b) I had reporting queires that did a lot of calculation for display (think OLAP). Otherwise, on the numerous servers I administrate, RAM spikes, and I/O bottlenecks, but the CPU stays almost flat. Of course, most of my apps are large databases (i.e. too big for RAM) with a heavy transaction-processing component. What kind of applications are you running? -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Tue Apr 6 17:22:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ECB71D1E086 for ; Tue, 6 Apr 2004 17:22:48 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34495-06 for ; Tue, 6 Apr 2004 17:22:48 -0300 (ADT) Received: from hotmail.com (law10-oe46.law10.hotmail.com [64.4.14.18]) by svr1.postgresql.org (Postfix) with ESMTP id 4E3D9D1D12F for ; Tue, 6 Apr 2004 17:22:46 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Tue, 6 Apr 2004 13:22:47 -0700 Received: from 67.81.102.201 by law10-oe46.law10.hotmail.com with DAV; Tue, 06 Apr 2004 20:22:47 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman2@hotmail.com] X-Sender: awerman2@hotmail.com From: "Aaron Werman" To: Cc: References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> <19781.1081230442@sss.pgh.pa.us> <200404061152.17997.josh@agliodbs.com> Subject: Re: possible improvement between G4 and G5 Date: Tue, 6 Apr 2004 16:22:46 -0400 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 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Message-ID: X-OriginalArrivalTime: 06 Apr 2004 20:22:47.0833 (UTC) FILETIME=[ED3CE090:01C41C14] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/69 X-Sequence-Number: 6369 ----- Original Message ----- From: "Josh Berkus" To: "Aaron Werman" ; "Qing Zhao" ; "Tom Lane" Cc: Sent: Tuesday, April 06, 2004 2:52 PM Subject: Re: [PERFORM] possible improvement between G4 and G5 > Aaron, > > > I'm surprised by this thought. I tend to hit CPU bottlenecks more often than > > I/O ones. In most applications, db I/O is a combination of buffer misses and > > logging, which are both reasonably constrained. > > Not my experience at all. In fact, the only times I've seen modern platforms > max out the CPU was when: > a) I had bad queries with bad plans, or > b) I had reporting queires that did a lot of calculation for display (think > OLAP). > > Otherwise, on the numerous servers I administrate, RAM spikes, and I/O > bottlenecks, but the CPU stays almost flat. > > Of course, most of my apps are large databases (i.e. too big for RAM) with a > heavy transaction-processing component. > > What kind of applications are you running? > > -- > -Josh Berkus > Aglio Database Solutions > San Francisco > > I do consulting, so they're all over the place and tend to be complex. Very few fit in RAM, but still are very buffered. These are almost all backed with very high end I/O subsystems, with dozens of spindles with battery backed up writethrough cache and gigs of buffers, which may be why I worry so much about CPU. I have had this issue with multiple servers. Consider an analysis db with 10G data. Of that, 98% of the access is read and only 2% write (that is normal for almost anything that is not order entry, even transaction processing with thorough cross validation). Almost all the queries access 10%, or 1G of the data. Of the reads, they average ~3 level b-trees, with the first 2 levels certainly cached, and the last ones often cached. Virtually all the I/O activity is logical reads against buffer. A system with a 100 transactions which on average access 200 rows does 98% of 200 rows x 100 transactions x 3 logical I/Os per read = 58,800 logical reads, of which actually maybe a hundred are physical reads. It also does 2% of 200 rows x 100 transactions x (1 table logical I/O and say 2 index logical writes) per write = 1,200 logical writes to log, of which there are 100 transaction commit synch writes, and in reality less than that because of queuing against logs (there are also 1,200 logical writes deferred to checkpoint, of which it is likely to only be 40 physical writes because of page overlaps). Transaction processing is a spectrum between activity logging, and database centric design. The former, where actions are stored in the database is totally I/O bound with the engine acting as a thin layer of logical to physical mapping. Database centric processing makes the engine a functional server of discrete actions - and is a big CPU hog. What my CPU tends to be doing is a combination of general processing, complex SQL processing: nested loops and sorting and hashing and triggers and SPs. I'm curious about you having flat CPU, which is not my experience. Are your apps mature and stable? /Aaron From pgsql-performance-owner@postgresql.org Tue Apr 6 17:26:01 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3CB01D1E0FD for ; Tue, 6 Apr 2004 17:25:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33440-10 for ; Tue, 6 Apr 2004 17:25:58 -0300 (ADT) Received: from mail4.speakeasy.net (mail4.speakeasy.net [216.254.0.204]) by svr1.postgresql.org (Postfix) with ESMTP id 6450CD1DB0E for ; Tue, 6 Apr 2004 17:25:55 -0300 (ADT) Received: (qmail 27061 invoked from network); 6 Apr 2004 20:25:56 -0000 Received: from dsl093-130-165.sfo4.dsl.speakeasy.net (HELO speakeasy.org) (kgeis@[66.93.130.165]) (envelope-sender ) by mail4.speakeasy.net (qmail-ldap-1.03) with AES256-SHA encrypted SMTP for ; 6 Apr 2004 20:25:56 -0000 Message-ID: <40731252.2030002@speakeasy.org> Date: Tue, 06 Apr 2004 13:25:54 -0700 From: Ken Geis User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: plan problem 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: 200404/70 X-Sequence-Number: 6370 I am trying to find an efficient way to draw a random sample from a complex query. I also want it to be easy to use within my application. So I've defined a view that encapsulates the query. The id in the "driving" table is exposed, and I run a query like: select * from stats_record_view where id in (select id from driver_stats order by random() limit 30000); driver_stats.id is unique, the primary key. The problem I'm having is that neither the ORDER BY nor the LIMIT change the uniqueness of that column, but the planner doesn't know that. It does a HashAggregate to make sure the results are unique. It thinks that 200 rows will come out of that operation, and then 200 rows is small enough that it thinks a Nested Loop is the best way to proceed from there. I can post more query plan, but I don't think it would be that very helpful. I'm considering just making a sample table and creating an analogous view around that. I'd like to be able to keep this as simple as possible though. Ken From pgsql-admin-owner@postgresql.org Tue Apr 6 18:10:10 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8BEDBD1E92A for ; Tue, 6 Apr 2004 18:10:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 51438-10 for ; Tue, 6 Apr 2004 18:10:06 -0300 (ADT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 003A8D1E8FC for ; Tue, 6 Apr 2004 18:10:04 -0300 (ADT) 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 i36LA3Sx094816 for ; Tue, 6 Apr 2004 21:10:03 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id i36LA0Qh094624 for pgsql-admin@postgresql.org; Tue, 6 Apr 2004 21:10:00 GMT From: Chris Browne X-Newsgroups: comp.databases.postgresql.admin Subject: Re: Raw devices vs. Filesystems Date: Tue, 06 Apr 2004 16:57:02 -0400 Organization: cbbrowne Computing Inc Lines: 23 Message-ID: <603c7gj04h.fsf@dev6.int.libertyrms.info> References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@news.hub.org User-Agent: Gnus/5.110002 (No Gnus v0.2) XEmacs/21.4 (Security Through Obscurity, linux) Cancel-Lock: sha1:15plkEF3LHVUmqYr2DqVhI1f/Z0= To: pgsql-admin@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/96 X-Sequence-Number: 13069 gsw@globexplorer.com ("Gregory S. Williamson") writes: > No point to beating a dead horse (other than the sheer joy of the > thing) since postgres does not have raw device support, but ... raw > devices, at least on solaris, are about 10 times as fast as cooked > file systems for Informix. This might still be a gain for postgres' > performance, but the portability issues remain. That claim seems really rather remarkable. It implies an entirely stunning degree of inefficiency in the implementation of filesystems on Solaris. The amount of indirection involved in walking through i-nodes and such is something I would expect to introduce some percentage of performance loss, but for it to introduce overhead of over 900% presumably implies that Sun (and/or Veritas) got something really horribly wrong. -- select 'cbbrowne' || '@' || 'cbbrowne.com'; http://www.ntlug.org/~cbbrowne/nonrdbms.html Rules of the Evil Overlord #1. "My Legions of Terror will have helmets with clear plexiglass visors, not face-concealing ones." From pgsql-performance-owner@postgresql.org Tue Apr 6 18:42:01 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E85CFD1B4F8 for ; Tue, 6 Apr 2004 18:41:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 67522-10 for ; Tue, 6 Apr 2004 18:41:58 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id C40CBD1B905 for ; Tue, 6 Apr 2004 18:41:55 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4785389; Tue, 06 Apr 2004 14:43:17 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: "Aaron Werman" Subject: Re: possible improvement between G4 and G5 Date: Tue, 6 Apr 2004 14:41:53 -0700 User-Agent: KMail/1.4.3 Cc: References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> <200404061152.17997.josh@agliodbs.com> In-Reply-To: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404061441.53598.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/71 X-Sequence-Number: 6371 Aaron, > I do consulting, so they're all over the place and tend to be complex. Very > few fit in RAM, but still are very buffered. These are almost all backed > with very high end I/O subsystems, with dozens of spindles with battery > backed up writethrough cache and gigs of buffers, which may be why I worry > so much about CPU. I have had this issue with multiple servers. Aha, I think this is the difference. I never seem to be able to get my clients to fork out for adequate disk support. They are always running off single or double SCSI RAID in the host server; not the sort of setup you have. > What my CPU tends to be doing is a combination of general processing, > complex SQL processing: nested loops and sorting and hashing and triggers > and SPs. I haven't noticed SPs to be particularly CPU-hoggish, more RAM. > I'm curious about you having flat CPU, which is not my experience. Are your > apps mature and stable? Well, "flat" was a bit of an exaggeration ... there are spikes ... but average CPU load is < 30%. I think the difference is that your clients listen to you about disk access. Mine are all too liable to purchase a quad-Xeon machine but with an Adaptec RAID-5 card with 4 drives, and *then* call me and ask for advice. As a result, most intensive operations don't tend to swamp the CPU because they are waiting for disk. I have noticed the limitiations on RAM for 64 vs. 32, as I find it easier to convince a client to get 8GB RAM than four-channel RAID with 12 drives, mostly because the former is cheaper. Linux 2.4 + Bigmem just doesn't cut it for making effective use of > 3GB of RAM. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Wed Apr 7 01:03:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 66841D1C9BB for ; Wed, 7 Apr 2004 01:03:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 83539-10 for ; Wed, 7 Apr 2004 01:03:54 -0300 (ADT) Received: from mail (unknown [202.108.142.212]) by svr1.postgresql.org (Postfix) with ESMTP id 45522D1C9C4 for ; Wed, 7 Apr 2004 01:03:53 -0300 (ADT) Received: from HUANGYQ ([61.149.154.25]) by mail.gthome.com (iPlanet Messaging Server 5.1 (built May 7 2001)) with ESMTPA id <0HVS00F9M905F7@mail.gthome.com> for pgsql-performance@postgresql.org; Wed, 07 Apr 2004 12:12:07 +0800 (CST) Date: Wed, 07 Apr 2004 12:00:37 +0800 From: huang yaqin Subject: Re: good pc but bad performance,why? To: Tom Lane Cc: "pgsql-performance@postgresql.org" Message-id: <0HVS00F9Q906F7@mail.gthome.com> MIME-version: 1.0 X-Mailer: Foxmail 4.2 [cn] Content-type: text/plain; charset=GB2312 Content-transfer-encoding: quoted-printable X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/72 X-Sequence-Number: 6372 hello=A3=A1 Thanks, you are right. I use "postmaster -o "-F" " to start my PG=A3=ACand performance improved = greatly. Best regards, huang yaqin >huang yaqin writes: >> I have some question when I use postgresql 7.4.1 on redhat adv server 2= .1 . >> I use IBM335 as server, it has 4 cpus, 1G RAM. but I got very bad perfor= mance. >> I can only do about 50 inserts per sencond. Event worse than my pc(PIII = 800,256M RAM), can anyone give me some advice?=09 > >If the cheap machine appears to be able to commit more transactions >per second than the better one, it's very likely because the cheap >machine has a disk that lies about write completion. Is the server >using SCSI disks by any chance? To a first approximation, IDE drives >lie by default, SCSI drives don't. > > regards, tom lane > > > > >Powered by MessageSoft SMG >SPAM, virus-free and secure email >http://www.messagesoft.com =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D= =3D =09=09=09 =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=D6=C2 =C0=F1=A3=A1 =09=09=09=09 =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1huang yaqin =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1hyq@gthome.com =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A12004-04-07 From pgsql-admin-owner@postgresql.org Wed Apr 7 02:26:06 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A6E61D1BB8C for ; Wed, 7 Apr 2004 02:26:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11650-07 for ; Wed, 7 Apr 2004 02:26:05 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 40FD4D1D08F for ; Wed, 7 Apr 2004 02:26:03 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i375Q2PU005720; Wed, 7 Apr 2004 01:26:03 -0400 (EDT) To: Chris Browne Cc: pgsql-admin@postgresql.org Subject: Re: Raw devices vs. Filesystems In-reply-to: <603c7gj04h.fsf@dev6.int.libertyrms.info> References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <603c7gj04h.fsf@dev6.int.libertyrms.info> Comments: In-reply-to Chris Browne message dated "Tue, 06 Apr 2004 16:57:02 -0400" Date: Wed, 07 Apr 2004 01:26:02 -0400 Message-ID: <5719.1081315562@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/100 X-Sequence-Number: 13073 Chris Browne writes: > That claim seems really rather remarkable. > It implies an entirely stunning degree of inefficiency in the > implementation of filesystems on Solaris. Solaris has a reputation for having stunning degrees of inefficiency in a number of places :-(. On the other hand I've also heard it praised for its ability to survive partial hardware failures (eg, N out of M CPUs down), so maybe that's the price you gotta pay. But to get back to the point of this discussion: to allow PG to use raw devices instead of filesystems, we'd first have to do a ton of portability work (since raw disk access is nowhere standard), and abandon our principle that Postgres does not run as root (since raw disk access is not permitted to non-root processes by any sane sysadmin). But that last is a mighty comforting principle to have, anytime someone complains that their el cheapo whitebox PC locks up as soon as they start to stress the database. I know I'd have wasted a lot more time chasing random hardware breakages if I couldn't say "system freezes and filesystem corruption are Clearly Not Our Fault". After that, we get to implement our own filesystem-equivalent management of disk space allocation, disk I/O scheduling, etc. Are we really smarter than all those kernel hackers doing this for a living? I doubt it. After that, we get to re-optimize all the existing Postgres behaviors that are designed to sit on top of a standard Unix buffering filesystem layer. After that, we might reap some performance benefits. Or maybe not. There's not a heck of a lot of hard evidence that we would --- and what there is traces to twenty-year-old assumptions about disk drive and OS behavior, which are quite unlikely to still apply today. Personally, I have a lot of more-promising projects to pursue... regards, tom lane From pgsql-performance-owner@postgresql.org Wed Apr 7 04:15:33 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C1646D1BB16 for ; Wed, 7 Apr 2004 04:15:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50922-01 for ; Wed, 7 Apr 2004 04:15:28 -0300 (ADT) Received: from ol-mailrelay-01.fh-oldenburg.de (ol-mailrelay-01.fh-oldenburg.de [139.13.186.99]) by svr1.postgresql.org (Postfix) with ESMTP id 74853D1BAD6 for ; Wed, 7 Apr 2004 04:15:25 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by ol-mailrelay-01.fh-oldenburg.de (Postfix) with ESMTP id 148FA38C3F7 for ; Wed, 7 Apr 2004 09:06:42 +0200 (CEST) Received: from ol-mailrelay-01.fh-oldenburg.de ([127.0.0.1]) by localhost (ol-mailrelay-01 [127.0.0.1]) (amavisd-new, port 10023) with ESMTP id 24456-03 for ; Wed, 7 Apr 2004 09:06:41 +0200 (CEST) Received: from webmail.fh-oldenburg.de (ol-mailrelay-01.fh-oldenburg.de [139.13.186.99]) by ol-mailrelay-01.fh-oldenburg.de (Postfix) with SMTP id B95F538C3F4 for ; Wed, 7 Apr 2004 09:06:41 +0200 (CEST) Received: from 195.243.253.146 (SquirrelMail authenticated user kehlenbh) by webmail.fh-oldenburg.de with HTTP; Wed, 7 Apr 2004 09:06:41 +0200 (CEST) Message-ID: <39651.195.243.253.146.1081321601.squirrel@webmail.fh-oldenburg.de> Date: Wed, 7 Apr 2004 09:06:41 +0200 (CEST) Subject: Re: performance comparission postgresql/ms-sql server From: "Heiko Kehlenbrink" To: pgsql-performance@postgresql.org User-Agent: SquirrelMail/1.4.2 MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 Importance: Normal X-Virus-Scanned: by amavisd-new X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/73 X-Sequence-Number: 6373 hi shridhar, > Heiko Kehlenbrink wrote: > >> hi list, >> >> i want to convince people to use postgresql instead of ms-sql server, so i >> set up a kind of comparission insert data / select data from postgresql / >> ms-sql server >> >> the table i use was pretty basic, >> >> id bigserial >> dist float8 >> x float8 >> y float8 >> z float8 >> >> i filled the table with a function which filled x,y,z with incremental increasing values (1,2,3,4,5,6...) and computing from that the dist value >> for every tupel (sqrt((x*x)+(y*y)+(z*z))). >> >> this works fine for both dbms >> >> postgresql needs 13:37 min for 10.000.000 tupel, >> ms-sql needs 1:01:27 h for 10.000.000 tupel. >> >> so far so good. >> >> i attached an index on the dist row and started to query the dbs with scripts which select a serial row of 100.000,200.000,500.000 tupels based >> on the dist row. >> i randomizly compute the start and the end distance and made a "select avg(dist) from table where dist > startdist and dist < enddist" > > Some basics to check quickly. > > 1. vacuum analyze the table before you start selecting. was done, > 2. for slow running queries, check explain analyze output and find out who takes > maximum time. hkehlenbrink@lin0493l:~> psql -d test -c 'explain analyse select avg(dist) from massive2 where dist > (1000000*sqrt(3.0))::float8 and dist < (1500000*sqrt(3.0))::float8;' NOTICE: QUERY PLAN: Aggregate (cost=14884.61..14884.61 rows=1 width=8) (actual time=3133.24..3133.24 rows=1 loops=1) -> Index Scan using massive2_dist on massive2 (cost=0.00..13648.17 rows=494573 width=8) (actual time=0.11..2061.38 rows=499999 loops=1) Total runtime: 3133.79 msec EXPLAIN seems to me that most time was needed for the index scanning... > 3. Check for typecasting. You need to typecast the query correctly e.g. > > select avg(dist) from table where dist >startdist::float8 and > dist > This might still end up with sequential scan depending upon the plan. but if > index scan is picked up, it might be plenty fast.. > nope, the dist row is float8 and the query-borders are float8 too, also the explain says that an index scann was done. > Post explain analyze for the queries if things don't improve. > see above.. > HTH > > Shridhar > best regards heiko > > From pgsql-admin-owner@postgresql.org Wed Apr 7 04:19:05 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7F4ECD1B911; Wed, 7 Apr 2004 04:19:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 49595-06; Wed, 7 Apr 2004 04:19:02 -0300 (ADT) Received: from elbereth.noviforum.si (unknown [193.189.169.66]) by svr1.postgresql.org (Postfix) with ESMTP id 5A853D1B515; Wed, 7 Apr 2004 04:19:00 -0300 (ADT) Received: from elbereth.noviforum.si (localhost [127.0.0.1]) by elbereth.noviforum.si (8.12.10/8.12.10) with ESMTP id i377IxHh014859; Wed, 7 Apr 2004 09:18:59 +0200 Received: (from gregab@localhost) by elbereth.noviforum.si (8.12.10/8.12.10/Submit) id i377IwPq014845; Wed, 7 Apr 2004 09:18:58 +0200 Date: Wed, 7 Apr 2004 09:18:58 +0200 From: Grega Bremec To: Tom Lane Cc: Chris Browne , pgsql-admin@postgresql.org, pgsql-performance@postgresql.org Subject: Re: Raw devices vs. Filesystems Message-ID: <20040407071858.GA7973@elbereth.noviforum.si> References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <603c7gj04h.fsf@dev6.int.libertyrms.info> <5719.1081315562@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="k1lZvvs/B4yU6o8G" Content-Disposition: inline In-Reply-To: <5719.1081315562@sss.pgh.pa.us> User-Agent: Mutt/1.4.2i Organization: Noviforum, Ltd., Software & Media X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/102 X-Sequence-Number: 13075 --k1lZvvs/B4yU6o8G Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable ...and on Wed, Apr 07, 2004 at 01:26:02AM -0400, Tom Lane used the keyboard: >=20 > After that, we get to implement our own filesystem-equivalent management > of disk space allocation, disk I/O scheduling, etc. Are we really > smarter than all those kernel hackers doing this for a living? I doubt i= t. >=20 > After that, we get to re-optimize all the existing Postgres behaviors > that are designed to sit on top of a standard Unix buffering filesystem > layer. >=20 > After that, we might reap some performance benefits. Or maybe not. > There's not a heck of a lot of hard evidence that we would --- and > what there is traces to twenty-year-old assumptions about disk drive > and OS behavior, which are quite unlikely to still apply today. >=20 > Personally, I have a lot of more-promising projects to pursue... >=20 Has anyone tried PostgreSQL on top of OCFS? Personally, I'm not sure it would even work, as Oracle clearly state that OCFS was _never_ meant to be a fully fledged UNIX filesystem with POSIX features such as correct timestamp updates, inode changes, etc., but OCFSv2 brings some features that might lead one into thinking they're about to make it suitable for uses beyond that of just having Oracle databases sitting on top of it. Furthermore, this filesystem would be a blazing one stop solution for all replication issues PostgreSQL currently suffers from, as its main design goal was to present "a consistent file system image across the servers in a cluster". Now, if both goals can be achieved in one go, hell, I'm willing to try it out myself in an attempt to extract off of it, some performance indicators that could be compared to other database performance tests sent to both this and the PERFORM mailing list. So, anyone? :) Cheers, --=20 Grega Bremec Senior Administrator Noviforum Ltd., Software & Media http://www.noviforum.si/ --k1lZvvs/B4yU6o8G Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (GNU/Linux) iD8DBQFAc6tiDo/EMYD4+osRAu+xAJ4t6r9JMoTTgMH2q5I3ab6A0pqRUwCfU8Ky o6Pu6BIKTMZw/dfgbdcY47M= =029G -----END PGP SIGNATURE----- --k1lZvvs/B4yU6o8G-- From pgsql-performance-owner@postgresql.org Wed Apr 7 05:33:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 48163D1B97B for ; Wed, 7 Apr 2004 05:33:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 76754-06 for ; Wed, 7 Apr 2004 05:33:17 -0300 (ADT) 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 4F96CD1B515 for ; Wed, 7 Apr 2004 05:33:15 -0300 (ADT) 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 1BB8Um-000Lzg-0X; Wed, 07 Apr 2004 09:33:16 +0100 Received: by mainbox.archonet.com (Postfix, from userid 529) id 0B831165B0; Wed, 7 Apr 2004 09:33:14 +0100 (BST) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 1C7B21643F; Wed, 7 Apr 2004 09:33:13 +0100 (BST) From: Richard Huxton To: huang yaqin , Tom Lane Subject: Re: good pc but bad performance,why? Date: Wed, 7 Apr 2004 09:33:12 +0100 User-Agent: KMail/1.5 Cc: "pgsql-performance@postgresql.org" References: <0HVS00F9Q906F7@mail.gthome.com> In-Reply-To: <0HVS00F9Q906F7@mail.gthome.com> MIME-Version: 1.0 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Message-Id: <200404070933.12544.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: 200404/75 X-Sequence-Number: 6375 On Wednesday 07 April 2004 05:00, huang yaqin wrote: > hello=A3=A1 > > Thanks, you are right. > I use "postmaster -o "-F" " to start my PG=A3=ACand performance improved > greatly. I don't think Tom was recommending turning fsync off. If you have a system= =20 crash/power glitch then the database can become corrupted. If you are happy the possibility if losing your data, write performance wil= l=20 improve noticably. --=20 Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Wed Apr 7 05:38:17 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4EE24D1B459 for ; Wed, 7 Apr 2004 05:38:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 82267-01 for ; Wed, 7 Apr 2004 05:38:16 -0300 (ADT) 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 EF093D1C4C4 for ; Wed, 7 Apr 2004 05:38:13 -0300 (ADT) 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 1BB8Zb-000CTT-0Y; Wed, 07 Apr 2004 09:38:15 +0100 Received: by mainbox.archonet.com (Postfix, from userid 529) id 1A17A1643F; Wed, 7 Apr 2004 09:38:14 +0100 (BST) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 19E3A1643F; Wed, 7 Apr 2004 09:38:12 +0100 (BST) From: Richard Huxton To: Ken Geis , pgsql-performance@postgresql.org Subject: Re: plan problem Date: Wed, 7 Apr 2004 09:38:11 +0100 User-Agent: KMail/1.5 References: <40731252.2030002@speakeasy.org> In-Reply-To: <40731252.2030002@speakeasy.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404070938.11482.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: 200404/76 X-Sequence-Number: 6376 On Tuesday 06 April 2004 21:25, Ken Geis wrote: > I am trying to find an efficient way to draw a random sample from a > complex query. I also want it to be easy to use within my application. > > So I've defined a view that encapsulates the query. The id in the > "driving" table is exposed, and I run a query like: > > select * from stats_record_view > where id in (select id from driver_stats > order by random() > limit 30000); How about a join? SELECT s.* FROM stats_record_view s JOIN (SELECT id FROM driver_stats ORDER BY random() LIMIT 30000) AS r ON s.id = r.id; Or, what about a cursor and fetch forward (or back?) a random number of rows before each fetch. That's probably not going to be so random though. Also worth checking the various list archives - this has come up in the past, but some time ago. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Wed Apr 7 06:00:44 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 024B5D1C9C4 for ; Wed, 7 Apr 2004 06:00:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 91514-02 for ; Wed, 7 Apr 2004 06:00:37 -0300 (ADT) Received: from mail (unknown [202.108.142.212]) by svr1.postgresql.org (Postfix) with ESMTP id 5DB39D1C995 for ; Wed, 7 Apr 2004 06:00:34 -0300 (ADT) Received: from HUANGYQ ([61.149.145.64]) by mail.gthome.com (iPlanet Messaging Server 5.1 (built May 7 2001)) with ESMTPA id <0HVS00G7EMQG35@mail.gthome.com> for pgsql-performance@postgresql.org; Wed, 07 Apr 2004 17:08:49 +0800 (CST) Date: Wed, 07 Apr 2004 16:56:56 +0800 From: huang yaqin Subject: Re: good pc but bad performance,why? To: Richard Huxton Cc: "pgsql-performance@postgresql.org" Message-id: <0HVS00G7FMQO35@mail.gthome.com> MIME-version: 1.0 X-Mailer: Foxmail 4.2 [cn] Content-type: text/plain; charset=GB2312 Content-transfer-encoding: quoted-printable X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/77 X-Sequence-Number: 6377 Hello, Richard Huxton, =09 You said turning fsync off may cause losing data, that's terrible. I use SCSI disk, and file system is ext3. I tune postgresql.conf and can't= get any improvement. So what can I do? Does SCSI disk and IDE disk have difference? Regards, Huang yaqin =3D=3D=3D=3D=3D=3D=3D 2004-04-07 09:33:00 =3D=3D=3D=3D=3D=3D=3D >On Wednesday 07 April 2004 05:00, huang yaqin wrote: >> hello=A3=A1 >> >> Thanks, you are right. >> I use "postmaster -o "-F" " to start my PG=A3=ACand performance improv= ed >> greatly. > >I don't think Tom was recommending turning fsync off. If you have a system >crash/power glitch then the database can become corrupted. > >If you are happy the possibility if losing your data, write performance wi= ll >improve noticably. > >-- > Richard Huxton > Archonet Ltd > > > > >Powered by MessageSoft SMG >SPAM, virus-free and secure email >http://www.messagesoft.com > >. =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D =3D= =3D =09=09=09 =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=D6=C2 =C0=F1=A3=A1 =09=09=09=09 =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1huang yaqin =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1hyq@gthome.com =A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A1=A12004-04-07 From pgsql-performance-owner@postgresql.org Wed Apr 7 06:03:33 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B27B3D1CC93 for ; Wed, 7 Apr 2004 06:03:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89770-08 for ; Wed, 7 Apr 2004 06:03:30 -0300 (ADT) Received: from mail2.speakeasy.net (mail2.speakeasy.net [216.254.0.202]) by svr1.postgresql.org (Postfix) with ESMTP id 56F76D1CACC for ; Wed, 7 Apr 2004 06:03:27 -0300 (ADT) Received: (qmail 15087 invoked from network); 7 Apr 2004 09:03:28 -0000 Received: from dsl093-130-165.sfo4.dsl.speakeasy.net (HELO speakeasy.org) (kgeis@[66.93.130.165]) (envelope-sender ) by mail2.speakeasy.net (qmail-ldap-1.03) with AES256-SHA encrypted SMTP for ; 7 Apr 2004 09:03:28 -0000 Message-ID: <4073C3DF.2070307@speakeasy.org> Date: Wed, 07 Apr 2004 02:03:27 -0700 From: Ken Geis User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031007 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton , pgsql-performance@postgresql.org Subject: Re: plan problem References: <40731252.2030002@speakeasy.org> <200404070938.11482.dev@archonet.com> In-Reply-To: <200404070938.11482.dev@archonet.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: 200404/78 X-Sequence-Number: 6378 Richard Huxton wrote: > On Tuesday 06 April 2004 21:25, Ken Geis wrote: > >>I am trying to find an efficient way to draw a random sample from a >>complex query. I also want it to be easy to use within my application. >> >>So I've defined a view that encapsulates the query. The id in the >>"driving" table is exposed, and I run a query like: >> >>select * from stats_record_view >> where id in (select id from driver_stats >> order by random() >> limit 30000); > > > How about a join? > > SELECT s.* > FROM > stats_record_view s > JOIN > (SELECT id FROM driver_stats ORDER BY random() LIMIT 30000) AS r > ON s.id = r.id; Yes, I tried this too after I sent the first mail, and this was somewhat better. I ended up adding a random column to the driving table, putting an index on it, and exposing that column in the view. Now I can say SELECT * FROM stats_record_view WHERE random < 0.093; For my application, it's OK if the same sample is picked time after time and it may change if data is added. ... > Also worth checking the various list archives - this has come up in the past, > but some time ago. There are some messages in the archives about how to get a random sample. I know how to do that, and that's not why I posted my message. Are you saying that the planner behavior I spoke of is in the archives? I wouldn't know what to search on to find that thread. Does anyone think that the planner issue has merit to address? Can someone help me figure out what code I would look at? Ken Geis From pgsql-performance-owner@postgresql.org Wed Apr 7 06:54:01 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2B889D1B893 for ; Wed, 7 Apr 2004 06:53:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08926-07 for ; Wed, 7 Apr 2004 06:54:00 -0300 (ADT) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 5F50FD1BA2E for ; Wed, 7 Apr 2004 06:53:57 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 79E8586FA; Wed, 7 Apr 2004 11:53:59 +0200 (CEST) Date: Wed, 7 Apr 2004 11:53:59 +0200 (CEST) From: Dennis Bjorklund To: huang yaqin Cc: Richard Huxton , "pgsql-performance@postgresql.org" Subject: Re: good pc but bad performance,why? In-Reply-To: <0HVS00G7FMQO35@mail.gthome.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/79 X-Sequence-Number: 6379 On Wed, 7 Apr 2004, huang yaqin wrote: > You said turning fsync off may cause losing data, that's terrible. I use > SCSI disk, and file system is ext3. I tune postgresql.conf and can't get > any improvement. So what can I do? Make sure you do as much as possible inside one transaction. If you want to do 1000 inserts, then do BEGIN; insert ....; insert; ... ; COMMIT; -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Wed Apr 7 07:41:19 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 36660D1C9F3 for ; Wed, 7 Apr 2004 07:41:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28071-04 for ; Wed, 7 Apr 2004 07:41:13 -0300 (ADT) Received: from gizmo08bw.bigpond.com (gizmo08bw.bigpond.com [144.140.70.18]) by svr1.postgresql.org (Postfix) with SMTP id 8F672D1BCB8 for ; Wed, 7 Apr 2004 07:41:09 -0300 (ADT) Received: (qmail 17836 invoked from network); 7 Apr 2004 10:31:59 -0000 Received: from unknown (HELO bwmam08.bigpond.com) (144.135.24.91) by gizmo08bw.bigpond.com with SMTP; 7 Apr 2004 10:31:59 -0000 Received: from cpe-203-45-197-156.qld.bigpond.net.au ([203.45.197.156]) by bwmam08.bigpond.com(MAM REL_3_4_2 135/13709782) with SMTP id 13709782; Wed, 07 Apr 2004 20:41:09 +1000 Message-ID: <001c01c41c8c$993eded0$fd08a8c0@steve> From: "Steven Butler" To: "huang yaqin" Cc: References: <0HVS00G7FMQO35@mail.gthome.com> Subject: Re: good pc but bad performance,why? Date: Wed, 7 Apr 2004 20:39:25 +1000 MIME-Version: 1.0 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1158 X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/80 X-Sequence-Number: 6380 It sounds almost like you're doing one insert per transaction. Try wrapping multiple inserts into a single transaction and see if that helps. This may not be appropriate for your application, but it does guarantee that committed transactions will not be lost. My apologies if you are already doing this. :) BEGIN; insert ... insert ... insert ... COMMIT; Regards, Steve Butler ----- Original Message ----- From: "huang yaqin" To: "Richard Huxton" Cc: Sent: Wednesday, April 07, 2004 6:56 PM Subject: Re: [PERFORM] good pc but bad performance,why? Hello, Richard Huxton, You said turning fsync off may cause losing data, that's terrible. I use SCSI disk, and file system is ext3. I tune postgresql.conf and can't get any improvement. So what can I do? Does SCSI disk and IDE disk have difference? Regards, Huang yaqin From pgsql-performance-owner@postgresql.org Wed Apr 7 08:30:15 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 73BC0D1D05D for ; Wed, 7 Apr 2004 08:30:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 41465-09 for ; Wed, 7 Apr 2004 08:30:12 -0300 (ADT) Received: from lamb.mcmillan.net.nz (218-101-13-8.paradise.net.nz [218.101.13.8]) by svr1.postgresql.org (Postfix) with ESMTP id 04890D1BAD5 for ; Wed, 7 Apr 2004 08:30:09 -0300 (ADT) Received: from lamb.mcmillan.net.nz (lamb.mcmillan.net.nz [127.0.0.1]) by lamb.mcmillan.net.nz (Postfix) with ESMTP id 4F737AD98581; Wed, 7 Apr 2004 23:29:43 +1200 (NZST) Subject: Re: good pc but bad performance,why? From: Andrew McMillan To: huang yaqin Cc: pgsql-performance@postgresql.org In-Reply-To: <0HVS00G7FMQO35@mail.gthome.com> References: <0HVS00G7FMQO35@mail.gthome.com> Content-Type: text/plain Message-Id: <1081337381.522.57.camel@lamb.mcmillan.net.nz> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Wed, 07 Apr 2004 23:29:42 +1200 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/81 X-Sequence-Number: 6381 On Wed, 2004-04-07 at 20:56, huang yaqin wrote: > Hello, Richard Huxton, > > You said turning fsync off may cause losing data, that's terrible. > I use SCSI disk, and file system is ext3. I tune postgresql.conf and can't get any improvement. So what can I do? > Does SCSI disk and IDE disk have difference? Yes, turning off fsync means that the database is not guaranteeing consistency of writes to disk any longer. On the other hand your IDE system probably never was, because IDE drives just typically turn on write caching in hardware without telling anyone. SCSI typically doesn't turn on write caching in the physical drive by default, as Tom Lane pointed out earlier. Good SCSI has a battery backed up cache, and then it is OK to turn on write caching, because the controller has enough battery to complete all writes in the event of a power failure. One thing I recommend is to use ext2 (or almost anything but ext3). There is no real need (or benefit) from having the database on a journalled filesystem - the journalling is only trying to give similar sorts of guarantees to what the fsync in PostgreSQL is doing. The suggestion someone else made regarding use of software raid is probably also a good one if you are trying to use the on-board RAID at the moment. Finally, I would say that because you are seeing poor performance on one box and great performance on another, you should look at the hardware, or at the hardware drivers, for the problem - not so much at PostgreSQL. Of course if it is application performance you want to achieve, we _can_ help here, but you will need to provide more details of what you are trying to do in your application, including; - confirmation that you have done a VACUUM and ANALYZE of all tables before you start - output from EXPLAIN ANALYZE for slow queries - anything else you think is useful. without that sort of detail we can only give vague suggestions, like "wrap everything in a transaction" - excellent advice, certainly, but you can read that in the manual. There are no magic bullets, but I am sure most of the people on this list have systems that regularly do way more than 50 inserts / second on server hardware. Regards, Andrew McMillan ------------------------------------------------------------------------- Andrew @ Catalyst .Net .NZ Ltd, PO Box 11-053, Manners St, Wellington WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St DDI: +64(4)916-7201 MOB: +64(21)635-694 OFFICE: +64(4)499-2267 http://survey.net.nz/ - any questions? ------------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Fri Apr 9 21:25:44 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E0641D1D066 for ; Wed, 7 Apr 2004 08:51:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 51172-08 for ; Wed, 7 Apr 2004 08:51:50 -0300 (ADT) Received: from smtp.pspl.co.in (unknown [202.54.11.65]) by svr1.postgresql.org (Postfix) with ESMTP id 116B2D1C50B for ; Wed, 7 Apr 2004 08:51:45 -0300 (ADT) Received: (from root@localhost) by smtp.pspl.co.in (8.12.9/8.12.9) id i37BpuoO026414 for ; Wed, 7 Apr 2004 17:21:56 +0530 X-Scanned: XAM28854 Scanned by XAMIME Received: from ps0499.intranet.pspl.co.in (PS0499.intranet.pspl.co.in [192.168.10.204]) (authenticated bits=0) by persistent.co.in (8.12.9/8.12.9) with ESMTP id i37Bpr26026372; Wed, 7 Apr 2004 17:21:54 +0530 From: Shridhar Daithankar To: Andrew McMillan Subject: Re: good pc but bad performance,why? Date: Wed, 7 Apr 2004 17:21:43 +0530 User-Agent: KMail/1.6.1 Cc: huang yaqin , pgsql-performance@postgresql.org References: <0HVS00G7FMQO35@mail.gthome.com> <1081337381.522.57.camel@lamb.mcmillan.net.nz> In-Reply-To: <1081337381.522.57.camel@lamb.mcmillan.net.nz> MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <200404071721.43059.shridhar_daithankar@persistent.co.in> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/124 X-Sequence-Number: 6424 On Wednesday 07 April 2004 16:59, Andrew McMillan wrote: > One thing I recommend is to use ext2 (or almost anything but ext3). > There is no real need (or benefit) from having the database on a > journalled filesystem - the journalling is only trying to give similar > sorts of guarantees to what the fsync in PostgreSQL is doing. That is not correct assumption. A journalling file system ensures file system consistency even at a cost of loss of some data. And postgresql can not guarantee recovery if WAL logs are corrupt. Some months back, there was a case reported where ext2 corrupted WAL and database. BAckup is only solution then.. Journalling file systems are usually very close to ext2 in performance, many a times lot better. With ext2, you are buying a huge risk. Unless there are good reason, I would not put a database on ext2. Performance isn't one ofthem.. Shridhar From pgsql-performance-owner@postgresql.org Wed Apr 7 08:54:45 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B46BDD1C4C4 for ; Wed, 7 Apr 2004 08:54:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55872-05 for ; Wed, 7 Apr 2004 08:54:45 -0300 (ADT) Received: from frodo.hserus.net (frodo.hserus.net [204.74.68.40]) by svr1.postgresql.org (Postfix) with ESMTP id CA8A3D1BC5C for ; Wed, 7 Apr 2004 08:54:41 -0300 (ADT) Received: from concord.pspl.co.in ([202.54.11.72]:61805 helo=ps0499.intranet.pspl.co.in) by frodo.hserus.net with asmtp (Cipher TLSv1:RC4-MD5:128) (Exim 4.31 #0) id 1BBBdk-000IKR-2P by authid with plain for ; Wed, 07 Apr 2004 17:24:44 +0530 From: Shridhar Daithankar Reply-To: shridhar@frodo.hserus.net To: pgsql-performance@postgresql.org Subject: Re: good pc but bad performance,why? Date: Wed, 7 Apr 2004 17:24:41 +0530 User-Agent: KMail/1.6.1 References: <0HVS00G7FMQO35@mail.gthome.com> <1081337381.522.57.camel@lamb.mcmillan.net.nz> <200404071721.43059.shridhar_daithankar@persistent.co.in> In-Reply-To: <200404071721.43059.shridhar_daithankar@persistent.co.in> MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <200404071724.41860.shridhar@frodo.hserus.net> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/82 X-Sequence-Number: 6382 Sending again bacuse of MUA error.. Chose a wrong address in From..:-( Shridhar On Wednesday 07 April 2004 17:21, Shridhar Daithankar wrote: > On Wednesday 07 April 2004 16:59, Andrew McMillan wrote: > > One thing I recommend is to use ext2 (or almost anything but ext3). > > There is no real need (or benefit) from having the database on a > > journalled filesystem - the journalling is only trying to give similar > > sorts of guarantees to what the fsync in PostgreSQL is doing. > > That is not correct assumption. A journalling file system ensures file > system consistency even at a cost of loss of some data. And postgresql can > not guarantee recovery if WAL logs are corrupt. Some months back, there was > a case reported where ext2 corrupted WAL and database. BAckup is only > solution then.. > > Journalling file systems are usually very close to ext2 in performance, > many a times lot better. With ext2, you are buying a huge risk. > > Unless there are good reason, I would not put a database on ext2. > Performance isn't one ofthem.. > > Shridhar From pgsql-performance-owner@postgresql.org Wed Apr 7 09:31:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 16C76D1D064 for ; Wed, 7 Apr 2004 09:31:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73273-03 for ; Wed, 7 Apr 2004 09:31:35 -0300 (ADT) 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 5F09FD1BC5C for ; Wed, 7 Apr 2004 09:31:31 -0300 (ADT) 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 1BBCDO-000BEf-0X; Wed, 07 Apr 2004 13:31:34 +0100 Received: by mainbox.archonet.com (Postfix, from userid 529) id 2EB95165A8; Wed, 7 Apr 2004 13:31:33 +0100 (BST) Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 922E11639A; Wed, 7 Apr 2004 13:31:30 +0100 (BST) From: Richard Huxton To: Ken Geis , pgsql-performance@postgresql.org Subject: Re: plan problem Date: Wed, 7 Apr 2004 13:31:30 +0100 User-Agent: KMail/1.5 References: <40731252.2030002@speakeasy.org> <200404070938.11482.dev@archonet.com> <4073C3DF.2070307@speakeasy.org> In-Reply-To: <4073C3DF.2070307@speakeasy.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404071331.30925.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: 200404/83 X-Sequence-Number: 6383 On Wednesday 07 April 2004 10:03, Ken Geis wrote: > Richard Huxton wrote: > > On Tuesday 06 April 2004 21:25, Ken Geis wrote: > >>I am trying to find an efficient way to draw a random sample from a > >>complex query. I also want it to be easy to use within my application. > >> > >>So I've defined a view that encapsulates the query. The id in the > >>"driving" table is exposed, and I run a query like: > >> > >>select * from stats_record_view > >> where id in (select id from driver_stats > >> order by random() > >> limit 30000); > > > > How about a join? > > > > SELECT s.* > > FROM > > stats_record_view s > > JOIN > > (SELECT id FROM driver_stats ORDER BY random() LIMIT 30000) AS r > > ON s.id = r.id; > > Yes, I tried this too after I sent the first mail, and this was somewhat > better. I ended up adding a random column to the driving table, putting > an index on it, and exposing that column in the view. Now I can say > > SELECT * FROM stats_record_view WHERE random < 0.093; > > For my application, it's OK if the same sample is picked time after time > and it may change if data is added. Fair enough - that'll certainly do it. > > Also worth checking the various list archives - this has come up in the > > past, but some time ago. > > There are some messages in the archives about how to get a random > sample. I know how to do that, and that's not why I posted my message. > Are you saying that the planner behavior I spoke of is in the > archives? I wouldn't know what to search on to find that thread. Does > anyone think that the planner issue has merit to address? Can someone > help me figure out what code I would look at? I was assuming after getting a random subset they'd see the same problem you are. If not, probably worth looking at. In which case, an EXPLAIN ANALYZE of your original query would be good. -- Richard Huxton Archonet Ltd From pgsql-admin-owner@postgresql.org Wed Apr 7 10:10:41 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 125CAD1B459 for ; Wed, 7 Apr 2004 10:10:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 94333-01 for ; Wed, 7 Apr 2004 10:10:41 -0300 (ADT) Received: from main.gmane.org (main.gmane.org [80.91.224.249]) by svr1.postgresql.org (Postfix) with ESMTP id 5E5EED1CB3F for ; Wed, 7 Apr 2004 10:10:37 -0300 (ADT) Received: from root by main.gmane.org with local (Exim 3.35 #1 (Debian)) id 1BBCpE-0001tD-00 for ; Wed, 07 Apr 2004 15:10:40 +0200 Received: from srv.protecting.net ([212.126.218.242]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Wed, 07 Apr 2004 15:10:40 +0200 Received: from hf320 by srv.protecting.net with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Wed, 07 Apr 2004 15:10:40 +0200 X-Injected-Via-Gmane: http://gmane.org/ To: pgsql-admin@postgresql.org From: Harald Fuchs Subject: Re: Raw devices vs. Filesystems Date: 07 Apr 2004 15:05:55 +0200 Organization: Linux Private Site Lines: 19 Message-ID: References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <603c7gj04h.fsf@dev6.int.libertyrms.info> <5719.1081315562@sss.pgh.pa.us> Reply-To: hf320@protecting.net Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII X-Complaints-To: usenet@sea.gmane.org X-Gmane-NNTP-Posting-Host: srv.protecting.net X-No-Archive: yes User-Agent: Gnus/5.0808 (Gnus v5.8.8) Emacs/20.7 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/105 X-Sequence-Number: 13078 In article <5719.1081315562@sss.pgh.pa.us>, Tom Lane writes: > But to get back to the point of this discussion: to allow PG to use raw > devices instead of filesystems, we'd first have to do a ton of > portability work (since raw disk access is nowhere standard), and > abandon our principle that Postgres does not run as root (since raw disk > access is not permitted to non-root processes by any sane sysadmin). Why not? In MySQL/InnoDB, you do a "chown mysql.daemon /dev/raw/raw1" (or whatever raw disk you want to access), and that's all. > After that, we get to implement our own filesystem-equivalent management > of disk space allocation, disk I/O scheduling, etc. Are we really > smarter than all those kernel hackers doing this for a living? I doubt it. Ditto. I don't have hard numbers for MySQL, but I didn't see any noticeable improvement when messing with raw disks (at least under Linux). From pgsql-performance-owner@postgresql.org Wed Apr 7 11:36:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 947C6D1CA50 for ; Wed, 7 Apr 2004 11:36:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27142-09 for ; Wed, 7 Apr 2004 11:36:46 -0300 (ADT) Received: from hotmail.com (law10-oe60.law10.hotmail.com [64.4.14.195]) by svr1.postgresql.org (Postfix) with ESMTP id BF362D1BAD5 for ; Wed, 7 Apr 2004 11:36:41 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Wed, 7 Apr 2004 07:36:45 -0700 Received: from 67.81.102.201 by law10-oe60.law10.hotmail.com with DAV; Wed, 07 Apr 2004 14:36:45 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman2@hotmail.com] X-Sender: awerman2@hotmail.com From: "Aaron Werman" Cc: References: <0HVS00F9Q906F7@mail.gthome.com> <200404070933.12544.dev@archonet.com> Subject: Re: good pc but bad performance,why? Date: Wed, 7 Apr 2004 10:36:48 -0400 MIME-Version: 1.0 Content-Type: text/plain; charset="GB2312" Content-Transfer-Encoding: 8bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Message-ID: X-OriginalArrivalTime: 07 Apr 2004 14:36:45.0889 (UTC) FILETIME=[C0916310:01C41CAD] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/84 X-Sequence-Number: 6384 fsync I'm thinking 50 inserts, if autocommiting is 50TPS = ~100 IO per second (50 WAL + checkpoint IO) = roughly the I/O rate of a single drive. Huang - Are you using a single drive for pg? If so, there is a safety problem of both the data and logs used for recovery on the same drive. If the drive crashes, there is nothing left for recovery. Also, there is a big contention issue, since the log is a fast sequential write, and checkpointing is random. If the log is on a separate drive, you'll probably see insert speed at disk sequential write speed, since the other drive(s) should hopefully be able to keep up when checkpointing. If they share the same drive, you'll see an initial burst of inserts, then a order of magnitude performance drop-off as soon as you checkpoint - because the disk is interleaving the log and data writes. fsync off is only appropriate for externally recoverable processes, such as loading an empty server from a file. /Aaron ----- Original Message ----- From: "Richard Huxton" To: "huang yaqin" ; "Tom Lane" Cc: Sent: Wednesday, April 07, 2004 4:33 AM Subject: Re: [PERFORM] good pc but bad performance,why? On Wednesday 07 April 2004 05:00, huang yaqin wrote: > hello�� > > Thanks, you are right. > I use "postmaster -o "-F" " to start my PG��and performance improved > greatly. I don't think Tom was recommending turning fsync off. If you have a system crash/power glitch then the database can become corrupted. If you are happy the possibility if losing your data, write performance will improve noticably. -- Richard Huxton Archonet Ltd ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster From pgsql-admin-owner@postgresql.org Wed Apr 7 13:09:35 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 06B67D1B9EB; Wed, 7 Apr 2004 13:09:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 79265-01; Wed, 7 Apr 2004 13:09:28 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id A5209D1B989; Wed, 7 Apr 2004 13:09:26 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4790870; Wed, 07 Apr 2004 09:10:47 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Grega Bremec , Tom Lane Subject: Re: [PERFORM] Raw devices vs. Filesystems Date: Wed, 7 Apr 2004 09:09:16 -0700 User-Agent: KMail/1.4.3 Cc: Chris Browne , pgsql-admin@postgresql.org, pgsql-performance@postgresql.org References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <5719.1081315562@sss.pgh.pa.us> <20040407071858.GA7973@elbereth.noviforum.si> In-Reply-To: <20040407071858.GA7973@elbereth.noviforum.si> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404070909.16123.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/109 X-Sequence-Number: 13082 Grega, > Furthermore, this filesystem would be a blazing one stop solution for > all replication issues PostgreSQL currently suffers from, as its main > design goal was to present "a consistent file system image across the > servers in a cluster". Does it work, though? Without Oracle admin tools? > Now, if both goals can be achieved in one go, hell, I'm willing to try > it out myself in an attempt to extract off of it, some performance > indicators that could be compared to other database performance tests > sent to both this and the PERFORM mailing list. Hey, any test you wanna run is fine with us. I'm pretty sure that OCFS belongs to Oracle, though, patent & copyright, so we couldn't actually use it in practice. If your intention in this test is to show the superiority of raw devices, let me give you a reality check: barring some major corporate backing getting involved, we can't possibly implement our own PG-FS for database support. We already have a TODO list which is far too long for our developer pool, and implementing a custom FS either takes a large team (OCFS) or several years of development (Reiser). Now, if you know somebody who might pay for one, then great .... -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-admin-owner@postgresql.org Wed Apr 7 13:37:47 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 03D00D1C951; Wed, 7 Apr 2004 13:37:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85699-09; Wed, 7 Apr 2004 13:37:42 -0300 (ADT) Received: from gp.word-to-the-wise.com (gp.word-to-the-wise.com [64.71.176.18]) by svr1.postgresql.org (Postfix) with ESMTP id 73E78D1BA90; Wed, 7 Apr 2004 13:37:41 -0300 (ADT) Received: by gp.word-to-the-wise.com (Postfix, from userid 500) id 14FC1900060; Wed, 7 Apr 2004 09:29:47 -0700 (PDT) Date: Wed, 7 Apr 2004 09:29:47 -0700 From: Steve Atkins To: pgsql-admin@postgresql.org Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] Raw devices vs. Filesystems Message-ID: <20040407162946.GA7271@gp.word-to-the-wise.com> References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <5719.1081315562@sss.pgh.pa.us> <20040407071858.GA7973@elbereth.noviforum.si> <200404070909.16123.josh@agliodbs.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200404070909.16123.josh@agliodbs.com> User-Agent: Mutt/1.4.1i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/111 X-Sequence-Number: 13084 On Wed, Apr 07, 2004 at 09:09:16AM -0700, Josh Berkus wrote: > If your intention in this test is to show the superiority of raw devices, let > me give you a reality check: barring some major corporate backing getting > involved, we can't possibly implement our own PG-FS for database support. We > already have a TODO list which is far too long for our developer pool, and > implementing a custom FS either takes a large team (OCFS) or several years of > development (Reiser). Is there any documentation as to what guarantees PostgreSQL requires from the filesystem, or what posix semantics can be relaxed? Cheers, Steve From pgsql-performance-owner@postgresql.org Wed Apr 7 14:07:58 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E4146D1C50B for ; Wed, 7 Apr 2004 14:07:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97441-08 for ; Wed, 7 Apr 2004 14:07:53 -0300 (ADT) Received: from bob.samurai.com (bob.samurai.com [205.207.28.75]) by svr1.postgresql.org (Postfix) with ESMTP id 7826CD1BB93 for ; Wed, 7 Apr 2004 14:07:52 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by bob.samurai.com (Postfix) with ESMTP id E982B20EC; Wed, 7 Apr 2004 13:07:50 -0400 (EDT) Received: from bob.samurai.com ([127.0.0.1]) by localhost (bob.samurai.com [127.0.0.1]) (amavisd-new, port 10024) with LMTP id 63453-01-9; Wed, 7 Apr 2004 13:07:46 -0400 (EDT) Received: from [192.168.1.104] (d226-89-59.home.cgocable.net [24.226.89.59]) (using TLSv1 with cipher RC4-MD5 (128/128 bits)) (No client certificate requested) by bob.samurai.com (Postfix) with ESMTP id CA51B1FC3; Wed, 7 Apr 2004 13:07:45 -0400 (EDT) Subject: Re: PostgreSQL and Linux 2.6 kernel. From: Neil Conway To: Josh Berkus Cc: Mike Nolan , Aaron Werman , Gary Doades , pgsql-performance@postgresql.org In-Reply-To: <200404050836.52115.josh@agliodbs.com> References: <200404040323.i343Nw3Z002747@gw.tssi.com> <200404050836.52115.josh@agliodbs.com> Content-Type: text/plain Message-Id: <1081357665.506.44.camel@tokyo> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Wed, 07 Apr 2004 13:07:45 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at mailbox.samurai.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/87 X-Sequence-Number: 6387 On Mon, 2004-04-05 at 11:36, Josh Berkus wrote: > Unfortunately, these days only Tom and Neil seem to be seriously working on > the query planner (beg pardon in advance if I've missed someone) Actually, Tom is the only person actively working on the planner -- while I hope to contribute to it in the future, I haven't done so yet. -Neil From pgsql-performance-owner@postgresql.org Wed Apr 7 16:10:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D268DD1BB16 for ; Wed, 7 Apr 2004 16:10:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 54324-09 for ; Wed, 7 Apr 2004 16:10:02 -0300 (ADT) Received: from sccrmhc11.comcast.net (sccrmhc11.comcast.net [204.127.202.55]) by svr1.postgresql.org (Postfix) with ESMTP id B37C3D1D05F for ; Wed, 7 Apr 2004 16:09:55 -0300 (ADT) Received: from game (c-24-13-191-126.client.comcast.net[24.13.191.126]) by comcast.net (sccrmhc11) with SMTP id <2004040719095701100rrcgue>; Wed, 7 Apr 2004 19:09:57 +0000 Message-ID: <00dd01c41cd3$ed6e1bf0$0201a8c0@game> Reply-To: "Nid" From: "Nid" To: Subject: Postgresql educational sources Date: Wed, 7 Apr 2004 14:10:01 -0500 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_00DA_01C41CAA.045EDB80" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1158 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 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_50, HTML_MESSAGE, RCVD_IN_SORBS X-Spam-Level: X-Archive-Number: 200404/88 X-Sequence-Number: 6388 This is a multi-part message in MIME format. ------=_NextPart_000_00DA_01C41CAA.045EDB80 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Rather than ask some general, unanswerable question on how to tune my datab= ase...I thought I ask where I might find an introduction to...or manual/tut= orial for the depths of managing a postgres db. Books? Websites? Assume = a basic to intermediate knowledge of DBs in general with a desire to learn = about postgres from the ground up. If it makes a difference I'm using a po= stgres db in a Red Hat Linux OS environment. Thanks! nid ------=_NextPart_000_00DA_01C41CAA.045EDB80 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Rather than ask some general, unanswerable= question=20 on how to tune my database...I thought I ask where I might find an introduc= tion=20 to...or manual/tutorial for the depths of managing a postgres db. =20 Books?  Websites?  Assume a basic to intermediate knowledge of DB= s in=20 general with a desire to learn about postgres from the ground up.  If = it=20 makes a difference I'm using a postgres db in a Red Hat Linux OS=20 environment.  Thanks!
 
nid
------=_NextPart_000_00DA_01C41CAA.045EDB80-- From pgsql-performance-owner@postgresql.org Wed Apr 7 16:54:46 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6562DD1BB4C for ; Wed, 7 Apr 2004 16:54:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80705-02 for ; Wed, 7 Apr 2004 16:54:42 -0300 (ADT) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id DEEB6D1BA98 for ; Wed, 7 Apr 2004 16:54:39 -0300 (ADT) 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 i37JqcjX010326; Wed, 7 Apr 2004 13:52:38 -0600 (MDT) Date: Wed, 7 Apr 2004 13:52:35 -0600 (MDT) From: "scott.marlowe" To: Andrew McMillan Cc: huang yaqin , Subject: Re: good pc but bad performance,why? In-Reply-To: <1081337381.522.57.camel@lamb.mcmillan.net.nz> 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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/89 X-Sequence-Number: 6389 On Wed, 7 Apr 2004, Andrew McMillan wrote: > On Wed, 2004-04-07 at 20:56, huang yaqin wrote: > > Hello, Richard Huxton, > > > > You said turning fsync off may cause losing data, that's terrible. > > I use SCSI disk, and file system is ext3. I tune postgresql.conf and can't get any improvement. So what can I do? > > Does SCSI disk and IDE disk have difference? > > Yes, turning off fsync means that the database is not guaranteeing > consistency of writes to disk any longer. On the other hand your IDE > system probably never was, because IDE drives just typically turn on > write caching in hardware without telling anyone. > > SCSI typically doesn't turn on write caching in the physical drive by > default, as Tom Lane pointed out earlier. Good SCSI has a battery > backed up cache, and then it is OK to turn on write caching, because the > controller has enough battery to complete all writes in the event of a > power failure. Actually, almost all SCSI drives turn on write caching by default, they just don't lie about fsync, so you still have a one update per revolution limit, but other things can be happening while that write is being commited due to the multi-threaded nature of both the SCSI interface and the kernel drivers associated with them It would appear the linux kernel hackers are trying to implement the multi-threaded features of the latest ATA spec, so that, at some future date, you could have IDE drives that cache AND tell the truth of their sync AND can do more than one thing at a time. > One thing I recommend is to use ext2 (or almost anything but ext3). > There is no real need (or benefit) from having the database on a > journalled filesystem - the journalling is only trying to give similar > sorts of guarantees to what the fsync in PostgreSQL is doing. Is this true? I was under the impression that without at least meta-data journaling postgresql could still be corrupted by power failure. > The suggestion someone else made regarding use of software raid is > probably also a good one if you are trying to use the on-board RAID at > the moment. Some onboard RAID controllers are fairly good (dell's 2600 series have an adaptec on board that can have battery backed cache that is ok, the lsi megaraid based one is faster under linux though.) But some of them are pretty poor performers. > Finally, I would say that because you are seeing poor performance on one > box and great performance on another, you should look at the hardware, > or at the hardware drivers, for the problem - not so much at PostgreSQL. More than likely, the biggest issue is that the SCSI drives are performing proper fsync, while the IDE drives are lying. Definitely a time to look at a good caching RAID controller. From pgsql-performance-owner@postgresql.org Wed Apr 7 17:28:15 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8CABCD1D062 for ; Wed, 7 Apr 2004 17:28:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 95511-02 for ; Wed, 7 Apr 2004 17:28:10 -0300 (ADT) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 71729D1C9DA for ; Wed, 7 Apr 2004 17:28:07 -0300 (ADT) 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 i37KRWjX014298; Wed, 7 Apr 2004 14:27:32 -0600 (MDT) Date: Wed, 7 Apr 2004 14:27:29 -0600 (MDT) From: "scott.marlowe" To: Nid Cc: Subject: Re: Postgresql educational sources In-Reply-To: <00dd01c41cd3$ed6e1bf0$0201a8c0@game> 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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/90 X-Sequence-Number: 6390 On Wed, 7 Apr 2004, Nid wrote: > Rather than ask some general, unanswerable question on how to tune my > database...I thought I ask where I might find an introduction to...or > manual/tutorial for the depths of managing a postgres db. Books? > Websites? Assume a basic to intermediate knowledge of DBs in general > with a desire to learn about postgres from the ground up. If it makes a > difference I'm using a postgres db in a Red Hat Linux OS environment. > Thanks! The online (adminstration) docs are quite good, and for tuning, look at the excellent tuning document on varlena: http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html From pgsql-performance-owner@postgresql.org Wed Apr 7 18:05:54 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 52AC9D1B893 for ; Wed, 7 Apr 2004 18:05:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 09569-05 for ; Wed, 7 Apr 2004 18:05:17 -0300 (ADT) Received: from hotmail.com (bay8-f42.bay8.hotmail.com [64.4.27.42]) by svr1.postgresql.org (Postfix) with ESMTP id 8C599D1B9EB for ; Wed, 7 Apr 2004 18:05:14 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Wed, 7 Apr 2004 14:05:15 -0700 Received: from 69.65.137.210 by by8fd.bay8.hotmail.msn.com with HTTP; Wed, 07 Apr 2004 21:05:15 GMT X-Originating-IP: [69.65.137.210] X-Originating-Email: [el_vigia_ec@hotmail.com] X-Sender: el_vigia_ec@hotmail.com From: "Jaime Casanova" To: pgsql-performance@postgresql.org Subject: statistics Date: Wed, 07 Apr 2004 21:05:15 +0000 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 07 Apr 2004 21:05:15.0524 (UTC) FILETIME=[06316C40:01C41CE4] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/91 X-Sequence-Number: 6391 What the statistics are? Where can i view it? where can i find info about its field and why are they valuable information to performance? thanx in advance, Jaime Casanova _________________________________________________________________ Protect your PC - get McAfee.com VirusScan Online http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963 From pgsql-performance-owner@postgresql.org Wed Apr 7 19:12:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 19E4BD1B4D3 for ; Wed, 7 Apr 2004 19:12:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37984-01 for ; Wed, 7 Apr 2004 19:12:53 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id BD2EBD1B459 for ; Wed, 7 Apr 2004 19:12:49 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i37MCck07003; Wed, 7 Apr 2004 18:12:38 -0400 (EDT) From: Bruce Momjian Message-Id: <200404072212.i37MCck07003@candle.pha.pa.us> Subject: Re: good pc but bad performance,why? In-Reply-To: To: "scott.marlowe" Date: Wed, 7 Apr 2004 18:12:38 -0400 (EDT) Cc: Andrew McMillan , huang yaqin , 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: 200404/92 X-Sequence-Number: 6392 scott.marlowe wrote: > > One thing I recommend is to use ext2 (or almost anything but ext3). > > There is no real need (or benefit) from having the database on a > > journalled filesystem - the journalling is only trying to give similar > > sorts of guarantees to what the fsync in PostgreSQL is doing. > > Is this true? I was under the impression that without at least meta-data > journaling postgresql could still be corrupted by power failure. It is false. ext2 isn't crash-safe, and PostgreSQL needs an intact file system for WAL recovery. -- 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 Apr 7 20:48:41 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C1FCAD1D090 for ; Wed, 7 Apr 2004 20:48:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70073-01 for ; Wed, 7 Apr 2004 20:48:40 -0300 (ADT) 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 AC568D1D088 for ; Wed, 7 Apr 2004 20:48:36 -0300 (ADT) Received: from tmsl-adsl.demon.co.uk ([80.177.114.181] helo=bacon.tmsl.demon.co.uk) by anchor-post-31.mail.demon.net with esmtp (Exim 3.35 #1) id 1BBMmd-000FuO-0V for pgsql-performance@postgresql.org; Thu, 08 Apr 2004 00:48:39 +0100 Date: Thu, 8 Apr 2004 00:48:35 +0100 From: Paul Thomas To: "pgsql-performance @ postgresql . org" Subject: Re: statistics Message-ID: <20040408004835.A8476@bacon> References: Mime-Version: 1.0 Content-Type: text/plain; format=flowed; charset=ISO-8859-1 In-Reply-To: ; from el_vigia_ec@hotmail.com on Wed, Apr 07, 2004 at 22:05:15 +0100 X-Mailer: Balsa 1.2.3 Lines: 60 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/93 X-Sequence-Number: 6393 On 07/04/2004 22:05 Jaime Casanova wrote: > What the statistics are? Where can i view it? where can i find info > about its field and why are they valuable information to performance? > > thanx in advance, > > Jaime Casanova OK. An idiot's guide to statistics by a full-time idiot... Let's start with a simple premise. I'm a RDBMS (forget that I'm actually an idiot for a moment...) and I've been asked for select * from foo where bar = 7; How do I go about fulfilling the reequest in the most efficient manner? (i.e., ASAP!) One way might be to read through the whole table and return only those rows which match the where criteron - a sequential scan on the table. But wait a minute, there is an index on column bar. Could I use this instead? Well, of course, I could use it but I have to keep sight of the goal of returning the data ASAP and I know that the act of reading index/reading table/... will have a performance penalty due to a lot more head movement on the disk. So how do I make chose between a sequential scan and an index scan? Let's lokk at a couple of extreme scenarios: 1) let's look at the condition where all or virtually all of the bar columns are populated wityh the value 7. In this case it would be more efficient to read sequentially through the table. 2) the opposite of (1) - very few of the bar columns have the value 7. In this case using the index could be a winner. So generalising, I need to be able to estimate whether doing a sequential scan is more efficient that an index scan and this comes down to 2 factors: a) the cost of moving the disk heads all over the place (random page cost) b) the spread of values in the selecting column(s) (a) is specfified in postgresql.conf (see archives for much discusion about what the value should be..) (b) is determined by the dastardly trick of actually sampling the data in the table!!! That's what analyze does. It samples your table(s) and uses the result to feeede into it's descision about when to flip between sequential and index scans. Hope this makes some kind of sense... -- Paul Thomas +------------------------------+---------------------------------------------+ | Thomas Micro Systems Limited | Software Solutions for Business | | Computer Consultants | http://www.thomas-micro-systems-ltd.co.uk | +------------------------------+---------------------------------------------+ From pgsql-performance-owner@postgresql.org Wed Apr 7 22:31:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A98F7D1BB7B for ; Wed, 7 Apr 2004 22:31:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97913-07 for ; Wed, 7 Apr 2004 22:31:40 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 5690CD1D06C for ; Wed, 7 Apr 2004 22:31:36 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i381VImA006789; Wed, 7 Apr 2004 21:31:18 -0400 (EDT) To: Bruce Momjian Cc: "scott.marlowe" , Andrew McMillan , huang yaqin , pgsql-performance@postgresql.org Subject: Re: good pc but bad performance,why? In-reply-to: <200404072212.i37MCck07003@candle.pha.pa.us> References: <200404072212.i37MCck07003@candle.pha.pa.us> Comments: In-reply-to Bruce Momjian message dated "Wed, 07 Apr 2004 18:12:38 -0400" Date: Wed, 07 Apr 2004 21:31:18 -0400 Message-ID: <6788.1081387878@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/94 X-Sequence-Number: 6394 Bruce Momjian writes: > scott.marlowe wrote: >>> There is no real need (or benefit) from having the database on a >>> journalled filesystem - the journalling is only trying to give similar >>> sorts of guarantees to what the fsync in PostgreSQL is doing. >> >> Is this true? I was under the impression that without at least meta-data >> journaling postgresql could still be corrupted by power failure. > It is false. ext2 isn't crash-safe, and PostgreSQL needs an intact file > system for WAL recovery. But it should be okay to set the filesystem to journal only its own metadata. There's no need for it to journal file contents. regards, tom lane From pgsql-performance-owner@postgresql.org Wed Apr 7 22:33:45 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9617BD1B4D3 for ; Wed, 7 Apr 2004 22:33:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96462-08 for ; Wed, 7 Apr 2004 22:33:45 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id E7896D1B4BF for ; Wed, 7 Apr 2004 22:33:40 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i381XYu13325; Wed, 7 Apr 2004 21:33:34 -0400 (EDT) From: Bruce Momjian Message-Id: <200404080133.i381XYu13325@candle.pha.pa.us> Subject: Re: good pc but bad performance,why? In-Reply-To: <6788.1081387878@sss.pgh.pa.us> To: Tom Lane Date: Wed, 7 Apr 2004 21:33:34 -0400 (EDT) Cc: "scott.marlowe" , Andrew McMillan , huang yaqin , 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: 200404/95 X-Sequence-Number: 6395 Tom Lane wrote: > Bruce Momjian writes: > > scott.marlowe wrote: > >>> There is no real need (or benefit) from having the database on a > >>> journalled filesystem - the journalling is only trying to give similar > >>> sorts of guarantees to what the fsync in PostgreSQL is doing. > >> > >> Is this true? I was under the impression that without at least meta-data > >> journaling postgresql could still be corrupted by power failure. > > > It is false. ext2 isn't crash-safe, and PostgreSQL needs an intact file > > system for WAL recovery. > > But it should be okay to set the filesystem to journal only its own > metadata. There's no need for it to journal file contents. Can you set ext2 to journal metadata? I didn't know it could do that. -- 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 Apr 7 23:13:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B080CD1D06A for ; Wed, 7 Apr 2004 23:13:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11913-05 for ; Wed, 7 Apr 2004 23:13:41 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 4D3C3D1C9C8 for ; Wed, 7 Apr 2004 23:13:37 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i382DYcb007295; Wed, 7 Apr 2004 22:13:34 -0400 (EDT) To: Bruce Momjian Cc: "scott.marlowe" , Andrew McMillan , huang yaqin , pgsql-performance@postgresql.org Subject: Re: good pc but bad performance,why? In-reply-to: <200404080133.i381XYu13325@candle.pha.pa.us> References: <200404080133.i381XYu13325@candle.pha.pa.us> Comments: In-reply-to Bruce Momjian message dated "Wed, 07 Apr 2004 21:33:34 -0400" Date: Wed, 07 Apr 2004 22:13:34 -0400 Message-ID: <7294.1081390414@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/96 X-Sequence-Number: 6396 Bruce Momjian writes: > Tom Lane wrote: >> But it should be okay to set the filesystem to journal only its own >> metadata. There's no need for it to journal file contents. > Can you set ext2 to journal metadata? I didn't know it could do that. No, ext2 has no journal at all AFAIK. But I believe ext3 has an option to journal or not journal file contents, and at least on a Postgres-only volume you'd want to turn that off. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Apr 8 00:13:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7989ED1D06E for ; Thu, 8 Apr 2004 00:13:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28571-07 for ; Thu, 8 Apr 2004 00:13:22 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id D8668D1D070 for ; Thu, 8 Apr 2004 00:13:22 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i383C7N28711; Wed, 7 Apr 2004 23:12:07 -0400 (EDT) From: Bruce Momjian Message-Id: <200404080312.i383C7N28711@candle.pha.pa.us> Subject: Re: good pc but bad performance,why? In-Reply-To: <7294.1081390414@sss.pgh.pa.us> To: Tom Lane Date: Wed, 7 Apr 2004 23:12:07 -0400 (EDT) Cc: "scott.marlowe" , Andrew McMillan , huang yaqin , 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: 200404/97 X-Sequence-Number: 6397 Tom Lane wrote: > Bruce Momjian writes: > > Tom Lane wrote: > >> But it should be okay to set the filesystem to journal only its own > >> metadata. There's no need for it to journal file contents. > > > Can you set ext2 to journal metadata? I didn't know it could do that. > > No, ext2 has no journal at all AFAIK. But I believe ext3 has an option > to journal or not journal file contents, and at least on a Postgres-only > volume you'd want to turn that off. Right, ext3 has that option. I don't think XFS needs it (it does meta-data only by default). -- 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-admin-owner@postgresql.org Thu Apr 8 01:33:15 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9B942D1D086; Thu, 8 Apr 2004 01:33:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 57703-02; Thu, 8 Apr 2004 01:33:09 -0300 (ADT) Received: from elbereth.noviforum.si (unknown [193.189.169.66]) by svr1.postgresql.org (Postfix) with ESMTP id A0AA1D1D09B; Thu, 8 Apr 2004 01:33:09 -0300 (ADT) Received: from elbereth.noviforum.si (localhost [127.0.0.1]) by elbereth.noviforum.si (8.12.10/8.12.10) with ESMTP id i384X5Hh028646; Thu, 8 Apr 2004 06:33:05 +0200 Received: (from gregab@localhost) by elbereth.noviforum.si (8.12.10/8.12.10/Submit) id i384X4xU028645; Thu, 8 Apr 2004 06:33:04 +0200 Date: Thu, 8 Apr 2004 06:33:04 +0200 From: Grega Bremec To: Josh Berkus Cc: pgsql-admin@postgresql.org, pgsql-performance@postgresql.org Subject: Re: [PERFORM] Raw devices vs. Filesystems Message-ID: <20040408043304.GA28539@elbereth.noviforum.si> References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <5719.1081315562@sss.pgh.pa.us> <20040407071858.GA7973@elbereth.noviforum.si> <200404070909.16123.josh@agliodbs.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="k1lZvvs/B4yU6o8G" Content-Disposition: inline In-Reply-To: <200404070909.16123.josh@agliodbs.com> User-Agent: Mutt/1.4.2i Organization: Noviforum, Ltd., Software & Media X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/135 X-Sequence-Number: 13108 --k1lZvvs/B4yU6o8G Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable ...and on Wed, Apr 07, 2004 at 09:09:16AM -0700, Josh Berkus used the keybo= ard: >=20 > Does it work, though? Without Oracle admin tools? Hello, Josh. :) Well, as I said, that's why I was asking - I'm willing to give it a go if nobody can prove me wrong. :) > > Now, if both goals can be achieved in one go, hell, I'm willing to try > > it out myself in an attempt to extract off of it, some performance > > indicators that could be compared to other database performance tests > > sent to both this and the PERFORM mailing list. >=20 > Hey, any test you wanna run is fine with us. I'm pretty sure that OCFS= =20 > belongs to Oracle, though, patent & copyright, so we couldn't actually us= e it=20 > in practice. I thought you knew - OCFS, OCFS-Tools and OCFSv2 have not only been open- source for quite a while now - they're released under the GPL. http://oss.oracle.com/projects/ocfs/ http://oss.oracle.com/projects/ocfs-tools/ http://oss.oracle.com/projects/ocfs2/ I don't know what that means to you (probably nothing good, as PostgreSQL is released under the BSD license), but it most definitely can be considered a good thing for the end user, as she can download it, compile, and set it up on her disks, without the need to pay Oracle royalties. :) > If your intention in this test is to show the superiority of raw devices,= let=20 > me give you a reality check: barring some major corporate backing getting= =20 > involved, we can't possibly implement our own PG-FS for database support.= We=20 > already have a TODO list which is far too long for our developer pool, an= d=20 > implementing a custom FS either takes a large team (OCFS) or several year= s of=20 > development (Reiser).=20=20=20 Not really - I was just thinking about something not-entirely-a-filesystem and POK!, OCFS sprang to mind. It omits many POSIX features that slow down a traditional filesystem, yet it does know the concept of inodes and most of all, it's _really_ heavy on caching. As such, it sounded quite promising to me, but trial, I think, is the best test. The question does spring up though, that Steve raised in another post - just for the record, what POSIX semantics can a postmaster live without in a filesystem? Cheers, --=20 Grega Bremec Senior Administrator Noviforum Ltd., Software & Media http://www.noviforum.si/ --k1lZvvs/B4yU6o8G Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (GNU/Linux) iD8DBQFAdNYADo/EMYD4+osRAjavAKCNkuX+bSoTf909fhLr0kt3WkcL0gCgkLoK 176SeZStwQ0aPWPRmK5g0Xs= =kDde -----END PGP SIGNATURE----- --k1lZvvs/B4yU6o8G-- From pgsql-performance-owner@postgresql.org Thu Apr 8 04:58:48 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2EED4D1CB3F for ; Thu, 8 Apr 2004 04:58:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 17860-02 for ; Thu, 8 Apr 2004 04:58:45 -0300 (ADT) Received: from ahmose.CICT_AD.NL (vnd-7521.mxs.adsl.euronet.nl [62.234.149.33]) by svr1.postgresql.org (Postfix) with ESMTP id 088CCD1C9FB for ; Thu, 8 Apr 2004 04:58:43 -0300 (ADT) Received: by ahmose.cict_ad.nl with Internet Mail Service (5.5.2657.72) id <2PZ2B05R>; Thu, 8 Apr 2004 09:59:19 +0200 Message-ID: <2A07EC2D0BC2774AAD6F74769F60D52A0832C6@ahmose.cict_ad.nl> From: "Priem, Alexander" To: pgsql-performance@postgresql.org Subject: data=writeback Date: Thu, 8 Apr 2004 09:59:18 +0200 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2657.72) Content-Type: text/plain X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/99 X-Sequence-Number: 6399 Hi everyone, I have done some reading on filesystems and I thought to optimize the settings for my PostgreSQL system. I use the ext3 filesystem and have the PostgreSQL data and WAL on different physical drives. I made some adjustments to my /etc/fstabd file, so it looks like this : LABEL=/ / ext3 noatime,data=ordered 1 1 LABEL=/boot /boot ext3 noatime,data=ordered 1 2 none /dev/pts devpts gid=5,mode=620 0 0 none /proc proc defaults 0 0 none /dev/shm tmpfs defaults 0 0 LABEL=/usr/local/pgsql /usr/local/pgsql ext3 noatime,data=writeback 1 2 LABEL=/usr/local/pgsql /usr/local/pgsql/wal ext3 noatime,data=ordered 1 2 /dev/sda5 swap swap defaults 0 0 /dev/cdrom /mnt/cdrom udf,iso9660 noauto,owner,kudzu,ro 0 0 /dev/fd0 /mnt/floppy auto noauto,owner,kudzu 0 0 Does this look OK? My knowledge of filesystems and their (journalling) options is not very broad... Thanks in advance, Alexander Priem. From pgsql-performance-owner@postgresql.org Thu Apr 8 05:54:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5A0BBD1BAF7 for ; Thu, 8 Apr 2004 05:54:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33130-06 for ; Thu, 8 Apr 2004 05:54:47 -0300 (ADT) Received: from lamb.mcmillan.net.nz (218-101-13-8.paradise.net.nz [218.101.13.8]) by svr1.postgresql.org (Postfix) with ESMTP id B8020D1BAAD for ; Thu, 8 Apr 2004 05:54:45 -0300 (ADT) Received: from lamb.mcmillan.net.nz (lamb.mcmillan.net.nz [127.0.0.1]) by lamb.mcmillan.net.nz (Postfix) with ESMTP id D2DCDAD98581; Thu, 8 Apr 2004 20:54:39 +1200 (NZST) Subject: Re: good pc but bad performance,why? From: Andrew McMillan To: Tom Lane Cc: pgsql-performance@postgresql.org In-Reply-To: <7294.1081390414@sss.pgh.pa.us> References: <200404080133.i381XYu13325@candle.pha.pa.us> <7294.1081390414@sss.pgh.pa.us> Content-Type: text/plain Message-Id: <1081414478.2428.13.camel@lamb.mcmillan.net.nz> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Thu, 08 Apr 2004 20:54:39 +1200 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/100 X-Sequence-Number: 6400 On Thu, 2004-04-08 at 14:13, Tom Lane wrote: > Bruce Momjian writes: > > Tom Lane wrote: > >> But it should be okay to set the filesystem to journal only its own > >> metadata. There's no need for it to journal file contents. > > > Can you set ext2 to journal metadata? I didn't know it could do that. > > No, ext2 has no journal at all AFAIK. But I believe ext3 has an option > to journal or not journal file contents, and at least on a Postgres-only > volume you'd want to turn that off. No, it certainly doesn't. To be honest I was not aware that PostgreSQL was susceptible to failure on non[metadata] journalled filesystems - I was [somewhat vaguely] of the understanding that it would work fine on any filesystem. And obviously, from my original post, we can see that I believed metadata journalling was wasted on it. Is the 'noatime' option worthwhile? Are you saying that PostgreSQL should always be run on a metadata journalled filesystem then, and that VFAT, ext2, etc are ++ungood? Thanks, Andrew McMillan. ------------------------------------------------------------------------- Andrew @ Catalyst .Net .NZ Ltd, PO Box 11-053, Manners St, Wellington WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St DDI: +64(4)916-7201 MOB: +64(21)635-694 OFFICE: +64(4)499-2267 A foolish consistency is the hobgoblin of little minds - Shaw ------------------------------------------------------------------------- From pgsql-performance-owner@postgresql.org Thu Apr 8 06:01:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 251C2D1D091 for ; Thu, 8 Apr 2004 06:01:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 39976-03 for ; Thu, 8 Apr 2004 06:01:26 -0300 (ADT) Received: from my.endian.it (unknown [62.146.87.34]) by svr1.postgresql.org (Postfix) with ESMTP id 350E1D1D08E for ; Thu, 8 Apr 2004 06:01:24 -0300 (ADT) Received: from [192.168.1.6] (host36-118.pool80180.interbusiness.it [80.180.118.36]) (authenticated (0 bits)) by my.endian.it (8.11.6/8.11.6) with ESMTP id i389XRl17246 for ; Thu, 8 Apr 2004 11:33:27 +0200 Subject: Re: data=writeback From: Chris To: pgsql-performance@postgresql.org In-Reply-To: <2A07EC2D0BC2774AAD6F74769F60D52A0832C6@ahmose.cict_ad.nl> References: <2A07EC2D0BC2774AAD6F74769F60D52A0832C6@ahmose.cict_ad.nl> Content-Type: text/plain Message-Id: <1081414889.5037.29.camel@dell> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 (1.4.5-7) Date: Thu, 08 Apr 2004 11:01:29 +0200 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/101 X-Sequence-Number: 6401 > LABEL=/usr/local/pgsql /usr/local/pgsql ext3 > noatime,data=writeback 1 2 > LABEL=/usr/local/pgsql /usr/local/pgsql/wal ext3 > noatime,data=ordered 1 2 The same label mounted on two different mount points is probably I typo? I'm not sure if data=writeback is ok. I was wondering about the same thing after reading the "good pc but bad performance,why?" thread. This is from man mount: writeback Data ordering is not preserved - data may be written into the main file system after its metadata has been commit- ted to the journal. This is rumoured to be the highest- throughput option. It guarantees internal file system integrity, however it can allow old data to appear in files after a crash and journal recovery. How does this relate to fflush()? Does fflush still garantee all data has ben written? Bye, Chris. From pgsql-performance-owner@postgresql.org Thu Apr 8 06:25:44 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1456BD1D09E for ; Thu, 8 Apr 2004 06:25:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 47768-03 for ; Thu, 8 Apr 2004 06:25:41 -0300 (ADT) Received: from ahmose.CICT_AD.NL (vnd-7521.mxs.adsl.euronet.nl [62.234.149.33]) by svr1.postgresql.org (Postfix) with ESMTP id 1C4AFD1BAEA for ; Thu, 8 Apr 2004 06:25:40 -0300 (ADT) Received: by ahmose.cict_ad.nl with Internet Mail Service (5.5.2657.72) id <2PZ2B06F>; Thu, 8 Apr 2004 11:26:17 +0200 Message-ID: <2A07EC2D0BC2774AAD6F74769F60D52A0832C8@ahmose.cict_ad.nl> From: "Priem, Alexander" To: 'Chris' , pgsql-performance@postgresql.org Subject: Re: data=writeback Date: Thu, 8 Apr 2004 11:26:17 +0200 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2657.72) Content-Type: text/plain X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/102 X-Sequence-Number: 6402 > > LABEL=/usr/local/pgsql /usr/local/pgsql ext3 > > noatime,data=writeback 1 2 > > LABEL=/usr/local/pgsql /usr/local/pgsql/wal ext3 > > noatime,data=ordered 1 2 > > The same label mounted on two different mount points is probably I typo? No, the same label mounted on two different mount points is not a typo. This is the way it is in my /etc/fstab. Note that I did not create this file myself, it was created by the RedHat Enterprise Linux 3 ES installer. I created different partitions for the data directory (/usr/local/pgsql) and the wal directory (/usr/local/pgsql/wal) using the installer and this is how the /etc/fstab file ended up. Why, is this bad? They use the same label, but use different mount points? Can this cause problems? From pgsql-performance-owner@postgresql.org Thu Apr 8 06:46:07 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9E4DFD1CCC5 for ; Thu, 8 Apr 2004 06:46:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 54959-03 for ; Thu, 8 Apr 2004 06:46:04 -0300 (ADT) Received: from ebrsrv01.ebravo.it (host166-198.pool21757.interbusiness.it [217.57.198.166]) by svr1.postgresql.org (Postfix) with ESMTP id 5C03BD1BAF7 for ; Thu, 8 Apr 2004 06:46:02 -0300 (ADT) Received: from streppone.it ([192.168.11.103]) (authenticated bits=0) by ebrsrv01.ebravo.it (8.12.1/8.12.5) with ESMTP id i389k2j4007706 for ; Thu, 8 Apr 2004 11:46:02 +0200 Message-ID: <40751ED5.1010208@streppone.it> Date: Thu, 08 Apr 2004 11:43:49 +0200 From: Cosimo Streppone User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: select count(*) on large tables 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: 200404/104 X-Sequence-Number: 6404 Hello, I've followed the last discussion about the particular case of "select count(*)"s on large tables being somewhat slow. I've seen also this issue already on the todo list, so I know it is not a simple question. This problem arises for me on very large tables, which I mean starting from 1 million rows and above. The alternative solution I tried, that has an optimal speed up, unfortunately is not a way out, and it is based on "EXPLAIN SELECT count(*)" output parsing, which is obviously *not* reliable. The times always get better doing a vacuum (and eventually reindex) of the table, and they slowly lower again. Is there an estimate time for this issue to be resolved? Can I help in some way (code, test cases, ...)? -- Cosimo From pgsql-performance-owner@postgresql.org Thu Apr 8 06:44:27 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 99E53D1D2EA for ; Thu, 8 Apr 2004 06:44:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52007-06 for ; Thu, 8 Apr 2004 06:44:23 -0300 (ADT) Received: from my.endian.it (unknown [62.146.87.34]) by svr1.postgresql.org (Postfix) with ESMTP id 20221D1D08A for ; Thu, 8 Apr 2004 06:44:22 -0300 (ADT) Received: from [192.168.1.6] (host36-118.pool80180.interbusiness.it [80.180.118.36]) (authenticated (0 bits)) by my.endian.it (8.11.6/8.11.6) with ESMTP id i38AGRl18561 for ; Thu, 8 Apr 2004 12:16:27 +0200 Subject: Re: data=writeback From: Chris To: pgsql-performance@postgresql.org In-Reply-To: <2A07EC2D0BC2774AAD6F74769F60D52A0832C8@ahmose.cict_ad.nl> References: <2A07EC2D0BC2774AAD6F74769F60D52A0832C8@ahmose.cict_ad.nl> Content-Type: text/plain Message-Id: <1081417469.5037.50.camel@dell> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 (1.4.5-7) Date: Thu, 08 Apr 2004 11:44:29 +0200 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/103 X-Sequence-Number: 6403 > > > LABEL=/usr/local/pgsql /usr/local/pgsql ext3 > > > noatime,data=writeback 1 2 > > > LABEL=/usr/local/pgsql /usr/local/pgsql/wal ext3 > > > noatime,data=ordered 1 2 > > > > The same label mounted on two different mount points is probably I typo? > > > No, the same label mounted on two different mount points is not a typo. This > is the way it is in my /etc/fstab. > > Note that I did not create this file myself, it was created by the RedHat > Enterprise Linux 3 ES installer. I created different partitions for the data > directory (/usr/local/pgsql) and the wal directory (/usr/local/pgsql/wal) > using the installer and this is how the /etc/fstab file ended up. > > Why, is this bad? They use the same label, but use different mount points? > Can this cause problems? Mmm... how can the mounter distinguish the two partitions? Maybe I'm missing a concept here, but I thought labels must uniquely identify partitions? Seems suspicious to me... Does it work? When you give just "mount" at the command line what output do you get? Bye, Chris. From pgsql-performance-owner@postgresql.org Thu Apr 8 07:09:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5BFEDD1C9BE for ; Thu, 8 Apr 2004 07:09:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 65534-01 for ; Thu, 8 Apr 2004 07:09:34 -0300 (ADT) Received: from ahmose.CICT_AD.NL (vnd-7521.mxs.adsl.euronet.nl [62.234.149.33]) by svr1.postgresql.org (Postfix) with ESMTP id 2BD14D1C4EB for ; Thu, 8 Apr 2004 07:09:33 -0300 (ADT) Received: by ahmose.cict_ad.nl with Internet Mail Service (5.5.2657.72) id <2PZ2B06Z>; Thu, 8 Apr 2004 12:10:11 +0200 Message-ID: <2A07EC2D0BC2774AAD6F74769F60D52A0832C9@ahmose.cict_ad.nl> From: "Priem, Alexander" To: 'Chris' , pgsql-performance@postgresql.org Subject: Re: data=writeback Date: Thu, 8 Apr 2004 12:10:10 +0200 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2657.72) Content-Type: text/plain X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/105 X-Sequence-Number: 6405 > > > > LABEL=/usr/local/pgsql /usr/local/pgsql ext3 > > > > noatime,data=writeback 1 2 > > > > LABEL=/usr/local/pgsql /usr/local/pgsql/wal ext3 > > > > noatime,data=ordered 1 2 > > > > > > The same label mounted on two different mount points is probably I > > > typo? > > > > > > No, the same label mounted on two different mount points is not a > > typo. This is the way it is in my /etc/fstab. > > > > Note that I did not create this file myself, it was created by the > > RedHat Enterprise Linux 3 ES installer. I created different partitions > > for the data directory (/usr/local/pgsql) and the wal directory > > (/usr/local/pgsql/wal) using the installer and this is how the > > /etc/fstab file ended up. > > > > Why, is this bad? They use the same label, but use different mount > > points? Can this cause problems? > > Mmm... how can the mounter distinguish the two partitions? > > Maybe I'm missing a concept here, but I thought labels must uniquely identify partitions? > > Seems suspicious to me... > > Does it work? When you give just "mount" at the command line what output do you get? > > Bye, Chris. When I give "mount" at the command line, everything looks just fine : /dev/sda2 on / type ext3 (rw,noatime,data=ordered) none on /proc type proc (rw) usbdevfs on /proc/bus/usb type usbdevfs (rw) /dev/sda1 on /boot type ext3 (rw,noatime,data=ordered) none on /dev/pts type devpts (rw,gid=5,mode=620) none on /dev/shm type tmpfs (rw) /dev/sdb1 on /usr/local/pgsql type ext3 (rw,noatime,data=writeback) /dev/sda3 on /usr/local/pgsql/wal type ext3 (rw,noatime,data=ordered) It looks like the labels are not really used, just the mount-points. Or could this cause other problems I am not aware of? Everything seems to be working just fine, for several months now... From pgsql-performance-owner@postgresql.org Thu Apr 8 07:54:31 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 662E4D1D080 for ; Thu, 8 Apr 2004 07:54:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75913-05 for ; Thu, 8 Apr 2004 07:54:30 -0300 (ADT) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 362B9D1D088 for ; Thu, 8 Apr 2004 07:54:28 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id BD53086FA; Thu, 8 Apr 2004 12:54:29 +0200 (CEST) Date: Thu, 8 Apr 2004 12:54:29 +0200 (CEST) From: Dennis Bjorklund To: Cosimo Streppone Cc: pgsql-performance@postgresql.org Subject: Re: select count(*) on large tables In-Reply-To: <40751ED5.1010208@streppone.it> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/106 X-Sequence-Number: 6406 On Thu, 8 Apr 2004, Cosimo Streppone wrote: > The alternative solution I tried, that has an optimal > speed up, unfortunately is not a way out, and it is based > on "EXPLAIN SELECT count(*)" output parsing, which > is obviously *not* reliable. Try this to get the estimate: SELECT relname, reltuples from pg_class order by relname; > The times always get better doing a vacuum (and eventually > reindex) of the table, and they slowly lower again. Yes, the estimate is updated by the analyze. > Is there an estimate time for this issue to be resolved? It's not so easy to "fix". The naive fixes makes other operations slower, most notably makes things less concurrent which is bad since it wont scale as good for many users then. You can always keep the count yourself and have some triggers that update the count on each insert and delete on the table. It will of course make all inserts and deletes slower, but if you count all rows often maybe it's worth it. Most people do not need to count all rows in a table anyway. You usually count all rows such as this and that (some condition). -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Thu Apr 8 08:54:32 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0B0B8D1D098 for ; Thu, 8 Apr 2004 08:54:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 90672-09 for ; Thu, 8 Apr 2004 08:54:29 -0300 (ADT) Received: from 3times25.net (66-23-202-190.clients.speedfactory.net [66.23.202.190]) by svr1.postgresql.org (Postfix) with ESMTP id 0FA5AD1B441 for ; Thu, 8 Apr 2004 08:54:27 -0300 (ADT) Received: from 3times25.net (localhost [127.0.0.1]) by 3times25.net (Postfix) with ESMTP id 7D5E74C167; Thu, 8 Apr 2004 07:56:53 -0400 (EDT) Message-ID: <40753E05.7030300@3times25.net> Date: Thu, 08 Apr 2004 07:56:53 -0400 From: Geoffrey User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Andrew McMillan Cc: Tom Lane , pgsql-performance@postgresql.org Subject: Re: good pc but bad performance,why? References: <200404080133.i381XYu13325@candle.pha.pa.us> <7294.1081390414@sss.pgh.pa.us> <1081414478.2428.13.camel@lamb.mcmillan.net.nz> In-Reply-To: <1081414478.2428.13.camel@lamb.mcmillan.net.nz> 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: 200404/107 X-Sequence-Number: 6407 Andrew McMillan wrote: > On Thu, 2004-04-08 at 14:13, Tom Lane wrote: > >> Bruce Momjian writes: >> >>> Tom Lane wrote: >>> >>>> But it should be okay to set the filesystem to journal only its >>>> own metadata. There's no need for it to journal file contents. >>>> >> >>> Can you set ext2 to journal metadata? I didn't know it could do >>> that. >> >> No, ext2 has no journal at all AFAIK. But I believe ext3 has an >> option to journal or not journal file contents, and at least on a >> Postgres-only volume you'd want to turn that off. > > > No, it certainly doesn't. You can mount ext3 filesystems as ext2 and they will function just as ext2. -- Until later, Geoffrey Registered Linux User #108567 Building secure systems in spite of Microsoft From pgsql-performance-owner@postgresql.org Thu Apr 8 09:29:03 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E8673D1D887 for ; Thu, 8 Apr 2004 09:29:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05175-07 for ; Thu, 8 Apr 2004 09:29:01 -0300 (ADT) Received: from my.endian.it (unknown [62.146.87.34]) by svr1.postgresql.org (Postfix) with ESMTP id 4F222D1D4D5 for ; Thu, 8 Apr 2004 09:28:59 -0300 (ADT) Received: from 1006.org (localhost.localdomain [127.0.0.1]) by my.endian.it (8.11.6/8.11.6) with SMTP id i38D16l21293 for ; Thu, 8 Apr 2004 15:01:06 +0200 Received: from host72-55.pool8248.interbusiness.it ([82.48.55.72]) (SquirrelMail authenticated user list@1006.org) by www.endian.it with HTTP; Thu, 8 Apr 2004 15:01:06 +0200 (CEST) Message-ID: <49386.82.48.55.72.1081429266.squirrel@www.endian.it> Date: Thu, 8 Apr 2004 15:01:06 +0200 (CEST) Subject: Re: data=writeback From: To: In-Reply-To: <2A07EC2D0BC2774AAD6F74769F60D52A0832C9@ahmose.cict_ad.nl> References: <2A07EC2D0BC2774AAD6F74769F60D52A0832C9@ahmose.cict_ad.nl> X-Priority: 3 Importance: Normal X-MSMail-Priority: Normal X-Mailer: SquirrelMail (version 1.2.7) 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: 200404/108 X-Sequence-Number: 6408 > When I give "mount" at the command line, everything looks just fine : > > /dev/sda2 on / type ext3 (rw,noatime,data=ordered) > none on /proc type proc (rw) > usbdevfs on /proc/bus/usb type usbdevfs (rw) > /dev/sda1 on /boot type ext3 (rw,noatime,data=ordered) > none on /dev/pts type devpts (rw,gid=5,mode=620) > none on /dev/shm type tmpfs (rw) > /dev/sdb1 on /usr/local/pgsql type ext3 (rw,noatime,data=writeback) > /dev/sda3 on /usr/local/pgsql/wal type ext3 (rw,noatime,data=ordered) > > It looks like the labels are not really used, just the mount-points. Or > could this cause other problems I am not aware of? Everything seems to > be working just fine, for several months now... Probably /dev/sdb1 and /dev/sda3 have the same labels and mount simply mounts them in a consistent way according to some logic we're not aware of. I'd say: if it works don't touch it ;) What remains unresolved is the question whether data=writeback is ok or not. We'll see if somebody has more information on that one... Bye, Chris. From pgsql-performance-owner@postgresql.org Thu Apr 8 11:09:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F0D04D1D09F for ; Thu, 8 Apr 2004 11:09:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 44223-10 for ; Thu, 8 Apr 2004 11:09:06 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 677F3D1D097 for ; Thu, 8 Apr 2004 11:09:03 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i38E976Y013442; Thu, 8 Apr 2004 10:09:07 -0400 (EDT) To: Cosimo Streppone Cc: pgsql-performance@postgresql.org Subject: Re: select count(*) on large tables In-reply-to: <40751ED5.1010208@streppone.it> References: <40751ED5.1010208@streppone.it> Comments: In-reply-to Cosimo Streppone message dated "Thu, 08 Apr 2004 11:43:49 +0200" Date: Thu, 08 Apr 2004 10:09:07 -0400 Message-ID: <13441.1081433347@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/109 X-Sequence-Number: 6409 Cosimo Streppone writes: > Is there an estimate time for this issue to be resolved? Approximately never. It's a fundamental feature of Postgres' design. As noted by Dennis, you can look at the pg_class statistics if a recent estimate is good enough, or you can build user-level tracking tools if you'd rather have fast count(*) than concurrent update capability. But don't sit around waiting for the developers to "fix this bug", because it isn't a bug and it isn't going to be fixed. regards, tom lane From pgsql-performance-owner@postgresql.org Thu Apr 8 11:52:38 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E80BFD1DB35 for ; Thu, 8 Apr 2004 11:52:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69849-02 for ; Thu, 8 Apr 2004 11:52:34 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 3EF09D1DB59 for ; Thu, 8 Apr 2004 11:52:30 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i38EqXV20923; Thu, 8 Apr 2004 10:52:33 -0400 (EDT) From: Bruce Momjian Message-Id: <200404081452.i38EqXV20923@candle.pha.pa.us> Subject: Re: data=writeback In-Reply-To: <49386.82.48.55.72.1081429266.squirrel@www.endian.it> To: list@1006.org Date: Thu, 8 Apr 2004 10:52:33 -0400 (EDT) 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: 200404/110 X-Sequence-Number: 6410 list@1006.org wrote: > > When I give "mount" at the command line, everything looks just fine : > > > > /dev/sda2 on / type ext3 (rw,noatime,data=ordered) > > none on /proc type proc (rw) > > usbdevfs on /proc/bus/usb type usbdevfs (rw) > > /dev/sda1 on /boot type ext3 (rw,noatime,data=ordered) > > none on /dev/pts type devpts (rw,gid=5,mode=620) > > none on /dev/shm type tmpfs (rw) > > /dev/sdb1 on /usr/local/pgsql type ext3 (rw,noatime,data=writeback) > > /dev/sda3 on /usr/local/pgsql/wal type ext3 (rw,noatime,data=ordered) > > > > It looks like the labels are not really used, just the mount-points. Or > > could this cause other problems I am not aware of? Everything seems to > > be working just fine, for several months now... > > Probably /dev/sdb1 and /dev/sda3 have the same labels and mount > simply mounts them in a consistent way according to some logic > we're not aware of. > > I'd say: if it works don't touch it ;) > > What remains unresolved is the question whether data=writeback is ok > or not. We'll see if somebody has more information on that one... Should be fine. We don't continue until fsync() writes all the data. We don't care what order it is written in, just that is all written before we continue. -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073 From pgsql-performance-owner@postgresql.org Thu Apr 8 11:56:16 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7BDB8D1DB81 for ; Thu, 8 Apr 2004 11:56:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68010-07 for ; Thu, 8 Apr 2004 11:56:11 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 9E9B9D1D8AE for ; Thu, 8 Apr 2004 11:56:07 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i38Eu5S21317; Thu, 8 Apr 2004 10:56:05 -0400 (EDT) From: Bruce Momjian Message-Id: <200404081456.i38Eu5S21317@candle.pha.pa.us> Subject: Re: good pc but bad performance,why? In-Reply-To: <1081414478.2428.13.camel@lamb.mcmillan.net.nz> To: Andrew McMillan Date: Thu, 8 Apr 2004 10:56:05 -0400 (EDT) Cc: Tom Lane , 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: 200404/111 X-Sequence-Number: 6411 Andrew McMillan wrote: > On Thu, 2004-04-08 at 14:13, Tom Lane wrote: > > Bruce Momjian writes: > > > Tom Lane wrote: > > >> But it should be okay to set the filesystem to journal only its own > > >> metadata. There's no need for it to journal file contents. > > > > > Can you set ext2 to journal metadata? I didn't know it could do that. > > > > No, ext2 has no journal at all AFAIK. But I believe ext3 has an option > > to journal or not journal file contents, and at least on a Postgres-only > > volume you'd want to turn that off. > > No, it certainly doesn't. > > To be honest I was not aware that PostgreSQL was susceptible to failure > on non[metadata] journalled filesystems - I was [somewhat vaguely] of > the understanding that it would work fine on any filesystem. We expect the filesystem to come back intact. If it doesn't from an ext2 crash, we can't WAL recover in all cases. > And obviously, from my original post, we can see that I believed > metadata journalling was wasted on it. No. UFS file systems don't do journaling, but do metadata fsync, which is all we need. > Is the 'noatime' option worthwhile? Are you saying that PostgreSQL noatime might help, not sure, but my guess is that most inode fsync's are modifications of mtime, which can't be turned off with amount option. > should always be run on a metadata journalled filesystem then, and that > VFAT, ext2, etc are ++ungood? Yep. Not sure about VFAT but we do need the filesystem to return after a crash, obviously, or we can't even get to the xlog directory or the /data files to do WAL recovery. -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073 From pgsql-performance-owner@postgresql.org Thu Apr 8 12:06:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5AF22D1D07E for ; Thu, 8 Apr 2004 12:06:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 67575-09 for ; Thu, 8 Apr 2004 12:06:30 -0300 (ADT) Received: from bramble.mmrd.com (unknown [65.217.53.66]) by svr1.postgresql.org (Postfix) with ESMTP id 75554D1CCBF for ; Thu, 8 Apr 2004 12:06:29 -0300 (ADT) 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 i38F7acM023307; Thu, 8 Apr 2004 11:07:37 -0400 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 i38F6Cl09664; Thu, 8 Apr 2004 11:06:13 -0400 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 FVMA763Q; Thu, 8 Apr 2004 11:06:10 -0400 Subject: Re: Postgresql educational sources From: Robert Treat To: "scott.marlowe" Cc: Nid , pgsql-performance@postgresql.org In-Reply-To: References: Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 Date: 08 Apr 2004 11:06:11 -0400 Message-Id: <1081436772.2470.55.camel@camel> Mime-Version: 1.0 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/112 X-Sequence-Number: 6412 On Wed, 2004-04-07 at 16:27, scott.marlowe wrote: > On Wed, 7 Apr 2004, Nid wrote: > > > Rather than ask some general, unanswerable question on how to tune my > > database...I thought I ask where I might find an introduction to...or > > manual/tutorial for the depths of managing a postgres db. Books? > > Websites? Assume a basic to intermediate knowledge of DBs in general > > with a desire to learn about postgres from the ground up. If it makes a > > difference I'm using a postgres db in a Red Hat Linux OS environment. > > Thanks! > > The online (adminstration) docs are quite good, and for tuning, look at > the excellent tuning document on varlena: > > http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html > Actually I might rather suggest looking at http://techdocs.postgresql.org/ which has a slew of links/articles/tutorials regarding development and administration of postgresql databases (including a link to Scott's aforementioned doc) Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Thu Apr 8 14:33:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 254DCD1D2A4 for ; Thu, 8 Apr 2004 14:33:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 24329-09 for ; Thu, 8 Apr 2004 14:33:32 -0300 (ADT) Received: from secure.icanx.com (secure.icanx.com [64.246.58.55]) by svr1.postgresql.org (Postfix) with ESMTP id E192CD1D09A for ; Thu, 8 Apr 2004 14:33:30 -0300 (ADT) Received: from 63-229-201-197.mpls.qwest.net ([63.229.201.197] helo=[192.168.2.102]) by secure.icanx.com with esmtp (Exim 4.24) id 1BBdPY-0001VQ-2D for pgsql-performance@postgresql.org; Thu, 08 Apr 2004 13:33:56 -0400 Mime-Version: 1.0 (Apple Message framework v613) Content-Transfer-Encoding: 7bit Message-Id: Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: Mark Lubratt Subject: tsearch query plan Date: Thu, 8 Apr 2004 12:33:28 -0500 X-Mailer: Apple Mail (2.613) X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - secure.icanx.com X-AntiAbuse: Original Domain - postgresql.org X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12] X-AntiAbuse: Sender Address Domain - indeq.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/113 X-Sequence-Number: 6413 Hello, I've just started using the tsearch2 system. I'm VERY impressed by the speed. I've got one question about query planning. Is it understandable to have the query plan estimation be off by a couple of orders of magnitude? Or, is it the fact that the cost estimation is small to begin with that the error between the actual and the estimated is "normal"? Here is my explain analyze run immediately after a vacuum full analyze: kjv=# vacuum full analyze; VACUUM kjv=# explain analyze select * from kjv where idxFTI @@ 'corinth'::tsquery; QUERY PLAN ------------------------------------------------------------------------ --------------------------------------------- Index Scan using idxfti_idx on kjv (cost=0.00..125.44 rows=32 width=193) (actual time=0.796..1.510 rows=6 loops=1) Index Cond: (idxfti @@ '\'corinth\''::tsquery) Filter: (idxfti @@ '\'corinth\''::tsquery) Total runtime: 1.679 ms (4 rows) Thanks! Mark From pgsql-performance-owner@postgresql.org Thu Apr 8 19:42:28 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1A8A5D1BAAD for ; Thu, 8 Apr 2004 19:42:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 46580-07 for ; Thu, 8 Apr 2004 19:42:26 -0300 (ADT) Received: from hotmail.com (bay8-f117.bay8.hotmail.com [64.4.27.117]) by svr1.postgresql.org (Postfix) with ESMTP id 790EAD1B4D3 for ; Thu, 8 Apr 2004 19:42:23 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Thu, 8 Apr 2004 15:42:25 -0700 Received: from 65.248.167.226 by by8fd.bay8.hotmail.msn.com with HTTP; Thu, 08 Apr 2004 22:42:25 GMT X-Originating-IP: [65.248.167.226] X-Originating-Email: [el_vigia_ec@hotmail.com] X-Sender: el_vigia_ec@hotmail.com From: "Jaime Casanova" To: paul@tmsl.demon.co.uk, pgsql-performance@postgresql.org Subject: Re: statistics Date: Thu, 08 Apr 2004 22:42:25 +0000 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 08 Apr 2004 22:42:25.0700 (UTC) FILETIME=[C3A96E40:01C41DBA] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/114 X-Sequence-Number: 6414 Thanks for the answer. I know the question was to primitive (it claims: i have no idea about databases). But i simply didn't find the answer and if a don't ask i won't learn. Someday i will talk with Tom Lane about how to improve the planner but until that day comes i have a lot of technical things to learn. _________________________________________________________________ Add photos to your messages with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/featuredemail From pgsql-performance-owner@postgresql.org Thu Apr 8 20:33:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9DB24D1BAAD for ; Thu, 8 Apr 2004 20:33:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 60163-05 for ; Thu, 8 Apr 2004 20:33:28 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 69855D1B49F for ; Thu, 8 Apr 2004 20:33:25 -0300 (ADT) Received: from [63.195.55.98] (HELO 192.168.1.29) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4801472; Thu, 08 Apr 2004 16:34:48 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Mark Lubratt , pgsql-performance@postgresql.org Subject: Re: tsearch query plan Date: Thu, 8 Apr 2004 16:33:40 -0700 User-Agent: KMail/1.5.4 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: <200404081633.40388.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/115 X-Sequence-Number: 6415 Mark, > I've got one question about query planning. Is it understandable to > have the query plan estimation be off by a couple of orders of > magnitude? Or, is it the fact that the cost estimation is small to > begin with that the error between the actual and the estimated is > "normal"? Well, your example is not "a couple orders of magnitude". 6 vs. 32 is actually pretty good accuracy. Now, 6 vs 192 would potentially be a problem, let alone 32 vs 13,471. -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Apr 9 00:28:44 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BFCE4D1CCC5 for ; Fri, 9 Apr 2004 00:28:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 29961-04 for ; Fri, 9 Apr 2004 00:28:39 -0300 (ADT) Received: from secure.icanx.com (secure.icanx.com [64.246.58.55]) by svr1.postgresql.org (Postfix) with ESMTP id 113E7D1CCC2 for ; Fri, 9 Apr 2004 00:28:40 -0300 (ADT) Received: from 63-229-201-197.mpls.qwest.net ([63.229.201.197] helo=[192.168.2.102]) by secure.icanx.com with esmtp (Exim 4.24) id 1BBmhU-0004nz-FQ; Thu, 08 Apr 2004 23:29:04 -0400 Mime-Version: 1.0 (Apple Message framework v613) In-Reply-To: <200404081633.40388.josh@agliodbs.com> References: <200404081633.40388.josh@agliodbs.com> Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Mark Lubratt Subject: Re: tsearch query plan Date: Thu, 8 Apr 2004 22:28:37 -0500 To: pgsql-performance@postgresql.org, josh@agliodbs.com X-Mailer: Apple Mail (2.613) X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - secure.icanx.com X-AntiAbuse: Original Domain - postgresql.org X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12] X-AntiAbuse: Sender Address Domain - indeq.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/116 X-Sequence-Number: 6416 On Apr 8, 2004, at 6:33 PM, Josh Berkus wrote: > Mark, > >> I've got one question about query planning. Is it understandable to >> have the query plan estimation be off by a couple of orders of >> magnitude? Or, is it the fact that the cost estimation is small to >> begin with that the error between the actual and the estimated is >> "normal"? > > Well, your example is not "a couple orders of magnitude". 6 vs. 32 is > actually pretty good accuracy. > > Now, 6 vs 192 would potentially be a problem, let alone 32 vs 13,471. > I guess I was looking more at the cost estimate and not so much at the rows estimate. I agree that the row estimate wasn't too bad. But the cost estimate seems way out of line. I'm somewhat new to examining explain analyze output and I'm looking at this as more of an education, since the speed is certainly good anyway. I just expected the cost estimate to be more in line especially immediately after an analyze. -Mark From pgsql-performance-owner@postgresql.org Fri Apr 9 02:26:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9FAE5D1D282 for ; Fri, 9 Apr 2004 02:26:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53271-08 for ; Fri, 9 Apr 2004 02:26:54 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 8C01DD1C9DA for ; Fri, 9 Apr 2004 02:26:54 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i395QqlA021350; Fri, 9 Apr 2004 01:26:52 -0400 (EDT) To: Mark Lubratt Cc: pgsql-performance@postgresql.org, josh@agliodbs.com Subject: Re: tsearch query plan In-reply-to: References: <200404081633.40388.josh@agliodbs.com> Comments: In-reply-to Mark Lubratt message dated "Thu, 08 Apr 2004 22:28:37 -0500" Date: Fri, 09 Apr 2004 01:26:51 -0400 Message-ID: <21349.1081488411@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/117 X-Sequence-Number: 6417 Mark Lubratt writes: > I guess I was looking more at the cost estimate and not so much at the > rows estimate. I agree that the row estimate wasn't too bad. But the > cost estimate seems way out of line. The cost estimates are not even in the same units as the actual runtime. Cost is in an arbitrary scale in which 1 unit = 1 sequential disk block fetch. It is unknown what this might equate to on your machine ... but it's quite unlikely that it's 1.0 millisecond. The thing to look at when considering EXPLAIN results is whether the ratios of different cost estimates are proportional to the actual runtimes. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 9 12:02:39 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 251C8D1D06F for ; Fri, 9 Apr 2004 12:02:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 23290-07 for ; Fri, 9 Apr 2004 12:02:14 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 5FFDAD1D078 for ; Fri, 9 Apr 2004 12:02:14 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i39F2CM7025870; Fri, 9 Apr 2004 11:02:12 -0400 (EDT) To: "Andrew Matthews" Cc: pgsql-performance@postgresql.org Subject: Re: Wierd issues In-reply-to: <79AEF92046759442A4AA04C2AB08C0B80834E2@exchange.corp.dslextreme.com> References: <79AEF92046759442A4AA04C2AB08C0B80834E2@exchange.corp.dslextreme.com> Comments: In-reply-to "Andrew Matthews" message dated "Mon, 05 Apr 2004 18:41:08 -0700" Date: Fri, 09 Apr 2004 11:02:11 -0400 Message-ID: <25869.1081522931@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/118 X-Sequence-Number: 6418 "Andrew Matthews" writes: > [ PG 7.3.4 much slower than 7.2.1 ] > > Both have same databases, Both have had vacume full ran on them. You did ANALYZE too, right? The bulk of the time is evidently going into the seqscan on users in each case: > -> Seq Scan on users u (cost=0.00..1938.51 rows=71283 width=4) (actual time=0.81..30119.58 rows=70809 loops=1) > -> Seq Scan on users u (cost=0.00..1888.85 rows=71548 width=4) (actual time=18.38..2277152.51 rows=71028 loops=1) > Filter: (get_pwd(username, '127.0.0.1'::character varying, '101'::character varying, 'MD5'::character varying) IS NOT NULL) I have to suspect that the inefficiency is inside this get_pwd() function, but you didn't tell us anything about that... regards, tom lane From pgsql-admin-owner@postgresql.org Fri Apr 9 13:02:35 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2D087D1BBA1; Fri, 9 Apr 2004 13:02:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 41593-09; Fri, 9 Apr 2004 13:02:29 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 88701D1BA90; Fri, 9 Apr 2004 13:02:28 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4806184; Fri, 09 Apr 2004 09:03:48 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Grega Bremec Subject: Re: [PERFORM] Raw devices vs. Filesystems Date: Fri, 9 Apr 2004 09:02:00 -0700 User-Agent: KMail/1.4.3 Cc: pgsql-admin@postgresql.org, pgsql-performance@postgresql.org References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <200404070909.16123.josh@agliodbs.com> <20040408043304.GA28539@elbereth.noviforum.si> In-Reply-To: <20040408043304.GA28539@elbereth.noviforum.si> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404090902.00934.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/182 X-Sequence-Number: 13155 Grega, > Well, as I said, that's why I was asking - I'm willing to give it a go > if nobody can prove me wrong. :) Why not? If you have time? > I thought you knew - OCFS, OCFS-Tools and OCFSv2 have not only been open- > source for quite a while now - they're released under the GPL. Keen! Wonder if we can make them regret it. Seriously, if Oracle opened this stuff, it's probably becuase they used some GPL components in it. It also probably means that it won't work for anything but Oracle ... > I don't know what that means to you (probably nothing good, as PostgreSQL > is released under the BSD license), Well, it just means that we can't ship OCFS with PostgreSQL. > The question does spring up though, that Steve raised in another post - > just for the record, what POSIX semantics can a postmaster live without in > a filesystem? You might want to ask that question again on Hackers. I don't know the answer, myself. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Apr 9 21:27:53 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 240EAD1D105 for ; Fri, 9 Apr 2004 13:12:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48057-08 for ; Fri, 9 Apr 2004 13:12:44 -0300 (ADT) Received: from exchange.rampageusa.net (exchange.dslextreme.com [66.51.198.16]) by svr1.postgresql.org (Postfix) with ESMTP id 6AC87D1D299 for ; Fri, 9 Apr 2004 13:12:43 -0300 (ADT) Received: from exstatica ([66.51.209.71]) by exchange.rampageusa.net with Microsoft SMTPSVC(6.0.3790.0); Fri, 9 Apr 2004 09:11:50 -0700 From: "Andrew Matthews" To: "'Tom Lane'" Cc: Subject: Re: Wierd issues Date: Fri, 9 Apr 2004 09:12:38 -0700 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.5510 Thread-Index: AcQeQ4U5gf8WxlkmSSeVfxxW2VCIBgACdxnQ X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 In-reply-to: <25869.1081522931@sss.pgh.pa.us> Message-ID: X-OriginalArrivalTime: 09 Apr 2004 16:11:51.0025 (UTC) FILETIME=[5DEB8E10:01C41E4D] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/125 X-Sequence-Number: 6425 Yes I did do analyze.... the here is the get_pwd function -- Function: public.get_pwd(varchar, varchar, varchar, varchar) -- DROP FUNCTION public.get_pwd(varchar, varchar, varchar, varchar); CREATE OR REPLACE FUNCTION public.get_pwd(varchar, varchar, varchar, varchar) RETURNS varchar AS ' DECLARE p_username ALIAS for $1; p_server ALIAS for $2; p_service ALIAS for $3; p_pwd_type ALIAS for $4; l_resource_id integer; l_server_id integer; l_service_id integer; l_allow_deny char(1); l_user_id integer; l_account_id integer; l_passwd varchar(40); begin -- get server identifier select id into l_server_id from servers s where address = p_server; if NOT FOUND then -- try to get default server select id into l_server_id from servers s where address = \'default\'; end if; if l_server_id isnull then return NULL; end if; -- get service identifier select id into l_service_id from services s where radius_service = p_service; if l_service_id isnull then return NULL; end if; -- get resource identifier (server/service combination) select id into l_resource_id from resources r where service_id = l_service_id and server_id = l_server_id; -- could not find resource via server_id, now look via server\'s group if any if l_resource_id isnull then select id into l_resource_id from resources r where service_id = l_service_id and server_group_id = (select server_group_id from servers where id = l_server_id); end if; -- could not determine resource user wants to access, so deny by returning NULL passwd if l_resource_id isnull then return NULL; end if; -- at this point we have a valid resource_id -- determine if valid username select u.id, u.account_id into l_user_id, l_account_id from users u, accounts a where u.username = upper(p_username) -- always uppercase in DB and u.del_id = 0 and u.status = \'A\' and a.status = \'A\' and u.account_id = a.id; -- if active user not found then return NULL for passwd if l_user_id isnull then return null; end if; -- user specific control select allow_deny into l_allow_deny from users_acl where resource_id = l_resource_id and user_id = l_user_id; if l_allow_deny = \'D\' then return NULL; elsif l_allow_deny isnull then -- no user-specific control select max(allow_deny) -- \'D\' is > \'A\' hence deny takes precedence if conflict across groups into l_allow_deny from users_acl where resource_id = l_resource_id and user_group_id in (select user_group_id from user_group_assignments where user_id = l_user_id); elsif l_allow_deny = \'A\' then -- do nothing; -- get and return passwd below end if; if l_allow_deny isnull or l_allow_deny = \'D\' then return NULL; elsif l_allow_deny = \'A\' then select password into l_passwd from user_pwds where password_type = upper(p_pwd_type) and user_id = l_user_id; return l_passwd; else return null; end if; end; ' LANGUAGE 'plpgsql' VOLATILE; -----Original Message----- From: Tom Lane [mailto:tgl@sss.pgh.pa.us] Sent: Friday, April 09, 2004 8:02 AM To: Andrew Matthews Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] Wierd issues "Andrew Matthews" writes: > [ PG 7.3.4 much slower than 7.2.1 ] > > Both have same databases, Both have had vacume full ran on them. You did ANALYZE too, right? The bulk of the time is evidently going into the seqscan on users in each case: > -> Seq Scan on users u (cost=0.00..1938.51 rows=71283 width=4) (actual time=0.81..30119.58 rows=70809 loops=1) > -> Seq Scan on users u (cost=0.00..1888.85 rows=71548 width=4) (actual time=18.38..2277152.51 rows=71028 loops=1) > Filter: (get_pwd(username, '127.0.0.1'::character varying, '101'::character varying, 'MD5'::character varying) IS NOT NULL) I have to suspect that the inefficiency is inside this get_pwd() function, but you didn't tell us anything about that... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 9 13:50:43 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 30712D1CCC2 for ; Fri, 9 Apr 2004 13:50:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 62545-05 for ; Fri, 9 Apr 2004 13:50:40 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 538FBD1CA77 for ; Fri, 9 Apr 2004 13:50:39 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i39GoZtX026676; Fri, 9 Apr 2004 12:50:35 -0400 (EDT) To: Ken Geis Cc: Richard Huxton , pgsql-performance@postgresql.org Subject: Re: plan problem In-reply-to: <4073C3DF.2070307@speakeasy.org> References: <40731252.2030002@speakeasy.org> <200404070938.11482.dev@archonet.com> <4073C3DF.2070307@speakeasy.org> Comments: In-reply-to Ken Geis message dated "Wed, 07 Apr 2004 02:03:27 -0700" Date: Fri, 09 Apr 2004 12:50:35 -0400 Message-ID: <26675.1081529435@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/120 X-Sequence-Number: 6420 Ken Geis writes: > Does anyone think that the planner issue has merit to address? Can > someone help me figure out what code I would look at? The planner doesn't currently attempt to "drill down" into a sub-select- in-FROM to find statistics about the variables emitted by the sub-select. So it's just falling back to a default estimate of the number of distinct values coming out of the sub-select. The "drilling down" part is not hard; the difficulty comes from trying to figure out whether and how the stats from the underlying column would need to be adjusted for the behavior of the sub-select itself. As an example, the result of (SELECT DISTINCT foo FROM bar) would usually have much different stats from the raw bar.foo column. In your example, the LIMIT clause potentially affects the stats by reducing the number of distinct values. Now in most situations where the sub-select wouldn't change the stats, there's no issue anyway because the planner will flatten the sub-select into the main query. So we really have to figure out the adjustment part before we can think about doing much here. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 9 14:28:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0C53FD1D1B8 for ; Fri, 9 Apr 2004 14:28:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 78801-01 for ; Fri, 9 Apr 2004 14:28:31 -0300 (ADT) Received: from fddlnint05.fds.com (fddlnint05.fds.com [208.15.91.52]) by svr1.postgresql.org (Postfix) with ESMTP id 632B5D1D070 for ; Fri, 9 Apr 2004 14:28:30 -0300 (ADT) Subject: Upgrading question (recycled transaction log) To: pgsql-performance@postgresql.org X-Mailer: Lotus Notes Release 6.0 September 26, 2002 Message-ID: From: "Patrick Hatcher" Date: Fri, 9 Apr 2004 10:18:12 -0700 X-MIMETrack: Serialize by Router on FDDLNINT05/FSG/SVR/FDD(Release 5.0.4 |June 8, 2000) at 04/09/2004 01:21:51 PM MIME-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/121 X-Sequence-Number: 6421 Doing an upgrade from 7.3.6 to 7.4.2 and I keep seeing the recycled transaction log about every 2 mins. For future upgrades, is there something that can be set so that I don't have as many recycles? It seems to slow down the importing of data. Here's my current settings: #--------------------------------------------------------------------------- # WRITE AHEAD LOG #--------------------------------------------------------------------------- # - Settings - #fsync = true # turns forced synchronization on or off #wal_sync_method = fsync # the default varies across platforms: # fsync, fdatasync, open_sync, or open_datasync wal_buffers = 32 # min 4, 8KB each # - Checkpoints - checkpoint_segments = 30 # in logfile segments, min 1, 16MB each 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_siblings = 5 # range 1-1000 TIA Patrick Hatcher From pgsql-performance-owner@postgresql.org Fri Apr 9 15:03:32 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C0F68D1D1B8 for ; Fri, 9 Apr 2004 15:03:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 90375-03 for ; Fri, 9 Apr 2004 15:03:29 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id C004BD1D07F for ; Fri, 9 Apr 2004 15:03:27 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i39I3RjZ027369; Fri, 9 Apr 2004 14:03:27 -0400 (EDT) To: "Patrick Hatcher" Cc: pgsql-performance@postgresql.org Subject: Re: Upgrading question (recycled transaction log) In-reply-to: References: Comments: In-reply-to "Patrick Hatcher" message dated "Fri, 09 Apr 2004 10:18:12 -0700" Date: Fri, 09 Apr 2004 14:03:27 -0400 Message-ID: <27368.1081533807@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/122 X-Sequence-Number: 6422 "Patrick Hatcher" writes: > Doing an upgrade from 7.3.6 to 7.4.2 and I keep seeing the recycled > transaction log about every 2 mins. For future upgrades, is there > something that can be set so that I don't have as many recycles? Increasing checkpoint_segments ... but you seem to have that pretty high already. regards, tom lane From pgsql-admin-owner@postgresql.org Fri Apr 9 17:10:48 2004 X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E5ACFD1B486 for ; Fri, 9 Apr 2004 17:10:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 32905-01 for ; Fri, 9 Apr 2004 17:10:45 -0300 (ADT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 47879D1BB49 for ; Fri, 9 Apr 2004 17:10:43 -0300 (ADT) 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 i39KAgSx032994 for ; Fri, 9 Apr 2004 20:10:42 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id i39JoZwQ029628 for pgsql-admin@postgresql.org; Fri, 9 Apr 2004 19:50:35 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.admin Subject: Re: [PERFORM] Raw devices vs. Filesystems Date: Fri, 09 Apr 2004 15:34:44 -0400 Organization: cbbrowne Computing Inc Lines: 31 Message-ID: References: <71E37EF6B7DCC1499CEA0316A25683280105786A@loki.wc.globexplorer.net> <200404070909.16123.josh@agliodbs.com> <20040408043304.GA28539@elbereth.noviforum.si> <200404090902.00934.josh@agliodbs.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@news.hub.org X-message-flag: Outlook is rather hackable, isn't it? X-Home-Page: http://www.cbbrowne.com/info/ X-Affero: http://svcs.affero.net/rm.php?r=cbbrowne User-Agent: Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Security Through Obscurity, linux) Cancel-Lock: sha1:Qy8p2wVqpwk1Hpbfyh9IDhTL0QM= To: pgsql-admin@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/186 X-Sequence-Number: 13159 josh@agliodbs.com (Josh Berkus) wrote: >> Well, as I said, that's why I was asking - I'm willing to give it a go >> if nobody can prove me wrong. :) > > Why not? If you have time? True enough. >> I thought you knew - OCFS, OCFS-Tools and OCFSv2 have not only been >> open- source for quite a while now - they're released under the >> GPL. > > Keen! Wonder if we can make them regret it. > > Seriously, if Oracle opened this stuff, it's probably becuase they > used some GPL components in it. It also probably means that it > won't work for anything but Oracle ... It could be that the experiment shows that OCFS isn't all that helpful. Or that it helps cover inadequacies in certain aspects of how Oracle accesses filesystems. If it _does_ show that it is helpful, then that may suggest a filesystem implementation strategy useful for the BSD folks. The main "failure case" would be if the exercise shows that using OCFS is pretty futile. -- select 'cbbrowne' || '@' || 'acm.org'; http://www3.sympatico.ca/cbbrowne/linux.html Do you know where your towel is? From pgsql-performance-owner@postgresql.org Fri Apr 9 16:45:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8894FD1BB49 for ; Fri, 9 Apr 2004 16:45:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14641-10 for ; Fri, 9 Apr 2004 16:45:55 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id C879AD1BA90 for ; Fri, 9 Apr 2004 16:45:53 -0300 (ADT) Received: from [63.195.55.98] (HELO 192.168.1.29) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4807612 for pgsql-performance@postgresql.org; Fri, 09 Apr 2004 12:47:17 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: pgsql-performance@postgresql.org Subject: Performance data for OpenFTS? Date: Fri, 9 Apr 2004 12:46:04 -0700 User-Agent: KMail/1.5.4 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404091246.04438.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/123 X-Sequence-Number: 6423 Hey, Has anyone done performance tests for OpenFTS on a really large database? I was speaking at PerlMongers and somebody asked. -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Mon Apr 12 06:27:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 30C28D1B4BE for ; Mon, 12 Apr 2004 06:27:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19488-06 for ; Mon, 12 Apr 2004 06:27:23 -0300 (ADT) Received: from mwinf0101.wanadoo.fr (smtp1.wanadoo.fr [193.252.22.30]) by svr1.postgresql.org (Postfix) with ESMTP id 876C5D1D070 for ; Mon, 12 Apr 2004 06:27:08 -0300 (ADT) Received: from [192.168.0.2] (AMontsouris-108-1-29-248.w81-53.abo.wanadoo.fr [81.53.158.248]) by mwinf0101.wanadoo.fr (SMTP Server) with ESMTP id AA0C0C0000B9 for ; Mon, 12 Apr 2004 11:26:57 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v613) Content-Transfer-Encoding: quoted-printable Message-Id: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Subject: Index Backward Scan fast / Index Scan slow ! Date: Mon, 12 Apr 2004 11:26:55 +0200 X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/126 X-Sequence-Number: 6426 Hi, I test many times the foolowing query. dps=3D# explain analyze select next_index_time from url order by=20=20 next_index_time desc limit 1; =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 QUERY PLAN ------------------------------------------------------------------------=20 ------------------------------------------------------------------------=20 ---- Limit (cost=3D0.00..2.62 rows=3D1 width=3D4) (actual time=3D56.615..56.6= 16=20=20 rows=3D1 loops=3D1) -> Index Scan Backward using url_next_index_time on url=20=20=20 (cost=3D0.00..768529.55 rows=3D293588 width=3D4) (actual time=3D56.610..56.= 610=20=20 rows=3D1 loops=3D1) Total runtime: 56.669 ms (3 rows) dps=3D# explain analyze select next_index_time from url order by=20=20 next_index_time asc limit 1; =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 QUERY PLAN ------------------------------------------------------------------------=20 ------------------------------------------------------------------------=20 - Limit (cost=3D0.00..2.62 rows=3D1 width=3D4) (actual=20=20 time=3D94879.636..94879.637 rows=3D1 loops=3D1) -> Index Scan using url_next_index_time on url=20=20=20 (cost=3D0.00..768529.55 rows=3D293588 width=3D4) (actual=20=20 time=3D94879.631..94879.631 rows=3D1 loops=3D1) Total runtime: 94879.688 ms (3 rows) How to optimize the last query ? (~ 2000 times slower than the first=20=20 one) I suppose there is some odd distribution of data in the index ? Is the solution to reindex data ? Cordialement, Jean-G=E9rard Pailloncy From pgsql-performance-owner@postgresql.org Mon Apr 12 08:52:22 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 54E6AD1CCB7 for ; Mon, 12 Apr 2004 08:52:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 57198-06 for ; Mon, 12 Apr 2004 08:52:23 -0300 (ADT) Received: from trade-india.com (unknown [61.16.154.82]) by svr1.postgresql.org (Postfix) with SMTP id EFB68D1DC69 for ; Mon, 12 Apr 2004 08:52:17 -0300 (ADT) Received: (qmail 15532 invoked from network); 12 Apr 2004 11:40:01 -0000 Received: from valium.trade-india-local.com (HELO trade-india.com) (192.168.0.72) by system66.trade-india-local.com with SMTP; 12 Apr 2004 11:40:01 -0000 Message-ID: <407A8369.9090205@trade-india.com> Date: Mon, 12 Apr 2004 17:24:17 +0530 From: Rajesh Kumar Mallah Organization: Infocom Network Limited User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Effect of too many columns 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: 200404/127 X-Sequence-Number: 6427 Greetings, Is there any performance penalty of having too many columns in a table in terms of read and write speeds. To order to keep operational queries simple (avoid joins) we plan to add columns in the main customer dimension table. Adding more columns also means increase in concurrency in the table as more and more applications will access the same table. Any ideas if its better to split the table application wise or is it ok? Regds mallah. From pgsql-performance-owner@postgresql.org Mon Apr 12 09:13:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8E48BD1BAF8 for ; Mon, 12 Apr 2004 09:13:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 61632-05 for ; Mon, 12 Apr 2004 09:14:00 -0300 (ADT) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id A9395D1BAA5 for ; Mon, 12 Apr 2004 09:13:56 -0300 (ADT) Received: (qmail 20507 invoked by uid 500); 12 Apr 2004 12:17:40 -0000 Date: Mon, 12 Apr 2004 07:17:40 -0500 From: Bruno Wolff III To: Rajesh Kumar Mallah Cc: pgsql-performance@postgresql.org Subject: Re: Effect of too many columns Message-ID: <20040412121740.GC20278@wolff.to> Mail-Followup-To: Rajesh Kumar Mallah , pgsql-performance@postgresql.org References: <407A8369.9090205@trade-india.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <407A8369.9090205@trade-india.com> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/128 X-Sequence-Number: 6428 On Mon, Apr 12, 2004 at 17:24:17 +0530, Rajesh Kumar Mallah wrote: > > Greetings, > > Is there any performance penalty of having too many columns in > a table in terms of read and write speeds. > > To order to keep operational queries simple (avoid joins) we plan to > add columns in the main customer dimension table. > > Adding more columns also means increase in concurrency in the table > as more and more applications will access the same table. > > Any ideas if its better to split the table application wise or is it ok? This is normally a bad idea. If you properly implement constraints in what is effectively a materialized view, you might end up with a slower system, depending on your mix of queries. (Generally updating will take more resources.) So you probably want to test your new design under a simulated normal load to see if it actually speeds things up in your case before making the change. From pgsql-performance-owner@postgresql.org Mon Apr 12 09:28:47 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 70ACED1BB8C for ; Mon, 12 Apr 2004 09:28:44 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69334-03 for ; Mon, 12 Apr 2004 09:28:46 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 7AB38D1BAAD for ; Mon, 12 Apr 2004 09:28:42 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3CCSJIh008580; Mon, 12 Apr 2004 08:28:19 -0400 (EDT) To: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Cc: pgsql-performance@postgresql.org Subject: Re: Index Backward Scan fast / Index Scan slow ! In-reply-to: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> References: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> Comments: In-reply-to =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= message dated "Mon, 12 Apr 2004 11:26:55 +0200" Date: Mon, 12 Apr 2004 08:28:19 -0400 Message-ID: <8579.1081772899@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/129 X-Sequence-Number: 6429 =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= writes: > How to optimize the last query ? (~ 2000 times slower than the first > one) > I suppose there is some odd distribution of data in the index ? Looks to me like a whole lot of dead rows at the left end of the index. Have you VACUUMed this table lately? It would be interesting to see what VACUUM VERBOSE has to say about it. > Is the solution to reindex data ? In 7.4 a VACUUM should be sufficient ... or at least, if it isn't I'd like to know why not before you destroy the evidence by reindexing. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 12 11:39:39 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 85B32D1D070 for ; Mon, 12 Apr 2004 11:39:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 13530-01 for ; Mon, 12 Apr 2004 11:39:37 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id C173BD1BB8C for ; Mon, 12 Apr 2004 11:39:23 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id <2A89N0WZ>; Mon, 12 Apr 2004 10:39:24 -0400 Message-ID: From: "Shea,Dan [CIS]" To: Postgres Performance Subject: Deleting certain duplicates Date: Mon, 12 Apr 2004 10:39:22 -0400 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: 200404/130 X-Sequence-Number: 6430 We have a large database which recently increased dramatically due to a change in our insert program allowing all entries. PWFPM_DEV=# select relname,relfilenode,reltuples from pg_class where relname = 'forecastelement'; relname | relfilenode | reltuples -----------------+-------------+------------- forecastelement | 361747866 | 4.70567e+08 Column | Type | Modifiers ----------------+-----------------------------+----------- version | character varying(99) | origin | character varying(10) | timezone | character varying(99) | region_id | character varying(20) | wx_element | character varying(99) | value | character varying(99) | flag | character(3) | units | character varying(99) | valid_time | timestamp without time zone | issue_time | timestamp without time zone | next_forecast | timestamp without time zone | reception_time | timestamp without time zone | The program is supposed to check to ensure that all fields but the reception_time are unique using a select statement, and if so, insert it. Due an error in a change, reception time was included in the select to check for duplicates. The reception_time is created by a program creating the dat file to insert. Essentially letting all duplicate files to be inserted. I tried the delete query below. PWFPM_DEV=# delete from forecastelement where oid not in (select min(oid) from forecastelement group by version,origin,timezone,region_id,wx_element,value,flag,units,valid_time,iss ue_time,next_forecast); It ran for 3 days creating what I assume is an index in pgsql_tmp of the group by statement. The query ended up failing with "dateERROR:write failed". Well the long weekend is over and we do not have the luxury of trying this again. So I was thinking maybe of doing the deletion in chunks, perhaps based on reception time. Are there any suggestions for a better way to do this, or using multiple queries to delete selectively a week at a time based on the reception_time. I would say there are a lot of duplicate entries between mid march to the first week of April. From pgsql-performance-owner@postgresql.org Mon Apr 12 12:18:47 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8146FD1E1DD for ; Mon, 12 Apr 2004 12:18:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27226-05 for ; Mon, 12 Apr 2004 12:18:43 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id A18D5D1E2A5 for ; Mon, 12 Apr 2004 12:18:42 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id <2A893B6P>; Mon, 12 Apr 2004 11:18:40 -0400 Message-ID: From: "Shea,Dan [CIS]" To: "Shea,Dan [CIS]" , Postgres Performance Subject: Re: Deleting certain duplicates Date: Mon, 12 Apr 2004 11:18:31 -0400 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: 200404/131 X-Sequence-Number: 6431 The index is Indexes: "forecastelement_rwv_idx" btree (region_id, wx_element, valid_time) -----Original Message----- From: Shea,Dan [CIS] [mailto:Dan.Shea@ec.gc.ca] Sent: Monday, April 12, 2004 10:39 AM To: Postgres Performance Subject: [PERFORM] Deleting certain duplicates We have a large database which recently increased dramatically due to a change in our insert program allowing all entries. PWFPM_DEV=# select relname,relfilenode,reltuples from pg_class where relname = 'forecastelement'; relname | relfilenode | reltuples -----------------+-------------+------------- forecastelement | 361747866 | 4.70567e+08 Column | Type | Modifiers ----------------+-----------------------------+----------- version | character varying(99) | origin | character varying(10) | timezone | character varying(99) | region_id | character varying(20) | wx_element | character varying(99) | value | character varying(99) | flag | character(3) | units | character varying(99) | valid_time | timestamp without time zone | issue_time | timestamp without time zone | next_forecast | timestamp without time zone | reception_time | timestamp without time zone | The program is supposed to check to ensure that all fields but the reception_time are unique using a select statement, and if so, insert it. Due an error in a change, reception time was included in the select to check for duplicates. The reception_time is created by a program creating the dat file to insert. Essentially letting all duplicate files to be inserted. I tried the delete query below. PWFPM_DEV=# delete from forecastelement where oid not in (select min(oid) from forecastelement group by version,origin,timezone,region_id,wx_element,value,flag,units,valid_time,iss ue_time,next_forecast); It ran for 3 days creating what I assume is an index in pgsql_tmp of the group by statement. The query ended up failing with "dateERROR:write failed". Well the long weekend is over and we do not have the luxury of trying this again. So I was thinking maybe of doing the deletion in chunks, perhaps based on reception time. Are there any suggestions for a better way to do this, or using multiple queries to delete selectively a week at a time based on the reception_time. I would say there are a lot of duplicate entries between mid march to the first week of April. ---------------------------(end of broadcast)--------------------------- TIP 8: explain analyze is your friend From pgsql-performance-owner@postgresql.org Mon Apr 12 12:40:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A6CF0D1D2C6 for ; Mon, 12 Apr 2004 12:40:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 43328-04 for ; Mon, 12 Apr 2004 12:40:32 -0300 (ADT) Received: from mail.autorevenue.com (ip-208.178.167.106.gblx.net [208.178.167.106]) by svr1.postgresql.org (Postfix) with ESMTP id 882DDD1D1B8 for ; Mon, 12 Apr 2004 12:40:31 -0300 (ADT) Received: from jeremydunn ([192.168.1.79]) (authenticated) by mail.autorevenue.com (8.11.6/8.11.6) with ESMTP id i3CFeVY24818 for ; Mon, 12 Apr 2004 11:40:31 -0400 Reply-To: From: "Jeremy Dunn" To: "Postgresql Performance" Subject: index v. seqscan for certain values Date: Mon, 12 Apr 2004 11:40:28 -0400 Organization: AutoRevenue Message-ID: <000901c420a4$7b733650$4f01a8c0@jeremydunn> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_000A_01C42082.F4619650" X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.2627 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/132 X-Sequence-Number: 6432 This is a multi-part message in MIME format. ------=_NextPart_000_000A_01C42082.F4619650 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit I've searched the archives and can't find an answer to this seemingly simple question. Apologies if it's too common. The table in question has ~1.3M rows. It has 85 columns, 5 of which have single-column indexes. The column in question (CID) has 183 distinct values. For these values, the largest has ~38,000 rows, and the smallest has 1 row. About 30 values have < 100 rows, and about 10 values have > 20,000 rows. The database is 7.2.3 running on RedHat 7.1. (we are in process of upgrading to PG 7.4.2) All of the query plan options are enabled, and the cpu costs are set to the default values. ( cpu_tuple_cost is 0.01, cpu_index_tuple_cost is 0.001). The database is VACUUM'd every night. The problem: A simply query: select count(*) from xxx where CID= where is a CID value which has relatively few rows, returns a plan using the index on that column. explain analyze select count(*) from xxx where cid=869366; Aggregate (cost=19136.33..19136.33 rows=1 width=0) (actual time=78.49..78.49 rows=1 loops=1) -> Index Scan using xxx_cid on emailrcpts (cost=0.00..19122.21 rows=5648 width=0) (actual time=63.40..78.46 rows=1 loops=1) Total runtime: 78.69 msec The same plan is true for values which have up to about 20,000 rows: explain analyze select count(*) from xxx where cid=6223341; Aggregate (cost=74384.19..74384.19 rows=1 width=0) (actual time=11614.89..11614.89 rows=1 loops=1) -> Index Scan using xxx_cid on emailrcpts (cost=0.00..74329.26 rows=21974 width=0) (actual time=35.75..11582.10 rows=20114 loops=1) Total runtime: 11615.05 msec However for the values that have > 20,000 rows, the plan changes to a sequential scan, which is proportionately much slower. explain analyze select count(*) from xxx where cid=7191032; Aggregate (cost=97357.61..97357.61 rows=1 width=0) (actual time=46427.81..46427.82 rows=1 loops=1) -> Seq Scan on xxx (cost=0.00..97230.62 rows=50792 width=0) (actual time=9104.45..46370.27 rows=37765 loops=1) Total runtime: 46428.00 msec The question: why does the planner consider a sequential scan to be better for these top 10 values? In terms of elapsed time it is more than twice as slow, proportionate to an index scan for the same number of rows. What I tried: A) alter table xxx alter column cid set statistics 500; analyze xxx; This does not affect the results. B) dropped/rebuilt the index, with no improvement. C) decreasing cpu_index_tuple_cost by a factor of up to 1000, with no success D) force an index scan for the larger values by using a very high value for cpu_tuple_cost (e.g. .5) but this doesn't seem like a wise thing to do. Your thoughts appreciated in advance! - Jeremy 7+ years experience in Oracle performance-tuning relatively new to postgresql ------=_NextPart_000_000A_01C42082.F4619650 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Message
I've sear= ched the=20 archives and can't find an answer to this seemingly simple question. = =20 Apologies if it's too common.
 
The table= in=20 question has ~1.3M rows.  It has 85 columns, 5 of which have single-co= lumn=20 indexes.
 
The colum= n in=20 question (CID) has 183 distinct values.  For these values, the largest= has=20 ~38,000 rows, and the smallest has 1 row.  About 30 values have < 1= 00=20 rows, and about 10 values have > 20,000 rows.
 
The datab= ase is=20 7.2.3 running on RedHat 7.1. (we are in process of upgrading to=20 PG 7.4.2)    All of the query plan options are enabled,= and=20 the cpu costs are set to the default values. ( cpu_tuple_cost is 0.01,= =20 cpu_index_tuple_cost is 0.001).  The database is VACUUM'd every=20 night.
 
The=20 problem:
A simply= =20 query:
    select count(*) from xxx where CID=3D<smalval>=
where=20 <smalval> is a CID value which has relatively few rows, returns a pla= n=20 using the index on that column.
 
 &nb= sp; explain=20 analyze select count(*) from xxx where cid=3D869366;
   Aggrega= te =20 (cost=3D19136.33..19136.33 rows=3D1 width=3D0) (actual time=3D78.49..78.49 = rows=3D1=20 loops=3D1)
     ->  Index Scan using xxx_cid= on=20 emailrcpts  (cost=3D0.00..19122.21 rows=3D5648 width=3D0) (actual=20 time=3D63.40..78.46 rows=3D1 loops=3D1)
   Total runtime: 78.6= 9=20 msec
 
The same = plan is=20 true for values which have up to about 20,000 rows:
 
 &nb= sp; explain=20 analyze select count(*) from xxx where cid=3D6223341;
   Aggrega= te =20 (cost=3D74384.19..74384.19 rows=3D1 width=3D0) (actual time=3D11614.89..116= 14.89 rows=3D1=20 loops=3D1)
     ->  Index Scan using xxx_cid= on=20 emailrcpts  (cost=3D0.00..74329.26 rows=3D21974 width=3D0) (actual=20 time=3D35.75..11582.10 rows=3D20114 loops=3D1)
   Total runtim= e: 11615.05=20 msec
However f= or the=20 values that have > 20,000 rows, the plan changes to a sequential scan, w= hich=20 is proportionately much slower.
 
&nb= sp; =20 explain analyze select count(*) from xxx where cid=3D7191032;
  = =20 Aggregate  (cost=3D97357.61..97357.61 rows=3D1 width=3D0) (actu= al=20 time=3D46427.81..46427.82 rows=3D1 loops=3D1)
    ->=20   Seq Scan on xxx=20 (cost=3D0.00..97230.62 rows=3D50792 width=3D0) (actual time=3D9104.4= 5..46370.27=20 rows=3D37765 loops=3D1)
  &nb= sp;=20 Total runtime: 46428.00 msec
 
 
The quest= ion: why=20 does the planner consider a sequential scan to be better for these top 10= =20 values?  In terms of elapsed time it is more than twice as slow,=20 proportionate to an index scan for the same number of rows.
 
What I=20 tried:
 
A) alter table xxx alter column cid set statistics 500;  &n= bsp;=20
    analyze xxx;
This does= not affect=20 the results.
 
B)=20  dropped/rebuilt the index, with no improvement.
 
C) decrea= sing=20 cpu_index_tuple_cost by a factor of up to 1000, with no=20 success
 
D) force = an index=20 scan for the larger values by using a very high value for cpu_tuple_cost (e= .g.=20 .5) but this doesn't seem like a wise thing to do.
 
Your thou= ghts=20 appreciated in advance!
 
-=20 Jeremy 
 
7+ years=20 experience in Oracle performance-tuning
relatively new to postgresql
------=_NextPart_000_000A_01C42082.F4619650-- From pgsql-performance-owner@postgresql.org Mon Apr 12 13:09:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A8D88D1D501 for ; Mon, 12 Apr 2004 13:09:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66924-04 for ; Mon, 12 Apr 2004 13:09:21 -0300 (ADT) Received: from internet.potentialtech.com (h-66-167-251-6.phlapafg.covad.net [66.167.251.6]) by svr1.postgresql.org (Postfix) with ESMTP id 31698D1BAAD for ; Mon, 12 Apr 2004 13:09:20 -0300 (ADT) Received: from potentialtech.com (pa-plum1c-102.pit.adelphia.net [24.53.179.102]) by internet.potentialtech.com (Postfix) with ESMTP id 34A2769A71; Mon, 12 Apr 2004 12:09:19 -0400 (EDT) Message-ID: <407ABF2B.80306@potentialtech.com> Date: Mon, 12 Apr 2004 12:09:15 -0400 From: Bill Moran User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031005 X-Accept-Language: en-us, en MIME-Version: 1.0 To: jdunn@autorevenue.com Cc: Postgresql Performance Subject: Re: index v. seqscan for certain values References: <000901c420a4$7b733650$4f01a8c0@jeremydunn> In-Reply-To: <000901c420a4$7b733650$4f01a8c0@jeremydunn> 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: 200404/133 X-Sequence-Number: 6433 Quick bit of input, since you didn't mention it. How often do you run ANALYZE? I found it interesting that a database I was doing tests on sped up by a factor of 20 after ANALYZE. If your data changes a lot, you should probably schedule ANALYZE to run with VACUUM. Jeremy Dunn wrote: > I've searched the archives and can't find an answer to this seemingly > simple question. Apologies if it's too common. > > The table in question has ~1.3M rows. It has 85 columns, 5 of which > have single-column indexes. > > The column in question (CID) has 183 distinct values. For these values, > the largest has ~38,000 rows, and the smallest has 1 row. About 30 > values have < 100 rows, and about 10 values have > 20,000 rows. > > The database is 7.2.3 running on RedHat 7.1. (we are in process of > upgrading to PG 7.4.2) All of the query plan options are enabled, and > the cpu costs are set to the default values. ( cpu_tuple_cost is 0.01, > cpu_index_tuple_cost is 0.001). The database is VACUUM'd every night. > > The problem: > A simply query: > select count(*) from xxx where CID= > where is a CID value which has relatively few rows, returns a > plan using the index on that column. > > explain analyze select count(*) from xxx where cid=869366; > Aggregate (cost=19136.33..19136.33 rows=1 width=0) (actual > time=78.49..78.49 rows=1 loops=1) > -> Index Scan using xxx_cid on emailrcpts (cost=0.00..19122.21 > rows=5648 width=0) (actual time=63.40..78.46 rows=1 loops=1) > Total runtime: 78.69 msec > > The same plan is true for values which have up to about 20,000 rows: > > explain analyze select count(*) from xxx where cid=6223341; > Aggregate (cost=74384.19..74384.19 rows=1 width=0) (actual > time=11614.89..11614.89 rows=1 loops=1) > -> Index Scan using xxx_cid on emailrcpts (cost=0.00..74329.26 > rows=21974 width=0) (actual time=35.75..11582.10 rows=20114 loops=1) > Total runtime: 11615.05 msec > However for the values that have > 20,000 rows, the plan changes to a > sequential scan, which is proportionately much slower. > > explain analyze select count(*) from xxx where cid=7191032; > Aggregate (cost=97357.61..97357.61 rows=1 width=0) (actual > time=46427.81..46427.82 rows=1 loops=1) > -> Seq Scan on xxx (cost=0.00..97230.62 rows=50792 width=0) > (actual time=9104.45..46370.27 rows=37765 loops=1) > Total runtime: 46428.00 msec > > > The question: why does the planner consider a sequential scan to be > better for these top 10 values? In terms of elapsed time it is more > than twice as slow, proportionate to an index scan for the same number > of rows. > > What I tried: > > A) alter table xxx alter column cid set statistics 500; > analyze xxx; > This does not affect the results. > > B) dropped/rebuilt the index, with no improvement. > > C) decreasing cpu_index_tuple_cost by a factor of up to 1000, with no > success > > D) force an index scan for the larger values by using a very high value > for cpu_tuple_cost (e.g. .5) but this doesn't seem like a wise thing to do. > > Your thoughts appreciated in advance! > > - Jeremy > > 7+ years experience in Oracle performance-tuning > relatively new to postgresql -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Mon Apr 12 14:08:11 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 403D3D1BA90 for ; Mon, 12 Apr 2004 14:08:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 84146-06 for ; Mon, 12 Apr 2004 14:08:09 -0300 (ADT) Received: from mail.autorevenue.com (ip-208.178.167.106.gblx.net [208.178.167.106]) by svr1.postgresql.org (Postfix) with ESMTP id C0D37D1B515 for ; Mon, 12 Apr 2004 14:08:07 -0300 (ADT) Received: from jeremydunn ([192.168.1.79]) (authenticated) by mail.autorevenue.com (8.11.6/8.11.6) with ESMTP id i3CH87Y26971 for ; Mon, 12 Apr 2004 13:08:07 -0400 Reply-To: From: "Jeremy Dunn" To: "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values Date: Mon, 12 Apr 2004 13:08:05 -0400 Organization: AutoRevenue Message-ID: <001801c420b0$b8b79b30$4f01a8c0@jeremydunn> 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 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 In-Reply-To: <407ABF2B.80306@potentialtech.com> Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/134 X-Sequence-Number: 6434 Sorry I should have written that we do VACUUM VERBOSE ANALYZE every night. - Jeremy -----Original Message----- From: pgsql-performance-owner@postgresql.org [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of Bill Moran Sent: Monday, April 12, 2004 12:09 PM To: jdunn@autorevenue.com Cc: Postgresql Performance Subject: Re: [PERFORM] index v. seqscan for certain values Quick bit of input, since you didn't mention it. How often do you run ANALYZE? I found it interesting that a database I was doing tests on sped up by a factor of 20 after ANALYZE. If your data changes a lot, you should probably schedule ANALYZE to run with VACUUM. Jeremy Dunn wrote: > I've searched the archives and can't find an answer to this seemingly > simple question. Apologies if it's too common. > > The table in question has ~1.3M rows. It has 85 columns, 5 of which > have single-column indexes. > > The column in question (CID) has 183 distinct values. For these > values, > the largest has ~38,000 rows, and the smallest has 1 row. About 30 > values have < 100 rows, and about 10 values have > 20,000 rows. > > The database is 7.2.3 running on RedHat 7.1. (we are in process of > upgrading to PG 7.4.2) All of the query plan options are enabled, and > the cpu costs are set to the default values. ( cpu_tuple_cost is 0.01, > cpu_index_tuple_cost is 0.001). The database is VACUUM'd every night. > > The problem: > A simply query: > select count(*) from xxx where CID= > where is a CID value which has relatively few rows, returns > a > plan using the index on that column. > > explain analyze select count(*) from xxx where cid=869366; > Aggregate (cost=19136.33..19136.33 rows=1 width=0) (actual > time=78.49..78.49 rows=1 loops=1) > -> Index Scan using xxx_cid on emailrcpts (cost=0.00..19122.21 > rows=5648 width=0) (actual time=63.40..78.46 rows=1 loops=1) > Total runtime: 78.69 msec > > The same plan is true for values which have up to about 20,000 rows: > > explain analyze select count(*) from xxx where cid=6223341; > Aggregate (cost=74384.19..74384.19 rows=1 width=0) (actual > time=11614.89..11614.89 rows=1 loops=1) > -> Index Scan using xxx_cid on emailrcpts (cost=0.00..74329.26 > rows=21974 width=0) (actual time=35.75..11582.10 rows=20114 loops=1) > Total runtime: 11615.05 msec > However for the values that have > 20,000 rows, the plan changes to a > sequential scan, which is proportionately much slower. > > explain analyze select count(*) from xxx where cid=7191032; > Aggregate (cost=97357.61..97357.61 rows=1 width=0) (actual > time=46427.81..46427.82 rows=1 loops=1) > -> Seq Scan on xxx (cost=0.00..97230.62 rows=50792 width=0) > (actual time=9104.45..46370.27 rows=37765 loops=1) > Total runtime: 46428.00 msec > > > The question: why does the planner consider a sequential scan to be > better for these top 10 values? In terms of elapsed time it is more > than twice as slow, proportionate to an index scan for the same number > of rows. > > What I tried: > > A) alter table xxx alter column cid set statistics 500; > analyze xxx; > This does not affect the results. > > B) dropped/rebuilt the index, with no improvement. > > C) decreasing cpu_index_tuple_cost by a factor of up to 1000, with no > success > > D) force an index scan for the larger values by using a very high > value > for cpu_tuple_cost (e.g. .5) but this doesn't seem like a wise thing to do. > > Your thoughts appreciated in advance! > > - Jeremy > > 7+ years experience in Oracle performance-tuning > relatively new to postgresql -- Bill Moran Potential Technologies http://www.potentialtech.com ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Mon Apr 12 14:40:02 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 141D2D1D300 for ; Mon, 12 Apr 2004 14:39:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 98945-05 for ; Mon, 12 Apr 2004 14:39:52 -0300 (ADT) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 6DFD2D1D282 for ; Mon, 12 Apr 2004 14:39:51 -0300 (ADT) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 3FCBA356E5; Mon, 12 Apr 2004 10:39:51 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 3E2DA3569F; Mon, 12 Apr 2004 10:39:51 -0700 (PDT) Date: Mon, 12 Apr 2004 10:39:51 -0700 (PDT) From: Stephan Szabo To: Jeremy Dunn Cc: Postgresql Performance Subject: Re: index v. seqscan for certain values In-Reply-To: <000901c420a4$7b733650$4f01a8c0@jeremydunn> Message-ID: <20040412103711.G16827@megazone.bigpanda.com> References: <000901c420a4$7b733650$4f01a8c0@jeremydunn> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/135 X-Sequence-Number: 6435 On Mon, 12 Apr 2004, Jeremy Dunn wrote: > explain analyze select count(*) from xxx where cid=6223341; > Aggregate (cost=74384.19..74384.19 rows=1 width=0) (actual > time=11614.89..11614.89 rows=1 loops=1) > -> Index Scan using xxx_cid on emailrcpts (cost=0.00..74329.26 > rows=21974 width=0) (actual time=35.75..11582.10 rows=20114 loops=1) > Total runtime: 11615.05 msec > > However for the values that have > 20,000 rows, the plan changes to a > sequential scan, which is proportionately much slower. > > explain analyze select count(*) from xxx where cid=7191032; > Aggregate (cost=97357.61..97357.61 rows=1 width=0) (actual > time=46427.81..46427.82 rows=1 loops=1) > -> Seq Scan on xxx (cost=0.00..97230.62 rows=50792 width=0) > (actual time=9104.45..46370.27 rows=37765 loops=1) > Total runtime: 46428.00 msec > > The question: why does the planner consider a sequential scan to be > better for these top 10 values? In terms of elapsed time it is more > than twice as slow, proportionate to an index scan for the same number > of rows. One thing to do is to set enable_seqscan=off and run the above and compare the estimated and real costs. It may be possible to lower random_page_cost to a still reasonable number in order to move the point of the switchover to seqscan. From pgsql-performance-owner@postgresql.org Mon Apr 12 14:51:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A9881D1C9FB for ; Mon, 12 Apr 2004 14:51:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04791-04 for ; Mon, 12 Apr 2004 14:51:30 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id AA3ADD1C9C7 for ; Mon, 12 Apr 2004 14:51:28 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3CHpSpd012038; Mon, 12 Apr 2004 13:51:28 -0400 (EDT) To: jdunn@autorevenue.com Cc: "Postgresql Performance" Subject: Re: index v. seqscan for certain values In-reply-to: <000901c420a4$7b733650$4f01a8c0@jeremydunn> References: <000901c420a4$7b733650$4f01a8c0@jeremydunn> Comments: In-reply-to "Jeremy Dunn" message dated "Mon, 12 Apr 2004 11:40:28 -0400" Date: Mon, 12 Apr 2004 13:51:28 -0400 Message-ID: <12037.1081792288@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/136 X-Sequence-Number: 6436 "Jeremy Dunn" writes: > The question: why does the planner consider a sequential scan to be > better for these top 10 values? At some point a seqscan *will* be better. In the limit, if the key being sought is common enough to occur on every page of the table, it's certain that a seqscan will require less I/O than an indexscan (because reading the index isn't actually saving you any heap fetches). In practice the breakeven point is less than that because Unix kernels are better at handling sequential than random access. Your gripe appears to be basically that the planner's idea of the breakeven point is off a bit. It looks to me like it's within about a factor of 2 of being right, though, which is not all that bad when it's using generic cost parameters. > A) alter table xxx alter column cid set statistics 500; > analyze xxx; > This does not affect the results. It probably improved the accuracy of the row count estimates, no? The estimate you show for cid=7191032 is off by more than 25% (37765 vs 50792), which seems like a lot of error for one of the most common values in the table. (I hope that was with default stats target and not 500.) That leads directly to a 25% overestimate of the cost of an indexscan, while having IIRC no impact on the cost of a seqscan. Since the cost ratio was more than 25%, this didn't change the selected plan, but you want to fix that error as best you can before you move on to tweaking cost parameters. > C) decreasing cpu_index_tuple_cost by a factor of up to 1000, with no > success Wrong thing. You should be tweaking random_page_cost. Looks to me like a value near 2 might be appropriate for your setup. Also it is likely appropriate to increase effective_cache_size, which is awfully small in the default configuration. I'd set that to something related to your available RAM before trying to home in on a suitable random_page_cost. AFAIK hardly anyone bothers with changing the cpu_xxx costs ... regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 12 16:02:16 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3FB29D1DC69 for ; Mon, 12 Apr 2004 16:02:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33042-04 for ; Mon, 12 Apr 2004 16:02:08 -0300 (ADT) Received: from mwinf0102.wanadoo.fr (smtp1.wanadoo.fr [193.252.22.30]) by svr1.postgresql.org (Postfix) with ESMTP id C9D1DD1D085 for ; Mon, 12 Apr 2004 16:02:05 -0300 (ADT) Received: from [192.168.0.2] (AMontsouris-108-1-29-248.w81-53.abo.wanadoo.fr [81.53.158.248]) by mwinf0102.wanadoo.fr (SMTP Server) with ESMTP id 0BA6A1BE7A42 for ; Mon, 12 Apr 2004 21:02:06 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v613) In-Reply-To: <8579.1081772899@sss.pgh.pa.us> References: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> <8579.1081772899@sss.pgh.pa.us> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Subject: Re: Index Backward Scan fast / Index Scan slow ! Date: Mon, 12 Apr 2004 21:02:02 +0200 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/137 X-Sequence-Number: 6437 Hi, >> How to optimize the last query ? (~ 2000 times slower than the first >> one) >> I suppose there is some odd distribution of data in the index ? > > Looks to me like a whole lot of dead rows at the left end of the index. > Have you VACUUMed this table lately? From pg_autovacuum: [2004-04-10 05:45:39 AM] Performing: ANALYZE "public"."url" [2004-04-10 11:13:25 AM] Performing: ANALYZE "public"."url" [2004-04-10 03:12:14 PM] Performing: VACUUM ANALYZE "public"."url" [2004-04-11 04:58:29 AM] Performing: ANALYZE "public"."url" [2004-04-11 03:48:25 PM] Performing: ANALYZE "public"."url" [2004-04-11 09:21:31 PM] Performing: ANALYZE "public"."url" [2004-04-12 03:24:06 AM] Performing: ANALYZE "public"."url" [2004-04-12 07:20:08 AM] Performing: VACUUM ANALYZE "public"."url" > It would be interesting to see > what VACUUM VERBOSE has to say about it. dps=3D# VACUUM VERBOSE url; INFO: vacuuming "public.url" INFO: index "url_pkey" now contains 348972 row versions in 2344 pages DETAIL: 229515 index row versions were removed. 41 index pages have been deleted, 41 are currently reusable. CPU 0.32s/1.40u sec elapsed 70.66 sec. INFO: index "url_crc" now contains 215141 row versions in 497 pages DETAIL: 108343 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.06s/0.96u sec elapsed 9.13 sec. INFO: index "url_seed" now contains 348458 row versions in 2987 pages DETAIL: 229515 index row versions were removed. 345 index pages have been deleted, 345 are currently reusable. CPU 0.40s/2.38u sec elapsed 74.26 sec. INFO: index "url_referrer" now contains 349509 row versions in 1964=20 pages DETAIL: 229515 index row versions were removed. 65 index pages have been deleted, 65 are currently reusable. CPU 0.34s/1.53u sec elapsed 127.37 sec. INFO: index "url_next_index_time" now contains 349519 row versions in=20 3534 pages DETAIL: 229515 index row versions were removed. 3071 index pages have been deleted, 2864 are currently reusable. CPU 0.32s/0.67u sec elapsed 76.25 sec. INFO: index "url_status" now contains 349520 row versions in 3465 pages DETAIL: 229515 index row versions were removed. 2383 index pages have been deleted, 2256 are currently reusable. CPU 0.35s/0.85u sec elapsed 89.25 sec. INFO: index "url_bad_since_time" now contains 349521 row versions in=20 2017 pages DETAIL: 229515 index row versions were removed. 38 index pages have been deleted, 38 are currently reusable. CPU 0.54s/1.46u sec elapsed 83.77 sec. INFO: index "url_hops" now contains 349620 row versions in 3558 pages DETAIL: 229515 index row versions were removed. 1366 index pages have been deleted, 1356 are currently reusable. CPU 0.43s/0.91u sec elapsed 132.14 sec. INFO: index "url_siteid" now contains 350551 row versions in 3409 pages DETAIL: 229515 index row versions were removed. 2310 index pages have been deleted, 2185 are currently reusable. CPU 0.35s/1.01u sec elapsed 85.08 sec. INFO: index "url_serverid" now contains 350552 row versions in 3469=20 pages DETAIL: 229515 index row versions were removed. 1014 index pages have been deleted, 1009 are currently reusable. CPU 0.54s/1.01u sec elapsed 120.40 sec. INFO: index "url_url" now contains 346563 row versions in 6494 pages DETAIL: 213608 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 1.35s/2.07u sec elapsed 285.05 sec. INFO: index "url_last_mod_time" now contains 346734 row versions in=20 1106 pages DETAIL: 213608 index row versions were removed. 27 index pages have been deleted, 17 are currently reusable. CPU 0.17s/0.95u sec elapsed 17.92 sec. INFO: "url": removed 229515 row versions in 4844 pages DETAIL: CPU 0.53s/1.26u sec elapsed 375.64 sec. INFO: "url": found 229515 removable, 310913 nonremovable row versions=20 in 26488 pages DETAIL: 29063 dead row versions cannot be removed yet. There were 3907007 unused item pointers. 192 pages are entirely empty. CPU 7.78s/17.09u sec elapsed 3672.29 sec. INFO: vacuuming "pg_toast.pg_toast_127397204" INFO: index "pg_toast_127397204_index" now contains 0 row versions in=20 1 pages DETAIL: 0 index pages have been deleted, 0 are currently reusable. CPU 0.00s/0.00u sec elapsed 0.06 sec. INFO: "pg_toast_127397204": found 0 removable, 0 nonremovable row=20 versions in 0 pages DETAIL: 0 dead row versions cannot be removed yet. There were 0 unused item pointers. 0 pages are entirely empty. CPU 0.00s/0.00u sec elapsed 0.07 sec. VACUUM >> Is the solution to reindex data ? > > In 7.4 a VACUUM should be sufficient ... or at least, if it isn't > I'd like to know why not before you destroy the evidence by reindexing. Yes, of course. Cordialement, Jean-G=E9rard Pailloncy From pgsql-performance-owner@postgresql.org Mon Apr 12 16:03:48 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E21A9D1B486 for ; Mon, 12 Apr 2004 16:03:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34274-08 for ; Mon, 12 Apr 2004 16:03:45 -0300 (ADT) Received: from mwinf0102.wanadoo.fr (smtp1.wanadoo.fr [193.252.22.30]) by svr1.postgresql.org (Postfix) with ESMTP id 11E68D1B4BE for ; Mon, 12 Apr 2004 16:03:44 -0300 (ADT) Received: from [192.168.0.2] (AMontsouris-108-1-29-248.w81-53.abo.wanadoo.fr [81.53.158.248]) by mwinf0102.wanadoo.fr (SMTP Server) with ESMTP id 03B201BE7B05 for ; Mon, 12 Apr 2004 21:03:45 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v613) Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed Resent-Date: Mon, 12 Apr 2004 21:03:42 +0200 Message-Id: <1D4DEBDC-8CB4-11D8-80FD-000A95DE2550@ifrance.com> Content-Transfer-Encoding: quoted-printable Resent-To: pgsql-performance@postgresql.org Resent-Message-Id: From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Subject: =?ISO-8859-1?Q?Re:__Index_Backward_Scan_fast_/_Index_Sc?= =?ISO-8859-1?Q?an_slow_!__=28Modifi=E9_par_Pailloncy_Jean-G=E9ra?= =?ISO-8859-1?Q?rd=29?= Resent-From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Date: Mon, 12 Apr 2004 21:02:02 +0200 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/138 X-Sequence-Number: 6438 Hi, > In 7.4 a VACUUM should be sufficient ... or at least, if it isn't Atfer VACUUM: dps=3D# explain analyze select next_index_time from url order by=20=20 next_index_time desc limit 1; =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 QUERY PLAN ------------------------------------------------------------------------=20 ------------------------------------------------------------------------=20 -- Limit (cost=3D0.00..2.62 rows=3D1 width=3D4) (actual time=3D0.098..0.099= =20=20 rows=3D1 loops=3D1) -> Index Scan Backward using url_next_index_time on url=20=20=20 (cost=3D0.00..814591.03 rows=3D310913 width=3D4) (actual time=3D0.096..0.09= 6=20=20 rows=3D1 loops=3D1) Total runtime: 0.195 ms (3 rows) dps=3D# explain analyze select next_index_time from url order by=20=20 next_index_time asc limit 1; =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 QUERY PLAN ------------------------------------------------------------------------=20 ------------------------------------------------------------------------=20 - Limit (cost=3D0.00..2.62 rows=3D1 width=3D4) (actual=20=20 time=3D13504.105..13504.106 rows=3D1 loops=3D1) -> Index Scan using url_next_index_time on url=20=20=20 (cost=3D0.00..814591.03 rows=3D310913 width=3D4) (actual=20=20 time=3D13504.099..13504.099 rows=3D1 loops=3D1) Total runtime: 13504.158 ms (3 rows) Better, but...... Cordialement, Jean-G=E9rard Pailloncy From pgsql-performance-owner@postgresql.org Mon Apr 12 16:05:07 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 06D73D1E822 for ; Mon, 12 Apr 2004 16:05:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33395-07 for ; Mon, 12 Apr 2004 16:05:05 -0300 (ADT) Received: from mail.autorevenue.com (ip-208.178.167.106.gblx.net [208.178.167.106]) by svr1.postgresql.org (Postfix) with ESMTP id 5FB0BD1DCA5 for ; Mon, 12 Apr 2004 16:05:03 -0300 (ADT) Received: from jeremydunn ([192.168.1.79]) (authenticated) by mail.autorevenue.com (8.11.6/8.11.6) with ESMTP id i3CJ54Y29381 for ; Mon, 12 Apr 2004 15:05:04 -0400 Reply-To: From: "Jeremy Dunn" To: "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values Date: Mon, 12 Apr 2004 15:05:02 -0400 Organization: AutoRevenue Message-ID: <001d01c420c1$0ed23290$4f01a8c0@jeremydunn> 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 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 In-Reply-To: <12037.1081792288@sss.pgh.pa.us> Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/139 X-Sequence-Number: 6439 > "Jeremy Dunn" writes: > > The question: why does the planner consider a sequential scan to be > > better for these top 10 values? > > At some point a seqscan *will* be better. In the limit, if > the key being sought is common enough to occur on every page > of the table, it's certain that a seqscan will require less > I/O than an indexscan (because reading the index isn't > actually saving you any heap fetches). In practice the > breakeven point is less than that because Unix kernels are > better at handling sequential than random access. > > Your gripe appears to be basically that the planner's idea of > the breakeven point is off a bit. It looks to me like it's > within about a factor of 2 of being right, though, which is > not all that bad when it's using generic cost parameters. Agreed. However, given that count(*) is a question that can be answered _solely_ using the index (without reference to the actual data blocks), I'd expect that the break-even point would be considerably higher than the < 3% (~38,000 / ~1.3M) I'm currently getting. Does PG not use solely the index in this situation?? > > A) alter table xxx alter column cid set statistics 500; > > analyze xxx; > > This does not affect the results. > > It probably improved the accuracy of the row count estimates, > no? The estimate you show for cid=7191032 is off by more than > 25% (37765 vs 50792), which seems like a lot of error for one > of the most common values in the table. (I hope that was > with default stats target and not 500.) That leads directly > to a 25% overestimate of the cost of an indexscan, while > having IIRC no impact on the cost of a seqscan. Since the > cost ratio was more than 25%, this didn't change the selected > plan, but you want to fix that error as best you can before > you move on to tweaking cost parameters. Actually it made them worse! Yes, this was the default statistics (10). When I just tried it again with a value of 300, analyze, then run the query, I get a *worse* result for an estimate. I don't understand this. alter table xxx alter column cid set statistics 300; analyze emailrcpts; set random_page_cost to 2; explain analyze select count(*) from xxx where cid=7191032; Aggregate (cost=20563.28..20563.28 rows=1 width=0) (actual time=7653.90..7653.90 rows=1 loops=1) -> Index Scan using xxx_cid on xxx (cost=0.00..20535.82 rows=10983 width=0) (actual time=72.24..7602.38 rows=37765 loops=1) Total runtime: 7654.14 msec Now it estimates I have only 10,983 rows (~3x too low) instead of the old estimate 50,792 (1.3x too high). Why is that ?? Anyway, a workable solution seems to be using a lower value for Random_Page_Cost. Thanks to everyone who replied with this answer. > Also it is likely appropriate to increase > effective_cache_size, which is awfully small in the default > configuration. I'd set that to something related to your > available RAM before trying to home in on a suitable random_page_cost. We have ours set to the default value of 1000, which does seem low for a system with 1GB of RAM. We'll up this once we figure out what's available. Then tweak the Random_Page_Cost appropriately at that point. I'd still like to understand the strangeness above, if anyone can shed light. - Jeremy From pgsql-performance-owner@postgresql.org Mon Apr 12 16:52:14 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4702AD1B486 for ; Mon, 12 Apr 2004 16:52:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55203-01 for ; Mon, 12 Apr 2004 16:52:13 -0300 (ADT) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id D5836D1D1B8 for ; Mon, 12 Apr 2004 16:52:10 -0300 (ADT) Received: (qmail 24169 invoked by uid 500); 12 Apr 2004 19:55:52 -0000 Date: Mon, 12 Apr 2004 14:55:52 -0500 From: Bruno Wolff III To: Jeremy Dunn Cc: 'Postgresql Performance' Subject: Re: index v. seqscan for certain values Message-ID: <20040412195552.GA24133@wolff.to> Mail-Followup-To: Jeremy Dunn , 'Postgresql Performance' References: <12037.1081792288@sss.pgh.pa.us> <001d01c420c1$0ed23290$4f01a8c0@jeremydunn> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <001d01c420c1$0ed23290$4f01a8c0@jeremydunn> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/140 X-Sequence-Number: 6440 On Mon, Apr 12, 2004 at 15:05:02 -0400, Jeremy Dunn wrote: > > Agreed. However, given that count(*) is a question that can be answered > _solely_ using the index (without reference to the actual data blocks), > I'd expect that the break-even point would be considerably higher than > the < 3% (~38,000 / ~1.3M) I'm currently getting. Does PG not use > solely the index in this situation?? That isn't true. In order to check visibility you need to look at the data blocks. From pgsql-performance-owner@postgresql.org Mon Apr 12 17:55:10 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id AE3E7D1D084 for ; Mon, 12 Apr 2004 17:55:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75529-07 for ; Mon, 12 Apr 2004 17:55:09 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id A39D4D1D070 for ; Mon, 12 Apr 2004 17:55:06 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3CKslMK013681; Mon, 12 Apr 2004 16:54:47 -0400 (EDT) To: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Cc: pgsql-performance@postgresql.org Subject: Re: =?ISO-8859-1?Q?Re:__Index_Backward_Scan_fast_/_Index_Sc?= =?ISO-8859-1?Q?an_slow_!__=28Modifi=E9_par_Pailloncy_Jean-G=E9ra?= =?ISO-8859-1?Q?rd=29?= In-reply-to: <1D4DEBDC-8CB4-11D8-80FD-000A95DE2550@ifrance.com> References: <1D4DEBDC-8CB4-11D8-80FD-000A95DE2550@ifrance.com> Comments: In-reply-to =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= message dated "Mon, 12 Apr 2004 21:02:02 +0200" Date: Mon, 12 Apr 2004 16:54:47 -0400 Message-ID: <13680.1081803287@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/141 X-Sequence-Number: 6441 =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= writes: >> In 7.4 a VACUUM should be sufficient ... or at least, if it isn't > Atfer VACUUM: > Better, but...... ... but not much :-(. Okay, could we see VACUUM VERBOSE results for this table? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 12 18:02:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7FE58D1C9BE for ; Mon, 12 Apr 2004 18:02:48 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81172-01 for ; Mon, 12 Apr 2004 18:02:48 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 3C7DBD1BAF8 for ; Mon, 12 Apr 2004 18:02:46 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3CL2l77013763; Mon, 12 Apr 2004 17:02:47 -0400 (EDT) To: jdunn@autorevenue.com Cc: "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values In-reply-to: <001d01c420c1$0ed23290$4f01a8c0@jeremydunn> References: <001d01c420c1$0ed23290$4f01a8c0@jeremydunn> Comments: In-reply-to "Jeremy Dunn" message dated "Mon, 12 Apr 2004 15:05:02 -0400" Date: Mon, 12 Apr 2004 17:02:47 -0400 Message-ID: <13762.1081803767@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/142 X-Sequence-Number: 6442 "Jeremy Dunn" writes: > Agreed. However, given that count(*) is a question that can be answered > _solely_ using the index (without reference to the actual data blocks), As Bruno noted, that is not the case in Postgres; we must visit the table rows anyway. > When I just tried it again with a value of 300, analyze, then run the > query, I get a *worse* result for an estimate. I don't understand this. That's annoying. How repeatable are these results --- if you do ANALYZE over again several times, how much does the row count estimate change each time? (It should change somewhat, since ANALYZE is taking a random sample, but one would like to think not a whole lot.) Is the variance more or less at the higher stats target? Take a look at a few different CID values to get a sense of the accuracy, don't look at just one ... (Actually, you might find it more profitable to look at the pg_stats entry for the CID column rather than reverse-engineering the stats via ANALYZE. Look at how well the most-common-values list and associated frequency numbers track reality.) Also, can you think of any reason for the distribution of CID values to be nonuniform within the table? For instance, do rows get inserted in order of increasing CID, or is there any clustering of rows with the same CID? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 12 18:24:32 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6F049D1D26C for ; Mon, 12 Apr 2004 18:24:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89033-06 for ; Mon, 12 Apr 2004 18:24:32 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 9297CD1D1C4 for ; Mon, 12 Apr 2004 18:24:23 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3CLNkjT013961; Mon, 12 Apr 2004 17:23:47 -0400 (EDT) To: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Cc: pgsql-performance@postgresql.org Subject: Re: Index Backward Scan fast / Index Scan slow ! In-reply-to: References: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> <8579.1081772899@sss.pgh.pa.us> Comments: In-reply-to =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= message dated "Mon, 12 Apr 2004 21:02:02 +0200" Date: Mon, 12 Apr 2004 17:23:46 -0400 Message-ID: <13960.1081805026@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/143 X-Sequence-Number: 6443 [ Ah, I just got to your message with the VACUUM VERBOSE results ... ] =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= writes: > dps=# VACUUM VERBOSE url; > INFO: index "url_next_index_time" now contains 349519 row versions in > 3534 pages > DETAIL: 229515 index row versions were removed. > 3071 index pages have been deleted, 2864 are currently reusable. > CPU 0.32s/0.67u sec elapsed 76.25 sec. Hm, this is odd. That says you've got 349519 live index entries in only 463 actively-used index pages, or an average of 754 per page, which AFAICS could not fit in an 8K page. Are you using a nondefault value of BLCKSZ? If so what? If you *are* using default BLCKSZ then this index must be corrupt, and what you probably need to do is REINDEX it. But before you do that, could you send me a copy of the index file? regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 12 21:17:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 18542D1E137 for ; Mon, 12 Apr 2004 21:17:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 47709-01 for ; Mon, 12 Apr 2004 21:17:36 -0300 (ADT) Received: from mail4.speakeasy.net (mail4.speakeasy.net [216.254.0.204]) by svr1.postgresql.org (Postfix) with ESMTP id 9866ED1E129 for ; Mon, 12 Apr 2004 21:17:32 -0300 (ADT) Received: (qmail 10051 invoked by uid 64014); 13 Apr 2004 00:17:36 -0000 Received: from kgeis@speakeasy.org by mail4.speakeasy.net with AmikaGuardian-Server-2.1.2; 13 Apr 2004 00:17:36 -0000 X-AmikaGuardian-Id: mail4.speakeasy.net108181545526110043 X-AmikaGuardian-Action: Do Nothing() Received: from adsl-67-113-18-226.dsl.snfc21.pacbell.net (HELO speakeasy.org) (kgeis@[67.113.18.226]) (envelope-sender ) by mail4.speakeasy.net (qmail-ldap-1.03) with AES256-SHA encrypted SMTP for ; 13 Apr 2004 00:17:35 -0000 Message-ID: <407B319B.7000400@speakeasy.org> Date: Mon, 12 Apr 2004 17:17:31 -0700 From: Ken Geis User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7b) Gecko/20040316 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Tracking down performance issue 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: 200404/144 X-Sequence-Number: 6444 I'm running 7.4.2 on an older Linux box (450MHzAMD K-6-III, 450M RAM) running kernel 2.6.5. My client is a Java/JDBC program on Windows. I'm having trouble seeing where the bottleneck in my performance is. The client uses about 30% CPU. The server uses 70% CPU plus 1.5% I/O wait. The I/O wait is very low because I'm doing a PK index scan where the index and data are on different disks and the table is clustered on the PK index. The network is 100Mb, and it's at 7% of capacity. I tried making the client operate on two threads on two database connections. That bumped the server utilization to 80% and barely changed the I/O wait. The throughput decreased by a third. The only thing I can think of is memory bandwidth. Does anyone have tips on how I can investigate more? From pgsql-performance-owner@postgresql.org Tue Apr 13 08:12:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 24943D1E180 for ; Tue, 13 Apr 2004 08:12:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52470-04 for ; Tue, 13 Apr 2004 08:12:40 -0300 (ADT) Received: from mwinf0204.wanadoo.fr (smtp2.wanadoo.fr [193.252.22.29]) by svr1.postgresql.org (Postfix) with ESMTP id E8EDFD1E084 for ; Tue, 13 Apr 2004 08:12:36 -0300 (ADT) Received: from [192.168.0.2] (AMontsouris-108-1-19-199.w80-15.abo.wanadoo.fr [80.15.148.199]) by mwinf0204.wanadoo.fr (SMTP Server) with ESMTP id 29481A0000B3; Tue, 13 Apr 2004 13:12:39 +0200 (CEST) In-Reply-To: <13960.1081805026@sss.pgh.pa.us> References: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> <8579.1081772899@sss.pgh.pa.us> <13960.1081805026@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v613) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <77EC9C09-8D3B-11D8-A5A0-000A95DE2550@ifrance.com> Content-Transfer-Encoding: quoted-printable Cc: pgsql-performance@postgresql.org From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Subject: Re: Index Backward Scan fast / Index Scan slow ! Date: Tue, 13 Apr 2004 13:12:36 +0200 To: Tom Lane X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/145 X-Sequence-Number: 6445 > Hm, this is odd. That says you've got 349519 live index entries in=20 > only > 463 actively-used index pages, or an average of 754 per page, which > AFAICS could not fit in an 8K page. Are you using a nondefault value=20 > of > BLCKSZ? If so what? Sorry, I forgot to specify I use BLCKSZ of 32768, the same blokck's=20 size for newfs, the same for RAID slice's size. I test the drive sometimes ago, and found a speed win if the slice size=20 the disk block size and the read block size was the same. I do not think that a different BLCKSZ should exhibit a slowdown as the=20 one I found. > If you *are* using default BLCKSZ then this index must be corrupt, and > what you probably need to do is REINDEX it. But before you do that, > could you send me a copy of the index file? Do you want the index file now, or may I try something before? Cordialement, Jean-G=E9rard Pailloncy From pgsql-performance-owner@postgresql.org Tue Apr 13 10:47:14 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 473C7D1E944 for ; Tue, 13 Apr 2004 10:47:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04431-05 for ; Tue, 13 Apr 2004 10:47:15 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 10398D1E129 for ; Tue, 13 Apr 2004 10:47:09 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3DDlCUq022097; Tue, 13 Apr 2004 09:47:12 -0400 (EDT) To: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Cc: pgsql-performance@postgresql.org Subject: Re: Index Backward Scan fast / Index Scan slow ! In-reply-to: <77EC9C09-8D3B-11D8-A5A0-000A95DE2550@ifrance.com> References: <8A2105D4-8C63-11D8-80FD-000A95DE2550@ifrance.com> <8579.1081772899@sss.pgh.pa.us> <13960.1081805026@sss.pgh.pa.us> <77EC9C09-8D3B-11D8-A5A0-000A95DE2550@ifrance.com> Comments: In-reply-to =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= message dated "Tue, 13 Apr 2004 13:12:36 +0200" Date: Tue, 13 Apr 2004 09:47:11 -0400 Message-ID: <22096.1081864031@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/146 X-Sequence-Number: 6446 =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= writes: >> Are you using a nondefault value of >> BLCKSZ? If so what? > Sorry, I forgot to specify I use BLCKSZ of 32768, Okay, the numbers are sensible then. The index density seems a bit low (754 entries/page where the theoretical ideal would be about 1365) but not really out-of-line. >> could you send me a copy of the index file? > Do you want the index file now, or may I try something before? If you're going to reindex, please do send me a copy of the file first. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 13 11:26:57 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ADDF5D1C9C8 for ; Tue, 13 Apr 2004 11:26:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28489-02 for ; Tue, 13 Apr 2004 11:26:56 -0300 (ADT) Received: from trade-india.com (ns5.trade-india.com [66.234.10.13]) by svr1.postgresql.org (Postfix) with SMTP id 7799BD1C9C4 for ; Tue, 13 Apr 2004 11:26:52 -0300 (ADT) Received: (qmail 14006 invoked from network); 13 Apr 2004 10:57:54 -0000 Received: from unknown (HELO trade-india.com) (203.145.130.142) by ns5.trade-india.com with SMTP; 13 Apr 2004 10:57:54 -0000 Message-ID: <407BF8B5.3010101@trade-india.com> Date: Tue, 13 Apr 2004 19:57:01 +0530 From: Rajesh Kumar Mallah User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Shea,Dan [CIS]" Cc: Postgres Performance Subject: Re: Deleting certain duplicates 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: 200404/147 X-Sequence-Number: 6447 Shea,Dan [CIS] wrote: >The index is >Indexes: > "forecastelement_rwv_idx" btree (region_id, wx_element, valid_time) > >-----Original Message----- >From: Shea,Dan [CIS] [mailto:Dan.Shea@ec.gc.ca] >Sent: Monday, April 12, 2004 10:39 AM >To: Postgres Performance >Subject: [PERFORM] Deleting certain duplicates > > >We have a large database which recently increased dramatically due to a >change in our insert program allowing all entries. >PWFPM_DEV=# select relname,relfilenode,reltuples from pg_class where relname >= 'forecastelement'; > relname | relfilenode | reltuples >-----------------+-------------+------------- > forecastelement | 361747866 | 4.70567e+08 > > Column | Type | Modifiers >----------------+-----------------------------+----------- > version | character varying(99) | > origin | character varying(10) | > timezone | character varying(99) | > region_id | character varying(20) | > wx_element | character varying(99) | > value | character varying(99) | > flag | character(3) | > units | character varying(99) | > valid_time | timestamp without time zone | > issue_time | timestamp without time zone | > next_forecast | timestamp without time zone | > reception_time | timestamp without time zone | > >The program is supposed to check to ensure that all fields but the >reception_time are unique using a select statement, and if so, insert it. >Due an error in a change, reception time was included in the select to check >for duplicates. The reception_time is created by a program creating the dat >file to insert. >Essentially letting all duplicate files to be inserted. > >I tried the delete query below. >PWFPM_DEV=# delete from forecastelement where oid not in (select min(oid) >from forecastelement group by >version,origin,timezone,region_id,wx_element,value,flag,units,valid_time,iss >ue_time,next_forecast); >It ran for 3 days creating what I assume is an index in pgsql_tmp of the >group by statement. >The query ended up failing with "dateERROR:write failed". >Well the long weekend is over and we do not have the luxury of trying this >again. >So I was thinking maybe of doing the deletion in chunks, perhaps based on >reception time. > > its more of an sql question though. to deduplicate on basis of version,origin,timezone,region_id,wx_element,value,flag,units,valid_time, issue_time,next_forecast You could do this. begin work; create temp_table as select distinct on (version,origin,timezone,region_id,wx_element,value,flag,units,valid_time, issue_time,next_forecast) * from forecastelement ; truncate table forecastelement ; drop index ; insert into forecastelement select * from temp_table ; commit; create indexes Analyze forecastelement ; note that distinct on will keep only one row out of all rows having distinct values of the specified columns. kindly go thru the distinct on manual before trying the queries. regds mallah. >Are there any suggestions for a better way to do this, or using multiple >queries to delete selectively a week at a time based on the reception_time. >I would say there are a lot of duplicate entries between mid march to the >first week of April. > > > >---------------------------(end of broadcast)--------------------------- >TIP 8: explain analyze is your friend > >---------------------------(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 Tue Apr 13 11:41:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C1FB2D1E944 for ; Tue, 13 Apr 2004 11:41:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 32517-05 for ; Tue, 13 Apr 2004 11:41:34 -0300 (ADT) Received: from mail.autorevenue.com (ip-208.178.167.106.gblx.net [208.178.167.106]) by svr1.postgresql.org (Postfix) with ESMTP id 91238D1E969 for ; Tue, 13 Apr 2004 11:41:28 -0300 (ADT) Received: from jeremydunn ([192.168.1.79]) (authenticated) by mail.autorevenue.com (8.11.6/8.11.6) with ESMTP id i3DEfWY18193 for ; Tue, 13 Apr 2004 10:41:32 -0400 Reply-To: From: "Jeremy Dunn" To: "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values Date: Tue, 13 Apr 2004 10:41:30 -0400 Organization: AutoRevenue Message-ID: <000801c42165$68a67750$4f01a8c0@jeremydunn> 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: <13762.1081803767@sss.pgh.pa.us> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/148 X-Sequence-Number: 6448 > > When I just tried it again with a value of 300, analyze, > then run the query, I get a *worse* result for an estimate. I don't understand > > this. > > That's annoying. How repeatable are these results --- if you > do ANALYZE over again several times, how much does the row > count estimate change each time? (It should change somewhat, > since ANALYZE is taking a random sample, but one would like > to think not a whole lot.) Is the variance more or less at > the higher stats target? Take a look at a few different CID > values to get a sense of the accuracy, don't look at just one ... Yes, it's repeatable. I tried a bunch of times, and there are only small variations in the stats for the higher stat targets. > (Actually, you might find it more profitable to look at the > pg_stats entry for the CID column rather than > reverse-engineering the stats via ANALYZE. Look at how well > the most-common-values list and associated frequency numbers > track reality.) I checked the accuracy of the stats for various values, and there is a wide variation. I see some values where the estimate is 1.75x the actual; and others where the estimate is .44x the actual. > Also, can you think of any reason for the distribution of CID > values to be nonuniform within the table? For instance, do > rows get inserted in order of increasing CID, or is there any > clustering of rows with the same CID? This is almost certainly the answer. The data is initially inserted in chunks for each CID, and later on there is a more normal distribution of insert/update/deletes across all CIDs; and then again a new CID will come with a large chunk of rows, etc. Interestingly, I tried increasing the stat size for the CID column to 2000, analyzing, and checking the accuracy of the stats again. Even with this relatively high value, the accuracy of the stats is not that close. The value giving .44x previously nows gives an estimate .77x of actual. Another value which was at 1.38x of actual is now at .71x of actual! Then just for kicks I set the statistics size to 100,000 (!), analyzed, and ran the query again. For the same CID I still got an estimated row count that is .71x the actual rows returned. Why is this not better? I wonder how high I'd have to set the statistics collector to get really good data, given the uneven data distribution of this table. Is there any other technique that works better to get good estimates, given uneven distribution of values? So I think this explains the inaccurate stats; and the solution as far as I'm concerned is to increase the two params mentioned yesterday (effective_cache_size & random_page_cost). Thanks again for the help! - Jeremy From pgsql-performance-owner@postgresql.org Tue Apr 13 14:55:53 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 67287D1DD2C for ; Tue, 13 Apr 2004 14:55:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 18160-06 for ; Tue, 13 Apr 2004 14:55:51 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 2AAE6D1DC7D for ; Tue, 13 Apr 2004 14:55:50 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3DHtnBn024671; Tue, 13 Apr 2004 13:55:50 -0400 (EDT) To: jdunn@autorevenue.com Cc: "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values In-reply-to: <000801c42165$68a67750$4f01a8c0@jeremydunn> References: <000801c42165$68a67750$4f01a8c0@jeremydunn> Comments: In-reply-to "Jeremy Dunn" message dated "Tue, 13 Apr 2004 10:41:30 -0400" Date: Tue, 13 Apr 2004 13:55:49 -0400 Message-ID: <24670.1081878949@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/149 X-Sequence-Number: 6449 "Jeremy Dunn" writes: > Interestingly, I tried increasing the stat size for the CID column to > 2000, analyzing, and checking the accuracy of the stats again. There's a hard limit of 1000, I believe. Didn't it give you a warning saying so? At 1000 the ANALYZE sample size would be 300000 rows, or about a quarter of your table. I would have thought this would give frequency estimates with much better precision than you seem to be seeing --- but my statistics are rusty enough that I'm not sure about it. Possibly the nonuniform clumping of CID has something to do with the poor results. Any stats majors on the list? regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 13 15:02:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6C8EDD1E98D for ; Tue, 13 Apr 2004 15:02:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 20498-08 for ; Tue, 13 Apr 2004 15:02:55 -0300 (ADT) Received: from bramble.mmrd.com (unknown [65.217.53.66]) by svr1.postgresql.org (Postfix) with ESMTP id 71FE0D1E98C for ; Tue, 13 Apr 2004 15:02:53 -0300 (ADT) 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 i3DI5GcM010301 for ; Tue, 13 Apr 2004 14:05:25 -0400 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 i3DI2el06362 for ; Tue, 13 Apr 2004 14:02:41 -0400 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 FVMA9AJ9; Tue, 13 Apr 2004 14:02:38 -0400 Subject: query slows down with more accurate stats From: Robert Treat To: pgsql-performance@postgresql.org Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 Date: 13 Apr 2004 14:02:39 -0400 Message-Id: <1081879359.25537.526.camel@camel> Mime-Version: 1.0 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/150 X-Sequence-Number: 6450 In the process of optimizing some queries, I have found the following query seems to degrade in performance the more accurate I make the statistics on the table... whether by using increased alter table ... set statistics or by using vacuum.. SELECT count( cl.caller_id ), npanxx.city, npanxx.state FROM cl LEFT OUTER JOIN npanxx on substr( cl.caller_id, 1, 3 ) = npanxx.npa and substr( cl.caller_id, 4, 3 ) = npanxx.nxx LEFT OUTER JOIN cp ON cl.caller_id = cp.caller_id WHERE cl.ivr_system_id = 130 AND cl.call_time > '2004-03-01 06:00:00.0 CST' AND cl.call_time < '2004-04-01 06:00:00.0 CST' AND cp.age >= 18 AND cp.age <= 24 AND cp.gender = 'm' GROUP BY npanxx.city, npanxx.state live=# analyze cl; ANALYZE live=# select reltuples from pg_class where relname = 'cl'; reltuples ----------- 53580 (1 row) live=# select count(*) from cl; count --------- 1140166 (1 row) The plan i get under these conditions is actually pretty good... HashAggregate (cost=28367.22..28367.66 rows=174 width=32) (actual time=1722.060..1722.176 rows=29 loops=1) -> Nested Loop (cost=0.00..28365.92 rows=174 width=32) (actual time=518.592..1716.254 rows=558 loops=1) -> Nested Loop Left Join (cost=0.00..20837.33 rows=1248 width=32) (actual time=509.991..1286.755 rows=13739 loops=1) -> Index Scan using cl_ivr_system_id on cl (cost=0.00..13301.15 rows=1248 width=14) (actual time=509.644..767.421 rows=13739 loops=1) Index Cond: (ivr_system_id = 130) Filter: ((call_time > '2004-03-01 07:00:00-05'::timestamp with time zone) AND (call_time < '2004-04-01 07:00:00-05'::timestamp with time zone)) -> Index Scan using npanxx_pkey on npanxx (cost=0.00..6.02 rows=1 width=32) (actual time=0.025..0.027 rows=1 loops=13739) Index Cond: ((substr(("outer".caller_id)::text, 1, 3) = (npanxx.npa)::text) AND (substr(("outer".caller_id)::text, 4, 3) = (npanxx.nxx)::text)) -> Index Scan using cp_pkey on cp (cost=0.00..6.02 rows=1 width=14) (actual time=0.027..0.027 rows=0 loops=13739) Index Cond: (("outer".caller_id)::text = (cp.caller_id)::text) Filter: ((age >= 18) AND (age <= 24) AND (gender = 'm'::bpchar)) Total runtime: 1722.489 ms (12 rows) but when i do live=# vacuum cl; VACUUM live=# select reltuples from pg_class where relname = 'cl'; reltuples ------------- 1.14017e+06 (1 row) (or alternatively increase the stats target on the table) I now get the following plan: HashAggregate (cost=80478.74..80482.41 rows=1471 width=32) (actual time=8132.261..8132.422 rows=29 loops=1) -> Merge Join (cost=79951.95..80467.70 rows=1471 width=32) (actual time=7794.338..8130.041 rows=558 loops=1) Merge Cond: ("outer"."?column4?" = "inner"."?column2?") -> Sort (cost=55719.06..55745.42 rows=10546 width=32) (actual time=4031.827..4052.526 rows=13739 loops=1) Sort Key: (cl.caller_id)::text -> Merge Right Join (cost=45458.30..55014.35 rows=10546 width=32) (actual time=2944.441..3796.787 rows=13739 loops=1) Merge Cond: ((("outer".npa)::text = "inner"."?column2?") AND (("outer".nxx)::text = "inner"."?column3?")) -> Index Scan using npanxx_pkey on npanxx (cost=0.00..8032.99 rows=132866 width=32) (actual time=0.200..461.767 rows=130262 loops=1) -> Sort (cost=45458.30..45484.67 rows=10546 width=14) (actual time=2942.994..2967.935 rows=13739 loops=1) Sort Key: substr((cl.caller_id)::text, 1, 3), substr((cl.caller_id)::text, 4, 3) -> Seq Scan on cl (cost=0.00..44753.60 rows=10546 width=14) (actual time=1162.423..2619.662 rows=13739 loops=1) Filter: ((ivr_system_id = 130) AND (call_time > '2004-03-01 07:00:00-05'::timestamp with time zone) AND (call_time < '2004-04-01 07:00:00-05'::timestamp with time zone)) -> Sort (cost=24232.89..24457.06 rows=89667 width=14) (actual time=3761.703..3900.340 rows=98010 loops=1) Sort Key: (cp.caller_id)::text -> Seq Scan on cp (cost=0.00..15979.91 rows=89667 width=14) (actual time=0.128..1772.215 rows=100302 loops=1) Filter: ((age >= 18) AND (age <= 24) AND (gender = 'm'::bpchar)) Total runtime: 8138.607 ms (17 rows) so i guess i am wondering if there is something I should be doing to help get the better plan at the more accurate stats levels and/or why it doesn't stick with the original plan (I noticed disabling merge joins does seem to push it back to the original plan). alternatively if anyone has any general suggestions on speeding up the query I'd be open to that too :-) Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Tue Apr 13 15:04:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9948ED1E180 for ; Tue, 13 Apr 2004 15:04:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19433-08 for ; Tue, 13 Apr 2004 15:04:25 -0300 (ADT) Received: from mail.autorevenue.com (ip-208.178.167.106.gblx.net [208.178.167.106]) by svr1.postgresql.org (Postfix) with ESMTP id C7892D1CCA7 for ; Tue, 13 Apr 2004 15:04:21 -0300 (ADT) Received: from jeremydunn ([192.168.1.79]) (authenticated) by mail.autorevenue.com (8.11.6/8.11.6) with ESMTP id i3DI4LY31234 for ; Tue, 13 Apr 2004 14:04:21 -0400 Reply-To: From: "Jeremy Dunn" To: "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values Date: Tue, 13 Apr 2004 14:04:19 -0400 Organization: AutoRevenue Message-ID: <000f01c42181$be1b6260$4f01a8c0@jeremydunn> 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: <24670.1081878949@sss.pgh.pa.us> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/151 X-Sequence-Number: 6451 > There's a hard limit of 1000, I believe. Didn't it give you > a warning saying so? No warning at 2000, and no warning at 100,000 either! Remember we are still on 7.2.x. The docs here http://www.postgresql.org/docs/7.2/static/sql-altertable.html don't say anything about a limit. This is good to know, if it's true. Can anyone confirm? - Jeremy From pgsql-performance-owner@postgresql.org Tue Apr 13 15:35:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F38E2D1D08B for ; Tue, 13 Apr 2004 15:35:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 35890-02 for ; Tue, 13 Apr 2004 15:35:26 -0300 (ADT) Received: from bramble.mmrd.com (unknown [65.217.53.66]) by svr1.postgresql.org (Postfix) with ESMTP id 93FC0D1D08A for ; Tue, 13 Apr 2004 15:35:24 -0300 (ADT) 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 i3DIbrcM010790; Tue, 13 Apr 2004 14:37:53 -0400 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 i3DIZHl07552; Tue, 13 Apr 2004 14:35:17 -0400 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 FVMA9A4C; Tue, 13 Apr 2004 14:35:15 -0400 Subject: Re: index v. seqscan for certain values From: Robert Treat To: jdunn@autorevenue.com Cc: "'Postgresql Performance'" In-Reply-To: <000f01c42181$be1b6260$4f01a8c0@jeremydunn> References: <000f01c42181$be1b6260$4f01a8c0@jeremydunn> Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 Date: 13 Apr 2004 14:35:16 -0400 Message-Id: <1081881316.25537.562.camel@camel> Mime-Version: 1.0 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/152 X-Sequence-Number: 6452 On Tue, 2004-04-13 at 14:04, Jeremy Dunn wrote: > > > There's a hard limit of 1000, I believe. Didn't it give you > > a warning saying so? > > No warning at 2000, and no warning at 100,000 either! > > Remember we are still on 7.2.x. The docs here > http://www.postgresql.org/docs/7.2/static/sql-altertable.html don't say > anything about a limit. > > This is good to know, if it's true. Can anyone confirm? > transform=# alter table data_pull alter column msg set statistics 100000; WARNING: lowering statistics target to 1000 ERROR: column "msg" of relation "data_pull" does not exist transform=# select version(); version ---------------------------------------------------------------- PostgreSQL 7.4beta4 on i686-pc-linux-gnu, compiled by GCC 2.96 (1 row) Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Tue Apr 13 16:10:53 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 95B34D1DA83 for ; Tue, 13 Apr 2004 15:50:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 41839-07 for ; Tue, 13 Apr 2004 15:50:05 -0300 (ADT) Received: from leo.supplyfx.com (h-67-100-199-42.lsanca54.covad.net [67.100.199.42]) by svr1.postgresql.org (Postfix) with ESMTP id 6A376D1E995 for ; Tue, 13 Apr 2004 15:49:59 -0300 (ADT) Received: from [10.0.100.21] (leo6.supplyfx.com [10.0.100.21]) by leo.supplyfx.com (Postfix) with ESMTP id CC92E105567; Tue, 13 Apr 2004 11:49:59 -0700 (PDT) In-Reply-To: <24670.1081878949@sss.pgh.pa.us> References: <000801c42165$68a67750$4f01a8c0@jeremydunn> <24670.1081878949@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v612) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> Content-Transfer-Encoding: 7bit Cc: Qing Zhao From: Qing Zhao Subject: configure shmmax on MAC OS X Date: Tue, 13 Apr 2004 11:49:43 -0700 To: 'Postgresql Performance' X-Mailer: Apple Mail (2.612) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/153 X-Sequence-Number: 6453 Hi, all, I have got a new MaC OS G5 with 8GB RAM. So i tried to increase the shmmax in Kernel so that I can take advantage of the RAM. I searched the web and read the manual for PG7.4 chapter 16.5.1. After that, I edited /etc/rc file: sysctl -w kern.sysv.shmmax=4294967296 // byte sysctl -w kern.sysv.shmmin=1 sysctl -w kern.sysv.shmmni=32 sysctl -w kern.sysv.shmseg=8 sysctl -w kern.sysv.shmall=1048576 //4kpage for 4G shared RAM. Then I changed postgresql.conf: shared_buffer=100000 //could be bigger? and restart the machine and postgres server. To my surprise, postgres server wouldn't start, saying that the requested shared memory exceeds kernel's shmmax. My suspision is that the change i made in /etc/rc does not take effect.Is there a way to check it? Is there an up limit for how much RAM can be allocated for shared buffer in MAC OS X? Or is there something wrong with my calculation in numbers? Thanks a lot! Qing From pgsql-performance-owner@postgresql.org Tue Apr 13 16:18:45 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 08788D1E97A for ; Tue, 13 Apr 2004 16:18:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55132-07 for ; Tue, 13 Apr 2004 16:18:43 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 76A80D1E935 for ; Tue, 13 Apr 2004 16:18:41 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3DJIglK025558; Tue, 13 Apr 2004 15:18:42 -0400 (EDT) To: Robert Treat Cc: pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats In-reply-to: <1081879359.25537.526.camel@camel> References: <1081879359.25537.526.camel@camel> Comments: In-reply-to Robert Treat message dated "13 Apr 2004 14:02:39 -0400" Date: Tue, 13 Apr 2004 15:18:42 -0400 Message-ID: <25556.1081883922@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/154 X-Sequence-Number: 6454 Robert Treat writes: > live=# analyze cl; > ANALYZE > live=# select reltuples from pg_class where relname = 'cl'; > reltuples > ----------- > 53580 > (1 row) > live=# vacuum cl; > VACUUM > live=# select reltuples from pg_class where relname = 'cl'; > reltuples > ------------- > 1.14017e+06 > (1 row) Well, the first problem is why is ANALYZE's estimate of the total row count so bad :-( ? I suspect you are running into the situation where the initial pages of the table are thinly populated and ANALYZE mistakenly assumes the rest are too. Manfred is working on a revised sampling method for ANALYZE that should fix this problem in 7.5 and beyond, but for now it seems like a VACUUM FULL might be in order. > so i guess i am wondering if there is something I should be doing to > help get the better plan at the more accurate stats levels and/or why it > doesn't stick with the original plan (I noticed disabling merge joins > does seem to push it back to the original plan). With the larger number of estimated rows it's figuring the nestloop will be too expensive. The row estimate for the cl scan went up from 1248 to 10546, so the estimated cost for the nestloop plan would go to about 240000 units vs 80000 for the mergejoin plan. This is obviously off rather badly when the true runtimes are 1.7 vs 8.1 seconds :-(. I think this is an example of a case where we really need better estimation of nestloop costs --- it's drastically overestimating the relative cost of the nestloop because it's not accounting for the cache benefits of the repeated index searches. You could probably force the nestloop to be chosen by lowering random_page_cost, but that's just a kluge solution ... the real problem is the model is wrong. I have a to-do item to work on this, and will try to bump up its priority a bit. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 13 16:25:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D9D70D1E986 for ; Tue, 13 Apr 2004 16:25:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 57187-10 for ; Tue, 13 Apr 2004 16:25:30 -0300 (ADT) Received: from visionlink.org (mail.visionlink.org [206.168.220.27]) by svr1.postgresql.org (Postfix) with SMTP id 84206D1E1F3 for ; Tue, 13 Apr 2004 16:25:27 -0300 (ADT) Received: from [65.100.170.137] (HELO [192.168.1.200]) by visionlink.org (Stalker SMTP Server 1.8b8) with ESMTP id S.0002538979; Tue, 13 Apr 2004 13:25:26 -0600 Mime-Version: 1.0 X-Sender: bohmer@mail.visionlink.org Message-Id: In-Reply-To: <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> References: <000801c42165$68a67750$4f01a8c0@jeremydunn> <24670.1081878949@sss.pgh.pa.us> <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> Date: Tue, 13 Apr 2004 13:25:25 -0600 To: Qing Zhao From: Jeff Bohmer Subject: Re: configure shmmax on MAC OS X Cc: pgsql-performance@postgresql.org Content-Type: text/plain; charset="us-ascii" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/155 X-Sequence-Number: 6455 On OS X, I've always made these changes in: /System/Library/StartupItems/SystemTuning/SystemTuning and manually checked it with sysctl after reboot. Works for me. 100k buffers is probably overkill. There can be a performance penalty with too many buffers. See this lists' archives for more. 10k would probably be a better start. - Jeff >Hi, all, > >I have got a new MaC OS G5 with 8GB RAM. So i tried to increase >the shmmax in Kernel so that I can take advantage of the RAM. > >I searched the web and read the manual for PG7.4 chapter 16.5.1. >After that, I edited /etc/rc file: > >sysctl -w kern.sysv.shmmax=4294967296 // byte >sysctl -w kern.sysv.shmmin=1 >sysctl -w kern.sysv.shmmni=32 >sysctl -w kern.sysv.shmseg=8 >sysctl -w kern.sysv.shmall=1048576 //4kpage > >for 4G shared RAM. > >Then I changed postgresql.conf: >shared_buffer=100000 //could be bigger? > >and restart the machine and postgres server. To my surprise, postgres server wouldn't >start, saying that the requested shared memory exceeds kernel's shmmax. > >My suspision is that the change i made in /etc/rc does not take effect.Is there a way >to check it? Is there an >up limit for how much RAM can be allocated for shared buffer in MAC OS X? Or >is there something wrong with my calculation in numbers? > >Thanks a lot! > >Qing > > >---------------------------(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 -- Jeff Bohmer VisionLink, Inc. _________________________________ 303.402.0170 www.visionlink.org _________________________________ People. Tools. Change. Community. From pgsql-performance-owner@postgresql.org Tue Apr 13 16:55:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 01E4BD1E935 for ; Tue, 13 Apr 2004 16:55:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73329-04 for ; Tue, 13 Apr 2004 16:55:16 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 2D9C5D1C9BE for ; Tue, 13 Apr 2004 16:55:14 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3DJtFjF025966; Tue, 13 Apr 2004 15:55:15 -0400 (EDT) To: Qing Zhao Cc: "'Postgresql Performance'" Subject: Re: configure shmmax on MAC OS X In-reply-to: <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> References: <000801c42165$68a67750$4f01a8c0@jeremydunn> <24670.1081878949@sss.pgh.pa.us> <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> Comments: In-reply-to Qing Zhao message dated "Tue, 13 Apr 2004 11:49:43 -0700" Date: Tue, 13 Apr 2004 15:55:15 -0400 Message-ID: <25965.1081886115@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/156 X-Sequence-Number: 6456 Qing Zhao writes: > My suspision is that the change i made in /etc/rc does not take > effect.Is there a way to check it? sysctl has an option to show the values currently in effect. I believe that /etc/rc is the correct place to set shmmax on OSX 10.3 or later ... but we have seen prior reports of people having trouble getting the setting to "take". There may be some other constraint involved. > sysctl -w kern.sysv.shmmax=4294967296 // byte Hmm, does sysctl work for values that exceed the range of int? There's no particularly good reason to try to set shmmax as high as you are trying anyhow; you really don't need more than a couple hundred meg in Postgres shared memory. It's better to leave the kernel to manage the bulk of your RAM. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 13 17:10:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 850B2D1D299 for ; Tue, 13 Apr 2004 17:10:48 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 79362-04 for ; Tue, 13 Apr 2004 17:10:43 -0300 (ADT) Received: from leo.supplyfx.com (h-67-100-199-42.lsanca54.covad.net [67.100.199.42]) by svr1.postgresql.org (Postfix) with ESMTP id 32C28D1C9FB for ; Tue, 13 Apr 2004 17:10:41 -0300 (ADT) Received: from [10.0.100.21] (leo6.supplyfx.com [10.0.100.21]) by leo.supplyfx.com (Postfix) with ESMTP id 7E1A5105808; Tue, 13 Apr 2004 13:10:42 -0700 (PDT) In-Reply-To: <25965.1081886115@sss.pgh.pa.us> References: <000801c42165$68a67750$4f01a8c0@jeremydunn> <24670.1081878949@sss.pgh.pa.us> <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> <25965.1081886115@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v612) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <993AFE3A-8D86-11D8-92CF-000A95AB8896@quotefx.net> Content-Transfer-Encoding: 7bit Cc: "'Postgresql Performance'" From: Qing Zhao Subject: Re: configure shmmax on MAC OS X Date: Tue, 13 Apr 2004 13:10:24 -0700 To: Tom Lane X-Mailer: Apple Mail (2.612) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/157 X-Sequence-Number: 6457 Tom: I used sysctl -A to see the kernel state, I got: kern.sysv.shmmax: -1 It looks the value is too big! Thanks! Qing On Apr 13, 2004, at 12:55 PM, Tom Lane wrote: > Qing Zhao writes: >> My suspision is that the change i made in /etc/rc does not take >> effect.Is there a way to check it? > > sysctl has an option to show the values currently in effect. > > I believe that /etc/rc is the correct place to set shmmax on OSX 10.3 > or > later ... but we have seen prior reports of people having trouble > getting the setting to "take". There may be some other constraint > involved. > >> sysctl -w kern.sysv.shmmax=4294967296 // byte > > Hmm, does sysctl work for values that exceed the range of int? > > There's no particularly good reason to try to set shmmax as high as you > are trying anyhow; you really don't need more than a couple hundred meg > in Postgres shared memory. It's better to leave the kernel to manage > the bulk of your RAM. > > regards, tom lane > From pgsql-performance-owner@postgresql.org Tue Apr 13 17:49:17 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D02F7D1DD17 for ; Tue, 13 Apr 2004 17:49:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 95521-03 for ; Tue, 13 Apr 2004 17:49:13 -0300 (ADT) Received: from mail6.speakeasy.net (mail6.speakeasy.net [216.254.0.206]) by svr1.postgresql.org (Postfix) with ESMTP id AFD06D1DA83 for ; Tue, 13 Apr 2004 17:49:10 -0300 (ADT) Received: (qmail 24287 invoked from network); 13 Apr 2004 20:49:12 -0000 Received: from mail.kinesis-cem.com (HELO pdarley) ([64.81.9.230]) (envelope-sender ) by mail6.speakeasy.net (qmail-ldap-1.03) with SMTP for ; 13 Apr 2004 20:49:12 -0000 From: "Peter Darley" To: "Pgsql-Performance" Subject: View columns calculated Date: Tue, 13 Apr 2004 13:49:53 -0700 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) 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=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/158 X-Sequence-Number: 6458 Folks, I have a question about views: I want to have a fairly wide view (lots of columns) where most of the columns have some heavyish calculations in them, but I'm concerned that it will have to calculate every column even when I'm not selecting them. So, the question is, if I have 5 columns in a view but only select 1 column, is the system smart enough to not calculate the unused columns, or am I taking a performance hit over a smaller view that doesn't have the extra 4 columns? Thanks, Peter Darley From pgsql-performance-owner@postgresql.org Tue Apr 13 17:59:31 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5C5B0D1E96B for ; Tue, 13 Apr 2004 17:59:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 99637-08 for ; Tue, 13 Apr 2004 17:59:27 -0300 (ADT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 62A72D1E98F for ; Tue, 13 Apr 2004 17:59:24 -0300 (ADT) Received: from commandprompt.com (dsl093-038-087.pdx1.dsl.speakeasy.net [66.93.38.87]) (authenticated) by hosting.commandprompt.com (8.11.6/8.11.6) with ESMTP id i3DKxIP32349; Tue, 13 Apr 2004 13:59:18 -0700 Message-ID: <407C54A1.6090908@commandprompt.com> Date: Tue, 13 Apr 2004 13:59:13 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: Qing Zhao , "'Postgresql Performance'" Subject: Re: configure shmmax on MAC OS X References: <000801c42165$68a67750$4f01a8c0@jeremydunn> <24670.1081878949@sss.pgh.pa.us> <53A6FF52-8D7B-11D8-92CF-000A95AB8896@quotefx.net> <25965.1081886115@sss.pgh.pa.us> In-Reply-To: <25965.1081886115@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: 200404/159 X-Sequence-Number: 6459 Hello, I found that if you SHMALL value was less than your SHMMAX value, the value wouldn't take. J Tom Lane wrote: > Qing Zhao writes: > >>My suspision is that the change i made in /etc/rc does not take >>effect.Is there a way to check it? > > > sysctl has an option to show the values currently in effect. > > I believe that /etc/rc is the correct place to set shmmax on OSX 10.3 or > later ... but we have seen prior reports of people having trouble > getting the setting to "take". There may be some other constraint > involved. > > >>sysctl -w kern.sysv.shmmax=4294967296 // byte > > > Hmm, does sysctl work for values that exceed the range of int? > > There's no particularly good reason to try to set shmmax as high as you > are trying anyhow; you really don't need more than a couple hundred meg > in Postgres shared memory. It's better to leave the kernel to manage > the bulk of your RAM. > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster From pgsql-performance-owner@postgresql.org Tue Apr 13 18:23:41 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8D580D1DCF5 for ; Tue, 13 Apr 2004 18:23:38 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08836-07 for ; Tue, 13 Apr 2004 18:23:38 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id B975CD1DB87 for ; Tue, 13 Apr 2004 18:23:35 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3DLNaUe026712; Tue, 13 Apr 2004 17:23:36 -0400 (EDT) To: "Peter Darley" Cc: "Pgsql-Performance" Subject: Re: View columns calculated In-reply-to: References: Comments: In-reply-to "Peter Darley" message dated "Tue, 13 Apr 2004 13:49:53 -0700" Date: Tue, 13 Apr 2004 17:23:36 -0400 Message-ID: <26711.1081891416@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/160 X-Sequence-Number: 6460 "Peter Darley" writes: > I have a question about views: I want to have a fairly wide view (lots of > columns) where most of the columns have some heavyish calculations in them, > but I'm concerned that it will have to calculate every column even when I'm > not selecting them. So, the question is, if I have 5 columns in a view but > only select 1 column, is the system smart enough to not calculate the unused > columns, It depends on what the rest of your view looks like. If the view is simple enough to be "flattened" into the parent query then the unused columns will disappear into the ether. If it's not flattenable then they will get evaluated. You can check by seeing whether an EXPLAIN shows a separate "subquery scan" node corresponding to the view. (Without bothering to look at the code, an unflattenable view is one that uses GROUP BY, DISTINCT, aggregates, ORDER BY, LIMIT, UNION, INTERSECT, EXCEPT, probably a couple other things.) regards, tom lane From pgsql-performance-owner@postgresql.org Wed Apr 14 09:39:23 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C1601D1E106 for ; Wed, 14 Apr 2004 09:39:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92244-10 for ; Wed, 14 Apr 2004 09:39:24 -0300 (ADT) Received: from mwinf0103.wanadoo.fr (smtp1.wanadoo.fr [193.252.22.30]) by svr1.postgresql.org (Postfix) with ESMTP id 8E456D1DC64 for ; Wed, 14 Apr 2004 09:39:20 -0300 (ADT) Received: from [192.168.0.2] (AMontsouris-108-1-33-122.w81-48.abo.wanadoo.fr [81.48.85.122]) by mwinf0103.wanadoo.fr (SMTP Server) with ESMTP id 9A0DC1BF94E8 for ; Wed, 14 Apr 2004 14:39:23 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v613) Content-Transfer-Encoding: quoted-printable Message-Id: Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed To: pgsql-performance@postgresql.org From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Subject: Date: Wed, 14 Apr 2004 14:39:21 +0200 X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/161 X-Sequence-Number: 6461 I run the following command three times to prevent cache/disk results. [...] dps=3D> explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D764518963 AND status IN (200,304,206) ORDER BY rec_id LIMIT 1; QUERY PLAN ------------------------------------------------------------------------=20 ----------------------------------------------------- Limit (cost=3D173.14..173.14 rows=3D1 width=3D4) (actual time=3D0.357..0= .358=20=20 rows=3D1 loops=3D1) -> Sort (cost=3D173.14..173.22 rows=3D32 width=3D4) (actual=20=20 time=3D0.354..0.354 rows=3D1 loops=3D1) Sort Key: rec_id -> Index Scan using url_crc on url (cost=3D0.00..172.34=20=20 rows=3D32 width=3D4) (actual time=3D0.039..0.271 rows=3D50 loops=3D1) Index Cond: (crc32 =3D 764518963) Filter: ((crc32 <> 0) AND ((status =3D 200) OR (status =3D= =20=20 304) OR (status =3D 206))) Total runtime: 0.410 ms (7 rows) dps=3D> explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D764518963 AND status IN (200,304,206) ORDER BY crc32,rec_id LIMIT= =20=20 1; QUERY PLAN ------------------------------------------------------------------------=20 ----------------------------------------------------- Limit (cost=3D173.14..173.14 rows=3D1 width=3D8) (actual time=3D0.378..0= .378=20=20 rows=3D1 loops=3D1) -> Sort (cost=3D173.14..173.22 rows=3D32 width=3D8) (actual=20=20 time=3D0.375..0.375 rows=3D1 loops=3D1) Sort Key: crc32, rec_id -> Index Scan using url_crc on url (cost=3D0.00..172.34=20=20 rows=3D32 width=3D8) (actual time=3D0.038..0.278 rows=3D50 loops=3D1) Index Cond: (crc32 =3D 764518963) Filter: ((crc32 <> 0) AND ((status =3D 200) OR (status =3D= =20=20 304) OR (status =3D 206))) Total runtime: 0.432 ms (7 rows) dps=3D> explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D419903683 AND status IN (200,304,206) ORDER BY rec_id LIMIT 1; QUERY PLAN ------------------------------------------------------------------------=20 ------------------------------------------------------ Limit (cost=3D0.00..37.03 rows=3D1 width=3D4) (actual time=3D156.712..15= 6.713=20=20 rows=3D1 loops=3D1) -> Index Scan using url_pkey on url (cost=3D0.00..14996.82 rows=3D405= =20=20 width=3D4) (actual time=3D156.707..156.707 rows=3D1 loops=3D1) Filter: ((crc32 <> 0) AND (crc32 =3D 419903683) AND ((status =3D= =20=20 200) OR (status =3D 304) OR (status =3D 206))) Total runtime: 156.769 ms (4 rows) dps=3D> explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D419903683 AND status IN (200,304,206) ORDER BY crc32,rec_id LIMIT= =20=20 1; QUERY PLAN ------------------------------------------------------------------------=20 -------------------------------------------------------- Limit (cost=3D1910.14..1910.14 rows=3D1 width=3D8) (actual=20=20 time=3D4.558..4.559 rows=3D1 loops=3D1) -> Sort (cost=3D1910.14..1911.15 rows=3D405 width=3D8) (actual=20=20 time=3D4.555..4.555 rows=3D1 loops=3D1) Sort Key: crc32, rec_id -> Index Scan using url_crc on url (cost=3D0.00..1892.60=20=20 rows=3D405 width=3D8) (actual time=3D0.042..2.935 rows=3D719 loops=3D1) Index Cond: (crc32 =3D 419903683) Filter: ((crc32 <> 0) AND ((status =3D 200) OR (status =3D= =20=20 304) OR (status =3D 206))) Total runtime: 4.636 ms (7 rows) The value 764518963 is not common, it appears 50 times in the table. The value 419903683 is the third most common value of the table url. dps=3D> select u.crc32, count(*) from url u group by u.crc32 order by=20= =20 count(*) desc; crc32 | count -------------+------ 0 | 82202 -946427862 | 10545 419903683 | 719 945866756 | 670 [...] How to setup pgsql to correctly select the good index for index scan ? I run Pgsql 7.4.x The database runs under pg_autovacuum daemon. And a VACUUM FULL VERBOSE ANALYZE was done 10 hours before. Cordialement, Jean-G=E9rard Pailloncy From pgsql-performance-owner@postgresql.org Wed Apr 14 10:23:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D54ABD1B49B for ; Wed, 14 Apr 2004 10:23:07 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05960-10 for ; Wed, 14 Apr 2004 10:23:09 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E5E57D1E949 for ; Wed, 14 Apr 2004 10:23:05 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3EDMgSV004924; Wed, 14 Apr 2004 09:22:42 -0400 (EDT) To: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Cc: pgsql-performance@postgresql.org Subject: Re: In-reply-to: References: Comments: In-reply-to =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= message dated "Wed, 14 Apr 2004 14:39:21 +0200" Date: Wed, 14 Apr 2004 09:22:42 -0400 Message-ID: <4923.1081948962@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/162 X-Sequence-Number: 6462 =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= writes: > I run the following command three times to prevent cache/disk results. Do you think that's actually representative of how your database will behave under load? If the DB is small enough to be completely cached in RAM, and you expect it to remain so, then it's sensible to optimize on the basis of fully-cached test cases. Otherwise I think you are optimizing the wrong thing. If you do want to plan on this basis, you want to set random_page_cost to 1, make sure effective_cache_size is large, and perhaps increase the cpu_xxx cost numbers. (What you're essentially doing here is reducing the estimated cost of a page fetch relative to CPU effort.) regards, tom lane From pgsql-performance-owner@postgresql.org Wed Apr 14 14:53:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 40B39D1E104 for ; Wed, 14 Apr 2004 14:53:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31479-07 for ; Wed, 14 Apr 2004 14:53:07 -0300 (ADT) Received: from trade-india.com (ns5.trade-india.com [66.234.10.13]) by svr1.postgresql.org (Postfix) with SMTP id 2AA8DD1E17E for ; Wed, 14 Apr 2004 14:53:05 -0300 (ADT) Received: (qmail 29057 invoked from network); 14 Apr 2004 14:23:59 -0000 Received: from unknown (HELO trade-india.com) (203.145.130.142) by ns5.trade-india.com with SMTP; 14 Apr 2004 14:23:59 -0000 Message-ID: <407D7A89.60508@trade-india.com> Date: Wed, 14 Apr 2004 23:23:13 +0530 From: Rajesh Kumar Mallah User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Postgres Performance Subject: select count(*) very slow on an already vacuumed table. Content-Type: multipart/alternative; boundary="------------070309060401080600010502" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/163 X-Sequence-Number: 6463 This is a multi-part message in MIME format. --------------070309060401080600010502 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Hi I have .5 million rows in a table. My problem is select count(*) takes ages. VACUUM FULL does not help. can anyone please tell me how to i enhance the performance of the setup. Regds mallah. postgresql.conf ---------------------- max_fsm_pages = 55099264 # min max_fsm_relations*16, 6 bytes each max_fsm_relations = 5000 tradein_clients=# explain analyze SELECT count(*) from eyp_rfi; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------- Aggregate (cost=78311.37..78311.37 rows=1 width=0) (actual time=42306.902..42306.903 rows=1 loops=1) -> Seq Scan on eyp_rfi (cost=0.00..77046.49 rows=505949 width=0) (actual time=0.032..41525.007 rows=505960 loops=1) Total runtime: 42306.995 ms (3 rows) tradein_clients=# SELECT count(*) from eyp_rfi; count -------- 505960 (1 row) tradein_clients=# VACUUM full verbose eyp_rfi; INFO: vacuuming "public.eyp_rfi" INFO: "eyp_rfi": found 0 removable, 505960 nonremovable row versions in 71987 pages DETAIL: 0 dead row versions cannot be removed yet. Nonremovable row versions range from 186 to 2036 bytes long. There were 42587 unused item pointers. Total free space (including removable row versions) is 21413836 bytes. 0 pages are or will become empty, including 0 at the end of the table. 38693 pages containing 19146684 free bytes are potential move destinations. CPU 2.62s/0.40u sec elapsed 38.45 sec. INFO: index "eyp_rfi_date" now contains 505960 row versions in 1197 pages DETAIL: 0 index row versions were removed. 4 index pages have been deleted, 4 are currently reusable. CPU 0.05s/0.29u sec elapsed 0.87 sec. INFO: index "eyp_rfi_receiver_uid" now contains 505960 row versions in 1163 pages DETAIL: 0 index row versions were removed. 1 index pages have been deleted, 1 are currently reusable. CPU 0.03s/0.42u sec elapsed 1.33 sec. INFO: index "eyp_rfi_inhouse" now contains 505960 row versions in 1208 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.04s/0.21u sec elapsed 1.20 sec. INFO: index "eyp_rfi_rfi_id_key" now contains 505960 row versions in 1201 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.03s/0.33u sec elapsed 0.81 sec. INFO: index "eyp_rfi_list_id_idx" now contains 505960 row versions in 1133 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.02s/0.43u sec elapsed 1.12 sec. INFO: index "eyp_rfi_status" now contains 505960 row versions in 1448 pages DETAIL: 0 index row versions were removed. 4 index pages have been deleted, 4 are currently reusable. CPU 0.05s/0.22u sec elapsed 1.08 sec. INFO: index "eyp_rfi_list_id" now contains 505960 row versions in 1133 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.02s/0.43u sec elapsed 1.00 sec. INFO: index "eyp_rfi_receiver_email" now contains 505960 row versions in 2801 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.16s/0.52u sec elapsed 10.38 sec. INFO: index "eyp_rfi_subj" now contains 80663 row versions in 463 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.03s/0.14u sec elapsed 3.20 sec. INFO: index "eyp_rfi_sender" now contains 505960 row versions in 3025 pages DETAIL: 0 index row versions were removed. 6 index pages have been deleted, 6 are currently reusable. CPU 0.10s/0.39u sec elapsed 4.99 sec. INFO: index "eyp_sender_uid_idx" now contains 505960 row versions in 1216 pages DETAIL: 0 index row versions were removed. 5 index pages have been deleted, 5 are currently reusable. CPU 0.04s/0.36u sec elapsed 2.61 sec. INFO: index "eyp_rfi_rec_uid_idx" now contains 505960 row versions in 1166 pages DETAIL: 0 index row versions were removed. 1 index pages have been deleted, 1 are currently reusable. CPU 0.05s/0.41u sec elapsed 2.04 sec. INFO: index "eyp_rfi_index" now contains 505960 row versions in 2051 pages DETAIL: 0 index row versions were removed. 7 index pages have been deleted, 7 are currently reusable. CPU 0.10s/0.28u sec elapsed 8.16 sec. INFO: "eyp_rfi": moved 0 row versions, truncated 71987 to 71987 pages DETAIL: CPU 2.03s/2.09u sec elapsed 95.24 sec. INFO: vacuuming "pg_toast.pg_toast_19609" INFO: "pg_toast_19609": found 0 removable, 105342 nonremovable row versions in 21038 pages DETAIL: 0 dead row versions cannot be removed yet. Nonremovable row versions range from 37 to 2034 bytes long. There were 145 unused item pointers. Total free space (including removable row versions) is 16551072 bytes. 0 pages are or will become empty, including 0 at the end of the table. 18789 pages containing 16512800 free bytes are potential move destinations. CPU 0.70s/0.09u sec elapsed 41.64 sec. INFO: index "pg_toast_19609_index" now contains 105342 row versions in 296 pages DETAIL: 0 index row versions were removed. 0 index pages have been deleted, 0 are currently reusable. CPU 0.00s/0.02u sec elapsed 0.63 sec. INFO: "pg_toast_19609": moved 0 row versions, truncated 21038 to 21038 pages DETAIL: CPU 0.01s/0.01u sec elapsed 10.03 sec. VACUUM tradein_clients=# explain analyze SELECT count(*) from eyp_rfi; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------- Aggregate (cost=78311.50..78311.50 rows=1 width=0) (actual time=50631.488..50631.489 rows=1 loops=1) -> Seq Scan on eyp_rfi (cost=0.00..77046.60 rows=505960 width=0) (actual time=0.030..49906.198 rows=505964 loops=1) Total runtime: 50631.658 ms (3 rows) --------------070309060401080600010502 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit
Hi
I have .5 million rows in a table. My problem is select count(*) takes ages.
VACUUM FULL does not help. can anyone please tell me
how to i enhance the performance of the setup.

Regds
mallah.

postgresql.conf
----------------------
max_fsm_pages = 55099264                # min max_fsm_relations*16, 6 bytes each
max_fsm_relations = 5000


tradein_clients=# explain analyze SELECT count(*) from eyp_rfi;
                                                       QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------
 Aggregate  (cost=78311.37..78311.37 rows=1 width=0) (actual time=42306.902..42306.903 rows=1 loops=1)
   ->  Seq Scan on eyp_rfi  (cost=0.00..77046.49 rows=505949 width=0) (actual time=0.032..41525.007 rows=505960 loops=1)
 Total runtime: 42306.995 ms
(3 rows)

tradein_clients=# SELECT count(*) from eyp_rfi;
 count
--------
 505960
(1 row)

tradein_clients=# VACUUM full verbose eyp_rfi;
INFO:  vacuuming "public.eyp_rfi"
INFO:  "eyp_rfi": found 0 removable, 505960 nonremovable row versions in 71987 pages
DETAIL:  0 dead row versions cannot be removed yet.
Nonremovable row versions range from 186 to 2036 bytes long.
There were 42587 unused item pointers.
Total free space (including removable row versions) is 21413836 bytes.
0 pages are or will become empty, including 0 at the end of the table.
38693 pages containing 19146684 free bytes are potential move destinations.
CPU 2.62s/0.40u sec elapsed 38.45 sec.
INFO:  index "eyp_rfi_date" now contains 505960 row versions in 1197 pages
DETAIL:  0 index row versions were removed.
4 index pages have been deleted, 4 are currently reusable.
CPU 0.05s/0.29u sec elapsed 0.87 sec.
INFO:  index "eyp_rfi_receiver_uid" now contains 505960 row versions in 1163 pages
DETAIL:  0 index row versions were removed.
1 index pages have been deleted, 1 are currently reusable.
CPU 0.03s/0.42u sec elapsed 1.33 sec.
INFO:  index "eyp_rfi_inhouse" now contains 505960 row versions in 1208 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.04s/0.21u sec elapsed 1.20 sec.
INFO:  index "eyp_rfi_rfi_id_key" now contains 505960 row versions in 1201 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.03s/0.33u sec elapsed 0.81 sec.
INFO:  index "eyp_rfi_list_id_idx" now contains 505960 row versions in 1133 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.02s/0.43u sec elapsed 1.12 sec.
INFO:  index "eyp_rfi_status" now contains 505960 row versions in 1448 pages
DETAIL:  0 index row versions were removed.
4 index pages have been deleted, 4 are currently reusable.
CPU 0.05s/0.22u sec elapsed 1.08 sec.
INFO:  index "eyp_rfi_list_id" now contains 505960 row versions in 1133 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.02s/0.43u sec elapsed 1.00 sec.
INFO:  index "eyp_rfi_receiver_email" now contains 505960 row versions in 2801 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.16s/0.52u sec elapsed 10.38 sec.
INFO:  index "eyp_rfi_subj" now contains 80663 row versions in 463 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.03s/0.14u sec elapsed 3.20 sec.
INFO:  index "eyp_rfi_sender" now contains 505960 row versions in 3025 pages
DETAIL:  0 index row versions were removed.
6 index pages have been deleted, 6 are currently reusable.
CPU 0.10s/0.39u sec elapsed 4.99 sec.
INFO:  index "eyp_sender_uid_idx" now contains 505960 row versions in 1216 pages
DETAIL:  0 index row versions were removed.
5 index pages have been deleted, 5 are currently reusable.
CPU 0.04s/0.36u sec elapsed 2.61 sec.
INFO:  index "eyp_rfi_rec_uid_idx" now contains 505960 row versions in 1166 pages
DETAIL:  0 index row versions were removed.
1 index pages have been deleted, 1 are currently reusable.
CPU 0.05s/0.41u sec elapsed 2.04 sec.
INFO:  index "eyp_rfi_index" now contains 505960 row versions in 2051 pages
DETAIL:  0 index row versions were removed.
7 index pages have been deleted, 7 are currently reusable.
CPU 0.10s/0.28u sec elapsed 8.16 sec.
INFO:  "eyp_rfi": moved 0 row versions, truncated 71987 to 71987 pages
DETAIL:  CPU 2.03s/2.09u sec elapsed 95.24 sec.
INFO:  vacuuming "pg_toast.pg_toast_19609"
INFO:  "pg_toast_19609": found 0 removable, 105342 nonremovable row versions in 21038 pages
DETAIL:  0 dead row versions cannot be removed yet.
Nonremovable row versions range from 37 to 2034 bytes long.
There were 145 unused item pointers.
Total free space (including removable row versions) is 16551072 bytes.
0 pages are or will become empty, including 0 at the end of the table.
18789 pages containing 16512800 free bytes are potential move destinations.
CPU 0.70s/0.09u sec elapsed 41.64 sec.
INFO:  index "pg_toast_19609_index" now contains 105342 row versions in 296 pages
DETAIL:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.02u sec elapsed 0.63 sec.
INFO:  "pg_toast_19609": moved 0 row versions, truncated 21038 to 21038 pages
DETAIL:  CPU 0.01s/0.01u sec elapsed 10.03 sec.
VACUUM
tradein_clients=# explain analyze SELECT count(*) from eyp_rfi;
                                                       QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------
 Aggregate  (cost=78311.50..78311.50 rows=1 width=0) (actual time=50631.488..50631.489 rows=1 loops=1)
   ->  Seq Scan on eyp_rfi  (cost=0.00..77046.60 rows=505960 width=0) (actual time=0.030..49906.198 rows=505964 loops=1)
 Total runtime: 50631.658 ms
(3 rows)

 --------------070309060401080600010502-- From pgsql-performance-owner@postgresql.org Wed Apr 14 16:09:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 604A7D1D16D for ; Wed, 14 Apr 2004 16:09:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 59189-06 for ; Wed, 14 Apr 2004 16:09:02 -0300 (ADT) 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 22E95D1D174 for ; Wed, 14 Apr 2004 16:09:00 -0300 (ADT) 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 1BDpkq-000PzM-0X; Wed, 14 Apr 2004 20:09:00 +0100 Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id DA96E1675B; Wed, 14 Apr 2004 20:08:59 +0100 (BST) From: Richard Huxton To: Rajesh Kumar Mallah , Postgres Performance Subject: Re: select count(*) very slow on an already vacuumed table. Date: Wed, 14 Apr 2004 20:08:58 +0100 User-Agent: KMail/1.5 References: <407D7A89.60508@trade-india.com> In-Reply-To: <407D7A89.60508@trade-india.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404142008.58533.dev@archonet.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/164 X-Sequence-Number: 6464 On Wednesday 14 April 2004 18:53, Rajesh Kumar Mallah wrote: > Hi > I have .5 million rows in a table. My problem is select count(*) takes > ages. VACUUM FULL does not help. can anyone please tell me > how to i enhance the performance of the setup. > SELECT count(*) from eyp_rfi; If this is the actual query you're running, and you need a guaranteed accurate result, then you only have one option: write a trigger function to update a table_count table with every insert/delete to eyp_rfi. There is loads of info on this (and why it isn't as simple as you might think) in the archives. First though: 1. Is this the actual query, or just a representation? 2. Do you need an accurate figure or just something "near enough"? -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Wed Apr 14 17:12:11 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D19FFD1D07E for ; Wed, 14 Apr 2004 17:12:06 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 86663-05 for ; Wed, 14 Apr 2004 17:12:06 -0300 (ADT) Received: from cmailm3.svr.pol.co.uk (cmailm3.svr.pol.co.uk [195.92.193.19]) by svr1.postgresql.org (Postfix) with ESMTP id 0ECBCD1CCB4 for ; Wed, 14 Apr 2004 17:12:04 -0300 (ADT) Received: from modem-3224.llama.dialup.pol.co.uk ([217.135.188.152] helo=stromboli) by cmailm3.svr.pol.co.uk with esmtp (Exim 4.14) id 1BDqjs-0005CC-Qj; Wed, 14 Apr 2004 21:12:05 +0100 From: "Simon Riggs" To: "Josh Berkus" , Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Wed, 14 Apr 2004 21:12:18 +0100 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: <200404050836.52115.josh@agliodbs.com> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/165 X-Sequence-Number: 6465 > Josh Berkus wrote: > Unfortunately, these days only Tom and Neil seem to be > seriously working on > the query planner (beg pardon in advance if I've missed > someone) so I think > the real answer is that we need another person interested in > this kind of > optimization before it's going to get much better. > Hmmmm. Interesting line of thought. Is the problem "a person interested" or is there another issue there? I was thinking the other day that maybe removing the ability to control join order through explicitly manipulating the FROM clause might actually be counter productive, in terms of longer term improvement of the optimizer. Treating the optimizer as a black box is something I'm very used to from other RDBMS. My question is, how can you explicitly re-write a query now to "improve" it? If there's no way of manipulating queries without actually re-writing the optimizer, we're now in a position where we aren't able to diagnose when the optimizer isn't working effectively. For my mind, all the people on this list are potential "optimizer developers" in the sense that we can all look at queries and see whether there is a problem with particular join plans. Providing good cases of poor optimization is just what's needed to assist those few that do understand the internals to continue improving things. I guess what I'm saying is it's not how many people you've got working on the optimizer, its how many accurate field reports of less-than perfect optimization reach them. In that case, PostgreSQL is likely in a better position than Microsoft, since the accessibility of the pg discussion lists makes such cases much more likely to get aired. Any thoughts? Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Wed Apr 14 17:42:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 44401D1DB0A for ; Wed, 14 Apr 2004 17:42:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96507-08 for ; Wed, 14 Apr 2004 17:42:54 -0300 (ADT) Received: from trade-india.com (ns5.trade-india.com [66.234.10.13]) by svr1.postgresql.org (Postfix) with SMTP id 56B4CD1D6CC for ; Wed, 14 Apr 2004 17:42:52 -0300 (ADT) Received: (qmail 27899 invoked from network); 14 Apr 2004 17:13:50 -0000 Received: from unknown (HELO trade-india.com) (203.145.130.142) by ns5.trade-india.com with SMTP; 14 Apr 2004 17:13:50 -0000 Message-ID: <407DA257.6010904@trade-india.com> Date: Thu, 15 Apr 2004 02:13:03 +0530 From: Rajesh Kumar Mallah User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: Postgres Performance Subject: Re: select count(*) very slow on an already vacuumed table. References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> In-Reply-To: <200404142008.58533.dev@archonet.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: 200404/166 X-Sequence-Number: 6466 Richard Huxton wrote: >On Wednesday 14 April 2004 18:53, Rajesh Kumar Mallah wrote: > > >>Hi >>I have .5 million rows in a table. My problem is select count(*) takes >>ages. VACUUM FULL does not help. can anyone please tell me >>how to i enhance the performance of the setup. >> >> > > > >>SELECT count(*) from eyp_rfi; >> >> > >If this is the actual query you're running, and you need a guaranteed accurate >result, then you only have one option: write a trigger function to update a >table_count table with every insert/delete to eyp_rfi. > > it is just an example. in general all the queries that involves eyp_rfi become slow. reloading the table makes the query faster. mallah. >There is loads of info on this (and why it isn't as simple as you might think) >in the archives. First though: >1. Is this the actual query, or just a representation? >2. Do you need an accurate figure or just something "near enough"? > > > From pgsql-performance-owner@postgresql.org Thu Apr 15 04:17:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E398FD1C9DA for ; Thu, 15 Apr 2004 04:17:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 90604-05 for ; Thu, 15 Apr 2004 04:17:51 -0300 (ADT) Received: from mail.unisoftbg.com (unknown [194.12.229.207]) by svr1.postgresql.org (Postfix) with ESMTP id A4F44D1BB49 for ; Thu, 15 Apr 2004 04:17:44 -0300 (ADT) Received: (qmail 4808 invoked by uid 507); 15 Apr 2004 07:25:00 -0000 Received: from unknown (HELO t1.unisoftbg.com) (pginfo%t1.unisoftbg.com@194.12.229.193) by 0 with SMTP; 15 Apr 2004 07:25:00 -0000 Message-ID: <407E25A6.ADFE4815@t1.unisoftbg.com> Date: Thu, 15 Apr 2004 08:03:18 +0200 From: pginfo X-Mailer: Mozilla 4.03 [en] (Win95; I) MIME-Version: 1.0 To: "pgsql-performance@postgresql.org" Subject: linux distro for better pg performance 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=1.5 tagged_above=0.0 required=5.0 tests=BAYES_40, RCVD_IN_DSBL, TO_ADDRESS_EQ_REAL X-Spam-Level: * X-Archive-Number: 200404/168 X-Sequence-Number: 6468 Hi, I am using pg from 3 y. and generaly I do not have big problems with it. I am searching for best pg distro to run pg (7.4.1). At the moment I am using RedHat AS 3.0, but I think it have some performance problems (I am not sure). My configuration: P4 2.8 GHz 1 GB RAM 120 GB IDE 7200 disk. Kernel version 2.4.21-4.EL (it is the installation vesrion for rh 3.0) . My problems: If I run some query with many reads, I see a massive disk transfer : procs memory swap io system cpu r b swpd free buff cache si so bi bo in cs us sy id wa 0 0 0 261724 3252 670748 0 0 0 4 105 19 0 0 100 0 0 0 0 261724 3252 670748 0 0 0 0 101 11 0 0 100 0 0 0 0 261724 3260 670748 0 0 0 4 104 19 0 0 100 0 0 1 0 259684 3268 674112 0 0 964 7 131 57 0 0 95 4 1 0 0 119408 3288 808540 0 0 27960 0 572 630 13 14 24 49 1 1 0 15896 3292 914436 0 0 7984 44744 531 275 11 18 24 47 0 2 0 16292 3296 924996 0 0 4145 6413 384 176 2 5 0 92 0 1 0 19928 3316 928844 0 0 11805 13335 497 388 5 9 5 81 0 3 0 19124 3296 924452 0 0 3153 19164 287 295 5 11 16 68 0 1 0 15956 3304 932984 0 0 536 6812 366 123 4 6 3 87 0 2 0 24956 3300 921416 0 0 1931 22936 And if I run top, I see a big iowait % (some times 70-80) and very low user % (10-15). I readet many docs about this problem, but do not find any solution. My question: If some one is using RH 3.0, pls post some result or suggestions for it performance with pg . What is the best linux distro for pg? Can I get better performance by using 15K SCSI disk ? Or it will be better to have more RAM (2 or 3 GB) ? regards, ivan. From pgsql-performance-owner@postgresql.org Thu Apr 15 04:08:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 446A7D1C9C8 for ; Thu, 15 Apr 2004 04:08:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85522-05 for ; Thu, 15 Apr 2004 04:08:46 -0300 (ADT) Received: from trade-india.com (unknown [61.16.154.82]) by svr1.postgresql.org (Postfix) with SMTP id 6E6D2D1C9BE for ; Thu, 15 Apr 2004 04:08:41 -0300 (ADT) Received: (qmail 11990 invoked from network); 15 Apr 2004 06:55:43 -0000 Received: from valium.trade-india-local.com (HELO trade-india.com) (192.168.0.72) by system66.trade-india-local.com with SMTP; 15 Apr 2004 06:55:43 -0000 Message-ID: <407E3563.6000408@trade-india.com> Date: Thu, 15 Apr 2004 12:40:27 +0530 From: Rajesh Kumar Mallah Organization: Infocom Network Limited User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: Postgres Performance Subject: Re: select count(*) very slow on an already vacuumed table. References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> In-Reply-To: <200404142008.58533.dev@archonet.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: 200404/167 X-Sequence-Number: 6467 The problem is that i want to know if i need a Hardware upgrade at the moment. Eg i have another table rfis which contains ~ .6 million records. SELECT count(*) from rfis where sender_uid > 0; +--------+ | count | +--------+ | 564870 | +--------+ Time: 117560.635 ms Which is approximate 4804 records per second. Is it an acceptable performance on the hardware below: RAM: 2 GB DISKS: ultra160 , 10 K , 18 GB Processor: 2* 2.0 Ghz Xeon What kind of upgrades shoud be put on the server for it to become reasonable fast. Regds mallah. Richard Huxton wrote: >On Wednesday 14 April 2004 18:53, Rajesh Kumar Mallah wrote: > > >>Hi >>I have .5 million rows in a table. My problem is select count(*) takes >>ages. VACUUM FULL does not help. can anyone please tell me >>how to i enhance the performance of the setup. >> >> > > > >>SELECT count(*) from eyp_rfi; >> >> > >If this is the actual query you're running, and you need a guaranteed accurate >result, then you only have one option: write a trigger function to update a >table_count table with every insert/delete to eyp_rfi. > >There is loads of info on this (and why it isn't as simple as you might think) >in the archives. First though: >1. Is this the actual query, or just a representation? >2. Do you need an accurate figure or just something "near enough"? > > > From pgsql-performance-owner@postgresql.org Thu Apr 15 04:32:57 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 88F58D1D06D for ; Thu, 15 Apr 2004 04:32:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 95713-04 for ; Thu, 15 Apr 2004 04:32:55 -0300 (ADT) Received: from trade-india.com (unknown [61.16.154.82]) by svr1.postgresql.org (Postfix) with SMTP id C0D56D1CCDA for ; Thu, 15 Apr 2004 04:32:51 -0300 (ADT) Received: (qmail 13040 invoked from network); 15 Apr 2004 07:19:56 -0000 Received: from valium.trade-india-local.com (HELO trade-india.com) (192.168.0.72) by system66.trade-india-local.com with SMTP; 15 Apr 2004 07:19:56 -0000 Message-ID: <407E3B10.5060107@trade-india.com> Date: Thu, 15 Apr 2004 13:04:40 +0530 From: Rajesh Kumar Mallah Organization: Infocom Network Limited User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: Postgres Performance Subject: Re: select count(*) very slow on an already vacuumed table. References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> In-Reply-To: <407E3563.6000408@trade-india.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: 200404/169 X-Sequence-Number: 6469 The relation size for this table is 1.7 GB tradein_clients=# SELECT public.relation_size ('general.rfis'); +------------------+ | relation_size | +------------------+ | 1,762,639,872 | +------------------+ (1 row) Regds mallah. Rajesh Kumar Mallah wrote: > > > The problem is that i want to know if i need a Hardware upgrade > at the moment. > > Eg i have another table rfis which contains ~ .6 million records. > > > SELECT count(*) from rfis where sender_uid > 0; > +--------+ > | count | > +--------+ > | 564870 | > +--------+ > Time: 117560.635 ms > > Which is approximate 4804 records per second. Is it an acceptable > performance on the hardware below: > > RAM: 2 GB > DISKS: ultra160 , 10 K , 18 GB > Processor: 2* 2.0 Ghz Xeon > > What kind of upgrades shoud be put on the server for it to become > reasonable fast. > > > Regds > mallah. > > > > > Richard Huxton wrote: > >> On Wednesday 14 April 2004 18:53, Rajesh Kumar Mallah wrote: >> >> >>> Hi >>> I have .5 million rows in a table. My problem is select count(*) takes >>> ages. VACUUM FULL does not help. can anyone please tell me >>> how to i enhance the performance of the setup. >>> >> >> >> >> >>> SELECT count(*) from eyp_rfi; >>> >> >> >> If this is the actual query you're running, and you need a guaranteed >> accurate result, then you only have one option: write a trigger >> function to update a table_count table with every insert/delete to >> eyp_rfi. >> >> There is loads of info on this (and why it isn't as simple as you >> might think) in the archives. First though: >> 1. Is this the actual query, or just a representation? >> 2. Do you need an accurate figure or just something "near enough"? >> >> >> > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) > From pgsql-performance-owner@postgresql.org Thu Apr 15 04:53:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D0925D1CCDA for ; Thu, 15 Apr 2004 04:53:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04793-02 for ; Thu, 15 Apr 2004 04:53:35 -0300 (ADT) 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 AB4E2D1CCB0 for ; Thu, 15 Apr 2004 04:53:33 -0300 (ADT) 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 1BE1gk-000Pgu-0U; Thu, 15 Apr 2004 08:53:34 +0100 Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 221CB1624A; Thu, 15 Apr 2004 08:53:33 +0100 (BST) From: Richard Huxton To: Rajesh Kumar Mallah Subject: Re: select count(*) very slow on an already vacuumed table. Date: Thu, 15 Apr 2004 08:53:32 +0100 User-Agent: KMail/1.5 Cc: Postgres Performance References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> In-Reply-To: <407E3563.6000408@trade-india.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404150853.32475.dev@archonet.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/170 X-Sequence-Number: 6470 On Thursday 15 April 2004 08:10, Rajesh Kumar Mallah wrote: > The problem is that i want to know if i need a Hardware upgrade > at the moment. > > Eg i have another table rfis which contains ~ .6 million records. > SELECT count(*) from rfis where sender_uid > 0; > Time: 117560.635 ms > > Which is approximate 4804 records per second. Is it an acceptable > performance on the hardware below: > > RAM: 2 GB > DISKS: ultra160 , 10 K , 18 GB > Processor: 2* 2.0 Ghz Xeon Hmm - doesn't seem good, does it? If you run it again, is it much faster (since the data should be cached then)? What does "vmstat 10" show while you're running the query? One thing you should have done is read the performance tuning guide at: http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php The default values are very conservative, and you will need to change them. > What kind of upgrades shoud be put on the server for it to become > reasonable fast. If you've only got one disk, then a second disk for OS/logging. Difficult to say more without knowing numbers of users/activity etc. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Thu Apr 15 07:29:44 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7B449D1D070 for ; Thu, 15 Apr 2004 07:29:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 58497-08 for ; Thu, 15 Apr 2004 07:29:44 -0300 (ADT) Received: from trade-india.com (unknown [61.16.154.82]) by svr1.postgresql.org (Postfix) with SMTP id 0CB4AD1C4C3 for ; Thu, 15 Apr 2004 07:29:39 -0300 (ADT) Received: (qmail 21893 invoked from network); 15 Apr 2004 10:16:42 -0000 Received: from valium.trade-india-local.com (HELO trade-india.com) (192.168.0.72) by system66.trade-india-local.com with SMTP; 15 Apr 2004 10:16:42 -0000 Message-ID: <407E647E.7090207@trade-india.com> Date: Thu, 15 Apr 2004 16:01:26 +0530 From: Rajesh Kumar Mallah Organization: Infocom Network Limited User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: Postgres Performance Subject: Re: select count(*) very slow on an already vacuumed table. References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> <200404150853.32475.dev@archonet.com> In-Reply-To: <200404150853.32475.dev@archonet.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: 200404/171 X-Sequence-Number: 6471 Richard Huxton wrote: >On Thursday 15 April 2004 08:10, Rajesh Kumar Mallah wrote: > > >>The problem is that i want to know if i need a Hardware upgrade >>at the moment. >> >>Eg i have another table rfis which contains ~ .6 million records. >> >> > > > >>SELECT count(*) from rfis where sender_uid > 0; >> >> > > > >>Time: 117560.635 ms >> >>Which is approximate 4804 records per second. Is it an acceptable >>performance on the hardware below: >> >>RAM: 2 GB >>DISKS: ultra160 , 10 K , 18 GB >>Processor: 2* 2.0 Ghz Xeon >> >> > >Hmm - doesn't seem good, does it? If you run it again, is it much faster >(since the data should be cached then)? What does "vmstat 10" show while >you're running the query? > >One thing you should have done is read the performance tuning guide at: > http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php >The default values are very conservative, and you will need to change them. > > Hi, Thanks for the interest . my config are not the default ones. i was running iostat while running the query. Looks like one of the disks doesnt' go past a read performance of 20 ,000 KBytes/sec while the other disk it goes as high as 40,000 . What i am ding currently is loading the table in both the disks and compare the table scan speeds. The performance is definitely better in the newly loaded table in the other disk . the load in server is 13 because i am simultaneously re-loading the data in other table. rt2=# SELECT count(*) from rfis where sender_uid > 0; +--------+ | count | +--------+ | 564870 | +--------+ (1 row) Time: 10288.359 ms rt2=# shall post the comparitive details under normal load soon regds mallah. > > >>What kind of upgrades shoud be put on the server for it to become >>reasonable fast. >> >> > >If you've only got one disk, then a second disk for OS/logging. Difficult to >say more without knowing numbers of users/activity etc. > > > From pgsql-performance-owner@postgresql.org Thu Apr 15 09:13:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7BDD7D1D06F for ; Thu, 15 Apr 2004 09:13:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01833-02 for ; Thu, 15 Apr 2004 09:13:35 -0300 (ADT) Received: from mail.unisoftbg.com (unknown [194.12.229.207]) by svr1.postgresql.org (Postfix) with ESMTP id 05806D1B515 for ; Thu, 15 Apr 2004 09:13:27 -0300 (ADT) Received: (qmail 15175 invoked by uid 507); 15 Apr 2004 12:20:47 -0000 Received: from unknown (HELO t1.unisoftbg.com) (pginfo%t1.unisoftbg.com@194.12.229.193) by 0 with SMTP; 15 Apr 2004 12:20:47 -0000 Message-ID: <407E6AF8.7643C61F@t1.unisoftbg.com> Date: Thu, 15 Apr 2004 12:59:04 +0200 From: pginfo X-Mailer: Mozilla 4.03 [en] (Win95; I) MIME-Version: 1.0 To: Rajesh Kumar Mallah Cc: Richard Huxton , Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> <200404150853.32475.dev@archonet.com> <407E7391.4060706@trade-india.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/174 X-Sequence-Number: 6474 Hi , I am not sure, but I remember the same problem. It was ot 7.3.x version and and I needet to reindex the table. I think after 7.4 vacuum also work correct with reindex. But I am not sure. regards, ivan. Rajesh Kumar Mallah wrote: > Hi, > > The problem was solved by reloading the Table. > the query now takes only 3 seconds. But that is > not a solution. > > The problem is that such phenomenon obscures our > judgement used in optimising queries and database. > > If a query runs slow we really cant tell if its a problem > with query itself , hardware or dead rows. > > I already did vacumm full on the table but it still did not > have that effect on performance. > In fact the last figures were after doing a vacuum full. > > Can there be any more elegent solution to this problem. > > Regds > Mallah. > > Richard Huxton wrote: > > >On Thursday 15 April 2004 08:10, Rajesh Kumar Mallah wrote: > > > > > >>The problem is that i want to know if i need a Hardware upgrade > >>at the moment. > >> > >>Eg i have another table rfis which contains ~ .6 million records. > >> > >> > > > > > > > >>SELECT count(*) from rfis where sender_uid > 0; > >> > >> > > > > > > > >>Time: 117560.635 ms > >> > >>Which is approximate 4804 records per second. Is it an acceptable > >>performance on the hardware below: > >> > >>RAM: 2 GB > >>DISKS: ultra160 , 10 K , 18 GB > >>Processor: 2* 2.0 Ghz Xeon > >> > >> > > > >Hmm - doesn't seem good, does it? If you run it again, is it much faster > >(since the data should be cached then)? What does "vmstat 10" show while > >you're running the query? > > > >One thing you should have done is read the performance tuning guide at: > > http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php > >The default values are very conservative, and you will need to change them. > > > > > > > >>What kind of upgrades shoud be put on the server for it to become > >>reasonable fast. > >> > >> > > > >If you've only got one disk, then a second disk for OS/logging. Difficult to > >say more without knowing numbers of users/activity etc. > > > > > > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org From pgsql-performance-owner@postgresql.org Thu Apr 15 08:34:04 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BEC07D1D391 for ; Thu, 15 Apr 2004 08:34:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80829-09 for ; Thu, 15 Apr 2004 08:34:03 -0300 (ADT) Received: from trade-india.com (unknown [61.16.154.82]) by svr1.postgresql.org (Postfix) with SMTP id DA233D1D085 for ; Thu, 15 Apr 2004 08:33:57 -0300 (ADT) Received: (qmail 24925 invoked from network); 15 Apr 2004 11:21:00 -0000 Received: from valium.trade-india-local.com (HELO trade-india.com) (192.168.0.72) by system66.trade-india-local.com with SMTP; 15 Apr 2004 11:21:00 -0000 Message-ID: <407E7391.4060706@trade-india.com> Date: Thu, 15 Apr 2004 17:05:45 +0530 From: Rajesh Kumar Mallah Organization: Infocom Network Limited User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> <200404150853.32475.dev@archonet.com> In-Reply-To: <200404150853.32475.dev@archonet.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: 200404/172 X-Sequence-Number: 6472 Hi, The problem was solved by reloading the Table. the query now takes only 3 seconds. But that is not a solution. The problem is that such phenomenon obscures our judgement used in optimising queries and database. If a query runs slow we really cant tell if its a problem with query itself , hardware or dead rows. I already did vacumm full on the table but it still did not have that effect on performance. In fact the last figures were after doing a vacuum full. Can there be any more elegent solution to this problem. Regds Mallah. Richard Huxton wrote: >On Thursday 15 April 2004 08:10, Rajesh Kumar Mallah wrote: > > >>The problem is that i want to know if i need a Hardware upgrade >>at the moment. >> >>Eg i have another table rfis which contains ~ .6 million records. >> >> > > > >>SELECT count(*) from rfis where sender_uid > 0; >> >> > > > >>Time: 117560.635 ms >> >>Which is approximate 4804 records per second. Is it an acceptable >>performance on the hardware below: >> >>RAM: 2 GB >>DISKS: ultra160 , 10 K , 18 GB >>Processor: 2* 2.0 Ghz Xeon >> >> > >Hmm - doesn't seem good, does it? If you run it again, is it much faster >(since the data should be cached then)? What does "vmstat 10" show while >you're running the query? > >One thing you should have done is read the performance tuning guide at: > http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php >The default values are very conservative, and you will need to change them. > > > >>What kind of upgrades shoud be put on the server for it to become >>reasonable fast. >> >> > >If you've only got one disk, then a second disk for OS/logging. Difficult to >say more without knowing numbers of users/activity etc. > > > From pgsql-performance-owner@postgresql.org Thu Apr 15 09:10:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6C396D1D06D for ; Thu, 15 Apr 2004 09:10:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01268-02 for ; Thu, 15 Apr 2004 09:10:04 -0300 (ADT) Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id A3ACAD1D065 for ; Thu, 15 Apr 2004 09:09:59 -0300 (ADT) Received: from aeccom.com (cayambe.core.aeccom.com [192.168.2.12]) by mail2.aeccom.com (Postfix) with ESMTP id 5BB4734 for ; Thu, 15 Apr 2004 14:10:01 +0200 (CEST) Message-ID: <407E7B99.3050306@aeccom.com> Date: Thu, 15 Apr 2004 14:10:01 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH, Berlin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031016 X-Accept-Language: de, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Toooo many context switches (maybe SLES8?) 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: 200404/173 X-Sequence-Number: 6473 Hi, we have a complex modperl database application using postgresql 7.4.1 on a new Dual Xeon MP Machine with SLES8 which seems to generate too much context switches (way more than 100.000) on higher load (meaning system load > 2). System response times significantly slow down then. We have tuned parameters for weeks now but could not come up with better results. It seems that we have had better performance on an older Dual XEON DP Machine running on RedHat 7.3. Here is the config: database machine on SuSE SLES 8: F-S Primergy RX600 2x XEON MP 2.5GHz 8GB RAM Hardware Raid 1+0 140GB Kernel 2.4.21-169-smp Postgresql 7.4.1 (self compiled) with max_connections = 170 shared_buffers = 40000 effective_cache_size = 800000 sort_mem = 30000 vacuum_mem = 420000 max_fsm_relations = 2000 max_fsm_pages = 200000 random_page_cost = 4 checkpoint_segments = 24 wal_buffers = 32 modperl application machine on RH 7.3: F-S Primergy RX200 2x XEON DP 2.4 GHz 4 GB RAM Kernel 2.4.18-10smp, RedHat 7.3 Apache 1.3.27 setup: MinSpareServers 15 MaxSpareServers 30 StartServers 15 MaxClients 80 MaxRequestsPerChild 100 vmstat 1 excerpt: procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 1 0 4868 242372 179488 6942316 0 0 12 8 18 9 6 2 92 0 2 1 4868 242204 179488 6942500 0 0 64 500 701 117921 35 18 48 0 0 1 4868 242032 179392 6941560 0 0 16 316 412 132295 28 25 47 0 1 0 4872 242396 179164 6933776 0 0 128 276 474 69708 21 24 56 0 3 0 4872 242536 179164 6933808 0 0 0 240 412 113643 27 27 46 0 2 0 4872 242872 179092 6931708 0 0 48 1132 521 127916 24 24 53 0 0 0 4876 242876 179092 6927512 0 0 48 532 504 117868 32 21 47 0 0 0 4876 242504 179096 6927560 0 0 0 188 412 127147 34 20 47 0 1 0 4876 242152 179096 6927856 0 0 96 276 529 117684 28 23 49 0 2 0 4876 242864 179096 6928384 0 0 88 560 507 135717 38 19 43 0 1 0 4876 242848 179096 6928520 0 0 64 232 433 151380 32 20 48 0 4 0 4876 242832 179144 6928916 0 0 16 10380 2913 112583 28 20 52 0 4 0 4876 242720 179144 6929240 0 0 196 0 329 154821 32 18 50 0 3 2 4876 243576 179144 6929408 0 0 0 460 451 160287 29 18 52 0 3 0 4876 243292 179180 6929468 0 0 16 436 614 51894 15 5 80 0 0 0 4876 243884 179180 6929580 0 0 0 236 619 154168 29 21 49 0 2 1 4876 243864 179180 6929860 0 0 128 380 493 155903 31 19 50 0 2 0 4876 244720 179180 6930276 0 0 16 1208 561 129336 27 16 56 0 2 0 4876 247204 179180 6930300 0 0 0 0 361 146268 33 20 47 0 3 0 4876 248620 179180 6930372 0 0 0 168 346 155915 32 12 56 0 2 0 4876 250476 179180 6930436 0 0 0 184 328 163842 35 20 46 0 0 0 4876 250496 179180 6930652 0 0 48 260 450 144930 31 15 53 0 1 0 4876 252236 179180 6930732 0 0 16 244 577 167259 35 15 50 0 0 0 4876 252236 179180 6930780 0 0 0 464 622 165488 31 15 54 0 1 0 4876 252268 179180 6930812 0 0 0 132 460 153381 34 15 52 0 2 0 4876 252268 179180 6930964 0 0 0 216 312 141009 31 19 50 0 1 0 4876 252264 179180 6930980 0 0 0 56 275 153143 33 20 47 0 2 0 4876 252212 179180 6931212 0 0 96 296 400 133982 32 18 50 0 1 0 4876 252264 179180 6931332 0 0 0 300 416 136034 32 18 50 0 1 1 4876 252264 179180 6931332 0 0 0 236 377 143300 34 22 44 0 4 0 4876 254876 179180 6931372 0 0 0 124 446 118117 34 20 45 0 1 0 4876 254876 179180 6931492 0 0 16 144 462 140499 38 16 46 0 2 0 4876 255860 179180 6931572 0 0 16 144 674 126250 33 20 47 0 1 0 4876 255860 179180 6931788 0 0 48 264 964 115679 36 13 51 0 3 0 4876 255864 179180 6931804 0 0 0 100 597 127619 36 19 46 0 5 1 4876 255864 179180 6931924 0 0 72 352 559 151620 34 18 48 0 2 0 4876 255860 179184 6932100 0 0 96 120 339 137821 34 20 47 0 0 0 4876 255860 179184 6932156 0 0 8 168 469 125281 36 21 43 0 2 0 4876 256092 179184 6932444 0 0 112 328 446 137939 34 19 48 0 2 0 4876 256092 179184 6932484 0 0 16 184 382 141800 35 16 49 0 3 0 4876 256464 179184 6932716 0 0 16 356 448 134238 30 18 51 0 5 0 4876 256464 179184 6932892 0 0 96 600 476 142838 34 20 46 0 1 0 4876 256464 179184 6933012 0 0 16 176 589 138546 35 22 43 0 2 0 4876 256436 179184 6933096 0 0 60 76 396 93110 42 17 41 0 1 0 4876 256464 179184 6933484 0 0 212 276 442 83060 45 11 44 0 5 0 4876 257612 179184 6933604 0 0 0 472 548 94158 39 17 45 0 0 0 4876 257560 179184 6933708 0 0 96 96 518 116764 38 19 43 0 1 0 4876 257612 179184 6933796 0 0 0 1768 729 139013 29 19 53 0 4 0 4876 257612 179184 6934188 0 0 296 108 332 134703 31 21 48 0 0 1 4876 258584 179184 6934380 0 0 0 492 405 141198 34 18 48 0 1 0 4876 258584 179184 6934492 0 0 0 176 575 134771 37 16 48 0 4 1 4876 257796 179184 6935724 0 0 1176 176 438 151240 33 20 48 0 1 0 4876 261448 179184 6935836 0 0 0 252 489 134348 29 19 51 0 2 0 4876 261448 179184 6935852 0 0 0 512 639 130875 34 16 49 0 2 1 4876 261724 179184 6935924 0 0 0 80 238 144970 33 20 47 0 From pgsql-performance-owner@postgresql.org Thu Apr 15 10:36:58 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 213A9D1D083 for ; Thu, 15 Apr 2004 10:36:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33854-06 for ; Thu, 15 Apr 2004 10:36:59 -0300 (ADT) Received: from internet.potentialtech.com (h-66-167-251-6.phlapafg.covad.net [66.167.251.6]) by svr1.postgresql.org (Postfix) with ESMTP id 5E8FBD1C9C4 for ; Thu, 15 Apr 2004 10:36:54 -0300 (ADT) Received: from potentialtech.com (pa-plum1c-102.pit.adelphia.net [24.53.179.102]) by internet.potentialtech.com (Postfix) with ESMTP id 345E869A71; Thu, 15 Apr 2004 09:36:57 -0400 (EDT) Message-ID: <407E8FF6.4050601@potentialtech.com> Date: Thu, 15 Apr 2004 09:36:54 -0400 From: Bill Moran User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031005 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Rajesh Kumar Mallah Cc: Richard Huxton , Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> <200404150853.32475.dev@archonet.com> <407E7391.4060706@trade-india.com> In-Reply-To: <407E7391.4060706@trade-india.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: 200404/175 X-Sequence-Number: 6475 Rajesh Kumar Mallah wrote: > > Hi, > > The problem was solved by reloading the Table. > the query now takes only 3 seconds. But that is > not a solution. If dropping/recreating the table improves things, then we can reasonably assume that the table is pretty active with updates/inserts. Correct? > The problem is that such phenomenon obscures our > judgement used in optimising queries and database. Lots of phenomenon obscure that ... > If a query runs slow we really cant tell if its a problem > with query itself , hardware or dead rows. > > I already did vacumm full on the table but it still did not > have that effect on performance. > In fact the last figures were after doing a vacuum full. If the data gets too fragmented, a vacuum may not be enough. Also, read up on the recommendations _against_ vacuum full (recommending only using vacuum on databases) With full, vacuum condenses the database, which may actually hurt performance. A regular vacuum just fixes things up, and may leave unused space lying around. However, this should apparently achieve a balance between usage and vacuum. See the docs, they are much better at describing this than I am. > Can there be any more elegent solution to this problem. As a guess, look into CLUSTER (a Postgres SQL command). CLUSTER will basically recreate the table while ordering rows based on an index. (this might benefit you in other ways as well) Don't forget to analyze after cluster. If the problem is caused by frequent updates/inserts, you may find that re-clustering the table on a certain schedule is worthwhile. Be warned, this suggestion is based on an educated guess, I make no guarantees that it will help your problem. Read the docs on cluster and come to your own conclusions. > > Regds > Mallah. > > > > > > Richard Huxton wrote: > >> On Thursday 15 April 2004 08:10, Rajesh Kumar Mallah wrote: >> >> >>> The problem is that i want to know if i need a Hardware upgrade >>> at the moment. >>> >>> Eg i have another table rfis which contains ~ .6 million records. >>> >> >> >> >> >>> SELECT count(*) from rfis where sender_uid > 0; >>> >> >> >> >> >>> Time: 117560.635 ms >>> >>> Which is approximate 4804 records per second. Is it an acceptable >>> performance on the hardware below: >>> >>> RAM: 2 GB >>> DISKS: ultra160 , 10 K , 18 GB >>> Processor: 2* 2.0 Ghz Xeon >>> >> >> Hmm - doesn't seem good, does it? If you run it again, is it much >> faster (since the data should be cached then)? What does "vmstat 10" >> show while you're running the query? >> >> One thing you should have done is read the performance tuning guide at: >> http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php >> The default values are very conservative, and you will need to change >> them. >> >>> What kind of upgrades shoud be put on the server for it to become >>> reasonable fast. >>> >> If you've only got one disk, then a second disk for OS/logging. >> Difficult to say more without knowing numbers of users/activity etc. -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Thu Apr 15 10:40:10 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 528ACD1D08B for ; Thu, 15 Apr 2004 10:40:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 34210-04 for ; Thu, 15 Apr 2004 10:40:10 -0300 (ADT) Received: from tetra.ehpg.net (tetra.ehpg.net [216.218.206.34]) by svr1.postgresql.org (Postfix) with ESMTP id 1AE91D1D085 for ; Thu, 15 Apr 2004 10:40:06 -0300 (ADT) Received: from adsl-68-120-225-71.dsl.irvnca.pacbell.net (ehpg.net) [68.120.225.71]by tetra.ehpg.netwith asmtp(Exim 4.21 #1 (Gentoo Linux 1.4))id 1BE76l-0007lt-96; Thu, 15 Apr 2004 06:40:47 -0700 Message-ID: <407E909F.9000101@ehpg.net> Date: Thu, 15 Apr 2004 06:39:43 -0700 From: "Gavin M. Roy" User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: pginfo Cc: "pgsql-performance@postgresql.org" Subject: Re: linux distro for better pg performance References: <407E25A6.ADFE4815@t1.unisoftbg.com> In-Reply-To: <407E25A6.ADFE4815@t1.unisoftbg.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Antivirus-Scanned: Clean X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/176 X-Sequence-Number: 6476 >I am searching for best pg distro to run pg (7.4.1). > > This is generally based upon opinion. Honestly though, your kernel version is more important for performance than the distro. Personally I use gentoo, love gentoo, and would recommend very few other distros (Slackware) for servers. RedHat and others seem to include kitchensinkd, when it's not needed. >At the moment I am using RedHat AS 3.0, but I think it have some >performance problems (I am not sure). >My configuration: >P4 2.8 GHz >1 GB RAM >120 GB IDE 7200 disk. > > Your IDE drive is the biggest hardward bottleneck here. RPM's and bus transfers are slower than SCSI or SATA. >Kernel version 2.4.21-4.EL (it is the installation vesrion for rh 3.0) . > > Jump to 2.6, it's much better for performance related issues, in my experience. >My problems: > >If I run some query with many reads, I see a massive disk transfer : >procs memory swap io >system cpu > r b swpd free buff cache si so bi bo in cs us sy >id wa > 0 0 0 261724 3252 670748 0 0 0 4 105 19 0 0 >100 0 > 0 0 0 261724 3252 670748 0 0 0 0 101 11 0 0 >100 0 > 0 0 0 261724 3260 670748 0 0 0 4 104 19 0 0 >100 0 > 0 1 0 259684 3268 674112 0 0 964 7 131 57 0 0 >95 4 > 1 0 0 119408 3288 808540 0 0 27960 0 572 630 13 14 >24 49 > 1 1 0 15896 3292 914436 0 0 7984 44744 531 275 11 18 >24 47 > 0 2 0 16292 3296 924996 0 0 4145 6413 384 176 2 >5 0 92 > 0 1 0 19928 3316 928844 0 0 11805 13335 497 388 5 >9 5 81 > 0 3 0 19124 3296 924452 0 0 3153 19164 287 295 5 11 >16 68 > 0 1 0 15956 3304 932984 0 0 536 6812 366 123 4 >6 3 87 > 0 2 0 24956 3300 921416 0 0 1931 22936 > >And if I run top, I see a big iowait % (some times 70-80) and very low >user % (10-15). > > again, this is your harddrive, and the kernel can play into that. >I readet many docs about this problem, but do not find any solution. > >My question: > >If some one is using RH 3.0, pls post some result or suggestions for it >performance with pg . > >What is the best linux distro for pg? > > There's no best, just personal preference. >Can I get better performance by using 15K SCSI disk ? > > Absolutely >Or it will be better to have more RAM (2 or 3 GB) ? > > Better to have a fast drive, but more ram can be helpful. >regards, >ivan. > > > HTH, Gavin From pgsql-performance-owner@postgresql.org Thu Apr 15 13:05:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 92D27D1D8AE for ; Thu, 15 Apr 2004 13:05:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 99399-03 for ; Thu, 15 Apr 2004 13:05:22 -0300 (ADT) Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82]) by svr1.postgresql.org (Postfix) with ESMTP id D48A6D1D391 for ; Thu, 15 Apr 2004 13:05:16 -0300 (ADT) Received: from [206.19.64.3] (account jconway HELO joeconway.com) by joeconway.com (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 1406805; Thu, 15 Apr 2004 08:59:34 -0700 Message-ID: <407EB1E5.4030600@joeconway.com> Date: Thu, 15 Apr 2004 09:01:41 -0700 From: Joe Conway User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Cc: pgsql-performance@postgresql.org Subject: Re: Toooo many context switches (maybe SLES8?) References: <407E7B99.3050306@aeccom.com> In-Reply-To: <407E7B99.3050306@aeccom.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/177 X-Sequence-Number: 6477 Dirk Lutzeb�ck wrote: > postgresql 7.4.1 > a new Dual Xeon MP > too much context switches (way more than 100.000) on higher load (meaning system > load > 2). I believe this was fixed in 7.4.2, although I can't seem to find it in the release notes. Joe From pgsql-performance-owner@postgresql.org Thu Apr 15 13:19:17 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 804BED1E2A7 for ; Thu, 15 Apr 2004 13:19:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04111-04 for ; Thu, 15 Apr 2004 13:19:12 -0300 (ADT) Received: from trade-india.com (ns5.trade-india.com [66.234.10.13]) by svr1.postgresql.org (Postfix) with SMTP id B9774D1E17E for ; Thu, 15 Apr 2004 13:19:07 -0300 (ADT) Received: (qmail 3156 invoked from network); 15 Apr 2004 12:50:07 -0000 Received: from unknown (HELO trade-india.com) (203.145.130.142) by ns5.trade-india.com with SMTP; 15 Apr 2004 12:50:07 -0000 Message-ID: <407EB605.70502@trade-india.com> Date: Thu, 15 Apr 2004 21:49:17 +0530 From: Rajesh Kumar Mallah User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Bill Moran Cc: Richard Huxton , Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> <200404150853.32475.dev@archonet.com> <407E7391.4060706@trade-india.com> <407E8FF6.4050601@potentialtech.com> In-Reply-To: <407E8FF6.4050601@potentialtech.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: 200404/178 X-Sequence-Number: 6478 Bill Moran wrote: > Rajesh Kumar Mallah wrote: > >> >> Hi, >> >> The problem was solved by reloading the Table. >> the query now takes only 3 seconds. But that is >> not a solution. > > > If dropping/recreating the table improves things, then we can reasonably > assume that the table is pretty active with updates/inserts. Correct? Yes the table results from an import process and under goes lots of inserts and updates , but thats before the vacuum full operation. the table is not accessed during vacuum. What i want to know is is there any wat to automate the dumping and reload of a table individually. will the below be safe and effective: begin work; create table new_tab AS select * from tab; truncate table tab; insert into tab select * from new_tab; drop table new_tab; commit; analyze tab; i havenot tried it but plan to do so. but i feel insert would take ages to update the indexes if any. BTW is there any way to disable checks and triggers on a table temporarily while loading data (is updating reltriggers in pg_class safe?) > >> The problem is that such phenomenon obscures our >> judgement used in optimising queries and database. > > > Lots of phenomenon obscure that ... > true. but there should not be too many. >> If a query runs slow we really cant tell if its a problem >> with query itself , hardware or dead rows. >> >> I already did vacumm full on the table but it still did not >> have that effect on performance. >> In fact the last figures were after doing a vacuum full. > > > If the data gets too fragmented, a vacuum may not be enough. Also, read > up on the recommendations _against_ vacuum full (recommending only using > vacuum on databases) With full, vacuum condenses the database, which may > actually hurt performance. A regular vacuum just fixes things up, and > may leave unused space lying around. However, this should apparently > achieve a balance between usage and vacuum. See the docs, they are much > better at describing this than I am. > i understand simultaneous vacuum and usage detoriates performance mostly. but this case is different. >> Can there be any more elegent solution to this problem. > > > As a guess, look into CLUSTER (a Postgres SQL command). CLUSTER will > basically recreate the table while ordering rows based on an index. > (this might benefit you in other ways as well) Don't forget to analyze > after cluster. If the problem is caused by frequent updates/inserts, > you may find that re-clustering the table on a certain schedule is > worthwhile. i could consider that option also. > > Be warned, this suggestion is based on an educated guess, I make no > guarantees that it will help your problem. Read the docs on cluster > and come to your own conclusions. Thanks . Regds mallah. > >> >> Regds >> Mallah. >> >> >> >> >> >> Richard Huxton wrote: >> >>> On Thursday 15 April 2004 08:10, Rajesh Kumar Mallah wrote: >>> >>> >>>> The problem is that i want to know if i need a Hardware upgrade >>>> at the moment. >>>> >>>> Eg i have another table rfis which contains ~ .6 million records. >>>> >>> >>> >>> >>> >>> >>>> SELECT count(*) from rfis where sender_uid > 0; >>>> >>> >>> >>> >>> >>> >>>> Time: 117560.635 ms >>>> >>>> Which is approximate 4804 records per second. Is it an acceptable >>>> performance on the hardware below: >>>> >>>> RAM: 2 GB >>>> DISKS: ultra160 , 10 K , 18 GB >>>> Processor: 2* 2.0 Ghz Xeon >>>> >>> >>> >>> Hmm - doesn't seem good, does it? If you run it again, is it much >>> faster (since the data should be cached then)? What does "vmstat 10" >>> show while you're running the query? >>> >>> One thing you should have done is read the performance tuning guide at: >>> http://www.varlena.com/varlena/GeneralBits/Tidbits/index.php >>> The default values are very conservative, and you will need to >>> change them. >>> >>>> What kind of upgrades shoud be put on the server for it to become >>>> reasonable fast. >>>> >>> >>> If you've only got one disk, then a second disk for OS/logging. >>> Difficult to say more without knowing numbers of users/activity etc. >> > > > From pgsql-performance-owner@postgresql.org Thu Apr 15 13:29:38 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EAE09D1E939 for ; Thu, 15 Apr 2004 13:29:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 10683-07 for ; Thu, 15 Apr 2004 13:29:33 -0300 (ADT) Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id E49B3D1E96B for ; Thu, 15 Apr 2004 13:29:26 -0300 (ADT) Received: from aeccom.com (cayambe.core.aeccom.com [192.168.2.12]) by mail2.aeccom.com (Postfix) with ESMTP id F2D1334; Thu, 15 Apr 2004 18:29:31 +0200 (CEST) Message-ID: <407EB86B.6080801@aeccom.com> Date: Thu, 15 Apr 2004 18:29:31 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH, Berlin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031016 X-Accept-Language: de, en MIME-Version: 1.0 To: Joe Conway Cc: pgsql-performance@postgresql.org Subject: Re: Toooo many context switches (maybe SLES8?) References: <407E7B99.3050306@aeccom.com> <407EB1E5.4030600@joeconway.com> In-Reply-To: <407EB1E5.4030600@joeconway.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/179 X-Sequence-Number: 6479 Joe, do you know where I should look in the 7.4.2 code to find this out? Dirk Joe Conway wrote: > Dirk Lutzeb�ck wrote: > >> postgresql 7.4.1 > >> a new Dual Xeon MP > >> too much context switches (way more than 100.000) on higher load >> (meaning system load > 2). > > > I believe this was fixed in 7.4.2, although I can't seem to find it in > the release notes. > > Joe > > From pgsql-performance-owner@postgresql.org Thu Apr 15 13:42:28 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7CE31D1E94D for ; Thu, 15 Apr 2004 13:42:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 10797-10 for ; Thu, 15 Apr 2004 13:42:28 -0300 (ADT) Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82]) by svr1.postgresql.org (Postfix) with ESMTP id CE411D1E947 for ; Thu, 15 Apr 2004 13:42:22 -0300 (ADT) Received: from [206.19.64.3] (account jconway HELO joeconway.com) by joeconway.com (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 1407858; Thu, 15 Apr 2004 09:36:40 -0700 Message-ID: <407EBA89.6080007@joeconway.com> Date: Thu, 15 Apr 2004 09:38:33 -0700 From: Joe Conway User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Cc: pgsql-performance@postgresql.org Subject: Re: Toooo many context switches (maybe SLES8?) References: <407E7B99.3050306@aeccom.com> <407EB1E5.4030600@joeconway.com> <407EB86B.6080801@aeccom.com> In-Reply-To: <407EB86B.6080801@aeccom.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/180 X-Sequence-Number: 6480 Dirk Lutzeb�ck wrote: > Joe, do you know where I should look in the 7.4.2 code to find this out? I think I was wrong. I just looked in CVS and found the commit I was thinking about: http://developer.postgresql.org/cvsweb.cgi/pgsql-server/src/backend/storage/lmgr/s_lock.c.diff?r1=1.22&r2=1.23 http://developer.postgresql.org/cvsweb.cgi/pgsql-server/src/include/storage/s_lock.h.diff?r1=1.123&r2=1.124 ========================= Revision 1.23 / (download) - [select for diffs] , Sat Dec 27 20:58:58 2003 UTC (3 months, 2 weeks ago) by tgl Changes since 1.22: +5 -1 lines Diff to previous 1.22 Improve spinlock code for recent x86 processors: insert a PAUSE instruction in the s_lock() wait loop, and use test before test-and-set in TAS() macro to avoid unnecessary bus traffic. Patch from Manfred Spraul, reworked a bit by Tom. ========================= I thought this had been committed to the 7.4 stable branch as well, but it appears not. Joe From pgsql-performance-owner@postgresql.org Thu Apr 15 14:17:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0558AD1E2A7 for ; Thu, 15 Apr 2004 14:17:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 26735-07 for ; Thu, 15 Apr 2004 14:17:12 -0300 (ADT) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id 7530ED1E1F3 for ; Thu, 15 Apr 2004 14:17:06 -0300 (ADT) Received: (qmail 15299 invoked by uid 500); 15 Apr 2004 17:20:59 -0000 Date: Thu, 15 Apr 2004 12:20:59 -0500 From: Bruno Wolff III To: Simon Riggs Cc: Josh Berkus , pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. Message-ID: <20040415172059.GA15119@wolff.to> Mail-Followup-To: Simon Riggs , Josh Berkus , pgsql-performance@postgresql.org References: <200404050836.52115.josh@agliodbs.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/181 X-Sequence-Number: 6481 On Wed, Apr 14, 2004 at 21:12:18 +0100, Simon Riggs wrote: > > I guess what I'm saying is it's not how many people you've got working > on the optimizer, its how many accurate field reports of less-than > perfect optimization reach them. In that case, PostgreSQL is likely in a > better position than Microsoft, since the accessibility of the pg > discussion lists makes such cases much more likely to get aired. > > Any thoughts? I have seen exactly this happen a number of times over the last several years. However there is still only one Tom Lane implementing the improvements. From pgsql-performance-owner@postgresql.org Thu Apr 15 14:39:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A38C9D1D6CC for ; Thu, 15 Apr 2004 14:39:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31037-10 for ; Thu, 15 Apr 2004 14:39:42 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 17B6AD1D2F7 for ; Thu, 15 Apr 2004 14:39:37 -0300 (ADT) Received: from [64.81.245.111] (HELO 192.168.1.102) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4862483; Thu, 15 Apr 2004 10:41:04 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Joe Conway , Dirk =?iso-8859-1?q?Lutzeb=E4ck?= Subject: Re: Toooo many context switches (maybe SLES8?) Date: Thu, 15 Apr 2004 10:40:01 -0700 User-Agent: KMail/1.5.4 Cc: pgsql-performance@postgresql.org References: <407E7B99.3050306@aeccom.com> <407EB1E5.4030600@joeconway.com> In-Reply-To: <407EB1E5.4030600@joeconway.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404151040.01160.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/182 X-Sequence-Number: 6482 Joe, > I believe this was fixed in 7.4.2, although I can't seem to find it in > the release notes. Depends on the cause of the issue. If it's the same issue that I'm currently struggling with, it's not fixed. -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Apr 15 14:44:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5F1CED1DB96 for ; Thu, 15 Apr 2004 14:44:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37464-04 for ; Thu, 15 Apr 2004 14:44:36 -0300 (ADT) 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 17CA6D1D8AE for ; Thu, 15 Apr 2004 14:44:31 -0300 (ADT) 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 1BEAuh-000Mo9-0U; Thu, 15 Apr 2004 18:44:36 +0100 Received: from client17.archonet.com (client17.archonet.com [192.168.1.17]) by mainbox.archonet.com (Postfix) with ESMTP id 0B9D516BFB; Thu, 15 Apr 2004 18:44:35 +0100 (BST) From: Richard Huxton To: Rajesh Kumar Mallah , Bill Moran Subject: Re: [ SOLVED ] select count(*) very slow on an already Date: Thu, 15 Apr 2004 18:44:33 +0100 User-Agent: KMail/1.5 Cc: Postgres Performance References: <407D7A89.60508@trade-india.com> <407E8FF6.4050601@potentialtech.com> <407EB605.70502@trade-india.com> In-Reply-To: <407EB605.70502@trade-india.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404151844.33672.dev@archonet.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/183 X-Sequence-Number: 6483 On Thursday 15 April 2004 17:19, Rajesh Kumar Mallah wrote: > Bill Moran wrote: > > Rajesh Kumar Mallah wrote: > >> Hi, > >> > >> The problem was solved by reloading the Table. > >> the query now takes only 3 seconds. But that is > >> not a solution. > > > > If dropping/recreating the table improves things, then we can reasonably > > assume that the table is pretty active with updates/inserts. Correct? > > Yes the table results from an import process and under goes lots > of inserts and updates , but thats before the vacuum full operation. > the table is not accessed during vacuum. What i want to know is > is there any wat to automate the dumping and reload of a table > individually. will the below be safe and effective: Shouldn't be necessary assuming you vacuum (not full) regularly. However, looking back at your original posting, the vacuum output doesn't seem to show any rows that need removing. # VACUUM full verbose eyp_rfi; INFO: vacuuming "public.eyp_rfi" INFO: "eyp_rfi": found 0 removable, 505960 nonremovable row versions in 71987 pages DETAIL: 0 dead row versions cannot be removed yet. Since your select count(*) showed 505960 rows, I can't see how dropping/replacing could make a difference on a sequential scan. Since we're not using any indexes I don't see how it could be related to that. > begin work; > create table new_tab AS select * from tab; > truncate table tab; > insert into tab select * from new_tab; > drop table new_tab; > commit; > analyze tab; > > i havenot tried it but plan to do so. > but i feel insert would take ages to update > the indexes if any. It will have to update them, which will take time. > BTW > > is there any way to disable checks and triggers on > a table temporarily while loading data (is updating > reltriggers in pg_class safe?) You can take a look at pg_restore and copy how it does it. -- Richard Huxton Archonet Ltd From pgsql-performance-owner@postgresql.org Tue Apr 20 10:58:15 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2EB0DD1D083 for ; Thu, 15 Apr 2004 14:48:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37464-05 for ; Thu, 15 Apr 2004 14:48:17 -0300 (ADT) Received: from ctg-msnexc01.staff.berbee.com (msn-office1.binc.net [64.73.12.254]) by svr1.postgresql.org (Postfix) with ESMTP id D46C1D1D07E for ; Thu, 15 Apr 2004 14:48:11 -0300 (ADT) Received: from localhost ([172.30.254.220] RDNS failed) by ctg-msnexc01.staff.berbee.com with Microsoft SMTPSVC(6.0.3790.0); Thu, 15 Apr 2004 12:48:16 -0500 From: "Jeremy M. Guthrie" Reply-To: jeremy.guthrie@berbee.com Organization: Berbee Information Networks To: pgsql-performance@postgresql.org Subject: Any way to 'analyze' indexes to get updated sizes? Date: Thu, 15 Apr 2004 12:48:14 -0500 User-Agent: KMail/1.6.1 MIME-Version: 1.0 Content-Disposition: inline Content-Type: Text/Plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Message-Id: <200404151248.16067.jeremy.guthrie@berbee.com> X-OriginalArrivalTime: 15 Apr 2004 17:48:16.0425 (UTC) FILETIME=[D4C40990:01C42311] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/297 X-Sequence-Number: 6597 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Is there a way to analyze indexes to provide updated sizes? Is a vacuum th= e=20 only way to determine the size of an index? Analyze updates the stats so I= =20 can see table space sizes but I cannot find an alternative to vacuum for=20 indexes. - --=20 - -------------------------------------------------- Jeremy M. Guthrie jeremy.guthrie@berbee.com Network Engineer Phone: 608-298-1061 Berbee Fax: 608-288-3007 5520 Research Park Drive NOC: 608-298-1102 Madison, WI 53711 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (GNU/Linux) iD8DBQFAfsreqtjaBHGZBeURAm3+AJ9F34SESTf8i/oEuKvKfXoh+NcOxwCcDcM9 HP5LHM3Qidb4wa2/rW5H0cI=3D =3DmJCz -----END PGP SIGNATURE----- From pgsql-performance-owner@postgresql.org Thu Apr 15 15:28:28 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 85393D1D2F7 for ; Thu, 15 Apr 2004 15:28:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52231-08 for ; Thu, 15 Apr 2004 15:28:27 -0300 (ADT) Received: from mx1.neopolitan.us (mx1.neopolitan.us [65.87.16.224]) by svr1.postgresql.org (Postfix) with ESMTP id 4D85AD1E97A for ; Thu, 15 Apr 2004 15:28:20 -0300 (ADT) Received: from [65.87.16.98] (HELO [10.0.0.210]) by mx1.neopolitan.us (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 3945316 for pgsql-performance@postgresql.org; Thu, 15 Apr 2004 11:28:26 -0700 Subject: Re: linux distro for better pg performance From: "J. Andrew Rogers" To: "pgsql-performance@postgresql.org" In-Reply-To: <407E909F.9000101@ehpg.net> References: <407E25A6.ADFE4815@t1.unisoftbg.com> <407E909F.9000101@ehpg.net> Content-Type: text/plain Organization: Message-Id: <1082053706.10823.38.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 15 Apr 2004 11:28:26 -0700 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/184 X-Sequence-Number: 6484 On Thu, 2004-04-15 at 06:39, Gavin M. Roy wrote: > Your IDE drive is the biggest hardward bottleneck here. RPM's and bus > transfers are slower than SCSI or SATA. Individual disk throughput generally has very little bearing on database performance compared to other factors. In fact, IDE bandwidth performance is perfectly adequate for databases, and for database purposes indistinguishable from SATA. I would say that average access and read/write completion times, especially under load, are by far the most limiting factors, and disk RPM is only one component of this. In fact, disk RPM is a very expensive way to get marginally better throughput in this regard, and I would suggest 10k rather than 15k drives for the money. There are really only two features that are worth buying in your disk subsystem which many people ignore: TCQ and independently managed I/O with a large battery-backed write-back cache. Currently, the only place to really get this is with SCSI RAID. You can get 10k SATA drives, so when you are buying SCSI you are really buying these features. Do these features make a difference? Far more than you would imagine. On one postgres server I just upgraded, we went from a 3Ware 8x7200-RPM RAID-10 configuration to an LSI 320-2 SCSI 3x10k RAID-5, with 256M cache, and got a 3-5x performance improvement in the disk subsystem under full database load. SCSI RAID can service a lot of I/O requests far more efficiently than current IDE/SATA RAID controllers, and it shows in the stats. Under these types of loads, the actually bandwidth utilized by the disks doesn't come anywhere close to even their rated performance, never mind the theoretical performance of the bus. Service times for IDE/SATA RAID increases dramatically under load, whereas SCSI tends not to under the same load. Considering that very good SCSI RAID controllers (e.g. the LSI 320-2 that I mention above) are only marginally more expensive than nominally equivalent IDE/SATA controller solutions, using SCSI RAID with 10k drives is pretty much the price-performance sweet spot if you use your disk system hard (like we do). For databases with low disk I/O intensity, stay with IDE/SATA and save a little money. For databases that have high disk I/O intensity, use SCSI. The price premium for SCSI is about 50%, but the performance difference is an integer factor under load. j. andrew rogers From pgsql-performance-owner@postgresql.org Thu Apr 15 15:30:06 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BE976D1E981 for ; Thu, 15 Apr 2004 15:30:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50970-09 for ; Thu, 15 Apr 2004 15:30:02 -0300 (ADT) Received: from secure.icanx.com (secure.icanx.com [64.246.58.55]) by svr1.postgresql.org (Postfix) with ESMTP id EB726D1E971 for ; Thu, 15 Apr 2004 15:29:55 -0300 (ADT) Received: from 63-229-201-197.mpls.qwest.net ([63.229.201.197] helo=[192.168.2.102]) by secure.icanx.com with esmtp (Exim 4.24) id 1BEBce-0003Dq-Nl for pgsql-performance@postgresql.org; Thu, 15 Apr 2004 14:30:00 -0400 Mime-Version: 1.0 (Apple Message framework v613) In-Reply-To: <200404151844.33672.dev@archonet.com> References: <407D7A89.60508@trade-india.com> <407E8FF6.4050601@potentialtech.com> <407EB605.70502@trade-india.com> <200404151844.33672.dev@archonet.com> Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Mark Lubratt Subject: Re: [ SOLVED ] select count(*) very slow on an already Date: Thu, 15 Apr 2004 13:29:57 -0500 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.613) X-AntiAbuse: This header was added to track abuse, please include it with any abuse report X-AntiAbuse: Primary Hostname - secure.icanx.com X-AntiAbuse: Original Domain - postgresql.org X-AntiAbuse: Originator/Caller UID/GID - [0 0] / [47 12] X-AntiAbuse: Sender Address Domain - indeq.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/185 X-Sequence-Number: 6485 On Apr 15, 2004, at 12:44 PM, Richard Huxton wrote: > On Thursday 15 April 2004 17:19, Rajesh Kumar Mallah wrote: >> Bill Moran wrote: > >> BTW >> >> is there any way to disable checks and triggers on >> a table temporarily while loading data (is updating >> reltriggers in pg_class safe?) > > You can take a look at pg_restore and copy how it does it. > > Does SET CONSTRAINT take care of checks within the transaction? Triggers would be a different matter... Mark From pgsql-performance-owner@postgresql.org Thu Apr 15 15:48:43 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 06C5DD1CCDA for ; Thu, 15 Apr 2004 15:48:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 60462-10 for ; Thu, 15 Apr 2004 15:48:44 -0300 (ADT) Received: from internet.potentialtech.com (h-66-167-251-6.phlapafg.covad.net [66.167.251.6]) by svr1.postgresql.org (Postfix) with ESMTP id 31662D1B4BE for ; Thu, 15 Apr 2004 15:48:39 -0300 (ADT) Received: from potentialtech.com (pa-plum1c-102.pit.adelphia.net [24.53.179.102]) by internet.potentialtech.com (Postfix) with ESMTP id 4CCEC69A71; Thu, 15 Apr 2004 14:48:45 -0400 (EDT) Message-ID: <407ED909.1020404@potentialtech.com> Date: Thu, 15 Apr 2004 14:48:41 -0400 From: Bill Moran User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031005 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Rajesh Kumar Mallah Cc: Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already References: <407D7A89.60508@trade-india.com> <200404142008.58533.dev@archonet.com> <407E3563.6000408@trade-india.com> <200404150853.32475.dev@archonet.com> <407E7391.4060706@trade-india.com> <407E8FF6.4050601@potentialtech.com> <407EB605.70502@trade-india.com> In-Reply-To: <407EB605.70502@trade-india.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: 200404/186 X-Sequence-Number: 6486 Rajesh Kumar Mallah wrote: > Bill Moran wrote: > >> Rajesh Kumar Mallah wrote: >> >>> Hi, >>> >>> The problem was solved by reloading the Table. >>> the query now takes only 3 seconds. But that is >>> not a solution. >> >> If dropping/recreating the table improves things, then we can reasonably >> assume that the table is pretty active with updates/inserts. Correct? > > Yes the table results from an import process and under goes lots > of inserts and updates , but thats before the vacuum full operation. > the table is not accessed during vacuum. What i want to know is > is there any wat to automate the dumping and reload of a table > individually. will the below be safe and effective: The CLUSTER command I described is one way of doing this. It essentially automates the task of copying the table, dropping the old one, and recreating it. >> If the data gets too fragmented, a vacuum may not be enough. Also, read >> up on the recommendations _against_ vacuum full (recommending only using >> vacuum on databases) With full, vacuum condenses the database, which may >> actually hurt performance. A regular vacuum just fixes things up, and >> may leave unused space lying around. However, this should apparently >> achieve a balance between usage and vacuum. See the docs, they are much >> better at describing this than I am. >> > i understand simultaneous vacuum and usage detoriates performance mostly. > but this case is different. Just want to make sure we're on the same page here. I'm not talking about vacuuming simultaneous with anything. I'm simply saying that "vacuum full" isn't always the best choice. You should probably only be doing "vacuum". The reason and details for this are in the admin docs. -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Thu Apr 15 16:24:43 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 850BBD1D2F7 for ; Thu, 15 Apr 2004 16:24:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77246-04 for ; Thu, 15 Apr 2004 16:24:44 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id C0E69D1B4BE for ; Thu, 15 Apr 2004 16:24:32 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id <2A89W2DH>; Thu, 15 Apr 2004 15:24:32 -0400 Message-ID: From: "Shea,Dan [CIS]" To: 'Bill Moran' Cc: Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already Date: Thu, 15 Apr 2004 15:24:32 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/187 X-Sequence-Number: 6487 Bill, if you had alot of updates and deletions and wanted to optimize your table, can you just issue the cluster command. Will the cluster command rewrite the table without the obsolete data that a vacuum flags or do you need to issue a vacuum first? Dan. -----Original Message----- From: Bill Moran [mailto:wmoran@potentialtech.com] Sent: Thursday, April 15, 2004 2:49 PM To: Rajesh Kumar Mallah Cc: Postgres Performance Subject: Re: [PERFORM] [ SOLVED ] select count(*) very slow on an already Rajesh Kumar Mallah wrote: > Bill Moran wrote: > >> Rajesh Kumar Mallah wrote: >> >>> Hi, >>> >>> The problem was solved by reloading the Table. >>> the query now takes only 3 seconds. But that is >>> not a solution. >> >> If dropping/recreating the table improves things, then we can reasonably >> assume that the table is pretty active with updates/inserts. Correct? > > Yes the table results from an import process and under goes lots > of inserts and updates , but thats before the vacuum full operation. > the table is not accessed during vacuum. What i want to know is > is there any wat to automate the dumping and reload of a table > individually. will the below be safe and effective: The CLUSTER command I described is one way of doing this. It essentially automates the task of copying the table, dropping the old one, and recreating it. >> If the data gets too fragmented, a vacuum may not be enough. Also, read >> up on the recommendations _against_ vacuum full (recommending only using >> vacuum on databases) With full, vacuum condenses the database, which may >> actually hurt performance. A regular vacuum just fixes things up, and >> may leave unused space lying around. However, this should apparently >> achieve a balance between usage and vacuum. See the docs, they are much >> better at describing this than I am. >> > i understand simultaneous vacuum and usage detoriates performance mostly. > but this case is different. Just want to make sure we're on the same page here. I'm not talking about vacuuming simultaneous with anything. I'm simply saying that "vacuum full" isn't always the best choice. You should probably only be doing "vacuum". The reason and details for this are in the admin docs. -- Bill Moran Potential Technologies http://www.potentialtech.com ---------------------------(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 Thu Apr 15 16:35:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BEBB3D1E99F for ; Thu, 15 Apr 2004 16:35:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77165-09 for ; Thu, 15 Apr 2004 16:35:25 -0300 (ADT) Received: from cmailg2.svr.pol.co.uk (cmailg2.svr.pol.co.uk [195.92.195.172]) by svr1.postgresql.org (Postfix) with ESMTP id 4BA61D1E9A2 for ; Thu, 15 Apr 2004 16:35:19 -0300 (ADT) Received: from modem-2043.lion.dialup.pol.co.uk ([217.135.167.251] helo=stromboli) by cmailg2.svr.pol.co.uk with esmtp (Exim 4.14) id 1BECdq-0006VF-LH; Thu, 15 Apr 2004 20:35:18 +0100 From: "Simon Riggs" To: "Bruno Wolff III" Cc: "Josh Berkus" , Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Thu, 15 Apr 2004 20:35:34 +0100 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) In-reply-to: <20040415172059.GA15119@wolff.to> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/188 X-Sequence-Number: 6488 > Bruno Wolff > Simon Riggs wrote: > > > > I guess what I'm saying is it's not how many people you've > got working > > on the optimizer, its how many accurate field reports of less-than > > perfect optimization reach them. In that case, PostgreSQL > is likely in a > > better position than Microsoft, since the accessibility of the pg > > discussion lists makes such cases much more likely to get aired. > > > > Any thoughts? > > I have seen exactly this happen a number of times over the > last several > years. However there is still only one Tom Lane implementing the > improvements. > ...and very few Mr.Microsofts too. [I'm uncomfortable with, and it was not my intent, to discuss such an issue with direct reference to particular individuals. There is no intent to critiscise or malign anybody named] Regards, Simon From pgsql-performance-owner@postgresql.org Thu Apr 15 16:38:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D8CBFD1D085 for ; Thu, 15 Apr 2004 16:38:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77321-07 for ; Thu, 15 Apr 2004 16:38:05 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 24945D1D06D for ; Thu, 15 Apr 2004 16:37:59 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3FJbL3C020499; Thu, 15 Apr 2004 15:37:21 -0400 (EDT) To: Joe Conway Cc: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= , pgsql-performance@postgresql.org Subject: Re: Toooo many context switches (maybe SLES8?) In-reply-to: <407EBA89.6080007@joeconway.com> References: <407E7B99.3050306@aeccom.com> <407EB1E5.4030600@joeconway.com> <407EB86B.6080801@aeccom.com> <407EBA89.6080007@joeconway.com> Comments: In-reply-to Joe Conway message dated "Thu, 15 Apr 2004 09:38:33 -0700" Date: Thu, 15 Apr 2004 15:37:21 -0400 Message-ID: <20498.1082057841@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/189 X-Sequence-Number: 6489 Joe Conway writes: >> Improve spinlock code for recent x86 processors: insert a PAUSE >> instruction in the s_lock() wait loop, and use test before test-and-set >> in TAS() macro to avoid unnecessary bus traffic. Patch from Manfred >> Spraul, reworked a bit by Tom. > I thought this had been committed to the 7.4 stable branch as well, but > it appears not. I am currently chasing what seems to be the same issue: massive context swapping on a dual Xeon system. I tried back-patching the above-mentioned patch ... it helps a little but by no means solves the problem ... regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 20 10:58:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3914BD1D25B for ; Thu, 15 Apr 2004 17:04:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88416-09 for ; Thu, 15 Apr 2004 17:04:30 -0300 (ADT) Received: from mailout11.sul.t-online.com (mailout11.sul.t-online.com [194.25.134.85]) by svr1.postgresql.org (Postfix) with ESMTP id 03E24D1D083 for ; Thu, 15 Apr 2004 17:04:24 -0300 (ADT) Received: from fwd03.aul.t-online.de by mailout11.sul.t-online.com with smtp id 1BED5M-00010G-06; Thu, 15 Apr 2004 22:03:44 +0200 Received: from aeccom.com (ZknwbMZrZe57WMjZzTNOinFhX4rdoiy+yeQm4pJ8W4cxP3xVE-aZ8n@[217.231.158.140]) by fwd03.sul.t-online.com with esmtp id 1BED5I-1IbFnE0; Thu, 15 Apr 2004 22:03:40 +0200 Message-ID: <407EEA9B.7010807@aeccom.com> Date: Thu, 15 Apr 2004 22:03:39 +0200 From: Dirk.Lutzebaeck@t-online.de (Dirk Lutzebaeck) Reply-To: lutzeb@aeccom.com 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: Tom Lane Cc: Joe Conway , pgsql-performance@postgresql.org Subject: Re: Toooo many context switches (maybe SLES8?) References: <407E7B99.3050306@aeccom.com> <407EB1E5.4030600@joeconway.com> <407EB86B.6080801@aeccom.com> <407EBA89.6080007@joeconway.com> <20498.1082057841@sss.pgh.pa.us> In-Reply-To: <20498.1082057841@sss.pgh.pa.us> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Seen: false X-ID: ZknwbMZrZe57WMjZzTNOinFhX4rdoiy+yeQm4pJ8W4cxP3xVE-aZ8n X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/298 X-Sequence-Number: 6598 Could this be related to the O(1) scheduler backpatches from 2.6 to 2.4 kernel on newer 2.4er distros (RedHat, SuSE)? Tom Lane wrote: >Joe Conway writes: > > >>>Improve spinlock code for recent x86 processors: insert a PAUSE >>>instruction in the s_lock() wait loop, and use test before test-and-set >>>in TAS() macro to avoid unnecessary bus traffic. Patch from Manfred >>>Spraul, reworked a bit by Tom. >>> >>> > > > >>I thought this had been committed to the 7.4 stable branch as well, but >>it appears not. >> >> > >I am currently chasing what seems to be the same issue: massive context >swapping on a dual Xeon system. I tried back-patching the above-mentioned >patch ... it helps a little but by no means solves the problem ... > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Thu Apr 15 17:13:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 360FED1D062 for ; Thu, 15 Apr 2004 17:13:28 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92020-08 for ; Thu, 15 Apr 2004 17:13:32 -0300 (ADT) Received: from internet.potentialtech.com (h-66-167-251-6.phlapafg.covad.net [66.167.251.6]) by svr1.postgresql.org (Postfix) with ESMTP id 83B0BD1D07E for ; Thu, 15 Apr 2004 17:13:26 -0300 (ADT) Received: from potentialtech.com (pa-plum1c-102.pit.adelphia.net [24.53.179.102]) by internet.potentialtech.com (Postfix) with ESMTP id 655C569A71; Thu, 15 Apr 2004 16:13:33 -0400 (EDT) Message-ID: <407EECEA.7010500@potentialtech.com> Date: Thu, 15 Apr 2004 16:13:30 -0400 From: Bill Moran User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031005 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Shea,Dan [CIS]" Cc: Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already 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: 200404/190 X-Sequence-Number: 6490 Shea,Dan [CIS] wrote: > Bill, if you had alot of updates and deletions and wanted to optimize your > table, can you just issue the cluster command. > Will the cluster command rewrite the table without the obsolete data that a > vacuum flags or do you need to issue a vacuum first? From the reference docs: "During the cluster operation, a temporary copy of the table is created that contains the table data in the index order. Temporary copies of each index on the table are created as well. Therefore, you need free space on disk at least equal to the sum of the table size and the index sizes. "CLUSTER preserves GRANT, inheritance, index, foreign key, and other ancillary information about the table. "Because the optimizer records statistics about the ordering of tables, it is advisable to run ANALYZE on the newly clustered table. Otherwise, the optimizer may make poor choices of query plans." The primary reason CLUSTER exists is to allow you to physically reorder a table based on a key. This should provide a performance improvement if data with the same key is accessed all at once. (i.e. if you do "SELECT * FROM table WHERE key=5" and it returns 100 rows, those 100 rows are guaranteed to be all on the same part of the disk after CLUSTER, thus a performance improvement should result.) Updates and inserts will add data in the next available space in a table with no regard for any keys, and _may_ require running all over the disk to retrieve the data in the previous example query. I doubt if CLUSTER is an end-all optimization tool. The specific reason I suggested it was because the original poster was asking for an easier way to drop/recreate a table (as prior experimentation had shown this to improve performance) I can't think of anything easier than "CLUSTER ON " Since CLUSTER recreates the table, it implicitly removes the dead tuples. However, it's going to be a LOT slower than vacuum, so if dead tuples are the main problem, vacuum is still the way to go. -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Thu Apr 15 17:37:11 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CF02FD1D08C for ; Thu, 15 Apr 2004 17:37:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 96731-10 for ; Thu, 15 Apr 2004 17:37:14 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id A4BB7D1C9BE for ; Thu, 15 Apr 2004 17:37:07 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4863796; Thu, 15 Apr 2004 13:38:36 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Tom Lane , Joe Conway Subject: Re: Toooo many context switches (maybe SLES8?) Date: Thu, 15 Apr 2004 13:37:00 -0700 User-Agent: KMail/1.4.3 Cc: Dirk =?iso-8859-1?q?Lutzeb=E4ck?= , pgsql-performance@postgresql.org References: <407E7B99.3050306@aeccom.com> <407EBA89.6080007@joeconway.com> <20498.1082057841@sss.pgh.pa.us> In-Reply-To: <20498.1082057841@sss.pgh.pa.us> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404151337.00273.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/191 X-Sequence-Number: 6491 Folks, > I am currently chasing what seems to be the same issue: massive context > swapping on a dual Xeon system. I tried back-patching the above-mentioned > patch ... it helps a little but by no means solves the problem ... BTW, I'm currently pursuing the possibility that this has something to do with the ServerWorks chipset on those motherboards. If anyone knows a high-end hardware+linux kernel geek I can corner, I'd appreciate it. Maybe I should contact OSDL ... -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Apr 15 17:40:01 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 96785D1E812 for ; Thu, 15 Apr 2004 17:39:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 07239-03 for ; Thu, 15 Apr 2004 17:39:59 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 74D12D1E1F3 for ; Thu, 15 Apr 2004 17:39:52 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4863804; Thu, 15 Apr 2004 13:41:22 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: "Simon Riggs" , Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Thu, 15 Apr 2004 13:39:45 -0700 User-Agent: KMail/1.4.3 References: In-Reply-To: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404151339.45615.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/192 X-Sequence-Number: 6492 Simon, > Is the problem "a person interested" or is there another issue there? IMHO, it's "a person interested". > Treating the optimizer as a black box is something I'm very used to from > other RDBMS. My question is, how can you explicitly re-write a query now > to "improve" it? If there's no way of manipulating queries without > actually re-writing the optimizer, we're now in a position where we > aren't able to diagnose when the optimizer isn't working effectively. Well, there is ... all of the various query cost parameters. > For my mind, all the people on this list are potential "optimizer > developers" in the sense that we can all look at queries and see whether > there is a problem with particular join plans. Providing good cases of > poor optimization is just what's needed to assist those few that do > understand the internals to continue improving things. ... which is what this list is for. But, ultimately, improvements on the planner are still bottlenecked by having only one developer actually hacking the changes. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Thu Apr 15 18:38:52 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9ABB5D1B515 for ; Thu, 15 Apr 2004 18:38:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 22561-05 for ; Thu, 15 Apr 2004 18:38:56 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id AF9D9D1B4BE for ; Thu, 15 Apr 2004 18:38:49 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 41D7717C465; Thu, 15 Apr 2004 17:38:24 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BEEYy-0002cV-00; Thu, 15 Apr 2004 17:38:24 -0400 To: Bruno Wolff III Cc: Simon Riggs , Josh Berkus , pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. References: <200404050836.52115.josh@agliodbs.com> <20040415172059.GA15119@wolff.to> In-Reply-To: <20040415172059.GA15119@wolff.to> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 15 Apr 2004 17:38:23 -0400 Message-ID: <873c74c46o.fsf@stark.xeocode.com> Lines: 11 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: 200404/193 X-Sequence-Number: 6493 Bruno Wolff III writes: > I have seen exactly this happen a number of times over the last several > years. However there is still only one Tom Lane implementing the > improvements. Ob: Well clearly the problem is we need more Tom Lanes. -- greg From pgsql-performance-owner@postgresql.org Thu Apr 15 19:21:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1C037D1CCB0 for ; Thu, 15 Apr 2004 19:21:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33709-09 for ; Thu, 15 Apr 2004 19:21:12 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 17146D1C9C8 for ; Thu, 15 Apr 2004 19:21:05 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3FMLBaA021757; Thu, 15 Apr 2004 18:21:11 -0400 (EDT) To: Greg Stark Cc: pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. In-reply-to: <873c74c46o.fsf@stark.xeocode.com> References: <200404050836.52115.josh@agliodbs.com> <20040415172059.GA15119@wolff.to> <873c74c46o.fsf@stark.xeocode.com> Comments: In-reply-to Greg Stark message dated "15 Apr 2004 17:38:23 -0400" Date: Thu, 15 Apr 2004 18:21:11 -0400 Message-ID: <21756.1082067671@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/195 X-Sequence-Number: 6495 Greg Stark writes: > Ob: Well clearly the problem is we need more Tom Lanes. ObHHGReference: "Haven't you heard? I come in six-packs!" regards, tom lane From pgsql-performance-owner@postgresql.org Thu Apr 15 19:19:01 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D67A1D1D083 for ; Thu, 15 Apr 2004 19:19:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37004-03 for ; Thu, 15 Apr 2004 19:19:06 -0300 (ADT) Received: from 3times25.net (66-23-202-190.clients.speedfactory.net [66.23.202.190]) by svr1.postgresql.org (Postfix) with ESMTP id DE2BBD1D06D for ; Thu, 15 Apr 2004 19:18:58 -0300 (ADT) Received: from 3times25.net (localhost [127.0.0.1]) by 3times25.net (Postfix) with ESMTP id 49A0C4B9F6; Thu, 15 Apr 2004 18:23:19 -0400 (EDT) Message-ID: <407F0B57.9040206@3times25.net> Date: Thu, 15 Apr 2004 18:23:19 -0400 From: Geoffrey User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Greg Stark Cc: Bruno Wolff III , Simon Riggs , Josh Berkus , pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. References: <200404050836.52115.josh@agliodbs.com> <20040415172059.GA15119@wolff.to> <873c74c46o.fsf@stark.xeocode.com> In-Reply-To: <873c74c46o.fsf@stark.xeocode.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: 200404/194 X-Sequence-Number: 6494 Greg Stark wrote: > Bruno Wolff III writes: > > >>I have seen exactly this happen a number of times over the last several >>years. However there is still only one Tom Lane implementing the >>improvements. > > > Ob: Well clearly the problem is we need more Tom Lanes. > my $pgGuru = "Tom Lane"; my @morepgGurus; my $howmany = 10; while($howmany--) { push @morepgGurus, $pgGuru; } -- Until later, Geoffrey Registered Linux User #108567 Building secure systems in spite of Microsoft From pgsql-performance-owner@postgresql.org Thu Apr 15 20:02:06 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 278B5D1D300 for ; Thu, 15 Apr 2004 20:02:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50208-03 for ; Thu, 15 Apr 2004 20:02:10 -0300 (ADT) Received: from email04.aon.at (WARSL404PIP2.highway.telekom.at [195.3.96.113]) by svr1.postgresql.org (Postfix) with SMTP id C9045D1D2F7 for ; Thu, 15 Apr 2004 20:02:02 -0300 (ADT) Received: (qmail 284980 invoked from network); 15 Apr 2004 23:02:10 -0000 Received: from m155p007.dipool.highway.telekom.at (HELO PASCAL) ([62.46.9.71]) (envelope-sender ) by qmail7rs.highway.telekom.at (qmail-ldap-1.03) with SMTP for ; 15 Apr 2004 23:02:10 -0000 From: Manfred Koizar To: Tom Lane Cc: jdunn@autorevenue.com, "'Postgresql Performance'" Subject: Re: index v. seqscan for certain values Date: Fri, 16 Apr 2004 01:04:24 +0200 Message-ID: <571u701m7fkc97oi1ul8qfuf8e94kvi8lg@email.aon.at> References: <000801c42165$68a67750$4f01a8c0@jeremydunn> <24670.1081878949@sss.pgh.pa.us> In-Reply-To: <24670.1081878949@sss.pgh.pa.us> X-Mailer: Forte Agent 1.93/32.576 English (American) 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: 200404/196 X-Sequence-Number: 6496 On Tue, 13 Apr 2004 13:55:49 -0400, Tom Lane wrote: >Possibly the >nonuniform clumping of CID has something to do with the poor results. It shouldn't. The sampling algorithm is designed to give each tuple the same chance of ending up in the sample, and tuples are selected independently. (IOW each one of the {N \chooose n} possible samples has the same probability.) There are known problems with nonuniform distribution of dead vs. live and large vs. small tuples, but AFAICS the order of values does not matter. Servus Manfred From pgsql-performance-owner@postgresql.org Thu Apr 15 20:30:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F18AFD1D07E for ; Thu, 15 Apr 2004 20:30:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 56339-06 for ; Thu, 15 Apr 2004 20:30:39 -0300 (ADT) Received: from email10.aon.at (warsl404pip7.highway.telekom.at [195.3.96.91]) by svr1.postgresql.org (Postfix) with SMTP id 801DCD1B8B7 for ; Thu, 15 Apr 2004 20:30:31 -0300 (ADT) Received: (qmail 143146 invoked from network); 15 Apr 2004 23:30:39 -0000 Received: from m155p007.dipool.highway.telekom.at (HELO PASCAL) ([62.46.9.71]) (envelope-sender ) by 172.18.5.237 (qmail-ldap-1.03) with SMTP for ; 15 Apr 2004 23:30:39 -0000 From: Manfred Koizar To: Tom Lane Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats Date: Fri, 16 Apr 2004 01:32:53 +0200 Message-ID: References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> In-Reply-To: <25556.1081883922@sss.pgh.pa.us> X-Mailer: Forte Agent 1.93/32.576 English (American) 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: 200404/197 X-Sequence-Number: 6497 [Just a quick note here; a more thorough discussion of my test results will be posted to -hackers] On Tue, 13 Apr 2004 15:18:42 -0400, Tom Lane wrote: >Well, the first problem is why is ANALYZE's estimate of the total row >count so bad :-( ? I suspect you are running into the situation where >the initial pages of the table are thinly populated and ANALYZE >mistakenly assumes the rest are too. Manfred is working on a revised >sampling method for ANALYZE that should fix this problem The new method looks very promising with respect to row count estimation: I got estimation errors of +/- 1% where the old method was off by up to 60%. (My test methods might be a bit biased though :-)) My biggest concern at the moment is that the new sampling method violates the contract of returning each possible sample with he same probability: getting several tuples from the same page is more likely than with the old method. Servus Manfred From pgsql-performance-owner@postgresql.org Thu Apr 15 21:18:47 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 273E3D1E971 for ; Thu, 15 Apr 2004 21:18:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 66774-10 for ; Thu, 15 Apr 2004 21:18:51 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id D50FFD1B8B7 for ; Thu, 15 Apr 2004 21:18:43 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3G0InlS022643; Thu, 15 Apr 2004 20:18:49 -0400 (EDT) To: Manfred Koizar Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats In-reply-to: References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> Comments: In-reply-to Manfred Koizar message dated "Fri, 16 Apr 2004 01:32:53 +0200" Date: Thu, 15 Apr 2004 20:18:49 -0400 Message-ID: <22642.1082074729@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/198 X-Sequence-Number: 6498 Manfred Koizar writes: > My biggest concern at the moment is that the new sampling method > violates the contract of returning each possible sample with he same > probability: getting several tuples from the same page is more likely > than with the old method. Hm, are you sure? I recall objecting to your original proposal because I thought that would happen, but after further thought it seemed not. Also, I'm not at all sure that the old method satisfies that constraint completely in the presence of nonuniform numbers of tuples per page, so we'd not necessarily be going backwards anyhow ... regards, tom lane From pgsql-performance-owner@postgresql.org Thu Apr 15 22:05:03 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3D183D1E990 for ; Thu, 15 Apr 2004 22:05:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81278-08 for ; Thu, 15 Apr 2004 22:05:07 -0300 (ADT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id C6CF5D1E98F for ; Thu, 15 Apr 2004 22:04:58 -0300 (ADT) Received: from commandprompt.com (clbb-248.saw.net [64.146.135.248]) (authenticated) by hosting.commandprompt.com (8.11.6/8.11.6) with ESMTP id i3G150u02206; Thu, 15 Apr 2004 18:05:01 -0700 Message-ID: <407F31D6.6010104@commandprompt.com> Date: Thu, 15 Apr 2004 18:07:34 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Geoffrey Cc: Greg Stark , Bruno Wolff III , Simon Riggs , Josh Berkus , pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. References: <200404050836.52115.josh@agliodbs.com> <20040415172059.GA15119@wolff.to> <873c74c46o.fsf@stark.xeocode.com> <407F0B57.9040206@3times25.net> In-Reply-To: <407F0B57.9040206@3times25.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: 200404/199 X-Sequence-Number: 6499 > > my $pgGuru = "Tom Lane"; my @morepgGurus; my $howmany = 10; > > while($howmany--) { push @morepgGurus, $pgGuru; } > This is just wrong... -- Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC and S/JDBC Postgresql support, programming shared hosting and dedicated hosting. +1-503-667-4564 - jd@commandprompt.com - http://www.commandprompt.com PostgreSQL Replicator -- production quality replication for PostgreSQL From pgsql-performance-owner@postgresql.org Thu Apr 15 23:27:16 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3CE90D1E947 for ; Thu, 15 Apr 2004 23:27:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 00702-07 for ; Thu, 15 Apr 2004 23:27:20 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 6C85CD1E1F3 for ; Thu, 15 Apr 2004 23:27:12 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id DF7C217C33C; Thu, 15 Apr 2004 22:27:21 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BEJ4b-0003ch-00; Thu, 15 Apr 2004 22:27:21 -0400 To: "Joshua D. Drake" Cc: Geoffrey , Greg Stark , Bruno Wolff III , Simon Riggs , Josh Berkus , pgsql-performance@postgresql.org Subject: Re: PostgreSQL and Linux 2.6 kernel. References: <200404050836.52115.josh@agliodbs.com> <20040415172059.GA15119@wolff.to> <873c74c46o.fsf@stark.xeocode.com> <407F0B57.9040206@3times25.net> <407F31D6.6010104@commandprompt.com> In-Reply-To: <407F31D6.6010104@commandprompt.com> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 15 Apr 2004 22:27:20 -0400 Message-ID: <87wu4gac8n.fsf@stark.xeocode.com> 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: 200404/200 X-Sequence-Number: 6500 "Joshua D. Drake" writes: > > while($howmany--) { push @morepgGurus, $pgGuru; } > > This is just wrong... yeah, it would have been much clearer written as: push @morepgGurus, ($pgGuru)x$howmany; Or at least the perlish: for (1..$howmany) instead of C style while syntax. Ok. I stop now. -- greg From pgsql-performance-owner@postgresql.org Fri Apr 16 04:24:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1CA25D1DB0A for ; Fri, 16 Apr 2004 04:24:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 91072-04 for ; Fri, 16 Apr 2004 04:24:31 -0300 (ADT) Received: from cmailm1.svr.pol.co.uk (cmailm1.svr.pol.co.uk [195.92.193.18]) by svr1.postgresql.org (Postfix) with ESMTP id 5A209D1D2DC for ; Fri, 16 Apr 2004 04:24:32 -0300 (ADT) Received: from modem-1925.llama.dialup.pol.co.uk ([217.135.183.133] helo=stromboli) by cmailm1.svr.pol.co.uk with esmtp (Exim 4.14) id 1BENiC-0002hc-DM; Fri, 16 Apr 2004 08:24:32 +0100 From: "Simon Riggs" To: "Josh Berkus" , Subject: Re: PostgreSQL and Linux 2.6 kernel. Date: Fri, 16 Apr 2004 08:24:46 +0100 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: <200404151339.45615.josh@agliodbs.com> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/201 X-Sequence-Number: 6501 >Josh Berkus > > Treating the optimizer as a black box is something I'm very > used to from > > other RDBMS. My question is, how can you explicitly > re-write a query now > > to "improve" it? If there's no way of manipulating queries without > > actually re-writing the optimizer, we're now in a position where we > > aren't able to diagnose when the optimizer isn't working > effectively. > > Well, there is ... all of the various query cost parameters. They are very blunt instruments for such a delicate task. Surely someone of your experience might have benefit from something more? My feeling is, I would, though I want those tools as *a developer* rather than for tuning specific queries for people, which is always so sensitive to upgrades etc. > But, ultimately, improvements on the planner are still > bottlenecked by having > only one developer actually hacking the changes. > Do we have a clear list of optimizations we'd like to be working on? The TODO items aren't very related to specific optimizations... The only ones I was aware of was deferred subselect evaluation for DBT-3. ...sounds like there's more to discuss here, so I'll duck out now and get back to my current project... Best Regards, Simon Riggs From pgsql-performance-owner@postgresql.org Fri Apr 16 05:22:14 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1445BD1C4C3 for ; Fri, 16 Apr 2004 05:22:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11556-02 for ; Fri, 16 Apr 2004 05:22:10 -0300 (ADT) Received: from trade-india.com (unknown [61.16.154.82]) by svr1.postgresql.org (Postfix) with SMTP id 149F9D1BCB8 for ; Fri, 16 Apr 2004 05:22:07 -0300 (ADT) Received: (qmail 7040 invoked from network); 16 Apr 2004 08:08:56 -0000 Received: from valium.trade-india-local.com (HELO trade-india.com) (192.168.0.72) by system66.trade-india-local.com with SMTP; 16 Apr 2004 08:08:56 -0000 Message-ID: <407F9816.2080407@trade-india.com> Date: Fri, 16 Apr 2004 13:53:50 +0530 From: Rajesh Kumar Mallah Organization: Infocom Network Limited User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Richard Huxton Cc: Bill Moran , Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already References: <407D7A89.60508@trade-india.com> <407E8FF6.4050601@potentialtech.com> <407EB605.70502@trade-india.com> <200404151844.33672.dev@archonet.com> In-Reply-To: <200404151844.33672.dev@archonet.com> Content-Type: multipart/alternative; boundary="------------020309060308030400050407" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/202 X-Sequence-Number: 6502 This is a multi-part message in MIME format. --------------020309060308030400050407 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit I am running an update on the same table update rfis set inquiry_status='APPROVED' where inquiry_status='a'; Its running for past 20 mins. and top output is below. The PID which is executing the query above is 6712. Can anyone tell me why it is in an uninterruptable sleep and does it relate to the apparent poor performance? Is it problem with the disk hardware. I know at nite this query will run reasonably fast. I am running on a decent hardware . Regds mallah. 1:41pm up 348 days, 21:10, 1 user, load average: 11.59, 13.69, 11.49 85 processes: 83 sleeping, 1 running, 0 zombie, 1 stopped CPU0 states: 8.1% user, 2.3% system, 0.0% nice, 89.0% idle CPU1 states: 3.3% user, 2.3% system, 0.0% nice, 93.2% idle CPU2 states: 7.4% user, 1.4% system, 0.0% nice, 90.0% idle CPU3 states: 9.3% user, 7.4% system, 0.0% nice, 82.2% idle Mem: 2064796K av, 2053964K used, 10832K free, 0K shrd, 22288K buff Swap: 2048244K av, 88660K used, 1959584K free 1801532K cached PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND * 6712 postgres 16 0 86592 84M 83920 D 11.1 4.1 1:36 postmaster* 13103 postgres 15 0 54584 53M 52556 S 3.5 2.6 0:01 postmaster 13034 root 16 0 1072 1072 848 R 2.1 0.0 0:02 top 13064 postgres 15 0 67256 65M 64516 D 2.1 3.2 0:01 postmaster 13088 postgres 16 0 43324 42M 40812 D 2.1 2.0 0:00 postmaster 13076 postgres 15 0 49016 47M 46628 S 1.9 2.3 0:00 postmaster 26931 postgres 15 0 84880 82M 83888 S 1.7 4.1 3:52 postmaster 13107 postgres 15 0 18400 17M 16488 S 1.5 0.8 0:00 postmaster 13068 postgres 15 0 44632 43M 42324 D 1.3 2.1 0:00 postmaster 13074 postgres 15 0 68852 67M 66508 D 1.3 3.3 0:00 postmaster 13108 postgres 15 0 11692 11M 10496 S 1.3 0.5 0:00 postmaster 13075 postgres 15 0 50860 49M 47680 S 1.1 2.4 0:04 postmaster 13066 postgres 15 0 56112 54M 53724 S 0.9 2.7 0:01 postmaster 13109 postgres 15 0 14528 14M 13272 S 0.9 0.7 0:00 postmaster 24454 postgres 15 0 2532 2380 1372 S 0.7 0.1 11:58 postmaster 12 root 15 0 0 0 0 SW 0.5 0.0 816:30 bdflush 24455 postgres 15 0 1600 1476 1380 S 0.5 0.0 9:11 postmaster 12528 postgres 15 0 84676 82M 79920 S 0.3 4.0 0:02 postmaster 12575 postgres 15 0 76660 74M 75796 D 0.3 3.7 0:09 postmaster 13038 postgres 15 0 48952 47M 46436 D 0.3 2.3 0:00 postmaster 13069 postgres 15 0 57464 56M 54852 S 0.3 2.7 0:00 postmaster 13102 postgres 15 0 17864 17M 16504 D 0.3 0.8 0:00 postmaster Richard Huxton wrote: >On Thursday 15 April 2004 17:19, Rajesh Kumar Mallah wrote: > > >>Bill Moran wrote: >> >> >>>Rajesh Kumar Mallah wrote: >>> >>> >>>>Hi, >>>> >>>>The problem was solved by reloading the Table. >>>>the query now takes only 3 seconds. But that is >>>>not a solution. >>>> >>>> >>>If dropping/recreating the table improves things, then we can reasonably >>>assume that the table is pretty active with updates/inserts. Correct? >>> >>> >>Yes the table results from an import process and under goes lots >>of inserts and updates , but thats before the vacuum full operation. >>the table is not accessed during vacuum. What i want to know is >>is there any wat to automate the dumping and reload of a table >>individually. will the below be safe and effective: >> >> > >Shouldn't be necessary assuming you vacuum (not full) regularly. However, >looking back at your original posting, the vacuum output doesn't seem to show >any rows that need removing. > ># VACUUM full verbose eyp_rfi; >INFO: vacuuming "public.eyp_rfi" >INFO: "eyp_rfi": found 0 removable, 505960 nonremovable row versions in >71987 pages >DETAIL: 0 dead row versions cannot be removed yet. > >Since your select count(*) showed 505960 rows, I can't see how >dropping/replacing could make a difference on a sequential scan. Since we're >not using any indexes I don't see how it could be related to that. > > > >>begin work; >>create table new_tab AS select * from tab; >>truncate table tab; >>insert into tab select * from new_tab; >>drop table new_tab; >>commit; >>analyze tab; >> >>i havenot tried it but plan to do so. >>but i feel insert would take ages to update >>the indexes if any. >> >> > >It will have to update them, which will take time. > > > >>BTW >> >>is there any way to disable checks and triggers on >>a table temporarily while loading data (is updating >>reltriggers in pg_class safe?) >> >> > >You can take a look at pg_restore and copy how it does it. > > > --------------020309060308030400050407 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit

I am running an update on the same table

update rfis set inquiry_status='APPROVED' where inquiry_status='a';

Its running for past 20 mins. and top output is below.
The PID which is executing the query above is 6712. Can anyone
tell me why it is in an uninterruptable sleep and does it relate
to the apparent poor performance? Is it problem with the disk
hardware. I know at nite this query will run reasonably fast.

I am running on a decent hardware .



Regds
mallah.



 1:41pm  up 348 days, 21:10,  1 user,  load average: 11.59, 13.69, 11.49
85 processes: 83 sleeping, 1 running, 0 zombie, 1 stopped
CPU0 states:  8.1% user,  2.3% system,  0.0% nice, 89.0% idle
CPU1 states:  3.3% user,  2.3% system,  0.0% nice, 93.2% idle
CPU2 states:  7.4% user,  1.4% system,  0.0% nice, 90.0% idle
CPU3 states:  9.3% user,  7.4% system,  0.0% nice, 82.2% idle
Mem:  2064796K av, 2053964K used,   10832K free,       0K shrd,   22288K buff
Swap: 2048244K av,   88660K used, 1959584K free                 1801532K cached

  PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
 6712 postgres  16   0 86592  84M 83920 D    11.1  4.1   1:36 postmaster
13103 postgres  15   0 54584  53M 52556 S     3.5  2.6   0:01 postmaster
13034 root      16   0  1072 1072   848 R     2.1  0.0   0:02 top
13064 postgres  15   0 67256  65M 64516 D     2.1  3.2   0:01 postmaster
13088 postgres  16   0 43324  42M 40812 D     2.1  2.0   0:00 postmaster
13076 postgres  15   0 49016  47M 46628 S     1.9  2.3   0:00 postmaster
26931 postgres  15   0 84880  82M 83888 S     1.7  4.1   3:52 postmaster
13107 postgres  15   0 18400  17M 16488 S     1.5  0.8   0:00 postmaster
13068 postgres  15   0 44632  43M 42324 D     1.3  2.1   0:00 postmaster
13074 postgres  15   0 68852  67M 66508 D     1.3  3.3   0:00 postmaster
13108 postgres  15   0 11692  11M 10496 S     1.3  0.5   0:00 postmaster
13075 postgres  15   0 50860  49M 47680 S     1.1  2.4   0:04 postmaster
13066 postgres  15   0 56112  54M 53724 S     0.9  2.7   0:01 postmaster
13109 postgres  15   0 14528  14M 13272 S     0.9  0.7   0:00 postmaster
24454 postgres  15   0  2532 2380  1372 S     0.7  0.1  11:58 postmaster
   12 root      15   0     0    0     0 SW    0.5  0.0 816:30 bdflush
24455 postgres  15   0  1600 1476  1380 S     0.5  0.0   9:11 postmaster
12528 postgres  15   0 84676  82M 79920 S     0.3  4.0   0:02 postmaster
12575 postgres  15   0 76660  74M 75796 D     0.3  3.7   0:09 postmaster
13038 postgres  15   0 48952  47M 46436 D     0.3  2.3   0:00 postmaster
13069 postgres  15   0 57464  56M 54852 S     0.3  2.7   0:00 postmaster
13102 postgres  15   0 17864  17M 16504 D     0.3  0.8   0:00 postmaster



















Richard Huxton wrote: 
On Thursday 15 April 2004 17:19, Rajesh Kumar Mallah wrote:
  
Bill Moran wrote:
    
Rajesh Kumar Mallah wrote:
      
Hi,

The problem was solved by reloading the Table.
the query now takes only 3 seconds. But that is
not a solution.
        
If dropping/recreating the table improves things, then we can reasonably
assume that the table is pretty active with updates/inserts.  Correct?
      
Yes the table results from an import process and under goes lots
of inserts and updates , but thats before the vacuum full operation.
the table is not accessed during vacuum. What i want to know is
is there any wat to automate the dumping and reload of a table
individually. will the below be safe and effective:
    

Shouldn't be necessary assuming you vacuum (not full) regularly. However, 
looking back at your original posting, the vacuum output doesn't seem to show 
any rows that need removing.

# VACUUM full verbose eyp_rfi;
INFO:  vacuuming "public.eyp_rfi"
INFO:  "eyp_rfi": found 0 removable, 505960 nonremovable row versions in 
71987 pages
DETAIL:  0 dead row versions cannot be removed yet.

Since your select count(*) showed 505960 rows, I can't see how 
dropping/replacing could make a difference on a sequential scan. Since we're 
not using any indexes I don't see how it could be related to that.

  
begin work;
create table new_tab AS select * from tab;
truncate table tab;
insert into tab select * from new_tab;
drop table new_tab;
commit;
analyze tab;

i havenot tried it but plan to do so.
but i feel insert would take ages to update
the indexes if any.
    

It will have to update them, which will take time.

  
BTW

is there any way to disable checks and triggers on
a table temporarily while loading data (is updating
reltriggers in pg_class safe?)
    

You can take a look at pg_restore and copy how it does it.

--------------020309060308030400050407-- From pgsql-performance-owner@postgresql.org Fri Apr 16 07:13:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 80028D1E129 for ; Fri, 16 Apr 2004 07:13:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48368-05 for ; Fri, 16 Apr 2004 07:13:56 -0300 (ADT) Received: from email01.aon.at (warsl404pip8.highway.telekom.at [195.3.96.102]) by svr1.postgresql.org (Postfix) with SMTP id 678BAD1E159 for ; Fri, 16 Apr 2004 07:13:56 -0300 (ADT) Received: (qmail 327992 invoked from network); 16 Apr 2004 10:13:56 -0000 Received: from m154p002.dipool.highway.telekom.at (HELO PASCAL) ([62.46.9.34]) (envelope-sender ) by qmail1rs.highway.telekom.at (qmail-ldap-1.03) with SMTP for ; 16 Apr 2004 10:13:56 -0000 From: Manfred Koizar To: Tom Lane Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats Date: Fri, 16 Apr 2004 12:16:11 +0200 Message-ID: References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> <22642.1082074729@sss.pgh.pa.us> In-Reply-To: <22642.1082074729@sss.pgh.pa.us> X-Mailer: Forte Agent 1.93/32.576 English (American) 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: 200404/203 X-Sequence-Number: 6503 On Thu, 15 Apr 2004 20:18:49 -0400, Tom Lane wrote: >> getting several tuples from the same page is more likely >> than with the old method. > >Hm, are you sure? Almost sure. Let's look at a corner case: What is the probability of getting a sample with no two tuples from the same page? To simplify the problem assume that each page contains the same number of tuples c. If the number of pages is B and the sample size is n, a perfect sampling method collects a sample where all tuples come from different pages with probability (in OpenOffice.org syntax): p = prod from{i = 0} to{n - 1} {{c(B - i)} over {cB - i}} or in C: p = 1.0; for (i = 0; i < n; ++i) p *= c*(B - i) / (c*B - i) This probability grows with increasing B. >Also, I'm not at all sure that the old method satisfies that constraint >completely in the presence of nonuniform numbers of tuples per page, >so we'd not necessarily be going backwards anyhow ... Yes, it boils down to a decision whether we want to replace one not quite perfect sampling method with another not quite perfect method. I'm still working on putting together the pros and cons ... Servus Manfred From pgsql-performance-owner@postgresql.org Fri Apr 16 10:03:30 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C5AE2D1B49B for ; Fri, 16 Apr 2004 10:03:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04641-04 for ; Fri, 16 Apr 2004 10:03:27 -0300 (ADT) Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id 896D5D1B8B7 for ; Fri, 16 Apr 2004 10:03:25 -0300 (ADT) Received: from aeccom.com (cayambe.core.aeccom.com [192.168.2.12]) by mail2.aeccom.com (Postfix) with ESMTP id 24A4259; Fri, 16 Apr 2004 15:03:28 +0200 (CEST) Message-ID: <407FD9A0.6040608@aeccom.com> Date: Fri, 16 Apr 2004 15:03:28 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH, Berlin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031016 X-Accept-Language: de, en MIME-Version: 1.0 To: Tom Lane , josh@agliodbs.com, pgsql-performance@postgresql.org Cc: Sven Geisler Subject: RESOLVED: Re: Wierd context-switching issue on Xeon References: <200311251419.36771.josh@agliodbs.com> <200404150827.03950.josh@agliodbs.com> <407EB5B0.3030706@aeccom.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <21280.1082064326@sss.pgh.pa.us> <407F9FE9.3060104@aeccom.com> <407FAF0F.2070304@aeccom.com> In-Reply-To: <407FAF0F.2070304@aeccom.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: 200404/204 X-Sequence-Number: 6504 Tom, Josh, I think we have the problem resolved after I found the following note from Tom: > A large number of semops may mean that you have excessive contention on some lockable > resource, but I don't have enough info to guess what resource. This was the key to look at: we were missing all indices on table which is used heavily and does lots of locking. After recreating the missing indices the production system performed normal. No, more excessive semop() calls, load way below 1.0, CS over 20.000 very rare, more in thousands realm and less. This is quite a relief but I am sorry that the problem was so stupid and you wasted some time although Tom said he had also seem excessive semop() calls on another Dual XEON system. Hyperthreading was turned off so far but will be turned on again the next days. I don't expect any problems then. I'm not sure if this semop() problem is still an issue but the database behaves a bit out of bounds in this situation, i.e. consuming system resources with semop() calls 95% while tables are locked very often and longer. Thanks for your help, Dirk At last here is the current vmstat 1 excerpt where the problem has been resolved: procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 1 0 2308 232508 201924 6976532 0 0 136 464 628 812 5 1 94 0 0 0 2308 232500 201928 6976628 0 0 96 296 495 484 4 0 95 0 0 1 2308 232492 201928 6976628 0 0 0 176 347 278 1 0 99 0 0 0 2308 233484 201928 6976596 0 0 40 580 443 351 8 2 90 0 1 0 2308 233484 201928 6976696 0 0 76 692 792 651 9 2 88 0 0 0 2308 233484 201928 6976696 0 0 0 20 132 34 0 0 100 0 0 0 2308 233484 201928 6976696 0 0 0 76 177 90 0 0 100 0 0 1 2308 233484 201928 6976696 0 0 0 216 321 250 4 0 96 0 0 0 2308 233484 201928 6976696 0 0 0 116 417 240 8 0 92 0 0 0 2308 233484 201928 6976784 0 0 48 600 403 270 8 0 92 0 0 0 2308 233464 201928 6976860 0 0 76 452 1064 2611 14 1 84 0 0 0 2308 233460 201932 6976900 0 0 32 256 587 587 12 1 87 0 0 0 2308 233460 201932 6976932 0 0 32 188 379 287 5 0 94 0 0 0 2308 233460 201932 6976932 0 0 0 0 103 8 0 0 100 0 0 0 2308 233460 201932 6976932 0 0 0 0 102 14 0 0 100 0 0 1 2308 233444 201948 6976932 0 0 0 348 300 180 1 0 99 0 1 0 2308 233424 201948 6976948 0 0 16 380 739 906 4 2 93 0 0 0 2308 233424 201948 6977032 0 0 68 260 724 987 7 0 92 0 0 0 2308 231924 201948 6977128 0 0 96 344 1130 753 11 1 88 0 1 0 2308 231924 201948 6977248 0 0 112 324 687 628 3 0 97 0 0 0 2308 231924 201948 6977248 0 0 0 192 575 430 5 0 95 0 1 0 2308 231924 201948 6977248 0 0 0 264 208 124 0 0 100 0 0 0 2308 231924 201948 6977264 0 0 16 272 380 230 3 2 95 0 0 0 2308 231924 201948 6977264 0 0 0 0 104 8 0 0 100 0 0 0 2308 231924 201948 6977264 0 0 0 48 258 92 1 0 99 0 0 0 2308 231816 201948 6977484 0 0 212 268 456 384 2 0 98 0 0 0 2308 231816 201948 6977484 0 0 0 88 453 770 0 0 99 0 0 0 2308 231452 201948 6977680 0 0 196 476 615 676 5 0 94 0 0 0 2308 231452 201948 6977680 0 0 0 228 431 400 2 0 98 0 0 0 2308 231452 201948 6977680 0 0 0 0 237 58 3 0 97 0 0 0 2308 231448 201952 6977680 0 0 0 0 365 84 2 0 97 0 0 0 2308 231448 201952 6977680 0 0 0 40 246 108 1 0 99 0 0 0 2308 231448 201952 6977776 0 0 96 352 606 1026 4 2 94 0 0 0 2308 231448 201952 6977776 0 0 0 240 295 266 5 0 95 0 From pgsql-performance-owner@postgresql.org Fri Apr 16 10:40:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B86FBD1BB93 for ; Fri, 16 Apr 2004 10:40:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 21839-03 for ; Fri, 16 Apr 2004 10:40:10 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id E9F54D1B4BE for ; Fri, 16 Apr 2004 10:40:04 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id <2A89X7DM>; Fri, 16 Apr 2004 09:40:07 -0400 Message-ID: From: "Shea,Dan [CIS]" To: 'Bill Moran' Cc: Postgres Performance Subject: Re: [ SOLVED ] select count(*) very slow on an already Date: Fri, 16 Apr 2004 09:40:06 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/205 X-Sequence-Number: 6505 Just a note, I was trying the cluster command and was short on space. I figured I had enough space for the new table and index. It failed on me twice. The reason is that I noticed for the command to complete, it needed the space of the new table and 2x the space of the new index. It looks like it creates the new table, then a new index. Afterwards it looked like it creates another index in the DB pgsql_tmp. So for me this is an important consideration, since the new index size was about 7GB. I had not anticipated the second index size so that is why it failed. I ended up creating a link of pgsql_tmp to another parttion to successfully complete. Dan. -----Original Message----- From: Bill Moran [mailto:wmoran@potentialtech.com] Sent: Thursday, April 15, 2004 4:14 PM To: Shea,Dan [CIS] Cc: Postgres Performance Subject: Re: [PERFORM] [ SOLVED ] select count(*) very slow on an already Shea,Dan [CIS] wrote: > Bill, if you had alot of updates and deletions and wanted to optimize your > table, can you just issue the cluster command. > Will the cluster command rewrite the table without the obsolete data that a > vacuum flags or do you need to issue a vacuum first? From the reference docs: "During the cluster operation, a temporary copy of the table is created that contains the table data in the index order. Temporary copies of each index on the table are created as well. Therefore, you need free space on disk at least equal to the sum of the table size and the index sizes. "CLUSTER preserves GRANT, inheritance, index, foreign key, and other ancillary information about the table. "Because the optimizer records statistics about the ordering of tables, it is advisable to run ANALYZE on the newly clustered table. Otherwise, the optimizer may make poor choices of query plans." The primary reason CLUSTER exists is to allow you to physically reorder a table based on a key. This should provide a performance improvement if data with the same key is accessed all at once. (i.e. if you do "SELECT * FROM table WHERE key=5" and it returns 100 rows, those 100 rows are guaranteed to be all on the same part of the disk after CLUSTER, thus a performance improvement should result.) Updates and inserts will add data in the next available space in a table with no regard for any keys, and _may_ require running all over the disk to retrieve the data in the previous example query. I doubt if CLUSTER is an end-all optimization tool. The specific reason I suggested it was because the original poster was asking for an easier way to drop/recreate a table (as prior experimentation had shown this to improve performance) I can't think of anything easier than "CLUSTER ON " Since CLUSTER recreates the table, it implicitly removes the dead tuples. However, it's going to be a LOT slower than vacuum, so if dead tuples are the main problem, vacuum is still the way to go. -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Fri Apr 16 10:49:41 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C5DC4D1B8B7 for ; Fri, 16 Apr 2004 10:49:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15586-10 for ; Fri, 16 Apr 2004 10:49:39 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E4621D1B515 for ; Fri, 16 Apr 2004 10:49:37 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GDncmY028604; Fri, 16 Apr 2004 09:49:38 -0400 (EDT) To: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Cc: josh@agliodbs.com, pgsql-performance@postgresql.org, Sven Geisler Subject: Re: RESOLVED: Re: Wierd context-switching issue on Xeon In-reply-to: <407FD9A0.6040608@aeccom.com> References: <200311251419.36771.josh@agliodbs.com> <200404150827.03950.josh@agliodbs.com> <407EB5B0.3030706@aeccom.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <21280.1082064326@sss.pgh.pa.us> <407F9FE9.3060104@aeccom.com> <407FAF0F.2070304@aeccom.com> <407FD9A0.6040608@aeccom.com> Comments: In-reply-to =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= message dated "Fri, 16 Apr 2004 15:03:28 +0200" Date: Fri, 16 Apr 2004 09:49:38 -0400 Message-ID: <28603.1082123378@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/206 X-Sequence-Number: 6506 =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= writes: > This was the key to look at: we were missing all indices on table which > is used heavily and does lots of locking. After recreating the missing > indices the production system performed normal. No, more excessive > semop() calls, load way below 1.0, CS over 20.000 very rare, more in > thousands realm and less. Hmm ... that's darn interesting. AFAICT the test case I am looking at for Josh's client has no such SQL-level problem ... but I will go back and double check ... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 11:34:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 447A6D1B4B2 for ; Fri, 16 Apr 2004 11:34:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 42366-02 for ; Fri, 16 Apr 2004 11:34:52 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id E1FADD1E106 for ; Fri, 16 Apr 2004 11:34:50 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GEYoLm029061; Fri, 16 Apr 2004 10:34:50 -0400 (EDT) To: Manfred Koizar Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats In-reply-to: References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> <22642.1082074729@sss.pgh.pa.us> Comments: In-reply-to Manfred Koizar message dated "Fri, 16 Apr 2004 12:16:11 +0200" Date: Fri, 16 Apr 2004 10:34:49 -0400 Message-ID: <29060.1082126089@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/207 X-Sequence-Number: 6507 Manfred Koizar writes: > If the number of pages is B and the sample size is n, a perfect sampling > method collects a sample where all tuples come from different pages with > probability (in OpenOffice.org syntax): > p = prod from{i = 0} to{n - 1} {{c(B - i)} over {cB - i}} So? You haven't proven that either sampling method fails to do the same. The desired property can also be phrased as "every tuple should be equally likely to be included in the final sample". What we actually have in the case of your revised algorithm is "every page is equally likely to be sampled, and of the pages included in the sample, every tuple is equally likely to be chosen". Given that there are B total pages of which we sample b pages that happen to contain T tuples (in any distribution), the probability that a particular tuple gets chosen is (b/B) * (n/T) assuming that the two selection steps are independent and unbiased. Now b, B, and n are not dependent on which tuple we are talking about. You could argue that a tuple on a heavily populated page is statistically likely to see a higher T when it's part of the page sample pool than a tuple on a near-empty page is likely to see, and therefore there is some bias against selection of the former tuple. But given a sample over a reasonably large number of pages, the contribution of any one page to T should be fairly small and so this effect ought to be small. In fact, because T directly determines our estimate of the total number of tuples in the relation, your experiments showing that the new method gives a reliable tuple count estimate directly prove that T is pretty stable regardless of exactly which pages get included in the sample. So I think this method is effectively unbiased at the tuple level. The variation in probability of selection of individual tuples can be no worse than the variation in the overall tuple count estimate. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 12:17:16 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 89691D1E96D for ; Fri, 16 Apr 2004 12:17:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 51321-08 for ; Fri, 16 Apr 2004 12:17:08 -0300 (ADT) Received: from flake.decibel.org (flake.decibel.org [66.143.173.58]) by svr1.postgresql.org (Postfix) with SMTP id 98BA2D1E970 for ; Fri, 16 Apr 2004 12:17:07 -0300 (ADT) Received: (qmail 45700 invoked by uid 1001); 16 Apr 2004 15:17:06 -0000 Date: Fri, 16 Apr 2004 10:17:06 -0500 From: "Jim C. Nasby" To: pgsql-performance@postgresql.org Subject: Poor performance of group by query Message-ID: <20040416151706.GE87362@nasby.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline X-Operating-System: FreeBSD 4.9-RELEASE-p3 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/208 X-Sequence-Number: 6508 Anyone have any ideas why this query would be so slow? stats=# explain analyze SELECT work_units, min(raw_rank) AS rank FROM Trank_work_overall GROUP BY work_units; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------- HashAggregate (cost=1050.12..1085.98 rows=14347 width=16) (actual time=163149.981..163227.758 rows=17849 loops=1) -> Seq Scan on trank_work_overall (cost=0.00..804.41 rows=49141 width=16) (actual time=0.071..328.682 rows=49091 loops=1) Total runtime: 163296.212 ms (3 rows) stats=# \d Trank_work_overall Table "pg_temp_1.trank_work_overall" Column | Type | Modifiers ------------+--------+----------- raw_rank | bigint | work_units | bigint | stats=# FreeBSD fritz.distributed.net 5.2.1-RELEASE FreeBSD 5.2.1-RELEASE #1: Wed Apr 7 18:42:52 CDT 2004 root@fritz.distributed.net:/usr/obj/usr/src/sys/FRITZ amd64 The machine is a dual opteron with 4G of memory. The query in question was not hitting the disk at all. PostgreSQL 7.4.2 compiled with -O3. Also, if I set enable_hashagg = false, it runs in less than a second. -- Jim C. Nasby, Database Consultant jim@nasby.net Member: Triangle Fraternity, Sports Car Club of America Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Fri Apr 16 12:28:09 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 78652D1B8DF for ; Fri, 16 Apr 2004 12:28:08 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 62499-03 for ; Fri, 16 Apr 2004 12:28:00 -0300 (ADT) Received: from mail.vistashare.com (www.vistashare.net [65.207.67.162]) by svr1.postgresql.org (Postfix) with ESMTP id 1FA36D1B4B2 for ; Fri, 16 Apr 2004 12:28:04 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by mail.vistashare.com (Postfix) with ESMTP id 770581180 for ; Fri, 16 Apr 2004 11:28:01 -0400 (EDT) From: Chris Kratz Organization: VistaShare To: pgsql-performance@postgresql.org Subject: Long running queries degrade performance Date: Fri, 16 Apr 2004 11:28:00 -0400 User-Agent: KMail/1.6.2 MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Message-Id: <200404161128.00851.chris.kratz@vistashare.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/209 X-Sequence-Number: 6509 Hello all, My apologies if this is not the right mailing list to ask this question, but we are wondering about general performance tuning principles for our main db server. We have a web app with a postgres backend. Most queries have subsecond response times through the web even with high usage. Every once in awhile someone will run either an ad-hoc query or some other long running db process. For some reason, it seems that a small number 3-4 of these jobs running in parallel absolutely floors our server. In monitoring the jobs, linux (Kernel 2.4) drops the long running jobs priority, but even so they seem to hog the system resources making subsequent requests for everyone else very slow. Our database at this point is almost entirely processor and memory bound because it isn't too large to fit most of the working data into memory yet. There is generally little disk activity when this occurs. These long running processes are almost always complex select statements, not generally inserts or updates. We continue to monitor and rework the bottlenecks, but what is a little scary to us is how easily the database becomes almost completely unresponsive with several large jobs running, especially since we have a large number of users. And it only takes one user trying to view a page with one of these selects clicking multiple times because it doesn't come back quickly to bring our system to it's knees for hours. We are looking to move to Kernel 2.6 and possibly a dedicated multiprocessor machine for postgres towards the end of this year. But, I am wondering if there is anything we can do now to increase the interactive performance while there are long running selects running as well. Are there ways to adjust the priority of backend processes, or things to tweak to maximize interactive throughput for the quick jobs while the long running ones run in the background? Or if worse comes to worse to actually kill long running processes without taking down the whole db as we have had to do on occasion. Our server is a modest 2.4Ghz P4 with mirrored UW SCSI drives and 1G of memory. The db on disk is around 800M and this machine also hosts our web app, so there is some contention for the processor. Does anyone have any suggestions or thoughts on things we could look at? Is a multiprocessor box the only answer, or are there other things we should be looking at hardware wise. Thank you for your time. -- Chris Kratz Systems Analyst/Programmer VistaShare LLC www.vistashare.com From pgsql-performance-owner@postgresql.org Fri Apr 16 12:37:25 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9D39BD1D08C for ; Fri, 16 Apr 2004 12:37:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 58571-09 for ; Fri, 16 Apr 2004 12:37:17 -0300 (ADT) Received: from bramble.mmrd.com (unknown [65.217.53.66]) by svr1.postgresql.org (Postfix) with ESMTP id 561F0D1D085 for ; Fri, 16 Apr 2004 12:37:21 -0300 (ADT) 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 i3GFeYcM009750; Fri, 16 Apr 2004 11:40:35 -0400 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 i3GFbFl10132; Fri, 16 Apr 2004 11:37:17 -0400 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 FVMA9Y9C; Fri, 16 Apr 2004 11:37:13 -0400 Subject: Re: query slows down with more accurate stats From: Robert Treat To: Tom Lane Cc: pgsql-performance@postgresql.org In-Reply-To: <25556.1081883922@sss.pgh.pa.us> References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> Content-Type: text/plain Content-Transfer-Encoding: 7bit X-Mailer: Ximian Evolution 1.0.8 Date: 16 Apr 2004 11:37:15 -0400 Message-Id: <1082129835.23419.869.camel@camel> Mime-Version: 1.0 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/210 X-Sequence-Number: 6510 On Tue, 2004-04-13 at 15:18, Tom Lane wrote: > Robert Treat writes: > Well, the first problem is why is ANALYZE's estimate of the total row > count so bad :-( ? I suspect you are running into the situation where > the initial pages of the table are thinly populated and ANALYZE > mistakenly assumes the rest are too. That was my thinking, which is somewhat confirmed after a vacuum full on the table; now analyze gives pretty accurate states. Of course the downside is that now the query is consistently slower. > > so i guess i am wondering if there is something I should be doing to > > help get the better plan at the more accurate stats levels and/or why it > > doesn't stick with the original plan (I noticed disabling merge joins > > does seem to push it back to the original plan). > > With the larger number of estimated rows it's figuring the nestloop will > be too expensive. The row estimate for the cl scan went up from 1248 > to 10546, so the estimated cost for the nestloop plan would go to about > 240000 units vs 80000 for the mergejoin plan. This is obviously off > rather badly when the true runtimes are 1.7 vs 8.1 seconds :-(. > > I think this is an example of a case where we really need better > estimation of nestloop costs --- it's drastically overestimating the > relative cost of the nestloop because it's not accounting for the cache > benefits of the repeated index searches. You could probably force the > nestloop to be chosen by lowering random_page_cost, but that's just a > kluge solution ... the real problem is the model is wrong. > Unfortunately playing with random_page_cost doesn't seem to be enough to get it to favor the nested loop... though setting it down to 2 does help overall. played with index_cpu_tuple_cost a bit but that seemed even less useful. aggravating when you know there is a better plan it could pick but no (clean) way to get it to do so... > I have a to-do item to work on this, and will try to bump up its > priority a bit. > I'll keep an eye out, thanks Tom. Robert Treat -- Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL From pgsql-performance-owner@postgresql.org Fri Apr 16 12:45:13 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2A4FED1E95F for ; Fri, 16 Apr 2004 12:45:10 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 67660-07 for ; Fri, 16 Apr 2004 12:44:59 -0300 (ADT) Received: from flake.decibel.org (flake.decibel.org [66.143.173.58]) by svr1.postgresql.org (Postfix) with SMTP id E9431D1DACE for ; Fri, 16 Apr 2004 12:45:02 -0300 (ADT) Received: (qmail 47123 invoked by uid 1001); 16 Apr 2004 15:45:02 -0000 Date: Fri, 16 Apr 2004 10:45:02 -0500 From: "Jim C. Nasby" To: pgsql-performance@postgresql.org Subject: Horribly slow hash join Message-ID: <20040416154502.GF87362@nasby.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline X-Operating-System: FreeBSD 4.9-RELEASE-p3 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/211 X-Sequence-Number: 6511 Note the time for the hash join step: ------------------------------------------------------------------------------------------------------------------------------------------ Hash Join (cost=357.62..26677.99 rows=93668 width=62) (actual time=741.159..443381.011 rows=49091 loops=1) Hash Cond: ("outer".work_today = "inner".work_units) -> Hash Join (cost=337.11..24784.11 rows=93668 width=54) (actual time=731.374..417188.519 rows=49091 loops=1) Hash Cond: ("outer".work_total = "inner".work_units) -> Seq Scan on email_rank (cost=0.00..22240.04 rows=254056 width=46) (actual time=582.145..1627.759 rows=49091 loops=1) Filter: (project_id = 8) -> Hash (cost=292.49..292.49 rows=17849 width=16) (actual time=148.944..148.944 rows=0 loops=1) -> Seq Scan on rank_tie_overall o (cost=0.00..292.49 rows=17849 width=16) (actual time=0.059..75.984 rows=17849 loops=1) -> Hash (cost=17.81..17.81 rows=1081 width=16) (actual time=8.996..8.996 rows=0 loops=1) -> Seq Scan on rank_tie_today d (cost=0.00..17.81 rows=1081 width=16) (actual time=0.080..4.635 rows=1081 loops=1) Total runtime: 619047.032 ms By comparison: stats=# set enable_hashjoin=false; SET stats=# explain analyze select * from email_rank, rank_tie_overall o, rank_tie_today d WHERE email_rank.work_today = d.work_units AND email_rank.work_total = o.work_units AND email_rank.project_id = :ProjectID; QUERY PLAN --------------------------------------------------------------------------------------------------------------------------------------------------------------------- Merge Join (cost=55391.69..56823.23 rows=93668 width=80) (actual time=2705.344..3349.318 rows=49091 loops=1) Merge Cond: ("outer".work_units = "inner".work_today) -> Index Scan using work_units_today on rank_tie_today d (cost=0.00..23.89 rows=1081 width=16) (actual time=0.150..4.874 rows=1081 loops=1) -> Sort (cost=55391.69..55625.86 rows=93668 width=64) (actual time=2705.153..2888.039 rows=49091 loops=1) Sort Key: email_rank.work_today -> Merge Join (cost=45047.64..47656.93 rows=93668 width=64) (actual time=1685.414..2494.342 rows=49091 loops=1) Merge Cond: ("outer".work_units = "inner".work_total) -> Index Scan using work_units_overall on rank_tie_overall o (cost=0.00..361.34 rows=17849 width=16) (actual time=0.122..79.383 rows=17849 loops=1) -> Sort (cost=45047.64..45682.78 rows=254056 width=48) (actual time=1685.228..1866.215 rows=49091 loops=1) Sort Key: email_rank.work_total -> Seq Scan on email_rank (cost=0.00..22240.04 rows=254056 width=48) (actual time=786.515..1289.101 rows=49091 loops=1) Filter: (project_id = 8) Total runtime: 3548.087 ms Even though the second case is only a select, it seems clear that something's wrong... -- Jim C. Nasby, Database Consultant jim@nasby.net Member: Triangle Fraternity, Sports Car Club of America Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Fri Apr 16 12:46:13 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DED26D1E96F for ; Fri, 16 Apr 2004 12:46:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 70059-08 for ; Fri, 16 Apr 2004 12:46:03 -0300 (ADT) Received: from gw.tssi.com (gw.tssi.com [198.147.197.1]) by svr1.postgresql.org (Postfix) with ESMTP id 4A4C1D1E96C for ; Fri, 16 Apr 2004 12:46:07 -0300 (ADT) 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 i3GFk48S010495; Fri, 16 Apr 2004 10:46:05 -0500 Received: (from nolan@localhost) by gw.tssi.com (8.12.6/8.12.6/Submit) id i3GFk3p3010490; Fri, 16 Apr 2004 10:46:03 -0500 From: Mike Nolan Message-Id: <200404161546.i3GFk3p3010490@gw.tssi.com> Subject: Re: Long running queries degrade performance To: chris.kratz@vistashare.com (Chris Kratz) Date: Fri, 16 Apr 2004 10:46:02 -0500 (CDT) Cc: pgsql-performance@postgresql.org In-Reply-To: <200404161128.00851.chris.kratz@vistashare.com> from "Chris Kratz" at Apr 16, 2004 11:28:00 AM 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: 200404/212 X-Sequence-Number: 6512 > We have a web app with a postgres backend. Most queries have subsecond > response times through the web even with high usage. Every once in awhile > someone will run either an ad-hoc query or some other long running db > process. Are you sure it is postgres where the delay is occurring? I ask this because I also have a web-based front end to postgres, and while most of the time the queries respond in about a second every now and then I see one that takes much longer, sometimes 10-15 seconds. I've seen this behavior on both my development system and on the production server. The same query a while later might respond quickly again. I'm not sure where to look for the delay, either, and it is intermittent enough that I'm not even sure what monitoring techniques to use. -- Mike Nolan From pgsql-performance-owner@postgresql.org Fri Apr 16 13:34:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DB0F2D1E17E for ; Fri, 16 Apr 2004 13:34:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 90550-02 for ; Fri, 16 Apr 2004 13:34:09 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 0B994D1DB0A for ; Fri, 16 Apr 2004 13:34:13 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GGYBfq000451; Fri, 16 Apr 2004 12:34:12 -0400 (EDT) To: "Jim C. Nasby" Cc: pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: <20040416154502.GF87362@nasby.net> References: <20040416154502.GF87362@nasby.net> Comments: In-reply-to "Jim C. Nasby" message dated "Fri, 16 Apr 2004 10:45:02 -0500" Date: Fri, 16 Apr 2004 12:34:11 -0400 Message-ID: <450.1082133251@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/213 X-Sequence-Number: 6513 "Jim C. Nasby" writes: > Note the time for the hash join step: Have you ANALYZEd these tables lately? It looks to me like it's hashing on some column that has only a small number of distinct values, so that the hash doesn't actually help to avoid comparisons. The planner should know better than to choose such a plan, but if it's working with obsolete stats ... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 13:37:04 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B0412D1E95F for ; Fri, 16 Apr 2004 13:37:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 91508-02 for ; Fri, 16 Apr 2004 13:36:55 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id D4A94D1E95D for ; Fri, 16 Apr 2004 13:36:58 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GGavuh000478; Fri, 16 Apr 2004 12:36:57 -0400 (EDT) To: "Jim C. Nasby" Cc: pgsql-performance@postgresql.org Subject: Re: Poor performance of group by query In-reply-to: <20040416151706.GE87362@nasby.net> References: <20040416151706.GE87362@nasby.net> Comments: In-reply-to "Jim C. Nasby" message dated "Fri, 16 Apr 2004 10:17:06 -0500" Date: Fri, 16 Apr 2004 12:36:57 -0400 Message-ID: <477.1082133417@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/214 X-Sequence-Number: 6514 "Jim C. Nasby" writes: > Anyone have any ideas why this query would be so slow? That seems very bizarre. Would you be willing to send me a dump of the table off-list? regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 13:40:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B934AD1E928 for ; Fri, 16 Apr 2004 13:40:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 86690-07 for ; Fri, 16 Apr 2004 13:40:24 -0300 (ADT) Received: from silmaril.syscor.priv (h24-108-208-253.sbm.shawcable.net [24.108.208.253]) by svr1.postgresql.org (Postfix) with ESMTP id 96F6BD1E298 for ; Fri, 16 Apr 2004 13:40:26 -0300 (ADT) Received: from syscor.com (strider.syscor.priv [192.168.1.3]) by silmaril.syscor.priv (8.12.8/8.12.8) with ESMTP id i3GGeZXR016250 for ; Fri, 16 Apr 2004 09:40:36 -0700 Message-ID: <40800CC2.3050707@syscor.com> Date: Fri, 16 Apr 2004 09:41:38 -0700 From: Ron St-Pierre User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Index Problem? 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: 200404/215 X-Sequence-Number: 6515 I am using postgres 7.4.1 and have a problem with a plpgsql function. When I run the function on the production server it takes approx 33 minutes to run. I dumped the DB and copied it to a similarly configured box and ran the function and it ran in about 10 minutes. Can anyone offer advice on tuning the function or my database? Here are the lengthy, gory details. F u n c t i o n It updates seven columns of a table 1 to 4 times daily. Current data = 42,000 rows, new data = 30,000 rows. CREATE TYPE employeeType AS (empID INTEGER, updateDate DATE, bDate INTEGER, val1 NUMERIC, val2 NUMERIC, val3 NUMERIC, val4 NUMERIC, favNum NUMERIC); CREATE OR REPLACE FUNCTION updateEmployeeData() RETURNS SETOF employeeType AS ' DECLARE rec RECORD; BEGIN FOR rec IN SELECT empID, updateDate, bDate, val1, val2 , val3, val4, favNum FROM newData LOOP RETURN NEXT rec; UPDATE currentData SET val1=rec.val1, val2=rec.val2, val3=rec.val2, val4=rec.val4, favNum=rec.favNum, updateDate=rec.updateDate WHERE empID=rec.empID; END LOOP; RETURN; END; ' LANGUAGE 'plpgsql'; The emp table has 60 columns, all indexed, about two-thirds are numeric, but they are not affected by this update. The other 50+ columns are updated in the middle of the night and the amount of time that update takes isn't a concern. Late last night I dumped the table, dropped it and re-created it from the dump (on the production server - when no one was looking). When I re-ran the function it took almost 11 minutes, which was pretty much in line with my results from the dev server. D e t a i l s v 7.4.1 Debian stable 1 GB ram shared_buffers = 2048 sort_mem = 1024 SHMMAX 360000000 (360,000,000) VACUUM FULL ANALYZE is run every night, and I ran it yesterday between running the function and it made no difference in running time. top shows the postmaster using minimal cpu (0-40%) and miniscule memory. vmstat shows a fair amount of IO (bo=1000->4000). Yesterday on the dev server we upgraded to the 2.6 kernel and unfortunately only noticed a small increase in update time (about one minute). So does anyone have any suggestions for me on speeding this up? Is it the index? The function is run daily during the mid afternoon to early evening and really drags the performance of the server down (it also hosts a web site). Thanks Ron From pgsql-performance-owner@postgresql.org Fri Apr 16 13:46:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D0BEED1E947 for ; Fri, 16 Apr 2004 13:46:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 84539-10 for ; Fri, 16 Apr 2004 13:46:42 -0300 (ADT) Received: from flake.decibel.org (flake.decibel.org [66.143.173.58]) by svr1.postgresql.org (Postfix) with SMTP id 141F8D1E8AD for ; Fri, 16 Apr 2004 13:46:45 -0300 (ADT) Received: (qmail 50223 invoked by uid 1001); 16 Apr 2004 16:46:44 -0000 Date: Fri, 16 Apr 2004 11:46:44 -0500 From: "Jim C. Nasby" To: Tom Lane Cc: pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join Message-ID: <20040416164644.GH87362@nasby.net> References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <450.1082133251@sss.pgh.pa.us> X-Operating-System: FreeBSD 4.9-RELEASE-p3 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/216 X-Sequence-Number: 6516 Yes, stats are up to date, and the values should be fairly unique. Combined with the hash aggregate problem I saw (see my other email to the list), do you think there could be some issue with the performance of the hash function on FreeBSD 5.2 on AMD64? I'll post the table you requested someplace you can grab it. On Fri, Apr 16, 2004 at 12:34:11PM -0400, Tom Lane wrote: > "Jim C. Nasby" writes: > > Note the time for the hash join step: > > Have you ANALYZEd these tables lately? > > It looks to me like it's hashing on some column that has only a small > number of distinct values, so that the hash doesn't actually help to > avoid comparisons. The planner should know better than to choose such > a plan, but if it's working with obsolete stats ... > > regards, tom lane > -- Jim C. Nasby, Database Consultant jim@nasby.net Member: Triangle Fraternity, Sports Car Club of America Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Fri Apr 16 13:57:57 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 79C9BD1CCCC for ; Fri, 16 Apr 2004 13:57:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 95481-04 for ; Fri, 16 Apr 2004 13:57:51 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 31332D1C9C8 for ; Fri, 16 Apr 2004 13:57:54 -0300 (ADT) Received: from [64.81.245.111] (HELO 192.168.1.102) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4875295; Fri, 16 Apr 2004 09:59:17 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Dirk =?iso-8859-1?q?Lutzeb=E4ck?= , Tom Lane , pgsql-performance@postgresql.org Subject: Re: RESOLVED: Re: Wierd context-switching issue on Xeon Date: Fri, 16 Apr 2004 09:58:14 -0700 User-Agent: KMail/1.5.4 Cc: Sven Geisler References: <200311251419.36771.josh@agliodbs.com> <407FAF0F.2070304@aeccom.com> <407FD9A0.6040608@aeccom.com> In-Reply-To: <407FD9A0.6040608@aeccom.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404160958.14902.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/217 X-Sequence-Number: 6517 Dirk, > I'm not sure if this semop() problem is still an issue but the database > behaves a bit out of bounds in this situation, i.e. consuming system > resources with semop() calls 95% while tables are locked very often and > longer. It would be helpful to us if you could test this with the indexes disabled on the non-Bigmem system. I'd like to eliminate Bigmem as a factor, if possible. -- -Josh Berkus ______AGLIO DATABASE SOLUTIONS___________________________ Josh Berkus Enterprise vertical business josh@agliodbs.com and data analysis solutions (415) 752-2387 and database optimization fax 651-9224 utilizing Open Source technology San Francisco From pgsql-performance-owner@postgresql.org Fri Apr 16 14:01:44 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D6F5AD1CCCC for ; Fri, 16 Apr 2004 14:01:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 00364-01 for ; Fri, 16 Apr 2004 14:01:36 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 4423CD1B515 for ; Fri, 16 Apr 2004 14:01:39 -0300 (ADT) Received: from [64.81.245.111] (HELO 192.168.1.102) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4875322; Fri, 16 Apr 2004 10:03:02 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Ron St-Pierre , pgsql-performance@postgresql.org Subject: Re: Index Problem? Date: Fri, 16 Apr 2004 10:01:52 -0700 User-Agent: KMail/1.5.4 References: <40800CC2.3050707@syscor.com> In-Reply-To: <40800CC2.3050707@syscor.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404161001.52989.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/218 X-Sequence-Number: 6518 Ron, > The emp table has 60 columns, all indexed, about two-thirds are numeric, > but they are not affected by this update. The other 50+ columns are > updated in the middle of the night and the amount of time that update > takes isn't a concern. Well, I'd say that you have an application design problem, but that's not what you asked for help with ;-) > Late last night I dumped the table, dropped it and re-created it from > the dump (on the production server - when no one was looking). When I > re-ran the function it took almost 11 minutes, which was pretty much in > line with my results from the dev server. Sounds like you need to run a REINDEX on the table -- and after that, dramatically increase your max_fsm_pages, and run lazy VACUUM immediately after the batch update to clean up. -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Apr 16 14:05:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8036CD1E0D8 for ; Fri, 16 Apr 2004 14:05:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 97634-08 for ; Fri, 16 Apr 2004 14:04:56 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 4C980D1DB0A for ; Fri, 16 Apr 2004 14:04:59 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GH4wk7000809; Fri, 16 Apr 2004 13:04:58 -0400 (EDT) To: "Jim C. Nasby" Cc: pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: <20040416164644.GH87362@nasby.net> References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> Comments: In-reply-to "Jim C. Nasby" message dated "Fri, 16 Apr 2004 11:46:44 -0500" Date: Fri, 16 Apr 2004 13:04:58 -0400 Message-ID: <808.1082135098@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/219 X-Sequence-Number: 6519 "Jim C. Nasby" writes: > Combined with the hash aggregate problem I saw (see my other email to > the list), do you think there could be some issue with the performance > of the hash function on FreeBSD 5.2 on AMD64? Yeah, I was wondering about that too. Hard to imagine what though. The hash function should be pretty platform-independent. regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 14:52:27 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E3E07D1B515 for ; Fri, 16 Apr 2004 14:52:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 12369-06 for ; Fri, 16 Apr 2004 14:52:19 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6C738D1E98D for ; Fri, 16 Apr 2004 14:52:16 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id <2A89YQS2>; Fri, 16 Apr 2004 13:52:16 -0400 Message-ID: From: "Shea,Dan [CIS]" To: 'Rajesh Kumar Mallah' Cc: Postgres Performance Subject: Re: Deleting certain duplicates Date: Fri, 16 Apr 2004 13:52:15 -0400 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: 200404/220 X-Sequence-Number: 6520 Thanks Mallah, I will keep this example in case I need it again sometime in the future. Unfortunately, I do not have enough free space at the moment to create a temp table. Dan -----Original Message----- From: Rajesh Kumar Mallah [mailto:mallah@trade-india.com] Sent: Tuesday, April 13, 2004 10:27 AM To: Shea,Dan [CIS] Cc: Postgres Performance Subject: Re: [PERFORM] Deleting certain duplicates Shea,Dan [CIS] wrote: >The index is >Indexes: > "forecastelement_rwv_idx" btree (region_id, wx_element, valid_time) > >-----Original Message----- >From: Shea,Dan [CIS] [mailto:Dan.Shea@ec.gc.ca] >Sent: Monday, April 12, 2004 10:39 AM >To: Postgres Performance >Subject: [PERFORM] Deleting certain duplicates > > >We have a large database which recently increased dramatically due to a >change in our insert program allowing all entries. >PWFPM_DEV=# select relname,relfilenode,reltuples from pg_class where relname >= 'forecastelement'; > relname | relfilenode | reltuples >-----------------+-------------+------------- > forecastelement | 361747866 | 4.70567e+08 > > Column | Type | Modifiers >----------------+-----------------------------+----------- > version | character varying(99) | > origin | character varying(10) | > timezone | character varying(99) | > region_id | character varying(20) | > wx_element | character varying(99) | > value | character varying(99) | > flag | character(3) | > units | character varying(99) | > valid_time | timestamp without time zone | > issue_time | timestamp without time zone | > next_forecast | timestamp without time zone | > reception_time | timestamp without time zone | > >The program is supposed to check to ensure that all fields but the >reception_time are unique using a select statement, and if so, insert it. >Due an error in a change, reception time was included in the select to check >for duplicates. The reception_time is created by a program creating the dat >file to insert. >Essentially letting all duplicate files to be inserted. > >I tried the delete query below. >PWFPM_DEV=# delete from forecastelement where oid not in (select min(oid) >from forecastelement group by >version,origin,timezone,region_id,wx_element,value,flag,units,valid_time,is s >ue_time,next_forecast); >It ran for 3 days creating what I assume is an index in pgsql_tmp of the >group by statement. >The query ended up failing with "dateERROR:write failed". >Well the long weekend is over and we do not have the luxury of trying this >again. >So I was thinking maybe of doing the deletion in chunks, perhaps based on >reception time. > > its more of an sql question though. to deduplicate on basis of version,origin,timezone,region_id,wx_element,value,flag,units,valid_time, issue_time,next_forecast You could do this. begin work; create temp_table as select distinct on (version,origin,timezone,region_id,wx_element,value,flag,units,valid_time, issue_time,next_forecast) * from forecastelement ; truncate table forecastelement ; drop index ; insert into forecastelement select * from temp_table ; commit; create indexes Analyze forecastelement ; note that distinct on will keep only one row out of all rows having distinct values of the specified columns. kindly go thru the distinct on manual before trying the queries. regds mallah. >Are there any suggestions for a better way to do this, or using multiple >queries to delete selectively a week at a time based on the reception_time. >I would say there are a lot of duplicate entries between mid march to the >first week of April. > > > >---------------------------(end of broadcast)--------------------------- >TIP 8: explain analyze is your friend > >---------------------------(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 Fri Apr 16 14:54:28 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CDC81D1B8DF for ; Fri, 16 Apr 2004 14:54:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08447-08 for ; Fri, 16 Apr 2004 14:54:19 -0300 (ADT) Received: from silmaril.syscor.priv (h24-108-208-253.sbm.shawcable.net [24.108.208.253]) by svr1.postgresql.org (Postfix) with ESMTP id 26A87D1B8B7 for ; Fri, 16 Apr 2004 14:54:22 -0300 (ADT) Received: from syscor.com (strider.syscor.priv [192.168.1.3]) by silmaril.syscor.priv (8.12.8/8.12.8) with ESMTP id i3GHsaXR016571 for ; Fri, 16 Apr 2004 10:54:37 -0700 Message-ID: <40801E1B.4020401@syscor.com> Date: Fri, 16 Apr 2004 10:55:39 -0700 From: Ron St-Pierre User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Index Problem? References: <40800CC2.3050707@syscor.com> <200404161001.52989.josh@agliodbs.com> In-Reply-To: <200404161001.52989.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: 200404/221 X-Sequence-Number: 6521 Josh Berkus wrote: >Ron, > > > >>The emp table has 60 columns, all indexed, about two-thirds are numeric, >>but they are not affected by this update. The other 50+ columns are >>updated in the middle of the night and the amount of time that update >>takes isn't a concern. >> >> > >Well, I'd say that you have an application design problem, but that's not what >you asked for help with ;-) > > Yeah I agree but I'm not allowed to remove those indexes. > > >>Late last night I dumped the table, dropped it and re-created it from >>the dump (on the production server - when no one was looking). When I >>re-ran the function it took almost 11 minutes, which was pretty much in >>line with my results from the dev server. >> >> > >Sounds like you need to run a REINDEX on the table -- and after that, >dramatically increase your max_fsm_pages, and run lazy VACUUM immediately >after the batch update to clean up. > > > On my dev server I increased max_fsm_pages from the default of 20000 to 40000, increased checkpoint_segments from 3 to 5, and the function ran in about 6-7 minutes which is a nice increase. According to the docs "Annotated postgresql.conf and Global User Configuration (GUC) Guide" on varlena I'll have to re-start postgres for the changes to take effect there (correct?). Also the docs on Varlena show the max_fsm_pages default to be 10,000 but my default was 20,000, looks like that needs updating. Thanks for your help Josh, I'll see after the weekend what the impact the changes will have on the production server. Ron From pgsql-performance-owner@postgresql.org Fri Apr 16 14:56:26 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2A69DD1D08C for ; Fri, 16 Apr 2004 14:56:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 18137-02 for ; Fri, 16 Apr 2004 14:56:18 -0300 (ADT) Received: from mail.vistashare.com (www.vistashare.net [65.207.67.162]) by svr1.postgresql.org (Postfix) with ESMTP id 34ACAD1C9C8 for ; Fri, 16 Apr 2004 14:56:21 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by mail.vistashare.com (Postfix) with ESMTP id CC0A5D804; Fri, 16 Apr 2004 13:56:21 -0400 (EDT) From: Chris Kratz Organization: VistaShare To: pgsql-performance@postgresql.org Subject: Re: Long running queries degrade performance Date: Fri, 16 Apr 2004 13:56:20 -0400 User-Agent: KMail/1.6.2 Cc: Mike Nolan References: <200404161546.i3GFk3p3010490@gw.tssi.com> In-Reply-To: <200404161546.i3GFk3p3010490@gw.tssi.com> MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <200404161356.20210.chris.kratz@vistashare.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/222 X-Sequence-Number: 6522 Fairly sure, when it is happening, postgres usually is taking up the top slots for cpu usage as reported by top. Perhaps there is a better way to monitor this? The other thing for us is that others talk about disks being the bottleneck whereas for us it is almost always the processor. I expected the drives to kill us early on (we have two uw scsi mirrored drives) but there is very little disk activity. The disks rarely do much during load for us (at this point). Most likely this is related more to data volume at this point. As far as in your case, is there a lot of disk activity happening? More likely you have a situation where something else is happening which blocks the current thread. We ran into two situations recently which exhibited this behavior. One was adding and dropping tables in a transaction which blocks any other transaction trying to do the same. And two threads inserting records with the same primary key value blocks the second till the first finishes. Both of these were triggered by users double clicking links in our web app and were fixed by a better implementation. Perhaps something like that is causing what you are seeing. -Chris On Friday 16 April 2004 11:46 am, Mike Nolan wrote: > > We have a web app with a postgres backend. Most queries have subsecond > > response times through the web even with high usage. Every once in > > awhile someone will run either an ad-hoc query or some other long running > > db process. > > Are you sure it is postgres where the delay is occurring? I ask this > because I also have a web-based front end to postgres, and while most of > the time the queries respond in about a second every now and then I see > one that takes much longer, sometimes 10-15 seconds. > > I've seen this behavior on both my development system and on the > production server. > > The same query a while later might respond quickly again. > > I'm not sure where to look for the delay, either, and it is intermittent > enough that I'm not even sure what monitoring techniques to use. > -- > Mike Nolan -- Chris Kratz Systems Analyst/Programmer VistaShare LLC www.vistashare.com From pgsql-performance-owner@postgresql.org Fri Apr 16 16:10:42 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ADA72D1C9DA for ; Fri, 16 Apr 2004 16:10:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 45973-05 for ; Fri, 16 Apr 2004 16:10:29 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 966EED1C9BE for ; Fri, 16 Apr 2004 16:10:31 -0300 (ADT) Received: from [64.81.245.111] (HELO 192.168.1.102) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4876284; Fri, 16 Apr 2004 12:11:55 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Ron St-Pierre , pgsql-performance@postgresql.org Subject: Re: Index Problem? Date: Fri, 16 Apr 2004 12:10:47 -0700 User-Agent: KMail/1.5.4 References: <40800CC2.3050707@syscor.com> <200404161001.52989.josh@agliodbs.com> <40801E1B.4020401@syscor.com> In-Reply-To: <40801E1B.4020401@syscor.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404161210.47995.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/223 X-Sequence-Number: 6523 Ron, > Yeah I agree but I'm not allowed to remove those indexes. It's not the indexes I'm talking about, it's the table. > On my dev server I increased max_fsm_pages from the default of 20000 to > 40000, A better way to set this would be to run VACUUM VERBOSE ANALYZE right after doing one of your update batches, and see how many dead pages are being reclaimed, and then set max_fsm_pages to that # + 50% (or more). increased checkpoint_segments from 3 to 5, and the function ran > in about 6-7 minutes which is a nice increase. According to the docs > "Annotated postgresql.conf and Global User Configuration (GUC) Guide" on > varlena I'll have to re-start postgres for the changes to take effect > there (correct?). Correct. > Also the docs on Varlena show the max_fsm_pages > default to be 10,000 but my default was 20,000, looks like that needs > updating. I don't think the default has been changed. Anyone? -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Fri Apr 16 17:05:37 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2E4B2D1D25B for ; Fri, 16 Apr 2004 17:05:33 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 63140-04 for ; Fri, 16 Apr 2004 17:05:27 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 83612D1C9C8 for ; Fri, 16 Apr 2004 17:05:28 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GK5SlO002409; Fri, 16 Apr 2004 16:05:28 -0400 (EDT) To: josh@agliodbs.com Cc: Ron St-Pierre , pgsql-performance@postgresql.org Subject: Re: Index Problem? In-reply-to: <200404161210.47995.josh@agliodbs.com> References: <40800CC2.3050707@syscor.com> <200404161001.52989.josh@agliodbs.com> <40801E1B.4020401@syscor.com> <200404161210.47995.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Fri, 16 Apr 2004 12:10:47 -0700" Date: Fri, 16 Apr 2004 16:05:27 -0400 Message-ID: <2408.1082145927@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/224 X-Sequence-Number: 6524 Josh Berkus writes: > A better way to set this would be to run VACUUM VERBOSE ANALYZE right after > doing one of your update batches, and see how many dead pages are being > reclaimed, and then set max_fsm_pages to that # + 50% (or more). Actually, since he's running 7.4, there's an even better way. Do a "VACUUM VERBOSE" (full-database vacuum --- doesn't matter whether you ANALYZE or not). At the end of the very voluminous output, you'll see something like INFO: free space map: 240 relations, 490 pages stored; 4080 total pages needed DETAIL: Allocated FSM size: 1000 relations + 20000 pages = 178 kB shared memory. Here, I would need max_fsm_relations = 240 and max_fsm_pages = 4080 to exactly cover the present freespace needs of my system. I concur with the suggestion to bump that up a good deal, of course, but that gives you a real number to start from. The DETAIL part of the message shows my current settings (which are the defaults) and what the FSM is costing me in shared memory space. If you have multiple active databases, the best approach to getting these numbers is to VACUUM in each one, adding VERBOSE when you do the last one. The FSM report is cluster-wide but you want to be sure the underlying info is up to date for all databases. >> Also the docs on Varlena show the max_fsm_pages >> default to be 10,000 but my default was 20,000, looks like that needs >> updating. > I don't think the default has been changed. Anyone? Yes, I kicked it up for 7.4 because FSM covers indexes too now. Both the 7.3 and 7.4 defaults are pretty arbitrary of course... regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 17:25:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9A59FD1C9BE for ; Fri, 16 Apr 2004 17:25:17 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 63942-09 for ; Fri, 16 Apr 2004 17:25:12 -0300 (ADT) Received: from gw.tssi.com (gw.tssi.com [198.147.197.1]) by svr1.postgresql.org (Postfix) with ESMTP id 42430D1D6CC for ; Fri, 16 Apr 2004 17:25:13 -0300 (ADT) 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 i3GKPE8S013712; Fri, 16 Apr 2004 15:25:14 -0500 Received: (from nolan@localhost) by gw.tssi.com (8.12.6/8.12.6/Submit) id i3GKPDe4013710; Fri, 16 Apr 2004 15:25:13 -0500 From: Mike Nolan Message-Id: <200404162025.i3GKPDe4013710@gw.tssi.com> Subject: Re: Long running queries degrade performance To: chris.kratz@vistashare.com (Chris Kratz) Date: Fri, 16 Apr 2004 15:25:12 -0500 (CDT) Cc: pgsql-performance@postgresql.org In-Reply-To: <200404161356.20210.chris.kratz@vistashare.com> from "Chris Kratz" at Apr 16, 2004 01:56:20 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: 200404/225 X-Sequence-Number: 6525 > Fairly sure, when it is happening, postgres usually is taking up the top slots > for cpu usage as reported by top. Perhaps there is a better way to monitor > this? Given the intermittent nature of the problem and its relative brevity (5-10 seconds), I don't know whether top offers the granularity needed to locate the bottleneck. > likely you have a situation where something else is happening which blocks > the current thread. It happens on my development system, and I'm the only one on it. I know I've seen it on the production server, but I think it is a bit more common on the development server, though that may be a case of which system I spend the most time on. (Also, the production server is 1300 miles away with a DSL connection, so I may just be seeing network delays some of the time there.) > Both of these were triggered by users double clicking links in our > web app and were fixed by a better implementation. Perhaps something like > that is causing what you are seeing. My web app traps double-clicks in javascript and ignores all but the first one. That's because some of the users have mice that give double-clicks even when they only want one click. -- Mike Nolan From pgsql-performance-owner@postgresql.org Fri Apr 16 17:51:39 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 727E8D1D08C for ; Fri, 16 Apr 2004 17:51:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 71868-09 for ; Fri, 16 Apr 2004 17:51:26 -0300 (ADT) Received: from mail.vistashare.com (www.vistashare.net [65.207.67.162]) by svr1.postgresql.org (Postfix) with ESMTP id B7AF8D1C9C8 for ; Fri, 16 Apr 2004 17:51:28 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by mail.vistashare.com (Postfix) with ESMTP id 72B11D87B; Fri, 16 Apr 2004 16:51:30 -0400 (EDT) From: Chris Kratz Organization: VistaShare To: pgsql-performance@postgresql.org Subject: Re: Long running queries degrade performance Date: Fri, 16 Apr 2004 16:51:29 -0400 User-Agent: KMail/1.6.2 Cc: Mike Nolan References: <200404162025.i3GKPDe4013710@gw.tssi.com> In-Reply-To: <200404162025.i3GKPDe4013710@gw.tssi.com> MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <200404161651.29332.chris.kratz@vistashare.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/226 X-Sequence-Number: 6526 On Friday 16 April 2004 4:25 pm, Mike Nolan wrote: > Given the intermittent nature of the problem and its relative brevity > (5-10 seconds), I don't know whether top offers the granularity needed to > locate the bottleneck. Our long running processes run on the order of multiple minutes (sometimes for over an hour) and it's expected because the sql can be quite complex over somewhat large datasets. But it's the bringing the server to it's knees, that I'm trying to figure out how to address if we can. In other words, let those long running processes run, but somehow still get decent performance for "quick" requests. Yours reminds me of what used to happen in our apps back when I worked in java and the garbage collector kicked in. Suddenly everything would stop for 10-15s and then continue on. Sort of makes you think the app froze for some reason. > It happens on my development system, and I'm the only one on it. I know > I've seen it on the production server, but I think it is a bit more > common on the development server, though that may be a case of which system > I spend the most time on. (Also, the production server is 1300 miles away > with a DSL connection, so I may just be seeing network delays some of > the time there.) Interesting. Have you tried running a processor monitor and seeing if you are getting a cpu or disk spike when you get the blips? Postgres has been pretty constant for us in it's average runtime for any particular query. We do get some fluctuation, but I've always attributed that to other things happening in the background. I sometimes run gkrellm off the server just to "see" what's happening on a macro scale. It's a great early indicator when we are getting slammed one way or another (network, memory, processor, disk, etc). Plus it shows a couple of seconds of history so you can see blips pretty easily. > My web app traps double-clicks in javascript and ignores all but the first > one. That's because some of the users have mice that give double-clicks > even when they only want one click. Hmmm, never thought of doing that. Might be interesting to do something like that in a few key places where we have problems. > -- > Mike Nolan -- Chris Kratz Systems Analyst/Programmer VistaShare LLC www.vistashare.com From pgsql-performance-owner@postgresql.org Fri Apr 16 18:12:54 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 998F6D1BCB8 for ; Fri, 16 Apr 2004 18:12:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80086-09 for ; Fri, 16 Apr 2004 18:12:48 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 5C486D1B8B7 for ; Fri, 16 Apr 2004 18:12:50 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GLCqUM002851; Fri, 16 Apr 2004 17:12:52 -0400 (EDT) To: Chris Kratz Cc: pgsql-performance@postgresql.org Subject: Re: Long running queries degrade performance In-reply-to: <200404161128.00851.chris.kratz@vistashare.com> References: <200404161128.00851.chris.kratz@vistashare.com> Comments: In-reply-to Chris Kratz message dated "Fri, 16 Apr 2004 11:28:00 -0400" Date: Fri, 16 Apr 2004 17:12:52 -0400 Message-ID: <2850.1082149972@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/227 X-Sequence-Number: 6527 Chris Kratz writes: > ... Or if worse comes to worse to actually kill long running > processes without taking down the whole db as we have had to do on occasion. A quick "kill -INT" suffices to issue a query cancel, which I think is what you want here. You could also consider putting an upper limit on how long things can run by means of statement_timeout. Those are just band-aids though. Not sure about the underlying problem. Ordinarily I'd guess that the big-hog queries are causing trouble by evicting everything the other queries need from cache. But since your database fits in RAM, that doesn't seem to hold water. What PG version are you running? regards, tom lane From pgsql-performance-owner@postgresql.org Fri Apr 16 18:26:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8A4A7D1E298 for ; Fri, 16 Apr 2004 18:26:32 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92168-03 for ; Fri, 16 Apr 2004 18:26:28 -0300 (ADT) Received: from mail.vistashare.com (www.vistashare.net [65.207.67.162]) by svr1.postgresql.org (Postfix) with ESMTP id BDD12D1E2A7 for ; Fri, 16 Apr 2004 18:26:30 -0300 (ADT) Received: from localhost (localhost.localdomain [127.0.0.1]) by mail.vistashare.com (Postfix) with ESMTP id 38CF19586 for ; Fri, 16 Apr 2004 17:26:33 -0400 (EDT) From: Chris Kratz Organization: VistaShare To: pgsql-performance@postgresql.org Subject: Re: Long running queries degrade performance Date: Fri, 16 Apr 2004 17:26:32 -0400 User-Agent: KMail/1.6.2 References: <200404161128.00851.chris.kratz@vistashare.com> <2850.1082149972@sss.pgh.pa.us> In-Reply-To: <2850.1082149972@sss.pgh.pa.us> MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <200404161726.32102.chris.kratz@vistashare.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/228 X-Sequence-Number: 6528 On Friday 16 April 2004 5:12 pm, Tom Lane wrote: > Chris Kratz writes: > > ... Or if worse comes to worse to actually kill long running > > processes without taking down the whole db as we have had to do on > > occasion. > > A quick "kill -INT" suffices to issue a query cancel, which I think is > what you want here. You could also consider putting an upper limit on > how long things can run by means of statement_timeout. Wow, that's exactly what I've been looking for. I thought I had scoured the manuals, but must have missed that one. I need to think about the statement_timeout, the might be a good idea to use as well. > Those are just band-aids though. Not sure about the underlying problem. > Ordinarily I'd guess that the big-hog queries are causing trouble by > evicting everything the other queries need from cache. But since your > database fits in RAM, that doesn't seem to hold water. That makes some sense, perhaps there is some other cache somewhere that is causing the problems. I am doing some tuning and have set the following items in our postgresql.conf: shared_buffers = 4096 max_fsm_relations = 1000 max_fsm_pages = 20000 sort_mem = 2048 effective_cache_size = 64000 I believe these are the only performance related items we've modified. One thing I did today, since we seem to run about 600M of memory available for file caches. The effective cache size used to be much lower, so perhaps that was causing some of the problems. > What PG version are you running? 7.3.4 with grand hopes to move to 7.4 this summer. > regards, tom lane -- Chris Kratz Systems Analyst/Programmer VistaShare LLC www.vistashare.com From pgsql-performance-owner@postgresql.org Fri Apr 16 18:45:33 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A97D5D1D803 for ; Fri, 16 Apr 2004 18:45:31 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 92168-09 for ; Fri, 16 Apr 2004 18:45:27 -0300 (ADT) Received: from web13125.mail.yahoo.com (web13125.mail.yahoo.com [216.136.174.143]) by svr1.postgresql.org (Postfix) with SMTP id 8289CD1B515 for ; Fri, 16 Apr 2004 18:45:28 -0300 (ADT) Message-ID: <20040416214529.10468.qmail@web13125.mail.yahoo.com> Received: from [63.78.248.48] by web13125.mail.yahoo.com via HTTP; Fri, 16 Apr 2004 14:45:29 PDT Date: Fri, 16 Apr 2004 14:45:29 -0700 (PDT) From: Litao Wu Subject: sunquery and estimated rows To: pgsql-performance@postgresql.org MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/229 X-Sequence-Number: 6529 Hi, When I included a subquery, the estimated rows (1240) is way too high as shown in the following example. Can someone explain why? Because of this behavior, some of our queries use hash join instead of nested loop. Thanks, select version(); version ------------------------------------------------------------- PostgreSQL 7.3.4 on i686-pc-linux-gnu, compiled by GCC 2.96 (1 row) \d test Table "public.test" Column | Type | Modifiers ---------+--------------------------+----------- id | integer | name | character varying(255) | d_id | integer | c_id | integer | r_id | integer | u_id | integer | scope | integer | active | integer | created | timestamp with time zone | typ | integer | Indexes: test_scope_idx btree (scope) reindex table test; vacuum full analyze test; select count(*) from test; count ------- 4959 (1 row) select count(*) from test where scope=10; count ------- 10 (1 row) explain analyze select * from test where scope=10; -- so far so good, estimate 12 rows, actual 10 rows QUERY PLAN ---------------------------------------------------------------------------------------------------------------------- Index Scan using test_scope_idx on test (cost=0.00..4.35 rows=12 width=59) (actual time=0.04..0.11 rows=10 loops=1) Index Cond: (scope = 10) Total runtime: 0.23 msec (3 rows) explain analyze select * from test where scope=(select 10); -- estimate rows is way too high, do not why???? QUERY PLAN ------------------------------------------------------------------------------------------------------------------------- Index Scan using test_scope_idx on test (cost=0.00..40.74 rows=1240 width=59) (actual time=0.06..0.13 rows=10 loops=1) Index Cond: (scope = $0) InitPlan -> Result (cost=0.00..0.01 rows=1 width=0) (actual time=0.01..0.01 rows=1 loops=1) Total runtime: 0.22 msec (5 rows) __________________________________ Do you Yahoo!? Yahoo! Tax Center - File online by April 15th http://taxes.yahoo.com/filing.html From pgsql-performance-owner@postgresql.org Fri Apr 16 18:53:38 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C81C6D1E993 for ; Fri, 16 Apr 2004 18:53:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01765-04 for ; Fri, 16 Apr 2004 18:53:33 -0300 (ADT) Received: from silmaril.syscor.priv (h24-108-208-253.sbm.shawcable.net [24.108.208.253]) by svr1.postgresql.org (Postfix) with ESMTP id B6110D1E981 for ; Fri, 16 Apr 2004 18:53:34 -0300 (ADT) Received: from syscor.com (strider.syscor.priv [192.168.1.3]) by silmaril.syscor.priv (8.12.8/8.12.8) with ESMTP id i3GLrqXR017338 for ; Fri, 16 Apr 2004 14:53:53 -0700 Message-ID: <4080562F.6050006@syscor.com> Date: Fri, 16 Apr 2004 14:54:55 -0700 From: Ron St-Pierre User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624 X-Accept-Language: en-us, en MIME-Version: 1.0 To: pgsql-performance@postgresql.org Subject: Re: Index Problem? References: <40800CC2.3050707@syscor.com> <200404161001.52989.josh@agliodbs.com> <40801E1B.4020401@syscor.com> <200404161210.47995.josh@agliodbs.com> <2408.1082145927@sss.pgh.pa.us> In-Reply-To: <2408.1082145927@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: 200404/230 X-Sequence-Number: 6530 Tom Lane wrote: >Josh Berkus writes: > > >>A better way to set this would be to run VACUUM VERBOSE ANALYZE right after >>doing one of your update batches, and see how many dead pages are being >>reclaimed, and then set max_fsm_pages to that # + 50% (or more). >> >> > >Actually, since he's running 7.4, there's an even better way. Do a >"VACUUM VERBOSE" (full-database vacuum --- doesn't matter whether you >ANALYZE or not). At the end of the very voluminous output, you'll see >something like > >INFO: free space map: 240 relations, 490 pages stored; 4080 total pages needed >DETAIL: Allocated FSM size: 1000 relations + 20000 pages = 178 kB shared memory. > >Here, I would need max_fsm_relations = 240 and max_fsm_pages = 4080 to >exactly cover the present freespace needs of my system. I concur with >the suggestion to bump that up a good deal, of course, but that gives >you a real number to start from. > >The DETAIL part of the message shows my current settings (which are the >defaults) and what the FSM is costing me in shared memory space. > > > Okay, after running the function VACUUM VERBOSE is telling me: INFO: free space map: 136 relations, 25014 pages stored; 22608 total pages needed DETAIL: Allocated FSM size: 1000 relations + 20000 pages = 178 kB shared memory. My max_fsm_pages was set to 20,000 and I reset it to 40,000 on the dev server and the function ran about 20-30% faster, so I'll try the same on the production server. Thanks for the analysis of the VACUUM info. Ron From pgsql-performance-owner@postgresql.org Fri Apr 16 19:24:11 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 20452D1DC5E for ; Fri, 16 Apr 2004 19:24:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 04663-10 for ; Fri, 16 Apr 2004 19:24:03 -0300 (ADT) Received: from email03.aon.at (warsl404pip6.highway.telekom.at [195.3.96.89]) by svr1.postgresql.org (Postfix) with SMTP id F103ED1DB96 for ; Fri, 16 Apr 2004 19:24:04 -0300 (ADT) Received: (qmail 365408 invoked from network); 16 Apr 2004 22:24:06 -0000 Received: from m154p002.dipool.highway.telekom.at (HELO PASCAL) ([62.46.9.34]) (envelope-sender ) by qmail3rs.highway.telekom.at (qmail-ldap-1.03) with SMTP for ; 16 Apr 2004 22:24:06 -0000 From: Manfred Koizar To: Tom Lane Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats Date: Sat, 17 Apr 2004 00:26:22 +0200 Message-ID: References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> <22642.1082074729@sss.pgh.pa.us> <29060.1082126089@sss.pgh.pa.us> In-Reply-To: <29060.1082126089@sss.pgh.pa.us> X-Mailer: Forte Agent 1.93/32.576 English (American) 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: 200404/231 X-Sequence-Number: 6531 On Fri, 16 Apr 2004 10:34:49 -0400, Tom Lane wrote: >> p = prod from{i = 0} to{n - 1} {{c(B - i)} over {cB - i}} > >So? You haven't proven that either sampling method fails to do the >same. On the contrary, I believe that above formula is more or less valid for both methods. The point is in what I said next: | This probability grows with increasing B. For the one-stage sampling method B is the number of pages of the whole table. With two-stage sampling we have to use n instead of B and get a smaller probability (for n < B, of course). So this merely shows that the two sampling methods are not equivalent. >The desired property can also be phrased as "every tuple should be >equally likely to be included in the final sample". Only at first sight. You really expect more from random sampling. Otherwise I'd just put one random tuple and its n - 1 successors (modulo N) into the sample. This satisfies your condition but you wouldn't call it a random sample. Random sampling is more like "every possible sample is equally likely to be collected", and two-stage sampling doesn't satisfy this condition. But if in your opinion the difference is not significant, I'll stop complaining against my own idea. Is there anybody else who cares? >You could argue that a tuple on a heavily populated page is >statistically likely to see a higher T when it's part of the page sample >pool than a tuple on a near-empty page is likely to see, and therefore >there is some bias against selection of the former tuple. But given a >sample over a reasonably large number of pages, the contribution of any >one page to T should be fairly small and so this effect ought to be >small. It is even better: Storing a certain number of tuples on heavily populated pages takes less pages than to store them on sparsely populated pages (due to tuple size or to dead tuples). So heavily populated pages are less likely to be selected in stage one, and this exactly offsets the effect of increasing T. >So I think this method is effectively unbiased at the tuple level. Servus Manfred From pgsql-performance-owner@postgresql.org Fri Apr 16 19:57:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8A120D1DACE for ; Fri, 16 Apr 2004 19:57:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 18431-06 for ; Fri, 16 Apr 2004 19:57:48 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id BBD34D1D5E4 for ; Fri, 16 Apr 2004 19:57:49 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 642E417C490; Fri, 16 Apr 2004 18:57:52 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BEcHQ-0000Qj-00; Fri, 16 Apr 2004 18:57:52 -0400 To: "Jim C. Nasby" Cc: pgsql-performance@postgresql.org Subject: Re: Poor performance of group by query References: <20040416151706.GE87362@nasby.net> In-Reply-To: <20040416151706.GE87362@nasby.net> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 16 Apr 2004 18:57:51 -0400 Message-ID: <87pta78r9s.fsf@stark.xeocode.com> Lines: 24 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: 200404/232 X-Sequence-Number: 6532 > stats=# explain analyze SELECT work_units, min(raw_rank) AS rank FROM Trank_work_overall GROUP BY work_units; > > ... > > raw_rank | bigint | > work_units | bigint | If you create a copy of the same table using regular integers does that run fast? And a copy of the table using bigints is still slow like the original? I know bigints are less efficient than integers because they're handled using dynamically allocated memory. This especially bites aggregate functions. But I don't see why it would be any slower for a hash aggregate than a regular aggregate. It's a pretty gross amount of time for 18k records. There was a thought a while back about making 64-bit machines handle 64-bit datatypes like bigints without pointers. That would help on your Opteron. -- greg From pgsql-performance-owner@postgresql.org Fri Apr 16 20:45:02 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9F15ED1E2A7 for ; Fri, 16 Apr 2004 20:45:00 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28618-06 for ; Fri, 16 Apr 2004 20:44:57 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 9E292D1D085 for ; Fri, 16 Apr 2004 20:44:58 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3GNj1lS003803; Fri, 16 Apr 2004 19:45:02 -0400 (EDT) To: Litao Wu Cc: pgsql-performance@postgresql.org Subject: Re: sunquery and estimated rows In-reply-to: <20040416214529.10468.qmail@web13125.mail.yahoo.com> References: <20040416214529.10468.qmail@web13125.mail.yahoo.com> Comments: In-reply-to Litao Wu message dated "Fri, 16 Apr 2004 14:45:29 -0700" Date: Fri, 16 Apr 2004 19:45:01 -0400 Message-ID: <3802.1082159101@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/233 X-Sequence-Number: 6533 Litao Wu writes: > When I included a subquery, the estimated rows (1240) > is way too high as shown in the following example. > select * from test > where scope=(select 10); The planner sees that as "where scope = " and falls back to a default estimate. It won't simplify a sub-select to a constant. (Some people consider that a feature ;-).) The estimate should still be derived from the statistics for the scope column, but it will just depend on the number of distinct values for the column and not on the specific comparison constant. regards, tom lane From pgsql-performance-owner@postgresql.org Sat Apr 17 10:12:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B9782D1D05C for ; Sat, 17 Apr 2004 10:12:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 12458-10 for ; Sat, 17 Apr 2004 10:12:24 -0300 (ADT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id B3236D1C9C8 for ; Sat, 17 Apr 2004 10:12:20 -0300 (ADT) 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 i3HDCKSx087303 for ; Sat, 17 Apr 2004 13:12:20 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id i3HCoLtF083930 for pgsql-performance@postgresql.org; Sat, 17 Apr 2004 12:50:21 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.performance Subject: Re: Long running queries degrade performance Date: Sat, 17 Apr 2004 07:59:23 -0400 Organization: cbbrowne Computing Inc Lines: 69 Message-ID: References: <200404161128.00851.chris.kratz@vistashare.com> <200404161546.i3GFk3p3010490@gw.tssi.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Complaints-To: usenet@news.hub.org X-message-flag: Outlook is rather hackable, isn't it? X-Home-Page: http://www.cbbrowne.com/info/ X-Affero: http://svcs.affero.net/rm.php?r=cbbrowne User-Agent: Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Security Through Obscurity, linux) Cancel-Lock: sha1:42F111rf9O1YatkhZzQ2UaW2rSY= To: pgsql-performance@postgresql.org X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/234 X-Sequence-Number: 6534 A long time ago, in a galaxy far, far away, nolan@gw.tssi.com (Mike Nolan) wrote: >> We have a web app with a postgres backend. Most queries have subsecond >> response times through the web even with high usage. Every once in awhile >> someone will run either an ad-hoc query or some other long running db >> process. > > Are you sure it is postgres where the delay is occurring? I ask this > because I also have a web-based front end to postgres, and while most of > the time the queries respond in about a second every now and then I see > one that takes much longer, sometimes 10-15 seconds. > > I've seen this behavior on both my development system and on the > production server. > > The same query a while later might respond quickly again. > > I'm not sure where to look for the delay, either, and it is > intermittent enough that I'm not even sure what monitoring > techniques to use. Well, a first thing to do is to see what query plans get set up for the queries. If the plans are varying over time, that suggests something's up with ANALYZEs. If the plans look a bit questionable, then you may be encountering the situation where cache is helping you on the _second_ query but not the first. I did some tuning yesterday involving the same sort of "symptoms," and that turned out to be what was happening. I'll describe (in vague detail ;-)) what I was seeing. - The table being queried was a "transaction" table, containing tens of thousands of records per day. - The query was pulling summary information about one or another customer's activity on that day. - The best index we had was on transaction date. Thus, the query would walk through the "txn date" index, pulling records into memory, and filtering them against the other selection criteria. The table is big, so that data is pretty widely scattered across many pages. The _first_ time the query is run, the data is all out on disk, and there are hundreds-to-thousands of page reads to collect it all. That took 10-15 seconds. The _second_ time it was run (as well as subsequent occasions), those pages were all in cache, so the query runs in under a second. What I wound up doing was to add an index on transaction date and customer ID, so that a query that specifies both criteria will look just for the few hundred (at most) records relevant to a particular customer. That's fast even the first time around. We had a really useful "hook" on this one because the developer noticed that the first time he queried for a particular day, it was slow. We could "repeat" the test easily by just changing to a day that we hadn't pulled into cache yet. -- (format nil "~S@~S" "cbbrowne" "acm.org") http://cbbrowne.com/info/lisp.html Referring to undocumented private communications allows one to claim virtually anything: "we discussed this idea in our working group last year, and concluded that it was totally brain-damaged". -- from the Symbolics Guidelines for Sending Mail From pgsql-performance-owner@postgresql.org Tue Apr 20 10:58:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8AC84D1DB96 for ; Sat, 17 Apr 2004 09:25:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 11282-02 for ; Sat, 17 Apr 2004 09:25:45 -0300 (ADT) Received: from mailhost2.tudelft.nl (mailhost2.tudelft.nl [130.161.180.2]) by svr1.postgresql.org (Postfix) with ESMTP id BCAD9D1DB17 for ; Sat, 17 Apr 2004 09:25:41 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by rav.antivirus (Postfix) with ESMTP id 8ACF7213A0; Sat, 17 Apr 2004 14:25:44 +0200 (MEST) Received: from listserv.tudelft.nl (listserv.tudelft.nl [130.161.180.33]) by mailhost2.tudelft.nl (Postfix) with ESMTP id 54BA9212F9; Sat, 17 Apr 2004 14:25:44 +0200 (MEST) Received: from oli.tudelft.nl (jochemd.tnw-s.tudelft.nl [145.94.90.156]) by listserv.tudelft.nl (8.12.10/8.12.8) with ESMTP id i3HCPfF7009695; Sat, 17 Apr 2004 14:25:43 +0200 (MEST) Message-ID: <40812244.5080001@oli.tudelft.nl> Date: Sat, 17 Apr 2004 14:25:40 +0200 From: Jochem van Dieten Organization: OnLine Internet User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: rstpierre@syscor.com Cc: pgsql-performance@postgresql.org Subject: Re: Index Problem? References: <40800CC2.3050707@syscor.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at tudelft.nl X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/299 X-Sequence-Number: 6599 Ron St-Pierre wrote: > I am using postgres 7.4.1 and have a problem with a plpgsql function. > When I run the function on the production server it takes approx 33 > minutes to run. I dumped the DB and copied it to a similarly configured > box and ran the function and it ran in about 10 minutes. Can anyone > offer advice on tuning the function or my database? Here are the > lengthy, gory details. > > F u n c t i o n > It updates seven columns of a table 1 to 4 times daily. Current data = > 42,000 rows, new data = 30,000 rows. > > CREATE TYPE employeeType AS (empID INTEGER, updateDate DATE, bDate > INTEGER, val1 NUMERIC, val2 NUMERIC, val3 NUMERIC, val4 NUMERIC, favNum > NUMERIC); > > CREATE OR REPLACE FUNCTION updateEmployeeData() RETURNS SETOF > employeeType AS ' > DECLARE > rec RECORD; > BEGIN > FOR rec IN SELECT empID, updateDate, bDate, val1, val2, val3, val4, favNum FROM newData LOOP > RETURN NEXT rec; > UPDATE currentData SET val1=rec.val1, val2=rec.val2, val3=rec.val2, val4=rec.val4, favNum=rec.favNum, updateDate=rec.updateDate > WHERE empID=rec.empID; > END LOOP; > RETURN; > END; > ' LANGUAGE 'plpgsql'; Can't you handle this with a simple update query? UPDATE currentData SET val1 = newData.val1, val2 = newData.val2, val3 = newData.val3, val4 = newData.val4, favNum = newData.favNum, updateDate = newData.updateDate FROM newData WHERE newDate.empID = currentData.empID Jochem -- I don't get it immigrants don't work and steal our jobs - Loesje From pgsql-performance-owner@postgresql.org Sat Apr 17 11:29:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 95C2BD1D05C for ; Sat, 17 Apr 2004 11:29:01 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33130-09 for ; Sat, 17 Apr 2004 11:29:04 -0300 (ADT) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id 75B56D1C9DA for ; Sat, 17 Apr 2004 11:28:59 -0300 (ADT) Received: (qmail 77089 invoked from network); 17 Apr 2004 14:29:07 -0000 Received: from unknown (HELO ?192.168.0.6?) (10.10.10.177) by 10.10.10.10 with SMTP; 17 Apr 2004 14:29:07 -0000 In-Reply-To: <407F9816.2080407@trade-india.com> References: <407D7A89.60508@trade-india.com> <407E8FF6.4050601@potentialtech.com> <407EB605.70502@trade-india.com> <200404151844.33672.dev@archonet.com> <407F9816.2080407@trade-india.com> Mime-Version: 1.0 (Apple Message framework v613) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <954A8D6A-907B-11D8-8E5D-000D9366F0C4@torgo.978.org> Content-Transfer-Encoding: 7bit Cc: Richard Huxton , Bill Moran , Postgres Performance From: Jeff Subject: Re: [ SOLVED ] select count(*) very slow on an already Date: Sat, 17 Apr 2004 10:29:06 -0400 To: Rajesh Kumar Mallah X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/235 X-Sequence-Number: 6535 On Apr 16, 2004, at 4:23 AM, Rajesh Kumar Mallah wrote: > > > I am running an update on the same table > > update rfis set inquiry_status='APPROVED' where inquiry_status='a'; > > Its running for past 20 mins. and top output is below. > The PID which is executing the query above is 6712. Can anyone > tell me why it is in an uninterruptable sleep and does it relate > to the apparent poor performance? Is it problem with the disk > hardware. I know at nite this query will run reasonably fast. > I've had this problem recently. The problem is simply that the disk cannot keep up. Most likely you don't see it at night because traffic is lower. There are only 2 solutions: 1. get more disks 2. write to the db less The machine I was running on had a single(!) disk. It was a quad xeon so there was plenty of cpu. I'd see 8-9 processes stuck in the "D" state. Doing a simple ls -l somefile would take 10-15 seconds and of course, db performance was abysmal. I had a lowly P2 with a few disks in it that was able to run circles around it for the simple fact the machine was not waiting for disk. Again, proof that disk is far more important than CPU in a db. good luck. -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Sat Apr 17 13:08:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DC171D1E8FB for ; Sat, 17 Apr 2004 13:08:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 62629-01 for ; Sat, 17 Apr 2004 13:08:18 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6B936D1E8A5 for ; Sat, 17 Apr 2004 13:08:17 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3HG8Fik010604; Sat, 17 Apr 2004 12:08:16 -0400 (EDT) To: "Jim C. Nasby" Cc: pgsql-performance@postgreSQL.org Subject: Re: Horribly slow hash join In-reply-to: <20040416184541.GK87362@nasby.net> References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> <808.1082135098@sss.pgh.pa.us> <20040416172523.GJ87362@nasby.net> <1263.1082138383@sss.pgh.pa.us> <20040416184541.GK87362@nasby.net> Comments: In-reply-to "Jim C. Nasby" message dated "Fri, 16 Apr 2004 13:45:41 -0500" Date: Sat, 17 Apr 2004 12:08:15 -0400 Message-ID: <10603.1082218095@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/236 X-Sequence-Number: 6536 [ resending because I fat-fingered the cc: to the list ] I see the problem: all the entries in your work_units column have the low 32 bits equal to zero. regression=# select distinct work_units % (2^32)::bigint from Trank_work_overall; ?column? ---------- 0 (1 row) The hash function for int8 only takes the low word into account, so all of the entries end up on the same hash chain, resulting in worst-case behavior. This applies to both your hash join and hash aggregate cases. We could change the hash function, perhaps, but then we'd just have different cases where there's a problem ... hashing will always fail on *some* set of inputs. (Also, I have been harboring some notions of supporting cross-type hash joins for integer types, which will not work unless small int8 values hash the same as int4 etc.) I guess the real issue is why are you encoding work_units like that? regards, tom lane From pgsql-performance-owner@postgresql.org Sat Apr 17 17:35:30 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 20B6AD1BCB8 for ; Sat, 17 Apr 2004 17:35:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27509-01 for ; Sat, 17 Apr 2004 17:35:27 -0300 (ADT) Received: from mx2.mundo-r.com (mx2.mundo-r.com [212.51.32.186]) by svr1.postgresql.org (Postfix) with ESMTP id F2F9ED1BB49 for ; Sat, 17 Apr 2004 17:35:24 -0300 (ADT) Received: from lobeira (cm248132.red.mundo-r.com [213.60.248.132]) by obelix.mundo-r.ggc (iPlanet Messaging Server 5.1 HotFix 1.6 (built Oct 18 2002)) with ESMTP id <0HWC00CG016MG1@obelix.mundo-r.ggc> for pgsql-performance@postgreSQL.org; Sat, 17 Apr 2004 22:35:26 +0200 (MEST) Date: Sat, 17 Apr 2004 22:35:09 +0200 From: =?iso-8859-1?Q?Marcos_Mart=EDnez=28R=29?= Subject: Re: Horribly slow hash join To: Tom Lane , "Jim C. Nasby" Cc: pgsql-performance@postgreSQL.org Message-id: <001001c424bb$7fb83740$0564a8c0@lobeira> MIME-version: 1.0 X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2800.1409 X-Mailer: Microsoft Outlook Express 6.00.2800.1409 Content-type: text/plain; charset=iso-8859-1 Content-transfer-encoding: 7BIT X-Priority: 3 X-MSMail-priority: Normal References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> <808.1082135098@sss.pgh.pa.us> <20040416172523.GJ87362@nasby.net> <1263.1082138383@sss.pgh.pa.us> <20040416184541.GK87362@nasby.net> <10603.1082218095@sss.pgh.pa.us> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/237 X-Sequence-Number: 6537 I didn't follow the conversation from the begining, bu I imagine that you could improve performance using the value (work_units % (2^32) ) instead of work_units. You could even make an index on this value. Like that, the HASH function will work well. This is not a good solution, but ... For example. create index ind1 on table1 ( work_units % (2^32) ); create index ind1 on table2 ( work_units % (2^32) ); Select * from table1 join table2 on (table1.work_units % (2^32) ) = (table2.work_units % (2^32) ) ----- Original Message ----- From: "Tom Lane" To: "Jim C. Nasby" Cc: Sent: Saturday, April 17, 2004 6:08 PM Subject: Re: [PERFORM] Horribly slow hash join > [ resending because I fat-fingered the cc: to the list ] > > I see the problem: all the entries in your work_units column have the > low 32 bits equal to zero. > > regression=# select distinct work_units % (2^32)::bigint from Trank_work_overall; > ?column? > ---------- > 0 > (1 row) > > The hash function for int8 only takes the low word into account, so all > of the entries end up on the same hash chain, resulting in worst-case > behavior. This applies to both your hash join and hash aggregate cases. > > We could change the hash function, perhaps, but then we'd just have > different cases where there's a problem ... hashing will always fail on > *some* set of inputs. (Also, I have been harboring some notions of > supporting cross-type hash joins for integer types, which will not work > unless small int8 values hash the same as int4 etc.) > > I guess the real issue is why are you encoding work_units like that? > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 6: Have you searched our list archives? > > http://archives.postgresql.org From pgsql-performance-owner@postgresql.org Sat Apr 17 20:04:42 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 561CBD1C4C3 for ; Sat, 17 Apr 2004 20:04:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 59865-01 for ; Sat, 17 Apr 2004 20:04:40 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id BE560D1BCB8 for ; Sat, 17 Apr 2004 20:04:37 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 7139B17C431; Sat, 17 Apr 2004 19:04:40 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BEyrY-0004mf-00; Sat, 17 Apr 2004 19:04:40 -0400 To: Tom Lane Cc: "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> <808.1082135098@sss.pgh.pa.us> <20040416172523.GJ87362@nasby.net> <1263.1082138383@sss.pgh.pa.us> <20040416184541.GK87362@nasby.net> <10603.1082218095@sss.pgh.pa.us> In-Reply-To: <10603.1082218095@sss.pgh.pa.us> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 17 Apr 2004 19:04:39 -0400 Message-ID: <878ygu8auw.fsf@stark.xeocode.com> 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: 200404/238 X-Sequence-Number: 6538 Tom Lane writes: > We could change the hash function, perhaps, but then we'd just have > different cases where there's a problem ... hashing will always fail on > *some* set of inputs. Sure, but completely ignoring part of the input seems like an unfortunate choice of hash function. > (Also, I have been harboring some notions of supporting cross-type hash > joins for integer types, which will not work unless small int8 values hash > the same as int4 etc.) The obvious way to modify the hash function is to xor the high 32 bits with the low 32 bits. That maintains the property you need and at least ensures that all the bits are taken into account. -- greg From pgsql-performance-owner@postgresql.org Sun Apr 18 00:34:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B47CCD1E94D for ; Sun, 18 Apr 2004 00:34:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 10187-07 for ; Sun, 18 Apr 2004 00:34:15 -0300 (ADT) Received: from linda-1.paradise.net.nz (bm-1a.paradise.net.nz [202.0.58.20]) by svr1.postgresql.org (Postfix) with ESMTP id 61C65D1E949 for ; Sun, 18 Apr 2004 00:34:14 -0300 (ADT) Received: from smtp-2.paradise.net.nz (smtp-2b.paradise.net.nz [202.0.32.211]) by linda-1.paradise.net.nz (Paradise.net.nz) with ESMTP id <0HWC002VOKL0AQ@linda-1.paradise.net.nz> for pgsql-performance@postgresql.org; Sun, 18 Apr 2004 15:34:13 +1200 (NZST) Received: from paradise.net.nz (203-96-145-144.adsl.paradise.net.nz [203.96.145.144]) by smtp-2.paradise.net.nz (Postfix) with ESMTP id 36AEB9E29A; Sun, 18 Apr 2004 15:34:12 +1200 (NZST) Date: Sun, 18 Apr 2004 15:35:58 +1200 From: Mark Kirkwood Subject: Re: Poor performance of group by query In-reply-to: <20040416151706.GE87362@nasby.net> To: "Jim C. Nasby" Cc: pgsql-performance@postgresql.org Message-id: <4081F79E.6070409@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:1.6) Gecko/20040404 References: <20040416151706.GE87362@nasby.net> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/239 X-Sequence-Number: 6539 It might be worth trying out a build with -O2, just to rule out any -O3 oddness. regards Mark Jim C. Nasby wrote: > PostgreSQL 7.4.2 compiled with -O3. > > > > From pgsql-performance-owner@postgresql.org Sun Apr 18 00:46:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E9F60D1DCF5 for ; Sun, 18 Apr 2004 00:46:04 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15239-01 for ; Sun, 18 Apr 2004 00:46:02 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 9297DD1DC7D for ; Sun, 18 Apr 2004 00:46:01 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3I3jxVJ015054; Sat, 17 Apr 2004 23:45:59 -0400 (EDT) To: Greg Stark Cc: "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: <878ygu8auw.fsf@stark.xeocode.com> References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> <808.1082135098@sss.pgh.pa.us> <20040416172523.GJ87362@nasby.net> <1263.1082138383@sss.pgh.pa.us> <20040416184541.GK87362@nasby.net> <10603.1082218095@sss.pgh.pa.us> <878ygu8auw.fsf@stark.xeocode.com> Comments: In-reply-to Greg Stark message dated "17 Apr 2004 19:04:39 -0400" Date: Sat, 17 Apr 2004 23:45:58 -0400 Message-ID: <15053.1082259958@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/240 X-Sequence-Number: 6540 Greg Stark writes: > Tom Lane writes: >> (Also, I have been harboring some notions of supporting cross-type hash >> joins for integer types, which will not work unless small int8 values hash >> the same as int4 etc.) > The obvious way to modify the hash function is to xor the high 32 bits with > the low 32 bits. That maintains the property you need No it doesn't ... regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 03:18:51 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EF406D1E298 for ; Sun, 18 Apr 2004 03:18:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 38135-10 for ; Sun, 18 Apr 2004 03:18:49 -0300 (ADT) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id DCD2AD1E2A7 for ; Sun, 18 Apr 2004 03:18:47 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id A74D888E9; Sun, 18 Apr 2004 08:18:47 +0200 (CEST) Date: Sun, 18 Apr 2004 08:18:47 +0200 (CEST) From: Dennis Bjorklund To: Tom Lane Cc: "Jim C. Nasby" , Subject: Re: Horribly slow hash join In-Reply-To: <10603.1082218095@sss.pgh.pa.us> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/241 X-Sequence-Number: 6541 On Sat, 17 Apr 2004, Tom Lane wrote: > *some* set of inputs. (Also, I have been harboring some notions of > supporting cross-type hash joins for integer types, which will not work > unless small int8 values hash the same as int4 etc.) The simple solution would be to always extend integers to 64 bits (or whatever the biggest integer is) before calculating the hash. It makes the hash function a little slower for smaller types, but it's mostly an operation in the cpu and no memory involved, so it's probably not noticable. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Sun Apr 18 03:43:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 00F05D1DC7D for ; Sun, 18 Apr 2004 03:43:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48056-03 for ; Sun, 18 Apr 2004 03:43:10 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 436D0D1D300 for ; Sun, 18 Apr 2004 03:43:09 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id D9A7417C144; Sun, 18 Apr 2004 02:43:09 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BF61F-0006Mm-00; Sun, 18 Apr 2004 02:43:09 -0400 To: Tom Lane Cc: Greg Stark , "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> <808.1082135098@sss.pgh.pa.us> <20040416172523.GJ87362@nasby.net> <1263.1082138383@sss.pgh.pa.us> <20040416184541.GK87362@nasby.net> <10603.1082218095@sss.pgh.pa.us> <878ygu8auw.fsf@stark.xeocode.com> <15053.1082259958@sss.pgh.pa.us> In-Reply-To: <15053.1082259958@sss.pgh.pa.us> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 18 Apr 2004 02:43:09 -0400 Message-ID: <873c719476.fsf@stark.xeocode.com> Lines: 23 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: 200404/242 X-Sequence-Number: 6542 Tom Lane writes: > Greg Stark writes: > > Tom Lane writes: > >> (Also, I have been harboring some notions of supporting cross-type hash > >> joins for integer types, which will not work unless small int8 values hash > >> the same as int4 etc.) > > > The obvious way to modify the hash function is to xor the high 32 bits with > > the low 32 bits. That maintains the property you need > > No it doesn't ... Eh? Oh, negative numbers? So low^high^sign. I wonder if it makes sense to have check the hash distribution after generating the table and if it's bad then throw it away and try again with a different hash function. The "different hash function" would probably just be a seed value changing. Probably way overkill though. -- greg From pgsql-performance-owner@postgresql.org Sun Apr 18 12:39:15 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 53D18D1DCF5 for ; Sun, 18 Apr 2004 12:39:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 62249-05 for ; Sun, 18 Apr 2004 12:39:13 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 65CF0D1DCC0 for ; Sun, 18 Apr 2004 12:39:12 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3IFdApS019682; Sun, 18 Apr 2004 11:39:10 -0400 (EDT) To: Dennis Bjorklund Cc: "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: References: Comments: In-reply-to Dennis Bjorklund message dated "Sun, 18 Apr 2004 08:18:47 +0200" Date: Sun, 18 Apr 2004 11:39:10 -0400 Message-ID: <19681.1082302750@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/243 X-Sequence-Number: 6543 Dennis Bjorklund writes: > On Sat, 17 Apr 2004, Tom Lane wrote: >> *some* set of inputs. (Also, I have been harboring some notions of >> supporting cross-type hash joins for integer types, which will not work >> unless small int8 values hash the same as int4 etc.) > The simple solution would be to always extend integers to 64 bits (or > whatever the biggest integer is) before calculating the hash. That creates portability issues though. We do not depend on there being a 64-bit-int type for anything except int8 itself, and I don't want to start doing so. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 12:46:41 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6E7FCD1B4B2 for ; Sun, 18 Apr 2004 12:46:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 68606-01 for ; Sun, 18 Apr 2004 12:46:38 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 1C2B3D1B482 for ; Sun, 18 Apr 2004 12:46:38 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3IFkZjr019740; Sun, 18 Apr 2004 11:46:36 -0400 (EDT) To: Greg Stark Cc: "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: <873c719476.fsf@stark.xeocode.com> References: <20040416154502.GF87362@nasby.net> <450.1082133251@sss.pgh.pa.us> <20040416164644.GH87362@nasby.net> <808.1082135098@sss.pgh.pa.us> <20040416172523.GJ87362@nasby.net> <1263.1082138383@sss.pgh.pa.us> <20040416184541.GK87362@nasby.net> <10603.1082218095@sss.pgh.pa.us> <878ygu8auw.fsf@stark.xeocode.com> <15053.1082259958@sss.pgh.pa.us> <873c719476.fsf@stark.xeocode.com> Comments: In-reply-to Greg Stark message dated "18 Apr 2004 02:43:09 -0400" Date: Sun, 18 Apr 2004 11:46:35 -0400 Message-ID: <19739.1082303195@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/244 X-Sequence-Number: 6544 Greg Stark writes: > Eh? Oh, negative numbers? So low^high^sign. [ thinks about it... ] Yeah, that would work. We can't backpatch it without breaking existing hash indexes on int8, but it'd be reasonable to change for 7.5 (since at the rate things are going, we won't have pg_upgrade for 7.5 anyway...) > I wonder if it makes sense to have check the hash distribution after > generating the table and if it's bad then throw it away and try again with a > different hash function. The "different hash function" would probably just be > a seed value changing. Probably way overkill though. Yeah, it'd be a pain trying to get all the type-specific hash functions doing that. I'm also unconvinced that a simple change of seed value would necessarily make the distribution better. In the worst case, if the real problem is that all the input values are identical, you can reseed all day long and it won't fix it. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 12:58:43 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 10A7AD1E928 for ; Sun, 18 Apr 2004 12:58:43 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 64800-08 for ; Sun, 18 Apr 2004 12:58:40 -0300 (ADT) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 4C38CD1E8FB for ; Sun, 18 Apr 2004 12:58:39 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id B470380DB; Sun, 18 Apr 2004 17:58:38 +0200 (CEST) Date: Sun, 18 Apr 2004 17:58:38 +0200 (CEST) From: Dennis Bjorklund To: Tom Lane Cc: "Jim C. Nasby" , Subject: Re: Horribly slow hash join In-Reply-To: <19681.1082302750@sss.pgh.pa.us> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/245 X-Sequence-Number: 6545 On Sun, 18 Apr 2004, Tom Lane wrote: > That creates portability issues though. We do not depend on there being > a 64-bit-int type for anything except int8 itself, and I don't want to > start doing so. What do you mean? int8 is supported on all platformas and if the hasfunction would convert all numbers to int8 before making the hash it would work. I don't see any portability problems. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Sun Apr 18 13:23:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2FD25D1E812 for ; Sun, 18 Apr 2004 13:23:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75020-02 for ; Sun, 18 Apr 2004 13:23:46 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 3ECBDD1E7EF for ; Sun, 18 Apr 2004 13:23:45 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3IGNd1C020085; Sun, 18 Apr 2004 12:23:39 -0400 (EDT) To: Dennis Bjorklund Cc: "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: References: Comments: In-reply-to Dennis Bjorklund message dated "Sun, 18 Apr 2004 17:58:38 +0200" Date: Sun, 18 Apr 2004 12:23:39 -0400 Message-ID: <20084.1082305419@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/246 X-Sequence-Number: 6546 Dennis Bjorklund writes: > What do you mean? int8 is supported on all platformas No it isn't. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 13:27:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 10BABD1D064 for ; Sun, 18 Apr 2004 13:27:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 72710-05 for ; Sun, 18 Apr 2004 13:27:10 -0300 (ADT) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 474E5D1D05C for ; Sun, 18 Apr 2004 13:27:09 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 19FAF88E9; Sun, 18 Apr 2004 18:27:09 +0200 (CEST) Date: Sun, 18 Apr 2004 18:27:09 +0200 (CEST) From: Dennis Bjorklund To: Tom Lane Cc: "Jim C. Nasby" , Subject: Re: Horribly slow hash join In-Reply-To: <20084.1082305419@sss.pgh.pa.us> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/247 X-Sequence-Number: 6547 On Sun, 18 Apr 2004, Tom Lane wrote: > > What do you mean? int8 is supported on all platformas > > No it isn't. So on platforms where it isn't you would use int4 as the biggest int then. I don't really see that as a problem. As long as you calculate the hash on the biggest int on that platform it should work. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Sun Apr 18 15:47:48 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D71CDD1CC8D for ; Sun, 18 Apr 2004 15:47:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 09515-06 for ; Sun, 18 Apr 2004 15:47:45 -0300 (ADT) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 7777CD1C515 for ; Sun, 18 Apr 2004 15:47:42 -0300 (ADT) Received: (qmail 625 invoked from network); 18 Apr 2004 18:47:38 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 18 Apr 2004 18:47:38 -0000 Subject: Re: Toooo many context switches (maybe SLES8?) From: Dave Cramer Reply-To: pg@fastcrypt.com To: Josh Berkus Cc: Tom Lane , Joe Conway , Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= , pgsql-performance@postgresql.org In-Reply-To: <200404151337.00273.josh@agliodbs.com> References: <407E7B99.3050306@aeccom.com> <407EBA89.6080007@joeconway.com> <20498.1082057841@sss.pgh.pa.us> <200404151337.00273.josh@agliodbs.com> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1082314193.1557.44.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Sun, 18 Apr 2004 14:49:53 -0400 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: 200404/248 X-Sequence-Number: 6548 Isn't this a linux kernel issue ? My understanding is that the scheduler doesn't know that 2 of the CPU's are actually the same underlying hardware and sometimes two contexts end up fighting for the same underlying chip? --dc-- On Thu, 2004-04-15 at 16:37, Josh Berkus wrote: > Folks, > > > I am currently chasing what seems to be the same issue: massive context > > swapping on a dual Xeon system. I tried back-patching the above-mentioned > > patch ... it helps a little but by no means solves the problem ... > > BTW, I'm currently pursuing the possibility that this has something to do with > the ServerWorks chipset on those motherboards. If anyone knows a high-end > hardware+linux kernel geek I can corner, I'd appreciate it. > > Maybe I should contact OSDL ... -- Dave Cramer 519 939 0336 ICQ # 14675561 From pgsql-performance-owner@postgresql.org Sun Apr 18 16:22:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4C565D1BC59 for ; Sun, 18 Apr 2004 16:22:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14908-09 for ; Sun, 18 Apr 2004 16:22:34 -0300 (ADT) Received: from web11.manitu.net (web11.manitu.net [217.11.48.111]) by svr1.postgresql.org (Postfix) with ESMTP id 2A049D1BB49 for ; Sun, 18 Apr 2004 16:22:32 -0300 (ADT) Received: from [192.168.0.1] (dsl-082-082-191-125.arcor-ip.net [82.82.191.125]) (authenticated) by web11.manitu.net (8.10.2-SOL3/8.10.2) with ESMTP id i3IJMMM04900; Sun, 18 Apr 2004 21:22:22 +0200 Subject: Re: sunquery and estimated rows From: Markus Bertheau To: Tom Lane Cc: Litao Wu , pgsql-performance@postgresql.org In-Reply-To: <3802.1082159101@sss.pgh.pa.us> References: <20040416214529.10468.qmail@web13125.mail.yahoo.com> <3802.1082159101@sss.pgh.pa.us> Content-Type: text/plain; charset=UTF-8 Message-Id: <1082316145.2191.0.camel@yarrow.bertheau.de> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 (1.4.6-1) Date: Sun, 18 Apr 2004 21:22:26 +0200 Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/249 X-Sequence-Number: 6549 =D0=92 =D0=A1=D0=B1=D1=82, 17.04.2004, =D0=B2 01:45, Tom Lane =D0=BF=D0=B8= =D1=88=D0=B5=D1=82: > The planner sees that as "where scope =3D " > and falls back to a default estimate. It won't simplify a sub-select > to a constant. (Some people consider that a feature ;-).) Why? Thanks --=20 Markus Bertheau From pgsql-performance-owner@postgresql.org Sun Apr 18 18:47:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BB186D1B8DF for ; Sun, 18 Apr 2004 18:47:48 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 38761-06 for ; Sun, 18 Apr 2004 18:47:48 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 61D49D1B4B2 for ; Sun, 18 Apr 2004 18:47:45 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3ILlfdb011438; Sun, 18 Apr 2004 17:47:41 -0400 (EDT) To: lutzeb@aeccom.com Cc: Josh Berkus , pgsql-performance@postgreSQL.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <407EFC95.5070805@aeccom.com> References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> Comments: In-reply-to Dirk.Lutzebaeck@t-online.de (Dirk Lutzebaeck) message dated "Thu, 15 Apr 2004 23:20:21 +0200" Date: Sun, 18 Apr 2004 17:47:41 -0400 Message-ID: <11437.1082324861@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/250 X-Sequence-Number: 6550 After some further digging I think I'm starting to understand what's up here, and the really fundamental answer is that a multi-CPU Xeon MP box sucks for running Postgres. I did a bunch of oprofile measurements on a machine belonging to one of Josh's clients, using a test case that involved heavy concurrent access to a relatively small amount of data (little enough to fit into Postgres shared buffers, so that no I/O or kernel calls were really needed once the test got going). I found that by nearly any measure --- elapsed time, bus transactions, or machine-clear events --- the spinlock acquisitions associated with grabbing and releasing the BufMgrLock took an unreasonable fraction of the time. I saw about 15% of elapsed time, 40% of bus transactions, and nearly 100% of pipeline-clear cycles going into what is essentially two instructions out of the entire backend. (Pipeline clears occur when the cache coherency logic detects a memory write ordering problem.) I am not completely clear on why this machine-level bottleneck manifests as a lot of context swaps at the OS level. I think what is happening is that because SpinLockAcquire is so slow, a process is much more likely than you'd normally expect to arrive at SpinLockAcquire while another process is also acquiring the spinlock. This puts the two processes into a "lockstep" condition where the second process is nearly certain to observe the BufMgrLock as locked, and be forced to suspend itself, even though the time the first process holds the BufMgrLock is not really very long at all. If you google for Xeon and "cache coherency" you'll find quite a bit of suggestive information about why this might be more true on the Xeon setup than others. A couple of interesting hits: http://www.theinquirer.net/?article=10797 says that Xeon MP uses a *slower* FSB than Xeon DP. This would translate directly to more time needed to transfer a dirty cache line from one processor to the other, which is the basic operation that we're talking about here. http://www.aceshardware.com/Spades/read.php?article_id=30000187 says that Opterons use a different cache coherency protocol that is fundamentally superior to the Xeon's, because dirty cache data can be transferred directly between two processor caches without waiting for main memory. So in the short term I think we have to tell people that Xeon MP is not the most desirable SMP platform to run Postgres on. (Josh thinks that the specific motherboard chipset being used in these machines might share some of the blame too. I don't have any evidence for or against that idea, but it's certainly possible.) In the long run, however, CPUs continue to get faster than main memory and the price of cache contention will continue to rise. So it seems that we need to give up the assumption that SpinLockAcquire is a cheap operation. In the presence of heavy contention it won't be. One thing we probably have got to do soon is break up the BufMgrLock into multiple finer-grain locks so that there will be less contention. However I am wary of doing this incautiously, because if we do it in a way that makes for a significant rise in the number of locks that have to be acquired to access a buffer, we might end up with a net loss. I think Neil Conway was looking into how the bufmgr might be restructured to reduce lock contention, but if he had come up with anything he didn't mention exactly what. Neil? regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 20:09:28 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ED1CED1E928 for ; Sun, 18 Apr 2004 20:09:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 61183-06 for ; Sun, 18 Apr 2004 20:09:29 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id F0651D1E949 for ; Sun, 18 Apr 2004 20:09:25 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3IN9RlN012051; Sun, 18 Apr 2004 19:09:27 -0400 (EDT) To: Markus Bertheau Cc: Litao Wu , pgsql-performance@postgresql.org Subject: Re: sunquery and estimated rows In-reply-to: <1082316145.2191.0.camel@yarrow.bertheau.de> References: <20040416214529.10468.qmail@web13125.mail.yahoo.com> <3802.1082159101@sss.pgh.pa.us> <1082316145.2191.0.camel@yarrow.bertheau.de> Comments: In-reply-to Markus Bertheau message dated "Sun, 18 Apr 2004 21:22:26 +0200" Date: Sun, 18 Apr 2004 19:09:26 -0400 Message-ID: <12050.1082329766@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/251 X-Sequence-Number: 6551 Markus Bertheau writes: > В Сбт, 17.04.2004, в 01:45, Tom Lane пишет: >> The planner sees that as "where scope = " >> and falls back to a default estimate. It won't simplify a sub-select >> to a constant. (Some people consider that a feature ;-).) > Why? It's the only way to prevent it from simplifying when you don't want it to. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 20:32:30 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 66519D1D064 for ; Sun, 18 Apr 2004 20:32:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 58765-09 for ; Sun, 18 Apr 2004 20:32:29 -0300 (ADT) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 916B0D1CACC for ; Sun, 18 Apr 2004 20:32:25 -0300 (ADT) Received: (qmail 10377 invoked from network); 18 Apr 2004 23:32:23 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 18 Apr 2004 23:32:24 -0000 Subject: Re: Wierd context-switching issue on Xeon From: Dave Cramer Reply-To: pg@fastcrypt.com To: Tom Lane Cc: lutzeb@aeccom.com, Josh Berkus , pgsql-performance@postgreSQL.org, Neil Conway In-Reply-To: <11437.1082324861@sss.pgh.pa.us> References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1082331281.1557.47.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Sun, 18 Apr 2004 19:34:41 -0400 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: 200404/252 X-Sequence-Number: 6552 So the the kernel/OS is irrelevant here ? this happens on any dual xeon? What about hypterthreading does it still happen if HTT is turned off ? Dave On Sun, 2004-04-18 at 17:47, Tom Lane wrote: > After some further digging I think I'm starting to understand what's up > here, and the really fundamental answer is that a multi-CPU Xeon MP box > sucks for running Postgres. > > I did a bunch of oprofile measurements on a machine belonging to one of > Josh's clients, using a test case that involved heavy concurrent access > to a relatively small amount of data (little enough to fit into Postgres > shared buffers, so that no I/O or kernel calls were really needed once > the test got going). I found that by nearly any measure --- elapsed > time, bus transactions, or machine-clear events --- the spinlock > acquisitions associated with grabbing and releasing the BufMgrLock took > an unreasonable fraction of the time. I saw about 15% of elapsed time, > 40% of bus transactions, and nearly 100% of pipeline-clear cycles going > into what is essentially two instructions out of the entire backend. > (Pipeline clears occur when the cache coherency logic detects a memory > write ordering problem.) > > I am not completely clear on why this machine-level bottleneck manifests > as a lot of context swaps at the OS level. I think what is happening is > that because SpinLockAcquire is so slow, a process is much more likely > than you'd normally expect to arrive at SpinLockAcquire while another > process is also acquiring the spinlock. This puts the two processes > into a "lockstep" condition where the second process is nearly certain > to observe the BufMgrLock as locked, and be forced to suspend itself, > even though the time the first process holds the BufMgrLock is not > really very long at all. > > If you google for Xeon and "cache coherency" you'll find quite a bit of > suggestive information about why this might be more true on the Xeon > setup than others. A couple of interesting hits: > > http://www.theinquirer.net/?article=10797 > says that Xeon MP uses a *slower* FSB than Xeon DP. This would > translate directly to more time needed to transfer a dirty cache line > from one processor to the other, which is the basic operation that we're > talking about here. > > http://www.aceshardware.com/Spades/read.php?article_id=30000187 > says that Opterons use a different cache coherency protocol that is > fundamentally superior to the Xeon's, because dirty cache data can be > transferred directly between two processor caches without waiting for > main memory. > > So in the short term I think we have to tell people that Xeon MP is not > the most desirable SMP platform to run Postgres on. (Josh thinks that > the specific motherboard chipset being used in these machines might > share some of the blame too. I don't have any evidence for or against > that idea, but it's certainly possible.) > > In the long run, however, CPUs continue to get faster than main memory > and the price of cache contention will continue to rise. So it seems > that we need to give up the assumption that SpinLockAcquire is a cheap > operation. In the presence of heavy contention it won't be. > > One thing we probably have got to do soon is break up the BufMgrLock > into multiple finer-grain locks so that there will be less contention. > However I am wary of doing this incautiously, because if we do it in a > way that makes for a significant rise in the number of locks that have > to be acquired to access a buffer, we might end up with a net loss. > > I think Neil Conway was looking into how the bufmgr might be > restructured to reduce lock contention, but if he had come up with > anything he didn't mention exactly what. Neil? > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) > > > > !DSPAM:4082feb7326901956819835! > > -- Dave Cramer 519 939 0336 ICQ # 14675561 From pgsql-performance-owner@postgresql.org Tue Apr 20 10:58:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 86EE5D1D09C for ; Sun, 18 Apr 2004 20:43:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73713-01 for ; Sun, 18 Apr 2004 20:43:53 -0300 (ADT) Received: from tht.net (vista.tht.net [216.126.88.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3B0C7D1D064 for ; Sun, 18 Apr 2004 20:43:50 -0300 (ADT) Received: from [134.22.69.199] (dyn-69-199.tor.dsl.tht.net [134.22.69.199]) by tht.net (Postfix) with ESMTP id D132876A2A; Sun, 18 Apr 2004 19:43:57 -0400 (EDT) Subject: Re: sunquery and estimated rows From: Rod Taylor To: Tom Lane Cc: Markus Bertheau , Litao Wu , Postgresql Performance In-Reply-To: <12050.1082329766@sss.pgh.pa.us> References: <20040416214529.10468.qmail@web13125.mail.yahoo.com> <3802.1082159101@sss.pgh.pa.us> <1082316145.2191.0.camel@yarrow.bertheau.de> <12050.1082329766@sss.pgh.pa.us> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="=-lUbIYknmc3AnUBbPQ3d/" Message-Id: <1082331774.47576.136.camel@jester> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Sun, 18 Apr 2004 19:42:55 -0400 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/300 X-Sequence-Number: 6600 --=-lUbIYknmc3AnUBbPQ3d/ Content-Type: text/plain Content-Transfer-Encoding: quoted-printable On Sun, 2004-04-18 at 19:09, Tom Lane wrote: > Markus Bertheau writes: > > , 17.04.2004, 01:45, Tom Lane : > >> The planner sees that as "where scope =3D " > >> and falls back to a default estimate. It won't simplify a sub-select > >> to a constant. (Some people consider that a feature ;-).) >=20 > > Why? >=20 > It's the only way to prevent it from simplifying when you don't want it > to. I'm having a difficult time coming up with a circumstance where that is beneficial except when stats are out of whack. Doesn't a prepared statement also falls back to the default estimate for variables. --=20 Rod Taylor Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL PGP Key: http://www.rbt.ca/signature.asc --=-lUbIYknmc3AnUBbPQ3d/ Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (FreeBSD) iD8DBQBAgxJ96DETLow6vwwRAsIkAJ91VWf31bzt826YLFp5TaIU2847SwCeP5bd EL/fSZkC394t55ypX0sD2GQ= =xhey -----END PGP SIGNATURE----- --=-lUbIYknmc3AnUBbPQ3d/-- From pgsql-performance-owner@postgresql.org Sun Apr 18 21:40:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2BD19D1E1F3 for ; Sun, 18 Apr 2004 21:40:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 86378-01 for ; Sun, 18 Apr 2004 21:40:35 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 3875ED1E0D8 for ; Sun, 18 Apr 2004 21:40:32 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id A582617C46F; Sun, 18 Apr 2004 20:40:35 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BFMpv-00016W-00; Sun, 18 Apr 2004 20:40:35 -0400 To: Tom Lane Cc: lutzeb@aeccom.com, Josh Berkus , pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> In-Reply-To: <11437.1082324861@sss.pgh.pa.us> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 18 Apr 2004 20:40:35 -0400 Message-ID: <87d6647qbg.fsf@stark.xeocode.com> Lines: 23 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: 200404/253 X-Sequence-Number: 6553 Tom Lane writes: > So in the short term I think we have to tell people that Xeon MP is not > the most desirable SMP platform to run Postgres on. (Josh thinks that > the specific motherboard chipset being used in these machines might > share some of the blame too. I don't have any evidence for or against > that idea, but it's certainly possible.) > > In the long run, however, CPUs continue to get faster than main memory > and the price of cache contention will continue to rise. So it seems > that we need to give up the assumption that SpinLockAcquire is a cheap > operation. In the presence of heavy contention it won't be. There's nothing about the way Postgres spinlocks are coded that affects this? Is it something the kernel could help with? I've been wondering whether there's any benefits postgres is missing out on by using its own hand-rolled locking instead of using the pthreads infrastructure that the kernel is often involved in. -- greg From pgsql-performance-owner@postgresql.org Sun Apr 18 23:16:52 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6DB6BD1B4BE for ; Sun, 18 Apr 2004 23:16:51 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 06866-01 for ; Sun, 18 Apr 2004 23:16:53 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id A3D42D1B482 for ; Sun, 18 Apr 2004 23:16:49 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3J2GrME013911; Sun, 18 Apr 2004 22:16:53 -0400 (EDT) To: Rod Taylor Cc: Markus Bertheau , Litao Wu , Postgresql Performance Subject: Re: sunquery and estimated rows In-reply-to: <1082331774.47576.136.camel@jester> References: <20040416214529.10468.qmail@web13125.mail.yahoo.com> <3802.1082159101@sss.pgh.pa.us> <1082316145.2191.0.camel@yarrow.bertheau.de> <12050.1082329766@sss.pgh.pa.us> <1082331774.47576.136.camel@jester> Comments: In-reply-to Rod Taylor message dated "Sun, 18 Apr 2004 19:42:55 -0400" Date: Sun, 18 Apr 2004 22:16:53 -0400 Message-ID: <13910.1082341013@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/254 X-Sequence-Number: 6554 Rod Taylor writes: >> It's the only way to prevent it from simplifying when you don't want it >> to. > I'm having a difficult time coming up with a circumstance where that is > beneficial except when stats are out of whack. Try trawling the archives --- I recall several cases in which people were using sub-selects for this purpose. In any case, I don't see the value of having the planner check to see if a sub-select is just a trivial arithmetic expression. The cases where people write that and expect it to be simplified are so few and far between that I can't believe it'd be a good use of planner cycles. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 23:20:25 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1DCF7D1E98C for ; Sun, 18 Apr 2004 23:20:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 03558-06 for ; Sun, 18 Apr 2004 23:20:27 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6EE19D1E997 for ; Sun, 18 Apr 2004 23:20:23 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3J2KMMc013961; Sun, 18 Apr 2004 22:20:22 -0400 (EDT) To: pg@fastcrypt.com Cc: lutzeb@aeccom.com, Josh Berkus , pgsql-performance@postgreSQL.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <1082331281.1557.47.camel@localhost.localdomain> References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <1082331281.1557.47.camel@localhost.localdomain> Comments: In-reply-to Dave Cramer message dated "Sun, 18 Apr 2004 19:34:41 -0400" Date: Sun, 18 Apr 2004 22:20:22 -0400 Message-ID: <13960.1082341222@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/255 X-Sequence-Number: 6555 Dave Cramer writes: > So the the kernel/OS is irrelevant here ? this happens on any dual xeon? I believe so. The context-switch behavior might possibly be a little more pleasant on other kernels, but the underlying spinlock problem is not dependent on the kernel. > What about hypterthreading does it still happen if HTT is turned off ? The problem comes from keeping the caches synchronized between multiple physical CPUs. AFAICS enabling HTT wouldn't make it worse, because a hyperthreaded processor still only has one cache. regards, tom lane From pgsql-performance-owner@postgresql.org Sun Apr 18 23:30:15 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 719CFD1B579 for ; Sun, 18 Apr 2004 23:30:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 10407-02 for ; Sun, 18 Apr 2004 23:30:15 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 5D9BDD1B515 for ; Sun, 18 Apr 2004 23:30:09 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3J2U9eT014057; Sun, 18 Apr 2004 22:30:09 -0400 (EDT) To: Greg Stark Cc: lutzeb@aeccom.com, Josh Berkus , pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <87d6647qbg.fsf@stark.xeocode.com> References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <87d6647qbg.fsf@stark.xeocode.com> Comments: In-reply-to Greg Stark message dated "18 Apr 2004 20:40:35 -0400" Date: Sun, 18 Apr 2004 22:30:08 -0400 Message-ID: <14056.1082341808@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/256 X-Sequence-Number: 6556 Greg Stark writes: > There's nothing about the way Postgres spinlocks are coded that affects this? No. AFAICS our spinlock sequences are pretty much equivalent to the way the Linux kernel codes its spinlocks, so there's no deep dark knowledge to be mined there. We could possibly use some more-efficient blocking mechanism than semop() once we've decided we have to block (it's a shame Linux still doesn't have cross-process POSIX semaphores). But the striking thing I learned from looking at the oprofile results is that most of the inefficiency comes at the very first TAS() operation, before we've even "spun" let alone decided we have to block. The s_lock() subroutine does not account for more than a few percent of the runtime in these tests, compared to 15% at the inline TAS() operations in LWLockAcquire and LWLockRelease. I interpret this to mean that once it's acquired ownership of the cache line, a Xeon can get through the "spinning" loop in s_lock() mighty quickly. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 19 00:12:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4A659D1EC90 for ; Mon, 19 Apr 2004 00:12:18 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 17503-05 for ; Mon, 19 Apr 2004 00:12:12 -0300 (ADT) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id 414DDD1B49B for ; Mon, 19 Apr 2004 00:12:06 -0300 (ADT) Received: (qmail 1925 invoked by uid 500); 19 Apr 2004 03:15:54 -0000 Date: Sun, 18 Apr 2004 22:15:54 -0500 From: Bruno Wolff III To: Dennis Bjorklund Cc: Tom Lane , "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join Message-ID: <20040419031554.GA1881@wolff.to> Mail-Followup-To: Dennis Bjorklund , Tom Lane , "Jim C. Nasby" , pgsql-performance@postgresql.org References: <20084.1082305419@sss.pgh.pa.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/257 X-Sequence-Number: 6557 On Sun, Apr 18, 2004 at 18:27:09 +0200, Dennis Bjorklund wrote: > On Sun, 18 Apr 2004, Tom Lane wrote: > > > > What do you mean? int8 is supported on all platformas > > > > No it isn't. > > So on platforms where it isn't you would use int4 as the biggest int then. > I don't really see that as a problem. As long as you calculate the hash on > the biggest int on that platform it should work. Another option would be to put the numbers into two int4s. For int4 or smaller types one of these would be zero. int8s would be split between the two. The hash function would then be defined on the two int4s. From pgsql-performance-owner@postgresql.org Mon Apr 19 00:20:07 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BD7D4D1E98D for ; Mon, 19 Apr 2004 00:20:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 19777-05 for ; Mon, 19 Apr 2004 00:20:03 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id DAD8DD1B49B for ; Mon, 19 Apr 2004 00:20:01 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3J3JuHv014395; Sun, 18 Apr 2004 23:19:56 -0400 (EDT) To: pg@fastcrypt.com Cc: lutzeb@aeccom.com, Josh Berkus , pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <13960.1082341222@sss.pgh.pa.us> References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <1082331281.1557.47.camel@localhost.localdomain> <13960.1082341222@sss.pgh.pa.us> Comments: In-reply-to Tom Lane message dated "Sun, 18 Apr 2004 22:20:22 -0400" Date: Sun, 18 Apr 2004 23:19:56 -0400 Message-ID: <14394.1082344796@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/258 X-Sequence-Number: 6558 >> What about hypterthreading does it still happen if HTT is turned off ? > The problem comes from keeping the caches synchronized between multiple > physical CPUs. AFAICS enabling HTT wouldn't make it worse, because a > hyperthreaded processor still only has one cache. Also, I forgot to say that the numbers I'm quoting *are* with HTT off. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 19 01:43:21 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E9B33D1DACE for ; Mon, 19 Apr 2004 01:43:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 47657-02 for ; Mon, 19 Apr 2004 01:43:19 -0300 (ADT) Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191]) by svr1.postgresql.org (Postfix) with ESMTP id 9DD36D1D167 for ; Mon, 19 Apr 2004 01:43:17 -0300 (ADT) Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1]) by zigo.dhs.org (Postfix) with ESMTP id 899E480DB; Mon, 19 Apr 2004 06:43:16 +0200 (CEST) Date: Mon, 19 Apr 2004 06:43:16 +0200 (CEST) From: Dennis Bjorklund To: Bruno Wolff III Cc: Tom Lane , "Jim C. Nasby" , Subject: Re: Horribly slow hash join In-Reply-To: <20040419031554.GA1881@wolff.to> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/259 X-Sequence-Number: 6559 On Sun, 18 Apr 2004, Bruno Wolff III wrote: > Another option would be to put the numbers into two int4s. For int4 or > smaller types one of these would be zero. int8s would be split between > the two. The hash function would then be defined on the two int4s. Sure, this is an internal calculation in the hash function. The only important thing is that the number 7 (for example) gives the same hash value no matter if it is an int2 or an int8 and that the hash function works well also for int8 numbers (which is does not today). At least that was the properties I understood that we wanted. We got side tracked into talking about what datatype exists in all platforms, that's not an issue at all. -- /Dennis Bj�rklund From pgsql-performance-owner@postgresql.org Mon Apr 19 02:29:24 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E5F68D1B8DF for ; Mon, 19 Apr 2004 02:29:22 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 53884-10 for ; Mon, 19 Apr 2004 02:29:21 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 7E17ED1B4BE for ; Mon, 19 Apr 2004 02:29:20 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 5D7DD17C428; Mon, 19 Apr 2004 01:29:20 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BFRLM-0001s9-00; Mon, 19 Apr 2004 01:29:20 -0400 To: Dennis Bjorklund Cc: Bruno Wolff III , Tom Lane , "Jim C. Nasby" , Subject: Re: Horribly slow hash join References: In-Reply-To: From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 19 Apr 2004 01:29:19 -0400 Message-ID: <874qrg7cy8.fsf@stark.xeocode.com> Lines: 50 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: 200404/260 X-Sequence-Number: 6560 Dennis Bjorklund writes: > On Sun, 18 Apr 2004, Bruno Wolff III wrote: > > > Another option would be to put the numbers into two int4s. For int4 or > > smaller types one of these would be zero. int8s would be split between > > the two. The hash function would then be defined on the two int4s. > > Sure, this is an internal calculation in the hash function. The only > important thing is that the number 7 (for example) gives the same hash > value no matter if it is an int2 or an int8 and that the hash function > works well also for int8 numbers (which is does not today). What's missing here is that the actual API for hash functions is that the data type provides a function that hashes to 32 bit integers. Then the hash code uses the 32 bit integer to crunch down to the actual number of buckets (using mod). The choice of 32 bit integers is purely arbitrary. As long as it's larger than than the number of buckets in any sane hash table it's fine. 32 bits is plenty. I question the use of mod to crunch the hash value down though. In the case of int4 the mapping to 32 bits is simply the identity. So the entire hash function ends up being simply "input mod #buckets". It seems way too easy to find real world data sets where many numbers will all be multiples of some number. If that common divisor shares any factors with the number of buckets, then the distribution will be very far from even with many empty buckets. If the hash tables were made a power of two then it would be possible to mix the bits of the 32 bit value and just mask off the unneeded bits. I've found one page via google that mentions mixing bits in a hash function, but I would look for a more serious treatment somewhere. http://burtleburtle.net/bob/hash/doobs.html Incidentally, this text claims mod is extremely slow compared to bit manipulations. I don't know that that kind of cycle counting is really is a factor for postgres though. Also, incidentally, this text is interesting: http://www.isthe.com/chongo/tech/comp/fnv/ -- greg From pgsql-performance-owner@postgresql.org Mon Apr 19 03:09:36 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 55E0ED1D2C1 for ; Mon, 19 Apr 2004 03:09:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 73070-01 for ; Mon, 19 Apr 2004 03:09:33 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 6D3FCD1D29D for ; Mon, 19 Apr 2004 03:09:32 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3J69Ugw016042; Mon, 19 Apr 2004 02:09:31 -0400 (EDT) To: Greg Stark Cc: Dennis Bjorklund , Bruno Wolff III , "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join In-reply-to: <874qrg7cy8.fsf@stark.xeocode.com> References: <874qrg7cy8.fsf@stark.xeocode.com> Comments: In-reply-to Greg Stark message dated "19 Apr 2004 01:29:19 -0400" Date: Mon, 19 Apr 2004 02:09:30 -0400 Message-ID: <16041.1082354970@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/261 X-Sequence-Number: 6561 Greg Stark writes: > If the hash tables were made a power of two then it would be possible to mix > the bits of the 32 bit value and just mask off the unneeded bits. I've found > one page via google that mentions mixing bits in a hash function, but I would > look for a more serious treatment somewhere. > http://burtleburtle.net/bob/hash/doobs.html > Incidentally, this text claims mod is extremely slow compared to bit > manipulations. Modding by a *non* power of 2 (esp. a prime) mixes the bits quite well, and is likely faster than any multiple-instruction way to do the same. The quoted article seems to be by someone who has spent a lot of time counting assembly cycles and none at all reading the last thirty years worth of CS literature. Knuth's treatment of hashing has some actual math to it... regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 19 04:28:00 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 43A33D1D086 for ; Mon, 19 Apr 2004 04:27:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 90727-08 for ; Mon, 19 Apr 2004 04:27:59 -0300 (ADT) Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id F21BCD1D096 for ; Mon, 19 Apr 2004 04:27:56 -0300 (ADT) Received: from aeccom.com (cayambe.core.aeccom.com [192.168.2.12]) by mail2.aeccom.com (Postfix) with ESMTP id 1E91F34; Mon, 19 Apr 2004 09:27:57 +0200 (CEST) Message-ID: <40837F7D.9050102@aeccom.com> Date: Mon, 19 Apr 2004 09:27:57 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH, Berlin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031016 X-Accept-Language: de, en MIME-Version: 1.0 To: josh@agliodbs.com Cc: Tom Lane , pgsql-performance@postgresql.org Subject: Re: RESOLVED: Re: Wierd context-switching issue on Xeon References: <200311251419.36771.josh@agliodbs.com> <407FAF0F.2070304@aeccom.com> <407FD9A0.6040608@aeccom.com> <200404160958.14902.josh@agliodbs.com> In-Reply-To: <200404160958.14902.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: 200404/262 X-Sequence-Number: 6562 Josh, I cannot reproduce the excessive semop() on a Dual XEON DP on a non-bigmem kernel, HT on. Interesting to know if the problem is related to XEON MP (as Tom wrote) or bigmem. Josh Berkus wrote: >Dirk, > > > >>I'm not sure if this semop() problem is still an issue but the database >>behaves a bit out of bounds in this situation, i.e. consuming system >>resources with semop() calls 95% while tables are locked very often and >>longer. >> >> > >It would be helpful to us if you could test this with the indexes disabled on >the non-Bigmem system. I'd like to eliminate Bigmem as a factor, if >possible. > > > From pgsql-performance-owner@postgresql.org Mon Apr 19 08:02:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id DB60BD1D064 for ; Mon, 19 Apr 2004 08:02:52 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 64527-01 for ; Mon, 19 Apr 2004 08:02:53 -0300 (ADT) Received: from kinetic.co.uk (mailhost.lei.co.uk [217.207.0.112]) by svr1.postgresql.org (Postfix) with ESMTP id 17557D1BAAE for ; Mon, 19 Apr 2004 08:02:50 -0300 (ADT) Received: from WorldClient ([217.207.0.112]) (authenticated user gerard.isdell@kinetic.co.uk) by kinetic.co.uk (kinetic.co.uk [217.207.0.112]) (MDaemon.PRO.v7.0.1.R) with ESMTP id md50000026316.msg for ; Mon, 19 Apr 2004 11:59:48 +0100 Received: from [217.207.0.3] via WorldClient with HTTP; Mon, 19 Apr 2004 11:59:48 +0100 Date: Mon, 19 Apr 2004 11:59:48 +0100 From: "Gerard Isdell" To: pgsql-performance@postgresql.org Subject: very slow simple query - outer join makes it quicker MIME-Version: 1.0 Content-Type: text/plain Message-ID: X-Mailer: WorldClient 7.0.1 X-Authenticated-Sender: gerard.isdell@kinetic.co.uk X-MDRemoteIP: 217.207.0.112 X-Return-Path: gerard.isdell@kinetic.co.uk X-MDaemon-Deliver-To: pgsql-performance@postgresql.org Reply-To: gerard.isdell@kinetic.co.uk X-Spam-Processed: kinetic.co.uk, Mon, 19 Apr 2004 11:59:49 +0100 X-MDAV-Processed: kinetic.co.uk, Mon, 19 Apr 2004 11:59:49 +0100 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/263 X-Sequence-Number: 6563 Hi Everyone I am new to this group and relatively new to Postgres, having used MSSQL 7 up until now. One table in my database in returning even the simplest of queries extremely slowly. The table is user table, and even the select userid from users takes over 20 seconds to run. There are about 2000 records in the table. The EXPLAIN ANALYZE on this table produces this output: Seq Scan on users (cost=0.00..89482.63 rows=1463 width=4) (actual time=68.836..40233.463 rows=1465 loops=1) Total runtime: 40234.965 ms SELECT USERID FROM USERS produces this: 1465 rows fetched (25.28 sec) The userid field is the primary key and has an index on it with this ddl: ALTER TABLE "public"."users" ADD CONSTRAINT "users_pkey" PRIMARY KEY ("userid"); There are other tables, such as the messages table, that have 10s of thousands of rows and they return records much more quickly. There must be something seriously wrong for simple queries like this to take so long. I should say that we are using the OpenFTS text search on the users table. In many cases to make the queries run at reasonable speeds I do an outer join on another table, and surprisingly these results come back very quickly Can anybody help me in diagnosing this problem. Gerard Isdell ************************************************************************* This e-mail and any attachments may contain confidential or privileged information. If you are not the intended recipient, please contact the sender immediately and do not use, store or disclose their contents. Any views expressed are those of the individual sender and not of Kinetic Information System Services Limited unless otherwise stated. www.kinetic.co.uk From pgsql-performance-owner@postgresql.org Mon Apr 19 23:29:23 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A8089D1D98A for ; Mon, 19 Apr 2004 23:29:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 12981-02 for ; Mon, 19 Apr 2004 23:29:21 -0300 (ADT) Received: from ganymede.hub.org (u46n208.hfx.eastlink.ca [24.222.46.208]) by svr1.postgresql.org (Postfix) with ESMTP id AF898D1B564 for ; Mon, 19 Apr 2004 23:29:14 -0300 (ADT) Received: by ganymede.hub.org (Postfix, from userid 1000) id A5EAB34288; Mon, 19 Apr 2004 23:29:18 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id A32F9341DD for ; Mon, 19 Apr 2004 23:29:18 -0300 (ADT) X-Received: from localhost ([unix socket]) by ganymede.hub.org (Cyrus v2.2.3) with LMTP; Mon, 19 Apr 2004 08:30:38 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id CF48F35BBD for ; Mon, 19 Apr 2004 08:30:27 -0300 (ADT) X-Received: from mail.postgresql.org [200.46.204.71] by localhost with IMAP (fetchmail-6.2.5) for scrappy@localhost (single-drop); Mon, 19 Apr 2004 08:30:27 -0300 (ADT) X-Received: from svr1.postgresql.org ([unix socket]) by svr1.postgresql.org (Cyrus v2.2.3) with LMTP; Mon, 19 Apr 2004 08:30:15 -0300 X-Sieve: CMU Sieve 2.2 X-Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9C69AD1D08C for ; Mon, 19 Apr 2004 08:30:14 -0300 (ADT) X-Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69136-07 for ; Mon, 19 Apr 2004 08:30:16 -0300 (ADT) X-Received: from minnie.omroep.nl (minnie.omroep.nl [145.58.30.4]) by svr1.postgresql.org (Postfix) with ESMTP id BA04FD1BCB8 for ; Mon, 19 Apr 2004 08:30:10 -0300 (ADT) X-Received: from tron.omroep.nl (tron.omroep.nl [145.58.31.20]) by minnie.omroep.nl (Postfix) with ESMTP id AB8271CF3E15 for ; Mon, 19 Apr 2004 13:30:13 +0200 (CDT) X-Received: from michiel.omroep.nl (michiel.omroep.nl [145.58.67.10]) by tron.omroep.nl (Postfix) with ESMTP id 9E8CE10056BD for ; Mon, 19 Apr 2004 13:30:13 +0200 (CEST) X-Received: from michiel by michiel.omroep.nl with local (Exim 3.35 #1 (Debian)) id 1BFWyb-0008UY-00 for ; Mon, 19 Apr 2004 13:30:13 +0200 Date: Mon, 19 Apr 2004 13:30:13 +0200 From: Michiel Meeuwissen To: performance@postgresql.org Subject: order by index, and inheritance Message-ID: <20040419113013.GA32186@michiel.omroep.nl> Reply-To: Michiel Meeuwissen Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Organization: Publieke Omroep []() User-Agent: Mutt/1.5.5.1+cvs20040105i X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-DCC: EATSERVER: ganymede.hub.org 1166; IP=ok Body=1 Fuz1=1 Fuz2=1 X-Spam-Pyzor: Reported 0 times. ReSent-Date: Mon, 19 Apr 2004 23:29:13 -0300 (ADT) Resent-From: "Marc G. Fournier" Resent-To: pgsql-performance@postgresql.org ReSent-Subject: order by index, and inheritance ReSent-Message-ID: <20040419232913.V780@ganymede.hub.org> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/283 X-Sequence-Number: 6583 I have a query which performs not so well: SELECT * FROM mm_mediasources ORDER BY number DESC LIMIT 20; costs nearly a minute. The table contains over 300 000 records. The table has two extensions, which are (a the moment) nearly empty, but have something to do with this, because: SELECT * FROM only mm_mediasources ORDER BY number DESC LIMIT 20; performs ok (8ms). The query plan is then as I would expect: media=# explain SELECT * FROM only mm_mediasources ORDER BY number DESC LIMIT 20; QUERY PLAN --------------------------------------------------------------------------------------------------------------------- Limit (cost=0.00..8.36 rows=20 width=105) -> Index Scan Backward using mediasource_object on mm_mediasources (cost=0.00..114641.05 rows=274318 width=105) The query plan of the original query, without 'only' does table scans: media=# explain SELECT * FROM mm_mediasources ORDER BY number DESC LIMIT 20; QUERY PLAN -------------------------------------------------------------------------------------------------------- Limit (cost=47248.70..47248.75 rows=20 width=105) -> Sort (cost=47248.70..47934.52 rows=274328 width=105) Sort Key: public.mm_mediasources.number -> Result (cost=0.00..8364.28 rows=274328 width=105) -> Append (cost=0.00..8364.28 rows=274328 width=105) -> Seq Scan on mm_mediasources (cost=0.00..8362.18 rows=274318 width=105) -> Seq Scan on mm_audiosources mm_mediasources (cost=0.00..1.01 rows=1 width=84) -> Seq Scan on mm_videosources mm_mediasources (cost=0.00..1.09 rows=9 width=89) and presumably because if that performs so lousy. Simply selecting on a number does work fast: media=# explain SELECT * FROM mm_mediasources where number = 606973 ; QUERY PLAN -------------------------------------------------------------------------------------------------------- Result (cost=0.00..6.13 rows=4 width=105) -> Append (cost=0.00..6.13 rows=4 width=105) -> Index Scan using mediasource_object on mm_mediasources (cost=0.00..4.00 rows=2 width=105) Index Cond: (number = 606973) -> Seq Scan on mm_audiosources mm_mediasources (cost=0.00..1.01 rows=1 width=84) Filter: (number = 606973) -> Seq Scan on mm_videosources mm_mediasources (cost=0.00..1.11 rows=1 width=89) Filter: (number = 606973) (3ms) I suppose seq scans are used on the extensions because they contain so few records. All tables have index on number. How do I force it to use them also when I use order by? I use psql 7.3.2 Michiel -- Michiel Meeuwissen | Mediapark C101 Hilversum | +31 (0)35 6772979 | I hate computers nl_NL eo_XX en_US | mihxil' | [] () | From pgsql-performance-owner@postgresql.org Mon Apr 19 09:02:21 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 997FDD1C952 for ; Mon, 19 Apr 2004 09:02:20 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 79041-05 for ; Mon, 19 Apr 2004 09:02:22 -0300 (ADT) Received: from tht.net (vista.tht.net [216.126.88.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8D489D1BB93 for ; Mon, 19 Apr 2004 09:02:18 -0300 (ADT) Received: from [134.22.69.199] (dyn-69-199.tor.dsl.tht.net [134.22.69.199]) by tht.net (Postfix) with ESMTP id E818C76A6D; Mon, 19 Apr 2004 08:02:18 -0400 (EDT) Subject: Re: very slow simple query - outer join makes it quicker From: Rod Taylor To: gerard.isdell@kinetic.co.uk Cc: Postgresql Performance In-Reply-To: References: Content-Type: text/plain Message-Id: <1082376074.24414.3.camel@jester> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Mon, 19 Apr 2004 08:01:15 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/264 X-Sequence-Number: 6564 > There are other tables, such as the messages table, that have 10s of > thousands of rows and they return records much more quickly. > There must be something seriously wrong for simple queries like this to take > so long. Have you run VACUUM recently? If not, run VACUUM FULL against the users table and see if that makes a difference. From pgsql-performance-owner@postgresql.org Mon Apr 19 09:05:27 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6CCDED1D08B for ; Mon, 19 Apr 2004 09:05:26 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 75036-09 for ; Mon, 19 Apr 2004 09:05:27 -0300 (ADT) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 43A74D1D08C for ; Mon, 19 Apr 2004 09:05:23 -0300 (ADT) Received: (qmail 8053 invoked from network); 19 Apr 2004 12:05:19 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 19 Apr 2004 12:05:19 -0000 Subject: Re: Horribly slow hash join From: Dave Cramer Reply-To: pg@fastcrypt.com To: Tom Lane Cc: Greg Stark , Dennis Bjorklund , Bruno Wolff III , "Jim C. Nasby" , pgsql-performance@postgresql.org In-Reply-To: <16041.1082354970@sss.pgh.pa.us> References: <874qrg7cy8.fsf@stark.xeocode.com> <16041.1082354970@sss.pgh.pa.us> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1082376464.1558.69.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Mon, 19 Apr 2004 08:07:44 -0400 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: 200404/265 X-Sequence-Number: 6565 Here's an interesting link that suggests that hyperthreading would be much worse. http://groups.google.com/groups?q=hyperthreading+dual+xeon+idle&start=10&hl=en&lr=&ie=UTF-8&c2coff=1&selm=aukkonen-FE5275.21093624062003%40shawnews.gv.shawcable.net&rnum=16 FWIW, I have anecdotal evidence that suggests that this is the case, on of my clients was seeing very large context switches with HTT turned on, and without it was much better. Dave On Mon, 2004-04-19 at 02:09, Tom Lane wrote: > Greg Stark writes: > > If the hash tables were made a power of two then it would be possible to mix > > the bits of the 32 bit value and just mask off the unneeded bits. I've found > > one page via google that mentions mixing bits in a hash function, but I would > > look for a more serious treatment somewhere. > > http://burtleburtle.net/bob/hash/doobs.html > > Incidentally, this text claims mod is extremely slow compared to bit > > manipulations. > > Modding by a *non* power of 2 (esp. a prime) mixes the bits quite well, > and is likely faster than any multiple-instruction way to do the same. > > The quoted article seems to be by someone who has spent a lot of time > counting assembly cycles and none at all reading the last thirty years > worth of CS literature. Knuth's treatment of hashing has some actual > math to it... > > regards, tom lane > > ---------------------------(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 > > > > !DSPAM:40837183123741526418863! > > -- Dave Cramer 519 939 0336 ICQ # 14675561 From pgsql-performance-owner@postgresql.org Mon Apr 19 09:12:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 461CAD1D08C for ; Mon, 19 Apr 2004 09:12:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 82621-08 for ; Mon, 19 Apr 2004 09:12:55 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id B0E71D1C9E1 for ; Mon, 19 Apr 2004 09:12:51 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 081CB17C4D1; Mon, 19 Apr 2004 08:12:55 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BFXdu-00034x-00; Mon, 19 Apr 2004 08:12:54 -0400 To: Tom Lane Cc: Greg Stark , Dennis Bjorklund , Bruno Wolff III , "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join References: <874qrg7cy8.fsf@stark.xeocode.com> <16041.1082354970@sss.pgh.pa.us> In-Reply-To: <16041.1082354970@sss.pgh.pa.us> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 19 Apr 2004 08:12:54 -0400 Message-ID: <87vfjw5fp5.fsf@stark.xeocode.com> Lines: 27 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: 200404/266 X-Sequence-Number: 6566 Tom Lane writes: > Greg Stark writes: > > If the hash tables were made a power of two then it would be possible to mix > > the bits of the 32 bit value and just mask off the unneeded bits. I've found > > one page via google that mentions mixing bits in a hash function, but I would > > look for a more serious treatment somewhere. > Modding by a *non* power of 2 (esp. a prime) mixes the bits quite well, > and is likely faster than any multiple-instruction way to do the same. Well a) any number that has any factors of two fails to mix in some bits. That's a lot more common than non powers of two. b) The postgres code makes no attempt to make the number of buckets a prime and c) Even if the number of buckets were prime then it seems it would still be too easy to find real-world data where all the data have that prime as a factor. As it is they only need to have common factors to lose. > The quoted article seems to be by someone who has spent a lot of time > counting assembly cycles and none at all reading the last thirty years > worth of CS literature. Yes, well I did note that. -- greg From pgsql-performance-owner@postgresql.org Mon Apr 19 09:16:37 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 6B948D1E991 for ; Mon, 19 Apr 2004 09:16:35 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88057-03 for ; Mon, 19 Apr 2004 09:16:36 -0300 (ADT) Received: from smtp.istop.com (dci.doncaster.on.ca [66.11.168.194]) by svr1.postgresql.org (Postfix) with ESMTP id 62E96D1E98A for ; Mon, 19 Apr 2004 09:16:33 -0300 (ADT) Received: from stark.xeocode.com (gsstark.mtl.istop.com [66.11.160.162]) by smtp.istop.com (Postfix) with ESMTP id 05BEF17C4CE; Mon, 19 Apr 2004 08:16:37 -0400 (EDT) Received: from localhost ([127.0.0.1] helo=stark.xeocode.com) by stark.xeocode.com with smtp (Exim 3.36 #1 (Debian)) id 1BFXhU-00036V-00; Mon, 19 Apr 2004 08:16:36 -0400 To: pg@fastcrypt.com Cc: Tom Lane , Greg Stark , Dennis Bjorklund , Bruno Wolff III , "Jim C. Nasby" , pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join References: <874qrg7cy8.fsf@stark.xeocode.com> <16041.1082354970@sss.pgh.pa.us> <1082376464.1558.69.camel@localhost.localdomain> In-Reply-To: <1082376464.1558.69.camel@localhost.localdomain> From: Greg Stark Organization: The Emacs Conspiracy; member since 1992 Date: 19 Apr 2004 08:16:36 -0400 Message-ID: <87pta45fiz.fsf@stark.xeocode.com> Lines: 10 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: 200404/267 X-Sequence-Number: 6567 Dave Cramer writes: > Here's an interesting link that suggests that hyperthreading would be > much worse. Uh, this is the wrong thread. -- greg From pgsql-performance-owner@postgresql.org Tue Apr 20 10:58:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D06F8D1D08E for ; Mon, 19 Apr 2004 09:27:45 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89415-05 for ; Mon, 19 Apr 2004 09:27:47 -0300 (ADT) Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id 8B102D1D099 for ; Mon, 19 Apr 2004 09:27:42 -0300 (ADT) Received: from andesW2K (dhcp00.core.aeccom.com [192.168.2.100]) by mail2.aeccom.com (Postfix) with SMTP id D27CC34; Mon, 19 Apr 2004 14:27:44 +0200 (CEST) Message-ID: <004201c42609$b7321390$6402a8c0@andesW2K> From: "Sven Geisler" To: , "Tom Lane" Cc: "Josh Berkus" , , "Neil Conway" References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> Subject: Re: Wierd context-switching issue on Xeon Date: Mon, 19 Apr 2004 14:27:44 +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 6.00.2800.1409 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409 X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/301 X-Sequence-Number: 6601 Hi Tom, Just to explain our hardware situation releated to the FSB of the XEON's. We have older XEON DP in operation with FSB 400 and 2.4 GHz. The XEON MP box runs with 2.5 GHz. The XEON MP box is a Fujitsu Siemens Primergy RX600 with ServerWorks GC LE as chipset. The box, which Dirk were use to compare the behavior, is our newest XEON DP system. This XEON DP box runs with 2.8 GHz and FSB 533 using the Intel 7501 chipset (Supermicro). I would agree to Jush. When PostgreSQL has an issue with the INTEL XEON MP hardware, this is more releated to the chipset. Back to the SQL-Level. We use SELECT FOR UPDATE as "semaphore". Should we try another implementation for this semahore on the client side to prevent this issue? Regards Sven. ----- Original Message ----- From: "Tom Lane" To: Cc: "Josh Berkus" ; ; "Neil Conway" Sent: Sunday, April 18, 2004 11:47 PM Subject: Re: [PERFORM] Wierd context-switching issue on Xeon > After some further digging I think I'm starting to understand what's up > here, and the really fundamental answer is that a multi-CPU Xeon MP box > sucks for running Postgres. > > I did a bunch of oprofile measurements on a machine belonging to one of > Josh's clients, using a test case that involved heavy concurrent access > to a relatively small amount of data (little enough to fit into Postgres > shared buffers, so that no I/O or kernel calls were really needed once > the test got going). I found that by nearly any measure --- elapsed > time, bus transactions, or machine-clear events --- the spinlock > acquisitions associated with grabbing and releasing the BufMgrLock took > an unreasonable fraction of the time. I saw about 15% of elapsed time, > 40% of bus transactions, and nearly 100% of pipeline-clear cycles going > into what is essentially two instructions out of the entire backend. > (Pipeline clears occur when the cache coherency logic detects a memory > write ordering problem.) > > I am not completely clear on why this machine-level bottleneck manifests > as a lot of context swaps at the OS level. I think what is happening is > that because SpinLockAcquire is so slow, a process is much more likely > than you'd normally expect to arrive at SpinLockAcquire while another > process is also acquiring the spinlock. This puts the two processes > into a "lockstep" condition where the second process is nearly certain > to observe the BufMgrLock as locked, and be forced to suspend itself, > even though the time the first process holds the BufMgrLock is not > really very long at all. > > If you google for Xeon and "cache coherency" you'll find quite a bit of > suggestive information about why this might be more true on the Xeon > setup than others. A couple of interesting hits: > > http://www.theinquirer.net/?article=10797 > says that Xeon MP uses a *slower* FSB than Xeon DP. This would > translate directly to more time needed to transfer a dirty cache line > from one processor to the other, which is the basic operation that we're > talking about here. > > http://www.aceshardware.com/Spades/read.php?article_id=30000187 > says that Opterons use a different cache coherency protocol that is > fundamentally superior to the Xeon's, because dirty cache data can be > transferred directly between two processor caches without waiting for > main memory. > > So in the short term I think we have to tell people that Xeon MP is not > the most desirable SMP platform to run Postgres on. (Josh thinks that > the specific motherboard chipset being used in these machines might > share some of the blame too. I don't have any evidence for or against > that idea, but it's certainly possible.) > > In the long run, however, CPUs continue to get faster than main memory > and the price of cache contention will continue to rise. So it seems > that we need to give up the assumption that SpinLockAcquire is a cheap > operation. In the presence of heavy contention it won't be. > > One thing we probably have got to do soon is break up the BufMgrLock > into multiple finer-grain locks so that there will be less contention. > However I am wary of doing this incautiously, because if we do it in a > way that makes for a significant rise in the number of locks that have > to be acquired to access a buffer, we might end up with a net loss. > > I think Neil Conway was looking into how the bufmgr might be > restructured to reduce lock contention, but if he had come up with > anything he didn't mention exactly what. Neil? > > regards, tom lane > > From pgsql-performance-owner@postgresql.org Mon Apr 19 09:30:17 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BAFCFD1CC8D for ; Mon, 19 Apr 2004 09:30:16 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 85835-08 for ; Mon, 19 Apr 2004 09:30:13 -0300 (ADT) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 681C3D1B515 for ; Mon, 19 Apr 2004 09:30:09 -0300 (ADT) Received: (qmail 12810 invoked from network); 19 Apr 2004 12:30:08 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 19 Apr 2004 12:30:08 -0000 Subject: Re: Wierd context-switching issue on Xeon From: Dave Cramer Reply-To: pg@fastcrypt.com To: Tom Lane Cc: lutzeb@aeccom.com, Josh Berkus , pgsql-performance@postgresql.org, Neil Conway In-Reply-To: <14394.1082344796@sss.pgh.pa.us> References: <200311251419.36771.josh@agliodbs.com> <200404151039.17763.josh@agliodbs.com> <407EE93F.3090705@aeccom.com> <200404151335.27762.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <1082331281.1557.47.camel@localhost.localdomain> <13960.1082341222@sss.pgh.pa.us> <14394.1082344796@sss.pgh.pa.us> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1082377953.1554.77.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Mon, 19 Apr 2004 08:32:33 -0400 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: 200404/268 X-Sequence-Number: 6568 Here's an interesting link that suggests that hyperthreading would be much worse. http://groups.google.com/groups?q=hyperthreading+dual+xeon+idle&start=10&hl=en&lr=&ie=UTF-8&c2coff=1&selm=aukkonen-FE5275.21093624062003%40shawnews.gv.shawcable.net&rnum=16 another which has some hints as to how it should be handled http://groups.google.com/groups?q=hyperthreading+dual+xeon+idle&start=10&hl=en&lr=&ie=UTF-8&c2coff=1&selm=u5tl1XD3BHA.2760%40tkmsftngp04&rnum=19 FWIW, I have anecdotal evidence that suggests that this is the case, on of my clients was seeing very large context switches with HTT turned on, and without it was much better. Dave On Sun, 2004-04-18 at 23:19, Tom Lane wrote: > >> What about hypterthreading does it still happen if HTT is turned off ? > > > The problem comes from keeping the caches synchronized between multiple > > physical CPUs. AFAICS enabling HTT wouldn't make it worse, because a > > hyperthreaded processor still only has one cache. > > Also, I forgot to say that the numbers I'm quoting *are* with HTT off. > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 8: explain analyze is your friend > > > > !DSPAM:40834781158911062514350! > > -- Dave Cramer 519 939 0336 ICQ # 14675561 From pgsql-performance-owner@postgresql.org Mon Apr 19 09:37:38 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7D314D1B9A6 for ; Mon, 19 Apr 2004 09:37:37 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88910-08 for ; Mon, 19 Apr 2004 09:37:37 -0300 (ADT) Received: from tht.net (vista.tht.net [216.126.88.2]) by svr1.postgresql.org (Postfix) with ESMTP id C6ECED1B4BE for ; Mon, 19 Apr 2004 09:37:33 -0300 (ADT) Received: from [134.22.69.199] (dyn-69-199.tor.dsl.tht.net [134.22.69.199]) by tht.net (Postfix) with ESMTP id BCE1F76A76; Mon, 19 Apr 2004 08:37:38 -0400 (EDT) Subject: Re: very slow simple query - outer join makes it quicker From: Rod Taylor To: gerard.isdell@kinetic.co.uk Cc: Postgresql Performance In-Reply-To: References: <1082376074.24414.3.camel@jester> Content-Type: text/plain Message-Id: <1082378194.24414.5.camel@jester> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Mon, 19 Apr 2004 08:36:34 -0400 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/269 X-Sequence-Number: 6569 On Mon, 2004-04-19 at 08:26, Gerard Isdell wrote: > Thank, that has worked. > > I've been running VACUUM regularly and thought that would have done it. > > Obviously the FULL makes a big difference It shouldn't. That FULL makes a significant difference says that you're not running regular VACUUM frequently enough and/or your fsm_* settings are too low. > -----Original Message----- > From: Rod Taylor > To: gerard.isdell@kinetic.co.uk > Cc: Postgresql Performance > Date: Mon, 19 Apr 2004 08:01:15 -0400 > Subject: Re: [PERFORM] very slow simple query - outer join makes it quicker > > > > There are other tables, such as the messages table, that have 10s of > > > thousands of rows and they return records much more quickly. > > > > > There must be something seriously wrong for simple queries like this > > to take > > > so long. > > > > Have you run VACUUM recently? > > > > If not, run VACUUM FULL against the users table and see if that makes a > > difference. > > > > > ************************************************************************* > This e-mail and any attachments may contain confidential or privileged > information. If you are not the intended recipient, please contact the > sender immediately and do not use, store or disclose their contents. > Any views expressed are those of the individual sender and not of Kinetic > Information System Services Limited unless otherwise stated. > > www.kinetic.co.uk From pgsql-performance-owner@postgresql.org Mon Apr 19 10:52:42 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id CAD1DD1C952 for ; Mon, 19 Apr 2004 10:52:40 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 22945-05 for ; Mon, 19 Apr 2004 10:52:41 -0300 (ADT) Received: from vt-pe2550-001.VANTAGE.vantage.com (unknown [64.80.203.242]) by svr1.postgresql.org (Postfix) with ESMTP id 5CC2AD1B9A6 for ; Mon, 19 Apr 2004 10:52:36 -0300 (ADT) Content-class: urn:content-classes:message Subject: Re: Wierd context-switching issue on Xeon MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0 Date: Mon, 19 Apr 2004 09:52:39 -0400 Message-ID: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509813F@vt-pe2550-001.vantage.vantage.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Wierd context-switching issue on Xeon Thread-Index: AcQlv1CsaIQyxA6cQsGxKSjJqw9QsQAVW1zS From: "Anjan Dave" To: "Greg Stark" , "Tom Lane" Cc: , "Josh Berkus" , , "Neil Conway" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/270 X-Sequence-Number: 6570 V2hhdCBhYm91dCBxdWFkLVhFT04gc2V0dXBzPyBDb3VsZCB0aGF0IGJlIHdv cnNlPyAoaGF2ZSBkdWFsLCBhbmQgcXVhZCBzZXR1cHMgYm90aCkgU2hhbGwg d2UgcmUtY29uc2lkZXIgWEVPTi1NUCBDUFUgbWFjaGluZXMgd2l0aCBoaWdo IGNhY2hlICg0TUIrKT8NCiANClZlcnkgZ2VuZXJhbGx5LCB3aGF0IG51bWJl ciB3b3VsZCBiZSBjb25zaWRlcmVkIGhpZ2gsIGVzcGVjaWFsbHksIGlmIGl0 IGNvaW5jaWRlcyB3aXRoIGV4cGVjdGVkIGhlYXZ5IGxvYWQ/DQogDQpOb3Qg c3VyZSBhIHNwZWNpZmljIGNoaXBzZXQgd2FzIG1lbnRpb25lZC4uLg0KIA0K VGhhbmtzLA0KQW5qYW4NCg0KCS0tLS0tT3JpZ2luYWwgTWVzc2FnZS0tLS0t IA0KCUZyb206IEdyZWcgU3RhcmsgW21haWx0bzpnc3N0YXJrQG1pdC5lZHVd IA0KCVNlbnQ6IFN1biA0LzE4LzIwMDQgODo0MCBQTSANCglUbzogVG9tIExh bmUgDQoJQ2M6IGx1dHplYkBhZWNjb20uY29tOyBKb3NoIEJlcmt1czsgcGdz cWwtcGVyZm9ybWFuY2VAcG9zdGdyZXNxbC5vcmc7IE5laWwgQ29ud2F5IA0K CVN1YmplY3Q6IFJlOiBbUEVSRk9STV0gV2llcmQgY29udGV4dC1zd2l0Y2hp bmcgaXNzdWUgb24gWGVvbg0KCQ0KCQ0KDQoNCglUb20gTGFuZSA8dGdsQHNz cy5wZ2gucGEudXM+IHdyaXRlczoNCgkNCgk+IFNvIGluIHRoZSBzaG9ydCB0 ZXJtIEkgdGhpbmsgd2UgaGF2ZSB0byB0ZWxsIHBlb3BsZSB0aGF0IFhlb24g TVAgaXMgbm90DQoJPiB0aGUgbW9zdCBkZXNpcmFibGUgU01QIHBsYXRmb3Jt IHRvIHJ1biBQb3N0Z3JlcyBvbi4gIChKb3NoIHRoaW5rcyB0aGF0DQoJPiB0 aGUgc3BlY2lmaWMgbW90aGVyYm9hcmQgY2hpcHNldCBiZWluZyB1c2VkIGlu IHRoZXNlIG1hY2hpbmVzIG1pZ2h0DQoJPiBzaGFyZSBzb21lIG9mIHRoZSBi bGFtZSB0b28uICBJIGRvbid0IGhhdmUgYW55IGV2aWRlbmNlIGZvciBvciBh Z2FpbnN0DQoJPiB0aGF0IGlkZWEsIGJ1dCBpdCdzIGNlcnRhaW5seSBwb3Nz aWJsZS4pDQoJPg0KCT4gSW4gdGhlIGxvbmcgcnVuLCBob3dldmVyLCBDUFVz IGNvbnRpbnVlIHRvIGdldCBmYXN0ZXIgdGhhbiBtYWluIG1lbW9yeQ0KCT4g YW5kIHRoZSBwcmljZSBvZiBjYWNoZSBjb250ZW50aW9uIHdpbGwgY29udGlu dWUgdG8gcmlzZS4gIFNvIGl0IHNlZW1zDQoJPiB0aGF0IHdlIG5lZWQgdG8g Z2l2ZSB1cCB0aGUgYXNzdW1wdGlvbiB0aGF0IFNwaW5Mb2NrQWNxdWlyZSBp cyBhIGNoZWFwDQoJPiBvcGVyYXRpb24uICBJbiB0aGUgcHJlc2VuY2Ugb2Yg aGVhdnkgY29udGVudGlvbiBpdCB3b24ndCBiZS4NCgkNCglUaGVyZSdzIG5v dGhpbmcgYWJvdXQgdGhlIHdheSBQb3N0Z3JlcyBzcGlubG9ja3MgYXJlIGNv ZGVkIHRoYXQgYWZmZWN0cyB0aGlzPw0KCQ0KCUlzIGl0IHNvbWV0aGluZyB0 aGUga2VybmVsIGNvdWxkIGhlbHAgd2l0aD8gSSd2ZSBiZWVuIHdvbmRlcmlu ZyB3aGV0aGVyDQoJdGhlcmUncyBhbnkgYmVuZWZpdHMgcG9zdGdyZXMgaXMg bWlzc2luZyBvdXQgb24gYnkgdXNpbmcgaXRzIG93biBoYW5kLXJvbGxlZA0K CWxvY2tpbmcgaW5zdGVhZCBvZiB1c2luZyB0aGUgcHRocmVhZHMgaW5mcmFz dHJ1Y3R1cmUgdGhhdCB0aGUga2VybmVsIGlzIG9mdGVuDQoJaW52b2x2ZWQg aW4uDQoJDQoJLS0NCglncmVnDQoJDQoJDQoJLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tKGVuZCBvZiBicm9hZGNhc3QpLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tDQoJVElQIDI6IHlvdSBjYW4gZ2V0IG9mZiBhbGwgbGlzdHMg YXQgb25jZSB3aXRoIHRoZSB1bnJlZ2lzdGVyIGNvbW1hbmQNCgkgICAgKHNl bmQgInVucmVnaXN0ZXIgWW91ckVtYWlsQWRkcmVzc0hlcmUiIHRvIG1ham9y ZG9tb0Bwb3N0Z3Jlc3FsLm9yZykNCgkNCg0K From pgsql-performance-owner@postgresql.org Mon Apr 19 13:00:25 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 2B12ED1E956 for ; Mon, 19 Apr 2004 13:00:21 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 77386-10 for ; Mon, 19 Apr 2004 13:00:19 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id D49F9D1E413 for ; Mon, 19 Apr 2004 13:00:16 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3JG0AQT025763; Mon, 19 Apr 2004 12:00:11 -0400 (EDT) To: Manfred Koizar Cc: Robert Treat , pgsql-performance@postgresql.org Subject: Re: query slows down with more accurate stats In-reply-to: References: <1081879359.25537.526.camel@camel> <25556.1081883922@sss.pgh.pa.us> <22642.1082074729@sss.pgh.pa.us> <29060.1082126089@sss.pgh.pa.us> Comments: In-reply-to Manfred Koizar message dated "Sat, 17 Apr 2004 00:26:22 +0200" Date: Mon, 19 Apr 2004 12:00:10 -0400 Message-ID: <25762.1082390410@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/271 X-Sequence-Number: 6571 Manfred Koizar writes: > Random sampling is more like "every possible sample is equally likely to > be collected", and two-stage sampling doesn't satisfy this condition. Okay, I finally see the point here: in the limit as the number of pages B goes to infinity, you'd expect the probability that each tuple in your sample came from a different page to go to 1. But this doesn't happen in the two-stage sampling method: the probability doesn't increase beyond the value it would have for B=n. On the average each sample page would supply one tuple, but the odds that this holds *exactly* would be pretty low. However the existing sampling method has glaring flaws of its own, in particular having to do with the fact that a tuple whose slot is preceded by N empty slots is N times more likely to be picked than one that has no empty-slot predecessors. The fact that the two-stage method artificially constrains the sample to come from only n pages seems like a minor problem by comparison; I'd happily accept it to get rid of the empty-slot bias. A possible compromise is to limit the number of pages sampled to something a bit larger than n, perhaps 2n or 3n. I don't have a feeling for the shape of the different-pages probability function; would this make a significant difference, or would it just waste cycles? regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 20 00:11:13 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B29A9D1CC8D for ; Tue, 20 Apr 2004 00:11:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 25282-03 for ; Tue, 20 Apr 2004 00:11:05 -0300 (ADT) Received: from web13121.mail.yahoo.com (web13121.mail.yahoo.com [216.136.174.83]) by svr1.postgresql.org (Postfix) with SMTP id E38C9D1DD91 for ; Tue, 20 Apr 2004 00:11:04 -0300 (ADT) Message-ID: <20040419162603.31251.qmail@web13121.mail.yahoo.com> Received: from [63.78.248.48] by web13121.mail.yahoo.com via HTTP; Mon, 19 Apr 2004 09:26:03 PDT Date: Mon, 19 Apr 2004 09:26:03 -0700 (PDT) From: Litao Wu Subject: Re: sunquery and estimated rows To: Tom Lane Cc: pgsql-performance@postgresql.org In-Reply-To: <1082316145.2191.0.camel@yarrow.bertheau.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/285 X-Sequence-Number: 6585 Well, the example shown is simplified version. Now, let's see a little 'real' example (still simplified version): Table test is same as before: \d test Table "public.test" Column | Type | Modifiers ---------+--------------------------+----------- id | integer | ... scope | integer | ... Indexes: test_scope_idx btree (scope) select count(*) from test; count ------- 4959 (1 row) select count(*) from test where scope=10; count ------- 10 (1 row) create table scope_def (scope int primary key, name varchar(30) unique); insert into scope_def values (10, 'TEST_SCOPE'); -- This is not a trivial arithmetic expression explain analyze select * from test where scope=(select scope from scope_def where name = 'TEST_SCOPE'); -- estimated row is 1653, returned rows is 10 QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------------ Index Scan using test_scope_idx on test (cost=0.00..49.91 rows=1653 width=59) (actual time=0.08..0.15 rows=10 loops=1) Index Cond: (scope = $0) InitPlan -> Index Scan using scope_def_name_key on scope_def (cost=0.00..4.82 rows=1 width=4) (actual time=0.04..0.04 rows=1 loops=1) Index Cond: (name = 'TEST_SCOPE'::character varying) Total runtime: 0.22 msec (6 rows) -- trivial arithmetic expression -- estimated row is 1653, returned rows is 10 explain analyze select * from test where scope=(select 10); QUERY PLAN ------------------------------------------------------------------------------------------------------------------------- Index Scan using test_scope_idx on test (cost=0.00..49.91 rows=1653 width=59) (actual time=0.06..0.14 rows=10 loops=1) Index Cond: (scope = $0) InitPlan -> Result (cost=0.00..0.01 rows=1 width=0) (actual time=0.01..0.01 rows=1 loops=1) Total runtime: 0.20 msec (5 rows) -- This is the plan I expect to see: estimated rows is -- close the actual returned rows. -- Do I have to devide the sub-select into two -- queries? explain analyze select * from test where scope=10; QUERY PLAN ---------------------------------------------------------------------------------------------------------------------- Index Scan using test_scope_idx on test (cost=0.00..3.77 rows=10 width=59) (actual time=0.05..0.12 rows=10 loops=1) Index Cond: (scope = 10) Total runtime: 0.18 msec (3 rows) -- Rewritten query using join in this case explain analyze select test.* from test JOIN scope_def using (scope) where scope_def.name = 'TEST_SCOPE'; QUERY PLAN ---------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=0.00..75.39 rows=5 width=63) (actual time=0.07..0.19 rows=10 loops=1) -> Index Scan using scope_def_name_key on scope_def (cost=0.00..4.82 rows=1 width=4) (actual time=0.04..0.04 rows=1 loops=1) Index Cond: (name = 'TEST_SCOPE'::character varying) -> Index Scan using test_scope_idx on test (cost=0.00..49.91 rows=1653 width=59) (actual time=0.02..0.09 rows=10 loops=1) Index Cond: (test.scope = "outer".scope) Total runtime: 0.28 msec (6 rows) __________________________________ Do you Yahoo!? Yahoo! Tax Center - File online by April 15th http://taxes.yahoo.com/filing.html From pgsql-performance-owner@postgresql.org Mon Apr 19 15:10:23 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B68ADD1E949 for ; Mon, 19 Apr 2004 14:39:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 26384-04 for ; Mon, 19 Apr 2004 14:39:02 -0300 (ADT) Received: from lorax.kcilink.com (lorax.kciLink.com [206.112.95.1]) by svr1.postgresql.org (Postfix) with ESMTP id CD11CD1E812 for ; Mon, 19 Apr 2004 14:39:01 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by lorax.kcilink.com (Postfix) with ESMTP id CEDED3F44 for ; Mon, 19 Apr 2004 13:39:01 -0400 (EDT) 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 25357-03-2 for ; Mon, 19 Apr 2004 13:39:01 -0400 (EDT) Received: by lorax.kcilink.com (Postfix, from userid 8) id 533453F1D; Mon, 19 Apr 2004 13:39:01 -0400 (EDT) To: pgsql-performance@postgresql.org Path: not-for-mail From: Vivek Khera Newsgroups: ml.postgres.performance Subject: Re: possible improvement between G4 and G5 Date: Mon, 19 Apr 2004 13:39:00 -0400 Organization: Khera Communications, Inc., Rockville, MD Lines: 26 Message-ID: References: <009901c41b48$8357c860$2500fa0a@CardServices.TCI.com> <200404061152.17997.josh@agliodbs.com> <200404061441.53598.josh@agliodbs.com> NNTP-Posting-Host: yertle.kcilink.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: lorax.kcilink.com 1082396341 55887 65.205.34.180 (19 Apr 2004 17:39:01 GMT) X-Complaints-To: daemon@kciLink.com NNTP-Posting-Date: Mon, 19 Apr 2004 17:39:01 +0000 (UTC) User-Agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Reasonable Discussion, berkeley-unix) Cancel-Lock: sha1:z/lJLpuNc9JdR8uVFivUP1NvE5Q= X-Virus-Scanned: by amavisd-new at kcilink.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/275 X-Sequence-Number: 6575 >>>>> "JB" == Josh Berkus writes: JB> Aaron, >> I do consulting, so they're all over the place and tend to be complex. Very >> few fit in RAM, but still are very buffered. These are almost all backed >> with very high end I/O subsystems, with dozens of spindles with battery >> backed up writethrough cache and gigs of buffers, which may be why I worry >> so much about CPU. I have had this issue with multiple servers. JB> Aha, I think this is the difference. I never seem to be able to JB> get my clients to fork out for adequate disk support. They are JB> always running off single or double SCSI RAID in the host server; JB> not the sort of setup you have. Even when I upgraded my system to a 14-spindle RAID5 with 128M cache and 4GB RAM on a dual Xeon system, I still wind up being I/O bound quite often. I think it depends on what your "working set" turns out to be. My workload really spans a lot more of the DB than I can end up caching. -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= 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 Mon Apr 19 14:52:52 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7F964D1E98D for ; Mon, 19 Apr 2004 14:50:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31056-02 for ; Mon, 19 Apr 2004 14:50:25 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id DD481D1E95F for ; Mon, 19 Apr 2004 14:50:23 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4900262; Mon, 19 Apr 2004 10:51:46 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: Tom Lane , lutzeb@aeccom.com Subject: Re: Wierd context-switching issue on Xeon Date: Mon, 19 Apr 2004 10:50:12 -0700 User-Agent: KMail/1.4.3 Cc: pgsql-performance@postgreSQL.org, Neil Conway References: <200311251419.36771.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> In-Reply-To: <11437.1082324861@sss.pgh.pa.us> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404191050.12833.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/272 X-Sequence-Number: 6572 Tom, > So in the short term I think we have to tell people that Xeon MP is not > the most desirable SMP platform to run Postgres on. (Josh thinks that > the specific motherboard chipset being used in these machines might > share some of the blame too. I don't have any evidence for or against > that idea, but it's certainly possible.) I have 3 reasons for thinking this: 1) the ServerWorks chipset is present in the fully documented cases that we have of this problem so far. This is notable becuase the SW is notorious for poor manufacturing quality, so much so that the company that made them is currently in receivership. These chips were so bad that Dell was forced to recall several hundred of it's 2650's, where the motherboards caught fire! 2) the main defect of the SW is the NorthBridge, which could conceivably adversely affect traffic between RAM and the processor cache. 3) XeonMP is a very popular platform thanks to Dell, and we are not seeing more problem reports than we are. The other thing I'd like your comment on, Tom, is that Dirk appears to have reported that when he installed a non-bigmem kernel, the issue went away. Dirk, is this correct? -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Mon Apr 19 15:02:34 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D75BDD1D52E for ; Mon, 19 Apr 2004 14:55:59 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 29582-05 for ; Mon, 19 Apr 2004 14:55:51 -0300 (ADT) Received: from mx1.neopolitan.us (mx1.neopolitan.us [65.87.16.224]) by svr1.postgresql.org (Postfix) with ESMTP id 3E6DDD1C9E1 for ; Mon, 19 Apr 2004 14:55:50 -0300 (ADT) Received: from [65.87.16.98] (HELO [10.0.0.210]) by mx1.neopolitan.us (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 3984841 for pgsql-performance@postgresql.org; Mon, 19 Apr 2004 10:55:50 -0700 Subject: Re: Wierd context-switching issue on Xeon From: "J. Andrew Rogers" To: pgsql-performance@postgresql.org In-Reply-To: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509813F@vt-pe2550-001.vantage.vantage.com> References: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509813F@vt-pe2550-001.vantage.vantage.com> Content-Type: text/plain Organization: Message-Id: <1082397350.30934.34.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 19 Apr 2004 10:55:50 -0700 Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/274 X-Sequence-Number: 6574 I decided to check the context-switching behavior here for baseline since we have a rather diverse set of postgres server hardware, though nothing using Xeon MP that is also running a postgres instance, and everything looks normal under load. Some platforms are better than others, but nothing is outside of what I would consider normal bounds. Our biggest database servers are Opteron SMP systems, and these servers are particularly well-behaved under load with Postgres 7.4.2. If there is a problem with the locking code and context-switching, it sure isn't manifesting on our Opteron SMP systems. Under rare confluences of process interaction, we occasionally see short spikes in the 2-3,000 cs/sec range. It typically peaks at a couple hundred cs/sec under load. Obviously this is going to be a function of our load profile a certain extent. The Opterons have proven to be very good database hardware in general for us. j. andrew rogers From pgsql-performance-owner@postgresql.org Mon Apr 19 15:02:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B5154D1DCA0 for ; Mon, 19 Apr 2004 15:00:12 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28608-10 for ; Mon, 19 Apr 2004 15:00:07 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id EFA2FD1E98A for ; Mon, 19 Apr 2004 15:00:05 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3JI01Mi026942; Mon, 19 Apr 2004 14:00:01 -0400 (EDT) To: Josh Berkus Cc: lutzeb@aeccom.com, pgsql-performance@postgreSQL.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <200404191050.12833.josh@agliodbs.com> References: <200311251419.36771.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <200404191050.12833.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Mon, 19 Apr 2004 10:50:12 -0700" Date: Mon, 19 Apr 2004 14:00:01 -0400 Message-ID: <26941.1082397601@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/273 X-Sequence-Number: 6573 Josh Berkus writes: > The other thing I'd like your comment on, Tom, is that Dirk appears to have > reported that when he installed a non-bigmem kernel, the issue went away. > Dirk, is this correct? I'd be really surprised if that had anything to do with it. AFAIR Dirk's test changed more than one variable and so didn't prove a connection. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 19 15:33:25 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 7FADED1B8DF for ; Mon, 19 Apr 2004 15:33:24 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 45171-08 for ; Mon, 19 Apr 2004 15:33:23 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 14ECED1B49B for ; Mon, 19 Apr 2004 15:33:21 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i3JIWxx10909; Mon, 19 Apr 2004 14:32:59 -0400 (EDT) From: Bruce Momjian Message-Id: <200404191832.i3JIWxx10909@candle.pha.pa.us> Subject: Re: Wierd context-switching issue on Xeon In-Reply-To: <200404191050.12833.josh@agliodbs.com> To: Josh Berkus Date: Mon, 19 Apr 2004 14:32:59 -0400 (EDT) Cc: Tom Lane , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway 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: 200404/276 X-Sequence-Number: 6576 Josh Berkus wrote: > Tom, > > > So in the short term I think we have to tell people that Xeon MP is not > > the most desirable SMP platform to run Postgres on. (Josh thinks that > > the specific motherboard chipset being used in these machines might > > share some of the blame too. I don't have any evidence for or against > > that idea, but it's certainly possible.) > > I have 3 reasons for thinking this: > 1) the ServerWorks chipset is present in the fully documented cases that we > have of this problem so far. This is notable becuase the SW is notorious > for poor manufacturing quality, so much so that the company that made them is > currently in receivership. These chips were so bad that Dell was forced to > recall several hundred of it's 2650's, where the motherboards caught fire! > 2) the main defect of the SW is the NorthBridge, which could conceivably > adversely affect traffic between RAM and the processor cache. > 3) XeonMP is a very popular platform thanks to Dell, and we are not seeing > more problem reports than we are. > > The other thing I'd like your comment on, Tom, is that Dirk appears to have > reported that when he installed a non-bigmem kernel, the issue went away. I have BSD on a SuperMicro dual Xeon, so if folks want another hardware/OS combination to test, I can give out logins to my machine. http://candle.pha.pa.us/main/hardware.html -- 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 Apr 19 17:16:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 9DD12D1B895 for ; Mon, 19 Apr 2004 17:16:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 89674-08 for ; Mon, 19 Apr 2004 17:16:02 -0300 (ADT) Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222]) by svr1.postgresql.org (Postfix) with ESMTP id 5A59FD1B49B for ; Mon, 19 Apr 2004 17:16:00 -0300 (ADT) 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 i3JKETu9019523; Mon, 19 Apr 2004 14:14:29 -0600 (MDT) Date: Mon, 19 Apr 2004 14:12:32 -0600 (MDT) From: "scott.marlowe" To: Bruce Momjian Cc: Josh Berkus , Tom Lane , , , Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-Reply-To: <200404191832.i3JIWxx10909@candle.pha.pa.us> 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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/277 X-Sequence-Number: 6577 On Mon, 19 Apr 2004, Bruce Momjian wrote: > Josh Berkus wrote: > > Tom, > > > > > So in the short term I think we have to tell people that Xeon MP is not > > > the most desirable SMP platform to run Postgres on. (Josh thinks that > > > the specific motherboard chipset being used in these machines might > > > share some of the blame too. I don't have any evidence for or against > > > that idea, but it's certainly possible.) > > > > I have 3 reasons for thinking this: > > 1) the ServerWorks chipset is present in the fully documented cases that we > > have of this problem so far. This is notable becuase the SW is notorious > > for poor manufacturing quality, so much so that the company that made them is > > currently in receivership. These chips were so bad that Dell was forced to > > recall several hundred of it's 2650's, where the motherboards caught fire! > > 2) the main defect of the SW is the NorthBridge, which could conceivably > > adversely affect traffic between RAM and the processor cache. > > 3) XeonMP is a very popular platform thanks to Dell, and we are not seeing > > more problem reports than we are. > > > > The other thing I'd like your comment on, Tom, is that Dirk appears to have > > reported that when he installed a non-bigmem kernel, the issue went away. > > I have BSD on a SuperMicro dual Xeon, so if folks want another > hardware/OS combination to test, I can give out logins to my machine. I can probably do some nighttime testing on a dual 2800MHz non-MP Xeon machine as well. It's a Dell 2600 series machine and very fast. It has the moderately fast 533MHz FSB so may not have as many problems as the MP type CPUs seem to be having. From pgsql-performance-owner@postgresql.org Tue Apr 20 10:58:39 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5F590D1B49B for ; Mon, 19 Apr 2004 17:41:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 94623-09 for ; Mon, 19 Apr 2004 17:41:08 -0300 (ADT) Received: from hotmail.com (bay9-dav33.bay9.hotmail.com [64.4.46.90]) by svr1.postgresql.org (Postfix) with ESMTP id 72086D1DB5B for ; Mon, 19 Apr 2004 17:41:05 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Mon, 19 Apr 2004 13:41:07 -0700 Received: from 67.81.102.201 by bay9-dav33.bay9.hotmail.com with DAV; Mon, 19 Apr 2004 20:41:06 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman@hotmail.com] X-Sender: awerman@hotmail.com From: "Aaron Werman" To: References: <200311251419.36771.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <200404191050.12833.josh@agliodbs.com> Subject: Re: possible improvement between G4 and G5 Date: Mon, 19 Apr 2004 16:41:22 -0400 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 6.00.2800.1409 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409 Message-ID: X-OriginalArrivalTime: 19 Apr 2004 20:41:07.0060 (UTC) FILETIME=[A3CC5B40:01C4264E] X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/302 X-Sequence-Number: 6602 There are a few things that you can do to help force yourself to be I/O bound. These include: - RAID 5 for write intensive applications, since multiple writes per synch write is good. (There is a special case for logging or other streaming sequential writes on RAID 5) - Data journaling file systems are helpful in stress testing your checkpoints - Using midsized battery backed up write through buffering controllers. In general, if you have a small cache, you see the problem directly, and a huge cache will balance out load and defer writes to quieter times. That is why a midsized cache is so useful in showing stress in your system only when it is being stressed. Only partly in jest, /Aaron BTW - I am truly curious about what happens to your system if you use separate RAID 0+1 for your logs, disk sorts, and at least the most active tables. This should reduce I/O load by an order of magnitude. "Vivek Khera" wrote in message news:x7smez7tqj.fsf@yertle.int.kciLink.com... > >>>>> "JB" == Josh Berkus writes: > > JB> Aaron, > >> I do consulting, so they're all over the place and tend to be complex. Very > >> few fit in RAM, but still are very buffered. These are almost all backed > >> with very high end I/O subsystems, with dozens of spindles with battery > >> backed up writethrough cache and gigs of buffers, which may be why I worry > >> so much about CPU. I have had this issue with multiple servers. > > JB> Aha, I think this is the difference. I never seem to be able to > JB> get my clients to fork out for adequate disk support. They are > JB> always running off single or double SCSI RAID in the host server; > JB> not the sort of setup you have. > > Even when I upgraded my system to a 14-spindle RAID5 with 128M cache > and 4GB RAM on a dual Xeon system, I still wind up being I/O bound > quite often. > > I think it depends on what your "working set" turns out to be. My > workload really spans a lot more of the DB than I can end up caching. > > -- > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= > 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/ > > ---------------------------(end of broadcast)--------------------------- > TIP 8: explain analyze is your friend > From pgsql-performance-owner@postgresql.org Tue Apr 20 10:59:48 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1BEC3D1E944 for ; Mon, 19 Apr 2004 17:50:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05121-05 for ; Mon, 19 Apr 2004 17:50:55 -0300 (ADT) Received: from main.gmane.org (main.gmane.org [80.91.224.249]) by svr1.postgresql.org (Postfix) with ESMTP id 6CD83D1E8A5 for ; Mon, 19 Apr 2004 17:50:52 -0300 (ADT) Received: from root by main.gmane.org with local (Exim 3.35 #1 (Debian)) id 1BFfjA-0007Rp-00 for ; Mon, 19 Apr 2004 22:50:52 +0200 Received: from tanagra.summersault.com ([12.161.105.149]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Mon, 19 Apr 2004 22:50:52 +0200 Received: from mark by tanagra.summersault.com with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Mon, 19 Apr 2004 22:50:52 +0200 X-Injected-Via-Gmane: http://gmane.org/ To: pgsql-performance@postgresql.org From: Mark Stosberg Subject: seeking consultant for high performance, complex searching with Postgres web app Date: Mon, 19 Apr 2004 20:42:06 +0000 (UTC) Lines: 40 Message-ID: X-Complaints-To: usenet@sea.gmane.org X-Gmane-NNTP-Posting-Host: tanagra.summersault.com X-Archive: encrypt User-Agent: slrn/0.9.8.0 (FreeBSD) X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/303 X-Sequence-Number: 6603 Hello, I work for Summersault, LLC. We've been using Postgres since the days of Postgres 6.5. We're focused on building database-driven websites using Perl and Postgres. We are currently seeking help developing a search system that needs to perform complex queries with high performance. Although we have strong skills in Perl and Postgres, we are new to the arena of complex, high-performance search systems. We are seeking to hire a consultant to help this as part of the re-vamp of the 1-800-Save-A-Pet.com website. 1-800-Save-A-Pet.com is a not-for-profit organization whose website finds homes for homeless pets, promoting pet adoption and saving thousands of animal lives. Summersault, LLC is a website development firm focused on creating highly customized database driven websites. The ideal consultant has expert experience with the PostgreSQL RDBMS and the Perl programming language, and is intimately familiar with the architecture and implementation of complex database queries for high-traffic web applications. The consultant should also have a strong background in creating solutions complementary to this work, e.g. assessing hardware requirements, designing a hosting and network infrastructure, and optimizing the algorithm based on real-world feedback. The consultant will work with Summersault developers as a part of a larger application development process. Interested persons or organizations should contact Chris Hardie of Summersault, LLC at chris@summersault.com for more information. Thanks! Mark -- . . . . . . . . . . . . . . . . . . . . . . . . . . . Mark Stosberg Principal Developer mark@summersault.com Summersault, LLC 765-939-9301 ext 202 database driven websites . . . . . http://www.summersault.com/ . . . . . . . . From pgsql-performance-owner@postgresql.org Mon Apr 19 18:02:56 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0EDF4D1D189 for ; Mon, 19 Apr 2004 18:02:55 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 09275-05 for ; Mon, 19 Apr 2004 18:02:53 -0300 (ADT) Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82]) by svr1.postgresql.org (Postfix) with ESMTP id 5D213D1D08B for ; Mon, 19 Apr 2004 18:02:50 -0300 (ADT) Received: from [206.19.64.3] (account jconway HELO joeconway.com) by joeconway.com (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 1439829; Mon, 19 Apr 2004 13:56:56 -0700 Message-ID: <40843E63.7050101@joeconway.com> Date: Mon, 19 Apr 2004 14:02:27 -0700 From: Joe Conway User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "scott.marlowe" Cc: Bruce Momjian , Josh Berkus , Tom Lane , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon 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: 200404/278 X-Sequence-Number: 6578 scott.marlowe wrote: > On Mon, 19 Apr 2004, Bruce Momjian wrote: >>I have BSD on a SuperMicro dual Xeon, so if folks want another >>hardware/OS combination to test, I can give out logins to my machine. > > I can probably do some nighttime testing on a dual 2800MHz non-MP Xeon > machine as well. It's a Dell 2600 series machine and very fast. It has > the moderately fast 533MHz FSB so may not have as many problems as the MP > type CPUs seem to be having. I've got a quad 2.8Ghz MP Xeon (IBM x445) that I could test on. Does anyone have a test set that can reliably reproduce the problem? Joe From pgsql-performance-owner@postgresql.org Tue Apr 20 11:01:38 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 12C99D1CC8D for ; Mon, 19 Apr 2004 18:58:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 28070-08 for ; Mon, 19 Apr 2004 18:58:55 -0300 (ADT) Received: from mailout05.sul.t-online.com (mailout05.sul.t-online.com [194.25.134.82]) by svr1.postgresql.org (Postfix) with ESMTP id 60CC0D1CACF for ; Mon, 19 Apr 2004 18:58:52 -0300 (ADT) Received: from fwd04.aul.t-online.de by mailout05.sul.t-online.com with smtp id 1BFgAB-0006lt-01; Mon, 19 Apr 2004 23:18:47 +0200 Received: from aeccom.com (Ek4eAcZGQeKPbfWeYc-9iHLKRPA1WT-9F+6OpXlIg+UbP1he0hDTYg@[217.231.172.140]) by fwd04.sul.t-online.com with esmtp id 1BFg9s-13rlK40; Mon, 19 Apr 2004 23:18:28 +0200 Message-ID: <40844223.9070200@aeccom.com> Date: Mon, 19 Apr 2004 23:18:27 +0200 From: Dirk.Lutzebaeck@t-online.de (Dirk Lutzebaeck) Reply-To: lutzeb@aeccom.com 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: Tom Lane Cc: Josh Berkus , pgsql-performance@postgreSQL.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon References: <200311251419.36771.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <200404191050.12833.josh@agliodbs.com> <26941.1082397601@sss.pgh.pa.us> In-Reply-To: <26941.1082397601@sss.pgh.pa.us> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Seen: false X-ID: Ek4eAcZGQeKPbfWeYc-9iHLKRPA1WT-9F+6OpXlIg+UbP1he0hDTYg X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/304 X-Sequence-Number: 6604 I would agree to Tom, that too much parameters are involved to blame bigmem. I have access to the following machines where the same application operates: a) Dual (4way) XEON MP, bigmem, HT off, ServerWorks chipset (a Fujitsu-Siemens Primergy) performs ok now because missing indexes were added but this is no proof that this behaviour occurs again under high load, context switches are moderate but have peaks to 40.000 b) Dual XEON DP, non-bigmem, HT on, ServerWorks chipset (a Dell machine I think) performs moderate because I see too much context switches here although the mentioned indexes are created, context switches go up to 30.000 often, I can see 50% semop calls c) Dual XEON DP, non-bigmem, HT on, E7500 Intel chipset (Supermicro) performs well and I could not observe context switch peaks here (one user active), almost no extra semop calls d) Dual XEON DP, bigmem, HT off, ServerWorks chipset (a Fujitsu-Siemens Primergy) performance unknown at the moment (is offline) but looks like a) in the past I can offer to do tests on those machines if somebody would provide me some test instructions to nail this problem down. Dirk Tom Lane wrote: >Josh Berkus writes: > > >>The other thing I'd like your comment on, Tom, is that Dirk appears to have >>reported that when he installed a non-bigmem kernel, the issue went away. >>Dirk, is this correct? >> >> > >I'd be really surprised if that had anything to do with it. AFAIR >Dirk's test changed more than one variable and so didn't prove a >connection. > > regards, tom lane > > > From pgsql-performance-owner@postgresql.org Mon Apr 19 18:54:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A6E96D1BAAE for ; Mon, 19 Apr 2004 18:54:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 27459-07 for ; Mon, 19 Apr 2004 18:54:49 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 1AE67D1B581 for ; Mon, 19 Apr 2004 18:54:47 -0300 (ADT) Received: from [64.81.245.111] (HELO 192.168.1.102) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4902060; Mon, 19 Apr 2004 14:56:14 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Joe Conway , "scott.marlowe" Subject: Re: Wierd context-switching issue on Xeon Date: Mon, 19 Apr 2004 14:55:04 -0700 User-Agent: KMail/1.5.4 Cc: Bruce Momjian , Tom Lane , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway References: <40843E63.7050101@joeconway.com> In-Reply-To: <40843E63.7050101@joeconway.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404191455.04068.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/279 X-Sequence-Number: 6579 Joe, > I've got a quad 2.8Ghz MP Xeon (IBM x445) that I could test on. Does > anyone have a test set that can reliably reproduce the problem? Unfortunately we can't seem to come up with one. So far we have 2 machines that exhibit the issue, and their databases are highly confidential (State of WA education data). It does seem to require a database which is in the many GB (> 10GB), and a situation where a small subset of the data is getting hit repeatedly by multiple processes. So you could try your own data warehouse, making sure that you have at least 4 connections hitting one query after another. -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Mon Apr 19 19:56:40 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 875FCD1B7F0 for ; Mon, 19 Apr 2004 19:56:39 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 50509-01 for ; Mon, 19 Apr 2004 19:56:39 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 8698CD1B4BE for ; Mon, 19 Apr 2004 19:56:36 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3JMtYNE029785; Mon, 19 Apr 2004 18:55:35 -0400 (EDT) To: josh@agliodbs.com Cc: Joe Conway , "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <200404191455.04068.josh@agliodbs.com> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Mon, 19 Apr 2004 14:55:04 -0700" Date: Mon, 19 Apr 2004 18:55:34 -0400 Message-ID: <29784.1082415334@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/280 X-Sequence-Number: 6580 Josh Berkus writes: >> I've got a quad 2.8Ghz MP Xeon (IBM x445) that I could test on. Does >> anyone have a test set that can reliably reproduce the problem? > Unfortunately we can't seem to come up with one. > It does seem to require a database which is in the many GB (> 10GB), and a > situation where a small subset of the data is getting hit repeatedly by > multiple processes. I do not think a large database is actually necessary; the test case Josh's client has is only hitting a relatively small amount of data. The trick seems to be to cause lots and lots of ReadBuffer/ReleaseBuffer activity without much else happening, and to do this from multiple backends concurrently. I believe the best way to make this happen is a lot of relatively simple (but not short) indexscan queries that in aggregate touch just a bit less than shared_buffers worth of data. I have not tried to make a self-contained test case, but based on what I know now I think it should be possible. I'll give this a shot later tonight --- it does seem that trying to reproduce the problem on different kinds of hardware is the next useful step we can take. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 19 21:02:52 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5389AD1BAF5 for ; Mon, 19 Apr 2004 21:02:50 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 64666-06 for ; Mon, 19 Apr 2004 21:02:51 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id D9C26D1D189 for ; Mon, 19 Apr 2004 21:02:47 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3K01udf001408; Mon, 19 Apr 2004 20:01:57 -0400 (EDT) To: josh@agliodbs.com Cc: Joe Conway , "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <200404191455.04068.josh@agliodbs.com> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Mon, 19 Apr 2004 14:55:04 -0700" MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----- =_aaaaaaaaaa0" Content-ID: <1376.1082419088.0@sss.pgh.pa.us> Date: Mon, 19 Apr 2004 20:01:56 -0400 Message-ID: <1407.1082419316@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/281 X-Sequence-Number: 6581 ------- =_aaaaaaaaaa0 Content-Type: text/plain; charset="us-ascii" Content-ID: <1376.1082419088.1@sss.pgh.pa.us> Here is a test case. To set up, run the "test_setup.sql" script once; then launch two copies of the "test_run.sql" script. (For those of you with more than two CPUs, see whether you need one per CPU to make trouble, or whether two test_runs are enough.) Check that you get a nestloops-with-index-scans plan shown by the EXPLAIN in test_run. In isolation, test_run.sql should do essentially no syscalls at all once it's past the initial ramp-up. On a machine that's functioning per expectations, multiple copies of test_run show a relatively low rate of semop() calls --- a few per second, at most --- and maybe a delaying select() here and there. What I actually see on Josh's client's machine is a context swap storm: "vmstat 1" shows CS rates around 170K/sec. strace'ing the backends shows a corresponding rate of semop() syscalls, with a few delaying select()s sprinkled in. top(1) shows system CPU percent of 25-30 and idle CPU percent of 16-20. I haven't bothered to check how long the test_run query takes, but if it ends while you're still examining the behavior, just start it again. Note the test case assumes you've got shared_buffers set to at least 1000; with smaller values, you may get some I/O syscalls, which will probably skew the results. regards, tom lane ------- =_aaaaaaaaaa0 Content-Type: text/plain; charset="us-ascii" Content-ID: <1376.1082419088.2@sss.pgh.pa.us> Content-Description: test_setup.sql drop table test_data; create table test_data(f1 int); insert into test_data values (random() * 100); insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; insert into test_data select random() * 100 from test_data; create index test_index on test_data(f1); vacuum verbose analyze test_data; checkpoint; ------- =_aaaaaaaaaa0 Content-Type: text/plain; charset="us-ascii" Content-ID: <1376.1082419088.3@sss.pgh.pa.us> Content-Description: test_run.sql -- force nestloop indexscan plan set enable_seqscan to 0; set enable_mergejoin to 0; set enable_hashjoin to 0; explain select count(*) from test_data a, test_data b, test_data c where a.f1 = b.f1 and b.f1 = c.f1; select count(*) from test_data a, test_data b, test_data c where a.f1 = b.f1 and b.f1 = c.f1; ------- =_aaaaaaaaaa0-- From pgsql-performance-owner@postgresql.org Mon Apr 19 21:53:58 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 76BA3D1DD51 for ; Mon, 19 Apr 2004 21:53:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80625-04 for ; Mon, 19 Apr 2004 21:53:59 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 74A3AD1DC7D for ; Mon, 19 Apr 2004 21:53:55 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3K0r9aj001932; Mon, 19 Apr 2004 20:53:09 -0400 (EDT) To: josh@agliodbs.com Cc: Joe Conway , "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-reply-to: <1407.1082419316@sss.pgh.pa.us> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> Comments: In-reply-to Tom Lane message dated "Mon, 19 Apr 2004 20:01:56 -0400" Date: Mon, 19 Apr 2004 20:53:09 -0400 Message-ID: <1931.1082422389@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/282 X-Sequence-Number: 6582 I wrote: > Here is a test case. Hmmm ... I've been able to reproduce the CS storm on a dual Athlon, which seems to pretty much let the Xeon per se off the hook. Anybody got a multiple Opteron to try? Totally non-Intel CPUs? It would be interesting to see results with non-Linux kernels, too. regards, tom lane From pgsql-performance-owner@postgresql.org Tue Apr 20 00:00:32 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 665C5D1BAAE for ; Tue, 20 Apr 2004 00:00:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 16732-06 for ; Tue, 20 Apr 2004 00:00:26 -0300 (ADT) Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82]) by svr1.postgresql.org (Postfix) with ESMTP id A2E14D1B973 for ; Tue, 20 Apr 2004 00:00:25 -0300 (ADT) Received: from [206.19.64.3] (account jconway HELO joeconway.com) by joeconway.com (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 1446269; Mon, 19 Apr 2004 19:54:27 -0700 Message-ID: <40849235.2070808@joeconway.com> Date: Mon, 19 Apr 2004 20:00:05 -0700 From: Joe Conway User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.6) Gecko/20040113 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: josh@agliodbs.com, "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> In-Reply-To: <1407.1082419316@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: 200404/284 X-Sequence-Number: 6584 Tom Lane wrote: > Here is a test case. To set up, run the "test_setup.sql" script once; > then launch two copies of the "test_run.sql" script. (For those of > you with more than two CPUs, see whether you need one per CPU to make > trouble, or whether two test_runs are enough.) Check that you get a > nestloops-with-index-scans plan shown by the EXPLAIN in test_run. Check. > In isolation, test_run.sql should do essentially no syscalls at all once > it's past the initial ramp-up. On a machine that's functioning per > expectations, multiple copies of test_run show a relatively low rate of > semop() calls --- a few per second, at most --- and maybe a delaying > select() here and there. > > What I actually see on Josh's client's machine is a context swap storm: > "vmstat 1" shows CS rates around 170K/sec. strace'ing the backends > shows a corresponding rate of semop() syscalls, with a few delaying > select()s sprinkled in. top(1) shows system CPU percent of 25-30 > and idle CPU percent of 16-20. Your test case works perfectly. I ran 4 concurrent psql sessions, on a quad Xeon (IBM x445, 2.8GHz, 4GB RAM), hyperthreaded. Heres what 'top' looks like: 177 processes: 173 sleeping, 3 running, 1 zombie, 0 stopped CPU states: cpu user nice system irq softirq iowait idle total 35.9% 0.0% 7.2% 0.0% 0.0% 0.0% 56.8% cpu00 19.6% 0.0% 4.9% 0.0% 0.0% 0.0% 75.4% cpu01 44.1% 0.0% 7.8% 0.0% 0.0% 0.0% 48.0% cpu02 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 100.0% cpu03 32.3% 0.0% 13.7% 0.0% 0.0% 0.0% 53.9% cpu04 21.5% 0.0% 10.7% 0.0% 0.0% 0.0% 67.6% cpu05 42.1% 0.0% 9.8% 0.0% 0.0% 0.0% 48.0% cpu06 100.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% cpu07 27.4% 0.0% 10.7% 0.0% 0.0% 0.0% 61.7% Mem: 4123700k av, 3933896k used, 189804k free, 0k shrd, 221948k buff 2492124k actv, 760612k in_d, 41416k in_c Swap: 2040244k av, 5632k used, 2034612k free 3113272k cached Note that cpu06 is not a postgres process. The output of vmstat looks like this: # vmstat 1 procs memory swap io system cpu r b swpd free buff cache si so bi bo in cs us sy id wa 4 0 5632 184264 221948 3113308 0 0 0 0 0 0 0 0 0 0 3 0 5632 184264 221948 3113308 0 0 0 0 112 211894 36 9 55 0 5 0 5632 184264 221948 3113308 0 0 0 0 125 222071 39 8 53 0 4 0 5632 184264 221948 3113308 0 0 0 0 110 215097 39 10 52 0 1 0 5632 184588 221948 3113308 0 0 0 96 139 187561 35 10 55 0 3 0 5632 184588 221948 3113308 0 0 0 0 114 241731 38 10 52 0 3 0 5632 184920 221948 3113308 0 0 0 0 132 257168 40 9 51 0 1 0 5632 184912 221948 3113308 0 0 0 0 114 251802 38 9 54 0 > Note the test case assumes you've got shared_buffers set to at least > 1000; with smaller values, you may get some I/O syscalls, which will > probably skew the results. shared_buffers ---------------- 16384 (1 row) I found that killing three of the four concurrent queries dropped context switches to about 70,000 to 100,000. Two or more sessions brings it up to 200K+. Joe From pgsql-performance-owner@postgresql.org Tue Apr 20 00:37:58 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id E656ED1D137 for ; Tue, 20 Apr 2004 00:37:57 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 33925-03 for ; Tue, 20 Apr 2004 00:37:54 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id CBD34D1D06C for ; Tue, 20 Apr 2004 00:37:53 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id ; Mon, 19 Apr 2004 23:37:48 -0400 Message-ID: From: "Shea,Dan [CIS]" To: pgsql-performance@postgresql.org Subject: Why will vacuum not end? Date: Mon, 19 Apr 2004 23:37:47 -0400 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: 200404/286 X-Sequence-Number: 6586 This vacuum is running a marathon. Why will it not end and show me free space map INFO? We have deleted a lot of data and I would like to be confident that these deletions will be used as free space, rather than creating more table files. PWFPM_DEV=# select now();vacuum verbose forecastelement;select now(); now ------------------------------- 2004-04-14 18:36:13.725285+00 (1 row) INFO: vacuuming "public.forecastelement" INFO: index "forecastelement_rwv_idx" now contains 473380072 row versions in 4986653 pages DETAIL: 5592106 index row versions were removed. 44688 index pages have been deleted, 1 are currently reusable. CPU 4942.30s/336.27u sec elapsed 74710.07 sec. INFO: "forecastelement": removed 5592106 row versions in 126370 pages DETAIL: CPU 58.43s/16.99u sec elapsed 366.24 sec. INFO: index "forecastelement_rwv_idx" now contains 472296119 row versions in 5027529 pages DETAIL: 5592097 index row versions were removed. 89120 index pages have been deleted, 0 are currently reusable. CPU 4260.08s/327.29u sec elapsed 59592.38 sec. INFO: "forecastelement": removed 5592097 row versions in 124726 pages DETAIL: CPU 33.38s/14.21u sec elapsed 210.36 sec. INFO: index "forecastelement_rwv_idx" now contains 467784889 row versions in 5037914 pages DETAIL: 5592089 index row versions were removed. 134286 index pages have been deleted, 0 are currently reusable. CPU 4185.86s/318.19u sec elapsed 57048.65 sec. INFO: "forecastelement": removed 5592089 row versions in 121657 pages DETAIL: CPU 51.19s/14.19u sec elapsed 238.31 sec. INFO: index "forecastelement_rwv_idx" now contains 462960132 row versions in 5039886 pages DETAIL: 5592067 index row versions were removed. 179295 index pages have been deleted, 0 are currently reusable. CPU 4002.76s/313.63u sec elapsed 54806.09 sec. INFO: "forecastelement": removed 5592067 row versions in 122510 pages DETAIL: CPU 25.32s/14.47u sec elapsed 187.73 sec. INFO: index "forecastelement_rwv_idx" now contains 457555142 row versions in 5041631 pages DETAIL: 5592085 index row versions were removed. 224480 index pages have been deleted, 0 are currently reusable. CPU 4149.42s/310.94u sec elapsed 55880.65 sec. INFO: "forecastelement": removed 5592085 row versions in 122500 pages DETAIL: CPU 16.70s/14.47u sec elapsed 180.27 sec. INFO: index "forecastelement_rwv_idx" now contains 452191660 row versions in 5044414 pages DETAIL: 5592089 index row versions were removed. 269665 index pages have been deleted, 0 are currently reusable. CPU 4213.10s/304.61u sec elapsed 55159.36 sec. INFO: "forecastelement": removed 5592089 row versions in 122663 pages DETAIL: CPU 37.28s/14.63u sec elapsed 206.96 sec. INFO: index "forecastelement_rwv_idx" now contains 446807778 row versions in 5046541 pages DETAIL: 5592077 index row versions were removed. 314747 index pages have been deleted, 0 are currently reusable. CPU 4039.49s/297.15u sec elapsed 55086.56 sec. INFO: "forecastelement": removed 5592077 row versions in 122558 pages DETAIL: CPU 20.21s/14.74u sec elapsed 227.53 sec. From pgsql-performance-owner@postgresql.org Tue Apr 20 00:47:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4B8B8D1C4C2 for ; Tue, 20 Apr 2004 00:47:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 37226-02 for ; Tue, 20 Apr 2004 00:47:50 -0300 (ADT) Received: from srvr3.iniquinet.com (srvr2.iniquinet.com [64.240.87.12]) by svr1.postgresql.org (Postfix) with SMTP id 6AEDDD1B482 for ; Tue, 20 Apr 2004 00:47:49 -0300 (ADT) Received: (qmail 14697 invoked by uid 104); 20 Apr 2004 03:47:47 -0000 Received: from Robert_Creager@LogicalChaos.org by srvr3.iniquinet.com by uid 101 with qmail-scanner-1.15 (clamscan: 0.65. spamassassin: 2.55. Clear:SA:0(-7.4/6.0):. Processed in 32.951673 secs); 20 Apr 2004 03:47:47 -0000 Received: from unknown (HELO thunder.mshome.net) (216.58.165.126) by srvr3.iniquinet.com with SMTP; 20 Apr 2004 03:47:14 -0000 Received: from localhost.localdomain (thunder.mshome.net [192.168.0.250]) by thunder.mshome.net (Postfix) with ESMTP id 998C994857; Mon, 19 Apr 2004 21:47:13 -0600 (MDT) Received: from logicalchaos.org (thunder.mshome.net [192.168.0.250]) by thunder.mshome.net (Postfix) with SMTP id 263B694857; Mon, 19 Apr 2004 21:47:10 -0600 (MDT) Date: Mon, 19 Apr 2004 21:47:02 -0600 From: Robert Creager To: Tom Lane Cc: josh@agliodbs.com, Joe Conway , "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon Message-Id: <20040419214702.70e5c9b6@thunder.mshome.net> In-Reply-To: <1931.1082422389@sss.pgh.pa.us> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us> Organization: Starlight Vision, LLC. X-Mailer: Sylpheed version 0.9.9claws (GTK+ 1.2.10; i586-mandrake-linux-gnu) Mime-Version: 1.0 Content-Type: multipart/signed; protocol="application/pgp-signature"; micalg="pgp-sha1"; boundary="Signature=_Mon__19_Apr_2004_21_47_02_-0600_TKk0qNlhGAUFCcfZ" X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200404/287 X-Sequence-Number: 6587 --Signature=_Mon__19_Apr_2004_21_47_02_-0600_TKk0qNlhGAUFCcfZ Content-Type: text/plain; charset=US-ASCII Content-Disposition: inline Content-Transfer-Encoding: 7bit When grilled further on (Mon, 19 Apr 2004 20:53:09 -0400), Tom Lane confessed: > I wrote: > > Here is a test case. > > Hmmm ... I've been able to reproduce the CS storm on a dual Athlon, > which seems to pretty much let the Xeon per se off the hook. Anybody > got a multiple Opteron to try? Totally non-Intel CPUs? > > It would be interesting to see results with non-Linux kernels, too. > Same problem on my dual AMD MP with 2.6.5 kernel using two sessions of your test, but maybe not quite as severe. The highest CS values I saw was 102k, with some non-db number crunching going on in parallel with the test. 'Average' about 80k with two instances. Using the anticipatory scheduler. A single instance pulls in around 200-300 CS, and no tests running around 200-300 CS (i.e. no CS difference). A snipet: procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- 3 0 284 90624 93452 1453740 0 0 0 0 1075 76548 83 17 0 0 6 0 284 125312 93452 1470196 0 0 0 0 1073 87702 78 22 0 0 3 0 284 178392 93460 1420208 0 0 76 298 1083 67721 77 24 0 0 4 0 284 177120 93460 1421500 0 0 1104 0 1054 89593 80 21 0 0 5 0 284 173504 93460 1425172 0 0 3584 0 1110 65536 81 19 0 0 4 0 284 169984 93460 1428708 0 0 3456 0 1098 66937 81 20 0 0 6 0 284 170944 93460 1428708 0 0 8 0 1045 66065 81 19 0 0 6 0 284 167288 93460 1428776 0 0 0 8 1097 75560 81 19 0 0 6 0 284 136296 93460 1458356 0 0 0 0 1036 80808 75 26 0 0 5 0 284 132864 93460 1461688 0 0 0 0 1007 76071 84 17 0 0 4 0 284 132880 93460 1461688 0 0 0 0 1079 86903 82 18 0 0 5 0 284 132880 93460 1461688 0 0 0 0 1078 79885 83 17 0 0 6 0 284 132648 93460 1461688 0 0 0 760 1228 66564 86 14 0 0 6 0 284 132648 93460 1461688 0 0 0 0 1047 69741 86 15 0 0 6 0 284 132672 93460 1461688 0 0 0 0 1057 79052 84 16 0 0 5 0 284 132672 93460 1461688 0 0 0 0 1054 81109 82 18 0 0 5 0 284 132736 93460 1461688 0 0 0 0 1043 91725 80 20 0 0 Cheers, Rob -- 21:33:03 up 3 days, 1:10, 3 users, load average: 5.05, 4.67, 4.22 Linux 2.6.5-01 #5 SMP Tue Apr 6 21:32:39 MDT 2004 --Signature=_Mon__19_Apr_2004_21_47_02_-0600_TKk0qNlhGAUFCcfZ Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iEYEARECAAYFAkCEnT4ACgkQLQ/DKuwDYzlP+gCfUEjce4uOlds2UYi74gQF4x74 RZoAn1NxiPBUks7VXrPJesieH371NWYp =cWZK -----END PGP SIGNATURE----- --Signature=_Mon__19_Apr_2004_21_47_02_-0600_TKk0qNlhGAUFCcfZ-- From pgsql-performance-owner@postgresql.org Tue Apr 20 00:57:35 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 1EB7BD1DC8D for ; Tue, 20 Apr 2004 00:57:34 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 35593-07 for ; Tue, 20 Apr 2004 00:57:32 -0300 (ADT) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 22AA7D1DCA0 for ; Tue, 20 Apr 2004 00:57:31 -0300 (ADT) 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 i3K3vEWL018842; Tue, 20 Apr 2004 11:57:14 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <4084A0D0.5060305@familyhealth.com.au> Date: Tue, 20 Apr 2004 12:02:24 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Shea,Dan [CIS]" Cc: pgsql-performance@postgresql.org Subject: Re: Why will vacuum not end? 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: 200404/288 X-Sequence-Number: 6588 > This vacuum is running a marathon. Why will it not end and show me free > space map INFO? We have deleted a lot of data and I would like to be > confident that these deletions will be used as free space, rather than > creating more table files. Does another postgres query running have a lock on that table? Chris From pgsql-performance-owner@postgresql.org Tue Apr 20 02:20:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5F2BBD1CCCF for ; Tue, 20 Apr 2004 02:20:09 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 60265-05 for ; Tue, 20 Apr 2004 02:20:08 -0300 (ADT) Received: from mtaw4.prodigy.net (mtaw4.prodigy.net [64.164.98.52]) by svr1.postgresql.org (Postfix) with ESMTP id BDC07D1C9E6 for ; Tue, 20 Apr 2004 02:20:06 -0300 (ADT) Received: from local2 (adsl-216-100-44-178.dsl.snfc21.pacbell.net [216.100.44.178]) by mtaw4.prodigy.net (8.12.10/8.12.10) with ESMTP id i3K5K55l012466 for ; Mon, 19 Apr 2004 22:20:05 -0700 (PDT) Date: Mon, 19 Apr 2004 22:18:21 -0700 (PDT) From: jelle X-X-Sender: jelle@localhost.localdomain Reply-To: jellej@pacbell.net To: pgsql-performance@postgresql.org Subject: Re: Wierd context-switching issue on Xeon In-Reply-To: <1931.1082422389@sss.pgh.pa.us> 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: 200404/289 X-Sequence-Number: 6589 Same problem with dual 1Ghz P3's running Postgres 7.4.2, linux 2.4.x, and 2GB ram, under load, with long transactions (i.e. 1 "cannot serialize" rollback per minute). 200K was the worst observed with vmstat. Finally moved DB to a single xeon box. From pgsql-performance-owner@postgresql.org Tue Apr 20 07:35:54 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id ED88FD1B7F0 for ; Tue, 20 Apr 2004 07:35:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 58446-06 for ; Tue, 20 Apr 2004 07:35:55 -0300 (ADT) Received: from server.pyrenet.fr (server.pyrenet.fr [194.250.190.1]) by svr1.postgresql.org (Postfix) with ESMTP id 8F332D1B49B for ; Tue, 20 Apr 2004 07:35:51 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by localhost.pyrenet.fr (Postfix) with ESMTP id 906D2295D1; Tue, 20 Apr 2004 12:35:53 +0200 (MET DST) Received: from server.pyrenet.fr ([127.0.0.1]) by localhost (server [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 03467-03; Tue, 20 Apr 2004 12:35:50 +0200 (MET DST) Received: from server.pyrenet.fr (server.pyrenet.fr [194.250.190.1]) by server.pyrenet.fr (Postfix) with ESMTP id 4AD71295D0; Tue, 20 Apr 2004 12:35:50 +0200 (MET DST) Date: Tue, 20 Apr 2004 12:35:50 +0200 (MET DST) From: ohp@pyrenet.fr Reply-To: ohp@pyrenet.fr To: Tom Lane Cc: josh@agliodbs.com, Joe Conway , "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon In-Reply-To: <1931.1082422389@sss.pgh.pa.us> Message-ID: References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at pyrenet.fr X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=NO_REAL_NAME X-Spam-Level: X-Archive-Number: 200404/290 X-Sequence-Number: 6590 Hi Tom, You still have an account on my Unixware Bi-Xeon hyperthreded machine. Feel free to use it for your tests. On Mon, 19 Apr 2004, Tom Lane wrote: > Date: Mon, 19 Apr 2004 20:53:09 -0400 > From: Tom Lane > To: josh@agliodbs.com > Cc: Joe Conway , scott.marlowe , > Bruce Momjian , lutzeb@aeccom.com, > pgsql-performance@postgresql.org, Neil Conway > Subject: Re: [PERFORM] Wierd context-switching issue on Xeon > > I wrote: > > Here is a test case. > > Hmmm ... I've been able to reproduce the CS storm on a dual Athlon, > which seems to pretty much let the Xeon per se off the hook. Anybody > got a multiple Opteron to try? Totally non-Intel CPUs? > > It would be interesting to see results with non-Linux kernels, too. > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster > -- Olivier PRENANT Tel: +33-5-61-50-97-00 (Work) 6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax) 31190 AUTERIVE +33-6-07-63-80-64 (GSM) FRANCE Email: ohp@pyrenet.fr ------------------------------------------------------------------------------ Make your life a dream, make your dream a reality. (St Exupery) From pgsql-performance-owner@postgresql.org Tue Apr 20 08:16:01 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F3B9ED1D08B for ; Tue, 20 Apr 2004 08:15:58 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 80842-02 for ; Tue, 20 Apr 2004 08:16:00 -0300 (ADT) Received: from biglumber.com (biglumber.com [207.228.252.42]) by svr1.postgresql.org (Postfix) with SMTP id 520C4D1B482 for ; Tue, 20 Apr 2004 08:15:56 -0300 (ADT) Received: (qmail 24833 invoked from network); 20 Apr 2004 11:15:58 -0000 Received: from unknown (HELO localhost) (207.228.252.42) by 0 with SMTP; 20 Apr 2004 11:15:58 -0000 From: "Greg Sabino Mullane" To: pgsql-performance@postgresql.org Subject: Re: Deleting certain duplicates X-PGP-Key: 2529 DF6A B8F7 9407 E944 45B4 BC9B 9067 1496 4AC8 X-Request-PGP: http://www.biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8 In-Reply-To: Date: Tue, 20 Apr 2004 11:15:58 -0000 X-Mailer: JoyMail 1.47 Message-ID: <82428c807d3c47fbcd0c013dbcab12b5@biglumber.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/291 X-Sequence-Number: 6591 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > So I was thinking maybe of doing the deletion in chunks, perhaps based on > reception time. > Are there any suggestions for a better way to do this, or using multiple > queries to delete selectively a week at a time based on the reception_time. > I would say there are a lot of duplicate entries between mid march to the > first week of April. You are on the right track, in that dividing up the table will help. However, you cannot divide on the reception_time as that is the unique column. Analyze your data and divide on a row with a fairly uniform distribution over the time period in question. Then copy a segment out, clean it up, and put it back in. Make sure there is an index on the column in question, of course. For example, if 1/10 of the table has a "units" of 12, you could do something like this: CREATE INDEX units_dev ON forecastelement (units); CREATE TEMPORARY TABLE units_temp AS SELECT * FROM forecastelement WHERE units='12'; CREATE INDEX units_oid_index ON units_temp(oid); (Delete out duplicate rows from units_temp using your previous query or something else) DELETE FROM forecastelement WHERE units='12'; INSERT INTO forecastelement SELECT * FROM units_temp; DELETE FROM units_temp; Repeat as needed until all rows are done. Subsequent runs can be done by doing a INSERT INTO units_temp SELECT * FROM forecastelement WHERE units='...' and skipping the CREATE INDEX steps. On the other hand, your original deletion query may work as is, with the addition of an oid index. Perhaps try an EXPLAIN on it. - -- Greg Sabino Mullane greg@turnstep.com PGP Key: 0x14964AC8 200404200706 -----BEGIN PGP SIGNATURE----- iD8DBQFAhQVWvJuQZxSWSsgRAvLEAKDCVcX3Llm8JgszI/BBC1SobtjVawCfVGKu ERcV5J2JolwgZRhMbXnNM90= =JqET -----END PGP SIGNATURE----- From pgsql-performance-owner@postgresql.org Tue Apr 20 09:46:13 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B7A56D1DD56 for ; Tue, 20 Apr 2004 09:46:11 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 01149-10 for ; Tue, 20 Apr 2004 09:46:12 -0300 (ADT) Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com [24.128.241.68]) by svr1.postgresql.org (Postfix) with SMTP id A7497D1DD51 for ; Tue, 20 Apr 2004 09:46:08 -0300 (ADT) Received: (qmail 34366 invoked from network); 20 Apr 2004 12:46:15 -0000 Received: from localhost (HELO ?10.124.7.139?) (threshar@127.0.0.1) by localhost with SMTP; 20 Apr 2004 12:46:15 -0000 In-Reply-To: <1407.1082419316@sss.pgh.pa.us> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v613) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit Cc: Bruce Momjian , Joe Conway , Neil Conway , josh@agliodbs.com, lutzeb@aeccom.com, pgsql-performance@postgresql.org, "scott.marlowe" From: Jeff Subject: Re: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 08:46:12 -0400 To: Tom Lane X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=2.7 tagged_above=0.0 required=5.0 tests=RCVD_IN_DYNABLOCK, RCVD_IN_SORBS X-Spam-Level: ** X-Archive-Number: 200404/292 X-Sequence-Number: 6592 On Apr 19, 2004, at 8:01 PM, Tom Lane wrote: [test case] Quad P3-700Mhz, ServerWorks, pg 7.4.2 - 1 process: 10-30 cs / second 2 process: 100k cs / sec 3 process: 140k cs / sec 8 process: 115k cs / sec Dual P2-450Mhz, non-serverworks (piix) - 1 process 15-20 / sec 2 process 30k / sec 3 (up to 7) process: 15k /sec (Yes, I verified with more processes the cs's drop) And finally, 6 cpu sun e4500, solaris 2.6, pg 7.4.2: 1 - 10 processes: hovered between 2-3k cs/second (there was other stuff running on the machine as well) Verrry interesting. I've got a dual G4 at home, but for convenience Apple doesn't ship a vmstat that tells context switches -- Jeff Trout http://www.jefftrout.com/ http://www.stuarthamm.net/ From pgsql-performance-owner@postgresql.org Tue Apr 20 09:58:20 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 3C3BED1BB93 for ; Tue, 20 Apr 2004 09:58:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08156-09 for ; Tue, 20 Apr 2004 09:58:21 -0300 (ADT) Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca [199.212.17.130]) by svr1.postgresql.org (Postfix) with ESMTP id 962FBD1D2B7 for ; Tue, 20 Apr 2004 09:58:08 -0300 (ADT) Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19) id ; Tue, 20 Apr 2004 08:58:09 -0400 Message-ID: From: "Shea,Dan [CIS]" To: 'Christopher Kings-Lynne' Cc: pgsql-performance@postgresql.org Subject: Re: Why will vacuum not end? Date: Tue, 20 Apr 2004 08:58:09 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/293 X-Sequence-Number: 6593 No, but data is constantly being inserted by userid scores. It is postgres runnimg the vacuum. Dan. -----Original Message----- From: Christopher Kings-Lynne [mailto:chriskl@familyhealth.com.au] Sent: Tuesday, April 20, 2004 12:02 AM To: Shea,Dan [CIS] Cc: pgsql-performance@postgresql.org Subject: Re: [PERFORM] Why will vacuum not end? > This vacuum is running a marathon. Why will it not end and show me free > space map INFO? We have deleted a lot of data and I would like to be > confident that these deletions will be used as free space, rather than > creating more table files. Does another postgres query running have a lock on that table? Chris From pgsql-performance-owner@postgresql.org Tue Apr 20 10:04:31 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F20B8D1DD59 for ; Tue, 20 Apr 2004 10:04:27 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 08341-08 for ; Tue, 20 Apr 2004 10:04:28 -0300 (ADT) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id AE033D1DD42 for ; Tue, 20 Apr 2004 10:04:24 -0300 (ADT) Received: (qmail 21581 invoked from network); 20 Apr 2004 13:04:20 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 20 Apr 2004 13:04:20 -0000 Subject: Re: Wierd context-switching issue on Xeon From: Dave Cramer Reply-To: pg@fastcrypt.com To: Jeff Cc: Tom Lane , Bruce Momjian , Joe Conway , Neil Conway , Josh Berkus , lutzeb@aeccom.com, pgsql-performance@postgresql.org, "scott.marlowe" In-Reply-To: References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1082466419.3069.132.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Tue, 20 Apr 2004 09:06:59 -0400 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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/294 X-Sequence-Number: 6594 Dual Athlon With one process running 30 cs/second with two process running 15000 cs/second Dave On Tue, 2004-04-20 at 08:46, Jeff wrote: > On Apr 19, 2004, at 8:01 PM, Tom Lane wrote: > [test case] > > Quad P3-700Mhz, ServerWorks, pg 7.4.2 - 1 process: 10-30 cs / second > 2 process: 100k cs / sec > 3 process: 140k cs / sec > 8 process: 115k cs / sec > > Dual P2-450Mhz, non-serverworks (piix) - 1 process 15-20 / sec > 2 process 30k / sec > 3 (up to 7) process: 15k /sec > > (Yes, I verified with more processes the cs's drop) > > And finally, > > 6 cpu sun e4500, solaris 2.6, pg 7.4.2: 1 - 10 processes: hovered > between 2-3k cs/second (there was other stuff running on the machine as > well) > > > Verrry interesting. > I've got a dual G4 at home, but for convenience Apple doesn't ship a > vmstat that tells context switches > > -- > Jeff Trout > http://www.jefftrout.com/ > http://www.stuarthamm.net/ > > > ---------------------------(end of broadcast)--------------------------- > TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/docs/faqs/FAQ.html > > > > !DSPAM:40851da1199651145780980! > > -- Dave Cramer 519 939 0336 ICQ # 14675561 From pgsql-performance-owner@postgresql.org Tue Apr 20 10:14:04 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id EA6CAD1D2B5 for ; Tue, 20 Apr 2004 10:14:02 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14734-09 for ; Tue, 20 Apr 2004 10:14:05 -0300 (ADT) Received: from internet.potentialtech.com (h-66-167-251-6.phlapafg.covad.net [66.167.251.6]) by svr1.postgresql.org (Postfix) with ESMTP id 377CED1D238 for ; Tue, 20 Apr 2004 10:14:01 -0300 (ADT) Received: from potentialtech.com (pa-plum1c-102.pit.adelphia.net [24.53.179.102]) by internet.potentialtech.com (Postfix) with ESMTP id A556669A71; Tue, 20 Apr 2004 09:14:04 -0400 (EDT) Message-ID: <40852218.2050204@potentialtech.com> Date: Tue, 20 Apr 2004 09:14:00 -0400 From: Bill Moran User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.4) Gecko/20031005 X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Shea,Dan [CIS]" Cc: 'Christopher Kings-Lynne' , pgsql-performance@postgresql.org Subject: Re: Why will vacuum not end? 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-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=RCVD_IN_RFCI, RCVD_IN_SORBS X-Spam-Level: X-Archive-Number: 200404/295 X-Sequence-Number: 6595 Shea,Dan [CIS] wrote: > No, but data is constantly being inserted by userid scores. It is postgres > runnimg the vacuum. > Dan. > > -----Original Message----- > From: Christopher Kings-Lynne [mailto:chriskl@familyhealth.com.au] > Sent: Tuesday, April 20, 2004 12:02 AM > To: Shea,Dan [CIS] > Cc: pgsql-performance@postgresql.org > Subject: Re: [PERFORM] Why will vacuum not end? > >>This vacuum is running a marathon. Why will it not end and show me free >>space map INFO? We have deleted a lot of data and I would like to be >>confident that these deletions will be used as free space, rather than >>creating more table files. > > Does another postgres query running have a lock on that table? This may be a dumb question (but only because I don't know the answer) Doesn't/shouldn't vacuum have some kind of timeout so if a table is locked it will give up eventually (loudly complaining when it does so)? -- Bill Moran Potential Technologies http://www.potentialtech.com From pgsql-performance-owner@postgresql.org Tue Apr 20 10:44:42 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 53643D1DB0C for ; Tue, 20 Apr 2004 10:44:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 39727-02 for ; Tue, 20 Apr 2004 10:44:42 -0300 (ADT) 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 CB157D1D2CC for ; Tue, 20 Apr 2004 10:44:37 -0300 (ADT) 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 E9FA29B774; Tue, 20 Apr 2004 14:44:40 +0100 (BST) From: "Matt Clark" To: "Tom Lane" , Cc: "Joe Conway" , "scott.marlowe" , "Bruce Momjian" , , , "Neil Conway" Subject: Re: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 14:44:40 +0100 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 9.0.6604 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409 In-Reply-To: <1407.1082419316@sss.pgh.pa.us> Importance: Normal X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/296 X-Sequence-Number: 6596 As a cross-ref to all the 7.4.x tests people have sent in, here's 7.2.3 (Redhat 7.3), Quad Xeon 700MHz/1MB L2 cache, 3GB RAM. Idle-ish (it's a production server) cs/sec ~5000 3 test queries running: procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 3 0 0 23380 577680 105912 2145140 0 0 0 0 107 116890 50 14 35 2 0 0 23380 577680 105912 2145140 0 0 0 0 114 118583 50 15 34 2 0 0 23380 577680 105912 2145140 0 0 0 0 107 115842 54 14 32 2 1 0 23380 577680 105920 2145140 0 0 0 32 156 117549 50 16 35 HTH Matt > -----Original Message----- > From: pgsql-performance-owner@postgresql.org > [mailto:pgsql-performance-owner@postgresql.org]On Behalf Of Tom Lane > Sent: 20 April 2004 01:02 > To: josh@agliodbs.com > Cc: Joe Conway; scott.marlowe; Bruce Momjian; lutzeb@aeccom.com; > pgsql-performance@postgresql.org; Neil Conway > Subject: Re: [PERFORM] Wierd context-switching issue on Xeon > > > Here is a test case. To set up, run the "test_setup.sql" script once; > then launch two copies of the "test_run.sql" script. (For those of > you with more than two CPUs, see whether you need one per CPU to make > trouble, or whether two test_runs are enough.) Check that you get a > nestloops-with-index-scans plan shown by the EXPLAIN in test_run. > > In isolation, test_run.sql should do essentially no syscalls at all once > it's past the initial ramp-up. On a machine that's functioning per > expectations, multiple copies of test_run show a relatively low rate of > semop() calls --- a few per second, at most --- and maybe a delaying > select() here and there. > > What I actually see on Josh's client's machine is a context swap storm: > "vmstat 1" shows CS rates around 170K/sec. strace'ing the backends > shows a corresponding rate of semop() syscalls, with a few delaying > select()s sprinkled in. top(1) shows system CPU percent of 25-30 > and idle CPU percent of 16-20. > > I haven't bothered to check how long the test_run query takes, but if it > ends while you're still examining the behavior, just start it again. > > Note the test case assumes you've got shared_buffers set to at least > 1000; with smaller values, you may get some I/O syscalls, which will > probably skew the results. > > regards, tom lane > > From pgsql-performance-owner@postgresql.org Tue Apr 20 12:39:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D8E05D1D08B for ; Tue, 20 Apr 2004 11:19:47 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 48887-09 for ; Tue, 20 Apr 2004 11:19:50 -0300 (ADT) Received: from mail.astudios.com (unknown [66.49.77.3]) by svr1.postgresql.org (Postfix) with ESMTP id 5111FD1C9E1 for ; Tue, 20 Apr 2004 11:19:45 -0300 (ADT) thread-index: AcQm4v/045T4I1TUSGSZhIXZa9046w== Received: from [10.4.2.163] ([66.194.26.150]) by mail.astudios.com with Microsoft SMTPSVC(6.0.3790.0); Tue, 20 Apr 2004 10:23:06 -0400 Content-Class: urn:content-classes:message Importance: normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132 From: "Chris Hoover" To: Subject: Use of subquery causes seq scan??? Date: Tue, 20 Apr 2004 10:20:05 -0400 User-Agent: KMail/1.6.1 References: <200404151248.16067.jeremy.guthrie@berbee.com> In-Reply-To: <200404151248.16067.jeremy.guthrie@berbee.com> MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Message-ID: <200404201020.05699.revoohc@sermonaudio.com> X-OriginalArrivalTime: 20 Apr 2004 14:23:07.0093 (UTC) FILETIME=[FFE5C850:01C426E2] X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/307 X-Sequence-Number: 6607 I need some help. I have a query that refuses to use the provided index an= d=20 is always sequentially scanning causing me large performance headaches. He= re=20 is the basic situation: Table A: inv_num int type char =2E =2E =2E pkey (inv_num, type) indx(inv_num) Table B (has the same primary key) Select * from table a where inv_num in (select inv_num from table b where ....) Doing this causes sequential scans of both tables. If I do a set=20 enable_seqscan to false before the query, I get an index scan of table b bu= t=20 still seq scan table a.=20 Is there anyway to force table a to use this index (or another) and not=20 sequentially scan the table? I'm running 7.3.4 on RedHat EL 2.1. Thanks, Chris From pgsql-performance-owner@postgresql.org Tue Apr 20 12:28:59 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 242CAD1C952 for ; Tue, 20 Apr 2004 11:29:05 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 56581-09 for ; Tue, 20 Apr 2004 11:29:02 -0300 (ADT) Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de [212.202.101.158]) by svr1.postgresql.org (Postfix) with ESMTP id A927CD1BB89 for ; Tue, 20 Apr 2004 11:28:58 -0300 (ADT) Received: from aeccom.com (cayambe.core.aeccom.com [192.168.2.12]) by mail2.aeccom.com (Postfix) with ESMTP id DCF8E34; Tue, 20 Apr 2004 16:29:01 +0200 (CEST) Message-ID: <408533AD.40904@aeccom.com> Date: Tue, 20 Apr 2004 16:29:01 +0200 From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Organization: AEC/communications GmbH, Berlin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031016 X-Accept-Language: de, en MIME-Version: 1.0 To: Tom Lane , Josh Berkus Cc: pgsql-performance@postgreSQL.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon References: <200311251419.36771.josh@agliodbs.com> <407EFC95.5070805@aeccom.com> <11437.1082324861@sss.pgh.pa.us> <200404191050.12833.josh@agliodbs.com> <26941.1082397601@sss.pgh.pa.us> <40844223.9070200@aeccom.com> In-Reply-To: <40844223.9070200@aeccom.com> Content-Type: text/plain; charset=us-ascii; format=flowed 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= X-Spam-Level: X-Archive-Number: 200404/306 X-Sequence-Number: 6606 Dirk Lutzebaeck wrote: > c) Dual XEON DP, non-bigmem, HT on, E7500 Intel chipset (Supermicro) > > performs well and I could not observe context switch peaks here (one > user active), almost no extra semop calls Did Tom's test here: with 2 processes I'll reach 200k+ CS with peaks to 300k CS. Bummer.. Josh, I don't think you can bash the ServerWorks chipset here nor bigmem. Dirk From pgsql-performance-owner@postgresql.org Tue Apr 20 11:48:38 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B448ED1CB0F for ; Tue, 20 Apr 2004 11:48:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 69090-02 for ; Tue, 20 Apr 2004 11:48:17 -0300 (ADT) Received: from net2.micro-automation.com (net2.micro-automation.com [64.7.141.29]) by svr1.postgresql.org (Postfix) with SMTP id 0BE17D1B967 for ; Tue, 20 Apr 2004 11:48:13 -0300 (ADT) Received: (qmail 10804 invoked from network); 20 Apr 2004 14:48:14 -0000 Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116) by net2.micro-automation.com with SMTP; 20 Apr 2004 14:48:14 -0000 Subject: Re: Toooo many context switches (maybe SLES8?) From: Dave Cramer Reply-To: pg@fastcrypt.com To: lutzeb@aeccom.com Cc: Tom Lane , Joe Conway , pgsql-performance@postgresql.org In-Reply-To: <407EEA9B.7010807@aeccom.com> References: <407E7B99.3050306@aeccom.com> <407EB1E5.4030600@joeconway.com> <407EB86B.6080801@aeccom.com> <407EBA89.6080007@joeconway.com> <20498.1082057841@sss.pgh.pa.us> <407EEA9B.7010807@aeccom.com> Content-Type: text/plain Organization: Cramer Consulting Message-Id: <1082472652.1554.155.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.6 Date: Tue, 20 Apr 2004 10:50:52 -0400 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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/305 X-Sequence-Number: 6605 Don't think so, mine is a vanilla kernel from kernel.org Dave On Thu, 2004-04-15 at 16:03, Dirk Lutzebaeck wrote: > Could this be related to the O(1) scheduler backpatches from 2.6 to 2.4 > kernel on newer 2.4er distros (RedHat, SuSE)? > > > Tom Lane wrote: > > >Joe Conway writes: > > > > > >>>Improve spinlock code for recent x86 processors: insert a PAUSE > >>>instruction in the s_lock() wait loop, and use test before test-and-set > >>>in TAS() macro to avoid unnecessary bus traffic. Patch from Manfred > >>>Spraul, reworked a bit by Tom. > >>> > >>> > > > > > > > >>I thought this had been committed to the 7.4 stable branch as well, but > >>it appears not. > >> > >> > > > >I am currently chasing what seems to be the same issue: massive context > >swapping on a dual Xeon system. I tried back-patching the above-mentioned > >patch ... it helps a little but by no means solves the problem ... > > > > regards, tom lane > > > > > > > > > > ---------------------------(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 > > > > !DSPAM:408535ce93801252113544! > > -- Dave Cramer 519 939 0336 ICQ # 14675561 From pgsql-performance-owner@postgresql.org Tue Apr 20 22:50:45 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 90446D1E98F; Tue, 20 Apr 2004 22:50:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 17447-03; Tue, 20 Apr 2004 22:50:21 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 9C7A1D1EC56; Tue, 20 Apr 2004 12:48:09 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i3KFmA814010; Tue, 20 Apr 2004 11:48:10 -0400 (EDT) From: Bruce Momjian Message-Id: <200404201548.i3KFmA814010@candle.pha.pa.us> Subject: pgbench written in Pro*C To: PostgreSQL-development Date: Tue, 20 Apr 2004 11:48:09 -0400 (EDT) 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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/323 X-Sequence-Number: 6623 I received a copy of pgbench rewritten in Pro*C, which is similar to embedded C. I think it was done so the same program could be tested on Oracle and PostgreSQL. Are folks interested in this code? Should it be put on gborg or in our /contrib/pgbench? -- 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 Tue Apr 20 22:51:26 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8FC6AD1EAA2 for ; Tue, 20 Apr 2004 22:51:23 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 16846-08 for ; Tue, 20 Apr 2004 22:51:19 -0300 (ADT) Received: from stubee.d2hosting.net (d2hosting.net [66.70.41.160]) by svr1.postgresql.org (Postfix) with ESMTP id C80EFD1ECD2 for ; Tue, 20 Apr 2004 12:49:29 -0300 (ADT) Received: from idigx.com (stubee.d2hosting.net [66.70.41.160]) by stubee.d2hosting.net (8.11.6/linuxconf) with ESMTP id i3KFnLM24882; Tue, 20 Apr 2004 10:49:26 -0500 Message-ID: <40854698.6070805@idigx.com> Date: Tue, 20 Apr 2004 10:49:44 -0500 From: Thomas Swan User-Agent: Mozilla Thunderbird 0.6a (Windows/20040419) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Randolf Richardson Cc: pgsql-performance@postgresql.org Subject: Re: Persistent Connections References: <20040124173209.933905A8170@kserver17.erfurt12.de> <4013A85C.1010406@chuckie.co.uk> In-Reply-To: X-Enigmail-Version: 0.83.6.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS X-Spam-Level: X-Archive-Number: 200404/325 X-Sequence-Number: 6625 Randolf Richardson wrote: >"nicky@chuckie.co.uk (Nick Barr)" stated in >comp.databases.postgresql.performance: > > > >>postgres@countup.de wrote: >> >> >[sNip] > > >>>Sorry I m a little bit confused about the persistent thing!! >>>Is it smart to use persistent connections at all if i expect 100K >>>Users to hit the script in an hour and the script calls up to 10-15 pg >>>functions? >>>I have at the mom one function but the server needs 500 ms, its a >>>little bit too much i think, and it crashed when i had 20K users >>> >>> >>Use the persistent connection but make sure the parameters in >>postgresql.conf match up with the Apache config. The specific settings >>are MaxClients in httpd.conf and max_connections in postgresql.conf. >>Make sure that max_connections is at least as big as MaxClients for >>every database that your PHP scripts connect to. >> >> > > Do you happen to have (or know where to get) some sample configuration >files for Apache 2 and PostgreSQL for this? The documentation I've found >so far is pretty sparse, and sample files would be very helpful. > > > > Beware that persistent connections in PHP behave a little differently than you would think. The connections stays open between an apache process and postgres. So each process has its own connection and you may not hit the same process on each request to the apache server. Temporary tables are not dropped automatically between refreshes on persistent connections. An example of this is to enable persistent connections and execute "CREATE TEMPORARY TABLE foo ( id INTEGER );" $conn = pg_pconnect( ... ); if (!$result = pg_query($conn, "CREATE TEMPORARY TABLE tmp_foo ( id INTEGER );")) { echo pg_result_error($result) ; } else { echo "created ok!"; } After a couple of refreshes you will get an error that states the table already exists. This was a pain to learn, especially while I was doing these operations inside of transactions. On most of my servers the connect time for postgresql was 6ms or less, so I disabled persistent connections altogether so that I could be assured that temporary tables and all php launched postgresql sessions were properly reset. As far as I know, there is no way to reset the sesssion ( cleaning up temporary tables, etc ) automatically with an SQL statement without closing the connection From pgsql-performance-owner@postgresql.org Tue Apr 20 22:51:22 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 949EAD1B88E for ; Tue, 20 Apr 2004 22:48:25 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 15062-05 for ; Tue, 20 Apr 2004 22:48:22 -0300 (ADT) Received: from trade-india.com (ns5.trade-india.com [66.234.10.13]) by svr1.postgresql.org (Postfix) with SMTP id 4857BD1D9C2 for ; Tue, 20 Apr 2004 13:17:54 -0300 (ADT) Received: (qmail 27939 invoked from network); 20 Apr 2004 12:48:42 -0000 Received: from unknown (HELO trade-india.com) (203.145.130.142) by ns5.trade-india.com with SMTP; 20 Apr 2004 12:48:42 -0000 Message-ID: <40854D38.6040403@trade-india.com> Date: Tue, 20 Apr 2004 21:48:00 +0530 From: Rajesh Kumar Mallah User-Agent: Mozilla Thunderbird 0.5 (X11/20040208) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Mark Stosberg Cc: pgsql-performance@postgresql.org Subject: Re: seeking consultant for high performance, complex searching 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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/324 X-Sequence-Number: 6624 Have you checked Tsearch2 http://www.sai.msu.su/~megera/postgres/gist/tsearch/V2/ is the most feature rich Full text Search system available for postgresql. We are also using the same system in the revamped version of our website. Regds Mallah. Mark Stosberg wrote: >Hello, > >I work for Summersault, LLC. We've been using Postgres since the days of >Postgres 6.5. We're focused on building database-driven websites using Perl and >Postgres. We are currently seeking help developing a search system that needs >to perform complex queries with high performance. Although we have strong >skills in Perl and Postgres, we are new to the arena of complex, >high-performance search systems. > >We are seeking to hire a consultant to help this as part of the re-vamp >of the 1-800-Save-A-Pet.com website. > >1-800-Save-A-Pet.com is a not-for-profit organization whose website >finds homes for homeless pets, promoting pet adoption and saving >thousands of animal lives. Summersault, LLC is a website development >firm focused on creating highly customized database driven websites. > >The ideal consultant has expert experience with the PostgreSQL RDBMS and >the Perl programming language, and is intimately familiar with the >architecture and implementation of complex database queries for >high-traffic web applications. The consultant should also have a strong >background in creating solutions complementary to this work, e.g. >assessing hardware requirements, designing a hosting and network >infrastructure, and optimizing the algorithm based on real-world >feedback. The consultant will work with Summersault developers as a >part of a larger application development process. > >Interested persons or organizations should contact Chris Hardie of >Summersault, LLC at chris@summersault.com for more information. > >Thanks! > > Mark > > > From pgsql-performance-owner@postgresql.org Tue Apr 20 22:50:12 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8C02BD1E13D for ; Tue, 20 Apr 2004 22:49:46 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 14739-08 for ; Tue, 20 Apr 2004 22:49:44 -0300 (ADT) Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251]) by svr1.postgresql.org (Postfix) with ESMTP id 824C0D1F388 for ; Tue, 20 Apr 2004 13:48:22 -0300 (ADT) Received: (from pgman@localhost) by candle.pha.pa.us (8.11.6/8.11.6) id i3KGmEV27394; Tue, 20 Apr 2004 12:48:14 -0400 (EDT) From: Bruce Momjian Message-Id: <200404201648.i3KGmEV27394@candle.pha.pa.us> Subject: Re: Wierd context-switching issue on Xeon In-Reply-To: <408533AD.40904@aeccom.com> To: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= Date: Tue, 20 Apr 2004 12:48:14 -0400 (EDT) Cc: Tom Lane , Josh Berkus , pgsql-performance@postgresql.org, Neil Conway X-Mailer: ELM [version 2.4ME+ PL108 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=UNKNOWN-8BIT X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/321 X-Sequence-Number: 6621 Dirk Lutzeb�ck wrote: > Dirk Lutzebaeck wrote: > > > c) Dual XEON DP, non-bigmem, HT on, E7500 Intel chipset (Supermicro) > > > > performs well and I could not observe context switch peaks here (one > > user active), almost no extra semop calls > > Did Tom's test here: with 2 processes I'll reach 200k+ CS with peaks to > 300k CS. Bummer.. Josh, I don't think you can bash the ServerWorks > chipset here nor bigmem. Dave Cramer reproduced the problem on my SuperMicro dual Xeon on BSD/OS. -- 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 Tue Apr 20 22:50:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BB0E2D1B895 for ; Tue, 20 Apr 2004 22:49:30 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 16846-02 for ; Tue, 20 Apr 2004 22:49:25 -0300 (ADT) Received: from vt-pe2550-001.VANTAGE.vantage.com (unknown [64.80.203.242]) by svr1.postgresql.org (Postfix) with ESMTP id 7CFA8D1F398 for ; Tue, 20 Apr 2004 13:48:59 -0300 (ADT) X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0 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: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 12:48:58 -0400 Message-ID: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [PERFORM] Wierd context-switching issue on Xeon Thread-Index: AcQm7otpHAx43BozTZu3cHFKf1c9lgAB8rAg From: "Anjan Dave" To: =?iso-8859-1?Q?Dirk_Lutzeb=E4ck?= , "Tom Lane" , "Josh Berkus" Cc: , "Neil Conway" X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/322 X-Sequence-Number: 6622 If this helps -=20 Quad 2.0GHz XEON with highest load we have seen on the applications, DB per= forming great -=20 procs memory swap io system = cpu r b w swpd free buff cache si so bi bo in cs us sy= id 1 0 0 1616 351820 66144 10813704 0 0 2 0 1 1 0 = 2 7 3 0 0 1616 349712 66144 10813736 0 0 8 1634 1362 4650 4 = 2 95 0 0 0 1616 347768 66144 10814120 0 0 188 1218 1158 4203 5 = 1 93 0 0 1 1616 346596 66164 10814184 0 0 8 1972 1394 4773 4 = 1 94 2 0 1 1616 345424 66164 10814272 0 0 20 1392 1184 4197 4 = 2 94 Around 4k CS/sec Chipset is Intel ServerWorks GC-HE. Linux Kernel 2.4.20-28.9bigmem #1 SMP Thanks, Anjan -----Original Message----- From: Dirk Lutzeb=E4ck [mailto:lutzeb@aeccom.com]=20 Sent: Tuesday, April 20, 2004 10:29 AM To: Tom Lane; Josh Berkus Cc: pgsql-performance@postgreSQL.org; Neil Conway Subject: Re: [PERFORM] Wierd context-switching issue on Xeon Dirk Lutzebaeck wrote: > c) Dual XEON DP, non-bigmem, HT on, E7500 Intel chipset (Supermicro) > > performs well and I could not observe context switch peaks here (one=20 > user active), almost no extra semop calls Did Tom's test here: with 2 processes I'll reach 200k+ CS with peaks to=20 300k CS. Bummer.. Josh, I don't think you can bash the ServerWorks=20 chipset here nor bigmem. Dirk ---------------------------(end of broadcast)--------------------------- TIP 6: Have you searched our list archives? http://archives.postgresql.org From pgsql-performance-owner@postgresql.org Tue Apr 20 22:31:22 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 418F5D1B4E1 for ; Tue, 20 Apr 2004 22:31:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 05233-08 for ; Tue, 20 Apr 2004 22:31:12 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id B02D7D1F46F for ; Tue, 20 Apr 2004 14:00:11 -0300 (ADT) Received: from [63.195.55.98] (HELO spooky) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4907935; Tue, 20 Apr 2004 10:01:35 -0700 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Organization: Aglio Database Solutions To: "Anjan Dave" , Dirk =?iso-8859-1?q?Lutzeb=E4ck?= , "Tom Lane" Subject: Re: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 09:59:52 -0700 User-Agent: KMail/1.4.3 Cc: , "Neil Conway" References: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com> In-Reply-To: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Message-Id: <200404200959.52813.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/320 X-Sequence-Number: 6620 Anjan, > Quad 2.0GHz XEON with highest load we have seen on the applications, DB > performing great - Can you run Tom's test? It takes a particular pattern of data access to reproduce the issue. -- Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Tue Apr 20 22:14:29 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id F108AD1EA95 for ; Tue, 20 Apr 2004 22:14:19 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 99387-07 for ; Tue, 20 Apr 2004 22:14:18 -0300 (ADT) Received: from mwinf0803.wanadoo.fr (smtp8.wanadoo.fr [193.252.22.23]) by svr1.postgresql.org (Postfix) with ESMTP id 6A13BD1F5AC for ; Tue, 20 Apr 2004 14:10:53 -0300 (ADT) Received: from [192.168.0.2] (AMontsouris-108-1-29-252.w81-53.abo.wanadoo.fr [81.53.158.252]) by mwinf0803.wanadoo.fr (SMTP Server) with ESMTP id ED94E18000EB for ; Tue, 20 Apr 2004 19:10:52 +0200 (CEST) Mime-Version: 1.0 (Apple Message framework v613) In-Reply-To: References: Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= Subject: Re: 225 times slower Date: Tue, 20 Apr 2004 19:10:50 +0200 To: pgsql-performance@postgresql.org X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS X-Spam-Level: X-Archive-Number: 200404/317 X-Sequence-Number: 6617 Hi, I apologize for the mistake. So, I dump the database, I reload it then VACUUM ANALYZE. For each statement: I then quit postgres, start it, execute one=20=20 command, then quit. Le 14 avr. 04, =E0 14:39, Pailloncy Jean-G=E9rard a =E9crit : dps=3D# explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D764518963 AND status IN (200,304,206) ORDER BY rec_id LIMIT 1; QUERY PLAN ------------------------------------------------------------------------=20 -------------------------------------------------------- Limit (cost=3D169.79..169.79 rows=3D1 width=3D4) (actual=20=20 time=3D502.397..502.398 rows=3D1 loops=3D1) -> Sort (cost=3D169.79..169.86 rows=3D30 width=3D4) (actual=20=20 time=3D502.393..502.393 rows=3D1 loops=3D1) Sort Key: rec_id -> Index Scan using url_crc on url (cost=3D0.00..169.05=20=20 rows=3D30 width=3D4) (actual time=3D43.545..490.895 rows=3D56 loops=3D1) Index Cond: (crc32 =3D 764518963) Filter: ((crc32 <> 0) AND ((status =3D 200) OR (status =3D= =20=20 304) OR (status =3D 206))) Total runtime: 502.520 ms (7 rows) dps=3D# \q dps=3D# explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D764518963 AND status IN (200,304,206) ORDER BY crc32,rec_id LIMIT= =20=20 1; QUERY PLAN ------------------------------------------------------------------------=20 ----------------------------------------------------- Limit (cost=3D169.79..169.79 rows=3D1 width=3D8) (actual time=3D5.893..5= .894=20=20 rows=3D1 loops=3D1) -> Sort (cost=3D169.79..169.86 rows=3D30 width=3D8) (actual=20=20 time=3D5.889..5.889 rows=3D1 loops=3D1) Sort Key: crc32, rec_id -> Index Scan using url_crc on url (cost=3D0.00..169.05=20=20 rows=3D30 width=3D8) (actual time=3D0.445..5.430 rows=3D56 loops=3D1) Index Cond: (crc32 =3D 764518963) Filter: ((crc32 <> 0) AND ((status =3D 200) OR (status =3D= =20=20 304) OR (status =3D 206))) Total runtime: 6.020 ms (7 rows) dps=3D# \q dps=3D# explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D419903683 AND status IN (200,304,206) ORDER BY rec_id LIMIT 1; QUERY PLAN ------------------------------------------------------------------------=20 ---------------------------------------------------------- Limit (cost=3D0.00..27.95 rows=3D1 width=3D4) (actual=20=20 time=3D11021.875..11021.876 rows=3D1 loops=3D1) -> Index Scan using url_pkey on url (cost=3D0.00..11625.49 rows=3D416= =20=20 width=3D4) (actual time=3D11021.868..11021.868 rows=3D1 loops=3D1) Filter: ((crc32 <> 0) AND (crc32 =3D 419903683) AND ((status =3D= =20=20 200) OR (status =3D 304) OR (status =3D 206))) Total runtime: 11021.986 ms (4 rows) dps=3D# \q dps=3D# explain analyze SELECT rec_id FROM url WHERE crc32!=3D0 AND=20=20 crc32=3D419903683 AND status IN (200,304,206) ORDER BY crc32,rec_id LIMIT= =20=20 1; QUERY PLAN ------------------------------------------------------------------------=20 --------------------------------------------------------- Limit (cost=3D2000.41..2000.41 rows=3D1 width=3D8) (actual=20=20 time=3D48.503..48.504 rows=3D1 loops=3D1) -> Sort (cost=3D2000.41..2001.45 rows=3D416 width=3D8) (actual=20=20 time=3D48.499..48.499 rows=3D1 loops=3D1) Sort Key: crc32, rec_id -> Index Scan using url_crc on url (cost=3D0.00..1982.31=20=20 rows=3D416 width=3D8) (actual time=3D4.848..45.452 rows=3D796 loops=3D1) Index Cond: (crc32 =3D 419903683) Filter: ((crc32 <> 0) AND ((status =3D 200) OR (status =3D= =20=20 304) OR (status =3D 206))) Total runtime: 48.656 ms (7 rows) dps=3D# \q So, with all fresh data, everything rebuild from scratch, on a backend=20= =20 that will done one and only one query, the results is strange. Why adding an ORDER BY clause on a column with one value speed up the=20=20 stuff 502ms to 6ms ? Why when crc32=3D419903683, which is one of the most often used value in=20= =20 the table, the query planner chose a plan so bad (225 times slower) ? Cordialement, Jean-G=E9rard Pailloncy From pgsql-performance-owner@postgresql.org Tue Apr 20 22:14:49 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id A1504D1D1D5 for ; Tue, 20 Apr 2004 22:14:29 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 99419-07 for ; Tue, 20 Apr 2004 22:14:25 -0300 (ADT) Received: from mx1.neopolitan.us (mx1.neopolitan.us [65.87.16.224]) by svr1.postgresql.org (Postfix) with ESMTP id 82E62D1F66C for ; Tue, 20 Apr 2004 14:17:22 -0300 (ADT) Received: from [65.87.16.98] (HELO [10.0.0.210]) by mx1.neopolitan.us (CommuniGate Pro SMTP 4.1.8) with ESMTP-TLS id 3996614; Tue, 20 Apr 2004 10:17:22 -0700 Subject: Re: Wierd context-switching issue on Xeon From: "J. Andrew Rogers" To: Tom Lane Cc: pgsql-performance@postgresql.org In-Reply-To: <1407.1082419316@sss.pgh.pa.us> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> Content-Type: text/plain Organization: Message-Id: <1082481442.997.4.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.2 (1.2.2-5) Date: 20 Apr 2004 10:17:22 -0700 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= X-Spam-Level: X-Archive-Number: 200404/318 X-Sequence-Number: 6618 I verified problem on a Dual Opteron server. I temporarily killed the normal load, so the server was largely idle when the test was run. Hardware: 2x Opteron 242 Rioworks HDAMA server board 4Gb RAM OS Kernel: RedHat9 + XFS 1 proc: 10-15 cs/sec 2 proc: 400,000-420,000 cs/sec j. andrew rogers From pgsql-performance-owner@postgresql.org Tue Apr 20 21:42:04 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B27B4D1E8CB for ; Tue, 20 Apr 2004 21:41:42 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 87736-05 for ; Tue, 20 Apr 2004 21:41:37 -0300 (ADT) Received: from wolff.to (wolff.to [66.93.249.74]) by svr1.postgresql.org (Postfix) with SMTP id E12FFD1EEBE for ; Tue, 20 Apr 2004 14:52:33 -0300 (ADT) Received: (qmail 11338 invoked by uid 500); 20 Apr 2004 17:56:32 -0000 Date: Tue, 20 Apr 2004 12:56:32 -0500 From: Bruno Wolff III To: Chris Hoover Cc: pgsql-performance@postgresql.org Subject: Re: Use of subquery causes seq scan??? Message-ID: <20040420175632.GB11203@wolff.to> Mail-Followup-To: Chris Hoover , pgsql-performance@postgresql.org References: <200404151248.16067.jeremy.guthrie@berbee.com> <200404201020.05699.revoohc@sermonaudio.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200404201020.05699.revoohc@sermonaudio.com> User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/316 X-Sequence-Number: 6616 Please don't reply to messages to start new threads. On Tue, Apr 20, 2004 at 10:20:05 -0400, Chris Hoover wrote: > I need some help. I have a query that refuses to use the provided index and > is always sequentially scanning causing me large performance headaches. Here > is the basic situation: > > Table A: > inv_num int > type char > . > . > . > pkey (inv_num, type) > indx(inv_num) > > Table B (has the same primary key) > > Select * > from table a > where inv_num in (select inv_num from table b where ....) > > Doing this causes sequential scans of both tables. If I do a set > enable_seqscan to false before the query, I get an index scan of table b but > still seq scan table a. > > Is there anyway to force table a to use this index (or another) and not > sequentially scan the table? > > I'm running 7.3.4 on RedHat EL 2.1. IN was slow in 7.3.x and before. The query will probably run much better as is in 7.4 and above. In 7.3 you want to rewrite it as a join or using EXISTS. From pgsql-performance-owner@postgresql.org Tue Apr 20 21:41:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id BBAD0D1BAFD for ; Tue, 20 Apr 2004 21:41:36 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 88358-02 for ; Tue, 20 Apr 2004 21:41:30 -0300 (ADT) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id D698FD1EF67 for ; Tue, 20 Apr 2004 14:57:59 -0300 (ADT) Received: from [64.81.245.111] (HELO 192.168.1.102) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8) with ESMTP id 4908282; Tue, 20 Apr 2004 10:59:24 -0700 From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: Dirk =?iso-8859-1?q?Lutzeb=E4ck?= , Tom Lane Subject: Re: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 10:58:18 -0700 User-Agent: KMail/1.5.4 Cc: pgsql-performance@postgreSQL.org References: <200311251419.36771.josh@agliodbs.com> <40844223.9070200@aeccom.com> <408533AD.40904@aeccom.com> In-Reply-To: <408533AD.40904@aeccom.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200404201058.18840.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0 tests=RCVD_NUMERIC_HELO X-Spam-Level: X-Archive-Number: 200404/315 X-Sequence-Number: 6615 Dirk, Tom, OK, off IRC, I have the following reports: Linux 2.4.21 or 2.4.20 on dual Pentium III : problem verified Linux 2.4.21 or 2.4.20 on dual Penitum II : problem cannot be reproduced Solaris 2.6 on 6 cpu e4500 (using 8 processes) : problem not reproduced -- -Josh Berkus Aglio Database Solutions San Francisco From pgsql-performance-owner@postgresql.org Tue Apr 20 20:35:06 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 4B4B3D1F11A for ; Tue, 20 Apr 2004 20:34:56 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55581-08 for ; Tue, 20 Apr 2004 20:34:52 -0300 (ADT) Received: from www.postgresql.com (www.postgresql.com [200.46.204.209]) by svr1.postgresql.org (Postfix) with ESMTP id E6FD1D20050 for ; Tue, 20 Apr 2004 19:00:32 -0300 (ADT) Received: from flake.decibel.org (flake.decibel.org [66.143.173.58]) by www.postgresql.com (Postfix) with SMTP id A51AFCF6658 for ; Tue, 20 Apr 2004 15:46:15 -0300 (ADT) Received: (qmail 55520 invoked by uid 1001); 20 Apr 2004 18:46:10 -0000 Date: Tue, 20 Apr 2004 13:46:10 -0500 From: "Jim C. Nasby" To: pgsql-performance@postgresql.org Subject: Re: Horribly slow hash join Message-ID: <20040420184610.GY87362@nasby.net> References: <20040416154502.GF87362@nasby.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20040416154502.GF87362@nasby.net> X-Operating-System: FreeBSD 4.9-RELEASE-p3 i386 X-Distributed: Join the Effort! http://www.distributed.net User-Agent: Mutt/1.5.6i X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/312 X-Sequence-Number: 6612 Dammit, I somehow deleted a bunch of replies to this. Did a TODO ever come out of this? -- Jim C. Nasby, Database Consultant jim@nasby.net Member: Triangle Fraternity, Sports Car Club of America Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?" From pgsql-performance-owner@postgresql.org Tue Apr 20 20:34:05 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5528ED1EAF2 for ; Tue, 20 Apr 2004 20:34:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 52979-07 for ; Tue, 20 Apr 2004 20:34:00 -0300 (ADT) Received: from tht.net (vista.tht.net [216.126.88.2]) by svr1.postgresql.org (Postfix) with ESMTP id 38475D1F976 for ; Tue, 20 Apr 2004 16:50:22 -0300 (ADT) Received: from [134.22.69.199] (dyn-69-199.tor.dsl.tht.net [134.22.69.199]) by tht.net (Postfix) with ESMTP id A3AC976AD5; Tue, 20 Apr 2004 15:50:15 -0400 (EDT) Subject: Re: Wierd context-switching issue on Xeon From: Rod Taylor To: Tom Lane Cc: Josh Berkus , Joe Conway , Scott Marlowe , Bruce Momjian , lutzeb@aeccom.com, Postgresql Performance , Neil Conway In-Reply-To: <1931.1082422389@sss.pgh.pa.us> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us> Content-Type: text/plain Message-Id: <1082490529.80320.71.camel@jester> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.4.5 Date: Tue, 20 Apr 2004 15:48:50 -0400 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= X-Spam-Level: X-Archive-Number: 200404/309 X-Sequence-Number: 6609 > It would be interesting to see results with non-Linux kernels, too. Dual Celeron 500Mhz (Abit BP6 mobo) - client & server on same machine 2 processes FreeBSD (5.2.1): 1800cs 3 processes FreeBSD: 14000cs 4 processes FreeBSD: 14500cs 2 processes Linux (2.4.18 kernel): 52000cs 3 processes Linux: 10000cs 4 processes Linux: 20000cs From pgsql-performance-owner@postgresql.org Tue Apr 20 20:18:32 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 8F244D1EC80 for ; Tue, 20 Apr 2004 20:18:13 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 46256-06 for ; Tue, 20 Apr 2004 20:18:07 -0300 (ADT) Received: from monsoon.he.net (monsoon.he.net [64.62.221.2]) by svr1.postgresql.org (Postfix) with SMTP id B945ED1F9FA for ; Tue, 20 Apr 2004 17:02:44 -0300 (ADT) Received: from [10.244.45.7] ([216.113.168.128]) by monsoon.he.net for ; Tue, 20 Apr 2004 13:02:39 -0700 In-Reply-To: <1931.1082422389@sss.pgh.pa.us> References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v613) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit Cc: pgsql-performance@postgresql.org From: Paul Tuckfield Subject: Re: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 13:02:43 -0700 To: Tom Lane X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/308 X-Sequence-Number: 6608 I tried to test how this is related to cache coherency, by forcing affinity of the two test_run.sql processes to the two cores (pipelines? threads) of a single hyperthreaded xeon processor in an smp xeon box. When the processes are allowed to run on distinct chips in the smp box, the CS storm happens. When they are "bound" to the two cores of a single hyperthreaded Xeon in the smp box, the CS storm *does* happen. I used the taskset command: taskset 01 -p taskset 01 -p I guess that 0 and 1 are the two cores (pipelines? hyper-threads?) on the first Xeon processor in the box. I did this on RedHat Fedora core1 on an intel motherboard (I'll get the part no if it matters) during storms : 300k CS/sec, 75% idle (on a dual xeon (four core)) machine (suggesting serializing/sleeping processes) no storm: 50k CS/sec, 50% idle (suggesting 2 cpu bound processes) Maybe there's a "hot block" that is bouncing back and forth between caches? or maybe the page holding semaphores? On Apr 19, 2004, at 5:53 PM, Tom Lane wrote: > I wrote: >> Here is a test case. > > Hmmm ... I've been able to reproduce the CS storm on a dual Athlon, > which seems to pretty much let the Xeon per se off the hook. Anybody > got a multiple Opteron to try? Totally non-Intel CPUs? > > It would be interesting to see results with non-Linux kernels, too. > > regards, tom lane > > ---------------------------(end of > broadcast)--------------------------- > TIP 4: Don't 'kill -9' the postmaster > From pgsql-performance-owner@postgresql.org Tue Apr 20 20:34:57 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 0677DD1F12B for ; Tue, 20 Apr 2004 20:34:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 56866-05 for ; Tue, 20 Apr 2004 20:34:43 -0300 (ADT) Received: from vt-pe2550-001.VANTAGE.vantage.com (unknown [64.80.203.242]) by svr1.postgresql.org (Postfix) with ESMTP id DA3B6D1FB84 for ; Tue, 20 Apr 2004 17:20:35 -0300 (ADT) X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C42714.F0CF1A43" Subject: Moving postgres to FC disks Date: Tue, 20 Apr 2004 16:20:36 -0400 Message-ID: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814D@vt-pe2550-001.vantage.vantage.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Moving postgres to FC disks Thread-Index: AcQnFPDFxLhs3dQcQhCbbbl4TZlISg== From: "Anjan Dave" To: X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=HTML_60_70, HTML_MESSAGE X-Spam-Level: X-Archive-Number: 200404/311 X-Sequence-Number: 6611 This is a multi-part message in MIME format. ------_=_NextPart_001_01C42714.F0CF1A43 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable I am planning to move the pg databases from the internal RAID to external Fiber Channel over SAN. =20 Question is -=20 =20 -With the db size being as big as, say, 30+GB, how do I move it on the new logical drive? (stop postgresql, and simply move it over somehow and make a link?) -Currently, the internal RAID volume is ext3 filesystem. Any recommendations for the filesystem on the new FC volume? Rieserfs? =20 DBs are 7.4.1(RH9), and 7.2.3 (RH8). =20 =20 Appreciate any pointers. =20 Thanks, Anjan ------_=_NextPart_001_01C42714.F0CF1A43 Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

I am planning to move the pg databases from the internal RAID to external Fiber Channel over SAN.

 

Question is –

 

-With the db size being as big as, say, 30+GB, how do I = move it on the new logical drive? (stop postgresql, and simply move it over some= how and make a link?)

-Currently, the internal RAID volume is ext3 filesystem.= Any recommendations for the filesystem on the new FC volume? Rieserfs?

 

DBs are 7.4.1(RH9), and 7.2.3 (RH8).

 

 

Appreciate any pointers.

 

Thanks,
Anjan

------_=_NextPart_001_01C42714.F0CF1A43-- From pgsql-performance-owner@postgresql.org Tue Apr 20 20:34:50 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 64BB8D1F11A for ; Tue, 20 Apr 2004 20:34:41 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 55788-07 for ; Tue, 20 Apr 2004 20:34:39 -0300 (ADT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id CEB66D1FB10 for ; Tue, 20 Apr 2004 18:58:29 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3KLwV2M022628; Tue, 20 Apr 2004 17:58:31 -0400 (EDT) To: "Chris Hoover" Cc: pgsql-performance@postgresql.org Subject: Re: Use of subquery causes seq scan??? In-reply-to: <200404201020.05699.revoohc@sermonaudio.com> References: <200404151248.16067.jeremy.guthrie@berbee.com> <200404201020.05699.revoohc@sermonaudio.com> Comments: In-reply-to "Chris Hoover" message dated "Tue, 20 Apr 2004 10:20:05 -0400" Date: Tue, 20 Apr 2004 17:58:31 -0400 Message-ID: <22627.1082498311@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/310 X-Sequence-Number: 6610 "Chris Hoover" writes: > Select * > from table a > where inv_num in (select inv_num from table b where ....) > I'm running 7.3.4 on RedHat EL 2.1. IN (SELECT) constructs pretty well suck in PG releases before 7.4. Update, or consult the FAQ about rewriting into an EXISTS form. regards, tom lane From pgsql-performance-owner@postgresql.org Mon Apr 26 14:46:18 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 5DFF3D1F864 for ; Tue, 20 Apr 2004 19:48:03 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 31860-01 for ; Tue, 20 Apr 2004 19:48:02 -0300 (ADT) Received: from mail2.fbab.net (spectre.fbab.net [212.214.165.139]) by svr1.postgresql.org (Postfix) with SMTP id BD532D1F879 for ; Tue, 20 Apr 2004 19:47:59 -0300 (ADT) Received: (qmail 13066 invoked by uid 136); 20 Apr 2004 22:48:00 -0000 Received: from mag@fbab.net by mail2.fbab.net by uid 133 with qmail-scanner-1.20rc1 (avp: 4.0.3.0. Clear:RC:0:. Processed in 2.472445 secs); 20 Apr 2004 22:48:00 -0000 Received: from unknown (HELO fbab.net) (magpool1@212.214.165.129) by mail2.fbab.net with SMTP; 20 Apr 2004 22:47:57 -0000 Message-ID: <4085A895.6080403@fbab.net> Date: Wed, 21 Apr 2004 00:47:49 +0200 From: "Magnus Naeslund(t)" User-Agent: Mozilla Thunderbird 0.5 (X11/20040321) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Tom Lane Cc: josh@agliodbs.com, Joe Conway , "scott.marlowe" , Bruce Momjian , lutzeb@aeccom.com, pgsql-performance@postgresql.org, Neil Conway Subject: Re: Wierd context-switching issue on Xeon References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us> In-Reply-To: <1931.1082422389@sss.pgh.pa.us> Content-Type: multipart/mixed; boundary="------------010806020208090107020602" X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/438 X-Sequence-Number: 6738 This is a multi-part message in MIME format. --------------010806020208090107020602 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Tom Lane wrote: > > Hmmm ... I've been able to reproduce the CS storm on a dual Athlon, > which seems to pretty much let the Xeon per se off the hook. Anybody > got a multiple Opteron to try? Totally non-Intel CPUs? > > It would be interesting to see results with non-Linux kernels, too. > > regards, tom lane I also tested on an dual Athlon MP Tyan Thunder motherboard (2xMP2800+, 2.5GB memory), and got the same high numbers. I then ran with kernel 2.6.5, it lowered them a little, but it's still some ping pong effect here. I wonder if this is some effect of the scheduler, maybe the shed frequency alone (100HZ vs 1000HZ). It would be interesting to see what a locking implementation ala FUTEX style would give on an 2.6 kernel, as i understood it that would work cross process with some work. The first file attached is kernel 2.4 running one process then starting up the other one. Same with second file, but with kernel 2.6... Regards Magnus --------------010806020208090107020602 Content-Type: text/plain; name="vmstat_1-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="vmstat_1-1" procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 1 0 0 1828408 27852 528852 0 0 0 0 317 557 50 0 50 0 1 0 0 1828408 27852 528852 0 0 0 0 293 491 50 0 49 0 1 0 0 1828400 27860 528852 0 0 0 16 399 709 50 0 50 0 1 0 0 1828400 27860 528852 0 0 0 0 350 593 50 0 49 0 2 0 0 1828400 27860 528852 0 0 0 0 349 608 50 0 50 0 1 0 0 1828400 27860 528852 0 0 0 0 109 412 50 0 50 0 1 0 0 1828400 27860 528852 0 0 0 0 101 92 50 0 50 0 1 0 0 1828392 27868 528852 0 0 0 16 104 96 50 0 50 0 1 0 0 1828392 27868 528852 0 0 0 0 101 103 50 0 50 0 2 0 0 1827408 27892 528852 0 0 8 48 113 61197 45 9 46 0 2 0 0 1827408 27892 528852 0 0 0 0 101 167237 41 27 32 0 procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 4 0 0 1827408 27892 528852 0 0 0 0 101 166145 39 25 36 0 2 0 0 1827400 27900 528852 0 0 0 48 105 149406 42 19 40 0 3 0 0 1827400 27900 528852 0 0 0 0 101 157559 43 26 32 0 2 0 0 1827400 27900 528852 0 0 0 0 101 163813 46 24 30 0 2 0 0 1827400 27900 528852 0 0 0 0 101 156872 44 26 30 0 2 0 0 1827400 27900 528852 0 0 0 0 103 160722 45 28 28 0 2 0 0 1827392 27908 528852 0 0 0 16 104 158644 41 23 37 0 3 0 0 1827392 27908 528852 0 0 0 0 101 157534 42 25 33 0 2 0 0 1827392 27908 528852 0 0 0 0 101 160007 37 28 35 0 3 0 0 1827392 27908 528852 0 0 0 0 101 161852 45 24 31 0 3 0 0 1827392 27908 528852 0 0 0 0 101 161616 42 25 33 0 2 0 0 1827392 27916 528852 0 0 0 68 114 152144 44 25 31 0 2 0 0 1827384 27916 528852 0 0 0 0 101 156485 35 28 37 0 --------------010806020208090107020602 Content-Type: text/plain; name="vmstat_1-2" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="vmstat_1-2" procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 1 0 0 2436044 8844 90028 0 0 0 16 1010 235 50 0 50 0 1 0 0 2436108 8844 90028 0 0 0 0 1024 404 50 0 50 0 1 0 0 2436108 8844 90028 0 0 0 0 1008 199 50 0 50 0 1 0 0 2436108 8844 90028 0 0 0 0 1017 272 50 0 50 0 1 0 0 2436108 8844 90028 0 0 0 0 1013 253 50 0 50 0 1 1 0 2436108 8852 90020 0 0 0 16 1019 282 51 0 49 1 2 0 0 2435068 8852 90020 0 0 0 0 1005 23929 45 4 50 0 2 0 0 2435068 8852 90020 0 0 0 20 1008 95501 33 14 53 0 procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 3 0 0 2435068 8852 90020 0 0 0 0 1002 103940 35 15 50 0 0 0 0 2435068 8852 90020 0 0 0 0 1003 104343 32 16 51 0 2 0 0 2435068 8860 90080 0 0 0 52 1006 102477 34 16 51 1 2 0 0 2435068 8860 90080 0 0 0 0 1002 92809 31 14 54 0 2 0 0 2435068 8860 90080 0 0 0 0 1002 100498 37 14 49 0 1 0 0 2435068 8860 90080 0 0 0 0 1002 108130 35 16 49 0 0 0 0 2435068 8860 90080 0 0 0 0 1002 94045 33 14 54 0 0 0 0 2435004 8868 90072 0 0 0 16 1005 104380 34 15 52 0 2 0 0 2435004 8868 90072 0 0 0 0 1002 100696 36 14 50 0 2 0 0 2435068 8868 90072 0 0 0 0 1002 98289 31 14 54 0 0 0 0 2435068 8868 90072 0 0 0 0 1002 97287 31 14 55 0 0 0 0 2435068 8868 90072 0 0 0 0 1002 92787 34 14 53 0 0 0 0 2435068 8876 90064 0 0 0 16 1005 98568 32 16 52 1 2 0 0 2435068 8876 90064 0 0 0 0 1003 107104 37 16 47 0 --------------010806020208090107020602-- From pgsql-performance-owner@postgresql.org Tue Apr 20 21:24:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id B2FE2D1F001 for ; Tue, 20 Apr 2004 21:24:53 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 81108-04 for ; Tue, 20 Apr 2004 21:24:53 -0300 (ADT) Received: from hosting.commandprompt.com (128.commandprompt.com [207.173.200.128]) by svr1.postgresql.org (Postfix) with ESMTP id 4B8E2D1EFB7 for ; Tue, 20 Apr 2004 21:24:50 -0300 (ADT) Received: from commandprompt.com (clbb-248.saw.net [64.146.135.248]) (authenticated) by hosting.commandprompt.com (8.11.6/8.11.6) with ESMTP id i3L0Oou25853; Tue, 20 Apr 2004 17:24:51 -0700 Message-ID: <4085BFF0.4050604@commandprompt.com> Date: Tue, 20 Apr 2004 17:27:28 -0700 From: "Joshua D. Drake" Organization: Command Prompt, Inc. User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Anjan Dave Cc: pgsql-performance@postgresql.org Subject: Re: Moving postgres to FC disks References: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814D@vt-pe2550-001.vantage.vantage.com> In-Reply-To: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814D@vt-pe2550-001.vantage.vantage.com> Content-Type: multipart/alternative; boundary="------------030009030401050901060103" X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.7 tagged_above=0.0 required=5.0 tests=HTML_50_60, HTML_MESSAGE, HTML_TITLE_EMPTY X-Spam-Level: X-Archive-Number: 200404/313 X-Sequence-Number: 6613 This is a multi-part message in MIME format. --------------030009030401050901060103 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit > -With the db size being as big as, say, 30+GB, how do I move it on the > new logical drive? (stop postgresql, and simply move it over somehow > and make a link?) > I would stop the database, move the data directory to the new volume using rsync then start up postgresql pointed at the new data directory. Providing everything is working correctly you can then remove the old data directory. > -Currently, the internal RAID volume is ext3 filesystem. Any > recommendations for the filesystem on the new FC volume? Rieserfs? > > > XFS > DBs are 7.4.1(RH9), and 7.2.3 (RH8). > > > > > > Appreciate any pointers. > > > > Thanks, > Anjan > -- Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC and S/JDBC Postgresql support, programming shared hosting and dedicated hosting. +1-503-667-4564 - jd@commandprompt.com - http://www.commandprompt.com PostgreSQL Replicator -- production quality replication for PostgreSQL --------------030009030401050901060103 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit

-With the db size being as big as, say, 30+GB, how do I move it on the new logical drive? (stop postgresql, and simply move it over somehow and make a link?)

I would stop the database, move the data directory to the new volume using rsync then start up postgresql pointed at the new data directory.
Providing everything is working correctly you can then remove the old data directory.

-Currently, the internal RAID volume is ext3 filesystem. Any recommendations for the filesystem on the new FC volume? Rieserfs?

 

XFS

DBs are 7.4.1(RH9), and 7.2.3 (RH8).

 

 

Appreciate any pointers.

 

Thanks,
Anjan



-- 
Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC and S/JDBC
Postgresql support, programming shared hosting and dedicated hosting.
+1-503-667-4564 - jd@commandprompt.com - http://www.commandprompt.com
PostgreSQL Replicator -- production quality replication for PostgreSQL
--------------030009030401050901060103-- From pgsql-performance-owner@postgresql.org Tue Apr 20 21:34:16 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D3461D1B974 for ; Tue, 20 Apr 2004 21:34:15 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 82083-10 for ; Tue, 20 Apr 2004 21:34:16 -0300 (ADT) Received: from monsoon.he.net (monsoon.he.net [64.62.221.2]) by svr1.postgresql.org (Postfix) with SMTP id 04A2BD1E9FF for ; Tue, 20 Apr 2004 21:34:13 -0300 (ADT) Received: from [10.244.45.7] ([216.113.168.128]) by monsoon.he.net for ; Tue, 20 Apr 2004 17:34:08 -0700 In-Reply-To: References: <40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com> <1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us> Mime-Version: 1.0 (Apple Message framework v613) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <9D274446-932B-11D8-BA67-000393BD6C3E@tuckfield.com> Content-Transfer-Encoding: 7bit Cc: Tom Lane , pgsql-performance@postgresql.org From: Paul Tuckfield Subject: Re: Wierd context-switching issue on Xeon Date: Tue, 20 Apr 2004 17:34:13 -0700 To: Paul Tuckfield X-Mailer: Apple Mail (2.613) X-Virus-Scanned: by amavisd-new at postgresql.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/314 X-Sequence-Number: 6614 Ooops, what I meant to say was that 2 threads bound to one (hyperthreaded) cpu does *NOT* cause the storm, even on an smp xeon. Therefore, the context switches may be a result of cache coherency related delays. (2 threads on one hyperthreaded cpu presumably have tightly coupled 1,l2 cache.) On Apr 20, 2004, at 1:02 PM, Paul Tuckfield wrote: > I tried to test how this is related to cache coherency, by forcing > affinity of the two test_run.sql processes to the two cores > (pipelines? threads) of a single hyperthreaded xeon processor in an > smp xeon box. > > When the processes are allowed to run on distinct chips in the smp > box, the CS storm happens. When they are "bound" to the two cores of > a single hyperthreaded Xeon in the smp box, the CS storm *does* > happen. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ er, meant *NOT HAPPEN* > > > > I used the taskset command: > taskset 01 -p > taskset 01 -p > > I guess that 0 and 1 are the two cores (pipelines? hyper-threads?) on > the first Xeon processor in the box. > > I did this on RedHat Fedora core1 on an intel motherboard (I'll get > the part no if it matters) > > during storms : 300k CS/sec, 75% idle (on a dual xeon (four core)) > machine (suggesting serializing/sleeping processes) > no storm: 50k CS/sec, 50% idle (suggesting 2 cpu bound processes) > > > Maybe there's a "hot block" that is bouncing back and forth between > caches? or maybe the page holding semaphores? > > On Apr 19, 2004, at 5:53 PM, Tom Lane wrote: > >> I wrote: >>> Here is a test case. >> >> Hmmm ... I've been able to reproduce the CS storm on a dual Athlon, >> which seems to pretty much let the Xeon per se off the hook. Anybody >> got a multiple Opteron to try? Totally non-Intel CPUs? >> >> It would be interesting to see results with non-Linux kernels, too. >> >> regards, tom lane >> >> ---------------------------(end of >> broadcast)--------------------------- >> TIP 4: Don't 'kill -9' the postmaster >> > > > ---------------------------(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 Apr 20 22:21:16 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id C8C0CD1DC22 for ; Tue, 20 Apr 2004 22:21:14 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 02987-07 for ; Tue, 20 Apr 2004 22:21:15 -0300 (ADT) Received: from houston.familyhealth.com.au (fhnet.arach.net.au [203.22.197.21]) by svr1.postgresql.org (Postfix) with ESMTP id 98490D1D52E for ; Tue, 20 Apr 2004 22:21:11 -0300 (ADT) 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 i3L1KWWL039645; Wed, 21 Apr 2004 09:20:47 +0800 (WST) (envelope-from chriskl@familyhealth.com.au) Message-ID: <4085CDC3.4090201@familyhealth.com.au> Date: Wed, 21 Apr 2004 09:26:27 +0800 From: Christopher Kings-Lynne User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207) X-Accept-Language: en-us, en MIME-Version: 1.0 To: "Shea,Dan [CIS]" Cc: pgsql-performance@postgresql.org Subject: Re: Why will vacuum not end? 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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200404/319 X-Sequence-Number: 6619 > No, but data is constantly being inserted by userid scores. It is postgres > runnimg the vacuum. > Dan. Well, inserts create some locks - perhaps that's the problem... Otherwise, check the pg_locks view to see if you can figure it out. Chris From pgsql-performance-owner@postgresql.org Tue Apr 20 23:14:55 2004 X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id 35F4DD1BAF5 for ; Tue, 20 Apr 2004 23:14:54 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 25195-10 for ; Tue, 20 Apr 2004 23:14:55 -0300 (ADT) Received: from hotmail.com (law10-oe51.law10.hotmail.com [64.4.14.40]) by svr1.postgresql.org (Postfix) with ESMTP id AF845D1B974 for ; Tue, 20 Apr 2004 23:14:51 -0300 (ADT) Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Tue, 20 Apr 2004 19:14:51 -0700 Received: from 67.81.102.201 by law10-oe51.law10.hotmail.com with DAV; Wed, 21 Apr 2004 02:14:50 +0000 X-Originating-IP: [67.81.102.201] X-Originating-Email: [awerman2@hotmail.com] X-Sender: awerman2@hotmail.com From: "Aaron Werman" To: "Joshua D. Drake" , "Anjan Dave" Cc: References: <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814D@vt-pe2550-001.vantage.vantage.com> <4085BFF0.4050604@commandprompt.com> Subject: Re: Moving postgres to FC disks Date: Tue, 20 Apr 2004 22:14:51 -0400 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0064_01C42724.E65B1010" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2800.1409 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409 Message-ID: X-OriginalArrivalTime: 21 Apr 2004 02:14:51.0302 (UTC) FILETIME=[6D96F460:01C42746] 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_40_50, HTML_MESSAGE, HTML_TITLE_EMPTY, RCVD_IN_NJABL, RCVD_IN_SORBS X-Spam-Level: * X-Archive-Number: 200404/326 X-Sequence-Number: 6626 This is a multi-part message in MIME format. ------=_NextPart_000_0064_01C42724.E65B1010 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I agree on not linking and adding non-SAN disk dependancy to your DB. I'm t= rying to understand your FS reasoning. I have never seen XFS run faster tha= n ReiserFS in any situation (or for that matter beat any FS in performance = except JFS). XFS has some nifty very large file features, but we're talking= about 30G and all modern FSs support >2G files.=20 My tendancy would be to stay on ext3, since it is the default RH FS. I woul= d review site preference and the SAN recommended FS and see if they add any= compelling points. /Aaron ----- Original Message -----=20 From: Joshua D. Drake=20 To: Anjan Dave=20 Cc: pgsql-performance@postgresql.org=20 Sent: Tuesday, April 20, 2004 8:27 PM Subject: Re: [PERFORM] Moving postgres to FC disks -With the db size being as big as, say, 30+GB, how do I move it on the = new logical drive? (stop postgresql, and simply move it over somehow and ma= ke a link?) I would stop the database, move the data directory to the new volume usin= g rsync then start up postgresql pointed at the new data directory. Providing everything is working correctly you can then remove the old dat= a directory. -Currently, the internal RAID volume is ext3 filesystem. Any recommenda= tions for the filesystem on the new FC volume? Rieserfs? XFS DBs are 7.4.1(RH9), and 7.2.3 (RH8). Appreciate any pointers. Thanks, Anjan --=20 Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC and S/JDBC Postgresql support, programming shared hosting and dedicated hosting. +1-503-667-4564 - jd@commandprompt.com - http://www.commandprompt.com PostgreSQL Replicator -- production quality replication for PostgreSQL ------=_NextPart_000_0064_01C42724.E65B1010 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
I agree on not linking and adding non-SAN = disk=20 dependancy to your DB. I'm trying to understand your FS reasoning. <= FONT=20 face=3DArial size=3D2>I have never seen XFS run faster than ReiserFS in any= =20 situation (or for that matter beat any FS in performance except JFS). XFS h= as=20 some nifty very large file features, but we're talking about 30G and all mo= dern=20 FSs support >2G files.
=  
My tendancy would be to stay on ext3, sinc= e it is=20 the default RH FS. I would review site preference and the SAN recommended F= S and=20 see if they add any compelling points.
 
/Aaron
----- Original Message -----
Fro= m:=20 Josh= ua D.=20 Drake
Cc: pgsql-performance@postgr= esql.org=20
Sent: Tuesday, April 20, 2004 8:27= =20 PM
Subject: Re: [PERFORM] Moving post= gres to=20 FC disks


-With t= he db=20 size being as big as, say, 30+GB, how do I move it on the new logical d= rive?=20 (stop postgresql, and simply move it over somehow and make a=20 link?)

I would stop the database, m= ove the=20 data directory to the new volume using rsync then start up postgresql poi= nted=20 at the new data directory.
Providing everything is working correctly y= ou=20 can then remove the old data directory.

-Curren= tly,=20 the internal RAID volume is ext3 filesystem. Any recommendations for th= e=20 filesystem on the new FC volume? Rieserfs?

XFS

DBs are= =20 7.4.1(RH9), and 7.2.3 (RH8).

Appreci= ate any=20 pointers.

Thanks,=
Anjan



--=20
Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC and S/JDBC
Postgresql support, programming shared hosting and dedicated hosting.
+1-503-667-4564 - jd@commandprompt.com - http://www.commandprompt.com
PostgreSQL Replicator -- production quality replication for PostgreSQL

------=_NextPart_000_0064_01C42724.E65B1010--

From pgsql-performance-owner@postgresql.org  Tue Apr 20 23:38:24 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 62510D1B91C
	for ;
	Tue, 20 Apr 2004 23:38:22 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 37921-01
	for ;
	Tue, 20 Apr 2004 23:38:22 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 2253DD1B974
	for ;
	Tue, 20 Apr 2004 23:38:19 -0300 (ADT)
Received: (qmail 31760 invoked from network); 21 Apr 2004 02:38:17 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 21 Apr 2004 02:38:17 -0000
Subject: Re: Wierd context-switching issue on Xeon
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Josh Berkus 
Cc: Anjan Dave ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	Tom Lane , pgsql-performance@postgreSQL.org,
	Neil Conway 
In-Reply-To: <200404200959.52813.josh@agliodbs.com>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1082515263.1558.214.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Tue, 20 Apr 2004 22:41:03 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/327
X-Sequence-Number: 6627

I modified the code in s_lock.c to remove the spins

#define SPINS_PER_DELAY         1

and it doesn't exhibit the behaviour

This effectively changes the code to 


while(TAS(lock))
	select(10000); // 10ms

Can anyone explain why executing TAS 100 times would increase context
switches ?

Dave


On Tue, 2004-04-20 at 12:59, Josh Berkus wrote:
> Anjan,
> 
> > Quad 2.0GHz XEON with highest load we have seen on the applications, DB
> > performing great -
> 
> Can you run Tom's test?   It takes a particular pattern of data access to 
> reproduce the issue.
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Tue Apr 20 23:51:23 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 6905BD1C518; Tue, 20 Apr 2004 23:51:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 40830-02; Tue, 20 Apr 2004 23:51:18 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 18A8BD1C511; Tue, 20 Apr 2004 23:51:15 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3L2pFvA025065;
	Tue, 20 Apr 2004 22:51:15 -0400 (EDT)
To: Bruce Momjian 
Cc: PostgreSQL-development ,
	pgsql-performance@postgresql.org
Subject: Re: pgbench written in Pro*C 
In-reply-to: <200404201548.i3KFmA814010@candle.pha.pa.us> 
References: <200404201548.i3KFmA814010@candle.pha.pa.us>
Comments: In-reply-to Bruce Momjian 
	message dated "Tue, 20 Apr 2004 11:48:09 -0400"
Date: Tue, 20 Apr 2004 22:51:15 -0400
Message-ID: <25064.1082515875@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/328
X-Sequence-Number: 6628

Bruce Momjian  writes:
> I received a copy of pgbench rewritten in Pro*C, which is similar to
> embedded C.  I think it was done so the same program could be tested on
> Oracle and PostgreSQL.

> Are folks interested in this code?  Should it be put on gborg or in our
> /contrib/pgbench?

If it requires non-free tools even to build, it is of no value.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 21 00:32:02 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id B516ED1D0B8; Wed, 21 Apr 2004 00:31:59 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 58060-01; Wed, 21 Apr 2004 00:31:57 -0300 (ADT)
Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 5AC55D1CCA8; Wed, 21 Apr 2004 00:31:56 -0300 (ADT)
Received: (from pgman@localhost)
	by candle.pha.pa.us (8.11.6/8.11.6) id i3L3Vuk20705;
	Tue, 20 Apr 2004 23:31:56 -0400 (EDT)
From: Bruce Momjian 
Message-Id: <200404210331.i3L3Vuk20705@candle.pha.pa.us>
Subject: Re: pgbench written in Pro*C
In-Reply-To: <25064.1082515875@sss.pgh.pa.us>
To: Tom Lane 
Date: Tue, 20 Apr 2004 23:31:56 -0400 (EDT)
Cc: PostgreSQL-development ,
	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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/329
X-Sequence-Number: 6629

Tom Lane wrote:
> Bruce Momjian  writes:
> > I received a copy of pgbench rewritten in Pro*C, which is similar to
> > embedded C.  I think it was done so the same program could be tested on
> > Oracle and PostgreSQL.
> 
> > Are folks interested in this code?  Should it be put on gborg or in our
> > /contrib/pgbench?
> 
> If it requires non-free tools even to build, it is of no value.

OK, it's only value would be if we could modify it so it compiled using
our ecpg and Pro*C and the comparison program could be run on both
databases.

I will tell the submitter to put it on gborg if they wish.

-- 
  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 Apr 21 00:47:55 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C58F4D1D2B7
	for ;
	Wed, 21 Apr 2004 00:47:53 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 55599-10
	for ;
	Wed, 21 Apr 2004 00:47:52 -0300 (ADT)
Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82])
	by svr1.postgresql.org (Postfix) with ESMTP id 251B5D1D267
	for ;
	Wed, 21 Apr 2004 00:47:51 -0300 (ADT)
Received: from [192.168.4.3] (account jconway HELO joeconway.com)
	by joeconway.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP-TLS id 1476184; Tue, 20 Apr 2004 20:42:02 -0700
Message-ID: <4085EEB2.5080007@joeconway.com>
Date: Tue, 20 Apr 2004 20:46:58 -0700
From: Joe Conway 
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.6) Gecko/20040113
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Tom Lane 
Cc: josh@agliodbs.com, "scott.marlowe" ,
	Bruce Momjian , lutzeb@aeccom.com,
	pgsql-performance@postgresql.org, Neil Conway 
Subject: Re: Wierd context-switching issue on Xeon
References: 
	<40843E63.7050101@joeconway.com>
	<200404191455.04068.josh@agliodbs.com>
	<1407.1082419316@sss.pgh.pa.us> <40849235.2070808@joeconway.com>
In-Reply-To: <40849235.2070808@joeconway.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/330
X-Sequence-Number: 6630

Joe Conway wrote:
>> In isolation, test_run.sql should do essentially no syscalls at all once
>> it's past the initial ramp-up.  On a machine that's functioning per
>> expectations, multiple copies of test_run show a relatively low rate of
>> semop() calls --- a few per second, at most --- and maybe a delaying
>> select() here and there.

Here's results for 7.4 on a dual Athlon server running fedora core:

CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
            total   86.0%    0.0%   52.4%   0.0%     0.0%    0.0%   61.2%
            cpu00   37.6%    0.0%   29.7%   0.0%     0.0%    0.0%   32.6%
            cpu01   48.5%    0.0%   22.7%   0.0%     0.0%    0.0%   28.7%

procs                      memory      swap          io     system 
    cpu
  r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs
  1  0 120448  25764  48300 1094576    0    0     0   124  170   187
  1  0 120448  25780  48300 1094576    0    0     0     0  152    89
  2  0 120448  25744  48300 1094580    0    0     0    60  141 78290
  2  0 120448  25752  48300 1094580    0    0     0     0  131 140326
  2  0 120448  25756  48300 1094576    0    0     0    40  122 140100
  2  0 120448  25764  48300 1094584    0    0     0    60  133 136595
  2  0 120448  24284  48300 1094584    0    0     0   200  138 135151

The jump in cs corresponds to starting the query in the second session.

Joe


From pgsql-performance-owner@postgresql.org  Wed Apr 21 03:37:03 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 113BBD1D08B
	for ;
	Wed, 21 Apr 2004 03:37:02 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 03547-10
	for ;
	Wed, 21 Apr 2004 03:36:59 -0300 (ADT)
Received: from mail.unisoftbg.com (unknown [194.12.229.207])
	by svr1.postgresql.org (Postfix) with ESMTP id 1D66FD1C967
	for ;
	Wed, 21 Apr 2004 03:36:55 -0300 (ADT)
Received: (qmail 22265 invoked by uid 507); 21 Apr 2004 06:44:01 -0000
Received: from unknown (HELO t1.unisoftbg.com)
	(pginfo%t1.unisoftbg.com@194.12.229.193)
	by 0 with SMTP; 21 Apr 2004 06:44:01 -0000
Message-ID: <40860509.C74D7457@t1.unisoftbg.com>
Date: Wed, 21 Apr 2004 07:22:17 +0200
From: pginfo 
X-Mailer: Mozilla 4.03 [en] (Win95; I)
MIME-Version: 1.0
To: "pgsql-performance@postgresql.org" 
Subject: Re: Wierd context-switching issue on Xeon
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=1.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_DSBL,
	TO_ADDRESS_EQ_REAL
X-Spam-Level: *
X-Archive-Number: 200404/331
X-Sequence-Number: 6631

Hi,

Dual Xeon P4 2.8
linux RedHat AS 3
kernel 2.4.21-4-EL-smp
2 GB ram

I can see the same problem:

procs                      memory      swap          io
system         cpu
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy
id wa
1  0      0  96212  61056 1720240    0    0     0     0  101    11 25  0
75  0
 1  0      0  96212  61056 1720240    0    0     0     0  108   139 25
0 75  0
 1  0      0  96212  61056 1720240    0    0     0     0  104   173 25
0 75  0
 1  0      0  96212  61056 1720240    0    0     0     0  102    11 25
0 75  0
 1  0      0  96212  61056 1720240    0    0     0     0  101    11 25
0 75  0
 2  0      0  96204  61056 1720240    0    0     0     0  110 53866 31
4 65  0
 2  0      0  96204  61056 1720240    0    0     0     0  101 83176 41
5 54  0
 2  0      0  96204  61056 1720240    0    0     0     0  102 86050 39
6 55  0
 2  0      0  96204  61056 1720240    0    0     0    49  113 73642 41
5 54  0
 2  0      0  96204  61056 1720240    0    0     0     0  102 84211 40
5 55  0
 2  0      0  96204  61056 1720240    0    0     0     0  101 105165 39
7 54  0
 2  0      0  96204  61056 1720240    0    0     0     0  103 97754 38
6 56  0
 2  0      0  96204  61056 1720240    0    0     0     0  103 113668 36
7 57  0
 2  0      0  96204  61056 1720240    0    0     0     0  103 112003 37
7 56  0

regards,
ivan.


From pgsql-performance-owner@postgresql.org  Wed Apr 21 04:17:27 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 494B7D1D163
	for ;
	Wed, 21 Apr 2004 04:17:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 21391-10
	for ;
	Wed, 21 Apr 2004 04:17:24 -0300 (ADT)
Received: from FE-mail03.sfg.albacom.net (FE-mail03.albacom.net
	[213.217.149.83])
	by svr1.postgresql.org (Postfix) with ESMTP id B62A2D1CCB1
	for ;
	Wed, 21 Apr 2004 04:17:22 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail03.sfg.albacom.net (7.0.009)
	id 40757CA50010913C for pgsql-performance@postgresql.org;
	Wed, 21 Apr 2004 09:17:24 +0200
Message-ID: <40862003.2090201@axa.it>
Date: Wed, 21 Apr 2004 09:17:23 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: slow seqscan
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/332
X-Sequence-Number: 6632

My first post to this list :)

Scenario:
I have a database used only with search queries with only one table that
holds about 450.000/500.000 records.
The table is well indexed so that most of the queries are executed with
index scan but since there is a big text field in the table (360chars)
some search operation (with certain filters) ends up with seq scans.
This table is not written during normal operation: twice per week there
is a batch program that insert about 35.000 records and updates another
40.000.

last friday morning, after that batch has been executed, the database 
started responding really slowly to queries (expecially seq scans), 
after a "vacuum full analize" things did get something better.
Yesterday the same: before the batch everything was perfect, after every 
query was really slow, I've vacuum it again and now is ok.
Since now the db was working fine, it's 4 month's old with two updates 
per week and I vacuum about once per month.

I am using version 7.3 do I need to upgrade to 7.4? also, I was thinking
about setting this table in a kind of  "read-only" mode to improve
performance, is this possible?

Thank you for your help
Edoardo Ceccarelli

From pgsql-performance-owner@postgresql.org  Wed Apr 21 04:48:14 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 4844AD1CCB1
	for ;
	Wed, 21 Apr 2004 04:48:09 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 35026-07
	for ;
	Wed, 21 Apr 2004 04:48:08 -0300 (ADT)
Received: from obie.mrkdns.com (obie.mrkdns.com [69.93.72.82])
	by svr1.postgresql.org (Postfix) with ESMTP id D58EBD1BA6C
	for ;
	Wed, 21 Apr 2004 04:48:05 -0300 (ADT)
Received: from [213.152.63.90] (helo=chuckie.co.uk)
	by obie.mrkdns.com with asmtp (Exim 4.30)
	id 1BGCSk-00018K-Bc; Wed, 21 Apr 2004 08:48:06 +0100
Message-ID: <4086272E.2080102@chuckie.co.uk>
Date: Wed, 21 Apr 2004 08:47:58 +0100
From: Nick Barr 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Edoardo Ceccarelli 
Cc: pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it>
In-Reply-To: <40862003.2090201@axa.it>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-AntiAbuse: This header was added to track abuse,
	please include it with any abuse report
X-AntiAbuse: Primary Hostname - obie.mrkdns.com
X-AntiAbuse: Original Domain - postgresql.org
X-AntiAbuse: Originator/Caller UID/GID - [0 0] / [47 12]
X-AntiAbuse: Sender Address Domain - chuckie.co.uk
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/334
X-Sequence-Number: 6634

Edoardo Ceccarelli wrote:

> My first post to this list :)
>
> Scenario:
> I have a database used only with search queries with only one table that
> holds about 450.000/500.000 records.
> The table is well indexed so that most of the queries are executed with
> index scan but since there is a big text field in the table (360chars)
> some search operation (with certain filters) ends up with seq scans.
> This table is not written during normal operation: twice per week there
> is a batch program that insert about 35.000 records and updates another
> 40.000.
>
> last friday morning, after that batch has been executed, the database 
> started responding really slowly to queries (expecially seq scans), 
> after a "vacuum full analize" things did get something better.
> Yesterday the same: before the batch everything was perfect, after 
> every query was really slow, I've vacuum it again and now is ok.
> Since now the db was working fine, it's 4 month's old with two updates 
> per week and I vacuum about once per month.
>
> I am using version 7.3 do I need to upgrade to 7.4? also, I was thinking
> about setting this table in a kind of  "read-only" mode to improve
> performance, is this possible?
>
> Thank you for your help
> Edoardo Ceccarelli
>
> ---------------------------(end of broadcast)---------------------------
> TIP 8: explain analyze is your friend

In general we are going to need more information, like what kind of 
search filters you are using on the text field and an EXPLAIN ANALYZE. 
But can you try and run the following, bearing in mind it will take a 
while to complete.

REINDEX TABLE 

 From what I remember there were issues with index space not being 
reclaimed in a vacuum. I believe this was fixed in 7.4. By not 
reclaiming the space the indexes grow larger and larger over time, 
causing PG to prefer a sequential scan over an index scan (I think).


Hope that helps

Nick



From pgsql-performance-owner@postgresql.org  Wed Apr 21 04:46:59 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9BFBCD1B562
	for ;
	Wed, 21 Apr 2004 04:46:57 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 38316-05
	for ;
	Wed, 21 Apr 2004 04:46:57 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP id 9E145D1B507
	for ;
	Wed, 21 Apr 2004 04:46:54 -0300 (ADT)
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
	i3L7klWL048815; Wed, 21 Apr 2004 15:46:47 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <40862857.4030008@familyhealth.com.au>
Date: Wed, 21 Apr 2004 15:52:55 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Edoardo Ceccarelli 
Cc: pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it>
In-Reply-To: <40862003.2090201@axa.it>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=HTML_MESSAGE,
	HTML_TAG_BALANCE_TABLE
X-Spam-Level: 
X-Archive-Number: 200404/333
X-Sequence-Number: 6633

Hi Edoardo,

> The table is well indexed so that most of the queries are executed with
> index scan but since there is a big text field in the table (360chars)
> some search operation (with certain filters) ends up with seq scans.

Please paste the exact SELECT query that uses a seqscan, plus the 
EXPLAIN ANALYZE of the SELECT, and the psql output of \d .

> This table is not written during normal operation: twice per week there
> is a batch program that insert about 35.000 records and updates another
> 40.000.

After such an update, you need to run VACUUM ANALYZE 
;  Run it 
before the update as well, if it doesn't take that long.

> last friday morning, after that batch has been executed, the database 
> started responding really slowly to queries (expecially seq scans), 
> after a "vacuum full analize" things did get something better.
> Yesterday the same: before the batch everything was perfect, after every 
> query was really slow, I've vacuum it again and now is ok.
> Since now the db was working fine, it's 4 month's old with two updates 
> per week and I vacuum about once per month.

You need to vacuum analyze (NOT full) once and HOUR, not once a month. 
Add this command to your crontab to run once an hour and verify that 
it's working:

vacuumdb -a -z -q

Otherwise, install the auto vacuum utility found in 
contrib/pg_autovacuum in the postgres source.  Set this up.  It will 
monitor postgres and run vacuums and analyzes when necessary.  You can 
then remove your cron job.

> I am using version 7.3 do I need to upgrade to 7.4? also, I was thinking
> about setting this table in a kind of  "read-only" mode to improve
> performance, is this possible?

There's no read only mode to improve performance.

Upgrading to 7.4 will more than likely improve the performance of your 
database in general.  Be careful to read the upgrade notes because there 
were a few incompatibilities.

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 21 05:31:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DD4A4D1CCA8
	for ;
	Wed, 21 Apr 2004 05:31:47 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 45864-06
	for ;
	Wed, 21 Apr 2004 05:31:48 -0300 (ADT)
Received: from obie.mrkdns.com (obie.mrkdns.com [69.93.72.82])
	by svr1.postgresql.org (Postfix) with ESMTP id 1B463D1BB36
	for ;
	Wed, 21 Apr 2004 05:31:46 -0300 (ADT)
Received: from [213.152.63.90] (helo=chuckie.co.uk)
	by obie.mrkdns.com with asmtp (Exim 4.30) id 1BGD92-0003PT-Rg
	for pgsql-performance@postgresql.org; Wed, 21 Apr 2004 09:31:48 +0100
Message-ID: <4086316B.4050304@chuckie.co.uk>
Date: Wed, 21 Apr 2004 09:31:39 +0100
From: Nick Barr 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: PgSQL Performance ML 
Subject: MySQL vs PG TPC-H benchmarks
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-AntiAbuse: This header was added to track abuse,
	please include it with any abuse report
X-AntiAbuse: Primary Hostname - obie.mrkdns.com
X-AntiAbuse: Original Domain - postgresql.org
X-AntiAbuse: Originator/Caller UID/GID - [0 0] / [47 12]
X-AntiAbuse: Sender Address Domain - chuckie.co.uk
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/335
X-Sequence-Number: 6635

Hi,

Has anyone had a look at:

http://people.ac.upc.es/zgomez/

I realize that MySQL & PG cannot really be compared (especially when you 
consider the issues that MySQL has with things like data integrity) but 
still surely PG would perform better than the stats show (i.e. #7 31.28 
seconds versus 42 minutes!!!).

On a side note it certainly looks like linux kernel 2.6 is quite a bit 
faster in comparision to 2.4.

Nick




From pgsql-performance-owner@postgresql.org  Wed Apr 21 05:34:54 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 777B4D1BACD
	for ;
	Wed, 21 Apr 2004 05:34:53 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 45089-08
	for ;
	Wed, 21 Apr 2004 05:34:52 -0300 (ADT)
Received: from FE-mail04.sfg.albacom.net (FE-mail04.albacom.net
	[213.217.149.84])
	by svr1.postgresql.org (Postfix) with ESMTP id ADF96D1BABD
	for ;
	Wed, 21 Apr 2004 05:34:50 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail04.sfg.albacom.net (7.0.009)
	id 40757DA1000F0A8A; Wed, 21 Apr 2004 10:34:48 +0200
Message-ID: <40863228.907@axa.it>
Date: Wed, 21 Apr 2004 10:34:48 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: Nick Barr 
Cc: pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
In-Reply-To: <4086272E.2080102@chuckie.co.uk>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/336
X-Sequence-Number: 6636


>
> In general we are going to need more information, like what kind of 
> search filters you are using on the text field and an EXPLAIN ANALYZE. 
> But can you try and run the following, bearing in mind it will take a 
> while to complete.
>
> REINDEX TABLE 
>
> From what I remember there were issues with index space not being 
> reclaimed in a vacuum. I believe this was fixed in 7.4. By not 
> reclaiming the space the indexes grow larger and larger over time, 
> causing PG to prefer a sequential scan over an index scan (I think).
>
>

The query is this:
SELECT *, oid FROM annuncio400
WHERE  rubric = 'DD' AND LOWER(testo) Like LOWER('cbr%')
OFFSET 0 LIMIT 11

dba400=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
'DD' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11;
                                                    QUERY 
PLAN                                                    
-------------------------------------------------------------------------------------------------------------------- 

Limit  (cost=0.00..3116.00 rows=11 width=546) (actual time=51.47..56.42 
rows=11 loops=1)
  ->  Seq Scan on annuncio400  (cost=0.00..35490.60 rows=125 width=546) 
(actual time=51.47..56.40 rows=12 loops=1)
        Filter: ((rubric = 'DD'::bpchar) AND (lower((testo)::text) ~~ 
'cbr%'::text))
Total runtime: 56.53 msec
(4 rows)


But the strangest thing ever is that if I change the filter with another 
one that represent a smaller amount of data  it uses the index scan!!!
check this (same table, same query, different rubric=MA index):

dba400=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
'MA' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11; 
                                                         QUERY 
PLAN                                                          
------------------------------------------------------------------------------------------------------------------------------- 

Limit  (cost=0.00..6630.72 rows=9 width=546) (actual time=42.74..42.74 
rows=0 loops=1)
  ->  Index Scan using rubric on annuncio400  (cost=0.00..6968.48 rows=9 
width=546) (actual time=42.73..42.73 rows=0 loops=1)
        Index Cond: (rubric = 'MA'::bpchar)
        Filter: (lower((testo)::text) ~~ 'cbr%'::text)
Total runtime: 42.81 msec
(5 rows)


Thanks for your help
Edoardo

>
>

From pgsql-performance-owner@postgresql.org  Wed Apr 21 05:48:08 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B783DD1BACD
	for ;
	Wed, 21 Apr 2004 05:48:07 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 51863-06
	for ;
	Wed, 21 Apr 2004 05:48:06 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP id B295BD1BABA
	for ;
	Wed, 21 Apr 2004 05:48:02 -0300 (ADT)
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
	i3L8llWL057799; Wed, 21 Apr 2004 16:47:48 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <408636A5.2050305@familyhealth.com.au>
Date: Wed, 21 Apr 2004 16:53:57 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Edoardo Ceccarelli 
Cc: Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it>
In-Reply-To: <40863228.907@axa.it>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/337
X-Sequence-Number: 6637

> dba400=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
> 'DD' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11;
>                                                    QUERY 
> PLAN                                                    
> -------------------------------------------------------------------------------------------------------------------- 
> 
> Limit  (cost=0.00..3116.00 rows=11 width=546) (actual time=51.47..56.42 
> rows=11 loops=1)
>  ->  Seq Scan on annuncio400  (cost=0.00..35490.60 rows=125 width=546) 
> (actual time=51.47..56.40 rows=12 loops=1)
>        Filter: ((rubric = 'DD'::bpchar) AND (lower((testo)::text) ~~ 
> 'cbr%'::text))
> Total runtime: 56.53 msec
> (4 rows)

What happens if you go:

CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(rubric, 
LOWER(testo));

or even just:

CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(LOWER(testo));

> But the strangest thing ever is that if I change the filter with another 
> one that represent a smaller amount of data  it uses the index scan!!!

What's strange about that?  The less data is going to be retrieved, the 
more likely postgres is to use the index.

I suggest maybe increasing the amount of stats recorded for your rubrik 
column:

ALTER TABLE annuncio400 ALTER rubrik SET STATISTICS 100;
ANALYZE annuncio400;

You could also try reducing the random_page_cost value in your 
postgresql.conf a little, say to 3 (if it's currently 4).  That will 
make postgres more likely to use index scans over seq scans.

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 21 06:41:23 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 19A43D1D206
	for ;
	Wed, 21 Apr 2004 06:41:21 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 81683-02
	for ;
	Wed, 21 Apr 2004 06:41:21 -0300 (ADT)
Received: from FE-mail03.sfg.albacom.net (FE-mail03.albacom.net
	[213.217.149.83])
	by svr1.postgresql.org (Postfix) with ESMTP id 6BCF3D1C511
	for ;
	Wed, 21 Apr 2004 06:41:19 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail03.sfg.albacom.net (7.0.009)
	id 40757CA50010E027; Wed, 21 Apr 2004 11:41:12 +0200
Message-ID: <408641B7.5090808@axa.it>
Date: Wed, 21 Apr 2004 11:41:11 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: Christopher Kings-Lynne 
Cc: Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
In-Reply-To: <408636A5.2050305@familyhealth.com.au>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/338
X-Sequence-Number: 6638


> What happens if you go:
>
> CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(rubric, 
> LOWER(testo));
>
> or even just:
>
> CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(LOWER(testo));
>
I wasn't able to make this 2 field index with lower:

dba400=# CREATE INDEX annuncio400_rubric_testo_idx ON 
annuncio400(rubric, LOWER(testo));
ERROR:  parser: parse error at or near "(" at character 71

seems impossible to creat 2 field indexes with lower function.

The other one does not make it use the index.


>> But the strangest thing ever is that if I change the filter with 
>> another one that represent a smaller amount of data  it uses the 
>> index scan!!!
>
>
> What's strange about that?  The less data is going to be retrieved, 
> the more likely postgres is to use the index.
>
can't understand this policy:

dba400=# SELECT count(*) from annuncio400 where rubric='DD';
 count
-------
  6753
(1 row)

dba400=# SELECT count(*) from annuncio400 where rubric='MA';
 count
-------
  2165
(1 row)

so it's using the index on 2000 rows and not for 6000?  it's not that 
big difference, isn't it?


> I suggest maybe increasing the amount of stats recorded for your 
> rubrik column:
>
> ALTER TABLE annuncio400 ALTER rubrik SET STATISTICS 100;
> ANALYZE annuncio400;
>
done, almost the same, still not using index

> You could also try reducing the random_page_cost value in your 
> postgresql.conf a little, say to 3 (if it's currently 4).  That will 
> make postgres more likely to use index scans over seq scans.
>

changed the setting on postgresql.conf, restarted the server,
nothing has changed.

what about setting this to false?
#enable_seqscan = true

thanks again
Edoardo

From pgsql-performance-owner@postgresql.org  Wed Apr 21 07:10:41 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C0768D1BA8F
	for ;
	Wed, 21 Apr 2004 07:10:16 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 83703-05
	for ;
	Wed, 21 Apr 2004 07:10:12 -0300 (ADT)
Received: from FE-mail03.sfg.albacom.net (FE-mail03.albacom.net
	[213.217.149.83])
	by svr1.postgresql.org (Postfix) with ESMTP id 10781D1BAA2
	for ;
	Wed, 21 Apr 2004 07:10:10 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail03.sfg.albacom.net (7.0.009)
	id 40757CA50010EF88; Wed, 21 Apr 2004 12:10:03 +0200
Message-ID: <4086487A.4090703@axa.it>
Date: Wed, 21 Apr 2004 12:10:02 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
Cc: Christopher Kings-Lynne ,
	Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
	<408641B7.5090808@axa.it>
In-Reply-To: <408641B7.5090808@axa.it>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/339
X-Sequence-Number: 6639

tried the

enable_seqscan = false

and I'm having all index scans, timing has improved from 600ms to 18ms

wondering what other implications I might expect.




Edoardo Ceccarelli ha scritto:

>
>> What happens if you go:
>>
>> CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(rubric, 
>> LOWER(testo));
>>
>> or even just:
>>
>> CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(LOWER(testo));
>>
> I wasn't able to make this 2 field index with lower:
>
> dba400=# CREATE INDEX annuncio400_rubric_testo_idx ON 
> annuncio400(rubric, LOWER(testo));
> ERROR:  parser: parse error at or near "(" at character 71
>
> seems impossible to creat 2 field indexes with lower function.
>
> The other one does not make it use the index.
>
>
>>> But the strangest thing ever is that if I change the filter with 
>>> another one that represent a smaller amount of data  it uses the 
>>> index scan!!!
>>
>>
>>
>> What's strange about that?  The less data is going to be retrieved, 
>> the more likely postgres is to use the index.
>>
> can't understand this policy:
>
> dba400=# SELECT count(*) from annuncio400 where rubric='DD';
> count
> -------
>  6753
> (1 row)
>
> dba400=# SELECT count(*) from annuncio400 where rubric='MA';
> count
> -------
>  2165
> (1 row)
>
> so it's using the index on 2000 rows and not for 6000?  it's not that 
> big difference, isn't it?
>
>
>> I suggest maybe increasing the amount of stats recorded for your 
>> rubrik column:
>>
>> ALTER TABLE annuncio400 ALTER rubrik SET STATISTICS 100;
>> ANALYZE annuncio400;
>>
> done, almost the same, still not using index
>
>> You could also try reducing the random_page_cost value in your 
>> postgresql.conf a little, say to 3 (if it's currently 4).  That will 
>> make postgres more likely to use index scans over seq scans.
>>
>
> changed the setting on postgresql.conf, restarted the server,
> nothing has changed.
>
> what about setting this to false?
> #enable_seqscan = true
>
> thanks again
> Edoardo
>
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
>    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)
>
>
>

From pgsql-performance-owner@postgresql.org  Wed Apr 21 07:15:31 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 280C4D1BAD1
	for ;
	Wed, 21 Apr 2004 07:15:28 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 94103-01
	for ;
	Wed, 21 Apr 2004 07:15:28 -0300 (ADT)
Received: from vscan02.westnet.com.au (vscan02.westnet.com.au [203.10.1.132])
	by svr1.postgresql.org (Postfix) with ESMTP id B82F2D1BABA
	for ;
	Wed, 21 Apr 2004 07:15:25 -0300 (ADT)
Received: from localhost (localhost.localdomain [127.0.0.1])
	by localhost (Postfix) with ESMTP id C65DA10066F;
	Wed, 21 Apr 2004 18:15:25 +0800 (WST)
Received: from familyhealth.com.au (dsl-202-72-133-22.wa.westnet.com.au
	[202.72.133.22])
	by vscan02.westnet.com.au (Postfix) with ESMTP id 42B05100738;
	Wed, 21 Apr 2004 18:15:24 +0800 (WST)
Message-ID: <408649C1.3000904@familyhealth.com.au>
Date: Wed, 21 Apr 2004 18:15:29 +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: Edoardo Ceccarelli 
Cc: Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
	<408641B7.5090808@axa.it> <4086487A.4090703@axa.it>
In-Reply-To: <4086487A.4090703@axa.it>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/340
X-Sequence-Number: 6640


> enable_seqscan = false
> 
> and I'm having all index scans, timing has improved from 600ms to 18ms
> 
> wondering what other implications I might expect.

Lots of really bad implications...it's really not a good idea.

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 21 08:19:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id EE330D1D2B7
	for ;
	Wed, 21 Apr 2004 08:18:56 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 14043-04
	for ;
	Wed, 21 Apr 2004 08:18:58 -0300 (ADT)
Received: from server.pyrenet.fr (server.pyrenet.fr [194.250.190.1])
	by svr1.postgresql.org (Postfix) with ESMTP id CD8E8D1D067
	for ;
	Wed, 21 Apr 2004 08:18:53 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by localhost.pyrenet.fr (Postfix) with ESMTP
	id 91FE011035; Wed, 21 Apr 2004 13:18:56 +0200 (MET DST)
Received: from server.pyrenet.fr ([127.0.0.1])
	by localhost (server [127.0.0.1]) (amavisd-new, port 10024) with ESMTP
	id 07841-09; Wed, 21 Apr 2004 13:18:53 +0200 (MET DST)
Received: from server.pyrenet.fr (server.pyrenet.fr [194.250.190.1])
	by server.pyrenet.fr (Postfix) with ESMTP
	id 0A9ED11034; Wed, 21 Apr 2004 13:18:53 +0200 (MET DST)
Date: Wed, 21 Apr 2004 13:18:53 +0200 (MET DST)
From: ohp@pyrenet.fr
Reply-To: ohp@pyrenet.fr
To: Tom Lane 
Cc: josh@agliodbs.com, Joe Conway ,
	"scott.marlowe" ,
	Bruce Momjian , lutzeb@aeccom.com,
	pgsql-performance@postgresql.org, Neil Conway 
Subject: Re: Wierd context-switching issue on Xeon 
In-Reply-To: <1407.1082419316@sss.pgh.pa.us>
Message-ID: 
References: 
	<40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com>
	<1407.1082419316@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Virus-Scanned: by amavisd-new at pyrenet.fr
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0 tests=NO_REAL_NAME
X-Spam-Level: 
X-Archive-Number: 200404/341
X-Sequence-Number: 6641

How long is this test supposed to run?

I've launched just 1 for testing, the plan seems horrible; the test is cpu
bound and hasn't finished yet after 17:02 min of CPU time, dual XEON 2.6G
Unixware 713

The machine is a Fujitsu-Siemens TX 200 server
 On Mon, 19 Apr 2004, Tom Lane wrote:

> Date: Mon, 19 Apr 2004 20:01:56 -0400
> From: Tom Lane 
> To: josh@agliodbs.com
> Cc: Joe Conway , scott.marlowe ,
>      Bruce Momjian , lutzeb@aeccom.com,
>      pgsql-performance@postgresql.org, Neil Conway 
> Subject: Re: [PERFORM] Wierd context-switching issue on Xeon
>
> Here is a test case.  To set up, run the "test_setup.sql" script once;
> then launch two copies of the "test_run.sql" script.  (For those of
> you with more than two CPUs, see whether you need one per CPU to make
> trouble, or whether two test_runs are enough.)  Check that you get a
> nestloops-with-index-scans plan shown by the EXPLAIN in test_run.
>
> In isolation, test_run.sql should do essentially no syscalls at all once
> it's past the initial ramp-up.  On a machine that's functioning per
> expectations, multiple copies of test_run show a relatively low rate of
> semop() calls --- a few per second, at most --- and maybe a delaying
> select() here and there.
>
> What I actually see on Josh's client's machine is a context swap storm:
> "vmstat 1" shows CS rates around 170K/sec.  strace'ing the backends
> shows a corresponding rate of semop() syscalls, with a few delaying
> select()s sprinkled in.  top(1) shows system CPU percent of 25-30
> and idle CPU percent of 16-20.
>
> I haven't bothered to check how long the test_run query takes, but if it
> ends while you're still examining the behavior, just start it again.
>
> Note the test case assumes you've got shared_buffers set to at least
> 1000; with smaller values, you may get some I/O syscalls, which will
> probably skew the results.
>
> 			regards, tom lane
>
>

-- 
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
------------------------------------------------------------------------------
Make your life a dream, make your dream a reality. (St Exupery)

From pgsql-performance-owner@postgresql.org  Wed Apr 21 08:50:27 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 8837FD1BABD
	for ;
	Wed, 21 Apr 2004 08:50:15 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 22221-07
	for ;
	Wed, 21 Apr 2004 08:50:16 -0300 (ADT)
Received: from FE-mail04.sfg.albacom.net (FE-mail04.albacom.net
	[213.217.149.84])
	by svr1.postgresql.org (Postfix) with ESMTP id AB941D1B482
	for ;
	Wed, 21 Apr 2004 08:50:13 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail04.sfg.albacom.net (7.0.009)
	id 40757DA1000F5C81; Wed, 21 Apr 2004 13:50:05 +0200
Message-ID: <40865FEC.9030500@axa.it>
Date: Wed, 21 Apr 2004 13:50:04 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: Christopher Kings-Lynne 
Cc: Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
	<408641B7.5090808@axa.it> <4086487A.4090703@axa.it>
	<408649C1.3000904@familyhealth.com.au>
In-Reply-To: <408649C1.3000904@familyhealth.com.au>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/342
X-Sequence-Number: 6642

just created a copy of the same database and it shows that is the 
analyze that's messing things:

Slow seqscan query executed on dba400

dba400=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
'DD' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11;
                                                     QUERY 
PLAN                                                    
--------------------------------------------------------------------------------------------------------------------
 Limit  (cost=0.00..3116.00 rows=11 width=546) (actual time=46.66..51.40 
rows=11 loops=1)
   ->  Seq Scan on annuncio400  (cost=0.00..35490.60 rows=125 width=546) 
(actual time=46.66..51.38 rows=12 loops=1)
         Filter: ((rubric = 'DD'::bpchar) AND (lower((testo)::text) ~~ 
'cbr%'::text))
 Total runtime: 51.46 msec
(4 rows)


fastest index scan query on dba400b (exact copy of dba400)


dba400b=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
'DD' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11;
                                                          QUERY 
PLAN                                                         
------------------------------------------------------------------------------------------------------------------------------
 Limit  (cost=0.00..7058.40 rows=9 width=546) (actual time=1.36..8.18 
rows=11 loops=1)
   ->  Index Scan using rubric on annuncio400  (cost=0.00..7369.42 
rows=9 width=546) (actual time=1.35..8.15 rows=12 loops=1)
         Index Cond: (rubric = 'DD'::bpchar)
         Filter: (lower((testo)::text) ~~ 'cbr%'::text)
 Total runtime: 8.28 msec
(5 rows)



what about this index you suggested? it gives me sintax error while 
trying to create it:

CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(rubric, 
LOWER(testo));


Thanks
Edoardo

Christopher Kings-Lynne ha scritto:

>
>> enable_seqscan = false
>>
>> and I'm having all index scans, timing has improved from 600ms to 18ms
>>
>> wondering what other implications I might expect.
>
>
> Lots of really bad implications...it's really not a good idea.
>
> Chris
>
>
> ---------------------------(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  Wed Apr 21 09:12:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 533F0D1DC20
	for ;
	Wed, 21 Apr 2004 09:10:56 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 28490-06
	for ;
	Wed, 21 Apr 2004 09:10:56 -0300 (ADT)
Received: from mail2.aeccom.com (port-212-202-101-158.reverse.qsc.de
	[212.202.101.158])
	by svr1.postgresql.org (Postfix) with ESMTP id A9097D1DC1C
	for ;
	Wed, 21 Apr 2004 09:10:52 -0300 (ADT)
Received: from aeccom.com (cayambe.core.aeccom.com [192.168.2.12])
	by mail2.aeccom.com (Postfix) with ESMTP
	id 5DD3434; Wed, 21 Apr 2004 14:10:55 +0200 (CEST)
Message-ID: <408664CF.9040507@aeccom.com>
Date: Wed, 21 Apr 2004 14:10:55 +0200
From: =?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= 
Organization: AEC/communications GmbH, Berlin
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5) Gecko/20031016
X-Accept-Language: de, en
MIME-Version: 1.0
To: ohp@pyrenet.fr
Cc: Tom Lane , josh@agliodbs.com,
	Joe Conway , "scott.marlowe" ,
	Bruce Momjian ,
	pgsql-performance@postgresql.org, Neil Conway 
Subject: Re: Wierd context-switching issue on Xeon
References: 
	<40843E63.7050101@joeconway.com>
	<200404191455.04068.josh@agliodbs.com>
	<1407.1082419316@sss.pgh.pa.us>
	
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/343
X-Sequence-Number: 6643

It is intended to run indefinately.

Dirk

ohp@pyrenet.fr wrote:

>How long is this test supposed to run?
>
>I've launched just 1 for testing, the plan seems horrible; the test is cpu
>bound and hasn't finished yet after 17:02 min of CPU time, dual XEON 2.6G
>Unixware 713
>
>The machine is a Fujitsu-Siemens TX 200 server
>  
>



From pgsql-performance-owner@postgresql.org  Wed Apr 21 09:19:21 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 653E6D1DB5C
	for ;
	Wed, 21 Apr 2004 09:19:18 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36767-05
	for ;
	Wed, 21 Apr 2004 09:19:18 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CA260D1DB23
	for ;
	Wed, 21 Apr 2004 09:19:14 -0300 (ADT)
Received: from [134.22.70.203] (dyn-70-203.tor.dsl.tht.net [134.22.70.203])
	by tht.net (Postfix) with ESMTP
	id DBA9376AC4; Wed, 21 Apr 2004 08:19:18 -0400 (EDT)
Subject: Re: MySQL vs PG TPC-H benchmarks
From: Rod Taylor 
To: Nick Barr 
Cc: PgSQL Performance ML 
In-Reply-To: <4086316B.4050304@chuckie.co.uk>
References: <4086316B.4050304@chuckie.co.uk>
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature";
	boundary="=-3w/jNW2iPKXXKOlc+nq6"
Message-Id: <1082549960.80320.108.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Wed, 21 Apr 2004 08:19:21 -0400
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/344
X-Sequence-Number: 6644

--=-3w/jNW2iPKXXKOlc+nq6
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

> I realize that MySQL & PG cannot really be compared (especially when you=
=20
> consider the issues that MySQL has with things like data integrity) but=
=20
> still surely PG would perform better than the stats show (i.e. #7 31.28=
=20
> seconds versus 42 minutes!!!).

We know that PostgreSQL 7.5 will perform much better than 7.4 did due to
the efforts of OSDN and Tom.

I've enquired as to whether they ran ANALYZE after the data load. They
don't explicitly mention it, and given the mention it took 2.5days to
load 1GB of data, they're not regular PostgreSQL users.

--=20
Rod Taylor 

Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL
PGP Key: http://www.rbt.ca/signature.asc

--=-3w/jNW2iPKXXKOlc+nq6
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (FreeBSD)

iD8DBQBAhmbI6DETLow6vwwRAhRrAJ9/D8zePxGF2h84HiHyMaaf65/2LwCeInjq
OJ+5gmGE3tuldo5Yg5U3hog=
=79aK
-----END PGP SIGNATURE-----

--=-3w/jNW2iPKXXKOlc+nq6--


From pgsql-performance-owner@postgresql.org  Wed Apr 21 09:22:38 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 7E6C2D1DA3C
	for ;
	Wed, 21 Apr 2004 09:22:28 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36767-07
	for ;
	Wed, 21 Apr 2004 09:22:25 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 0E4DDD1BABF
	for ;
	Wed, 21 Apr 2004 09:22:22 -0300 (ADT)
Received: from [134.22.70.203] (dyn-70-203.tor.dsl.tht.net [134.22.70.203])
	by tht.net (Postfix) with ESMTP
	id 45E3476A11; Wed, 21 Apr 2004 08:22:27 -0400 (EDT)
Subject: Re: MySQL vs PG TPC-H benchmarks
From: Rod Taylor 
To: Nick Barr 
Cc: PgSQL Performance ML 
In-Reply-To: <1082549960.80320.108.camel@jester>
References: <4086316B.4050304@chuckie.co.uk>
	<1082549960.80320.108.camel@jester>
Content-Type: text/plain
Message-Id: <1082550148.80320.111.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Wed, 21 Apr 2004 08:22:29 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/345
X-Sequence-Number: 6645

On Wed, 2004-04-21 at 08:19, Rod Taylor wrote:
> > I realize that MySQL & PG cannot really be compared (especially when you 
> > consider the issues that MySQL has with things like data integrity) but 
> > still surely PG would perform better than the stats show (i.e. #7 31.28 
> > seconds versus 42 minutes!!!).
> 
> We know that PostgreSQL 7.5 will perform much better than 7.4 did due to
> the efforts of OSDN and Tom.

OSDL not OSDN.

> I've enquired as to whether they ran ANALYZE after the data load. They
> don't explicitly mention it, and given the mention it took 2.5days to
> load 1GB of data, they're not regular PostgreSQL users.


From pgsql-performance-owner@postgresql.org  Wed Apr 21 09:56:03 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A5EE2D1D2AA
	for ;
	Wed, 21 Apr 2004 09:55:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 45681-07
	for ;
	Wed, 21 Apr 2004 09:55:26 -0300 (ADT)
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 2AB4DD1D26C
	for ;
	Wed, 21 Apr 2004 09:55:22 -0300 (ADT)
Received: from tmsl-adsl.demon.co.uk ([80.177.114.181]
	helo=bacon.tmsl.demon.co.uk)
	by anchor-post-35.mail.demon.net with esmtp (Exim 3.35 #1)
	id 1BGHG7-0002yb-0Z; Wed, 21 Apr 2004 13:55:23 +0100
Date: Wed, 21 Apr 2004 13:55:21 +0100
From: Paul Thomas 
To: Nick Barr 
Cc: "pgsql-performance @ postgresql . org" 
Subject: Re: MySQL vs PG TPC-H benchmarks
Message-ID: <20040421135521.A23015@bacon>
References: <4086316B.4050304@chuckie.co.uk>
Mime-Version: 1.0
Content-Type: text/plain; format=flowed; charset=ISO-8859-1
In-Reply-To: <4086316B.4050304@chuckie.co.uk>;
	from nicky@chuckie.co.uk on Wed, Apr 21, 2004 at 09:31:39 +0100
X-Mailer: Balsa 1.2.3
Lines: 34
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/346
X-Sequence-Number: 6646


On 21/04/2004 09:31 Nick Barr wrote:
> Hi,
> 
> Has anyone had a look at:
> 
> http://people.ac.upc.es/zgomez/
> 
> I realize that MySQL & PG cannot really be compared (especially when you 
> consider the issues that MySQL has with things like data integrity) but 
> still surely PG would perform better than the stats show (i.e. #7 31.28 
> seconds versus 42 minutes!!!).

Looks like he's using the default postgresql.conf settings in which case 
I'm not suprised at pg looking so slow. His stated use of foreign keys 
invalidates the tests anyway as MyISAM tables don't support FKs so we're 
probably seeing FK check overheads in pg that are simply ignore by MySQL. 
In an honest test, MySQL should be reported as failing those tests.

Perhaps one of the advocay team will pick up the batton?
> 
> On a side note it certainly looks like linux kernel 2.6 is quite a bit 
> faster in comparision to 2.4.

Yes, I've seen other benchmarks which also show that.

-- 
Paul Thomas
+------------------------------+---------------------------------------------+
| Thomas Micro Systems Limited | Software Solutions for 
Business             |
| Computer Consultants         | 
http://www.thomas-micro-systems-ltd.co.uk   |
+------------------------------+---------------------------------------------+

From pgsql-performance-owner@postgresql.org  Wed Apr 21 10:32:03 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 0347DD1D2AA
	for ;
	Wed, 21 Apr 2004 10:31:09 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 57905-09
	for ;
	Wed, 21 Apr 2004 10:31:05 -0300 (ADT)
Received: from smtp.casablanca.sk (smtp.casablanca.sk [81.0.239.6])
	by svr1.postgresql.org (Postfix) with ESMTP id 766B4D1D26C
	for ;
	Wed, 21 Apr 2004 10:31:00 -0300 (ADT)
Received: from omikron.sk (cepr.nustep.sk [81.0.222.49])
	by smtp.casablanca.sk (8.12.9/8.12.9) with SMTP id i3LDXBI3028005
	for ; Wed, 21 Apr 2004 15:33:11 +0200
Received: (qmail 11404 invoked from network); 21 Apr 2004 13:31:02 -0000
Received: from unknown (HELO stratos) (158.195.31.19)
	by 0 with SMTP; 21 Apr 2004 13:31:02 -0000
Message-ID: <037701c427a4$e41a38a0$0200a8c0@stratos>
From: "Cestmir Hybl" 
To: "Paul Thomas" ,
	"Nick Barr" 
Cc: 
References: <4086316B.4050304@chuckie.co.uk> <20040421135521.A23015@bacon>
Subject: Re: MySQL vs PG TPC-H benchmarks
Date: Wed, 21 Apr 2004 15:31:02 +0200
Organization: NUSTEP s.r.o.
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 6.00.2800.1158
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/347
X-Sequence-Number: 6647

> Looks like he's using the default postgresql.conf settings in which case
> I'm not suprised at pg looking so slow.

The question also is, IMHO, why the hell, postgreSQL still comes out of the
box with so stupid configuration defaults, totally underestimated for todays
average hardware configuration (1+GHz, 0.5+GB RAM, fast FSB, fast HDD).

It seems to me better strategy to force that 1% of users to "downgrade" cfg.
than vice-versa.

regards
ch


From pgsql-performance-owner@postgresql.org  Wed Apr 21 11:09:35 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D9931D1DD1F
	for ;
	Wed, 21 Apr 2004 11:08:13 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 72738-10
	for ;
	Wed, 21 Apr 2004 11:08:15 -0300 (ADT)
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 7F122D1DD1B
	for ;
	Wed, 21 Apr 2004 11:08:11 -0300 (ADT)
Received: from tmsl-adsl.demon.co.uk ([80.177.114.181]
	helo=bacon.tmsl.demon.co.uk)
	by anchor-post-33.mail.demon.net with esmtp (Exim 3.35 #1)
	id 1BGIOd-0002eN-0X; Wed, 21 Apr 2004 15:08:15 +0100
Date: Wed, 21 Apr 2004 15:08:09 +0100
From: Paul Thomas 
To: Cestmir Hybl 
Cc: pgsql-performance@postgresql.org
Subject: Re: MySQL vs PG TPC-H benchmarks
Message-ID: <20040421150809.C23015@bacon>
References: <4086316B.4050304@chuckie.co.uk> <20040421135521.A23015@bacon>
	<037701c427a4$e41a38a0$0200a8c0@stratos>
Mime-Version: 1.0
Content-Type: text/plain; format=flowed; charset=ISO-8859-1
In-Reply-To: <037701c427a4$e41a38a0$0200a8c0@stratos>;
	from cestmirl@freeside.sk on Wed, Apr 21, 2004 at 14:31:02 +0100
X-Mailer: Balsa 1.2.3
Lines: 29
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/348
X-Sequence-Number: 6648

On 21/04/2004 14:31 Cestmir Hybl wrote:
> > Looks like he's using the default postgresql.conf settings in which
> case
> > I'm not suprised at pg looking so slow.
> 
> The question also is, IMHO, why the hell, postgreSQL still comes out of
> the
> box with so stupid configuration defaults, totally underestimated for
> todays
> average hardware configuration (1+GHz, 0.5+GB RAM, fast FSB, fast HDD).
> 
> It seems to me better strategy to force that 1% of users to "downgrade"
> cfg.
> than vice-versa.
> 
> regards
> ch
> 

This has been discussed many times before. Check the archives.

-- 
Paul Thomas
+------------------------------+---------------------------------------------+
| Thomas Micro Systems Limited | Software Solutions for 
Business             |
| Computer Consultants         | 
http://www.thomas-micro-systems-ltd.co.uk   |
+------------------------------+---------------------------------------------+

From pgsql-performance-owner@postgresql.org  Wed Apr 21 11:37:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6ABE9D1BB64
	for ;
	Wed, 21 Apr 2004 11:31:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 79866-10
	for ;
	Wed, 21 Apr 2004 11:31:27 -0300 (ADT)
Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210])
	by svr1.postgresql.org (Postfix) with ESMTP id 44F9FD1BABD
	for ;
	Wed, 21 Apr 2004 11:31:23 -0300 (ADT)
Received: by megazone.bigpanda.com (Postfix, from userid 1001)
	id 380D935A6C; Wed, 21 Apr 2004 07:31:27 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1])
	by megazone.bigpanda.com (Postfix) with ESMTP
	id 3672F35A68; Wed, 21 Apr 2004 07:31:27 -0700 (PDT)
Date: Wed, 21 Apr 2004 07:31:27 -0700 (PDT)
From: Stephan Szabo 
To: Edoardo Ceccarelli 
Cc: Christopher Kings-Lynne ,
	Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
In-Reply-To: <408641B7.5090808@axa.it>
Message-ID: <20040421072156.K55057@megazone.bigpanda.com>
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
	<408641B7.5090808@axa.it>
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/349
X-Sequence-Number: 6649


On Wed, 21 Apr 2004, Edoardo Ceccarelli wrote:

>
> > What happens if you go:
> >
> > CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(rubric,
> > LOWER(testo));
> >
> > or even just:
> >
> > CREATE INDEX annuncio400_rubric_testo_idx ON annuncio400(LOWER(testo));
> >
> I wasn't able to make this 2 field index with lower:
>
> dba400=# CREATE INDEX annuncio400_rubric_testo_idx ON
> annuncio400(rubric, LOWER(testo));
> ERROR:  parser: parse error at or near "(" at character 71

That's a 7.4 feature I think (and I think the version with two columns
may need extra parens around the lower()). I think the only way to do
something equivalent in 7.3 is to make a function that concatenates the
two in some fashion after having applied the lower to the one part and
then using that in the queries as well.  Plus, if you're not in "C"
locale, I'm not sure that it'd help in 7.3 anyway.

> >> But the strangest thing ever is that if I change the filter with
> >> another one that represent a smaller amount of data  it uses the
> >> index scan!!!
> >
> >
> > What's strange about that?  The less data is going to be retrieved,
> > the more likely postgres is to use the index.
> >
> can't understand this policy:
>
> dba400=# SELECT count(*) from annuncio400 where rubric='DD';
>  count
> -------
>   6753
> (1 row)
>
> dba400=# SELECT count(*) from annuncio400 where rubric='MA';
>  count
> -------
>   2165
> (1 row)
>
> so it's using the index on 2000 rows and not for 6000?  it's not that
> big difference, isn't it?

It's a question of how many pages it thinks it's going to have to retrieve
in order to handle the request.  If it say needs (or think it needs) to
retrieve 50% of the pages, then given a random_page_cost of 4, it's going
to expect the index scan to be about twice the cost.

Generally speaking one good way to compare is to try the query with
explain analyze and then change parameters like enable_seqscan and try the
query with explain analyze again and compare the estimated rows and costs.
That'll give an idea of how it expects the two versions of the query to
compare speed wise.

From pgsql-performance-owner@postgresql.org  Wed Apr 21 12:06:29 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 434EAD1B47D
	for ;
	Wed, 21 Apr 2004 12:02:48 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 95716-06
	for ;
	Wed, 21 Apr 2004 12:02:46 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id D702CD1DD93
	for ;
	Wed, 21 Apr 2004 12:02:45 -0300 (ADT)
Received: (qmail 9877 invoked from network); 21 Apr 2004 15:02:37 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 21 Apr 2004 15:02:37 -0000
Subject: Re: Wierd context-switching issue on Xeon
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: =?ISO-8859-1?Q?Dirk=5FLutzeb=E4ck?= 
Cc: ohp@pyrenet.fr, Tom Lane ,
	Josh Berkus , Joe Conway ,
	"scott.marlowe" ,
	Bruce Momjian ,
	pgsql-performance@postgresql.org, Neil Conway 
In-Reply-To: <408664CF.9040507@aeccom.com>
References: 
	<40843E63.7050101@joeconway.com> <200404191455.04068.josh@agliodbs.com>
	<1407.1082419316@sss.pgh.pa.us>
	
	<408664CF.9040507@aeccom.com>
Content-Type: text/plain; charset=ISO-8859-1
Organization: Cramer Consulting
Message-Id: <1082559931.1557.235.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Wed, 21 Apr 2004 11:05:31 -0400
Content-Transfer-Encoding: 8bit
X-AntiVirus: scanned for viruses by AMaViS 0.2.1 (http://amavis.org/)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No,
 hits=2.0 tagged_above=0.0 required=5.0 tests=DRASTIC_REDUCED
X-Spam-Level: *
X-Archive-Number: 200404/350
X-Sequence-Number: 6650

After some testing if you use the current head code for s_lock.c which
has some mods in it to alleviate this situation, and change
SPINS_PER_DELAY to 10 you can drastically reduce the cs with tom's test.
I am seeing a slight degradation in throughput using pgbench -c 10 -t
1000 but it might be liveable, considering the alternative is unbearable
in some situations.

Can anyone else replicate my results?

Dave
On Wed, 2004-04-21 at 08:10, Dirk_Lutzeb�ck wrote:
> It is intended to run indefinately.
> 
> Dirk
> 
> ohp@pyrenet.fr wrote:
> 
> >How long is this test supposed to run?
> >
> >I've launched just 1 for testing, the plan seems horrible; the test is cpu
> >bound and hasn't finished yet after 17:02 min of CPU time, dual XEON 2.6G
> >Unixware 713
> >
> >The machine is a Fujitsu-Siemens TX 200 server
> >  
> >
> 
> 
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
>     (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)
> 
> 
> 
> !DSPAM:40866735106778584283649!
> 
> 
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Wed Apr 21 12:46:11 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B9061D1B482
	for ;
	Wed, 21 Apr 2004 12:34:21 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 11447-04
	for ;
	Wed, 21 Apr 2004 12:34:20 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id 4D361D1B47D
	for ;
	Wed, 21 Apr 2004 12:34:19 -0300 (ADT)
thread-index: AcQntpQqm5wXS2HIRF2GRclnFZOqtg==
Content-Class: urn:content-classes:message
Importance: normal
Received: from [10.4.2.163] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Wed, 21 Apr 2004 11:37:39 -0400
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
From: "Chris Hoover" 
To: 
Subject: Help understanding stat tables
Date: Wed, 21 Apr 2004 11:34:16 -0400
User-Agent: KMail/1.6.1
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404211134.16876.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 21 Apr 2004 15:37:39.0562 (UTC)
	FILETIME=[941C24A0:01C427B6]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/351
X-Sequence-Number: 6651

I just want to make sure that I am interpreting this data correctly.

=46rom pg_statio_user_tables, I have pulled relname, heap_blks_read,=20
heap_blks_hit.  I get several rows like this:
relname				heap_bkls_read	heap_blks_hit
 clmhdr                       	8607161			196547165=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=
=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09=09


So this means that I am getting over a 100% cache hit ratio for this table,=
=20
right?  If not, please help me understand what these numbers mean.

Thanks,

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 21 13:42:36 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 43E9FD1BA8F
	for ;
	Wed, 21 Apr 2004 13:23:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 25305-07
	for ;
	Wed, 21 Apr 2004 13:23:23 -0300 (ADT)
Received: from smtp1.aruba.it (smtp1.aruba.it [62.149.128.200])
	by svr1.postgresql.org (Postfix) with SMTP id 75B59D1DF75
	for ;
	Wed, 21 Apr 2004 13:23:22 -0300 (ADT)
Received: (qmail 14567 invoked from network); 21 Apr 2004 16:23:10 -0000
Received: from unknown (HELO axa.it) (80.183.237.70)
	by smtp1.aruba.it with SMTP; 21 Apr 2004 16:23:10 -0000
Message-ID: <40869FF0.6090100@axa.it>
Date: Wed, 21 Apr 2004 18:23:12 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: Stephan Szabo 
Cc: Christopher Kings-Lynne ,
	Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
	<408641B7.5090808@axa.it>
	<20040421072156.K55057@megazone.bigpanda.com>
In-Reply-To: <20040421072156.K55057@megazone.bigpanda.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Spam-Rating: smtp1.aruba.it 1.6.2 0/1000/N
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=RCVD_IN_NJABL,
	RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/354
X-Sequence-Number: 6654


>>can't understand this policy:
>>
>>dba400=# SELECT count(*) from annuncio400 where rubric='DD';
>> count
>>-------
>>  6753
>>(1 row)
>>
>>dba400=# SELECT count(*) from annuncio400 where rubric='MA';
>> count
>>-------
>>  2165
>>(1 row)
>>
>>so it's using the index on 2000 rows and not for 6000?  it's not that
>>big difference, isn't it?
>>    
>>
>
>It's a question of how many pages it thinks it's going to have to retrieve
>in order to handle the request.  If it say needs (or think it needs) to
>retrieve 50% of the pages, then given a random_page_cost of 4, it's going
>to expect the index scan to be about twice the cost.
>
>Generally speaking one good way to compare is to try the query with
>explain analyze and then change parameters like enable_seqscan and try the
>query with explain analyze again and compare the estimated rows and costs.
>That'll give an idea of how it expects the two versions of the query to
>compare speed wise.
>
>
>  
>
Ok then how do you explain this?
just created a copy of the same database

Slow seqscan query executed on dba400

dba400=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
'DD' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11;
                                                    QUERY 
PLAN                                                    
-------------------------------------------------------------------------------------------------------------------- 

Limit  (cost=0.00..3116.00 rows=11 width=546) (actual time=46.66..51.40 
rows=11 loops=1)
  ->  Seq Scan on annuncio400  (cost=0.00..35490.60 rows=125 width=546) 
(actual time=46.66..51.38 rows=12 loops=1)
        Filter: ((rubric = 'DD'::bpchar) AND (lower((testo)::text) ~~ 
'cbr%'::text))
Total runtime: 51.46 msec
(4 rows)


fastest index scan query on dba400b (exact copy of dba400)


dba400b=# explain analyze SELECT *, oid FROM annuncio400 WHERE  rubric = 
'DD' AND LOWER(testo) Like LOWER('cbr%') OFFSET 0 LIMIT 11;
                                                         QUERY 
PLAN                                                         
------------------------------------------------------------------------------------------------------------------------------ 

Limit  (cost=0.00..7058.40 rows=9 width=546) (actual time=1.36..8.18 
rows=11 loops=1)
  ->  Index Scan using rubric on annuncio400  (cost=0.00..7369.42 rows=9 
width=546) (actual time=1.35..8.15 rows=12 loops=1)
        Index Cond: (rubric = 'DD'::bpchar)
        Filter: (lower((testo)::text) ~~ 'cbr%'::text)
Total runtime: 8.28 msec
(5 rows)


anyway, shall I try to lower the random_page value since I get an index 
scan? I mean that in my case I've already noted that with index scan 
that query get executed in 1/10 of the seqscan speed.

Thank you
Edoardo

From pgsql-performance-owner@postgresql.org  Wed Apr 21 13:42:04 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1BB4AD1DF5D
	for ;
	Wed, 21 Apr 2004 13:25:33 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 25457-06
	for ;
	Wed, 21 Apr 2004 13:25:31 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id A77CCD1BAA2
	for ;
	Wed, 21 Apr 2004 13:25:00 -0300 (ADT)
thread-index: AcQnvZuLnERkrY/KQWur/4uiE99PJA==
Received: from [10.4.2.163] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Wed, 21 Apr 2004 12:27:58 -0400
Content-Class: urn:content-classes:message
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
From: "Chris Hoover" 
To: 
Subject: Re: Help understanding stat tables
Date: Wed, 21 Apr 2004 12:24:35 -0400
User-Agent: KMail/1.6.1
References: <200404211134.16876.revoohc@sermonaudio.com>
In-Reply-To: <200404211134.16876.revoohc@sermonaudio.com>
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404211224.35696.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 21 Apr 2004 16:27:58.0437 (UTC)
	FILETIME=[9B7FE950:01C427BD]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/353
X-Sequence-Number: 6653

I think I have figured my problem out.

I was taking heap_blks_hit / heap_blks_read for my hit pct.

It should be heap_blks_hit/(heap_blks_read+heap_blks_hit), correct?

Thanks
On Wednesday 21 April 2004 11:34, Chris Hoover wrote:
> I just want to make sure that I am interpreting this data correctly.
>
> From pg_statio_user_tables, I have pulled relname, heap_blks_read,
> heap_blks_hit.  I get several rows like this:
> relname				heap_bkls_read	heap_blks_hit
>  clmhdr                       	8607161			196547165
>
>
> So this means that I am getting over a 100% cache hit ratio for this tabl=
e,
> right?  If not, please help me understand what these numbers mean.
>
> Thanks,
>
> Chris
>
>
> ---------------------------(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 Apr 21 13:35:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 0C062D1BAA2
	for ;
	Wed, 21 Apr 2004 13:29:13 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 28269-08
	for ;
	Wed, 21 Apr 2004 13:29:10 -0300 (ADT)
Received: from gin.mims.be (gin.mims.be [217.145.39.2])
	by svr1.postgresql.org (Postfix) with SMTP id 659AFD1B4C3
	for ;
	Wed, 21 Apr 2004 13:29:08 -0300 (ADT)
Received: (qmail 50561 invoked from network); 21 Apr 2004 16:29:05 -0000
Received: from localhost by gin.mims.be (uid 82) with SMTP (mims-scanner-1.3,
	0.378714 s); 21 Apr 2004 16:29:05 -0000
Received: from 213-193-172-189.adsl.easynet.be (HELO frmims) (213.193.172.189)
	by gin.mims.be with SMTP; 21 Apr 2004 16:29:05 -0000
Message-ID: <009901c427bd$c4752410$c601a8c0@MIMS>
From: =?iso-8859-1?Q?Fr=E9d=E9ric_Robinet?= 
To: 
Subject: Shared buffers, Sort memory, Effective Cache Size
Date: Wed, 21 Apr 2004 18:29:06 +0200
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2800.1409
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409
X-Virus-Flag: NO
X-Virus-Scanner: F-Prot v4.4.1 (engine v3.14.11)
X-Virus-Definitions: SIGN.DEF (21 April 2004), SIGN2.DEF (21 April 2004),
	MACRO.DEF (21 April 2004)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/352
X-Sequence-Number: 6652

Hello,

I have a bi-PIII server with 2Gb of RAM with Debian and a PostgreSQL 7.4
running on.  What are the bests settings for shared buffers, sort memory and
effective cache size?

My main database have a small/mid range size: some tables may have 1 or 2
millions of records.

Thanks

Fr�d�ric Robinet
Fr@Mims.be


From pgsql-performance-owner@postgresql.org  Mon Apr 26 14:47:13 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 719D7D1E029
	for ;
	Wed, 21 Apr 2004 14:01:35 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 41794-04
	for ;
	Wed, 21 Apr 2004 14:01:33 -0300 (ADT)
Received: from leo.supplyfx.com (adsl-67-119-155-141.dsl.lsan03.pacbell.net
	[67.119.155.141])
	by svr1.postgresql.org (Postfix) with ESMTP id 8D0A9D1DFA1
	for ;
	Wed, 21 Apr 2004 14:01:32 -0300 (ADT)
Received: from [10.0.100.21] (leo6.supplyfx.com [10.0.100.21])
	by leo.supplyfx.com (Postfix) with ESMTP
	id 7037E330BC; Wed, 21 Apr 2004 10:01:32 -0700 (PDT)
In-Reply-To: <009901c427bd$c4752410$c601a8c0@MIMS>
References: <009901c427bd$c4752410$c601a8c0@MIMS>
Mime-Version: 1.0 (Apple Message framework v613)
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Message-Id: <8937BC90-93B5-11D8-AC12-000A95AB8896@supplyfx.com>
Content-Transfer-Encoding: quoted-printable
Cc: 
From: Qing Zhao 
Subject: Re: Shared buffers, Sort memory, Effective Cache Size
Date: Wed, 21 Apr 2004 10:01:30 -0700
To: =?ISO-8859-1?Q?Fr=E9d=E9ric_Robinet?= 
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=2.6 tagged_above=0.0 required=5.0
	tests=RCVD_IN_DYNABLOCK, RCVD_IN_SORBS
X-Spam-Level: **
X-Archive-Number: 200404/439
X-Sequence-Number: 6739

Hello,

I have recently configured my PG7.3 on a G5 (8GB RAM) with
shmmax set to 512MB and shared_buffer=3D50000, sort_mem=3D4096
and effective cache size =3D 10000.  It seems working great so far but
I am wondering if I should make effctive cache size larger myself.

Tnaks!

Qing
On Apr 21, 2004, at 9:29 AM, Fr=E9d=E9ric Robinet wrote:

> Hello,
>
> I have a bi-PIII server with 2Gb of RAM with Debian and a PostgreSQL=20
> 7.4
> running on.  What are the bests settings for shared buffers, sort=20
> memory and
> effective cache size?
>
> My main database have a small/mid range size: some tables may have 1=20
> or 2
> millions of records.
>
> Thanks
>
> Fr=E9d=E9ric Robinet
> Fr@Mims.be
>
>
> ---------------------------(end of=20
> broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
>


From pgsql-performance-owner@postgresql.org  Wed Apr 21 14:30:25 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D4664D1BABD
	for ;
	Wed, 21 Apr 2004 14:30:19 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 51908-03
	for ;
	Wed, 21 Apr 2004 14:30:13 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 09197D1BA6C
	for ;
	Wed, 21 Apr 2004 14:30:11 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4915415; Wed, 21 Apr 2004 10:31:33 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: pg@fastcrypt.com, =?iso-8859-1?q?Dirk=5FLutzeb=E4ck?= 
Subject: Re: Wierd context-switching issue on Xeon
Date: Wed, 21 Apr 2004 10:29:43 -0700
User-Agent: KMail/1.4.3
Cc: ohp@pyrenet.fr, Tom Lane ,
	Joe Conway , "scott.marlowe" ,
	Bruce Momjian ,
	pgsql-performance@postgresql.org, Neil Conway 
References: 
	<408664CF.9040507@aeccom.com>
	<1082559931.1557.235.camel@localhost.localdomain>
In-Reply-To: <1082559931.1557.235.camel@localhost.localdomain>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404211029.43675.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No,
 hits=2.0 tagged_above=0.0 required=5.0 tests=DRASTIC_REDUCED
X-Spam-Level: *
X-Archive-Number: 200404/355
X-Sequence-Number: 6655

Dave,

> After some testing if you use the current head code for s_lock.c which
> has some mods in it to alleviate this situation, and change
> SPINS_PER_DELAY to 10 you can drastically reduce the cs with tom's test.
> I am seeing a slight degradation in throughput using pgbench -c 10 -t
> 1000 but it might be liveable, considering the alternative is unbearable
> in some situations.
>
> Can anyone else replicate my results?

Can you produce a patch against 7.4.1?   I'd like to test your fix against a 
real-world database.


-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-advocacy-owner@postgresql.org  Wed Apr 21 14:32:00 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A8F0CD1BABA
	for ;
	Wed, 21 Apr 2004 14:31:59 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 47883-10
	for ;
	Wed, 21 Apr 2004 14:31:58 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id E4C9AD1BA6C
	for ;
	Wed, 21 Apr 2004 14:31:56 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4915417 for pgsql-advocacy@postgresql.org;
	Wed, 21 Apr 2004 10:33:20 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
Subject: Fwd: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
Date: Wed, 21 Apr 2004 10:31:31 -0700
User-Agent: KMail/1.4.3
To: pgsql-advocacy@postgresql.org
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404211031.31007.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/111
X-Sequence-Number: 4083



----------  Forwarded Message  ----------

Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
Date: Wed, 21 Apr 2004 13:55:21 +0100
From: Paul Thomas 
To: Nick Barr 
Cc: "pgsql-performance @ postgresql . org" 

On 21/04/2004 09:31 Nick Barr wrote:
> Hi,
>
> Has anyone had a look at:
>
> http://people.ac.upc.es/zgomez/
>
> I realize that MySQL & PG cannot really be compared (especially when you
> consider the issues that MySQL has with things like data integrity) but
> still surely PG would perform better than the stats show (i.e. #7 31.28
> seconds versus 42 minutes!!!).

Looks like he's using the default postgresql.conf settings in which case
I'm not suprised at pg looking so slow. His stated use of foreign keys
invalidates the tests anyway as MyISAM tables don't support FKs so we're
probably seeing FK check overheads in pg that are simply ignore by MySQL.
In an honest test, MySQL should be reported as failing those tests.

Perhaps one of the advocay team will pick up the batton?

> On a side note it certainly looks like linux kernel 2.6 is quite a bit
> faster in comparision to 2.4.

Yes, I've seen other benchmarks which also show that.

--
Paul Thomas
+------------------------------+---------------------------------------------
+

| Thomas Micro Systems Limited | Software Solutions for

Business             |

| Computer Consultants         |

http://www.thomas-micro-systems-ltd.co.uk   |
+------------------------------+---------------------------------------------
+

---------------------------(end of broadcast)---------------------------
TIP 2: you can get off all lists at once with the unregister command
    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)

-------------------------------------------------------

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-advocacy-owner@postgresql.org  Wed Apr 21 14:47:37 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id D304BD1BA6C; Wed, 21 Apr 2004 14:47:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 55141-10; Wed, 21 Apr 2004 14:47:31 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP
	id C8F26D1B52F; Wed, 21 Apr 2004 14:47:29 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4915510; Wed, 21 Apr 2004 10:48:53 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: pgsql-advocacy@postgresql.org
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
Date: Wed, 21 Apr 2004 10:47:03 -0700
User-Agent: KMail/1.4.3
References: <200404211031.31007.josh@agliodbs.com>
In-Reply-To: <200404211031.31007.josh@agliodbs.com>
Cc: pgsql-performance@postgreSQL.org
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404211047.03808.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/112
X-Sequence-Number: 4084

Folks,

I've sent a polite e-mail to Mr. Gomez offering our help.  Please, nobody 
flame him!

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Wed Apr 21 15:19:59 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CBA9CD1CCD5
	for ;
	Wed, 21 Apr 2004 15:19:52 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 73995-03
	for ;
	Wed, 21 Apr 2004 15:19:47 -0300 (ADT)
Received: from monsoon.he.net (monsoon.he.net [64.62.221.2])
	by svr1.postgresql.org (Postfix) with SMTP id A1FC6D1D10E
	for ;
	Wed, 21 Apr 2004 15:19:45 -0300 (ADT)
Received: from [10.244.45.7] ([216.113.168.128]) by monsoon.he.net for
	; Wed, 21 Apr 2004 11:19:42 -0700
In-Reply-To: <1082515263.1558.214.camel@localhost.localdomain>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
Mime-Version: 1.0 (Apple Message framework v613)
Content-Type: text/plain; charset=US-ASCII; format=flowed
Message-Id: <78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
Content-Transfer-Encoding: 7bit
Cc: Anjan Dave , Josh Berkus ,
	Neil Conway ,
	=?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org, Tom Lane 
From: Paul Tuckfield 
Subject: Re: Wierd context-switching issue on Xeon
Date: Wed, 21 Apr 2004 11:19:46 -0700
To: pg@fastcrypt.com
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/357
X-Sequence-Number: 6657

Dave:

Why would test and set increase context swtches:
Note that it *does not increase* context swtiches when the two threads 
are on the two cores of a single Xeon processor. (use taskset to force 
affinity on linux)

Scenario:
If the two test and set processes are testing and setting the same bit 
as each other, then they'll see worst case cache coherency misses.  
They'll ping a cache line back and forth between CPUs.  Another case, 
might be that they're tesing and setting different bits or words, but 
those bits or words are always in the same cache line, again causing 
worst case cache coherency and misses.  The fact that tis doesn't 
happen when the threads are bound to the 2 cores of a single Xeon 
suggests it's because they're now sharing L1 cache. No pings/bounces.


I wonder do the threads stall so badly when pinging cache lines back 
and forth,  that the kernel sees it as an opportunity to put the 
process to sleep? or do these worst case misses cause an interrupt?

My question is:  What is it that the two threads waiting for when they 
spin? Is it exactly the same resource, or two resources that happen to 
have test-and-set flags in the same cache line?

On Apr 20, 2004, at 7:41 PM, Dave Cramer wrote:

> I modified the code in s_lock.c to remove the spins
>
> #define SPINS_PER_DELAY         1
>
> and it doesn't exhibit the behaviour
>
> This effectively changes the code to
>
>
> while(TAS(lock))
> 	select(10000); // 10ms
>
> Can anyone explain why executing TAS 100 times would increase context
> switches ?
>
> Dave
>
>
> On Tue, 2004-04-20 at 12:59, Josh Berkus wrote:
>> Anjan,
>>
>>> Quad 2.0GHz XEON with highest load we have seen on the applications, 
>>> DB
>>> performing great -
>>
>> Can you run Tom's test?   It takes a particular pattern of data 
>> access to
>> reproduce the issue.
> -- 
> Dave Cramer
> 519 939 0336
> ICQ # 14675561
>
>
> ---------------------------(end of 
> broadcast)---------------------------
> TIP 8: explain analyze is your friend
>


From pgsql-performance-owner@postgresql.org  Wed Apr 21 15:20:49 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9890FD1B967
	for ;
	Wed, 21 Apr 2004 15:20:38 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 74453-05
	for ;
	Wed, 21 Apr 2004 15:20:36 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id C4979D1D085
	for ;
	Wed, 21 Apr 2004 15:20:33 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3LIKP1v013306;
	Wed, 21 Apr 2004 14:20:25 -0400 (EDT)
To: Edoardo Ceccarelli 
Cc: Christopher Kings-Lynne ,
	Nick Barr , pgsql-performance@postgresql.org
Subject: Re: slow seqscan 
In-reply-to: <408641B7.5090808@axa.it> 
References: <40862003.2090201@axa.it> <4086272E.2080102@chuckie.co.uk>
	<40863228.907@axa.it> <408636A5.2050305@familyhealth.com.au>
	<408641B7.5090808@axa.it>
Comments: In-reply-to Edoardo Ceccarelli 
	message dated "Wed, 21 Apr 2004 11:41:11 +0200"
Date: Wed, 21 Apr 2004 14:20:25 -0400
Message-ID: <13305.1082571625@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/358
X-Sequence-Number: 6658

Edoardo Ceccarelli  writes:
> I wasn't able to make this 2 field index with lower:

> dba400=# CREATE INDEX annuncio400_rubric_testo_idx ON 
> annuncio400(rubric, LOWER(testo));
> ERROR:  parser: parse error at or near "(" at character 71

> seems impossible to creat 2 field indexes with lower function.

You need 7.4 to do that; previous releases don't support multi-column
functional indexes.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 21 15:51:43 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 44F0FD1D2AA
	for ;
	Wed, 21 Apr 2004 15:51:37 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 82348-09
	for ;
	Wed, 21 Apr 2004 15:51:36 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 53100D1D26C
	for ;
	Wed, 21 Apr 2004 15:51:34 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3LIpV4g014265;
	Wed, 21 Apr 2004 14:51:31 -0400 (EDT)
To: Paul Tuckfield 
Cc: pg@fastcrypt.com, Anjan Dave ,
	Josh Berkus , Neil Conway ,
	=?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon 
In-reply-to: <78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com> 
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
Comments: In-reply-to Paul Tuckfield 
	message dated "Wed, 21 Apr 2004 11:19:46 -0700"
Date: Wed, 21 Apr 2004 14:51:31 -0400
Message-ID: <14264.1082573491@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/359
X-Sequence-Number: 6659

Paul Tuckfield  writes:
> I wonder do the threads stall so badly when pinging cache lines back 
> and forth,  that the kernel sees it as an opportunity to put the 
> process to sleep? or do these worst case misses cause an interrupt?

No; AFAICS the kernel could not even be aware of that behavior.

The context swap storm is happening because of contention at the next
level up (LWLocks rather than spinlocks).  It could be an independent
issue that just happens to be triggered by the same sort of access
pattern.  I put forward a hypothesis that the cache miss storm caused by
the test-and-set ops induces the context swap storm by making the code
more likely to be executing in certain places at certain times ... but
it's only a hypothesis.

Yesterday evening I had pretty well convinced myself that they were
indeed independent issues: profiling on a single-CPU machine was telling
me that the test case I proposed spends over 10% of its time inside
ReadBuffer, which certainly seems like enough to explain a high rate of
contention on the BufMgrLock, without any assumptions about funny
behavior at the hardware level.  However, your report and Dave's suggest
that there really is some linkage.  So I'm still confused.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 21 15:57:30 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 77CA2D1D26A
	for ;
	Wed, 21 Apr 2004 15:57:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 83545-10
	for ;
	Wed, 21 Apr 2004 15:57:25 -0300 (ADT)
Received: from outbound.mailhop.org (outbound.mailhop.org [63.208.196.171])
	by svr1.postgresql.org (Postfix) with ESMTP id B08E9D1D263
	for ;
	Wed, 21 Apr 2004 15:57:23 -0300 (ADT)
Received: from ool-4353b51a.dyn.optonline.net ([67.83.181.26] helo=zeut.net)
	by outbound.mailhop.org with asmtp (Exim 4.20)
	id 1BGMuL-0006K9-K8; Wed, 21 Apr 2004 14:57:17 -0400
Message-ID: <4086C40C.5020703@zeut.net>
Date: Wed, 21 Apr 2004 14:57:16 -0400
From: "Matthew T. O'Connor" 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Paul Thomas 
Cc: Nick Barr , "pgsql-performance @ postgresql . org"
	
Subject: Re: MySQL vs PG TPC-H benchmarks
References: <4086316B.4050304@chuckie.co.uk> <20040421135521.A23015@bacon>
In-Reply-To: <20040421135521.A23015@bacon>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Mail-Handler: MailHop Outbound by DynDNS.org
X-Report-Abuse-To: abuse@dyndns.org
X-MHO-User: zeut
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=RCVD_IN_NJABL,
	RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/360
X-Sequence-Number: 6660

Paul Thomas wrote:

> Looks like he's using the default postgresql.conf settings in which 
> case I'm not suprised at pg looking so slow. His stated use of foreign 
> keys invalidates the tests anyway as MyISAM tables don't support FKs 
> so we're probably seeing FK check overheads in pg that are simply 
> ignore by MySQL. In an honest test, MySQL should be reported as 
> failing those tests.


Either failures, or they should not have been using MyISAM, they should 
have used the table format that supports FK's.  This is just not apples 
to apples.



From pgsql-performance-owner@postgresql.org  Wed Apr 21 16:09:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1270BD1D263
	for ;
	Wed, 21 Apr 2004 16:04:52 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 84383-10
	for ;
	Wed, 21 Apr 2004 16:04:52 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 210D1D1BABF
	for ;
	Wed, 21 Apr 2004 16:04:50 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3LJ4pxi014420;
	Wed, 21 Apr 2004 15:04:51 -0400 (EDT)
To: "Chris Hoover" 
Cc: pgsql-performance@postgresql.org
Subject: Re: Help understanding stat tables 
In-reply-to: <200404211224.35696.revoohc@sermonaudio.com> 
References: <200404211134.16876.revoohc@sermonaudio.com>
	<200404211224.35696.revoohc@sermonaudio.com>
Comments: In-reply-to "Chris Hoover" 
	message dated "Wed, 21 Apr 2004 12:24:35 -0400"
Date: Wed, 21 Apr 2004 15:04:50 -0400
Message-ID: <14419.1082574290@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/361
X-Sequence-Number: 6661

"Chris Hoover"  writes:
> I was taking heap_blks_hit / heap_blks_read for my hit pct.
> It should be heap_blks_hit/(heap_blks_read+heap_blks_hit), correct?

Right.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 21 16:13:54 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id AEE2ED1D290
	for ;
	Wed, 21 Apr 2004 16:10:42 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 93929-04
	for ;
	Wed, 21 Apr 2004 16:10:41 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 11398D1D238
	for ;
	Wed, 21 Apr 2004 16:10:39 -0300 (ADT)
Received: (qmail 28183 invoked from network); 21 Apr 2004 19:10:33 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 21 Apr 2004 19:10:33 -0000
Subject: Re: Wierd context-switching issue on Xeon
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Paul Tuckfield 
Cc: Anjan Dave , Josh Berkus ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org, Tom Lane 
In-Reply-To: <78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1082574808.1558.243.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Wed, 21 Apr 2004 15:13:28 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/362
X-Sequence-Number: 6662

FYI,

I am doing my testing on non hyperthreading dual athlons. 

Also, the test and set is attempting to set the same resource, and not
simply a bit. It's really an lock;xchg in assemblelr.

Also we are using the PAUSE mnemonic, so we should not be seeing any
cache coherency issues, as the cache is being taken out of the picture
AFAICS ?

Dave

On Wed, 2004-04-21 at 14:19, Paul Tuckfield wrote:
> Dave:
> 
> Why would test and set increase context swtches:
> Note that it *does not increase* context swtiches when the two threads 
> are on the two cores of a single Xeon processor. (use taskset to force 
> affinity on linux)
> 
> Scenario:
> If the two test and set processes are testing and setting the same bit 
> as each other, then they'll see worst case cache coherency misses.  
> They'll ping a cache line back and forth between CPUs.  Another case, 
> might be that they're tesing and setting different bits or words, but 
> those bits or words are always in the same cache line, again causing 
> worst case cache coherency and misses.  The fact that tis doesn't 
> happen when the threads are bound to the 2 cores of a single Xeon 
> suggests it's because they're now sharing L1 cache. No pings/bounces.
> 
> 
> I wonder do the threads stall so badly when pinging cache lines back 
> and forth,  that the kernel sees it as an opportunity to put the 
> process to sleep? or do these worst case misses cause an interrupt?
> 
> My question is:  What is it that the two threads waiting for when they 
> spin? Is it exactly the same resource, or two resources that happen to 
> have test-and-set flags in the same cache line?
> 
> On Apr 20, 2004, at 7:41 PM, Dave Cramer wrote:
> 
> > I modified the code in s_lock.c to remove the spins
> >
> > #define SPINS_PER_DELAY         1
> >
> > and it doesn't exhibit the behaviour
> >
> > This effectively changes the code to
> >
> >
> > while(TAS(lock))
> > 	select(10000); // 10ms
> >
> > Can anyone explain why executing TAS 100 times would increase context
> > switches ?
> >
> > Dave
> >
> >
> > On Tue, 2004-04-20 at 12:59, Josh Berkus wrote:
> >> Anjan,
> >>
> >>> Quad 2.0GHz XEON with highest load we have seen on the applications, 
> >>> DB
> >>> performing great -
> >>
> >> Can you run Tom's test?   It takes a particular pattern of data 
> >> access to
> >> reproduce the issue.
> > -- 
> > Dave Cramer
> > 519 939 0336
> > ICQ # 14675561
> >
> >
> > ---------------------------(end of 
> > broadcast)---------------------------
> > TIP 8: explain analyze is your friend
> >
> 
> 
> ---------------------------(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
> 
> 
> 
> !DSPAM:4086c4d0263544680737483!
> 
> 
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-advocacy-owner@postgresql.org  Wed Apr 21 16:39:05 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id AD29AD1DA3C
	for ;
	Wed, 21 Apr 2004 16:38:57 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 01561-04
	for ;
	Wed, 21 Apr 2004 16:38:57 -0300 (ADT)
Received: from smtp016.mail.yahoo.com (smtp016.mail.yahoo.com
	[216.136.174.113])
	by svr1.postgresql.org (Postfix) with SMTP id DD163D1D614
	for ;
	Wed, 21 Apr 2004 16:38:53 -0300 (ADT)
Received: from unknown (HELO europa.janwieck.net) (janwieck@68.80.245.191 with
	login)
	by smtp016.mail.yahoo.com with SMTP; 21 Apr 2004 19:38:54 -0000
Received: from Yahoo.com (ismtp.afilias.com [216.217.55.254]) (authenticated)
	by europa.janwieck.net (8.11.6/8.11.6) with ESMTP id i3LJcrj11150;
	Wed, 21 Apr 2004 15:38:53 -0400
Message-ID: <4086CDC7.5000204@Yahoo.com>
Date: Wed, 21 Apr 2004 15:38:47 -0400
From: Jan Wieck 
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.4) Gecko/20030624
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Josh Berkus 
Cc: pgsql-advocacy@postgresql.org, pgsql-performance@postgresql.org
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
References: <200404211031.31007.josh@agliodbs.com>
	<200404211047.03808.josh@agliodbs.com>
In-Reply-To: <200404211047.03808.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-Spam-Status: No, hits=2.6 tagged_above=0.0 required=5.0
	tests=RCVD_IN_DYNABLOCK, RCVD_IN_SORBS
X-Spam-Level: **
X-Archive-Number: 200404/114
X-Sequence-Number: 4086

Josh Berkus wrote:

> Folks,
> 
> I've sent a polite e-mail to Mr. Gomez offering our help.  Please, nobody 
> flame him!
> 

Please keep in mind that the entire test has, other than a similar 
database schema and query types maybe, nothing to do with a TPC-H. I 
don't see any kind of SUT. Foreign key support on the DB level is not 
required by any of the TPC benchmarks. But the System Under Test, which 
is the combination of middleware application and database together with 
all computers and network components these parts are running on, must 
implement all the required semantics, like ACID properties, referential 
integrity &c. One could implement a TPC-H with flat files, it's just a 
major pain in the middleware.

A proper TPC benchmark implementation would for example be a complete 
PHP+DB application, where the user interaction is done by an emulated 
"browser" and what is measured is the http response times, not anything 
going on between PHP and the DB. Assuming that all requirements of the 
TPC specification are implemented by either using available DB features, 
or including appropriate workarounds in the PHP code, that would very 
well lead to something that can compare PHP+MySQL vs. PHP+PostgreSQL.

All TPC benchmarks I have seen are performed by timing such a system 
after a considerable rampup time, giving the DB system a chance to 
properly populate caches and so forth. Rebooting the machine just before 
the test is the wrong thing here and will especially kill any advanced 
cache algorithms like ARC.


Jan

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck@Yahoo.com #


From pgsql-performance-owner@postgresql.org  Wed Apr 21 17:47:02 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 630E5D1DA3C
	for ;
	Wed, 21 Apr 2004 17:47:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 31545-05
	for ;
	Wed, 21 Apr 2004 17:46:59 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 72A8BD1D8BF
	for ;
	Wed, 21 Apr 2004 17:46:57 -0300 (ADT)
Received: (qmail 22072 invoked from network); 21 Apr 2004 20:46:51 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 21 Apr 2004 20:46:51 -0000
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Paul Tuckfield 
Cc: Anjan Dave , Josh Berkus ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org, Tom Lane 
In-Reply-To: <1082574808.1558.243.camel@localhost.localdomain>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
	<1082574808.1558.243.camel@localhost.localdomain>
Content-Type: multipart/mixed; boundary="=-sFlYaGnQ+RAFjH6Z5gg/"
Organization: Cramer Consulting
Message-Id: <1082580588.13371.260.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Wed, 21 Apr 2004 16:49:48 -0400
X-AntiVirus: scanned for viruses by AMaViS 0.2.1 (http://amavis.org/)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/364
X-Sequence-Number: 6664

--=-sFlYaGnQ+RAFjH6Z5gg/
Content-Type: text/plain
Content-Transfer-Encoding: 7bit

attached.
-- 
Dave Cramer
519 939 0336
ICQ # 14675561

--=-sFlYaGnQ+RAFjH6Z5gg/
Content-Disposition: attachment; filename=spinlock.patch
Content-Type: text/x-patch; name=spinlock.patch; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

Index: backend/storage/lmgr/s_lock.c
===================================================================
RCS file: /usr/local/cvs/pgsql-server/src/backend/storage/lmgr/s_lock.c,v
retrieving revision 1.16
diff -c -r1.16 s_lock.c
*** backend/storage/lmgr/s_lock.c	8 Aug 2003 21:42:00 -0000	1.16
--- backend/storage/lmgr/s_lock.c	21 Apr 2004 20:27:34 -0000
***************
*** 76,82 ****
  	 * The select() delays are measured in centiseconds (0.01 sec) because 10
  	 * msec is a common resolution limit at the OS level.
  	 */
! #define SPINS_PER_DELAY		100
  #define NUM_DELAYS			1000
  #define MIN_DELAY_CSEC		1
  #define MAX_DELAY_CSEC		100
--- 76,82 ----
  	 * The select() delays are measured in centiseconds (0.01 sec) because 10
  	 * msec is a common resolution limit at the OS level.
  	 */
! #define SPINS_PER_DELAY		10
  #define NUM_DELAYS			1000
  #define MIN_DELAY_CSEC		1
  #define MAX_DELAY_CSEC		100
***************
*** 88,93 ****
--- 88,94 ----
  
  	while (TAS(lock))
  	{
+ 		__asm__ __volatile__ ( "rep;nop": : :"memory");
  		if (++spins > SPINS_PER_DELAY)
  		{
  			if (++delays > NUM_DELAYS)
Index: include/storage/s_lock.h
===================================================================
RCS file: /usr/local/cvs/pgsql-server/src/include/storage/s_lock.h,v
retrieving revision 1.115.2.1
diff -c -r1.115.2.1 s_lock.h
*** include/storage/s_lock.h	4 Nov 2003 09:43:56 -0000	1.115.2.1
--- include/storage/s_lock.h	21 Apr 2004 20:26:25 -0000
***************
*** 103,110 ****
  	register slock_t _res = 1;
  
  	__asm__ __volatile__(
! 		"	lock			\n"
  		"	xchgb	%0,%1	\n"
  :		"=q"(_res), "=m"(*lock)
  :		"0"(_res));
  	return (int) _res;
--- 103,113 ----
  	register slock_t _res = 1;
  
  	__asm__ __volatile__(
! 		"       cmpb $0,%1      \n"
! 		"       jne     1f      \n"
! 		"	lock		\n"
  		"	xchgb	%0,%1	\n"
+ 		"       1:\n"
  :		"=q"(_res), "=m"(*lock)
  :		"0"(_res));
  	return (int) _res;

--=-sFlYaGnQ+RAFjH6Z5gg/--


From pgsql-performance-owner@postgresql.org  Wed Apr 21 18:50:22 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 15230D1D501
	for ;
	Wed, 21 Apr 2004 18:50:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 55255-03
	for ;
	Wed, 21 Apr 2004 18:50:14 -0300 (ADT)
Received: from jinx.internetstaff.com (jinx.internetstaff.com
 [63.214.174.240])
	by svr1.postgresql.org (Postfix) with ESMTP id 907C8D1D08B
	for ;
	Wed, 21 Apr 2004 18:50:11 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by jinx.internetstaff.com (Postfix) with ESMTP id 97A874BC0B6
	for ;
	Wed, 21 Apr 2004 14:50:13 -0700 (MST)
Received: from jinx.internetstaff.com ([127.0.0.1])
	by localhost (jinx [127.0.0.1]) (amavisd-new, port 10024) with LMTP
	id 05056-01-18 for ;
	Wed, 21 Apr 2004 14:50:13 -0700 (MST)
Received: from [10.10.4.32] (unknown [65.161.175.209])
	by jinx.internetstaff.com (Postfix) with ESMTP id 4937C4BC0B2
	for ;
	Wed, 21 Apr 2004 14:50:13 -0700 (MST)
Subject: Re: Moving postgres to FC disks
From: Cott Lang 
To: pgsql-performance@postgresql.org
In-Reply-To: <4085BFF0.4050604@commandprompt.com>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814D@vt-pe2550-001.vantage.vantage.com>
	<4085BFF0.4050604@commandprompt.com>
Content-Type: text/plain
Message-Id: <1082584212.2356.14.camel@blackbox.vcommerce.com>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 (1.4.6-1) 
Date: Wed, 21 Apr 2004 14:50:12 -0700
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at internetstaff.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/365
X-Sequence-Number: 6665

On Tue, 2004-04-20 at 17:27, Joshua D. Drake wrote:

> > -Currently, the internal RAID volume is ext3 filesystem. Any
> > recommendations for the filesystem on the new FC volume? Rieserfs?
> > 
> > 
> XFS

What Linux distributions are popular in here for PG+XFS?  

I'm very disappointed that Redhat Enterprise 3 doesn't appear to support
XFS/JFS, or anything else. Suse Server 8 seems very dated, at least from
the eval I downloaded. I'm curious as to where other people have gone
with the death of RH9. I'd have gone on to Redhat 3 if I wasn't
interested in getting some of the benefits of XFS at the same time ...




From pgsql-performance-owner@postgresql.org  Mon Apr 26 14:51:25 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 90717D1CCD5
	for ;
	Wed, 21 Apr 2004 19:02:35 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 54475-08
	for ;
	Wed, 21 Apr 2004 19:02:36 -0300 (ADT)
Received: from is.rice.edu (is.rice.edu [128.42.42.24])
	by svr1.postgresql.org (Postfix) with ESMTP id ADB76D1CCCA
	for ;
	Wed, 21 Apr 2004 19:02:33 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by localhost.is.rice.edu (Postfix) with ESMTP
	id 24C7B4190C; Wed, 21 Apr 2004 17:02:36 -0500 (CDT)
Received: from is.rice.edu ([127.0.0.1])
	by localhost (it.is.rice.edu [127.0.0.1]) (amavisd-new, port 10024)
	with ESMTP id 25811-10; Wed, 21 Apr 2004 17:02:32 -0500 (CDT)
Received: by is.rice.edu (Postfix, from userid 18612)
	id 4F31F4196A; Wed, 21 Apr 2004 17:02:32 -0500 (CDT)
Date: Wed, 21 Apr 2004 17:02:31 -0500
From: Kenneth Marshall 
To: Tom Lane 
Cc: pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon
Message-ID: <20040421220231.GA18010@it.is.rice.edu>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
	<14264.1082573491@sss.pgh.pa.us>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <14264.1082573491@sss.pgh.pa.us>
User-Agent: Mutt/1.4.2i
X-Virus-Scanned: by amavis-20030314-p2 at is.rice.edu
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/441
X-Sequence-Number: 6741

On Wed, Apr 21, 2004 at 02:51:31PM -0400, Tom Lane wrote:
> The context swap storm is happening because of contention at the next
> level up (LWLocks rather than spinlocks).  It could be an independent
> issue that just happens to be triggered by the same sort of access
> pattern.  I put forward a hypothesis that the cache miss storm caused by
> the test-and-set ops induces the context swap storm by making the code
> more likely to be executing in certain places at certain times ... but
> it's only a hypothesis.
> 
If the context swap storm derives from LWLock contention, maybe using
a random order to assign buffer locks in buf_init.c would prevent
simple adjacency of buffer allocation to cause the storm. Just offsetting
the assignment by the cacheline size should work. I notice that when
initializing the buffers in shared memory, both the buf->meta_data_lock
and the buf->cntx_lock are immediately adjacent in memory. I am not
familiar enough with the flow through postgres to see if there could
be "fighting" for those two locks. If so, offsetting those by the cache
line size would also stop the context swap storm.

--Ken

From pgsql-performance-owner@postgresql.org  Wed Apr 21 22:46:16 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6CF8CD1DB23
	for ;
	Wed, 21 Apr 2004 22:45:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 26062-08
	for ;
	Wed, 21 Apr 2004 22:45:55 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 6AE10D1DA3C
	for ;
	Wed, 21 Apr 2004 22:45:51 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3M1js0j017857;
	Wed, 21 Apr 2004 21:45:54 -0400 (EDT)
To: Kenneth Marshall 
Cc: pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon 
In-reply-to: <20040421220231.GA18010@it.is.rice.edu> 
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
	<14264.1082573491@sss.pgh.pa.us>
	<20040421220231.GA18010@it.is.rice.edu>
Comments: In-reply-to Kenneth Marshall 
	message dated "Wed, 21 Apr 2004 17:02:31 -0500"
Date: Wed, 21 Apr 2004 21:45:54 -0400
Message-ID: <17856.1082598354@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/366
X-Sequence-Number: 6666

Kenneth Marshall  writes:
> If the context swap storm derives from LWLock contention, maybe using
> a random order to assign buffer locks in buf_init.c would prevent
> simple adjacency of buffer allocation to cause the storm.

Good try, but no cigar ;-).  The test cases I've been looking at take
only shared locks on the per-buffer locks, so that's not where the
context swaps are coming from.  The swaps have to be caused by the
BufMgrLock, because that's the only exclusive lock being taken.

I did try increasing the allocated size of the spinlocks to 128 bytes
to see if it would do anything.  It didn't ...

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 21 23:36:17 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 38739D1D72D
	for ;
	Wed, 21 Apr 2004 23:35:39 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 38444-05
	for ;
	Wed, 21 Apr 2004 23:35:41 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id B7375D1D707
	for ;
	Wed, 21 Apr 2004 23:35:36 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3M2ZUBQ018323;
	Wed, 21 Apr 2004 22:35:30 -0400 (EDT)
To: pg@fastcrypt.com
Cc: Paul Tuckfield ,
	Anjan Dave , Josh Berkus ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1 
In-reply-to: <1082580588.13371.260.camel@localhost.localdomain> 
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
	<1082574808.1558.243.camel@localhost.localdomain>
	<1082580588.13371.260.camel@localhost.localdomain>
Comments: In-reply-to Dave Cramer 
	message dated "Wed, 21 Apr 2004 16:49:48 -0400"
Date: Wed, 21 Apr 2004 22:35:30 -0400
Message-ID: <18322.1082601330@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/367
X-Sequence-Number: 6667

Dave Cramer  writes:
> diff -c -r1.16 s_lock.c
> *** backend/storage/lmgr/s_lock.c	8 Aug 2003 21:42:00 -0000	1.16
> --- backend/storage/lmgr/s_lock.c	21 Apr 2004 20:27:34 -0000
> ***************
> *** 76,82 ****
>   	 * The select() delays are measured in centiseconds (0.01 sec) because 10
>   	 * msec is a common resolution limit at the OS level.
>   	 */
> ! #define SPINS_PER_DELAY		100
>   #define NUM_DELAYS			1000
>   #define MIN_DELAY_CSEC		1
>   #define MAX_DELAY_CSEC		100
> --- 76,82 ----
>   	 * The select() delays are measured in centiseconds (0.01 sec) because 10
>   	 * msec is a common resolution limit at the OS level.
>   	 */
> ! #define SPINS_PER_DELAY		10
>   #define NUM_DELAYS			1000
>   #define MIN_DELAY_CSEC		1
>   #define MAX_DELAY_CSEC		100


As far as I can tell, this does reduce the rate of semop's
significantly, but it does so by bringing the overall processing rate
to a crawl :-(.  I see 97% CPU idle time when using this patch.
I believe what is happening is that the select() delay in s_lock.c is
being hit frequently because the spin loop isn't allowed to run long
enough to let the other processor get out of the spinlock.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 21 23:56:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 8B6CED1D707
	for ;
	Wed, 21 Apr 2004 23:53:55 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 46573-03
	for ;
	Wed, 21 Apr 2004 23:53:55 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 7A9BCD1CCB1
	for ;
	Wed, 21 Apr 2004 23:53:50 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4918812; Wed, 21 Apr 2004 19:55:20 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: Tom Lane , pg@fastcrypt.com
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Wed, 21 Apr 2004 19:53:24 -0700
User-Agent: KMail/1.4.3
Cc: Paul Tuckfield ,
	Anjan Dave , Neil Conway ,
	Dirk =?iso-8859-1?q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<1082580588.13371.260.camel@localhost.localdomain>
	<18322.1082601330@sss.pgh.pa.us>
In-Reply-To: <18322.1082601330@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404211953.24898.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/368
X-Sequence-Number: 6668

Tom,

> As far as I can tell, this does reduce the rate of semop's
> significantly, but it does so by bringing the overall processing rate
> to a crawl :-(.  I see 97% CPU idle time when using this patch.
> I believe what is happening is that the select() delay in s_lock.c is
> being hit frequently because the spin loop isn't allowed to run long
> enough to let the other processor get out of the spinlock.

Also, I tested it on production data, and it reduces the CSes by about 40%.  
An improvement, but not a magic bullet.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Thu Apr 22 03:44:56 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 2586BD1DB96
	for ;
	Thu, 22 Apr 2004 03:04:53 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 94654-09
	for ;
	Thu, 22 Apr 2004 03:04:43 -0300 (ADT)
Received: from www.postgresql.com (www.postgresql.com [200.46.204.209])
	by svr1.postgresql.org (Postfix) with ESMTP id 15BD5D1EB82
	for ;
	Thu, 22 Apr 2004 03:02:32 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by www.postgresql.com (Postfix) with SMTP id BAD7DCF64A3
	for ;
	Thu, 22 Apr 2004 00:03:49 -0300 (ADT)
Received: (qmail 9328 invoked from network); 22 Apr 2004 03:03:41 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 22 Apr 2004 03:03:41 -0000
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Tom Lane 
Cc: Paul Tuckfield ,
	Anjan Dave , Josh Berkus ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
In-Reply-To: <18322.1082601330@sss.pgh.pa.us>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
	<1082574808.1558.243.camel@localhost.localdomain>
	<1082580588.13371.260.camel@localhost.localdomain>
	<18322.1082601330@sss.pgh.pa.us>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1082603201.1556.272.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Wed, 21 Apr 2004 23:06:41 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/375
X-Sequence-Number: 6675

Yeah, I did some more testing myself, and actually get better numbers
with increasing spins per delay to 1000, but my suspicion is that it is
highly dependent on finding the right delay for the processor you are
on.

My hypothesis is that if you spin approximately the same or more time
than the average time it takes to get finished with the shared resource
then this should reduce cs.

Certainly more ideas are required here.

Dave 
On Wed, 2004-04-21 at 22:35, Tom Lane wrote:
> Dave Cramer  writes:
> > diff -c -r1.16 s_lock.c
> > *** backend/storage/lmgr/s_lock.c	8 Aug 2003 21:42:00 -0000	1.16
> > --- backend/storage/lmgr/s_lock.c	21 Apr 2004 20:27:34 -0000
> > ***************
> > *** 76,82 ****
> >   	 * The select() delays are measured in centiseconds (0.01 sec) because 10
> >   	 * msec is a common resolution limit at the OS level.
> >   	 */
> > ! #define SPINS_PER_DELAY		100
> >   #define NUM_DELAYS			1000
> >   #define MIN_DELAY_CSEC		1
> >   #define MAX_DELAY_CSEC		100
> > --- 76,82 ----
> >   	 * The select() delays are measured in centiseconds (0.01 sec) because 10
> >   	 * msec is a common resolution limit at the OS level.
> >   	 */
> > ! #define SPINS_PER_DELAY		10
> >   #define NUM_DELAYS			1000
> >   #define MIN_DELAY_CSEC		1
> >   #define MAX_DELAY_CSEC		100
> 
> 
> As far as I can tell, this does reduce the rate of semop's
> significantly, but it does so by bringing the overall processing rate
> to a crawl :-(.  I see 97% CPU idle time when using this patch.
> I believe what is happening is that the select() delay in s_lock.c is
> being hit frequently because the spin loop isn't allowed to run long
> enough to let the other processor get out of the spinlock.
> 
> 			regards, tom lane
> 
> 
> 
> !DSPAM:40872f7e21492906114513!
> 
> 
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Thu Apr 22 03:04:11 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 83028D1DD99
	for ;
	Thu, 22 Apr 2004 03:03:49 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 94654-05
	for ;
	Thu, 22 Apr 2004 03:03:43 -0300 (ADT)
Received: from www.postgresql.com (www.postgresql.com [200.46.204.209])
	by svr1.postgresql.org (Postfix) with ESMTP id 4FD2BD1EAAC
	for ;
	Thu, 22 Apr 2004 03:01:56 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by www.postgresql.com (Postfix) with ESMTP id 98C40CF64A0
	for ;
	Thu, 22 Apr 2004 00:10:44 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3M3Ah8A018621;
	Wed, 21 Apr 2004 23:10:43 -0400 (EDT)
To: Paul Tuckfield 
Cc: pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon 
In-reply-to: <9D274446-932B-11D8-BA67-000393BD6C3E@tuckfield.com> 
References: 
	<40843E63.7050101@joeconway.com>
	<200404191455.04068.josh@agliodbs.com>
	<1407.1082419316@sss.pgh.pa.us> <1931.1082422389@sss.pgh.pa.us>
	
	<9D274446-932B-11D8-BA67-000393BD6C3E@tuckfield.com>
Comments: In-reply-to Paul Tuckfield 
	message dated "Tue, 20 Apr 2004 17:34:13 -0700"
Date: Wed, 21 Apr 2004 23:10:43 -0400
Message-ID: <18620.1082603443@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/374
X-Sequence-Number: 6674

Paul Tuckfield  writes:
>> I used the taskset command:
>> taskset 01 -p 
>> taskset 01 -p 
>> 
>> I guess that 0 and 1 are the two cores (pipelines? hyper-threads?) on 
>> the first Xeon processor in the box.

AFAICT, what you've actually done here is to bind both backends to the
first logical processor of the first Xeon.  If you'd used 01 and 02
as the affinity masks then you'd have bound them to the two cores of
that Xeon, but what you actually did simply reduces the system to a
uniprocessor.  In that situation the context swap rate will be normally
one swap per scheduler timeslice, and at worst two swaps per timeslice
(if a process is swapped away from while it holds a lock the other one
wants).  It doesn't prove a lot about our SMP problem though.

I don't have access to a Xeon with both taskset and hyperthreading
enabled, so I can't check what happens when you do the taskset correctly
... could you retry?

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 22 04:17:06 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CDF4FD1EAC3
	for ;
	Thu, 22 Apr 2004 03:01:58 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 86178-07
	for ;
	Thu, 22 Apr 2004 03:01:44 -0300 (ADT)
Received: from www.postgresql.com (www.postgresql.com [200.46.204.209])
	by svr1.postgresql.org (Postfix) with ESMTP id BEC5AD1E0F1
	for ;
	Thu, 22 Apr 2004 03:01:32 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by www.postgresql.com (Postfix) with SMTP id 1CC98CF6671
	for ;
	Thu, 22 Apr 2004 00:15:53 -0300 (ADT)
Received: (qmail 11674 invoked from network); 22 Apr 2004 03:15:46 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 22 Apr 2004 03:15:46 -0000
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Josh Berkus 
Cc: Tom Lane ,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
In-Reply-To: <200404211953.24898.josh@agliodbs.com>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<1082580588.13371.260.camel@localhost.localdomain>
	<18322.1082601330@sss.pgh.pa.us> <200404211953.24898.josh@agliodbs.com>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1082603927.1558.279.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Wed, 21 Apr 2004 23:18:47 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/376
X-Sequence-Number: 6676

More data....

On a dual xeon with HTT enabled:

I tried increasing the NUM_SPINS to 1000 and it works better.

NUM_SPINLOCKS	CS	ID	pgbench

100		250K	59%	230 TPS
1000		125K	55%	228 TPS

This is certainly heading in the right direction ? Although it looks
like it is highly dependent on the system you are running on.

--dc--	 



On Wed, 2004-04-21 at 22:53, Josh Berkus wrote:
> Tom,
> 
> > As far as I can tell, this does reduce the rate of semop's
> > significantly, but it does so by bringing the overall processing rate
> > to a crawl :-(.  I see 97% CPU idle time when using this patch.
> > I believe what is happening is that the select() delay in s_lock.c is
> > being hit frequently because the spin loop isn't allowed to run long
> > enough to let the other processor get out of the spinlock.
> 
> Also, I tested it on production data, and it reduces the CSes by about 40%.  
> An improvement, but not a magic bullet.
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Thu Apr 22 03:03:36 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DAF42D1DF80
	for ;
	Thu, 22 Apr 2004 03:03:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 94710-04
	for ;
	Thu, 22 Apr 2004 03:03:10 -0300 (ADT)
Received: from www.postgresql.com (www.postgresql.com [200.46.204.209])
	by svr1.postgresql.org (Postfix) with ESMTP id CA090D1E9E9
	for ;
	Thu, 22 Apr 2004 03:01:48 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by www.postgresql.com (Postfix) with ESMTP id 67520CF6879
	for ;
	Thu, 22 Apr 2004 01:15:15 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3M4FDVO019516;
	Thu, 22 Apr 2004 00:15:13 -0400 (EDT)
To: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= 
Cc: pgsql-performance@postgresql.org
Subject: Re: 225 times slower 
In-reply-to:  
References: 
	
Comments: In-reply-to =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?=
	
	message dated "Tue, 20 Apr 2004 19:10:50 +0200"
Date: Thu, 22 Apr 2004 00:15:13 -0400
Message-ID: <19515.1082607313@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/373
X-Sequence-Number: 6673

=?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?=  writes:
> dps=# explain analyze SELECT rec_id FROM url WHERE crc32!=0 AND  
> crc32=419903683 AND status IN (200,304,206) ORDER BY rec_id LIMIT 1;
>                                                              QUERY PLAN
> ------------------------------------------------------------------------ 
> ----------------------------------------------------------
>   Limit  (cost=0.00..27.95 rows=1 width=4) (actual  
> time=11021.875..11021.876 rows=1 loops=1)
>     ->  Index Scan using url_pkey on url  (cost=0.00..11625.49 rows=416  
> width=4) (actual time=11021.868..11021.868 rows=1 loops=1)
>           Filter: ((crc32 <> 0) AND (crc32 = 419903683) AND ((status =  
> 200) OR (status = 304) OR (status = 206)))
>   Total runtime: 11021.986 ms
> (4 rows)
> dps=# \q

The planner is guessing that scanning in rec_id order will produce a
matching row fairly quickly (sooner than selecting all the matching rows
and sorting them would do).  It's wrong in this case, but I'm not sure
it could do better without very detailed cross-column statistics.  Am I
right to guess that the rows that match the WHERE clause are not evenly
distributed in the rec_id order, but rather there are no such rows till
you get well up in the ordering?

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 22 01:24:14 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1C66AD1BB36
	for ;
	Thu, 22 Apr 2004 01:23:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 46049-08
	for ;
	Thu, 22 Apr 2004 01:23:30 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 5F84BD1BACC
	for ;
	Thu, 22 Apr 2004 01:23:29 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3M4NPoK019597;
	Thu, 22 Apr 2004 00:23:25 -0400 (EDT)
To: pg@fastcrypt.com
Cc: Josh Berkus ,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1 
In-reply-to: <1082603927.1558.279.camel@localhost.localdomain> 
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<1082580588.13371.260.camel@localhost.localdomain>
	<18322.1082601330@sss.pgh.pa.us>
	<200404211953.24898.josh@agliodbs.com>
	<1082603927.1558.279.camel@localhost.localdomain>
Comments: In-reply-to Dave Cramer 
	message dated "Wed, 21 Apr 2004 23:18:47 -0400"
Date: Thu, 22 Apr 2004 00:23:24 -0400
Message-ID: <19596.1082607804@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/369
X-Sequence-Number: 6669

Dave Cramer  writes:
> I tried increasing the NUM_SPINS to 1000 and it works better.

Doesn't surprise me.  The value of 100 is about right on the assumption
that the spinlock instruction per se is not too much more expensive than
any other instruction.  What I was seeing from oprofile suggested that
the spinlock instruction cost about 100x more than an ordinary
instruction :-( ... so maybe 200 or so would be good on a Xeon.

> This is certainly heading in the right direction ? Although it looks
> like it is highly dependent on the system you are running on.

Yeah.  I don't know a reasonable way to tune this number automatically
for particular systems ... but at the very least we'd need to find a way
to distinguish uniprocessor from multiprocessor, because on a
uniprocessor the optimal value is surely 1.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 22 01:37:36 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B24D0D1D0B8
	for ;
	Thu, 22 Apr 2004 01:37:34 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 67990-05
	for ;
	Thu, 22 Apr 2004 01:37:33 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP id D0F6AD1BCC2
	for ;
	Thu, 22 Apr 2004 01:37:31 -0300 (ADT)
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
	i3M4bFWL087263; Thu, 22 Apr 2004 12:37:15 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <40874D97.5030401@familyhealth.com.au>
Date: Thu, 22 Apr 2004 12:44:07 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Tom Lane 
Cc: pg@fastcrypt.com, Josh Berkus ,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	=?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<1082580588.13371.260.camel@localhost.localdomain>
	<18322.1082601330@sss.pgh.pa.us>
	<200404211953.24898.josh@agliodbs.com>
	<1082603927.1558.279.camel@localhost.localdomain>
	<19596.1082607804@sss.pgh.pa.us>
In-Reply-To: <19596.1082607804@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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/370
X-Sequence-Number: 6670

> Yeah.  I don't know a reasonable way to tune this number automatically
> for particular systems ... but at the very least we'd need to find a way
> to distinguish uniprocessor from multiprocessor, because on a
> uniprocessor the optimal value is surely 1.

 From TODO:

* Add code to detect an SMP machine and handle spinlocks accordingly 
from distributted.net, http://www1.distributed.net/source, in 
client/common/cpucheck.cpp

Chris


From pgsql-performance-owner@postgresql.org  Thu Apr 22 01:56:34 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B907AD1D096
	for ;
	Thu, 22 Apr 2004 01:56:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 74325-04
	for ;
	Thu, 22 Apr 2004 01:56:31 -0300 (ADT)
Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251])
	by svr1.postgresql.org (Postfix) with ESMTP id 5FCE3D1D0B8
	for ;
	Thu, 22 Apr 2004 01:56:29 -0300 (ADT)
Received: (from pgman@localhost)
	by candle.pha.pa.us (8.11.6/8.11.6) id i3M4uFv04710;
	Thu, 22 Apr 2004 00:56:15 -0400 (EDT)
From: Bruce Momjian 
Message-Id: <200404220456.i3M4uFv04710@candle.pha.pa.us>
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
In-Reply-To: <19596.1082607804@sss.pgh.pa.us>
To: Tom Lane 
Date: Thu, 22 Apr 2004 00:56:15 -0400 (EDT)
Cc: pg@fastcrypt.com, Josh Berkus ,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	=?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/371
X-Sequence-Number: 6671

Tom Lane wrote:
> Dave Cramer  writes:
> > I tried increasing the NUM_SPINS to 1000 and it works better.
> 
> Doesn't surprise me.  The value of 100 is about right on the assumption
> that the spinlock instruction per se is not too much more expensive than
> any other instruction.  What I was seeing from oprofile suggested that
> the spinlock instruction cost about 100x more than an ordinary
> instruction :-( ... so maybe 200 or so would be good on a Xeon.
> 
> > This is certainly heading in the right direction ? Although it looks
> > like it is highly dependent on the system you are running on.
> 
> Yeah.  I don't know a reasonable way to tune this number automatically
> for particular systems ... but at the very least we'd need to find a way
> to distinguish uniprocessor from multiprocessor, because on a
> uniprocessor the optimal value is surely 1.

Have you looked at the code pointed to by our TODO item:
	
	* Add code to detect an SMP machine and handle spinlocks accordingly
	  from distributted.net, http://www1.distributed.net/source,
	  in client/common/cpucheck.cpp

For BSDOS it has:

    #if (CLIENT_OS == OS_FREEBSD) || (CLIENT_OS == OS_BSDOS) || \
        (CLIENT_OS == OS_OPENBSD) || (CLIENT_OS == OS_NETBSD)
    { /* comment out if inappropriate for your *bsd - cyp (25/may/1999) */
      int ncpus; size_t len = sizeof(ncpus);
      int mib[2]; mib[0] = CTL_HW; mib[1] = HW_NCPU;
      if (sysctl( &mib[0], 2, &ncpus, &len, NULL, 0 ) == 0)
      //if (sysctlbyname("hw.ncpu", &ncpus, &len, NULL, 0 ) == 0)
        cpucount = ncpus;
    }

and I can confirm that on my computer it works:

	hw.ncpu = 2

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

From pgsql-performance-owner@postgresql.org  Thu Apr 22 02:13:59 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CEE5AD1B44F
	for ;
	Thu, 22 Apr 2004 02:13:57 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 75956-09
	for ;
	Thu, 22 Apr 2004 02:13:57 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 0F88AD1B441
	for ;
	Thu, 22 Apr 2004 02:13:56 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3M5Dl3J020172;
	Thu, 22 Apr 2004 01:13:47 -0400 (EDT)
To: Bruce Momjian 
Cc: pg@fastcrypt.com, Josh Berkus ,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	=?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1 
In-reply-to: <200404220456.i3M4uFv04710@candle.pha.pa.us> 
References: <200404220456.i3M4uFv04710@candle.pha.pa.us>
Comments: In-reply-to Bruce Momjian 
	message dated "Thu, 22 Apr 2004 00:56:15 -0400"
Date: Thu, 22 Apr 2004 01:13:47 -0400
Message-ID: <20171.1082610827@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/372
X-Sequence-Number: 6672

Bruce Momjian  writes:
> For BSDOS it has:

>     #if (CLIENT_OS == OS_FREEBSD) || (CLIENT_OS == OS_BSDOS) || \
>         (CLIENT_OS == OS_OPENBSD) || (CLIENT_OS == OS_NETBSD)
>     { /* comment out if inappropriate for your *bsd - cyp (25/may/1999) */
>       int ncpus; size_t len = sizeof(ncpus);
>       int mib[2]; mib[0] = CTL_HW; mib[1] = HW_NCPU;
>       if (sysctl( &mib[0], 2, &ncpus, &len, NULL, 0 ) == 0)
>       //if (sysctlbyname("hw.ncpu", &ncpus, &len, NULL, 0 ) == 0)
>         cpucount = ncpus;
>     }

Multiplied by how many platforms?  Ewww...

I was wondering about some sort of dynamic adaptation, roughly along the
lines of "whenever a spin loop successfully gets the lock after
spinning, decrease the allowed loop count by one; whenever we fail to
get the lock after spinning, increase by 100; if the loop count reaches,
say, 10000, decide we are on a uniprocessor and irreversibly set it to
1."  As written this would tend to incur a select() delay once per
hundred spinlock acquisitions, which is way too much, but I think we
could make it work with a sufficiently slow adaptation rate.  The tricky
part is that a slow adaptation rate means we can't have every backend
figuring this out for itself --- the right value would have to be
maintained globally, and I'm not sure how to do that without adding a
lot of overhead.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Mon Apr 26 14:55:13 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id EDFD9D1C4EC
	for ;
	Sat, 24 Apr 2004 01:32:23 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 27329-04
	for ;
	Sat, 24 Apr 2004 01:32:22 -0300 (ADT)
Received: from ganymede.hub.org (u46n208.hfx.eastlink.ca [24.222.46.208])
	by svr1.postgresql.org (Postfix) with ESMTP id C4428D1B43A
	for ;
	Sat, 24 Apr 2004 01:32:21 -0300 (ADT)
Received: by ganymede.hub.org (Postfix, from userid 1000)
	id 8068434356; Sat, 24 Apr 2004 01:32:21 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id 7F68E34323
	for ;
	Sat, 24 Apr 2004 01:32:21 -0300 (ADT)
X-Received: from localhost ([unix socket])
	by ganymede.hub.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 08:05:48 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id 3B82836B79
	for ; Thu, 22 Apr 2004 08:05:41 -0300 (ADT)
X-Received: from mail.postgresql.org [200.46.204.71]
	by localhost with IMAP (fetchmail-6.2.5)
	for scrappy@localhost (single-drop);
	Thu, 22 Apr 2004 08:05:41 -0300 (ADT)
X-Received: from svr1.postgresql.org ([unix socket])
	by svr1.postgresql.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 08:02:15 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B70F1D1DFA5
	for ; Thu, 22 Apr 2004 08:02:15 -0300 (ADT)
X-Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 98010-06 for ;
	Thu, 22 Apr 2004 08:02:15 -0300 (ADT)
X-Received: from minnie.omroep.nl (minnie.omroep.nl [145.58.30.4])
	by svr1.postgresql.org (Postfix) with ESMTP id 86711D1E0FE
	for ; Thu, 22 Apr 2004 08:02:12 -0300 (ADT)
X-Received: from tron.omroep.nl (tron.omroep.nl [145.58.31.20])
	by minnie.omroep.nl (Postfix) with ESMTP
	id DA6601CF5524; Thu, 22 Apr 2004 13:02:14 +0200 (CDT)
X-Received: from michiel.omroep.nl (michiel.omroep.nl [145.58.67.10])
	by tron.omroep.nl (Postfix) with ESMTP
	id 926C710056BD; Thu, 22 Apr 2004 13:02:14 +0200 (CEST)
X-Received: from michiel by michiel.omroep.nl with local (Exim 3.35 #1
	(Debian)) id 1BGbyA-0006Vo-00; Thu, 22 Apr 2004 13:02:14 +0200
Date: Thu, 22 Apr 2004 13:02:14 +0200
From: Michiel Meeuwissen 
To: Rod Taylor 
Cc: performance@postgresql.org
Subject: Re: order by index, and inheritance
Message-ID: <20040422110214.GH6870@michiel.omroep.nl>
Reply-To: Michiel Meeuwissen 
References: <20040419113013.GA32186@michiel.omroep.nl>
	<1082455997.57436.10.camel@jester>
	<20040420104518.GF1618@michiel.omroep.nl>
	<1082461767.57436.93.camel@jester>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1082461767.57436.93.camel@jester>
Organization: Publieke Omroep []()
User-Agent: Mutt/1.5.5.1+cvs20040105i
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-DCC: : 
X-Spam-Pyzor: 
ReSent-Date: Sat, 24 Apr 2004 01:32:14 -0300 (ADT)
Resent-From: "Marc G. Fournier" 
Resent-To: pgsql-performance@postgresql.org
ReSent-Subject: Re: [PERFORM] order by index, and inheritance
ReSent-Message-ID: <20040424013214.A42925@ganymede.hub.org>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/444
X-Sequence-Number: 6744

Rod Taylor  wrote:
> The scan is picking the best method for grabbing everything within the
> table, since it is not aware that we do not require everything.

Hmm. That is a bit silly. Why does it use the index if select only from
mm_mediasources?

> You can explicitly tell it what you want to do via:
> 
> SELECT *
>   FROM (SELECT * FROM mm_mediasources ORDER BY number DESC LIMIT 20 
>   UNION SELECT * FROM  ORDER BY number DESC LIMIT 20) AS tab
>  ORDER BY number DESC LIMIT 20

I think you meant 'only mm_mediasources', and btw order by and limit are not
accepted before union, so the above query does not compile.

I can't figure out any acceptable work-around. Even if something as the
above would work, it still would be hardly elegant, and you can as well have
no support for inheritance (actually, you can _better_ have no inheritance,
because at least it is clear what works then).

Michiel

btw. Why are these messages not appearing on the list itself?

-- 
Michiel Meeuwissen
Mediapark C101 Hilversum
+31 (0)35 6772979
nl_NL eo_XX en_US
mihxil'
 [] ()

From pgsql-performance-owner@postgresql.org  Thu Apr 22 09:37:05 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 623EED1B560
	for ;
	Thu, 22 Apr 2004 09:36:34 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 31012-05
	for ;
	Thu, 22 Apr 2004 09:36:34 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id EA8A2D1B43A
	for ;
	Thu, 22 Apr 2004 09:36:30 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3MCaTOc025059;
	Thu, 22 Apr 2004 08:36:29 -0400 (EDT)
To: pg@fastcrypt.com
Cc: Paul Tuckfield ,
	Anjan Dave , Josh Berkus ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1 
In-reply-to: <1082603201.1556.272.camel@localhost.localdomain> 
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404200959.52813.josh@agliodbs.com>
	<1082515263.1558.214.camel@localhost.localdomain>
	<78427208-93C0-11D8-BA67-000393BD6C3E@tuckfield.com>
	<1082574808.1558.243.camel@localhost.localdomain>
	<1082580588.13371.260.camel@localhost.localdomain>
	<18322.1082601330@sss.pgh.pa.us>
	<1082603201.1556.272.camel@localhost.localdomain>
Comments: In-reply-to Dave Cramer 
	message dated "Wed, 21 Apr 2004 23:06:41 -0400"
Date: Thu, 22 Apr 2004 08:36:28 -0400
Message-ID: <25058.1082637388@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/377
X-Sequence-Number: 6677

Dave Cramer  writes:
> My hypothesis is that if you spin approximately the same or more time
> than the average time it takes to get finished with the shared resource
> then this should reduce cs.

The only thing we use spinlocks for nowadays is to protect LWLocks, so
the "average time" involved is fairly small and stable --- or at least
that was the design intention.  What we seem to be seeing is that on SMP
machines, cache coherency issues cause the TAS step itself to be
expensive and variable.  However, in the experiments I did, strace'ing
showed that actual spin timeouts (manifested by the execution of a
delaying select()) weren't actually that common; the big source of
context switches is semop(), which indicates contention at the LWLock
level rather than the spinlock level.  So while tuning the spinlock
limit count might be a useful thing to do in general, I think it will
have only negligible impact on the particular problems we're discussing
in this thread.

			regards, tom lane

From pgsql-advocacy-owner@postgresql.org  Mon Apr 26 11:21:09 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 5B0E9D1DC96
	for ;
	Thu, 22 Apr 2004 09:53:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 39106-03
	for ;
	Thu, 22 Apr 2004 09:53:19 -0300 (ADT)
Received: from web60601.mail.yahoo.com (web60601.mail.yahoo.com
	[216.109.118.221])
	by svr1.postgresql.org (Postfix) with SMTP id 05A5FD1D079
	for ;
	Thu, 22 Apr 2004 09:53:15 -0300 (ADT)
Message-ID: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
Received: from [200.17.210.82] by web60601.mail.yahoo.com via HTTP;
	Thu, 22 Apr 2004 05:53:18 PDT
Date: Thu, 22 Apr 2004 05:53:18 -0700 (PDT)
From: Eduardo Almeida 
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
To: Jan Wieck , Josh Berkus 
Cc: pgsql-advocacy@postgresql.org, pgsql-performance@postgresql.org
In-Reply-To: <4086CDC7.5000204@Yahoo.com>
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/209
X-Sequence-Number: 4181

Folks,

I�m doing the 100GB TPC-H and I�ll show the previous
results to our community (Postgres) in 3 weeks before
finishing the study.

My intention is to carry through a test with a VLDB in
a low cost platform (PostgreSQL, Linux and cheap HW)
and not to compare with another DBMS.

So far I can tell you that the load time on PG 7.4.2
with kernel 2.6.5 on Opteron 64 model 240 in RAID 0
with 8 disks (960 GB) loaded the database in less than
24 hours. 
About 7hs:30min to load the data and 16:09:25 to
create the indexes

The Power test still running and that�s why I�ll not
present anything so far. Now I�ll just send to the
list my environment configuration.

- The configuration of the machine is:
Dual opteron 64 bits model 240
4GB RAM
960 GB on RAID 0
Mandrake Linux 64 with Kernel 2.6.5  (I compiled a
kernel for this test)
Java SDK  java version "1.4.2_04"
PostgreSQL JDBC pg74.1jdbc3.jar

- The TPC-H configuration is:
TPC-H 2.0.0
100GB
load using flat files
Refresh functions using java

- The PostgreSQL 7.4.2 configuration is:

add_missing_from               | on
 australian_timezones           | off
 authentication_timeout         | 60
 check_function_bodies          | on
 checkpoint_segments            | 128
 checkpoint_timeout             | 300
 checkpoint_warning             | 30
 client_encoding                | SQL_ASCII
 client_min_messages            | notice
 commit_delay                   | 0
 commit_siblings                | 5
 cpu_index_tuple_cost           | 0.001
 cpu_operator_cost              | 0.0025
 cpu_tuple_cost                 | 0.01
 DateStyle                      | ISO, MDY
 db_user_namespace              | off
 deadlock_timeout               | 1000
 debug_pretty_print             | off
 debug_print_parse              | off
 debug_print_plan               | off
 debug_print_rewritten          | off
 default_statistics_target      | 10
 default_transaction_isolation  | read committed
 default_transaction_read_only  | off
 dynamic_library_path           | $libdir
 effective_cache_size           | 150000
 enable_hashagg                 | on
 enable_hashjoin                | on
 enable_indexscan               | on
 enable_mergejoin               | on
 enable_nestloop                | on
 enable_seqscan                 | on
 enable_sort                    | on
 enable_tidscan                 | on
 explain_pretty_print           | on
 extra_float_digits             | 0
 from_collapse_limit            | 8
 fsync                          | off
 geqo                           | on
 geqo_effort                    | 1
 geqo_generations               | 0
geqo_pool_size                 | 0
 geqo_selection_bias            | 2
 geqo_threshold                 | 11
 join_collapse_limit            | 8
 krb_server_keyfile             | unset
 lc_collate                     | en_US
 lc_ctype                       | en_US
 lc_messages                    | C
 lc_monetary                    | C
 lc_numeric                     | C
 lc_time                        | C
 log_connections                | off
 log_duration                   | off
 log_error_verbosity            | default
 log_executor_stats             | off
 log_hostname                   | off
 log_min_duration_statement     | -1
 log_min_error_statement        | panic
 log_min_messages               | notice
 log_parser_stats               | off
 log_pid                        | off
 log_planner_stats              | off
 log_source_port                | off
 log_statement                  | off
 log_statement_stats            | off
 log_timestamp                  | off
 max_connections                | 10
 max_expr_depth                 | 10000
 max_files_per_process          | 1000
 max_fsm_pages                  | 20000
 max_fsm_relations              | 1000
 max_locks_per_transaction      | 64
 password_encryption            | on
 port                           | 5432
 pre_auth_delay                 | 0
 preload_libraries              | unset
 random_page_cost               | 1.25
 regex_flavor                   | advanced
 rendezvous_name                | unset
 search_path                    | $user,public
 server_encoding                | SQL_ASCII
 server_version                 | 7.4.2
 shared_buffers                 | 40000
 silent_mode                    | off
sort_mem                       | 65536
 sql_inheritance                | on
 ssl                            | off
 statement_timeout              | 10000000
 stats_block_level              | off
 stats_command_string           | off
 stats_reset_on_server_start    | on
 stats_row_level                | off
 stats_start_collector          | on
 superuser_reserved_connections | 2
 syslog                         | 0
 syslog_facility                | LOCAL0
 syslog_ident                   | postgres
 tcpip_socket                   | on
 TimeZone                       | unknown
 trace_notify                   | off
 transaction_isolation          | read committed
 transaction_read_only          | off
 transform_null_equals          | off
 unix_socket_directory          | unset
 unix_socket_group              | unset
 unix_socket_permissions        | 511
 vacuum_mem                     | 65536
 virtual_host                   | unset
 wal_buffers                    | 32
 wal_debug                      | 0
 wal_sync_method                | fdatasync
 zero_damaged_pages             | off
(113 rows)


suggestions, doubts and commentaries are very welcome

regards 
______________________________
Eduardo Cunha de Almeida
Administra��o de Banco de Dados
UFPR - CCE 
+55-41-361-3321
eduardo.almeida@ufpr.br
edalmeida@yahoo.com

--- Jan Wieck  wrote:
> Josh Berkus wrote:
> 
> > Folks,
> > 
> > I've sent a polite e-mail to Mr. Gomez offering
> our help.  Please, nobody 
> > flame him!
> > 
> 
> Please keep in mind that the entire test has, other
> than a similar 
> database schema and query types maybe, nothing to do
> with a TPC-H. I 
> don't see any kind of SUT. Foreign key support on
> the DB level is not 
> required by any of the TPC benchmarks. But the
> System Under Test, which 
> is the combination of middleware application and
> database together with 
> all computers and network components these parts are
> running on, must 
> implement all the required semantics, like ACID
> properties, referential 
> integrity &c. One could implement a TPC-H with flat
> files, it's just a 
> major pain in the middleware.
> 
> A proper TPC benchmark implementation would for
> example be a complete 
> PHP+DB application, where the user interaction is
> done by an emulated 
> "browser" and what is measured is the http response
> times, not anything 
> going on between PHP and the DB. Assuming that all
> requirements of the 
> TPC specification are implemented by either using
> available DB features, 
> or including appropriate workarounds in the PHP
> code, that would very 
> well lead to something that can compare PHP+MySQL
> vs. PHP+PostgreSQL.
> 
> All TPC benchmarks I have seen are performed by
> timing such a system 
> after a considerable rampup time, giving the DB
> system a chance to 
> properly populate caches and so forth. Rebooting the
> machine just before 
> the test is the wrong thing here and will especially
> kill any advanced 
> cache algorithms like ARC.
> 
> 
> Jan
> 
> -- 
>
#======================================================================#
> # It's easier to get forgiveness for being wrong
> than for being right. #
> # Let's break this rule - forgive me.               
>                   #
> #==================================================
> JanWieck@Yahoo.com #
> 
> 
> ---------------------------(end of
> broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
> 
>               
> http://www.postgresql.org/docs/faqs/FAQ.htmlIP 5:
> Have you checked our extensive FAQ?
> 
>               
http://www.postgresql.org/docs/faqs/FAQ.html



	
		
__________________________________
Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25�
http://photos.yahoo.com/ph/print_splash

From pgsql-advocacy-owner@postgresql.org  Mon Apr 26 11:17:46 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 3C1D2D1BACD; Thu, 22 Apr 2004 10:42:58 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 56629-02; Thu, 22 Apr 2004 10:42:55 -0300 (ADT)
Received: from elbereth.noviforum.si (unknown [193.189.169.66])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 5A783D1B579; Thu, 22 Apr 2004 10:42:50 -0300 (ADT)
Received: from elbereth.noviforum.si (localhost [127.0.0.1])
	by elbereth.noviforum.si (8.12.10/8.12.10) with ESMTP id i3MDgnHh014369;
	Thu, 22 Apr 2004 15:42:49 +0200
Received: (from gregab@localhost)
	by elbereth.noviforum.si (8.12.10/8.12.10/Submit) id i3MDgn2C014368;
	Thu, 22 Apr 2004 15:42:49 +0200
Date: Thu, 22 Apr 2004 15:42:49 +0200
From: Grega Bremec 
To: Eduardo Almeida 
Cc: Jan Wieck , Josh Berkus ,
	pgsql-advocacy@postgresql.org, pgsql-performance@postgresql.org
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
Message-ID: <20040422134249.GA14342@elbereth.noviforum.si>
References: <4086CDC7.5000204@Yahoo.com>
	<20040422125318.26053.qmail@web60601.mail.yahoo.com>
Mime-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature"; boundary="xHFwDpU9dbj6ez1V"
Content-Disposition: inline
In-Reply-To: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
User-Agent: Mutt/1.4.2i
Organization: Noviforum, Ltd., Software & Media
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/206
X-Sequence-Number: 4178

--xHFwDpU9dbj6ez1V
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

...and on Thu, Apr 22, 2004 at 05:53:18AM -0700, Eduardo Almeida used the k=
eyboard:
>=20
> - The configuration of the machine is:
> Dual opteron 64 bits model 240
> 4GB RAM
> 960 GB on RAID 0
> Mandrake Linux 64 with Kernel 2.6.5  (I compiled a
> kernel for this test)
> Java SDK  java version "1.4.2_04"
> PostgreSQL JDBC pg74.1jdbc3.jar
>=20
> - The TPC-H configuration is:
> TPC-H 2.0.0
> 100GB
> load using flat files
> Refresh functions using java
>=20

I'll just add for the reference, to those that aren't aware of it, the Java
virtual machine for x86_64 only exists in the 1.5 branch so far, and it's so
utterly unstable that most every notable shuffling around in the memory
crashes it. :)

Hence the 1.4.2_04 is a 32-bit application running in 32-bit mode.

I won't be getting into how much this affects the benchmarks as I didn't
really get into how CPU- and memory-intensive the refresh functions are in
these, so as I said - let's keep it a reference.

Cheers,
--=20
    Grega Bremec
    Senior Administrator
    Noviforum Ltd., Software & Media
    http://www.noviforum.si/

--xHFwDpU9dbj6ez1V
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQFAh8vZDo/EMYD4+osRAtE8AJ9p6blMEGRKc8wHc9ti+YcKBwk/+QCfaswM
TvcECDmUWoW4yiuQX1WNhoY=
=F8Wo
-----END PGP SIGNATURE-----

--xHFwDpU9dbj6ez1V--

From pgsql-performance-owner@postgresql.org  Sat Apr 24 01:28:56 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 77571D1B470
	for ;
	Sat, 24 Apr 2004 01:28:53 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 03923-10
	for ;
	Sat, 24 Apr 2004 01:28:51 -0300 (ADT)
Received: from ganymede.hub.org (u46n208.hfx.eastlink.ca [24.222.46.208])
	by svr1.postgresql.org (Postfix) with ESMTP id 2FF7ED1B43A
	for ;
	Sat, 24 Apr 2004 01:28:51 -0300 (ADT)
Received: by ganymede.hub.org (Postfix, from userid 1000)
	id CF15C341C6; Sat, 24 Apr 2004 01:28:50 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id CC53D33C79
	for ;
	Sat, 24 Apr 2004 01:28:50 -0300 (ADT)
X-Received: from localhost ([unix socket])
	by ganymede.hub.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 10:51:19 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id B8E5E37452
	for ; Thu, 22 Apr 2004 10:50:57 -0300 (ADT)
X-Received: from mail.postgresql.org [200.46.204.71]
	by localhost with IMAP (fetchmail-6.2.5)
	for scrappy@localhost (single-drop);
	Thu, 22 Apr 2004 10:50:57 -0300 (ADT)
X-Received: from svr1.postgresql.org ([unix socket])
	by svr1.postgresql.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 10:46:54 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DE67CD1C50D
	for ; Thu, 22 Apr 2004 10:46:53 -0300 (ADT)
X-Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 58271-05 for ;
	Thu, 22 Apr 2004 10:46:55 -0300 (ADT)
X-Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 08D54D1BACD
	for ; Thu, 22 Apr 2004 10:46:51 -0300 (ADT)
X-Received: from [134.22.70.203] (dyn-70-203.tor.dsl.tht.net [134.22.70.203])
	by tht.net (Postfix) with ESMTP
	id 329A376A31; Thu, 22 Apr 2004 09:46:56 -0400 (EDT)
Subject: Re: order by index, and inheritance
From: Rod Taylor 
To: Michiel Meeuwissen 
Cc: performance@postgresql.org
In-Reply-To: <20040422110214.GH6870@michiel.omroep.nl>
References: <20040419113013.GA32186@michiel.omroep.nl>
	<1082455997.57436.10.camel@jester>
	<20040420104518.GF1618@michiel.omroep.nl>
	<1082461767.57436.93.camel@jester>
	<20040422110214.GH6870@michiel.omroep.nl>
Content-Type: text/plain
Message-Id: <1082641607.91360.2.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Thu, 22 Apr 2004 09:46:48 -0400
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-DCC: : 
X-Spam-Pyzor: 
ReSent-Date: Sat, 24 Apr 2004 01:28:46 -0300 (ADT)
Resent-From: "Marc G. Fournier" 
Resent-To: pgsql-performance@postgresql.org
ReSent-Subject: Re: [PERFORM] order by index, and inheritance
ReSent-Message-ID: <20040424012846.R42925@ganymede.hub.org>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/422
X-Sequence-Number: 6722

On Thu, 2004-04-22 at 07:02, Michiel Meeuwissen wrote:
> Rod Taylor  wrote:
> > The scan is picking the best method for grabbing everything within the
> > table, since it is not aware that we do not require everything.
> 
> Hmm. That is a bit silly. Why does it use the index if select only from
> mm_mediasources?
> 
> > You can explicitly tell it what you want to do via:
> > 
> > SELECT *
> >   FROM (SELECT * FROM mm_mediasources ORDER BY number DESC LIMIT 20 
> >   UNION SELECT * FROM  ORDER BY number DESC LIMIT 20) AS tab
> >  ORDER BY number DESC LIMIT 20
> 
> I think you meant 'only mm_mediasources', and btw order by and limit are not
> accepted before union, so the above query does not compile.

Yes, I did mean only. Try putting another set of brackets around the
selects to get ORDER BY, etc. accepted. You can add another layer of
subselects in the from if that doesn't work.


From pgsql-advocacy-owner@postgresql.org  Mon Apr 26 11:22:57 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 574A6D1E12B
	for ;
	Thu, 22 Apr 2004 10:59:09 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 65063-03
	for ;
	Thu, 22 Apr 2004 10:59:11 -0300 (ADT)
Received: from web60607.mail.yahoo.com (web60607.mail.yahoo.com
	[216.109.118.245])
	by svr1.postgresql.org (Postfix) with SMTP id F004DD1DC74
	for ;
	Thu, 22 Apr 2004 10:59:06 -0300 (ADT)
Message-ID: <20040422135910.24475.qmail@web60607.mail.yahoo.com>
Received: from [200.17.210.82] by web60607.mail.yahoo.com via HTTP;
	Thu, 22 Apr 2004 06:59:10 PDT
Date: Thu, 22 Apr 2004 06:59:10 -0700 (PDT)
From: Eduardo Almeida 
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
To: Grega Bremec 
Cc: Jan Wieck , Josh Berkus ,
	pgsql-advocacy@postgresql.org, pgsql-performance@postgresql.org
In-Reply-To: <20040422134249.GA14342@elbereth.noviforum.si>
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/210
X-Sequence-Number: 4182

Grega,

That�s why I used java 32bits and needed to compile
the kernel 2.6.5 with the 32bits modules.
To reference, Sun has java 64bits just to IA64 and
Solaris Sparc 64 not to Opteron.

regards,
Eduardo
--- Grega Bremec  wrote:
> ...and on Thu, Apr 22, 2004 at 05:53:18AM -0700,
> Eduardo Almeida used the keyboard:
> > 
> > - The configuration of the machine is:
> > Dual opteron 64 bits model 240
> > 4GB RAM
> > 960 GB on RAID 0
> > Mandrake Linux 64 with Kernel 2.6.5  (I compiled a
> > kernel for this test)
> > Java SDK  java version "1.4.2_04"
> > PostgreSQL JDBC pg74.1jdbc3.jar
> > 
> > - The TPC-H configuration is:
> > TPC-H 2.0.0
> > 100GB
> > load using flat files
> > Refresh functions using java
> > 
> 
> I'll just add for the reference, to those that
> aren't aware of it, the Java
> virtual machine for x86_64 only exists in the 1.5
> branch so far, and it's so
> utterly unstable that most every notable shuffling
> around in the memory
> crashes it. :)
> 
> Hence the 1.4.2_04 is a 32-bit application running
> in 32-bit mode.
> 
> I won't be getting into how much this affects the
> benchmarks as I didn't
> really get into how CPU- and memory-intensive the
> refresh functions are in
> these, so as I said - let's keep it a reference.
> 
> Cheers,
> -- 
>     Grega Bremec
>     Senior Administrator
>     Noviforum Ltd., Software & Media
>     http://www.noviforum.si/
> 

> ATTACHMENT part 2 application/pgp-signature 




	
		
__________________________________
Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25�
http://photos.yahoo.com/ph/print_splash

From pgsql-performance-owner@postgresql.org  Mon Apr 26 14:55:03 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E045BD1B498
	for ;
	Sat, 24 Apr 2004 01:29:33 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 27547-01
	for ;
	Sat, 24 Apr 2004 01:29:32 -0300 (ADT)
Received: from ganymede.hub.org (u46n208.hfx.eastlink.ca [24.222.46.208])
	by svr1.postgresql.org (Postfix) with ESMTP id F231CD1B43A
	for ;
	Sat, 24 Apr 2004 01:29:30 -0300 (ADT)
Received: by ganymede.hub.org (Postfix, from userid 1000)
	id B3D2033CD9; Sat, 24 Apr 2004 01:29:30 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id B2CAA33C79
	for ;
	Sat, 24 Apr 2004 01:29:30 -0300 (ADT)
X-Received: from localhost ([unix socket])
	by ganymede.hub.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 11:40:56 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id C4084375F5
	for ; Thu, 22 Apr 2004 11:40:38 -0300 (ADT)
X-Received: from mail.postgresql.org [200.46.204.71]
	by localhost with IMAP (fetchmail-6.2.5)
	for scrappy@localhost (single-drop);
	Thu, 22 Apr 2004 11:40:38 -0300 (ADT)
X-Received: from svr1.postgresql.org ([unix socket])
	by svr1.postgresql.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 11:40:27 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 00B86D1CCB1
	for ; Thu, 22 Apr 2004 11:40:27 -0300 (ADT)
X-Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 70106-10 for ;
	Thu, 22 Apr 2004 11:40:24 -0300 (ADT)
X-Received: from minnie.omroep.nl (minnie.omroep.nl [145.58.30.4])
	by svr1.postgresql.org (Postfix) with ESMTP id A4E13D1CACD
	for ; Thu, 22 Apr 2004 11:40:19 -0300 (ADT)
X-Received: from tron.omroep.nl (tron.omroep.nl [145.58.31.20])
	by minnie.omroep.nl (Postfix) with ESMTP
	id BFA351CD0D27; Thu, 22 Apr 2004 16:40:23 +0200 (CDT)
X-Received: from michiel.omroep.nl (michiel.omroep.nl [145.58.67.10])
	by tron.omroep.nl (Postfix) with ESMTP
	id B0F6610056BD; Thu, 22 Apr 2004 16:40:23 +0200 (CEST)
X-Received: from michiel by michiel.omroep.nl with local (Exim 3.35 #1
	(Debian)) id 1BGfNH-0007OS-00; Thu, 22 Apr 2004 16:40:23 +0200
Date: Thu, 22 Apr 2004 16:40:23 +0200
From: Michiel Meeuwissen 
To: Rod Taylor 
Cc: performance@postgresql.org
Subject: Re: order by index, and inheritance
Message-ID: <20040422144023.GI6870@michiel.omroep.nl>
Reply-To: Michiel Meeuwissen 
References: <20040419113013.GA32186@michiel.omroep.nl>
	<1082455997.57436.10.camel@jester>
	<20040420104518.GF1618@michiel.omroep.nl>
	<1082461767.57436.93.camel@jester>
	<20040422110214.GH6870@michiel.omroep.nl>
	<1082641607.91360.2.camel@jester>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1082641607.91360.2.camel@jester>
Organization: Publieke Omroep []()
User-Agent: Mutt/1.5.5.1+cvs20040105i
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-DCC: : 
X-Spam-Pyzor: 
ReSent-Date: Sat, 24 Apr 2004 01:29:25 -0300 (ADT)
Resent-From: "Marc G. Fournier" 
Resent-To: pgsql-performance@postgresql.org
ReSent-Subject: Re: [PERFORM] order by index, and inheritance
ReSent-Message-ID: <20040424012925.V42925@ganymede.hub.org>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/443
X-Sequence-Number: 6743

Rod Taylor  wrote:
> On Thu, 2004-04-22 at 07:02, Michiel Meeuwissen wrote:
> > Rod Taylor  wrote:
> > > The scan is picking the best method for grabbing everything within the
> > > table, since it is not aware that we do not require everything.
> > 
> > Hmm. That is a bit silly. Why does it use the index if select only from
> > mm_mediasources?
> > 
> > > You can explicitly tell it what you want to do via:
> > > 
> > > SELECT *
> > >   FROM (SELECT * FROM mm_mediasources ORDER BY number DESC LIMIT 20 
> > >   UNION SELECT * FROM  ORDER BY number DESC LIMIT 20) AS tab
> > >  ORDER BY number DESC LIMIT 20
> > 
> > I think you meant 'only mm_mediasources', and btw order by and limit are not
> > accepted before union, so the above query does not compile.
> 
> Yes, I did mean only. Try putting another set of brackets around the
> selects to get ORDER BY, etc. accepted. You can add another layer of
> subselects in the from if that doesn't work.


Ok, I can get it working:

select number,url 
   from ( select number,url from  (select number,url from only mm_mediasources order by number desc limit 20) as A 
          union select number,url from (select number,url from mm_audiosources order by number desc limit 20) as B
          union select number,url from (select number,url from mm_videosources order by number desc limit 20) as C
        ) as TAB order by number desc limit 20;

This indeeds performs good (about 10000 times faster then select number,url
from mm_mediasources order by number desc limit 20) . But hardly beautiful,
and quite useless too because of course I am now going to want to use an
offset (limit 20 offset 20, you see..), which seems more or less impossible
in this way, isn't it.

select number,url 
   from ( select number,url from  (select number,url from only mm_mediasources order by number desc limit 100020) as A 
          union select number,url from (select number,url from mm_audiosources order by number desc limit 100020) as B
          union select number,url from (select number,url from mm_videosources order by number desc limit 100020) as C
        ) as TAB order by number desc limit 20 offset 100000;

This would be it, I think, but this performs, expectedly, quit bad again,
though still 5 times faster then  select url,number from mm_mediasources order by number desc limit 20 offset 100000;


I'm thinking of dropping inheritance all together and using foreign keys or
so for the extra fields, to simulate inheritance. That might perhaps work a whole lot better?

Thanks anyway,

-- 
Michiel Meeuwissen
Mediapark C101 Hilversum
+31 (0)35 6772979
nl_NL eo_XX en_US
mihxil'
 [] ()

From pgsql-performance-owner@postgresql.org  Sat Apr 24 01:29:46 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9BEF3D1CCA2
	for ;
	Sat, 24 Apr 2004 01:29:40 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 05183-06
	for ;
	Sat, 24 Apr 2004 01:29:39 -0300 (ADT)
Received: from ganymede.hub.org (u46n208.hfx.eastlink.ca [24.222.46.208])
	by svr1.postgresql.org (Postfix) with ESMTP id D7F49D1CB1D
	for ;
	Sat, 24 Apr 2004 01:29:38 -0300 (ADT)
Received: by ganymede.hub.org (Postfix, from userid 1000)
	id 52E8B3429B; Sat, 24 Apr 2004 01:29:38 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id 521D0341C6
	for ;
	Sat, 24 Apr 2004 01:29:38 -0300 (ADT)
X-Received: from localhost ([unix socket])
	by ganymede.hub.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 11:50:38 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (localhost [127.0.0.1])
	by ganymede.hub.org (Postfix) with ESMTP id 94BA537628
	for ; Thu, 22 Apr 2004 11:50:35 -0300 (ADT)
X-Received: from mail.postgresql.org [200.46.204.71]
	by localhost with IMAP (fetchmail-6.2.5)
	for scrappy@localhost (single-drop);
	Thu, 22 Apr 2004 11:50:35 -0300 (ADT)
X-Received: from svr1.postgresql.org ([unix socket])
	by svr1.postgresql.org (Cyrus v2.2.3) with LMTP;
	Thu, 22 Apr 2004 11:46:45 -0300
X-Sieve: CMU Sieve 2.2
X-Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 842ECD1D09C
	for ; Thu, 22 Apr 2004 11:46:44 -0300 (ADT)
X-Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 84786-01 for ;
	Thu, 22 Apr 2004 11:46:47 -0300 (ADT)
X-Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DEC58D1CCB1
	for ; Thu, 22 Apr 2004 11:46:42 -0300 (ADT)
X-Received: from [134.22.70.203] (dyn-70-203.tor.dsl.tht.net [134.22.70.203])
	by tht.net (Postfix) with ESMTP
	id 1677976AA2; Thu, 22 Apr 2004 10:46:50 -0400 (EDT)
Subject: Re: order by index, and inheritance
From: Rod Taylor 
To: Michiel Meeuwissen 
Cc: performance@postgresql.org
In-Reply-To: <20040422144023.GI6870@michiel.omroep.nl>
References: <20040419113013.GA32186@michiel.omroep.nl>
	<1082455997.57436.10.camel@jester>
	<20040420104518.GF1618@michiel.omroep.nl>
	<1082461767.57436.93.camel@jester>
	<20040422110214.GH6870@michiel.omroep.nl>
	<1082641607.91360.2.camel@jester>
	<20040422144023.GI6870@michiel.omroep.nl>
Content-Type: text/plain
Message-Id: <1082645200.91360.24.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Thu, 22 Apr 2004 10:46:40 -0400
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-DCC: : 
X-Spam-Pyzor: 
ReSent-Date: Sat, 24 Apr 2004 01:29:33 -0300 (ADT)
Resent-From: "Marc G. Fournier" 
Resent-To: pgsql-performance@postgresql.org
ReSent-Subject: Re: [PERFORM] order by index, and inheritance
ReSent-Message-ID: <20040424012933.R42925@ganymede.hub.org>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/423
X-Sequence-Number: 6723


> This indeeds performs good (about 10000 times faster then select number,url
> from mm_mediasources order by number desc limit 20) . But hardly beautiful,
> and quite useless too because of course I am now going to want to use an
> offset (limit 20 offset 20, you see..), which seems more or less impossible
> in this way, isn't it.

Yes, and the offset is a good reason why PostgreSQL will not be able to
do it by itself either.

Is "number" unique across the board?

If so, instead of the offset you could use WHERE number > $lastValue.


From pgsql-advocacy-owner@postgresql.org  Thu Apr 22 14:34:06 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CB5A6D1D3BE
	for ;
	Thu, 22 Apr 2004 12:54:50 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 06636-03
	for ;
	Thu, 22 Apr 2004 12:54:49 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 024E2D1D067
	for ;
	Thu, 22 Apr 2004 12:54:49 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3MFsmGD007445;
	Thu, 22 Apr 2004 11:54:48 -0400 (EDT)
To: Eduardo Almeida 
Cc: Jan Wieck , Josh Berkus ,
	pgsql-advocacy@postgresql.org, pgsql-performance@postgresql.org
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks 
In-reply-to: <20040422125318.26053.qmail@web60601.mail.yahoo.com> 
References: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
Comments: In-reply-to Eduardo Almeida 
	message dated "Thu, 22 Apr 2004 05:53:18 -0700"
Date: Thu, 22 Apr 2004 11:54:48 -0400
Message-ID: <7444.1082649288@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/125
X-Sequence-Number: 4097

Eduardo Almeida  writes:
> About 7hs:30min to load the data and 16:09:25 to
> create the indexes

You could probably improve the index-create time by temporarily
increasing sort_mem.  It wouldn't be unreasonable to give CREATE INDEX
several hundred meg to work in.  (You don't want sort_mem that big
normally, because there may be many sorts happening in parallel,
but in a data-loading context there'll just be one active sort.)

			regards, tom lane

From pgsql-advocacy-owner@postgresql.org  Thu Apr 22 13:44:02 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 4631FD1D2B7
	for ;
	Thu, 22 Apr 2004 13:20:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 17346-04
	for ;
	Thu, 22 Apr 2004 13:19:57 -0300 (ADT)
Received: from smtp011.mail.yahoo.com (smtp011.mail.yahoo.com
 [216.136.173.31])
	by svr1.postgresql.org (Postfix) with SMTP id E1E75D1CCAD
	for ;
	Thu, 22 Apr 2004 13:19:55 -0300 (ADT)
Received: from unknown (HELO europa.janwieck.net) (janwieck@68.80.245.191 with
	login)
	by smtp011.mail.yahoo.com with SMTP; 22 Apr 2004 16:19:55 -0000
Received: from Yahoo.com (ismtp.afilias.com [216.217.55.254]) (authenticated)
	by europa.janwieck.net (8.11.6/8.11.6) with ESMTP id i3MGJrj15217;
	Thu, 22 Apr 2004 12:19:53 -0400
Message-ID: <4087F0A3.6040009@Yahoo.com>
Date: Thu, 22 Apr 2004 12:19:47 -0400
From: Jan Wieck 
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.4) Gecko/20030624
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Eduardo Almeida 
Cc: Josh Berkus , pgsql-advocacy@postgresql.org,
	pgsql-performance@postgresql.org
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
References: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
In-Reply-To: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 8bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=2.6 tagged_above=0.0 required=5.0
	tests=RCVD_IN_DYNABLOCK, RCVD_IN_SORBS
X-Spam-Level: **
X-Archive-Number: 200404/124
X-Sequence-Number: 4096

Eduardo Almeida wrote:

> Folks,
> 
> I�m doing the 100GB TPC-H and I�ll show the previous
> results to our community (Postgres) in 3 weeks before
> finishing the study.
> 
> My intention is to carry through a test with a VLDB in
> a low cost platform (PostgreSQL, Linux and cheap HW)
> and not to compare with another DBMS.

QphH and Price/QphH will be enought for us to see where in the list we 
are. Unfortunately there are only Sybase and MS SQL results published in 
the 100 GB category. The 300 GB has DB2 as well. Oracle starts at 1 TB 
and in the 10 TB category Oracle and DB2 are the only players left.


Jan

> 
> So far I can tell you that the load time on PG 7.4.2
> with kernel 2.6.5 on Opteron 64 model 240 in RAID 0
> with 8 disks (960 GB) loaded the database in less than
> 24 hours. 
> About 7hs:30min to load the data and 16:09:25 to
> create the indexes
> 
> The Power test still running and that�s why I�ll not
> present anything so far. Now I�ll just send to the
> list my environment configuration.
> 
> - The configuration of the machine is:
> Dual opteron 64 bits model 240
> 4GB RAM
> 960 GB on RAID 0
> Mandrake Linux 64 with Kernel 2.6.5  (I compiled a
> kernel for this test)
> Java SDK  java version "1.4.2_04"
> PostgreSQL JDBC pg74.1jdbc3.jar
> 
> - The TPC-H configuration is:
> TPC-H 2.0.0
> 100GB
> load using flat files
> Refresh functions using java
> 
> - The PostgreSQL 7.4.2 configuration is:
> 
> add_missing_from               | on
>  australian_timezones           | off
>  authentication_timeout         | 60
>  check_function_bodies          | on
>  checkpoint_segments            | 128
>  checkpoint_timeout             | 300
>  checkpoint_warning             | 30
>  client_encoding                | SQL_ASCII
>  client_min_messages            | notice
>  commit_delay                   | 0
>  commit_siblings                | 5
>  cpu_index_tuple_cost           | 0.001
>  cpu_operator_cost              | 0.0025
>  cpu_tuple_cost                 | 0.01
>  DateStyle                      | ISO, MDY
>  db_user_namespace              | off
>  deadlock_timeout               | 1000
>  debug_pretty_print             | off
>  debug_print_parse              | off
>  debug_print_plan               | off
>  debug_print_rewritten          | off
>  default_statistics_target      | 10
>  default_transaction_isolation  | read committed
>  default_transaction_read_only  | off
>  dynamic_library_path           | $libdir
>  effective_cache_size           | 150000
>  enable_hashagg                 | on
>  enable_hashjoin                | on
>  enable_indexscan               | on
>  enable_mergejoin               | on
>  enable_nestloop                | on
>  enable_seqscan                 | on
>  enable_sort                    | on
>  enable_tidscan                 | on
>  explain_pretty_print           | on
>  extra_float_digits             | 0
>  from_collapse_limit            | 8
>  fsync                          | off
>  geqo                           | on
>  geqo_effort                    | 1
>  geqo_generations               | 0
> geqo_pool_size                 | 0
>  geqo_selection_bias            | 2
>  geqo_threshold                 | 11
>  join_collapse_limit            | 8
>  krb_server_keyfile             | unset
>  lc_collate                     | en_US
>  lc_ctype                       | en_US
>  lc_messages                    | C
>  lc_monetary                    | C
>  lc_numeric                     | C
>  lc_time                        | C
>  log_connections                | off
>  log_duration                   | off
>  log_error_verbosity            | default
>  log_executor_stats             | off
>  log_hostname                   | off
>  log_min_duration_statement     | -1
>  log_min_error_statement        | panic
>  log_min_messages               | notice
>  log_parser_stats               | off
>  log_pid                        | off
>  log_planner_stats              | off
>  log_source_port                | off
>  log_statement                  | off
>  log_statement_stats            | off
>  log_timestamp                  | off
>  max_connections                | 10
>  max_expr_depth                 | 10000
>  max_files_per_process          | 1000
>  max_fsm_pages                  | 20000
>  max_fsm_relations              | 1000
>  max_locks_per_transaction      | 64
>  password_encryption            | on
>  port                           | 5432
>  pre_auth_delay                 | 0
>  preload_libraries              | unset
>  random_page_cost               | 1.25
>  regex_flavor                   | advanced
>  rendezvous_name                | unset
>  search_path                    | $user,public
>  server_encoding                | SQL_ASCII
>  server_version                 | 7.4.2
>  shared_buffers                 | 40000
>  silent_mode                    | off
> sort_mem                       | 65536
>  sql_inheritance                | on
>  ssl                            | off
>  statement_timeout              | 10000000
>  stats_block_level              | off
>  stats_command_string           | off
>  stats_reset_on_server_start    | on
>  stats_row_level                | off
>  stats_start_collector          | on
>  superuser_reserved_connections | 2
>  syslog                         | 0
>  syslog_facility                | LOCAL0
>  syslog_ident                   | postgres
>  tcpip_socket                   | on
>  TimeZone                       | unknown
>  trace_notify                   | off
>  transaction_isolation          | read committed
>  transaction_read_only          | off
>  transform_null_equals          | off
>  unix_socket_directory          | unset
>  unix_socket_group              | unset
>  unix_socket_permissions        | 511
>  vacuum_mem                     | 65536
>  virtual_host                   | unset
>  wal_buffers                    | 32
>  wal_debug                      | 0
>  wal_sync_method                | fdatasync
>  zero_damaged_pages             | off
> (113 rows)
> 
> 
> suggestions, doubts and commentaries are very welcome
> 
> regards 
> ______________________________
> Eduardo Cunha de Almeida
> Administra��o de Banco de Dados
> UFPR - CCE 
> +55-41-361-3321
> eduardo.almeida@ufpr.br
> edalmeida@yahoo.com
> 
> --- Jan Wieck  wrote:
>> Josh Berkus wrote:
>> 
>> > Folks,
>> > 
>> > I've sent a polite e-mail to Mr. Gomez offering
>> our help.  Please, nobody 
>> > flame him!
>> > 
>> 
>> Please keep in mind that the entire test has, other
>> than a similar 
>> database schema and query types maybe, nothing to do
>> with a TPC-H. I 
>> don't see any kind of SUT. Foreign key support on
>> the DB level is not 
>> required by any of the TPC benchmarks. But the
>> System Under Test, which 
>> is the combination of middleware application and
>> database together with 
>> all computers and network components these parts are
>> running on, must 
>> implement all the required semantics, like ACID
>> properties, referential 
>> integrity &c. One could implement a TPC-H with flat
>> files, it's just a 
>> major pain in the middleware.
>> 
>> A proper TPC benchmark implementation would for
>> example be a complete 
>> PHP+DB application, where the user interaction is
>> done by an emulated 
>> "browser" and what is measured is the http response
>> times, not anything 
>> going on between PHP and the DB. Assuming that all
>> requirements of the 
>> TPC specification are implemented by either using
>> available DB features, 
>> or including appropriate workarounds in the PHP
>> code, that would very 
>> well lead to something that can compare PHP+MySQL
>> vs. PHP+PostgreSQL.
>> 
>> All TPC benchmarks I have seen are performed by
>> timing such a system 
>> after a considerable rampup time, giving the DB
>> system a chance to 
>> properly populate caches and so forth. Rebooting the
>> machine just before 
>> the test is the wrong thing here and will especially
>> kill any advanced 
>> cache algorithms like ARC.
>> 
>> 
>> Jan
>> 
>> -- 
>>
> #======================================================================#
>> # It's easier to get forgiveness for being wrong
>> than for being right. #
>> # Let's break this rule - forgive me.               
>>                   #
>> #==================================================
>> JanWieck@Yahoo.com #
>> 
>> 
>> ---------------------------(end of
>> broadcast)---------------------------
>> TIP 5: Have you checked our extensive FAQ?
>> 
>>               
>> http://www.postgresql.org/docs/faqs/FAQ.htmlIP 5:
>> Have you checked our extensive FAQ?
>> 
>>               
> http://www.postgresql.org/docs/faqs/FAQ.html
> 
> 
> 
> 	
> 		
> __________________________________
> Do you Yahoo!?
> Yahoo! Photos: High-quality 4x6 digital prints for 25�
> http://photos.yahoo.com/ph/print_splash


-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck@Yahoo.com #


From pgsql-advocacy-owner@postgresql.org  Mon Apr 26 11:23:41 2004
X-Original-To: pgsql-advocacy-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 8D18CD1DA3C
	for ;
	Thu, 22 Apr 2004 14:11:07 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 38476-02
	for ;
	Thu, 22 Apr 2004 14:11:05 -0300 (ADT)
Received: from web60607.mail.yahoo.com (web60607.mail.yahoo.com
	[216.109.118.245])
	by svr1.postgresql.org (Postfix) with SMTP id 8D53CD1D26C
	for ;
	Thu, 22 Apr 2004 14:11:03 -0300 (ADT)
Message-ID: <20040422171034.91737.qmail@web60607.mail.yahoo.com>
Received: from [200.17.210.82] by web60607.mail.yahoo.com via HTTP;
	Thu, 22 Apr 2004 10:10:34 PDT
Date: Thu, 22 Apr 2004 10:10:34 -0700 (PDT)
From: Eduardo Almeida 
Subject: Re: [PERFORM] MySQL vs PG TPC-H benchmarks
To: Jan Wieck 
Cc: Josh Berkus , pgsql-advocacy@postgresql.org,
	pgsql-performance@postgresql.org
In-Reply-To: <4087F0A3.6040009@Yahoo.com>
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/212
X-Sequence-Number: 4184

Folks,

I forgot to mention that I used Shell scripts to load
the data and use Java just to run the refresh
functions.

Talking about sort_mem config, I used 65000 but in the
TPCH specification they said that you are not able to
change the configs when you start the benchmark, is
that a big problem to use 65000? In the TPCH 100GB we
run 5 streams in parallel for the throughput test!  To
power test I think is not a problem because it runs
one query after another.

Another thing is that I put statement_timeout =
10000000 

Some queries may exceed this timeout and I�ll send the
EXPLAIN for this ones.
             
The last thing is that Jan forgets to mention that
Teradata doesn�t show up now but in older lists shows
3TB and 10TB results.

regards
Eduardo
--- Jan Wieck  wrote:
> Eduardo Almeida wrote:
> 
> > Folks,
> > 
> > I�m doing the 100GB TPC-H and I�ll show the
> previous
> > results to our community (Postgres) in 3 weeks
> before
> > finishing the study.
> > 
> > My intention is to carry through a test with a
> VLDB in
> > a low cost platform (PostgreSQL, Linux and cheap
> HW)
> > and not to compare with another DBMS.
> 
> QphH and Price/QphH will be enought for us to see
> where in the list we 
> are. Unfortunately there are only Sybase and MS SQL
> results published in 
> the 100 GB category. The 300 GB has DB2 as well.
> Oracle starts at 1 TB 
> and in the 10 TB category Oracle and DB2 are the
> only players left.
> 
> 
> Jan
> 
> > 
> > So far I can tell you that the load time on PG
> 7.4.2
> > with kernel 2.6.5 on Opteron 64 model 240 in RAID
> 0
> > with 8 disks (960 GB) loaded the database in less
> than
> > 24 hours. 
> > About 7hs:30min to load the data and 16:09:25 to
> > create the indexes
> > 
> > The Power test still running and that�s why I�ll
> not
> > present anything so far. Now I�ll just send to the
> > list my environment configuration.
> > 
> > - The configuration of the machine is:
> > Dual opteron 64 bits model 240
> > 4GB RAM
> > 960 GB on RAID 0
> > Mandrake Linux 64 with Kernel 2.6.5  (I compiled a
> > kernel for this test)
> > Java SDK  java version "1.4.2_04"
> > PostgreSQL JDBC pg74.1jdbc3.jar
> > 
> > - The TPC-H configuration is:
> > TPC-H 2.0.0
> > 100GB
> > load using flat files
> > Refresh functions using java
> > 
> > - The PostgreSQL 7.4.2 configuration is:
> > 
> > add_missing_from               | on
> >  australian_timezones           | off
> >  authentication_timeout         | 60
> >  check_function_bodies          | on
> >  checkpoint_segments            | 128
> >  checkpoint_timeout             | 300
> >  checkpoint_warning             | 30
> >  client_encoding                | SQL_ASCII
> >  client_min_messages            | notice
> >  commit_delay                   | 0
> >  commit_siblings                | 5
> >  cpu_index_tuple_cost           | 0.001
> >  cpu_operator_cost              | 0.0025
> >  cpu_tuple_cost                 | 0.01
> >  DateStyle                      | ISO, MDY
> >  db_user_namespace              | off
> >  deadlock_timeout               | 1000
> >  debug_pretty_print             | off
> >  debug_print_parse              | off
> >  debug_print_plan               | off
> >  debug_print_rewritten          | off
> >  default_statistics_target      | 10
> >  default_transaction_isolation  | read committed
> >  default_transaction_read_only  | off
> >  dynamic_library_path           | $libdir
> >  effective_cache_size           | 150000
> >  enable_hashagg                 | on
> >  enable_hashjoin                | on
> >  enable_indexscan               | on
> >  enable_mergejoin               | on
> >  enable_nestloop                | on
> >  enable_seqscan                 | on
> >  enable_sort                    | on
> >  enable_tidscan                 | on
> >  explain_pretty_print           | on
> >  extra_float_digits             | 0
> >  from_collapse_limit            | 8
> >  fsync                          | off
> >  geqo                           | on
> >  geqo_effort                    | 1
> >  geqo_generations               | 0
> > geqo_pool_size                 | 0
> >  geqo_selection_bias            | 2
> >  geqo_threshold                 | 11
> >  join_collapse_limit            | 8
> >  krb_server_keyfile             | unset
> >  lc_collate                     | en_US
> >  lc_ctype                       | en_US
> >  lc_messages                    | C
> >  lc_monetary                    | C
> >  lc_numeric                     | C
> >  lc_time                        | C
> >  log_connections                | off
> >  log_duration                   | off
> >  log_error_verbosity            | default
> >  log_executor_stats             | off
> >  log_hostname                   | off
> >  log_min_duration_statement     | -1
> >  log_min_error_statement        | panic
> >  log_min_messages               | notice
> >  log_parser_stats               | off
> >  log_pid                        | off
> >  log_planner_stats              | off
> >  log_source_port                | off
> >  log_statement                  | off
> >  log_statement_stats            | off
> >  log_timestamp                  | off
> >  max_connections                | 10
> >  max_expr_depth                 | 10000
> >  max_files_per_process          | 1000
> >  max_fsm_pages                  | 20000
> >  max_fsm_relations              | 1000
> >  max_locks_per_transaction      | 64
> >  password_encryption            | on
> >  port                           | 5432
> >  pre_auth_delay                 | 0
> >  preload_libraries              | unset
> >  random_page_cost               | 1.25
> >  regex_flavor                   | advanced
> >  rendezvous_name                | unset
> >  search_path                    | $user,public
> >  server_encoding                | SQL_ASCII
> >  server_version                 | 7.4.2
> >  shared_buffers                 | 40000
> >  silent_mode                    | off
> > sort_mem                       | 65536
> >  sql_inheritance                | on
> >  ssl                            | off
> >  statement_timeout              | 10000000
> >  stats_block_level              | off
> >  stats_command_string           | off
> >  stats_reset_on_server_start    | on
> >  stats_row_level                | off
> >  stats_start_collector          | on
> >  superuser_reserved_connections | 2
> >  syslog                         | 0
> >  syslog_facility                | LOCAL0
> >  syslog_ident                   | postgres
> >  tcpip_socket                   | on
> >  TimeZone                       | unknown
> >  trace_notify                   | off
> >  transaction_isolation          | read committed
> >  transaction_read_only          | off
> >  transform_null_equals          | off
> >  unix_socket_directory          | unset
> >  unix_socket_group              | unset
> >  unix_socket_permissions        | 511
> >  vacuum_mem                     | 65536
> >  virtual_host                   | unset
> >  wal_buffers                    | 32
> >  wal_debug                      | 0
> >  wal_sync_method                | fdatasync
> >  zero_damaged_pages             | off
> > (113 rows)
> > 
> > 
> > suggestions, doubts and commentaries are very
> welcome
> > 
> > regards 
> > ______________________________
> > Eduardo Cunha de Almeida
> > Administra��o de Banco de Dados
> > UFPR - CCE 
> > +55-41-361-3321
> > eduardo.almeida@ufpr.br
> > edalmeida@yahoo.com
> > 
> > --- Jan Wieck  wrote:
> >> Josh Berkus wrote:
> >> 
> >> > Folks,
> >> > 
> >> > I've sent a polite e-mail to Mr. Gomez offering
> >> our help.  Please, nobody 
> >> > flame him!
> >> > 
> >> 
> >> Please keep in mind that the entire test has,
> other
> 
=== message truncated ===



	
		
__________________________________
Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25�
http://photos.yahoo.com/ph/print_splash

From pgsql-performance-owner@postgresql.org  Thu Apr 22 16:25:21 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 13792D1DF8B
	for ;
	Thu, 22 Apr 2004 14:37:47 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 45280-07
	for ;
	Thu, 22 Apr 2004 14:37:46 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 002A7D1E0F2
	for ;
	Thu, 22 Apr 2004 14:37:44 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4922627; Thu, 22 Apr 2004 10:39:10 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: Tom Lane ,
	Bruce Momjian 
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Thu, 22 Apr 2004 10:37:10 -0700
User-Agent: KMail/1.4.3
Cc: pg@fastcrypt.com, Paul Tuckfield ,
	Anjan Dave , Neil Conway ,
	Dirk =?iso-8859-1?q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
References: <200404220456.i3M4uFv04710@candle.pha.pa.us>
	<20171.1082610827@sss.pgh.pa.us>
In-Reply-To: <20171.1082610827@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404221037.10097.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/384
X-Sequence-Number: 6684

Tom,

> The tricky
> part is that a slow adaptation rate means we can't have every backend
> figuring this out for itself --- the right value would have to be
> maintained globally, and I'm not sure how to do that without adding a
> lot of overhead.

This may be a moot point, since you've stated that changing the loop timing 
won't solve the problem, but what about making the test part of make?   I 
don't think too many systems are going to change processor architectures once 
in production, and those that do can be told to re-compile.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Thu Apr 22 14:54:13 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E3178D1E124
	for ;
	Thu, 22 Apr 2004 14:46:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 53004-02
	for ;
	Thu, 22 Apr 2004 14:46:25 -0300 (ADT)
Received: from mail.deg.cc (mail.deg.cc [198.70.16.196])
	by svr1.postgresql.org (Postfix) with ESMTP id 5D451D1E104
	for ;
	Thu, 22 Apr 2004 14:46:24 -0300 (ADT)
Received: from deg.cc (barney.deg.cc [198.70.16.205])
	by mail.deg.cc (Postfix) with ESMTP id D6A3955F7D
	for ;
	Thu, 22 Apr 2004 13:46:17 -0400 (EDT)
Message-ID: <4088062E.2050901@deg.cc>
Date: Thu, 22 Apr 2004 13:51:42 -0400
From: Pallav Kalva 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4.2) Gecko/20040308
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Setting Shared Buffers , Effective Cache, Sort Mem Parameters
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/383
X-Sequence-Number: 6683

Hi

    We are in the process of building a new machine for our production 
database. Below you will see some of the harware specs for the machine. 
I need some help with setting these parameters (shared buffers, 
effective cache, sort mem) in the pg_conf file.  Also can anyone explain 
the difference between shared buffers and effective cache , how these 
are allocated in the main memory (the docs are not clear on this).

Here are the Hardware details:
Operating System: Red Hat 9
Database Ver: Postgres 7.4
CPU'S : 4
RAM : 4 gig
Datafile layout : RAID 1+0
Transaction log : on different RAID1 Array
RAID Stripe Size: 8k


Thanks!
Pallav


From pgsql-performance-owner@postgresql.org  Thu Apr 22 18:02:24 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 57232D1DF75
	for ;
	Thu, 22 Apr 2004 14:56:03 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 54670-07
	for ;
	Thu, 22 Apr 2004 14:56:02 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id BFB1ED1DF5D
	for ;
	Thu, 22 Apr 2004 14:56:00 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3MHtr7h008584;
	Thu, 22 Apr 2004 13:55:53 -0400 (EDT)
To: Josh Berkus 
Cc: Bruce Momjian , pg@fastcrypt.com,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	Dirk =?iso-8859-1?q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1 
In-reply-to: <200404221037.10097.josh@agliodbs.com> 
References: <200404220456.i3M4uFv04710@candle.pha.pa.us>
	<20171.1082610827@sss.pgh.pa.us>
	<200404221037.10097.josh@agliodbs.com>
Comments: In-reply-to Josh Berkus 
	message dated "Thu, 22 Apr 2004 10:37:10 -0700"
Date: Thu, 22 Apr 2004 13:55:53 -0400
Message-ID: <8583.1082656553@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/390
X-Sequence-Number: 6690

Josh Berkus  writes:
> This may be a moot point, since you've stated that changing the loop timing 
> won't solve the problem, but what about making the test part of make?   I 
> don't think too many systems are going to change processor architectures once
> in production, and those that do can be told to re-compile.

Having to recompile to run on single- vs dual-processor machines doesn't
seem like it would fly.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 22 17:45:27 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C9937D1DF89
	for ;
	Thu, 22 Apr 2004 15:11:24 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 58883-07
	for ;
	Thu, 22 Apr 2004 15:11:24 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id E8978D1DF87
	for ;
	Thu, 22 Apr 2004 15:11:22 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4922814; Thu, 22 Apr 2004 11:12:48 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: Tom Lane 
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Thu, 22 Apr 2004 11:11:43 -0700
User-Agent: KMail/1.5.4
Cc: Bruce Momjian , pg@fastcrypt.com,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	Dirk =?iso-8859-1?q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
References: <200404220456.i3M4uFv04710@candle.pha.pa.us>
	<200404221037.10097.josh@agliodbs.com>
	<8583.1082656553@sss.pgh.pa.us>
In-Reply-To: <8583.1082656553@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404221111.43151.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/389
X-Sequence-Number: 6689

Tom,

> Having to recompile to run on single- vs dual-processor machines doesn't
> seem like it would fly.

Oh, I don't know.  Many applications require compiling for a target 
architecture; SQL Server, for example, won't use a 2nd processor without 
re-installation.   I'm not sure about Oracle.

It certainly wasn't too long ago that Linux gurus were esposing re-compiling 
the kernel for the machine.

And it's not like they would *have* to re-compile to use PostgreSQL after 
adding an additional processor.  Just if they wanted to maximize peformance 
benefit.

Also, this is a fairly rare circumstance, I think; to judge by my clients, 
once a database server is in production nobody touches the hardware.

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Thu Apr 22 17:31:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 4116AD1D52E
	for ;
	Thu, 22 Apr 2004 15:21:46 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 58883-10
	for ;
	Thu, 22 Apr 2004 15:21:46 -0300 (ADT)
Received: from web11.manitu.net (web11.manitu.net [217.11.48.111])
	by svr1.postgresql.org (Postfix) with ESMTP id 52E61D1D2B7
	for ;
	Thu, 22 Apr 2004 15:21:44 -0300 (ADT)
Received: from [192.168.0.1] (dsl-082-082-191-254.arcor-ip.net
 [82.82.191.254])
	(authenticated)
	by web11.manitu.net (8.10.2-SOL3/8.10.2) with ESMTP id i3MIKhM15894;
	Thu, 22 Apr 2004 20:20:43 +0200
Subject: Re: [pgsql-advocacy] MySQL vs PG TPC-H benchmarks
From: Markus Bertheau 
To: Tom Lane 
Cc: Eduardo Almeida ,
	Jan Wieck , Josh Berkus ,
	pgsql-performance@postgresql.org
In-Reply-To: <7444.1082649288@sss.pgh.pa.us>
References: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
	<7444.1082649288@sss.pgh.pa.us>
Content-Type: text/plain; charset=UTF-8
Message-Id: <1082658047.4524.0.camel@yarrow.bertheau.de>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 (1.4.6-1) 
Date: Thu, 22 Apr 2004 20:20:47 +0200
Content-Transfer-Encoding: quoted-printable
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/388
X-Sequence-Number: 6688

=D0=92 =D0=A7=D1=82=D0=B2, 22.04.2004, =D0=B2 17:54, Tom Lane =D0=BF=D0=B8=
=D1=88=D0=B5=D1=82:
> Eduardo Almeida  writes:
> > About 7hs:30min to load the data and 16:09:25 to
> > create the indexes
>=20
> You could probably improve the index-create time by temporarily
> increasing sort_mem.  It wouldn't be unreasonable to give CREATE INDEX
> several hundred meg to work in.  (You don't want sort_mem that big
> normally, because there may be many sorts happening in parallel,
> but in a data-loading context there'll just be one active sort.)

Doesn't this provide a reason for CREATE INDEX not to honour sort_mem?

--=20
Markus Bertheau 


From pgsql-performance-owner@postgresql.org  Thu Apr 22 17:24:08 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id ADF17D1CCAD
	for ;
	Thu, 22 Apr 2004 15:31:56 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 70736-02
	for ;
	Thu, 22 Apr 2004 15:31:54 -0300 (ADT)
Received: from candle.pha.pa.us (candle.pha.pa.us [207.106.42.251])
	by svr1.postgresql.org (Postfix) with ESMTP id 3C84AD1CCA8
	for ;
	Thu, 22 Apr 2004 15:31:52 -0300 (ADT)
Received: (from pgman@localhost)
	by candle.pha.pa.us (8.11.6/8.11.6) id i3MIVMl22821;
	Thu, 22 Apr 2004 14:31:22 -0400 (EDT)
From: Bruce Momjian 
Message-Id: <200404221831.i3MIVMl22821@candle.pha.pa.us>
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
In-Reply-To: <200404221111.43151.josh@agliodbs.com>
To: josh@agliodbs.com
Date: Thu, 22 Apr 2004 14:31:22 -0400 (EDT)
Cc: Tom Lane , pg@fastcrypt.com,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	=?ISO-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/387
X-Sequence-Number: 6687

Josh Berkus wrote:
> Tom,
> 
> > Having to recompile to run on single- vs dual-processor machines doesn't
> > seem like it would fly.
> 
> Oh, I don't know.  Many applications require compiling for a target 
> architecture; SQL Server, for example, won't use a 2nd processor without 
> re-installation.   I'm not sure about Oracle.
> 
> It certainly wasn't too long ago that Linux gurus were esposing re-compiling 
> the kernel for the machine.
> 
> And it's not like they would *have* to re-compile to use PostgreSQL after 
> adding an additional processor.  Just if they wanted to maximize peformance 
> benefit.
> 
> Also, this is a fairly rare circumstance, I think; to judge by my clients, 
> once a database server is in production nobody touches the hardware.

A much simpler solution would be for the postmaster to run a test during
startup.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

From pgsql-performance-owner@postgresql.org  Thu Apr 22 17:05:17 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1312CD1DCAC
	for ;
	Thu, 22 Apr 2004 15:47:19 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 68438-07
	for ;
	Thu, 22 Apr 2004 15:47:18 -0300 (ADT)
Received: from mwinf0303.wanadoo.fr (smtp3.wanadoo.fr [193.252.22.28])
	by svr1.postgresql.org (Postfix) with ESMTP id 58826D1D10E
	for ;
	Thu, 22 Apr 2004 15:47:16 -0300 (ADT)
Received: from [192.168.0.2] (AMontsouris-108-1-20-220.w217-128.abo.wanadoo.fr
	[217.128.143.220])
	by mwinf0303.wanadoo.fr (SMTP Server) with ESMTP id 449AB5001438
	for ;
	Thu, 22 Apr 2004 20:47:16 +0200 (CEST)
Mime-Version: 1.0 (Apple Message framework v613)
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Resent-Date: Thu, 22 Apr 2004 20:47:16 +0200
Message-Id: <79BD69C1-948D-11D8-ACE1-000A95DE2550@ifrance.com>
Content-Transfer-Encoding: quoted-printable
Resent-To: pgsql-performance@postgresql.org
Resent-Message-Id: <6B54817E-948D-11D8-ACE1-000A95DE2550@ifrance.com>
From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= 
Subject: Re: 225 times slower 
Resent-From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= 
Date: Thu, 22 Apr 2004 20:46:51 +0200
To: Tom Lane 
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/386
X-Sequence-Number: 6686

> The planner is guessing that scanning in rec_id order will produce a
> matching row fairly quickly (sooner than selecting all the matching=20
> rows
> and sorting them would do).  It's wrong in this case, but I'm not sure
> it could do better without very detailed cross-column statistics.

> Am I
> right to guess that the rows that match the WHERE clause are not evenly
> distributed in the rec_id order, but rather there are no such rows till
> you get well up in the ordering?

I must agree that the data are not evenly distributed....

For table url:
count         271.395
min rec_id          1
max rec_id  3.386.962

dps=3D> select * from url where crc32=3D419903683;
count             852
min rec_id    264.374
max rec_id  2.392.046

I do
dps=3D> select ctid, rec_id from url where crc32=3D419903683 order by=20
crc32,rec_id;
And then in a text edit extract the "page_id" from ctid
and there is 409 distinct pages for the 852 rows.
There is 4592 pages for the tables url.

dps=3D> select (rec_id/25), count(*) from url where crc32=3D419903683 group=
=20
by rec_id/25 having count(*)>4 order by count(*) desc;
  ?column? | count
----------+-------
     30289 |    25
     11875 |    24
     11874 |    24
     11876 |    24
     28154 |    23
     26164 |    21
     26163 |    21
     55736 |    21
     40410 |    20
     47459 |    20
     30290 |    20
     28152 |    20
     26162 |    19
     30291 |    19
     37226 |    19
     60357 |    18
     28150 |    18
     12723 |    17
     40413 |    17
     40412 |    16
     33167 |    15
     40415 |    15
     12961 |    15
     40414 |    15
     28151 |    14
     63961 |    14
     26165 |    13
     11873 |    13
     63960 |    12
     37225 |    12
     37224 |    12
     20088 |    11
     30288 |    11
     91450 |    11
     20087 |    11
     26892 |    10
     47458 |    10
     40411 |    10
     91451 |    10
     12722 |    10
     28153 |     9
     43488 |     9
     60358 |     7
     60356 |     7
     11877 |     7
     33168 |     6
     91448 |     6
     26161 |     6
     40409 |     5
     28155 |     5
     28318 |     5
     30292 |     5
     26891 |     5
     95666 |     5
(54 rows)



An other question, with VACUUM VERBOSE ANALYZE, I see:
> INFO:  "url": removed 568107 row versions in 4592 pages
> DETAIL:  CPU 0.51s/1.17u sec elapsed 174.74 sec.
And I run pg_autovacuum.
Does the big number (568107) of removed row indicates I should set a=20
higher max_fsm_pages ?

 > grep fsm /var/pgsql/postgresql.conf
max_fsm_pages =3D 60000           # min max_fsm_relations*16, 6 bytes each
max_fsm_relations =3D 200         # min 100, ~50 bytes each

dps=3D> VACUUM VERBOSE ANALYSE url;
INFO:  vacuuming "public.url"
INFO:  index "url_crc" now contains 211851 row versions in 218 pages
DETAIL:  129292 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/1.38u sec elapsed 5.71 sec.
INFO:  index "url_seed" now contains 272286 row versions in 644 pages
DETAIL:  568107 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.10s/2.96u sec elapsed 13.06 sec.
INFO:  index "url_referrer" now contains 272292 row versions in 603=20
pages
DETAIL:  568107 index row versions were removed.
4 index pages have been deleted, 0 are currently reusable.
CPU 0.10s/2.98u sec elapsed 22.30 sec.
INFO:  index "url_next_index_time" now contains 272292 row versions in=20
684 pages
DETAIL:  568107 index row versions were removed.
42 index pages have been deleted, 0 are currently reusable.
CPU 0.07s/1.80u sec elapsed 9.50 sec.
INFO:  index "url_status" now contains 272298 row versions in 638 pages
DETAIL:  568107 index row versions were removed.
12 index pages have been deleted, 0 are currently reusable.
CPU 0.03s/2.18u sec elapsed 13.66 sec.
INFO:  index "url_bad_since_time" now contains 272317 row versions in=20
611 pages
DETAIL:  568107 index row versions were removed.
4 index pages have been deleted, 0 are currently reusable.
CPU 0.07s/2.40u sec elapsed 10.99 sec.
INFO:  index "url_hops" now contains 272317 row versions in 637 pages
DETAIL:  568107 index row versions were removed.
5 index pages have been deleted, 0 are currently reusable.
CPU 0.04s/2.24u sec elapsed 12.46 sec.
INFO:  index "url_siteid" now contains 272321 row versions in 653 pages
DETAIL:  568107 index row versions were removed.
13 index pages have been deleted, 0 are currently reusable.
CPU 0.14s/2.05u sec elapsed 11.63 sec.
INFO:  index "url_serverid" now contains 272321 row versions in 654=20
pages
DETAIL:  568107 index row versions were removed.
8 index pages have been deleted, 0 are currently reusable.
CPU 0.10s/2.27u sec elapsed 11.45 sec.
INFO:  index "url_url" now contains 272065 row versions in 1892 pages
DETAIL:  193884 index row versions were removed.
5 index pages have been deleted, 0 are currently reusable.
CPU 0.39s/1.50u sec elapsed 36.99 sec.
INFO:  index "url_last_mod_time" now contains 272071 row versions in=20
317 pages
DETAIL:  193884 index row versions were removed.
7 index pages have been deleted, 0 are currently reusable.
CPU 0.03s/1.38u sec elapsed 5.61 sec.
INFO:  index "url_pkey" now contains 272086 row versions in 328 pages
DETAIL:  193884 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.05s/1.60u sec elapsed 60.64 sec.
INFO:  "url": removed 568107 row versions in 4592 pages
DETAIL:  CPU 0.51s/1.17u sec elapsed 174.74 sec.
INFO:  "url": found 568107 removable, 272027 nonremovable row versions=20
in 4614 pages
DETAIL:  402 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 1.98s/26.08u sec elapsed 466.27 sec.
INFO:  vacuuming "pg_toast.pg_toast_137628026"
INFO:  index "pg_toast_137628026_index" now contains 0 row versions in=20
1 pages
DETAIL:  0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.06 sec.
INFO:  "pg_toast_137628026": found 0 removable, 0 nonremovable row=20
versions in 0 pages
DETAIL:  0 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.07 sec.
INFO:  analyzing "public.url"
INFO:  "url": 4624 pages, 150000 rows sampled, 577419 estimated total=20
rows
VACUUM

Cordialement,
Jean-G=E9rard Pailloncy


From pgsql-performance-owner@postgresql.org  Thu Apr 22 16:47:02 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E7D75D1D10E
	for ;
	Thu, 22 Apr 2004 16:22:55 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 91082-04
	for ;
	Thu, 22 Apr 2004 16:22:56 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 2B495D1D09C
	for ;
	Thu, 22 Apr 2004 16:22:54 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3MJMpav009414;
	Thu, 22 Apr 2004 15:22:52 -0400 (EDT)
To: Markus Bertheau 
Cc: Eduardo Almeida ,
	Jan Wieck , Josh Berkus ,
	pgsql-performance@postgresql.org
Subject: Re: [pgsql-advocacy] MySQL vs PG TPC-H benchmarks 
In-reply-to: <1082658047.4524.0.camel@yarrow.bertheau.de> 
References: <20040422125318.26053.qmail@web60601.mail.yahoo.com>
	<7444.1082649288@sss.pgh.pa.us>
	<1082658047.4524.0.camel@yarrow.bertheau.de>
Comments: In-reply-to Markus Bertheau 
	message dated "Thu, 22 Apr 2004 20:20:47 +0200"
Date: Thu, 22 Apr 2004 15:22:51 -0400
Message-ID: <9413.1082661771@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/385
X-Sequence-Number: 6685

Markus Bertheau  writes:
>> You could probably improve the index-create time by temporarily
>> increasing sort_mem.  It wouldn't be unreasonable to give CREATE INDEX
>> several hundred meg to work in.  (You don't want sort_mem that big
>> normally, because there may be many sorts happening in parallel,
>> but in a data-loading context there'll just be one active sort.)

> Doesn't this provide a reason for CREATE INDEX not to honour sort_mem?

Already done for 7.5.

http://archives.postgresql.org/pgsql-committers/2004-02/msg00025.php

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 22 20:12:09 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id AF422D1CCCA
	for ;
	Thu, 22 Apr 2004 18:20:38 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 29661-07
	for ;
	Thu, 22 Apr 2004 18:20:39 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id A5156D1CC7F
	for ;
	Thu, 22 Apr 2004 18:20:36 -0300 (ADT)
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 i3MLK1Il023568;
	Thu, 22 Apr 2004 15:20:02 -0600 (MDT)
Date: Thu, 22 Apr 2004 15:16:41 -0600 (MDT)
From: "scott.marlowe" 
To: Pallav Kalva 
Cc: 
Subject: Re: Setting Shared Buffers , Effective Cache, Sort Mem
In-Reply-To: <4088062E.2050901@deg.cc>
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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/396
X-Sequence-Number: 6696

On Thu, 22 Apr 2004, Pallav Kalva wrote:

> Hi
> 
>     We are in the process of building a new machine for our production 
> database. Below you will see some of the harware specs for the machine. 
> I need some help with setting these parameters (shared buffers, 
> effective cache, sort mem) in the pg_conf file.  Also can anyone explain 
> the difference between shared buffers and effective cache , how these 
> are allocated in the main memory (the docs are not clear on this).
> 
> Here are the Hardware details:
> Operating System: Red Hat 9
> Database Ver: Postgres 7.4
> CPU'S : 4
> RAM : 4 gig
> Datafile layout : RAID 1+0
> Transaction log : on different RAID1 Array
> RAID Stripe Size: 8k

Read this first:

http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html

Basically shared buffers are the "play area" for the database backends to 
toss data in the air and munge it together.  The effective cache size 
reflects the approximate amount of space your operating system is using to 
buffer Postgresql data.  On a dedicated database machine this is about the 
same as the size of the kernel buffer shown in top.  On a mixed machine, 
you'll have to see how much of what data is getting buffered to get a 
guesstimate of how much kernel cache is being used for pgsql and how much 
for other processes.  Then divide that number in bytes by 8192, the 
default block size.  On a machine with 1.2 gigs of kernel cache, that'd be 
about 150,000 blocks.

Buffer sizes from 1000 to 10000 blocks are common.  Block sizes from 10000 
to 50000 can somtimes increase performance, but those sizes only really 
make sense for machines with lots of ram, and very large datasets being 
operated on.


From pgsql-performance-owner@postgresql.org  Thu Apr 22 20:09:03 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 23BD0D1E0F1
	for ;
	Thu, 22 Apr 2004 18:21:40 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 33981-02
	for ;
	Thu, 22 Apr 2004 18:21:36 -0300 (ADT)
Received: from mx-2.sollentuna.net (mx-2.sollentuna.net [195.84.163.199])
	by svr1.postgresql.org (Postfix) with ESMTP id A6361D1DFA5
	for ;
	Thu, 22 Apr 2004 18:21:33 -0300 (ADT)
Received: from ALGOL.sollentuna.se (janus-en.sollentuna.se [195.84.163.194])
	by mx-2.sollentuna.net (Postfix) with ESMTP
	id 415348F2B4; Thu, 22 Apr 2004 23:21:34 +0200 (CEST)
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Thu, 22 Apr 2004 23:21:33 +0200
Message-ID: <6BCB9D8A16AC4241919521715F4D8BCE171658@algol.sollentuna.se>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [PERFORM] Wierd context-switching issue on Xeon patch for 7.4.1
Thread-Index: AcQorMHfolQ6yLncSkOOjBB/+NYT/gAAindg
From: "Magnus Hagander" 
To: , "Tom Lane" 
Cc: "Bruce Momjian" , ,
	"Paul Tuckfield" ,
	"Anjan Dave" , "Neil Conway" ,
	=?iso-8859-1?Q?Dirk_Lutzeb=E4ck?= ,
	
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/395
X-Sequence-Number: 6695

>> Having to recompile to run on single- vs dual-processor=20
>machines doesn't
>> seem like it would fly.
>
>Oh, I don't know.  Many applications require compiling for a target=20
>architecture; SQL Server, for example, won't use a 2nd=20
>processor without=20
>re-installation.   I'm not sure about Oracle.

Uh, that is not quite true - at leasdt not for current versions. SQL Server=
 will pick up and use whatever processors the underlying OS supports. Now, =
depending on how you install the OS (Windows, that is) you may have ended u=
p with a kernel and HAL that does not support multiprocessor. In this case,=
 you have to change HAL. But you certainly don't have to reinstalsl SQL Ser=
ver or Windows. Just a reboot (pretty normal when you add a CPU...)

Now, there can be licensing issues if you are in per-processor licensing, b=
ut that's a completely different issue. Also, the "Standard Edition" only u=
ses up to 4 CPUs, but again, that's a different issue.

//Magnus

From pgsql-performance-owner@postgresql.org  Thu Apr 22 20:07:59 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 42A5DD1CCCA
	for ;
	Thu, 22 Apr 2004 18:22:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 25992-09
	for ;
	Thu, 22 Apr 2004 18:22:23 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A515ED1CCA8
	for ;
	Thu, 22 Apr 2004 18:22:20 -0300 (ADT)
Received: from [134.22.70.203] (dyn-70-203.tor.dsl.tht.net [134.22.70.203])
	by tht.net (Postfix) with ESMTP
	id 1756376A2B; Thu, 22 Apr 2004 17:22:26 -0400 (EDT)
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Rod Taylor 
To: Tom Lane 
Cc: Josh Berkus ,
	Bruce Momjian , pg@fastcrypt.com,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	Postgresql Performance 
In-Reply-To: <8583.1082656553@sss.pgh.pa.us>
References: <200404220456.i3M4uFv04710@candle.pha.pa.us>
	<20171.1082610827@sss.pgh.pa.us> <200404221037.10097.josh@agliodbs.com>
	<8583.1082656553@sss.pgh.pa.us>
Content-Type: text/plain
Message-Id: <1082668930.94537.0.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Thu, 22 Apr 2004 17:22:11 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/394
X-Sequence-Number: 6694

On Thu, 2004-04-22 at 13:55, Tom Lane wrote:
> Josh Berkus  writes:
> > This may be a moot point, since you've stated that changing the loop timing 
> > won't solve the problem, but what about making the test part of make?   I 
> > don't think too many systems are going to change processor architectures once
> > in production, and those that do can be told to re-compile.
> 
> Having to recompile to run on single- vs dual-processor machines doesn't
> seem like it would fly.

Is it something the postmaster could quickly determine and set a global
during the startup cycle?



From pgsql-performance-owner@postgresql.org  Thu Apr 22 19:28:43 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E5105D1B560
	for ;
	Thu, 22 Apr 2004 18:57:16 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 45778-04
	for ;
	Thu, 22 Apr 2004 18:57:17 -0300 (ADT)
Received: from audiogram.mail.pas.earthlink.net
	(audiogram.mail.pas.earthlink.net [207.217.120.253])
	by svr1.postgresql.org (Postfix) with ESMTP id F375BD1B470
	for ;
	Thu, 22 Apr 2004 18:57:14 -0300 (ADT)
Received: from dpcbw098074.direcpc.com ([65.88.98.74] helo=earthlink.net)
	by audiogram.mail.pas.earthlink.net with asmtp (Exim 3.36 #4)
	id 1BGmC1-0006L2-00
	for pgsql-performance@postgresql.org; Thu, 22 Apr 2004 14:57:15 -0700
Message-ID: <40883FAB.3070109@earthlink.net>
Date: Thu, 22 Apr 2004 17:56:59 -0400
From: Sean Shanny 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Looking for ideas on how to speed up warehouse loading
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-ELNK-Trace: 
 6ca696a38bf90a1a1de288fa2098067a1aa676d7e74259b7b3291a7d08dfec792ac252445810def34516d470db6023d3350badd9bab72f9c350badd9bab72f9c
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/392
X-Sequence-Number: 6692

To all,

Essentials:  Running 7.4.1 on OSX on a loaded G5 with dual procs, 8GB 
memory, direct attached via fibre channel to a fully optioned 3.5TB 
XRaid (14 spindles, 2 sets of 7 in RAID 5) box running RAID 50.

Background:  We are loading what are essentially xml based access logs 
from about 20+ webservers daily, about 6GB of raw data.  We have a 
classic star schema.  All the ETL tools are custom java code or standard 
*nix tools like sort, uniq etc...

The problem:  We have about 46 million rows in a table with the 
following schema:

Table "public.d_referral"
       Column       |  Type   | Modifiers
--------------------+---------+-----------
 id                 | integer | not null
 referral_raw_url   | text    | not null
 job_control_number | integer | not null
Indexes:
    "d_referral_pkey" primary key, btree (id)
    "idx_referral_url" btree (referral_raw_url)

This is one of our dimension tables.  Part of the daily ETL process is 
to match all the new referral URL's against existing data in the 
d_referral table.  Some of the values in referral_raw_url can be 5000 
characters long :-( .  The avg length is :  109.57 characters.

I sort and uniq all the incoming referrals and load them into a temp table.

Table "public.referral_temp"
 Column | Type | Modifiers
--------+------+-----------
 url    | text | not null
Indexes:
    "referral_temp_pkey" primary key, btree (url)

I then do a left join

SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER JOIN d_referral t1 
ON t2.url = t1.referral_raw_url ORDER BY t1.id

This is the output from an explain analyze (Please note that I do a set 
enable_index_scan = false prior to issuing this because it takes forever 
using indexes.):

explain analyze SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER 
JOIN d_referral t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id;
                                                                      
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------------------
 Sort  (cost=4012064.81..4013194.45 rows=451856 width=115) (actual 
time=1297320.823..1297739.813 rows=476176 loops=1)
   Sort Key: t1.id
   ->  Hash Left Join  (cost=1052345.95..3969623.10 rows=451856 
width=115) (actual time=1146650.487..1290230.590 rows=476176 loops=1)
         Hash Cond: ("outer".url = "inner".referral_raw_url)
         ->  Seq Scan on referral_temp t2  (cost=0.00..6645.56 
rows=451856 width=111) (actual time=20.285..1449.634 rows=476176 loops=1)
         ->  Hash  (cost=729338.16..729338.16 rows=46034716 width=124) 
(actual time=1146440.710..1146440.710 rows=0 loops=1)
               ->  Seq Scan on d_referral t1  (cost=0.00..729338.16 
rows=46034716 width=124) (actual time=14.502..-1064277.123 rows=46034715 
loops=1)
 Total runtime: 1298153.193 ms
(8 rows)



What I would like to know is if there are better ways to do the join?  I 
need to get all the rows back from the referral_temp table as they are 
used for assigning FK's for the fact table later in processing.  When I 
iterate over the values that I get back those with t1.id = null I assign 
a new FK and push both into the d_referral table as new entries as well 
as a text file for later use.  The matching records are written to a 
text file for later use. 

If we cannot improve the join performance my question becomes are there 
better tools to match up the 46 million and growing at the rate of 1 
million every 3 days, strings outside of postgresql?  We don't want to 
have to invest in zillions of dollars worth of hardware but if we have 
to we will.  I just want to make sure we have all the non hardware 
possibilities for improvement covered before we start investing in large 
disk arrays. 

Thanks.

--sean

From pgsql-performance-owner@postgresql.org  Thu Apr 22 19:57:00 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D6A20D1CACD
	for ;
	Thu, 22 Apr 2004 19:48:13 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 60200-07
	for ;
	Thu, 22 Apr 2004 19:48:14 -0300 (ADT)
Received: from email06.aon.at (WARSL402PIP3.highway.telekom.at [195.3.96.75])
	by svr1.postgresql.org (Postfix) with SMTP id 85B5CD1BB74
	for ;
	Thu, 22 Apr 2004 19:48:11 -0300 (ADT)
Received: (qmail 141892 invoked from network); 22 Apr 2004 22:48:12 -0000
Received: from m169p017.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.11.17]) (envelope-sender )
	by qmail6rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 22 Apr 2004 22:48:12 -0000
From: Manfred Koizar 
To: Pallav Kalva 
Cc: pgsql-performance@postgresql.org
Subject: Re: Setting Shared Buffers , Effective Cache, Sort Mem Parameters
Date: Fri, 23 Apr 2004 00:50:37 +0200
Message-ID: 
References: <4088062E.2050901@deg.cc>
In-Reply-To: <4088062E.2050901@deg.cc>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/393
X-Sequence-Number: 6693

On Thu, 22 Apr 2004 13:51:42 -0400, Pallav Kalva  wrote:
>I need some help with setting these parameters (shared buffers, 
>effective cache, sort mem) in the pg_conf file.

It really depends on the kind of queries you intend to run, the number
of concurrent active connections, the size of the working set (active
part of the database), what else is running on the machine, and and and
...

Setting shared_buffers to 10000, effective_cache_size to 400000 (80% of
installed RAM), and sort_mem to a few thousand might be a good start.

>  Also can anyone explain 
>the difference between shared buffers and effective cache , how these 
>are allocated in the main memory (the docs are not clear on this).

Shared_buffers directly controls how many pages are allocated as
internal cache.  Effective_cache_size doesn't allocate anything, it is
just a hint to the planner how much cache is available on the system
level.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Mon Apr 26 14:51:28 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DCF19D1B470
	for ;
	Thu, 22 Apr 2004 20:13:35 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 66747-09
	for ;
	Thu, 22 Apr 2004 20:13:37 -0300 (ADT)
Received: from news.hub.org (news.hub.org [200.46.204.72])
	by svr1.postgresql.org (Postfix) with ESMTP id 9924CD1E104
	for ;
	Thu, 22 Apr 2004 20:13:34 -0300 (ADT)
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 i3MNDVSx074795
	for ; Thu, 22 Apr 2004 23:13:31 GMT
	(envelope-from news@news.hub.org)
Received: (from news@localhost)
	by news.hub.org (8.12.9/8.12.9/Submit) id i3MN4EWp073325
	for pgsql-performance@postgresql.org; Thu, 22 Apr 2004 23:04:14 GMT
From: CoL 
X-Newsgroups: comp.databases.postgresql.performance
Subject: Re: Looking for ideas on how to speed up warehouse loading
Date: Fri, 23 Apr 2004 01:05:20 +0200
Organization: Hub.Org Networking Services
Lines: 19
Message-ID: 
References: <40883FAB.3070109@earthlink.net>
Reply-To: col@mportal.hu
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:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
In-Reply-To: <40883FAB.3070109@earthlink.net>
To: pgsql-performance@postgresql.org
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/442
X-Sequence-Number: 6742

hi,

Sean Shanny wrote, On 4/22/2004 23:56:
> 
> SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER JOIN d_referral t1 
> ON t2.url = t1.referral_raw_url ORDER BY t1.id

index on url (text) has no sense. Try to use and md5 (char(32) column) 
which contains the md5 hash of url field. and join these ones. You can 
have a better index on this char 32 field.

do not forget to analyze the tables after data load, and you can fine 
tune you postgresql.conf, default_statistics_target for better index 
info, and others.
check this info pages:
http://www.varlena.com/varlena/GeneralBits/Tidbits/annotated_conf_e.html
http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html

C.

From pgsql-performance-owner@postgresql.org  Thu Apr 22 20:54:57 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C897FD1BC54
	for ;
	Thu, 22 Apr 2004 20:31:05 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 73749-08
	for ;
	Thu, 22 Apr 2004 20:31:07 -0300 (ADT)
Received: from mynah.mail.pas.earthlink.net (mynah.mail.pas.earthlink.net
	[207.217.120.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 60D2BD1BABA
	for ;
	Thu, 22 Apr 2004 20:31:03 -0300 (ADT)
Received: from dpcbw098074.direcpc.com ([65.88.98.74] helo=earthlink.net)
	by mynah.mail.pas.earthlink.net with asmtp (Exim 3.36 #4)
	id 1BGneo-0006jt-00
	for pgsql-performance@postgresql.org; Thu, 22 Apr 2004 16:31:03 -0700
Message-ID: <408855AD.4090500@earthlink.net>
Date: Thu, 22 Apr 2004 19:30:53 -0400
From: Sean Shanny 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: Looking for ideas on how to speed up warehouse loading
References: <40883FAB.3070109@earthlink.net>
In-Reply-To: <40883FAB.3070109@earthlink.net>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-ELNK-Trace: 
 6ca696a38bf90a1a1de288fa2098067a1aa676d7e74259b7b3291a7d08dfec79f9464016687b596ee7bf1ce92ae5b783350badd9bab72f9c350badd9bab72f9c
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/397
X-Sequence-Number: 6697

I should have included this as well:

 show all;
              name              |    setting
--------------------------------+----------------
 add_missing_from               | on
 australian_timezones           | off
 authentication_timeout         | 60
 check_function_bodies          | on
 checkpoint_segments            | 64
 checkpoint_timeout             | 30
 checkpoint_warning             | 30
 client_encoding                | UNICODE
 client_min_messages            | notice
 commit_delay                   | 0
 commit_siblings                | 5
 cpu_index_tuple_cost           | 0.001
 cpu_operator_cost              | 0.0025
 cpu_tuple_cost                 | 0.01
 DateStyle                      | ISO, MDY
 db_user_namespace              | off
 deadlock_timeout               | 1000
 debug_pretty_print             | off
 debug_print_parse              | off
 debug_print_plan               | off
 debug_print_rewritten          | off
 default_statistics_target      | 1000
 default_transaction_isolation  | read committed
 default_transaction_read_only  | off
 dynamic_library_path           | $libdir
 effective_cache_size           | 400000
 enable_hashagg                 | on
 enable_hashjoin                | on
 enable_indexscan               | on
 enable_mergejoin               | on
 enable_nestloop                | on
 enable_seqscan                 | on
 enable_sort                    | on
 enable_tidscan                 | on
 explain_pretty_print           | on
 extra_float_digits             | 0
 from_collapse_limit            | 8
 fsync                          | on
 geqo                           | on
 geqo_effort                    | 1
 geqo_generations               | 0
 geqo_pool_size                 | 0
 geqo_selection_bias            | 2
 geqo_threshold                 | 11
 join_collapse_limit            | 8
 krb_server_keyfile             | unset
 lc_collate                     | C
 lc_ctype                       | C
 lc_messages                    | C
 lc_monetary                    | C
 lc_numeric                     | C
 lc_time                        | C
 log_connections                | off
 log_duration                   | off
 log_error_verbosity            | default
 log_executor_stats             | off
 log_hostname                   | off
 log_min_duration_statement     | -1
 log_min_error_statement        | panic
 log_min_messages               | notice
 log_parser_stats               | off
 log_pid                        | off
 log_planner_stats              | off
 log_source_port                | off
 log_statement                  | off
 log_statement_stats            | off
 log_timestamp                  | on
 max_connections                | 100
 max_expr_depth                 | 10000
 max_files_per_process          | 1000
 max_fsm_pages                  | 20000
 max_fsm_relations              | 1000
 max_locks_per_transaction      | 64
 password_encryption            | on
 port                           | 5432
 pre_auth_delay                 | 0
 preload_libraries              | unset
 random_page_cost               | 4
 regex_flavor                   | advanced
 rendezvous_name                | unset
 search_path                    | $user,public
 server_encoding                | UNICODE
 server_version                 | 7.4.1
 shared_buffers                 | 4000
 silent_mode                    | off
 sort_mem                       | 64000
 sql_inheritance                | on
 ssl                            | off
 statement_timeout              | 0
 stats_block_level              | on
 stats_command_string           | on
 stats_reset_on_server_start    | off
 stats_row_level                | on
 stats_start_collector          | on
 superuser_reserved_connections | 2
 syslog                         | 0
 syslog_facility                | LOCAL0
 syslog_ident                   | postgres
 tcpip_socket                   | on
 TimeZone                       | unknown
 trace_notify                   | off
 transaction_isolation          | read committed
 transaction_read_only          | off
 transform_null_equals          | off
 unix_socket_directory          | unset
 unix_socket_group              | unset
 unix_socket_permissions        | 511
 vacuum_mem                     | 64000
 virtual_host                   | unset
 wal_buffers                    | 1024
 wal_debug                      | 0
 wal_sync_method                | open_sync
 zero_damaged_pages             | off


Sean Shanny wrote:

> To all,
>
> Essentials:  Running 7.4.1 on OSX on a loaded G5 with dual procs, 8GB 
> memory, direct attached via fibre channel to a fully optioned 3.5TB 
> XRaid (14 spindles, 2 sets of 7 in RAID 5) box running RAID 50.
>
> Background:  We are loading what are essentially xml based access logs 
> from about 20+ webservers daily, about 6GB of raw data.  We have a 
> classic star schema.  All the ETL tools are custom java code or 
> standard *nix tools like sort, uniq etc...
>
> The problem:  We have about 46 million rows in a table with the 
> following schema:
>
> Table "public.d_referral"
>       Column       |  Type   | Modifiers
> --------------------+---------+-----------
> id                 | integer | not null
> referral_raw_url   | text    | not null
> job_control_number | integer | not null
> Indexes:
>    "d_referral_pkey" primary key, btree (id)
>    "idx_referral_url" btree (referral_raw_url)
>
> This is one of our dimension tables.  Part of the daily ETL process is 
> to match all the new referral URL's against existing data in the 
> d_referral table.  Some of the values in referral_raw_url can be 5000 
> characters long :-( .  The avg length is :  109.57 characters.
>
> I sort and uniq all the incoming referrals and load them into a temp 
> table.
>
> Table "public.referral_temp"
> Column | Type | Modifiers
> --------+------+-----------
> url    | text | not null
> Indexes:
>    "referral_temp_pkey" primary key, btree (url)
>
> I then do a left join
>
> SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER JOIN d_referral 
> t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id
>
> This is the output from an explain analyze (Please note that I do a 
> set enable_index_scan = false prior to issuing this because it takes 
> forever using indexes.):
>
> explain analyze SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER 
> JOIN d_referral t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id;
>                                                                      
> QUERY PLAN
> ------------------------------------------------------------------------------------------------------------------------------------------------------ 
>
> Sort  (cost=4012064.81..4013194.45 rows=451856 width=115) (actual 
> time=1297320.823..1297739.813 rows=476176 loops=1)
>   Sort Key: t1.id
>   ->  Hash Left Join  (cost=1052345.95..3969623.10 rows=451856 
> width=115) (actual time=1146650.487..1290230.590 rows=476176 loops=1)
>         Hash Cond: ("outer".url = "inner".referral_raw_url)
>         ->  Seq Scan on referral_temp t2  (cost=0.00..6645.56 
> rows=451856 width=111) (actual time=20.285..1449.634 rows=476176 loops=1)
>         ->  Hash  (cost=729338.16..729338.16 rows=46034716 width=124) 
> (actual time=1146440.710..1146440.710 rows=0 loops=1)
>               ->  Seq Scan on d_referral t1  (cost=0.00..729338.16 
> rows=46034716 width=124) (actual time=14.502..-1064277.123 
> rows=46034715 loops=1)
> Total runtime: 1298153.193 ms
> (8 rows)
>
>
>
> What I would like to know is if there are better ways to do the join?  
> I need to get all the rows back from the referral_temp table as they 
> are used for assigning FK's for the fact table later in processing.  
> When I iterate over the values that I get back those with t1.id = null 
> I assign a new FK and push both into the d_referral table as new 
> entries as well as a text file for later use.  The matching records 
> are written to a text file for later use.
> If we cannot improve the join performance my question becomes are 
> there better tools to match up the 46 million and growing at the rate 
> of 1 million every 3 days, strings outside of postgresql?  We don't 
> want to have to invest in zillions of dollars worth of hardware but if 
> we have to we will.  I just want to make sure we have all the non 
> hardware possibilities for improvement covered before we start 
> investing in large disk arrays.
> Thanks.
>
> --sean
>
> ---------------------------(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 Apr 22 21:56:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D3237D1E11A
	for ;
	Thu, 22 Apr 2004 21:54:22 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 97491-10
	for ;
	Thu, 22 Apr 2004 21:54:24 -0300 (ADT)
Received: from mx.tripadvisor.com (unknown [151.203.96.35])
	by svr1.postgresql.org (Postfix) with ESMTP id 35E53D1E0ED
	for ;
	Thu, 22 Apr 2004 21:54:20 -0300 (ADT)
Received: from [192.168.0.102] (h0080c8aa8d54.ne.client2.attbi.com
	[24.147.6.138])
	by mx.tripadvisor.com (8.12.10/8.12.10) with SMTP id i3N0sFQ8093722;
	Thu, 22 Apr 2004 20:54:15 -0400 (EDT)
	(envelope-from nshanny@tripadvisor.com)
In-Reply-To: <408855AD.4090500@earthlink.net>
References: <40883FAB.3070109@earthlink.net> <408855AD.4090500@earthlink.net>
Mime-Version: 1.0 (Apple Message framework v613)
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
Message-Id: 
Content-Transfer-Encoding: 7bit
Cc: pgsql-performance@postgresql.org
From: Nicholas Shanny 
Subject: Re: Looking for ideas on how to speed up warehouse loading
Date: Thu, 22 Apr 2004 20:54:15 -0400
To: Sean Shanny 
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/398
X-Sequence-Number: 6698

One other thing: we are running with a block size of 32K.


Nick Shanny
(Brother of above person)

On Apr 22, 2004, at 7:30 PM, Sean Shanny wrote:

> I should have included this as well:
>
> show all;
>              name              |    setting
> --------------------------------+----------------
> add_missing_from               | on
> australian_timezones           | off
> authentication_timeout         | 60
> check_function_bodies          | on
> checkpoint_segments            | 64
> checkpoint_timeout             | 30
> checkpoint_warning             | 30
> client_encoding                | UNICODE
> client_min_messages            | notice
> commit_delay                   | 0
> commit_siblings                | 5
> cpu_index_tuple_cost           | 0.001
> cpu_operator_cost              | 0.0025
> cpu_tuple_cost                 | 0.01
> DateStyle                      | ISO, MDY
> db_user_namespace              | off
> deadlock_timeout               | 1000
> debug_pretty_print             | off
> debug_print_parse              | off
> debug_print_plan               | off
> debug_print_rewritten          | off
> default_statistics_target      | 1000
> default_transaction_isolation  | read committed
> default_transaction_read_only  | off
> dynamic_library_path           | $libdir
> effective_cache_size           | 400000
> enable_hashagg                 | on
> enable_hashjoin                | on
> enable_indexscan               | on
> enable_mergejoin               | on
> enable_nestloop                | on
> enable_seqscan                 | on
> enable_sort                    | on
> enable_tidscan                 | on
> explain_pretty_print           | on
> extra_float_digits             | 0
> from_collapse_limit            | 8
> fsync                          | on
> geqo                           | on
> geqo_effort                    | 1
> geqo_generations               | 0
> geqo_pool_size                 | 0
> geqo_selection_bias            | 2
> geqo_threshold                 | 11
> join_collapse_limit            | 8
> krb_server_keyfile             | unset
> lc_collate                     | C
> lc_ctype                       | C
> lc_messages                    | C
> lc_monetary                    | C
> lc_numeric                     | C
> lc_time                        | C
> log_connections                | off
> log_duration                   | off
> log_error_verbosity            | default
> log_executor_stats             | off
> log_hostname                   | off
> log_min_duration_statement     | -1
> log_min_error_statement        | panic
> log_min_messages               | notice
> log_parser_stats               | off
> log_pid                        | off
> log_planner_stats              | off
> log_source_port                | off
> log_statement                  | off
> log_statement_stats            | off
> log_timestamp                  | on
> max_connections                | 100
> max_expr_depth                 | 10000
> max_files_per_process          | 1000
> max_fsm_pages                  | 20000
> max_fsm_relations              | 1000
> max_locks_per_transaction      | 64
> password_encryption            | on
> port                           | 5432
> pre_auth_delay                 | 0
> preload_libraries              | unset
> random_page_cost               | 4
> regex_flavor                   | advanced
> rendezvous_name                | unset
> search_path                    | $user,public
> server_encoding                | UNICODE
> server_version                 | 7.4.1
> shared_buffers                 | 4000
> silent_mode                    | off
> sort_mem                       | 64000
> sql_inheritance                | on
> ssl                            | off
> statement_timeout              | 0
> stats_block_level              | on
> stats_command_string           | on
> stats_reset_on_server_start    | off
> stats_row_level                | on
> stats_start_collector          | on
> superuser_reserved_connections | 2
> syslog                         | 0
> syslog_facility                | LOCAL0
> syslog_ident                   | postgres
> tcpip_socket                   | on
> TimeZone                       | unknown
> trace_notify                   | off
> transaction_isolation          | read committed
> transaction_read_only          | off
> transform_null_equals          | off
> unix_socket_directory          | unset
> unix_socket_group              | unset
> unix_socket_permissions        | 511
> vacuum_mem                     | 64000
> virtual_host                   | unset
> wal_buffers                    | 1024
> wal_debug                      | 0
> wal_sync_method                | open_sync
> zero_damaged_pages             | off
>
>
> Sean Shanny wrote:
>
>> To all,
>>
>> Essentials:  Running 7.4.1 on OSX on a loaded G5 with dual procs, 8GB  
>> memory, direct attached via fibre channel to a fully optioned 3.5TB  
>> XRaid (14 spindles, 2 sets of 7 in RAID 5) box running RAID 50.
>>
>> Background:  We are loading what are essentially xml based access  
>> logs from about 20+ webservers daily, about 6GB of raw data.  We have  
>> a classic star schema.  All the ETL tools are custom java code or  
>> standard *nix tools like sort, uniq etc...
>>
>> The problem:  We have about 46 million rows in a table with the  
>> following schema:
>>
>> Table "public.d_referral"
>>       Column       |  Type   | Modifiers
>> --------------------+---------+-----------
>> id                 | integer | not null
>> referral_raw_url   | text    | not null
>> job_control_number | integer | not null
>> Indexes:
>>    "d_referral_pkey" primary key, btree (id)
>>    "idx_referral_url" btree (referral_raw_url)
>>
>> This is one of our dimension tables.  Part of the daily ETL process  
>> is to match all the new referral URL's against existing data in the  
>> d_referral table.  Some of the values in referral_raw_url can be 5000  
>> characters long :-( .  The avg length is :  109.57 characters.
>>
>> I sort and uniq all the incoming referrals and load them into a temp  
>> table.
>>
>> Table "public.referral_temp"
>> Column | Type | Modifiers
>> --------+------+-----------
>> url    | text | not null
>> Indexes:
>>    "referral_temp_pkey" primary key, btree (url)
>>
>> I then do a left join
>>
>> SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER JOIN d_referral  
>> t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id
>>
>> This is the output from an explain analyze (Please note that I do a  
>> set enable_index_scan = false prior to issuing this because it takes  
>> forever using indexes.):
>>
>> explain analyze SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER  
>> JOIN d_referral t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id;
>>                                                                       
>> QUERY PLAN
>> ---------------------------------------------------------------------- 
>> ---------------------------------------------------------------------- 
>> ----------
>> Sort  (cost=4012064.81..4013194.45 rows=451856 width=115) (actual  
>> time=1297320.823..1297739.813 rows=476176 loops=1)
>>   Sort Key: t1.id
>>   ->  Hash Left Join  (cost=1052345.95..3969623.10 rows=451856  
>> width=115) (actual time=1146650.487..1290230.590 rows=476176 loops=1)
>>         Hash Cond: ("outer".url = "inner".referral_raw_url)
>>         ->  Seq Scan on referral_temp t2  (cost=0.00..6645.56  
>> rows=451856 width=111) (actual time=20.285..1449.634 rows=476176  
>> loops=1)
>>         ->  Hash  (cost=729338.16..729338.16 rows=46034716 width=124)  
>> (actual time=1146440.710..1146440.710 rows=0 loops=1)
>>               ->  Seq Scan on d_referral t1  (cost=0.00..729338.16  
>> rows=46034716 width=124) (actual time=14.502..-1064277.123  
>> rows=46034715 loops=1)
>> Total runtime: 1298153.193 ms
>> (8 rows)
>>
>>
>>
>> What I would like to know is if there are better ways to do the join?  
>>  I need to get all the rows back from the referral_temp table as they  
>> are used for assigning FK's for the fact table later in processing.   
>> When I iterate over the values that I get back those with t1.id =  
>> null I assign a new FK and push both into the d_referral table as new  
>> entries as well as a text file for later use.  The matching records  
>> are written to a text file for later use.
>> If we cannot improve the join performance my question becomes are  
>> there better tools to match up the 46 million and growing at the rate  
>> of 1 million every 3 days, strings outside of postgresql?  We don't  
>> want to have to invest in zillions of dollars worth of hardware but  
>> if we have to we will.  I just want to make sure we have all the non  
>> hardware possibilities for improvement covered before we start  
>> investing in large disk arrays.
>> Thanks.
>>
>> --sean
>>
>> ---------------------------(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
>>
>
> ---------------------------(end of  
> broadcast)---------------------------
> TIP 8: explain analyze is your friend
>


From pgsql-performance-owner@postgresql.org  Thu Apr 22 23:03:52 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D9464D1E14F
	for ;
	Thu, 22 Apr 2004 23:03:12 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 16770-03
	for ;
	Thu, 22 Apr 2004 23:03:15 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id B6A64D1E147
	for ;
	Thu, 22 Apr 2004 23:03:09 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3N23Cav014460;
	Thu, 22 Apr 2004 22:03:12 -0400 (EDT)
To: Sean Shanny 
Cc: pgsql-performance@postgresql.org
Subject: Re: Looking for ideas on how to speed up warehouse loading 
In-reply-to: <40883FAB.3070109@earthlink.net> 
References: <40883FAB.3070109@earthlink.net>
Comments: In-reply-to Sean Shanny 
	message dated "Thu, 22 Apr 2004 17:56:59 -0400"
Date: Thu, 22 Apr 2004 22:03:12 -0400
Message-ID: <14459.1082685792@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/399
X-Sequence-Number: 6699

Sean Shanny  writes:
> explain analyze SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER 
> JOIN d_referral t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id;
                                                                      
> QUERY PLAN
> ------------------------------------------------------------------------------------------------------------------------------------------------------
>  Sort  (cost=4012064.81..4013194.45 rows=451856 width=115) (actual 
> time=1297320.823..1297739.813 rows=476176 loops=1)
>    Sort Key: t1.id
>    ->  Hash Left Join  (cost=1052345.95..3969623.10 rows=451856 
> width=115) (actual time=1146650.487..1290230.590 rows=476176 loops=1)
>          Hash Cond: ("outer".url = "inner".referral_raw_url)
>          ->  Seq Scan on referral_temp t2  (cost=0.00..6645.56 
> rows=451856 width=111) (actual time=20.285..1449.634 rows=476176 loops=1)
>          ->  Hash  (cost=729338.16..729338.16 rows=46034716 width=124) 
> (actual time=1146440.710..1146440.710 rows=0 loops=1)
>                ->  Seq Scan on d_referral t1  (cost=0.00..729338.16 
> rows=46034716 width=124) (actual time=14.502..-1064277.123 rows=46034715 
> loops=1)
>  Total runtime: 1298153.193 ms
> (8 rows)

> What I would like to know is if there are better ways to do the join?

What have you got sort_mem set to?  You might try increasing it to a gig
or so, since you seem to have plenty of RAM in that box ...

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 22 23:28:59 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A569FD1E148
	for ;
	Thu, 22 Apr 2004 23:27:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 18953-09
	for ;
	Thu, 22 Apr 2004 23:27:56 -0300 (ADT)
Received: from vt-pe2550-001.VANTAGE.vantage.com (unknown [64.80.203.242])
	by svr1.postgresql.org (Postfix) with ESMTP id BA0F5D1E144
	for ;
	Thu, 22 Apr 2004 23:27:51 -0300 (ADT)
X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: multipart/mixed;
 boundary="----_=_NextPart_001_01C428DA.95C5DCB2"
Subject: Re: Wierd context-switching issue on Xeon
Date: Thu, 22 Apr 2004 22:27:55 -0400
Message-ID: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF785098157@vt-pe2550-001.vantage.vantage.com>
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
Thread-Topic: [PERFORM] Wierd context-switching issue on Xeon
Thread-Index: AcQnQtAtXqbOYp+aQp2pIsOo9mab0wBlwiJV
From: "Anjan Dave" 
To: "Josh Berkus" ,
	=?utf-8?Q?Dirk_Lutzeb=C3=A4ck?= ,
	"Tom Lane" 
Cc: ,
	"Neil Conway" 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/400
X-Sequence-Number: 6700

This is a multi-part message in MIME format.

------_=_NextPart_001_01C428DA.95C5DCB2
Content-Type: text/plain;
	charset="utf-8"
Content-Transfer-Encoding: base64
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------_=_NextPart_001_01C428DA.95C5DCB2
Content-Type: text/richtext;
	name="Context-Switching.rtf"
Content-Transfer-Encoding: base64
Content-Description: Context-Switching.rtf
Content-Disposition: attachment;
	filename="Context-Switching.rtf"
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------_=_NextPart_001_01C428DA.95C5DCB2--

From pgsql-performance-owner@postgresql.org  Fri Apr 23 01:39:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 74907D1D077
	for ;
	Fri, 23 Apr 2004 01:39:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 56323-04
	for ;
	Fri, 23 Apr 2004 01:39:16 -0300 (ADT)
Received: from joeconway.com (66-146-172-82.skyriver.net [66.146.172.82])
	by svr1.postgresql.org (Postfix) with ESMTP id A384DD1CCCA
	for ;
	Fri, 23 Apr 2004 01:39:15 -0300 (ADT)
Received: from [192.168.4.3] (account jconway HELO joeconway.com)
	by joeconway.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP-TLS id 1519993; Thu, 22 Apr 2004 21:33:27 -0700
Message-ID: <40889DAD.6080703@joeconway.com>
Date: Thu, 22 Apr 2004 21:38:05 -0700
From: Joe Conway 
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.6) Gecko/20040113
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Sean Shanny 
Cc: pgsql-performance@postgresql.org
Subject: Re: Looking for ideas on how to speed up warehouse loading
References: <40883FAB.3070109@earthlink.net>
In-Reply-To: <40883FAB.3070109@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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/401
X-Sequence-Number: 6701

Sean Shanny wrote:
> explain analyze SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER 
> JOIN d_referral t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id;

> What I would like to know is if there are better ways to do the join?  I 
> need to get all the rows back from the referral_temp table as they are 
> used for assigning FK's for the fact table later in processing.  When I 
> iterate over the values that I get back those with t1.id = null I assign 
> a new FK and push both into the d_referral table as new entries as well 
> as a text file for later use.  The matching records are written to a 
> text file for later use.

Would something like this work any better (without disabling index scans):

SELECT t1.id, t2.url
FROM referral_temp t2, d_referral t1
WHERE t1.referral_raw_url = t2.url;



SELECT t1.id, t2.url
FROM referral_temp t2
WHERE NOT EXISTS
(select 1 FROM d_referral t1 WHERE t1.referral_raw_url = t2.url);



?

Joe

From pgsql-performance-owner@postgresql.org  Fri Apr 23 09:19:43 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 58242D1C9E8
	for ;
	Fri, 23 Apr 2004 09:19:35 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 86772-07
	for ;
	Fri, 23 Apr 2004 09:19:35 -0300 (ADT)
Received: from hotmail.com (bay18-dav4.bay18.hotmail.com [65.54.187.184])
	by svr1.postgresql.org (Postfix) with ESMTP id 79ACFD1C9B7
	for ;
	Fri, 23 Apr 2004 09:19:31 -0300 (ADT)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	Fri, 23 Apr 2004 05:19:30 -0700
Received: from 67.81.102.201 by bay18-dav4.bay18.hotmail.com with DAV;
	Fri, 23 Apr 2004 12:19:30 +0000
X-Originating-IP: [67.81.102.201]
X-Originating-Email: [awerman2@hotmail.com]
X-Sender: awerman2@hotmail.com
From: "Aaron Werman" 
To: "Joe Conway" ,
	"Sean Shanny" 
Cc: 
References: <40883FAB.3070109@earthlink.net> <40889DAD.6080703@joeconway.com>
Subject: Re: Looking for ideas on how to speed up warehouse loading
Date: Fri, 23 Apr 2004 08:19:36 -0400
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 6.00.2800.1409
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409
Message-ID: 
X-OriginalArrivalTime: 23 Apr 2004 12:19:30.0871 (UTC)
	FILETIME=[3AB95C70:01C4292D]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=RCVD_IN_NJABL,
	RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/402
X-Sequence-Number: 6702

By definition, it is equivalent to:

SELECT t1.id, t2.url FROM referral_temp t2 LEFT /*OUTER*/ JOIN d_referral t1
ON t2.url = t1.referral_raw_url
union all
SELECT null, url FROM referral_temp WHERE url is null
ORDER BY 1;



/Aaron

----- Original Message ----- 
From: "Joe Conway" 
To: "Sean Shanny" 
Cc: 
Sent: Friday, April 23, 2004 12:38 AM
Subject: Re: [PERFORM] Looking for ideas on how to speed up warehouse
loading


> Sean Shanny wrote:
> > explain analyze SELECT t1.id, t2.url FROM referral_temp t2 LEFT OUTER
> > JOIN d_referral t1 ON t2.url = t1.referral_raw_url ORDER BY t1.id;
>
> > What I would like to know is if there are better ways to do the join?  I
> > need to get all the rows back from the referral_temp table as they are
> > used for assigning FK's for the fact table later in processing.  When I
> > iterate over the values that I get back those with t1.id = null I assign
> > a new FK and push both into the d_referral table as new entries as well
> > as a text file for later use.  The matching records are written to a
> > text file for later use.
>
> Would something like this work any better (without disabling index scans):
>
> SELECT t1.id, t2.url
> FROM referral_temp t2, d_referral t1
> WHERE t1.referral_raw_url = t2.url;
>
> 
>
> SELECT t1.id, t2.url
> FROM referral_temp t2
> WHERE NOT EXISTS
> (select 1 FROM d_referral t1 WHERE t1.referral_raw_url = t2.url);
>
> 
>
> ?
>
> Joe
>
> ---------------------------(end of broadcast)---------------------------
> TIP 7: don't forget to increase your free space map settings
>

From pgsql-performance-owner@postgresql.org  Fri Apr 23 10:32:09 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9CAA9D1BA98
	for ;
	Fri, 23 Apr 2004 10:31:19 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 16875-01
	for ;
	Fri, 23 Apr 2004 10:31:21 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id 1EBB5D1B98D
	for ;
	Fri, 23 Apr 2004 10:31:16 -0300 (ADT)
thread-index: AcQpN7xPOA4ilzVLQ6yUbo1GCcCURw==
Content-Class: urn:content-classes:message
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
Received: from [10.4.2.155] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Fri, 23 Apr 2004 09:34:42 -0400
From: "Chris Hoover" 
To: 
Subject: Help with performance problems
Date: Fri, 23 Apr 2004 09:31:17 -0400
User-Agent: KMail/1.6.1
Cc: 
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404230931.17850.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 23 Apr 2004 13:34:43.0156 (UTC)
	FILETIME=[BC416940:01C42937]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/403
X-Sequence-Number: 6703

I need some help.  I have 5 db servers running our database servers, and th=
ey=20
all are having various degrees of performance problems.  The problems we ar=
e=20
experiencing are:

1.  General slowness
2.  High loads

All of our db's are running on Dell Poweredge 2650 with 2 P4 Xeons (2.8 ->=
=20
3.06 GHz) with 8 to 12 GB of memory.  The databases are running on attached=
=20
Dell Powervault 220s running raid5.

The databases were created and taken into production before I started worki=
ng=20
here and are very flat.  Most of the major tables have a combined primary k=
ey=20
using an int field and a single char field.  There are some additional=20
indexes on some tables.  Most queries I see in the  logs are running at les=
s=20
than .01 seconds with many significantly slower.

We are trying to narrow down the performance problem to either the db or th=
e=20
hardware.  As the dba, I need to  try and get these db's tuned to the best=
=20
possible way considering the current db state.  We are in the beginning of =
a=20
complete db redesign and application re-write, but the completion and=20
deployment of the new db and app are quite a ways off.

Anyway, we are running the following:
PE 2650 w/ 2 cpus (2.8-3.06) - HT on
8-12 GB memory
OS on raid 0
DB's on Powervaults 220S using raid 5 (over 6 disks)
Each Postgresql cluster has 2 db up to almost 170db's (project to level out=
=20
the num of db's/cluster is being started)
DB's are no bigger than a few GB in size (largest is about 11GB according t=
o a=20
du -h)
Running RH ES 2.1

Here is the postgresql.conf from the server with the 11GB db:

max_connections =3D 64
shared_buffers =3D 32768		# 256MB=3D32768(buffs)*8192(bytes/buff)
max_fsm_relations =3D 1000	# min 10, fsm is free space map, ~40 bytes
max_fsm_pages =3D 10000		# min 1000, fsm is free space map, ~6 bytes
sort_mem =3D 4096			# 256MB=3D4096(bytes/proc)*64(procs or conns)
checkpoint_segments =3D 16	# in logfile segments, min 1, 16MB each
checkpoint_timeout =3D 30		# range 30-3600, in seconds
effective_cache_size =3D 131072	# typically 8KB each
log_connections =3D true
log_pid =3D true
log_statement =3D true
log_duration =3D true
log_timestamp =3D true
stats_start_collector =3D true
stats_reset_on_server_start =3D true
stats_command_string =3D true
stats_row_level =3D true
stats_block_level =3D true
LC_MESSAGES =3D 'en_US'
LC_MONETARY =3D 'en_US'
LC_NUMERIC =3D 'en_US'
LC_TIME =3D 'en_US'

Here is top (server running pretty good right now)
  9:28am  up 25 days, 16:02,  2 users,  load average: 0.54, 0.33, 0.22
94 processes: 91 sleeping, 3 running, 0 zombie, 0 stopped
CPU0 states: 64.0% user,  0.1% system,  0.0% nice, 34.0% idle
CPU1 states: 29.0% user,  9.0% system,  0.0% nice, 60.0% idle
CPU2 states:  2.0% user,  0.1% system,  0.0% nice, 96.0% idle
CPU3 states:  0.0% user,  0.0% system,  0.0% nice, 100.0% idle
Mem:  7720072K av, 7711648K used,    8424K free,  265980K shrd,  749888K bu=
ff
Swap: 2096440K av,   22288K used, 2074152K free                 6379304K=20
cached

Here is top from another server (with the most db's):=20
 9:31am  up 25 days, 16:05,  5 users,  load average: 2.34, 3.39, 4.28
147 processes: 145 sleeping, 2 running, 0 zombie, 0 stopped
CPU0 states:  6.0% user,  1.0% system,  0.0% nice, 91.0% idle
CPU1 states:  9.0% user,  4.0% system,  0.0% nice, 85.0% idle
CPU2 states:  9.0% user,  3.0% system,  0.0% nice, 86.0% idle
CPU3 states:  9.0% user,  4.0% system,  0.0% nice, 85.0% idle
Mem:  7721096K av, 7708040K used,   13056K free,  266132K shrd, 3151336K bu=
ff
Swap: 2096440K av,   24208K used, 2072232K free                 3746596K=20
cached

Thanks for any help/advice,

Chris


From pgsql-performance-owner@postgresql.org  Fri Apr 23 13:16:02 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 31DE8D1B911
	for ;
	Fri, 23 Apr 2004 12:16:11 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 48035-08
	for ;
	Fri, 23 Apr 2004 12:16:09 -0300 (ADT)
Received: from hotmail.com (bay18-dav43.bay18.hotmail.com [65.54.187.223])
	by svr1.postgresql.org (Postfix) with ESMTP id 99028D1B91B
	for ;
	Fri, 23 Apr 2004 12:16:08 -0300 (ADT)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	Fri, 23 Apr 2004 08:16:07 -0700
Received: from 67.81.102.201 by bay18-dav43.bay18.hotmail.com with DAV;
	Fri, 23 Apr 2004 15:16:07 +0000
X-Originating-IP: [67.81.102.201]
X-Originating-Email: [awerman2@hotmail.com]
X-Sender: awerman2@hotmail.com
From: "Aaron Werman" 
To: "Chris Hoover" ,
	
Cc: 
References: <200404230931.17850.revoohc@sermonaudio.com>
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 11:16:13 -0400
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 6.00.2800.1409
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1409
Message-ID: 
X-OriginalArrivalTime: 23 Apr 2004 15:16:07.0654 (UTC)
	FILETIME=[E6E62060:01C42945]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.2 tagged_above=0.0 required=5.0 tests=RCVD_IN_NJABL,
	RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/404
X-Sequence-Number: 6704

Your second server has queuing (load averages are highish), only 2 processes
running, and almost all cycles are idle. You need to track down your
bottleneck. Have you looked at iostat/vmstat? I think it would be useful to
post these, ideally both before and after full vacuum analyze.

/Aaron


----- Original Message ----- 
From: "Chris Hoover" 
To: 
Cc: 
Sent: Friday, April 23, 2004 9:31 AM
Subject: [PERFORM] Help with performance problems


I need some help.  I have 5 db servers running our database servers, and
they
all are having various degrees of performance problems.  The problems we are
experiencing are:

1.  General slowness
2.  High loads

All of our db's are running on Dell Poweredge 2650 with 2 P4 Xeons (2.8 ->
3.06 GHz) with 8 to 12 GB of memory.  The databases are running on attached
Dell Powervault 220s running raid5.

The databases were created and taken into production before I started
working
here and are very flat.  Most of the major tables have a combined primary
key
using an int field and a single char field.  There are some additional
indexes on some tables.  Most queries I see in the  logs are running at less
than .01 seconds with many significantly slower.

We are trying to narrow down the performance problem to either the db or the
hardware.  As the dba, I need to  try and get these db's tuned to the best
possible way considering the current db state.  We are in the beginning of a
complete db redesign and application re-write, but the completion and
deployment of the new db and app are quite a ways off.

Anyway, we are running the following:
PE 2650 w/ 2 cpus (2.8-3.06) - HT on
8-12 GB memory
OS on raid 0
DB's on Powervaults 220S using raid 5 (over 6 disks)
Each Postgresql cluster has 2 db up to almost 170db's (project to level out
the num of db's/cluster is being started)
DB's are no bigger than a few GB in size (largest is about 11GB according to
a
du -h)
Running RH ES 2.1

Here is the postgresql.conf from the server with the 11GB db:

max_connections = 64
shared_buffers = 32768 # 256MB=32768(buffs)*8192(bytes/buff)
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
sort_mem = 4096 # 256MB=4096(bytes/proc)*64(procs or conns)
checkpoint_segments = 16 # in logfile segments, min 1, 16MB each
checkpoint_timeout = 30 # range 30-3600, in seconds
effective_cache_size = 131072 # typically 8KB each
log_connections = true
log_pid = true
log_statement = true
log_duration = true
log_timestamp = true
stats_start_collector = true
stats_reset_on_server_start = true
stats_command_string = true
stats_row_level = true
stats_block_level = true
LC_MESSAGES = 'en_US'
LC_MONETARY = 'en_US'
LC_NUMERIC = 'en_US'
LC_TIME = 'en_US'

Here is top (server running pretty good right now)
  9:28am  up 25 days, 16:02,  2 users,  load average: 0.54, 0.33, 0.22
94 processes: 91 sleeping, 3 running, 0 zombie, 0 stopped
CPU0 states: 64.0% user,  0.1% system,  0.0% nice, 34.0% idle
CPU1 states: 29.0% user,  9.0% system,  0.0% nice, 60.0% idle
CPU2 states:  2.0% user,  0.1% system,  0.0% nice, 96.0% idle
CPU3 states:  0.0% user,  0.0% system,  0.0% nice, 100.0% idle
Mem:  7720072K av, 7711648K used,    8424K free,  265980K shrd,  749888K
buff
Swap: 2096440K av,   22288K used, 2074152K free                 6379304K
cached

Here is top from another server (with the most db's):
 9:31am  up 25 days, 16:05,  5 users,  load average: 2.34, 3.39, 4.28
147 processes: 145 sleeping, 2 running, 0 zombie, 0 stopped
CPU0 states:  6.0% user,  1.0% system,  0.0% nice, 91.0% idle
CPU1 states:  9.0% user,  4.0% system,  0.0% nice, 85.0% idle
CPU2 states:  9.0% user,  3.0% system,  0.0% nice, 86.0% idle
CPU3 states:  9.0% user,  4.0% system,  0.0% nice, 85.0% idle
Mem:  7721096K av, 7708040K used,   13056K free,  266132K shrd, 3151336K
buff
Swap: 2096440K av,   24208K used, 2072232K free                 3746596K
cached

Thanks for any help/advice,

Chris


---------------------------(end of broadcast)---------------------------
TIP 2: you can get off all lists at once with the unregister command
    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)

From pgsql-performance-owner@postgresql.org  Fri Apr 23 13:29:06 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 71789D1C9B4
	for ;
	Fri, 23 Apr 2004 13:29:05 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 75194-10
	for ;
	Fri, 23 Apr 2004 13:29:03 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id 5B870D1BB4C
	for ;
	Fri, 23 Apr 2004 13:29:02 -0300 (ADT)
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 i3NGSRIl004047;
	Fri, 23 Apr 2004 10:28:27 -0600 (MDT)
Date: Fri, 23 Apr 2004 10:25:01 -0600 (MDT)
From: "scott.marlowe" 
To: Sean Shanny 
Cc: 
Subject: Re: Looking for ideas on how to speed up warehouse loading
In-Reply-To: <408855AD.4090500@earthlink.net>
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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/405
X-Sequence-Number: 6705


On Thu, 22 Apr 2004, Sean Shanny wrote:

> I should have included this as well:
>  fsync                          | on
>  shared_buffers                 | 4000
>  sort_mem                       | 64000

For purposes of loading only, you can try turning off fsync, assuming this 
is a virgin load and you can just re-initdb should bad things happen (OS, 
postgresql crash, power plug pulled, etc...)

Also increasing sort_mem and shared_buffers might help.  Especially 
sort_mem.  But turn it back down to something reasonable after the import.

And turn fsync back on after the import too.  Note you have to restart 
postgresql to make fsync = off take effect.


From pgsql-performance-owner@postgresql.org  Fri Apr 23 13:43:41 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 0E824D1C9F3
	for ;
	Fri, 23 Apr 2004 13:43:33 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 86750-05
	for ;
	Fri, 23 Apr 2004 13:43:32 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id DE36AD1BAD5
	for ;
	Fri, 23 Apr 2004 13:43:30 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4929127; Fri, 23 Apr 2004 09:44:55 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: "Chris Hoover" ,
	
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 09:42:48 -0700
User-Agent: KMail/1.4.3
Cc: 
References: <200404230931.17850.revoohc@sermonaudio.com>
In-Reply-To: <200404230931.17850.revoohc@sermonaudio.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404230942.48290.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/406
X-Sequence-Number: 6706

Chris,

> I need some help.  I have 5 db servers running our database servers, and
> they all are having various degrees of performance problems.  The problems
> we are experiencing are:

I'mm confused.  You're saying "general slowness" but say that most queries run 
in under .01 seconds.   And you say "high loads" but the TOP snapshots you 
provide show servers with 2 CPUs idle.   

Are you sure you actually *have* a performance issue?

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Fri Apr 23 14:16:12 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id BFD5DD1B98D
	for ;
	Fri, 23 Apr 2004 14:16:08 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 97714-03
	for ;
	Fri, 23 Apr 2004 14:16:06 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id CE93ED1B91B
	for ;
	Fri, 23 Apr 2004 14:16:04 -0300 (ADT)
thread-index: AcQpVyJlIsB8KxyHSGGj2MXxoKHPug==
Content-Class: urn:content-classes:message
Received: from [10.4.2.155] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Fri, 23 Apr 2004 13:19:28 -0400
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
From: "Chris Hoover" 
To: "Josh Berkus" 
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 13:16:03 -0400
User-Agent: KMail/1.6.1
Cc: 
References: <200404230931.17850.revoohc@sermonaudio.com>
	<200404230942.48290.josh@agliodbs.com>
In-Reply-To: <200404230942.48290.josh@agliodbs.com>
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404231316.03447.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 23 Apr 2004 17:19:28.0843 (UTC)
	FILETIME=[2259F1B0:01C42957]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/407
X-Sequence-Number: 6707

I know the numbers look ok, but we are definetly suffering.  Also, if I try=
 to=20
run any sort of vacuum or other db activity during normal business hours,=
=20
load goes through the roof.  I have seen loads of over 10 when trying to=20
vacuum the larger cluster and would have to kill the vacuums due to=20
complaints.=20=20

I think this is probably related to the hardware configuration, but I want =
to=20
make sure that there are no changes I could make configuration wise to the =
db=20
that might lighten the problem.

I'm especially want to make sure that I have the memory parameters set to g=
ood=20
numbers for my db's so that I can minimize thrashing between the postgres=
=20
memory pools and the hard drive.  I am thinking that this may be a big issu=
e=20
here?

Thanks for any help,

Chris
On Friday 23 April 2004 12:42, Josh Berkus wrote:
> Chris,
>
> > I need some help.  I have 5 db servers running our database servers, and
> > they all are having various degrees of performance problems.  The
> > problems we are experiencing are:
>
> I'mm confused.  You're saying "general slowness" but say that most queries
> run in under .01 seconds.   And you say "high loads" but the TOP snapshots
> you provide show servers with 2 CPUs idle.
>
> Are you sure you actually *have* a performance issue?


From pgsql-performance-owner@postgresql.org  Fri Apr 23 14:25:38 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 42A21D1D63E
	for ;
	Fri, 23 Apr 2004 14:25:20 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 96510-08
	for ;
	Fri, 23 Apr 2004 14:25:17 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id A03CAD1D370
	for ;
	Fri, 23 Apr 2004 14:25:16 -0300 (ADT)
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 i3NHOxIl010294;
	Fri, 23 Apr 2004 11:24:59 -0600 (MDT)
Date: Fri, 23 Apr 2004 11:21:32 -0600 (MDT)
From: "scott.marlowe" 
To: Chris Hoover 
Cc: 
Subject: Re: Help with performance problems
In-Reply-To: <200404230931.17850.revoohc@sermonaudio.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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/408
X-Sequence-Number: 6708

On Fri, 23 Apr 2004, Chris Hoover wrote:

> DB's on Powervaults 220S using raid 5 (over 6 disks)

What controller is this, the adaptec?  We've found it to be slower than 
the LSI megaraid based controller, but YMMV.

> Running RH ES 2.1

Are you running the latest kernel for ES 2.1?  Early 2.4 kernels are 
pretty pokey and have some odd behaviour under load that later 2.4 
kernels seemed to fix.

> Here is the postgresql.conf from the server with the 11GB db:
> 
> max_connections = 64
> shared_buffers = 32768		# 256MB=32768(buffs)*8192(bytes/buff)
> 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

IF you're doing lots of updates and such, you might want these higher.
Have you vacuumed full the databases since taking over?

> sort_mem = 4096			# 256MB=4096(bytes/proc)*64(procs or conns)

Sorry, that's wrong.  sort_mem is measure in kbytes.  i.e. 8192 means 8 
megs sort_mem.  Try setting it a bit higher (you've got LOTS of ram in these 
boxes) to something like 16 or 32 meg.

> checkpoint_segments = 16	# in logfile segments, min 1, 16MB each
> checkpoint_timeout = 30		# range 30-3600, in seconds
> effective_cache_size = 131072	# typically 8KB each

This still looks low.  On one machine you're showing kernel cache of about 
.7 gig, on the other it's 6 gig.  6 gigs of kernel cache would be a 
setting of 800000.  It's more of a nudge factor than an exact science, so 
don't worry too much.

If you've got fast I/O look at lowering random page cost to something 
between 1 and 2.  We use 1.3 to 1.4 on most of our machines with fast 
drives under them.

I'd use vmstat to see if you're I/O bound.  

also, look for index bloat.  Before 7.4 it was a serious problem.  With 
7.4 regular vacuuming should reclaim most lost space, but there are corner 
cases where you still might need to re-index.


From pgsql-performance-owner@postgresql.org  Fri Apr 23 14:53:24 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CF359D1B4B6
	for ;
	Fri, 23 Apr 2004 14:53:22 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 15815-01
	for ;
	Fri, 23 Apr 2004 14:53:20 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id C24A2D1C4EC
	for ;
	Fri, 23 Apr 2004 14:53:18 -0300 (ADT)
thread-index: AcQpXFZpF1kw81j3T6iDrHUVhr260A==
Content-Class: urn:content-classes:message
Received: from [10.4.2.155] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Fri, 23 Apr 2004 13:56:43 -0400
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
From: "Chris Hoover" 
To: 
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 13:53:17 -0400
User-Agent: KMail/1.6.1
References: <200404230931.17850.revoohc@sermonaudio.com>
	<200404231316.03447.revoohc@sermonaudio.com>
	<4089561B.8010001@speedfc.com>
In-Reply-To: <4089561B.8010001@speedfc.com>
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404231353.17234.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 23 Apr 2004 17:56:43.0515 (UTC)
	FILETIME=[5651D4B0:01C4295C]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/409
X-Sequence-Number: 6709

Sorry for the confusion here.  I can't run any sort of vacuum durin the day=
=20
due to performance hits.  However, I have run vacuums at night.  Several=20
nights a week I run a vacuumdb -f -z  on all of the clusters.  I can take=
=20
serveral hours to complete, but it does complete.

During the day, I have tried to run a vacuumdb -v and a vacuumdb -z -v duri=
ng=20
the day since I read it is supposed to help performance, but as I said, it=
=20
causes to much of a stress on the system.

I did change the vacuumdb script to do set the vacuum_mem to 512 when=20
vacuuming to try and help the situation (from the script: ${PATHNAME}psql=
=20
$PSQLOPT $ECHOOPT -c "SET vacuum_mem=3D524288;SET autocommit TO 'on';VACUUM=
=20
$full $verbose $analyze $table" -d $db ), and I reset it to 8192 at the end.

Anyway, thank you for the ideas so far, and any additional will be greatly=
=20
appreciated.

Chris
On Friday 23 April 2004 13:44, Kevin Barnard wrote:
> Chris Hoover wrote:
> >I know the numbers look ok, but we are definetly suffering.  Also, if I
> > try to run any sort of vacuum or other db activity during normal busine=
ss
> > hours, load goes through the roof.  I have seen loads of over 10 when
> > trying to vacuum the larger cluster and would have to kill the vacuums
> > due to complaints.
>
> This is your problem then.  You have to regularly vacuum the DB.  You
> might want to dump and reload or schedule a vacuum full.  If you don't
> it doesn't matter what you do you will never get decent performance.
> Make sure you vacuum as a superuser this way you get system tables as wel=
l.
>
> Killing a vacuum is bad it tends to make the situation worse.  If you
> need to vaccuum one table at a time.
>
> >I think this is probably related to the hardware configuration, but I wa=
nt
> > to make sure that there are no changes I could make configuration wise =
to
> > the db that might lighten the problem.
> >
> >I'm especially want to make sure that I have the memory parameters set to
> > good numbers for my db's so that I can minimize thrashing between the
> > postgres memory pools and the hard drive.  I am thinking that this may =
be
> > a big issue here?
>
> Get the vacuum done and don't worry about the hardware or the settings
> until afterwords.


From pgsql-performance-owner@postgresql.org  Fri Apr 23 15:01:40 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E7D47D1BA98
	for ;
	Fri, 23 Apr 2004 15:01:34 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 15115-06
	for ;
	Fri, 23 Apr 2004 15:01:33 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id 6D806D1BAD5
	for ;
	Fri, 23 Apr 2004 15:01:30 -0300 (ADT)
thread-index: AcQpXXtoDiNBWFjzQYyCZJaeKYDeqg==
Content-Class: urn:content-classes:message
Received: from [10.4.2.155] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Fri, 23 Apr 2004 14:04:54 -0400
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
From: "Chris Hoover" 
To: "scott.marlowe" 
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 14:01:29 -0400
User-Agent: KMail/1.6.1
Cc: 
References: 
In-Reply-To: 
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404231401.29144.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 23 Apr 2004 18:04:55.0031 (UTC)
	FILETIME=[7B493870:01C4295D]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/410
X-Sequence-Number: 6710

On Friday 23 April 2004 13:21, scott.marlowe wrote:
> On Fri, 23 Apr 2004, Chris Hoover wrote:
> > DB's on Powervaults 220S using raid 5 (over 6 disks)
>
> What controller is this, the adaptec?  We've found it to be slower than
> the LSI megaraid based controller, but YMMV.
>
We are using the perc3/di controller.  Believe it is using the megaraid=20
driver.
> > Running RH ES 2.1
>
> Are you running the latest kernel for ES 2.1?  Early 2.4 kernels are
> pretty pokey and have some odd behaviour under load that later 2.4
> kernels seemed to fix.
>
I'm not sure we are at the latest and greatest for 2.1, but I am trying to =
get=20
there.  Management won't let me do the upgrade w/o first testing/proving it=
=20
will not cause any more issues.  Due to all of the current issues, and the=
=20
criticality of these systems to our bottom line, they are being very carefu=
l=20
with any change that may impact our users further.

We are waiting on our datacenter to plug in our test server and powervault =
so=20
that we can test the upgrades the the latest RH 2.1 kernel.
> > Here is the postgresql.conf from the server with the 11GB db:
> >
> > max_connections =3D 64
> > shared_buffers =3D 32768		# 256MB=3D32768(buffs)*8192(bytes/buff)
> > max_fsm_relations =3D 1000	# min 10, fsm is free space map, ~40 bytes
> > max_fsm_pages =3D 10000		# min 1000, fsm is free space map, ~6 bytes
>
> IF you're doing lots of updates and such, you might want these higher.
> Have you vacuumed full the databases since taking over?
>
> > sort_mem =3D 4096			# 256MB=3D4096(bytes/proc)*64(procs or conns)
>
> Sorry, that's wrong.  sort_mem is measure in kbytes.  i.e. 8192 means 8
> megs sort_mem.  Try setting it a bit higher (you've got LOTS of ram in
> these boxes) to something like 16 or 32 meg.
>
> > checkpoint_segments =3D 16	# in logfile segments, min 1, 16MB each
> > checkpoint_timeout =3D 30		# range 30-3600, in seconds
> > effective_cache_size =3D 131072	# typically 8KB each
>
> This still looks low.  On one machine you're showing kernel cache of about
> .7 gig, on the other it's 6 gig.  6 gigs of kernel cache would be a
> setting of 800000.  It's more of a nudge factor than an exact science, so
> don't worry too much.
I believe changing this requires a restart of the cluster (correct?).  If s=
o,=20
I'll try bumping up the effective_cache_size over the weekend.

Also, will all of the memory available to these machines, should I be runni=
ng=20
with larger shared_buffers?  It seems like 256M is a bit small.
>
> If you've got fast I/O look at lowering random page cost to something
> between 1 and 2.  We use 1.3 to 1.4 on most of our machines with fast
> drives under them.
>
> I'd use vmstat to see if you're I/O bound.
>
If we end up being I/O bound, should the random page cost be set higher?

> also, look for index bloat.  Before 7.4 it was a serious problem.  With
> 7.4 regular vacuuming should reclaim most lost space, but there are corner
> cases where you still might need to re-index.
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
>
>                http://archives.postgresql.org
Thanks for the help,

Chris


From pgsql-performance-owner@postgresql.org  Fri Apr 23 15:16:13 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C2BE4D1C9F3
	for ;
	Fri, 23 Apr 2004 15:16:08 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 22035-03
	for ;
	Fri, 23 Apr 2004 15:16:07 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id B1E40D1C9B7
	for ;
	Fri, 23 Apr 2004 15:16:05 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4929755; Fri, 23 Apr 2004 11:17:30 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: "Chris Hoover" ,
	
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 11:15:23 -0700
User-Agent: KMail/1.4.3
References: <200404230931.17850.revoohc@sermonaudio.com>
	<4089561B.8010001@speedfc.com>
	<200404231353.17234.revoohc@sermonaudio.com>
In-Reply-To: <200404231353.17234.revoohc@sermonaudio.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404231115.23198.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/411
X-Sequence-Number: 6711

Chris,

> Sorry for the confusion here.  I can't run any sort of vacuum durin the day
> due to performance hits.  However, I have run vacuums at night.  Several
> nights a week I run a vacuumdb -f -z  on all of the clusters.  I can take
> serveral hours to complete, but it does complete.

Well, here's your first problem:  since your FSM pages is low, and you're only 
vacuuming once a day, you've got to have some serious table and index bloat.   
SO you're going to need to do VACUUM FULL on all of your databases, and then 
REINDEX on all of your indexes.

After that, raise your max_fsm_pages to something useful, like 1,000,000.   Of 
course, data on your real rate of updates would help more.

If you're getting severe disk choke when you vacuum, you probably are I/O 
bound.   You may want to try something which allows you to vacuum one table 
at a time, either pg_autovacuum or a custom script.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Fri Apr 23 15:19:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 11580D1BACD
	for ;
	Fri, 23 Apr 2004 15:19:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 23856-02
	for ;
	Fri, 23 Apr 2004 15:19:15 -0300 (ADT)
Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca
	[199.212.17.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 845C9D1BAD5
	for ;
	Fri, 23 Apr 2004 15:19:05 -0300 (ADT)
Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19)
	id ; Fri, 23 Apr 2004 14:19:05 -0400
Message-ID: 
From: "Shea,Dan [CIS]" 
To: 'Christopher Kings-Lynne' 
Cc: pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Fri, 23 Apr 2004 14:19:04 -0400
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/412
X-Sequence-Number: 6712


PWFPM_DEV=# select * from pg_locks;
 relation | database | transaction |  pid  |           mode           |
granted
----------+----------+-------------+-------+--------------------------+-----
----
    17472 |    17347 |             |  2618 | ShareUpdateExclusiveLock | t
          |          |    10858533 | 28778 | ExclusiveLock            | t
    17472 |    17347 |             |  2618 | ShareUpdateExclusiveLock | t
          |          |    10803814 |  2618 | ExclusiveLock            | t
    16759 |    17347 |             | 28778 | AccessShareLock          | t
(5 rows)

PWFPM_DEV=#

17347 is the database PWFPM_DEV iod, The pids are below

[root@murphy root]# ps -ef |grep 28778|grep -v "grep"
postgres 28778   504  0 18:06 ?        00:00:00 postgres: scores PWFPM_DEV
[local] idle
[root@murphy root]# ps -ef |grep 2618|grep -v "grep"
postgres  2618   504  8 Apr22 ?        02:31:00 postgres: postgres PWFPM_DEV
[local] VACUUM
[root@murphy root]#
A vacuum is running now.  I restarted the database, set vacuum_mem =
'196608'; and started a new vacuum.  I also stopped inserting into the
database.
I hoping I will get some results.

PWFPM_DEV=# select now();vacuum verbose analyze forecastelement;select
now();
              now
-------------------------------
 2004-04-22 13:38:02.083592+00
(1 row)

INFO:  vacuuming "public.forecastelement"
INFO:  index "forecastelement_rwv_idx" now contains 391385895 row versions
in 5051132 pages
DETAIL:  27962015 index row versions were removed.
771899 index pages have been deleted, 496872 are currently reusable.
CPU 4499.54s/385.76u sec elapsed 55780.91 sec.
INFO:  "forecastelement": removed 33554117 row versions in 737471 pages
DETAIL:  CPU 135.61s/83.99u sec elapsed 1101.26 sec.
-----Original Message-----
From: Christopher Kings-Lynne [mailto:chriskl@familyhealth.com.au]
Sent: Tuesday, April 20, 2004 9:26 PM
To: Shea,Dan [CIS]
Cc: pgsql-performance@postgresql.org
Subject: Re: [PERFORM] Why will vacuum not end?


> No, but data is constantly being inserted by userid scores.  It is
postgres
> runnimg the vacuum.
> Dan.

Well, inserts create some locks - perhaps that's the problem...

Otherwise, check the pg_locks view to see if you can figure it out.

Chris


---------------------------(end of broadcast)---------------------------
TIP 2: you can get off all lists at once with the unregister command
    (send "unregister YourEmailAddressHere" to majordomo@postgresql.org)

From pgsql-performance-owner@postgresql.org  Fri Apr 23 15:48:47 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F354ED1C4EB
	for ;
	Fri, 23 Apr 2004 15:48:45 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36150-01
	for ;
	Fri, 23 Apr 2004 15:48:43 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 7845AD1BCB9
	for ;
	Fri, 23 Apr 2004 15:48:41 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4929948; Fri, 23 Apr 2004 11:50:05 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: "Shea,Dan [CIS]" ,
	'Christopher Kings-Lynne' 
Subject: Re: Why will vacuum not end?
Date: Fri, 23 Apr 2004 11:47:58 -0700
User-Agent: KMail/1.4.3
Cc: pgsql-performance@postgresql.org
References: 
In-Reply-To: 
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404231147.58279.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/413
X-Sequence-Number: 6713

Guys,

> Well, inserts create some locks - perhaps that's the problem...
>
> Otherwise, check the pg_locks view to see if you can figure it out.

FWIW, I've had this happen a couple of times, too.   Unfortunately, it's 
happend in the middle of the day so that I had to cancel the processes and 
get the system back to normal in too much of a hurry to consider documenting 
it.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Fri Apr 23 15:56:12 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B6C9AD1B8C0
	for ;
	Fri, 23 Apr 2004 15:56:10 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 27570-10
	for ;
	Fri, 23 Apr 2004 15:56:09 -0300 (ADT)
Received: from silmaril.syscor.priv (h24-108-208-253.sbm.shawcable.net
	[24.108.208.253])
	by svr1.postgresql.org (Postfix) with ESMTP id 189DCD1C50D
	for ;
	Fri, 23 Apr 2004 15:56:06 -0300 (ADT)
Received: from syscor.com (strider.syscor.priv [192.168.1.3])
	by silmaril.syscor.priv (8.12.8/8.12.8) with ESMTP id i3NIu8XR014585;
	Fri, 23 Apr 2004 11:56:09 -0700
Message-ID: <40896713.2010703@syscor.com>
Date: Fri, 23 Apr 2004 11:57:23 -0700
From: Ron St-Pierre 
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.4) Gecko/20030624
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: revoohc@sermonaudio.com, pgsql-performance@postgresql.org
Subject: Re: Help with performance problems
References: <200404230931.17850.revoohc@sermonaudio.com>
	<4089561B.8010001@speedfc.com>
	<200404231353.17234.revoohc@sermonaudio.com>
	<200404231115.23198.josh@agliodbs.com>
In-Reply-To: <200404231115.23198.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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/414
X-Sequence-Number: 6714

Josh Berkus wrote:

>Chris,
>
>  
>
>>Sorry for the confusion here.  I can't run any sort of vacuum durin the day
>>due to performance hits.  However, I have run vacuums at night.  Several
>>nights a week I run a vacuumdb -f -z  on all of the clusters.  I can take
>>serveral hours to complete, but it does complete.
>>    
>>
>
>Well, here's your first problem:  since your FSM pages is low, and you're only 
>vacuuming once a day, you've got to have some serious table and index bloat.   
>SO you're going to need to do VACUUM FULL on all of your databases, and then 
>REINDEX on all of your indexes.
>
>After that, raise your max_fsm_pages to something useful, like 1,000,000.   Of 
>course, data on your real rate of updates would help more.
>
>If you're getting severe disk choke when you vacuum, you probably are I/O 
>bound.   You may want to try something which allows you to vacuum one table 
>at a time, either pg_autovacuum or a custom script.
>
>  
>
Tom and Josh recently gave me some help about setting the fsm settings 
which was quite useful. The full message is at
http://archives.postgresql.org/pgsql-performance/2004-04/msg00229.php
and the 'most interesting' posrtion was:

 Actually, since he's running 7.4, there's an even better way.  Do a
 "VACUUM VERBOSE" (full-database vacuum --- doesn't matter whether you
 ANALYZE or not).  At the end of the very voluminous output, you'll see
 something like


 INFO:  free space map: 240 relations, 490 pages stored; 4080 total pages needed
 DETAIL:  Allocated FSM size: 1000 relations + 20000 pages = 178 kB shared memory.


 Here, I would need max_fsm_relations = 240 and max_fsm_pages = 4080 to
 exactly cover the present freespace needs of my system.  I concur with
 the suggestion to bump that up a good deal, of course, but that gives
 you a real number to start from.


 The DETAIL part of the message shows my current settings (which are the
 defaults) and what the FSM is costing me in shared memory space.

Good luck
Ron




From pgsql-performance-owner@postgresql.org  Fri Apr 23 16:27:31 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F03BBD1D073
	for ;
	Fri, 23 Apr 2004 16:27:28 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 49811-01
	for ;
	Fri, 23 Apr 2004 16:27:27 -0300 (ADT)
Received: from mail.astudios.com (unknown [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id 2414ED1C9E8
	for ;
	Fri, 23 Apr 2004 16:27:25 -0300 (ADT)
thread-index: AcQpaXvsoqCzdrFlSuChyRKvDuPyjQ==
Content-Class: urn:content-classes:message
Received: from [10.4.2.155] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Fri, 23 Apr 2004 15:30:49 -0400
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
From: "Chris Hoover" 
To: "Ron St-Pierre" 
Subject: Re: Help with performance problems
Date: Fri, 23 Apr 2004 15:27:24 -0400
User-Agent: KMail/1.6.1
Cc: 
References: <200404230931.17850.revoohc@sermonaudio.com>
	<200404231115.23198.josh@agliodbs.com>
	<40896713.2010703@syscor.com>
In-Reply-To: <40896713.2010703@syscor.com>
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404231527.24230.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 23 Apr 2004 19:30:49.0968 (UTC)
	FILETIME=[7BDE2F00:01C42969]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/415
X-Sequence-Number: 6715

On Friday 23 April 2004 14:57, Ron St-Pierre wrote:
Does this apply to 7.3.4 also?
>  Actually, since he's running 7.4, there's an even better way.  Do a
>  "VACUUM VERBOSE" (full-database vacuum --- doesn't matter whether you
>  ANALYZE or not).  At the end of the very voluminous output, you'll see
>  something like
>
>
>  INFO:  free space map: 240 relations, 490 pages stored; 4080 total pages
> needed DETAIL:  Allocated FSM size: 1000 relations + 20000 pages =3D 178 =
kB
> shared memory.
>
>
>  Here, I would need max_fsm_relations =3D 240 and max_fsm_pages =3D 4080 =
to
>  exactly cover the present freespace needs of my system.  I concur with
>  the suggestion to bump that up a good deal, of course, but that gives
>  you a real number to start from.
>
>
>  The DETAIL part of the message shows my current settings (which are the
>  defaults) and what the FSM is costing me in shared memory space.
>
> Good luck
> Ron
>
>
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 7: don't forget to increase your free space map settings


From pgsql-performance-owner@postgresql.org  Fri Apr 23 17:02:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 38FE0D1B498
	for ;
	Fri, 23 Apr 2004 17:02:46 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 56008-07
	for ;
	Fri, 23 Apr 2004 17:02:46 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id A62B1D1B47D
	for ;
	Fri, 23 Apr 2004 17:02:42 -0300 (ADT)
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 i3NK2JIl026378;
	Fri, 23 Apr 2004 14:02:19 -0600 (MDT)
Date: Fri, 23 Apr 2004 13:58:51 -0600 (MDT)
From: "scott.marlowe" 
To: Chris Hoover 
Cc: 
Subject: Re: Help with performance problems
In-Reply-To: <200404231401.29144.revoohc@sermonaudio.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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/416
X-Sequence-Number: 6716

On Fri, 23 Apr 2004, Chris Hoover wrote:

> On Friday 23 April 2004 13:21, scott.marlowe wrote:
> > On Fri, 23 Apr 2004, Chris Hoover wrote:
> > > DB's on Powervaults 220S using raid 5 (over 6 disks)
> >
> > What controller is this, the adaptec?  We've found it to be slower than
> > the LSI megaraid based controller, but YMMV.
> >
> We are using the perc3/di controller.  Believe it is using the megaraid 
> driver.

No, that's the adaptec, the PERC3/DC is the lsi megaraid.  See if there 
are newer drivers for the RAID card.  In terms of performance, the adaptec 
and lsi drivers have improved considerably in later versions.  In terms of 
stability they've largely gotten better with a few in between releases on 
the megaraid getting poor grades.  The latest / greatest from Dell is 
pretty up to date.

> > > Running RH ES 2.1
> >
> > Are you running the latest kernel for ES 2.1?  Early 2.4 kernels are
> > pretty pokey and have some odd behaviour under load that later 2.4
> > kernels seemed to fix.
> >
> I'm not sure we are at the latest and greatest for 2.1, but I am trying to get 
> there.  Management won't let me do the upgrade w/o first testing/proving it 
> will not cause any more issues.  Due to all of the current issues, and the 
> criticality of these systems to our bottom line, they are being very careful 
> with any change that may impact our users further.

Understood.  It's why my production box is still running a 2.4 kernel on 
rh 7.2 with pg 7.2.  They just work, but for us stability AND performance 
are both good with our load.

You can install a new kernel and set up the machine to still boot off of 
the old one, and test on the weekend to see how it behaves under 
simulated load.  Mining the logs for slow queries is a good way to build 
one.

while we don't upgrade our production server's applications to the latest 
and greatest all the time (i.e. php or postgresql or openldap) we always 
run the latest security patches, and I think the latest kernels had 
security fixes for ES 2.1, so NOT upgrading it dangerous.  Late model 
linux kernels (the 2.0.x and 2.2.x where x>20) tend to be VERY stable and 
very conservatively backported and upgraded, so running a new one isn't 
usually a big risk.

> > > Here is the postgresql.conf from the server with the 11GB db:
> > >
> > > max_connections = 64
> > > shared_buffers = 32768		# 256MB=32768(buffs)*8192(bytes/buff)
> > > 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
> >
> > IF you're doing lots of updates and such, you might want these higher.
> > Have you vacuumed full the databases since taking over?
> >
> > > sort_mem = 4096			# 256MB=4096(bytes/proc)*64(procs or conns)
> >
> > Sorry, that's wrong.  sort_mem is measure in kbytes.  i.e. 8192 means 8
> > megs sort_mem.  Try setting it a bit higher (you've got LOTS of ram in
> > these boxes) to something like 16 or 32 meg.
> >
> > > checkpoint_segments = 16	# in logfile segments, min 1, 16MB each
> > > checkpoint_timeout = 30		# range 30-3600, in seconds
> > > effective_cache_size = 131072	# typically 8KB each
> >
> > This still looks low.  On one machine you're showing kernel cache of about
> > .7 gig, on the other it's 6 gig.  6 gigs of kernel cache would be a
> > setting of 800000.  It's more of a nudge factor than an exact science, so
> > don't worry too much.
> I believe changing this requires a restart of the cluster (correct?).  If so, 
> I'll try bumping up the effective_cache_size over the weekend.
> 
> Also, will all of the memory available to these machines, should I be running 
> with larger shared_buffers?  It seems like 256M is a bit small.

No, you probably shouldn't.  PostgreSQL doesn't "cache" in the classical 
sense.   If all backends close, the stuff they had in their buffers 
disappears in a flash.  So, it's generally considered better to let the 
kernel do the bulk of the caching, and having the buffer area be large 
enough to hold a large portion, if not all, of your working set of data.  
But between the cache management which is dirt simple and works but seems 
to have performance issues with large numbers of buffers, and the fact 
that all the memory in it disappears when the last backend using it.

for instance, in doing the following seq scan select:

explain analyze select * from test;

where test is a ~10 megabyte table, the first time I ran it it took 5 
seconds to run.  The second time took it 2.5, the third 1.9, and it 
levelled out around there.  Starting up another backend and running the 
same query got a 1.9 second response also.  Shutting down both 
connections, and running the query again, with only the kernel for 
caching, I got 1.9.

That's on a 2.4.2[2-4] kernel.

> > If you've got fast I/O look at lowering random page cost to something
> > between 1 and 2.  We use 1.3 to 1.4 on most of our machines with fast
> > drives under them.
> >
> > I'd use vmstat to see if you're I/O bound.
> >
> If we end up being I/O bound, should the random page cost be set higher?

Not necessarily.  Often times on a machine with a lot of memory, you are 
better off using index scans where disk seek time would be expensive, but 
with indexes in ram, the page cost in comparison to seq pages is almost 1, 
with a slight overhead cost.  So, lowering the random page cost favors 
indexes, generally.  If your I/O subsystem is doing a lot of seq scans, 
when only part of the data set is ever really being worked on, this tends 
to flush out the kernel cache, and we wind up going back to disk over and 
over.  On the other hand, if your data is normally going to be 
sequentially accessed, then you'll have to invest in better RAID hardware 
/ more drives etc...

but with 12 gigs on one box, and an already reasonably fast I/O subsystem 
in place, I'd think a lower random page cost would help, not hurt 
performance.

Have you explain analyzed your slower queries?



From pgsql-performance-owner@postgresql.org  Fri Apr 23 18:35:16 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 7DC4CD1D2BC
	for ;
	Fri, 23 Apr 2004 18:35:14 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 90610-01
	for ;
	Fri, 23 Apr 2004 18:35:14 -0300 (ADT)
Received: from silmaril.syscor.priv (h24-108-208-253.sbm.shawcable.net
	[24.108.208.253])
	by svr1.postgresql.org (Postfix) with ESMTP id B30D3D1D26C
	for ;
	Fri, 23 Apr 2004 18:35:10 -0300 (ADT)
Received: from syscor.com (strider.syscor.priv [192.168.1.3])
	by silmaril.syscor.priv (8.12.8/8.12.8) with ESMTP id i3NLZGXR015000;
	Fri, 23 Apr 2004 14:35:17 -0700
Message-ID: <40898C5F.7070004@syscor.com>
Date: Fri, 23 Apr 2004 14:36:31 -0700
From: Ron St-Pierre 
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.4) Gecko/20030624
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Chris Hoover ,
	pgsql-performance@postgresql.org
Subject: Re: Help with performance problems
References: <200404230931.17850.revoohc@sermonaudio.com>
	<200404231115.23198.josh@agliodbs.com>
	<40896713.2010703@syscor.com>
	<200404231527.24230.revoohc@sermonaudio.com>
In-Reply-To: <200404231527.24230.revoohc@sermonaudio.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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/417
X-Sequence-Number: 6717

Chris Hoover wrote:

>On Friday 23 April 2004 14:57, Ron St-Pierre wrote:
>Does this apply to 7.3.4 also?
>
No it doesn't, I didn't look back through the thread far enough to see 
what you were running. I tried it on 7.3.4 and none of the summary info 
listed below was returned.  FWIW one of our DBs was slowing down 
considerably on an update (30+ minutes) and after I changed 
max_fsm_pages from the 7.4 default of 20,000 to 50,000, it completed in 
about eight minutes.

Ron

>  
>
>> Actually, since he's running 7.4, there's an even better way.  Do a
>> "VACUUM VERBOSE" (full-database vacuum --- doesn't matter whether you
>> ANALYZE or not).  At the end of the very voluminous output, you'll see
>> something like
>>
>>
>> INFO:  free space map: 240 relations, 490 pages stored; 4080 total pages
>>needed DETAIL:  Allocated FSM size: 1000 relations + 20000 pages = 178 kB
>>shared memory.
>>
>>
>> Here, I would need max_fsm_relations = 240 and max_fsm_pages = 4080 to
>> exactly cover the present freespace needs of my system.  I concur with
>> the suggestion to bump that up a good deal, of course, but that gives
>> you a real number to start from.
>>
>>
>> The DETAIL part of the message shows my current settings (which are the
>> defaults) and what the FSM is costing me in shared memory space.
>>
>>Good luck
>>Ron
>>
>>
>>
>>
>>---------------------------(end of broadcast)---------------------------
>>TIP 7: don't forget to increase your free space map settings
>>    
>>
>
>
>
>  
>



From pgsql-performance-owner@postgresql.org  Fri Apr 23 18:42:36 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D5C7DD1D08A
	for ;
	Fri, 23 Apr 2004 18:42:33 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 88557-07
	for ;
	Fri, 23 Apr 2004 18:42:34 -0300 (ADT)
Received: from email06.aon.at (WARSL402PIP3.highway.telekom.at [195.3.96.75])
	by svr1.postgresql.org (Postfix) with SMTP id 5256AD1D076
	for ;
	Fri, 23 Apr 2004 18:42:31 -0300 (ADT)
Received: (qmail 364292 invoked from network); 23 Apr 2004 21:42:30 -0000
Received: from m161p025.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.10.25]) (envelope-sender )
	by qmail6rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 23 Apr 2004 21:42:30 -0000
From: Manfred Koizar 
To: Pallav Kalva 
Cc: pgsql-performance@postgresql.org
Subject: Re: Setting Shared Buffers , Effective Cache, Sort Mem Parameters
Date: Fri, 23 Apr 2004 23:44:51 +0200
Message-ID: 
References: <4088062E.2050901@deg.cc>
	
	<4089261A.9070700@deg.cc>
In-Reply-To: <4089261A.9070700@deg.cc>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/418
X-Sequence-Number: 6718

On Fri, 23 Apr 2004 10:20:10 -0400, Pallav Kalva  wrote:
> the database sizes is around 2- 4 gig and 
>there are 5 of them. this machine is
>     mainly for the databases and nothing is running on them.

Did I understand correctly that you run (or plan to run) five
postmasters?  Is there a special reason that you cannot put all your
tables into one database?

>    setting shared buffers to 10000 allocates (81Mb)  and effective 
>cache to 400000 would be around (3gig)
>    does this means that if all of the 81mb of the shared memory gets 
>allocated it will use rest from the effective
>    cache of (3g-81mb) ?

Simply said, if Postgres wants to access a block, it first looks whether
this block is already in shared buffers which should be the case, if the
block is one of the last 10000 blocks accessed.  Otherwise the block has
to be read in.  If the OS has the block in its cache, reading it is just
a (fast) memory operation, else it involves a (slow) physical disk read.

The number of database pages residing in the OS cache is totally out of
control of Postgres.  Effective_cache_size tells the query planner how
many database pages can be *expected* to be present in the OS cache.

>increasing the shared buffers space to 2g

Setting shared_buffers to half your available memory is the worst thing
you can do.  You would end up caching exactly the same set of blocks in
the internal buffers and in the OS cache, thus effectively making one of
the caches useless.

Better keep shared_buffers low and let the OS do its job.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Fri Apr 23 19:04:32 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F03A3D1D25B
	for ;
	Fri, 23 Apr 2004 19:04:28 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 93235-09
	for ;
	Fri, 23 Apr 2004 19:04:27 -0300 (ADT)
Received: from mta4.rcsntx.swbell.net (mta4.rcsntx.swbell.net [151.164.30.28])
	by svr1.postgresql.org (Postfix) with ESMTP id 7E113D1D08F
	for ;
	Fri, 23 Apr 2004 19:04:24 -0300 (ADT)
Received: from one.cheapcomplexdevices.com
	(adsl-68-122-9-222.dsl.pltn13.pacbell.net [68.122.9.222])
	by mta4.rcsntx.swbell.net (8.12.10/8.12.10) with ESMTP id
	i3NM3xY4004066; Fri, 23 Apr 2004 17:04:00 -0500 (CDT)
Received: from localhost (one [127.0.0.1])
	by one.cheapcomplexdevices.com (Postfix) with ESMTP
	id D39AA216F7D; Fri, 23 Apr 2004 15:10:20 -0700 (PDT)
Date: Fri, 23 Apr 2004 15:10:20 -0700 (PDT)
From: Ron Mayer 
X-X-Sender: mayer@one
To: Manfred Koizar 
Cc: Pallav Kalva , pgsql-performance@postgresql.org
Subject: Re: Setting Shared Buffers , Effective Cache, Sort Mem
In-Reply-To: 
Message-ID: 
References: <4088062E.2050901@deg.cc>
	
	<4089261A.9070700@deg.cc>
	
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/419
X-Sequence-Number: 6719


On Fri, 23 Apr 2004, Manfred Koizar wrote:
>
> Setting shared_buffers to half your available memory is the worst thing
> you can do.  You would end up caching exactly the same set of blocks in
> the internal buffers and in the OS cache, thus effectively making one of
> the caches useless.

One minor detail...  You wouldn't really cache the _exact_ same blocks
because cache-hits in shared-buffers (on the most frequently accessed
pages) would let the OS cache some other pages in it's cache.

But in my experience Manfred's right that there's no benefit and
some penalty to making shared_buffers so large it takes a significant
piece away from the OS's caching.

From pgsql-performance-owner@postgresql.org  Mon Apr 26 14:51:02 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 30A0CD1CCDC
	for ;
	Fri, 23 Apr 2004 19:21:21 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 01500-02
	for ;
	Fri, 23 Apr 2004 19:21:21 -0300 (ADT)
Received: from uno.mnl.com (uno.mnl.com [63.97.246.49])
	by svr1.postgresql.org (Postfix) with SMTP id C196BD1B560
	for ;
	Fri, 23 Apr 2004 19:21:18 -0300 (ADT)
Received: (qmail 12062 invoked by uid 1000); 23 Apr 2004 22:21:21 -0000
Message-ID: <20040423222121.12061.qmail@uno.mnl.com>
From: brad-pgperf@duttonbros.com
To: pgsql-performance@postgresql.org
Subject: index usage
Date: Fri, 23 Apr 2004 15:21:21 -0700
Mime-Version: 1.0
Content-Type: text/plain; format=flowed; charset="utf-8"
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0 tests=NO_REAL_NAME
X-Spam-Level: 
X-Archive-Number: 200404/440
X-Sequence-Number: 6740

Hi, 

I have a query which I think should be using an index all of the time but 
postgres only uses the index part of the time.  The index 
(ticket_crm_map_crm_id_suppid) has the where clause column (crm_id) listed 
first followed by the selected column (support_person_id).  Wouldn't the 
most efficient plan be to scan the index each time because the only columns 
needed are in the index?  Below is the table, 2 queries showing the 
difference in plans, followed by the record distribution of ticket_crm_map.  
I first did a 'vacuum analyze' to update the statistics. 

Thanks,
Brad 


athenapost=> \d ticket_crm_map
                                  Table "public.ticket_crm_map"
        Column         |            Type             |                 
Modifiers
 ------------------------+-----------------------------+--------------------- 
 -----------------------
tcrm_map_id            | integer                     | not null
ticket_id              | integer                     | not null
crm_id                 | integer                     | not null
support_person_id      | integer                     | not null
escalated_to_person_id | integer                     | not null
status                 | character varying(50)       | not null default 
'Open'::character varying
close_date             | timestamp without time zone |
updated_date           | timestamp without time zone |
updated_by             | character varying(255)      |
created_date           | timestamp without time zone |
created_by             | character varying(255)      |
additional_info        | text                        |
subject                | character varying(255)      |
Indexes:
   "ticket_crm_map_pkey" primary key, btree (tcrm_map_id)
   "ticket_crm_map_crm_id_key" unique, btree (crm_id, ticket_id)
   "ticket_crm_map_crm_id_suppid" btree (crm_id, support_person_id)
   "ticket_crm_map_status" btree (status)
   "ticket_crm_map_ticket_id" btree (ticket_id)
Foreign-key constraints:
   "$1" FOREIGN KEY (ticket_id) REFERENCES ticket(ticket_id)
   "$2" FOREIGN KEY (crm_id) REFERENCES company_crm(crm_id)
   "$3" FOREIGN KEY (support_person_id) REFERENCES person(person_id)
   "$4" FOREIGN KEY (escalated_to_person_id) REFERENCES person(person_id)
   "$5" FOREIGN KEY (status) REFERENCES ticket_status(status) 

athenapost=> explain analyze select distinct support_person_id from 
ticket_crm_map where crm_id = 7;
                                                                           
QUERY PLAN
 ---------------------------------------------------------------------------- 
 ---------------------------------------------------------------------------- 
 ----------
Unique  (cost=1262.99..1265.27 rows=1 width=4) (actual time=15.335..18.245 
rows=20 loops=1)
  ->  Sort  (cost=1262.99..1264.13 rows=456 width=4) (actual 
time=15.332..16.605 rows=2275 loops=1)
        Sort Key: support_person_id
        ->  Index Scan using ticket_crm_map_crm_id_suppid on ticket_crm_map  
(cost=0.00..1242.85 rows=456 width=4) (actual time=0.055..11.281 rows=2275 
loops=1)
              Index Cond: (crm_id = 7)
Total runtime: 18.553 ms
(6 rows) 

Time: 20.598 ms
athenapost=> explain analyze select distinct support_person_id from 
ticket_crm_map where crm_id = 1;
                                                          QUERY PLAN
 ---------------------------------------------------------------------------- 
 -----------------------------------------------------
Unique  (cost=10911.12..11349.26 rows=32 width=4) (actual 
time=659.102..791.517 rows=24 loops=1)
  ->  Sort  (cost=10911.12..11130.19 rows=87628 width=4) (actual 
time=659.090..713.285 rows=93889 loops=1)
        Sort Key: support_person_id
        ->  Seq Scan on ticket_crm_map  (cost=0.00..3717.25 rows=87628 
width=4) (actual time=0.027..359.299 rows=93889 loops=1)
              Filter: (crm_id = 1)
Total runtime: 814.601 ms
(6 rows) 

Time: 817.095 ms
athenapost=> select count(*), crm_id from ticket_crm_map group by crm_id;
count | crm_id
 -------+--------
 2554 |     63
  129 |     25
   17 |     24
  110 |     23
   74 |     22
   69 |     21
    2 |     20
   53 |     82
   10 |     17
   16 |     81
46637 |     16
   14 |     80
    2 |     15
 1062 |     79
   87 |     78
   93 |     77
   60 |     44
  363 |     76
  225 |     10
    4 |     74
   83 |      9
   27 |     73
  182 |      8
 2275 |      7
   15 |     71
  554 |      6
   44 |     70
  631 |      5
   37 |      4
  190 |      3
  112 |      2
93889 |      1
(32 rows) 

Time: 436.697 ms

From pgsql-performance-owner@postgresql.org  Fri Apr 23 23:50:34 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F2660D1B43A
	for ;
	Fri, 23 Apr 2004 23:50:29 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 72995-08
	for ;
	Fri, 23 Apr 2004 23:50:31 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 98745D1BB16
	for ;
	Fri, 23 Apr 2004 23:50:26 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3O2oEQD006799;
	Fri, 23 Apr 2004 22:50:14 -0400 (EDT)
To: Ron Mayer 
Cc: Manfred Koizar , Pallav Kalva ,
	pgsql-performance@postgresql.org
Subject: Re: Setting Shared Buffers , Effective Cache, Sort Mem 
In-reply-to:  
References: <4088062E.2050901@deg.cc>
	
	<4089261A.9070700@deg.cc>
	
	
Comments: In-reply-to Ron Mayer 
	message dated "Fri, 23 Apr 2004 15:10:20 -0700"
Date: Fri, 23 Apr 2004 22:50:14 -0400
Message-ID: <6798.1082775014@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/420
X-Sequence-Number: 6720

Ron Mayer  writes:
> [ on setting shared_buffers = half of RAM ]

> One minor detail...  You wouldn't really cache the _exact_ same blocks
> because cache-hits in shared-buffers (on the most frequently accessed
> pages) would let the OS cache some other pages in it's cache.

> But in my experience Manfred's right that there's no benefit and
> some penalty to making shared_buffers so large it takes a significant
> piece away from the OS's caching.

True, it'd probably not be the *exact* worst case.  But it'd be a good
approximation.  In practice you should either bet on the kernel doing
most of the caching (in which case you set shared_buffers pretty low)
or bet on Postgres doing most of the caching (in which case you set
shared_buffers to eat most of RAM).

The conventional wisdom at this point is to bet the first way; no one
has shown performance benefits from setting shared_buffers higher than
the low tens of thousands.  (Most of the mail list traffic on this
predates the existence of pgsql-performance, so check the other list
archives too if you go looking for discussion.)

It's possible that Jan's recent buffer-management improvements will
change the story as of 7.5.  I kinda doubt it myself, but it'd be worth
re-running any experiments you've done when you start working with 7.5.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Sat Apr 24 00:00:52 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 766A2D1DCD2
	for ;
	Fri, 23 Apr 2004 23:58:22 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 85298-06
	for ;
	Fri, 23 Apr 2004 23:58:25 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 73125D1DCCF
	for ;
	Fri, 23 Apr 2004 23:58:20 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3O2wO45006866;
	Fri, 23 Apr 2004 22:58:25 -0400 (EDT)
To: "Chris Hoover" 
Cc: pgsql-performance@postgresql.org
Subject: Re: Help with performance problems 
In-reply-to: <200404230931.17850.revoohc@sermonaudio.com> 
References: <200404230931.17850.revoohc@sermonaudio.com>
Comments: In-reply-to "Chris Hoover" 
	message dated "Fri, 23 Apr 2004 09:31:17 -0400"
Date: Fri, 23 Apr 2004 22:58:24 -0400
Message-ID: <6865.1082775504@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/421
X-Sequence-Number: 6721

"Chris Hoover"  writes:
> Here is the postgresql.conf from the server with the 11GB db:

> max_fsm_pages = 10000		# min 1000, fsm is free space map, ~6 bytes

It's unlikely that that's enough for an 11Gb database, especially if
you're only vacuuming a few times a week.  You should make your next run
be a "vacuum verbose" and look at the output to get an idea of what sort
of table bloat you're seeing, but I'll bet it's bad ...

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Sat Apr 24 11:45:44 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D9FE7D1B4AE
	for ;
	Sat, 24 Apr 2004 11:45:43 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 06886-07
	for ;
	Sat, 24 Apr 2004 11:45:46 -0300 (ADT)
Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca
	[199.212.17.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 49E2CD1B4D0
	for ;
	Sat, 24 Apr 2004 11:45:37 -0300 (ADT)
Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19)
	id ; Sat, 24 Apr 2004 10:45:41 -0400
Message-ID: 
From: "Shea,Dan [CIS]" 
To: 'Josh Berkus' 
Cc: pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sat, 24 Apr 2004 10:45:40 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/424
X-Sequence-Number: 6724

Josh, how long should a vacuum take on a 87 GB table with a 39 GB index?

I do not think that the verbose option of vacuum is verbose enough.
The vacuum keeps redoing the index, but there is no indication as to why it
is doing this.  

I see alot of activity with transaction logs being recycled (15 to 30 every
3 to 20 minutes).  
Is the vacuum causing this?


-----Original Message-----
From: Josh Berkus [mailto:josh@agliodbs.com]
Sent: Friday, April 23, 2004 2:48 PM
To: Shea,Dan [CIS]; 'Christopher Kings-Lynne'
Cc: pgsql-performance@postgresql.org
Subject: Re: [PERFORM] Why will vacuum not end?


Guys,

> Well, inserts create some locks - perhaps that's the problem...
>
> Otherwise, check the pg_locks view to see if you can figure it out.

FWIW, I've had this happen a couple of times, too.   Unfortunately, it's 
happend in the middle of the day so that I had to cancel the processes and 
get the system back to normal in too much of a hurry to consider documenting

it.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Sat Apr 24 12:36:17 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A09DCD1B931
	for ;
	Sat, 24 Apr 2004 12:36:15 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 40924-02
	for ;
	Sat, 24 Apr 2004 12:36:14 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id B6C7AD1B927
	for ;
	Sat, 24 Apr 2004 12:36:13 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4937421; Sat, 24 Apr 2004 08:37:37 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: "Shea,Dan [CIS]" 
Subject: Re: Why will vacuum not end?
Date: Sat, 24 Apr 2004 08:35:21 -0700
User-Agent: KMail/1.4.3
Cc: pgsql-performance@postgresql.org
References: 
In-Reply-To: 
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404240835.21914.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/425
X-Sequence-Number: 6725

Dan,

> Josh, how long should a vacuum take on a 87 GB table with a 39 GB index?

Depends:
-- What's your disk support?
-- VACUUM, VACUUM ANALYZE, or VACUUM FULL?
-- What's your vacuum_mem setting?
-- What are checkpoint and wal settings?

> I see alot of activity with transaction logs being recycled (15 to 30 every
> 3 to 20 minutes).
> Is the vacuum causing this?

Probably, yes.   How many checkpoint_buffers do you allow?

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Sat Apr 24 12:38:37 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 2BD8FD1B514
	for ;
	Sat, 24 Apr 2004 12:38:36 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 40774-04
	for ;
	Sat, 24 Apr 2004 12:38:34 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 52D0DD1B4F7
	for ;
	Sat, 24 Apr 2004 12:38:33 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4937428; Sat, 24 Apr 2004 08:39:58 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: Tom Lane ,
	Ron Mayer 
Subject: Re: Setting Shared Buffers , Effective Cache, Sort Mem
Date: Sat, 24 Apr 2004 08:37:42 -0700
User-Agent: KMail/1.4.3
Cc: Manfred Koizar , Pallav Kalva ,
	pgsql-performance@postgresql.org
References: <4088062E.2050901@deg.cc> 
	<6798.1082775014@sss.pgh.pa.us>
In-Reply-To: <6798.1082775014@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404240837.42398.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/426
X-Sequence-Number: 6726

Tom,

> It's possible that Jan's recent buffer-management improvements will
> change the story as of 7.5.  I kinda doubt it myself, but it'd be worth
> re-running any experiments you've done when you start working with 7.5.

Yes, Jan has indicated to me that he expects to make much heavier use of 
shared buffers under ARC.    But 7.5 still seems to be too unstable for me to 
test this assertion on a large database.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Sat Apr 24 14:55:01 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1ABADD1B4C1
	for ;
	Sat, 24 Apr 2004 14:55:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 72650-08
	for ;
	Sat, 24 Apr 2004 14:54:59 -0300 (ADT)
Received: from email01.aon.at (WARSL402PIP8.highway.telekom.at [195.3.96.97])
	by svr1.postgresql.org (Postfix) with SMTP id CB25BD1B4D7
	for ;
	Sat, 24 Apr 2004 14:54:56 -0300 (ADT)
Received: (qmail 285016 invoked from network); 24 Apr 2004 17:54:55 -0000
Received: from m160p001.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.9.225]) (envelope-sender )
	by qmail1rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 24 Apr 2004 17:54:55 -0000
From: Manfred Koizar 
To: "Shea,Dan [CIS]" 
Cc: 'Josh Berkus' ,
	pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sat, 24 Apr 2004 19:57:22 +0200
Message-ID: 
References: 
In-Reply-To: 
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/427
X-Sequence-Number: 6727

On Sat, 24 Apr 2004 10:45:40 -0400, "Shea,Dan [CIS]" 
wrote:
>[...] 87 GB table with a 39 GB index?

>The vacuum keeps redoing the index, but there is no indication as to why it
>is doing this.  

If VACUUM finds a dead tuple, if does not immediately remove index
entries pointing to that tuple.  It instead collects such tuple ids and
later does a bulk delete, i.e. scans the whole index and removes all
index items pointing to one of those tuples.  The number of tuple ids
that can be remembered is controlled by vacuum_mem: it is

	VacuumMem * 1024 / 6

Whenever this number of dead tuples has been found, VACUUM scans the
index (which takes ca. 60000 seconds, more than 16 hours), empties the
list and continues to scan the heap ...

 From the number of dead tuples you can estimate how often your index
will be scanned.  If dead tuples are evenly distributed, expect there to
be 15 index scans with your current vacuum_mem setting of 196608.  So
your VACUUM will run for 11 days :-(

OTOH this would mean that there are 500 million dead tuples.  Do you
think this is possible?

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Sat Apr 24 16:48:27 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 61076D1B8A9
	for ;
	Sat, 24 Apr 2004 16:48:24 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 06502-07
	for ;
	Sat, 24 Apr 2004 16:48:24 -0300 (ADT)
Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca
	[199.212.17.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 39B22D1B8A0
	for ;
	Sat, 24 Apr 2004 16:48:22 -0300 (ADT)
Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19)
	id ; Sat, 24 Apr 2004 15:48:20 -0400
Message-ID: 
From: "Shea,Dan [CIS]" 
To: 'Josh Berkus' 
Cc: pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sat, 24 Apr 2004 15:48:19 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/428
X-Sequence-Number: 6728

Manfred is indicating the reason it is taking so long is due to the number
of dead tuples in my index and the vacuum_mem setting.  
The last delete that I did before starting a vacuum had 219,177,133
deletions.
Dan.
>Dan,

>> Josh, how long should a vacuum take on a 87 GB table with a 39 GB index?

>Depends:
>-- What's your disk support?

>-- VACUUM, VACUUM ANALYZE, or VACUUM FULL?
VACUUM ANALYZE
>-- What's your vacuum_mem setting?
set vacuum_mem = '196608'
#fsync = true                   # turns forced synchronization on or off
#wal_sync_method = fsync  
>-- What are checkpoint and wal settings?
wal_buffers = 64 
checkpoint_segments = 30 
checkpoint_timeout = 300

>> I see alot of activity with transaction logs being recycled (15 to 30
every
>> 3 to 20 minutes).
>> Is the vacuum causing this?

>Probably, yes.   How many checkpoint_buffers do you allow?
I am not sure what the checkpoint_buffers are, we are running 7.4.0?
>-- 
>Josh Berkus
>Aglio Database Solutions
>San Francisco

From pgsql-performance-owner@postgresql.org  Sat Apr 24 16:58:10 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C908ED1B4DF
	for ;
	Sat, 24 Apr 2004 16:58:09 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 06012-09
	for ;
	Sat, 24 Apr 2004 16:58:09 -0300 (ADT)
Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca
	[199.212.17.130])
	by svr1.postgresql.org (Postfix) with ESMTP id ABC45D1B4F1
	for ;
	Sat, 24 Apr 2004 16:58:07 -0300 (ADT)
Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19)
	id ; Sat, 24 Apr 2004 15:58:09 -0400
Message-ID: 
From: "Shea,Dan [CIS]" 
To: 'Manfred Koizar' 
Cc: 'Josh Berkus' ,
	pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sat, 24 Apr 2004 15:58:08 -0400
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/429
X-Sequence-Number: 6729

There were defintely 219,177,133 deletions.  
The deletions are most likely from the beginning, it was based on the
reception_time of the data.
I would rather not use re-index, unless it is faster then using vacuum.
What do you think would be the best way to get around this?
Increase vacuum_mem to a higher amount 1.5 to 2 GB or try a re-index (rather
not re-index so that data can be queried without soing a seqscan).
Once the index is cleaned up, how does vacuum handle the table?  
Does it take as long as the index or is it faster?



-----Original Message-----
From: Manfred Koizar [mailto:mkoi-pg@aon.at]
Sent: Saturday, April 24, 2004 1:57 PM
To: Shea,Dan [CIS]
Cc: 'Josh Berkus'; pgsql-performance@postgresql.org
Subject: Re: [PERFORM] Why will vacuum not end?


On Sat, 24 Apr 2004 10:45:40 -0400, "Shea,Dan [CIS]" 
wrote:
>[...] 87 GB table with a 39 GB index?

>The vacuum keeps redoing the index, but there is no indication as to why it
>is doing this.  

If VACUUM finds a dead tuple, if does not immediately remove index
entries pointing to that tuple.  It instead collects such tuple ids and
later does a bulk delete, i.e. scans the whole index and removes all
index items pointing to one of those tuples.  The number of tuple ids
that can be remembered is controlled by vacuum_mem: it is

	VacuumMem * 1024 / 6

Whenever this number of dead tuples has been found, VACUUM scans the
index (which takes ca. 60000 seconds, more than 16 hours), empties the
list and continues to scan the heap ...

 From the number of dead tuples you can estimate how often your index
will be scanned.  If dead tuples are evenly distributed, expect there to
be 15 index scans with your current vacuum_mem setting of 196608.  So
your VACUUM will run for 11 days :-(

OTOH this would mean that there are 500 million dead tuples.  Do you
think this is possible?

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Sat Apr 24 21:02:57 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 46177D1B553
	for ;
	Sat, 24 Apr 2004 21:02:55 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 64044-08
	for ;
	Sat, 24 Apr 2004 21:02:56 -0300 (ADT)
Received: from email04.aon.at (WARSL402PIP2.highway.telekom.at [195.3.96.74])
	by svr1.postgresql.org (Postfix) with SMTP id 033EAD1B53B
	for ;
	Sat, 24 Apr 2004 21:02:53 -0300 (ADT)
Received: (qmail 97716 invoked from network); 25 Apr 2004 00:02:55 -0000
Received: from m160p001.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.9.225]) (envelope-sender )
	by qmail7rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 25 Apr 2004 00:02:55 -0000
From: Manfred Koizar 
To: "Shea,Dan [CIS]" 
Cc: 'Josh Berkus' ,
	pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sun, 25 Apr 2004 02:05:22 +0200
Message-ID: 
References: 
In-Reply-To: 
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/430
X-Sequence-Number: 6730

On Sat, 24 Apr 2004 15:48:19 -0400, "Shea,Dan [CIS]" 
wrote:
>Manfred is indicating the reason it is taking so long is due to the number
>of dead tuples in my index and the vacuum_mem setting.  


Not dead tuples in the index, but dead tuples in the table.


>The last delete that I did before starting a vacuum had 219,177,133
>deletions.

Ok, with vacuum_mem = 196608 the bulk delete batch size is ca. 33.5 M
tuple ids.  219 M dead tuples will cause 7 index scans.  The time for an
index scan is more or less constant, 60000 seconds in your case.  So
yes, a larger vacuum_mem will help, but only if you really have as much
*free* memory.  Forcing the machine into swapping would make things
worse.

BTW, VACUUM frees millions of index pages, is your FSM large enough?

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Sat Apr 24 21:26:11 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 71FBFD1B520
	for ;
	Sat, 24 Apr 2004 21:26:09 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 72670-03
	for ;
	Sat, 24 Apr 2004 21:26:11 -0300 (ADT)
Received: from email07.aon.at (WARSL402PIP8.highway.telekom.at [195.3.96.97])
	by svr1.postgresql.org (Postfix) with SMTP id 450BAD1B514
	for ;
	Sat, 24 Apr 2004 21:26:07 -0300 (ADT)
Received: (qmail 693236 invoked from network); 25 Apr 2004 00:26:10 -0000
Received: from m160p001.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.9.225]) (envelope-sender )
	by 172.18.5.236 (qmail-ldap-1.03) with SMTP
	for ; 25 Apr 2004 00:26:10 -0000
From: Manfred Koizar 
To: "Shea,Dan [CIS]" 
Cc: 'Josh Berkus' ,
	pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sun, 25 Apr 2004 02:28:37 +0200
Message-ID: 
References: 
In-Reply-To: 
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/431
X-Sequence-Number: 6731

On Sat, 24 Apr 2004 15:58:08 -0400, "Shea,Dan [CIS]" 
wrote:
>There were defintely 219,177,133 deletions.  
>The deletions are most likely from the beginning, it was based on the
>reception_time of the data.
>I would rather not use re-index, unless it is faster then using vacuum.

I don't know whether it would be faster.  But if you decide to reindex,
make sure sort_mem is *huge*!

>What do you think would be the best way to get around this?
>Increase vacuum_mem to a higher amount 1.5 to 2 GB or try a re-index (rather
>not re-index so that data can be queried without soing a seqscan).

Just out of curiosity:  What kind of machine is this running on?  And
how long does a seq scan take?

>Once the index is cleaned up, how does vacuum handle the table?  

If you are lucky VACUUM frees half the index pages.  And if we assume
that the most time spent scanning an index goes into random page
accesses, future VACUUMs will take "only" 30000 seconds per index scan.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Sat Apr 24 22:50:07 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id EC113D1B436
	for ;
	Sat, 24 Apr 2004 22:50:01 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 85001-07
	for ;
	Sat, 24 Apr 2004 22:50:03 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id D2F89D1B432
	for ;
	Sat, 24 Apr 2004 22:49:58 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4939836; Sat, 24 Apr 2004 18:51:26 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: "Shea,Dan [CIS]" ,
	'Manfred Koizar' 
Subject: Re: Why will vacuum not end?
Date: Sat, 24 Apr 2004 18:49:07 -0700
User-Agent: KMail/1.4.3
Cc: pgsql-performance@postgresql.org
References: 
In-Reply-To: 
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404241849.07896.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/432
X-Sequence-Number: 6732

Dan,

> There were defintely 219,177,133 deletions.
> The deletions are most likely from the beginning, it was based on the
> reception_time of the data.

You need to run VACUUM more often, I think.    Vacuuming out 219 million dead 
tuples is going to take a long time no matter how you look at it.   If you 
vacuum more often, presumably there will be less for Vacuum to do each time.

-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Sun Apr 25 04:13:43 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 90554D1B48A
	for ;
	Sun, 25 Apr 2004 04:13:42 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 58789-05
	for ;
	Sun, 25 Apr 2004 04:13:42 -0300 (ADT)
Received: from lamb.mcmillan.net.nz (218-101-45-219.paradise.net.nz
	[218.101.45.219])
	by svr1.postgresql.org (Postfix) with ESMTP id DA14ED1B4AE
	for ;
	Sun, 25 Apr 2004 04:13:39 -0300 (ADT)
Received: from lamb.mcmillan.net.nz (lamb.mcmillan.net.nz [127.0.0.1])
	by lamb.mcmillan.net.nz (Postfix) with ESMTP
	id 2025FAD985AC; Sun, 25 Apr 2004 19:13:36 +1200 (NZST)
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Andrew McMillan 
To: Josh Berkus 
Cc: Tom Lane ,
	Bruce Momjian , pg@fastcrypt.com,
	Paul Tuckfield , Anjan Dave ,
	Neil Conway ,
	Dirk =?ISO-8859-1?Q?Lutzeb=E4ck?= ,
	pgsql-performance@postgresql.org
In-Reply-To: <200404221037.10097.josh@agliodbs.com>
References: <200404220456.i3M4uFv04710@candle.pha.pa.us>
	<20171.1082610827@sss.pgh.pa.us> <200404221037.10097.josh@agliodbs.com>
Content-Type: text/plain
Message-Id: <1082877215.8041.20.camel@lamb.mcmillan.net.nz>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.5.7 
Date: Sun, 25 Apr 2004 19:13:35 +1200
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=
X-Spam-Level: 
X-Archive-Number: 200404/433
X-Sequence-Number: 6733

On Thu, 2004-04-22 at 10:37 -0700, Josh Berkus wrote:
> Tom,
> 
> > The tricky
> > part is that a slow adaptation rate means we can't have every backend
> > figuring this out for itself --- the right value would have to be
> > maintained globally, and I'm not sure how to do that without adding a
> > lot of overhead.
> 
> This may be a moot point, since you've stated that changing the loop timing 
> won't solve the problem, but what about making the test part of make?   I 
> don't think too many systems are going to change processor architectures once 
> in production, and those that do can be told to re-compile.

Sure they do - PostgreSQL is regularly provided as a pre-compiled
distribution.  I haven't compiled PostgreSQL for years, and we have it
running on dozens of machines, some SMP, some not, but most running
Debian Linux.

Even having a compiler _installed_ on one of our client's database
servers would usually be considered against security procedures, and
would get a black mark when the auditors came through.

Regards,
					Andrew McMillan
-------------------------------------------------------------------------
Andrew @ Catalyst .Net .NZ  Ltd,  PO Box 11-053,  Manners St,  Wellington
WEB: http://catalyst.net.nz/             PHYS: Level 2, 150-154 Willis St
DDI: +64(4)916-7201       MOB: +64(21)635-694      OFFICE: +64(4)499-2267
                     Planning an election?  Call us!
-------------------------------------------------------------------------


From pgsql-performance-owner@postgresql.org  Sun Apr 25 10:05:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 07458D1B46D
	for ;
	Sun, 25 Apr 2004 10:05:18 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 29541-02
	for ;
	Sun, 25 Apr 2004 10:05:20 -0300 (ADT)
Received: from ECCMCBH.cmc.int.ec.gc.ca (ecdor130.cmc.ec.gc.ca
	[199.212.17.130])
	by svr1.postgresql.org (Postfix) with ESMTP id C1004D1B496
	for ;
	Sun, 25 Apr 2004 10:05:13 -0300 (ADT)
Received: by eccmcbh.cmc.ec.gc.ca with Internet Mail Service (5.5.2653.19)
	id ; Sun, 25 Apr 2004 09:05:16 -0400
Message-ID: 
From: "Shea,Dan [CIS]" 
To: 'Manfred Koizar' 
Cc: pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sun, 25 Apr 2004 09:05:11 -0400
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/434
X-Sequence-Number: 6734

It is set at max_fsm_pages = 1500000 .

We are running a 
DELL PowerEdge 6650 with 4 CPU's
Mem:  3611320k av from top.
The database is on a shared device (SAN) raid5, 172 GB.
Qlogic Fibre optic cards(desc: "QLogic Corp.|QLA2312 Fibre Channel Adapter")
connected to the Dell version of an EMC SAN (FC4700 I believe).

I have set vacuum_mem = 917504;
and started another vacuum verbose on the table in question.
Tried to set vacuum_mem to 1114112 and vacuum failed, then tried 917504 and
vacuum started.

PWFPM_DEV=# set vacuum_mem = '1114112';
SET
PWFPM_DEV=# show vacuum_mem;
 vacuum_mem
------------
 1114112
(1 row)

PWFPM_DEV=# vacuum verbose  forecastelement;

INFO:  vacuuming "public.forecastelement"
ERROR:  invalid memory alloc request size 1140850686
PWFPM_DEV=# set vacuum_mem = 917504;
SET
PWFPM_DEV=# show vacuum_mem;
 vacuum_mem
------------
 917504
(1 row)

PWFPM_DEV=# select now();vacuum verbose  forecastelement;select now();
              now
-------------------------------
 2004-04-25 01:40:23.367123+00
(1 row)

INFO:  vacuuming "public.forecastelement"

I performed a query that used a seqscan

PWFPM_DEV=# explain analyze select count(*) from forecastelement;
                                                                  QUERY PLAN
----------------------------------------------------------------------------
-------------------------------------------------------------------
 Aggregate  (cost=16635987.60..16635987.60 rows=1 width=0) (actual
time=13111152.844..13111152.847 rows=1 loops=1)
   ->  Seq Scan on forecastelement  (cost=0.00..15403082.88 rows=493161888
width=0) (actual time=243.562..12692714.422 rows=264422681 loops=1)
 Total runtime: 13111221.978 ms
(3 rows)

Dan.

-----Original Message-----
From: Manfred Koizar [mailto:mkoi-pg@aon.at]
Sent: Saturday, April 24, 2004 8:29 PM
To: Shea,Dan [CIS]
Cc: 'Josh Berkus'; pgsql-performance@postgresql.org
Subject: Re: [PERFORM] Why will vacuum not end?


On Sat, 24 Apr 2004 15:58:08 -0400, "Shea,Dan [CIS]" 
wrote:
>There were defintely 219,177,133 deletions.  
>The deletions are most likely from the beginning, it was based on the
>reception_time of the data.
>I would rather not use re-index, unless it is faster then using vacuum.

I don't know whether it would be faster.  But if you decide to reindex,
make sure sort_mem is *huge*!

>What do you think would be the best way to get around this?
>Increase vacuum_mem to a higher amount 1.5 to 2 GB or try a re-index
(rather
>not re-index so that data can be queried without soing a seqscan).

Just out of curiosity:  What kind of machine is this running on?  And
how long does a seq scan take?

>Once the index is cleaned up, how does vacuum handle the table?  

If you are lucky VACUUM frees half the index pages.  And if we assume
that the most time spent scanning an index goes into random page
accesses, future VACUUMs will take "only" 30000 seconds per index scan.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Sun Apr 25 17:24:32 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 414EBD1B478
	for ;
	Sun, 25 Apr 2004 17:24:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 21904-10
	for ;
	Sun, 25 Apr 2004 17:24:30 -0300 (ADT)
Received: from email03.aon.at (WARSL402PIP6.highway.telekom.at [195.3.96.93])
	by svr1.postgresql.org (Postfix) with SMTP id 38489D1B468
	for ;
	Sun, 25 Apr 2004 17:24:27 -0300 (ADT)
Received: (qmail 109770 invoked from network); 25 Apr 2004 20:24:27 -0000
Received: from m149p013.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.141]) (envelope-sender )
	by qmail3rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 25 Apr 2004 20:24:27 -0000
From: Manfred Koizar 
To: Tom Lane 
Cc: Robert Treat ,
	pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org
Subject: Number of pages in a random sample (was: query slows down with more
	accurate stats)
Date: Sun, 25 Apr 2004 22:26:56 +0200
Message-ID: 
References: <1081879359.25537.526.camel@camel>
 <25556.1081883922@sss.pgh.pa.us>
	
	<22642.1082074729@sss.pgh.pa.us>
	
	<29060.1082126089@sss.pgh.pa.us>
	
	<25762.1082390410@sss.pgh.pa.us>
In-Reply-To: <25762.1082390410@sss.pgh.pa.us>
X-Mailer: Forte Agent 1.93/32.576 English (American)
MIME-Version: 1.0
Content-Type: multipart/mixed;
	boundary="--=_go7o80pp4j494vn888uhe9n4n4njhifgts.MFSBCHJLHS"
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/435
X-Sequence-Number: 6735

----=_go7o80pp4j494vn888uhe9n4n4njhifgts.MFSBCHJLHS
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

On Mon, 19 Apr 2004 12:00:10 -0400, Tom Lane  wrote:
>A possible compromise is to limit the number of pages sampled to
>something a bit larger than n, perhaps 2n or 3n.  I don't have a feeling
>for the shape of the different-pages probability function; would this
>make a significant difference, or would it just waste cycles?

I would have replied earlier, if I had a good answer.  What I have so
far contains at least one, probably two flaws.  Knowing not much more
than the four basic arithmetic operations I was not able to improve my
model.  So I post what I have:

As usual we assume a constant number c of tuples per page.  If we have a
table of size B pages and want to collect a sample of n tuples, the
number of possible samples is (again in OOo syntax)

	left( binom{cB}{n} right)

If we select an arbitrary page, the number of possible samples that do
NOT contain any tuple from this page is

	left( binom {c (B-1)} {n} right)

Let's forget about our actual implementations of sampling methods and
pretend we have a perfect random sampling method.  So the probability
Pnot(c, B, n) that a certain page is not represented in a random sample
is

	left( binom {c (B-1)} {n} right) over left( binom{cB}{n} right)

which can be transformed into the more computing-friendly form

	prod from{i=0} to{n-1} {{cB-c - i} over {cB - i}}

Clearly the probability that a certain page *is* represented in a sample
is

	Pyes(c, B, n) = 1 - Pnot(c, B, n)

The next step assumes that these probabilities are independent for
different pages, which in reality they are not.  We simply estimate the
number of pages represented in a random sample as 

	numPag(c, B, n) = B * Pyes(c, B, n)

Here are some results for n = 3000:

	B  \ c->    10 |   100 |   200
	-------+-------+-------+-------
	   100 |  ---  |   100 |   100
	  1000 |   972 |   953 |   951
	  2000 |  1606 |  1559 |  1556
	  3000 |  1954 |  1902 |  1899
	  6000 |  2408 |  2366 |  2363
	  9000 |  2588 |  2555 |  2553
	 20000 |  2805 |  2788 |  2787
	 30000 |  2869 |  2856 |  2856
	100000 |  2960 |  2956 |  2956

This doesn't look to depend heavily on the number of tuples per page,
which sort of justifies the assumption that c is constant.

In the next step I tried to estimate the number of pages that contain
exactly 1, 2, ... tuples of the sample.  My naive procedure works as
follows (I'm not sure whether it is even valid as a rough approximation,
constructive criticism is very welcome):

For c=100, B=3000, n=3000 we expect 1902 pages to contain at least 1
tuple of the sample.  There are 1098 more tuples than pages, these
tuples lie somewhere in those 1902 pages from the first step.
numPag(99, 1902, 1098) = 836 pages contain at least a second tuple.
So the number of pages containing exactly 1 tuple is 1902 - 836 = 1066.
Repeating these steps we get 611 pages with 2 tuples, 192 with 3, 30
with 4, and 3 pages with 5 tuples.

Here are some more numbers for c = 100 and n = 3000:
	
	   B   | pages with 1, 2, ... tuples
	-------+--------------------------------------------------------
	   100 |  1 to 24 tuples: 0, then 1, 2, 4, 10, 18, 26, 24, 11, 4
	  1000 |  108, 201, 268, 229, 113, 29, 5
	  2000 |  616, 555, 292,  83, 12, 1
	  3000 | 1066, 611, 192,  30,  3
	  6000 | 1809, 484,  68,   5
	  9000 | 2146, 374,  32,   2
	 20000 | 2584, 196,   8
	 30000 | 2716, 138,   3
	100000 | 2912,  44

A small C program to experimentally confirm or refute these calculations
is attached.  Its results are fairly compatible with above numbers,
IMHO.

Servus
 Manfred

----=_go7o80pp4j494vn888uhe9n4n4njhifgts.MFSBCHJLHS
Content-Type: text/plain; charset=us-ascii; name=samsim.c
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment; filename=samsim.c

/*
** samsim.c  -  sampling simulator
*/
#include 
#include 
#include 
#include 

typedef int bool;


#define MAX_RANDOM_VALUE  (0x7FFFFFFF)

static void initrandom()
{
	struct timeval tv;

	gettimeofday(&tv, NULL);
	srandom(tv.tv_sec ^ tv.tv_usec);
}/*initrandom*/

/* Select a random value R uniformly distributed in 0 < R < 1 */
static double
random_fract(void)
{
	long		z;

	/* random() can produce endpoint values, try again if so */
	do
	{
		z = random();
	} while (z <= 0 || z >= MAX_RANDOM_VALUE);
	return (double) z / (double) MAX_RANDOM_VALUE;
}

/*
** data structure for (modified) Algorithm S from Knuth 3.4.2
*/
typedef struct
{
	long	N;				/* number of tuples, known in advance */
	int		n;				/* sample size */
	long	t;				/* current tuple number */
	int		m;				/* tuples selected so far */
} SamplerData;
typedef SamplerData *Sampler;

static void Sampler_Init(Sampler bs, long N, int samplesize);
static bool Sampler_HasMore(Sampler bs);
static long Sampler_Next(Sampler bs);

/*
** Sampler_Init -- prepare for random sampling
*/
static void
Sampler_Init(Sampler bs, long N, int samplesize)
{
	bs->N = N;			/* table size */
	bs->n = samplesize;
	bs->t = 0;					/* tuples scanned so far */
	bs->m = 0;					/* tuples selected so far */
}

static bool
Sampler_HasMore(Sampler bs)
{
	return (bs->t < bs->N) && (bs->m < bs->n);
}

static long
Sampler_Next(Sampler bs)
{
	long	K = bs->N - bs->t;		/* remaining tuples */
	int		k = bs->n - bs->m;		/* tuples still to sample */
	double	p;				/* probability to skip the next tuple */
	double	V;				/* random */

	/* Assert(Sampler_HasMore(bs)); */

	if (k >= K)
	{
		/* need all the rest */
		bs->t += 1;
		bs->m += 1;
		return bs->t - 1;
	}
	
	p = 1.0 - (double) k / (double) K;
	V = random_fract();
	while (V < p)
	{
		bs->t += 1;
		K -= 1;
		/*
		** Assert(K > 0)
		** because we startet with K > k > 0,
		** and when K == k, the loop terminates
		*/
		p *= 1.0 - (double) k / (double) K;
	}

	/* select */
	bs->t += 1;
	bs->m += 1;
	return bs->t - 1;
}

static void usage()
{
	fprintf(stderr, "usage: samsim c B n\n"
	"where c = tuples/page\n"
	"      B = # of pages\n"
	"      n = sample size\n");
}/*usage*/

static void dumpstats(long *stat, long cnt)
{
	long i;

	for (i = 1; i <= cnt; ++i) {
		printf("%ld%c", stat[i], (i < cnt) ? '\t' : '\n');
	}/*for*/
}/*dumpstats*/

static int samsim(long c, long B, long n)
{
	SamplerData s;
	long oldblock = -1;
	long blockhits = 0;
	long maxhits = -1;
	long *stat = calloc(c + 1, sizeof(long));

	if (stat == NULL) {
		fprintf(stderr, "cannot allocate %ld numbers\n", c);
		return 2;
	}/*if*/

	Sampler_Init(&s, c * B, n);
	if (!Sampler_HasMore(&s)) {
		fprintf(stderr, "empty sample\n");
		return 3;
	}/*if*/

	while (Sampler_HasMore(&s)) {
		long t = Sampler_Next(&s);
		long blocknr = t / c;

		if (blocknr == oldblock) {
			++blockhits;
		} else {
			if (oldblock >= 0) {
				stat[blockhits] += 1;
				if (blockhits > maxhits)
					maxhits = blockhits;
			}/*if*/

			oldblock = blocknr;
			blockhits = 1;
		}/*else*/
	}/*while*/
	stat[blockhits] += 1;

	dumpstats(stat, maxhits);
	free(stat);

	return 0;		/* success */
}/*samsim*/

int main(int argc, char* argv[])
{
	long c, B, n;

	if (argc != 4) {
		usage();
		exit(1);
	}/*if*/

	c = atol(argv[1]);
	B = atol(argv[2]);
	n = atol(argv[3]);

	if ((c <= 0) || (B <= 0) || (n <= 0)) {
		usage();
		exit(1);
	}/*if*/

	initrandom();
	return samsim(c, B, n);
}/*main*/

----=_go7o80pp4j494vn888uhe9n4n4njhifgts.MFSBCHJLHS--

From pgsql-performance-owner@postgresql.org  Sun Apr 25 17:44:28 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F2456D1B587
	for ;
	Sun, 25 Apr 2004 17:44:26 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 33483-07
	for ;
	Sun, 25 Apr 2004 17:44:27 -0300 (ADT)
Received: from email06.aon.at (WARSL402PIP3.highway.telekom.at [195.3.96.75])
	by svr1.postgresql.org (Postfix) with SMTP id B43B8D1B586
	for ;
	Sun, 25 Apr 2004 17:44:24 -0300 (ADT)
Received: (qmail 218026 invoked from network); 25 Apr 2004 20:44:26 -0000
Received: from m149p013.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.141]) (envelope-sender )
	by qmail6rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 25 Apr 2004 20:44:26 -0000
From: Manfred Koizar 
To: "Shea,Dan [CIS]" 
Cc: pgsql-performance@postgresql.org
Subject: Re: Why will vacuum not end?
Date: Sun, 25 Apr 2004 22:46:54 +0200
Message-ID: <2v7o80hmttci6k5m3ht6ov760r5jbckmcc@email.aon.at>
References: 
In-Reply-To: 
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/436
X-Sequence-Number: 6736

On Sun, 25 Apr 2004 09:05:11 -0400, "Shea,Dan [CIS]" 
wrote:
>It is set at max_fsm_pages = 1500000 .

This might be too low.  Your index has ca. 5 M pages, you are going to
delete half of its entries, and what you delete is a contiguous range of
values.  So up to 2.5 M index pages might be freed (minus inner nodes
and pages not completely empty).  And there will be lots of free heap
pages too ...

I wrote:
>If you are lucky VACUUM frees half the index pages.  And if we assume
>that the most time spent scanning an index goes into random page
>accesses, future VACUUMs will take "only" 30000 seconds per index scan.

After a closer look at the code and after having slept over it I'm not
so sure any more that the number of tuple ids to be removed has only
minor influence on the time spent for a bulk delete run.  After the
current VACUUM has finished would you be so kind to run another VACUUM
VERBOSE with only a few dead tuples and post the results here?

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Mon Apr 26 06:02:01 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 8031DD1B472
	for ;
	Mon, 26 Apr 2004 06:02:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 38484-07
	for ;
	Mon, 26 Apr 2004 06:01:59 -0300 (ADT)
Received: from mail.dsvr.co.uk (mail.dsvr.co.uk [212.69.192.8])
	by svr1.postgresql.org (Postfix) with ESMTP id 38A0CD1B46C
	for ;
	Mon, 26 Apr 2004 06:01:57 -0300 (ADT)
Received: from [212.69.216.20] (helo=dsvr.net)
	by mail.dsvr.co.uk with esmtp (Exim 4.30) id 1BI1zy-0004CR-42
	for pgsql-performance@postgresql.org; Mon, 26 Apr 2004 10:01:58 +0100
Message-ID: <408CD006.3090004@dsvr.net>
Date: Mon, 26 Apr 2004 10:01:58 +0100
From: Rob Fielding 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8a) Gecko/20040418
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: OT: Help with performance problems
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/437
X-Sequence-Number: 6737

scott.marlowe wrote:
> On Fri, 23 Apr 2004, Chris Hoover wrote:
> 
> 
>>DB's on Powervaults 220S using raid 5 (over 6 disks)
> 
> 
> What controller is this, the adaptec?  We've found it to be slower than 
> the LSI megaraid based controller, but YMMV.

Wow, really? You got any more details of the chipset, mobo and kernel 
driver ?

I've been taken to my wits end wrestling with an LSI MegaRAID 320-1 
controller on a supermicro board all weekend. I just couldn't get 
anything more than 10MB/sec out of it with megaraid driver v1 OR v2 in 
Linux 2.4.26, nor the version in 2.6.6-rc2. After 2 days of humming the 
Adaptec mantra I gave in and switched the array straight onto the 
onboard Adaptec 160 controller (same cable and everything). Software 
RAID 5 gets me over 40MB sec for a nominal cpu hit - more than 4 times 
what I could get out of the MegaRAID controller :( Even the 2nd SCSI-2 
channel gets 40MB/sec max (pg_xlog :)

And HOW LONG does it take to detect drives during POST....ohhhh never 
mind ... I really just wanna rant :) There should be a free counseling 
service for enraged sysops.

-- 

Rob Fielding
rob@dsvr.net

www.dsvr.co.uk              Development             Designer Servers Ltd

From pgsql-hackers-owner@postgresql.org  Mon Apr 26 12:10:34 2004
X-Original-To: pgsql-hackers-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id A62C0D1B51A; Mon, 26 Apr 2004 12:08:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 82600-08; Mon, 26 Apr 2004 12:08:21 -0300 (ADT)
Received: from relay2.EECS.Berkeley.EDU (relay2.EECS.Berkeley.EDU
	[169.229.60.28]) by svr1.postgresql.org (Postfix) with ESMTP
	id 9057DD1B464; Mon, 26 Apr 2004 12:08:20 -0300 (ADT)
Received: from relay3.EECS.Berkeley.EDU (localhost [127.0.0.1])
	by relay2.EECS.Berkeley.EDU (8.12.10/8.9.3) with ESMTP id
	i3QF8INk024799; Mon, 26 Apr 2004 08:08:19 -0700 (PDT)
Received: from datafix.CS.Berkeley.EDU (datafix.CS.Berkeley.EDU
	[128.32.37.185])
	by relay3.EECS.Berkeley.EDU (8.12.10/8.9.3) with ESMTP id
	i3QF8Htj015313; Mon, 26 Apr 2004 08:08:17 -0700 (PDT)
Reply-To: sailesh@cs.berkeley.edu
X-URL: http://www.cs.berkeley.edu/~sailesh
X-Attribution: Sailesh
To: Manfred Koizar 
Cc: Tom Lane , Robert Treat ,
	pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org
Subject: Re: Number of pages in a random sample
References: <1081879359.25537.526.camel@camel>
 <25556.1081883922@sss.pgh.pa.us>
	
	<22642.1082074729@sss.pgh.pa.us>
	
	<29060.1082126089@sss.pgh.pa.us>
	
	<25762.1082390410@sss.pgh.pa.us>
	
From: Sailesh Krishnamurthy 
Date: Mon, 26 Apr 2004 08:08:16 -0700
In-Reply-To:  (Manfred
	Koizar's message of "Sun, 25 Apr 2004 22:26:56 +0200")
Message-ID: 
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/839
X-Sequence-Number: 52761


I have not been following this thread carefully. Just in case you are
interested in further reading, you could check this paper:

 "A Bi-Level Bernoulli Scheme for Database Sampling"
 Peter Haas, Christian Koenig (SIGMOD 2004) 

-- 
Pip-pip
Sailesh
http://www.cs.berkeley.edu/~sailesh



From pgsql-admin-owner@postgresql.org  Mon Apr 26 17:55:51 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id DAB2ED1B432; Mon, 26 Apr 2004 13:38:04 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 21040-06; Mon, 26 Apr 2004 13:38:01 -0300 (ADT)
Received: from FE-mail03.sfg.albacom.net (FE-mail03.albacom.net
	[213.217.149.83]) by svr1.postgresql.org (Postfix) with ESMTP
	id 903E5D1B4E6; Mon, 26 Apr 2004 13:38:00 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail03.sfg.albacom.net (7.0.009)
	id 40757CA500244891; Mon, 26 Apr 2004 18:37:56 +0200
Message-ID: <408D3AE8.5020006@axa.it>
Date: Mon, 26 Apr 2004 18:38:00 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: pgsql-jdbc@postgresql.org, pgsql-performance@postgresql.org,
	pgsql-admin@postgresql.org
Subject: is a good practice to create an index on the oid?
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/313
X-Sequence-Number: 13286

I am using the oid of the table as the main key and I've found that is 
not indexed (maybe because I have declared another primary key in the table)

it is a good practice to create an index like this on the oid of a table?
CREATE INDEX idoid annuncio400 USING btree (oid);


does it work as a normal index?

Thank you
Edoardo

From pgsql-performance-owner@postgresql.org  Mon Apr 26 17:24:00 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 5FF50D1B498
	for ;
	Mon, 26 Apr 2004 14:11:08 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36409-03
	for ;
	Mon, 26 Apr 2004 14:11:07 -0300 (ADT)
Received: from uno.mnl.com (uno.mnl.com [63.97.246.49])
	by svr1.postgresql.org (Postfix) with SMTP id E6E40D1B448
	for ;
	Mon, 26 Apr 2004 14:11:05 -0300 (ADT)
Received: (qmail 45601 invoked by uid 1000); 26 Apr 2004 17:11:03 -0000
Message-ID: <20040426171103.45600.qmail@uno.mnl.com>
From: brad-pgperf@duttonbros.com
To: pgsql-performance@postgresql.org
Subject: planner/optimizer question
Date: Mon, 26 Apr 2004 10:11:03 -0700
Mime-Version: 1.0
Content-Type: text/plain; format=flowed; charset="utf-8"
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0 tests=NO_REAL_NAME
X-Spam-Level: 
X-Archive-Number: 200404/446
X-Sequence-Number: 6746

Hi, 

I have a query which I think should be using an index all of the time but 
postgres only uses the index part of the time.  The index 
(ticket_crm_map_crm_id_suppid) has the where clause column (crm_id) listed 
first followed by the selected column (support_person_id).  Wouldn't the 
most efficient plan be to scan the index regardless of crm_id because the 
only columns needed are in the index?  Below is the table, 2 queries showing 
the difference in plans, followed by the record distribution of 
ticket_crm_map.  I first did a 'vacuum analyze' and am running postgres 
7.4.2. 

Thanks,
Brad 


athenapost=> \d ticket_crm_map
                                Table "public.ticket_crm_map"
      Column         |            Type             |                 
Modifiers
 ------------------------+-----------------------------+--------------------- 
 -----------------------
tcrm_map_id            | integer                     | not null
ticket_id              | integer                     | not null
crm_id                 | integer                     | not null
support_person_id      | integer                     | not null
escalated_to_person_id | integer                     | not null
status                 | character varying(50)       | not null default 
'Open'::character varying
close_date             | timestamp without time zone |
updated_date           | timestamp without time zone |
updated_by             | character varying(255)      |
created_date           | timestamp without time zone |
created_by             | character varying(255)      |
additional_info        | text                        |
subject                | character varying(255)      |
Indexes:
 "ticket_crm_map_pkey" primary key, btree (tcrm_map_id)
 "ticket_crm_map_crm_id_key" unique, btree (crm_id, ticket_id)
 "ticket_crm_map_crm_id_suppid" btree (crm_id, support_person_id)
 "ticket_crm_map_status" btree (status)
 "ticket_crm_map_ticket_id" btree (ticket_id)
Foreign-key constraints:
 "$1" FOREIGN KEY (ticket_id) REFERENCES ticket(ticket_id)
 "$2" FOREIGN KEY (crm_id) REFERENCES company_crm(crm_id)
 "$3" FOREIGN KEY (support_person_id) REFERENCES person(person_id)
 "$4" FOREIGN KEY (escalated_to_person_id) REFERENCES person(person_id)
 "$5" FOREIGN KEY (status) REFERENCES ticket_status(status) 

athenapost=> explain analyze select distinct support_person_id from 
ticket_crm_map where crm_id = 7;
                                                                         
QUERY PLAN
 ---------------------------------------------------------------------------- 
 ---------------------------------------------------------------------------- 
 ----------
Unique  (cost=1262.99..1265.27 rows=1 width=4) (actual time=15.335..18.245 
rows=20 loops=1)
 ->  Sort  (cost=1262.99..1264.13 rows=456 width=4) (actual 
time=15.332..16.605 rows=2275 loops=1)
      Sort Key: support_person_id
      ->  Index Scan using ticket_crm_map_crm_id_suppid on ticket_crm_map  
(cost=0.00..1242.85 rows=456 width=4) (actual time=0.055..11.281 rows=2275 
loops=1)
            Index Cond: (crm_id = 7)
Total runtime: 18.553 ms
(6 rows) 

Time: 20.598 ms
athenapost=> explain analyze select distinct support_person_id from 
ticket_crm_map where crm_id = 1;
                                                        QUERY PLAN
 ---------------------------------------------------------------------------- 
 -----------------------------------------------------
Unique  (cost=10911.12..11349.26 rows=32 width=4) (actual 
time=659.102..791.517 rows=24 loops=1)
 ->  Sort  (cost=10911.12..11130.19 rows=87628 width=4) (actual 
time=659.090..713.285 rows=93889 loops=1)
      Sort Key: support_person_id
      ->  Seq Scan on ticket_crm_map  (cost=0.00..3717.25 rows=87628 
width=4) (actual time=0.027..359.299 rows=93889 loops=1)
            Filter: (crm_id = 1)
Total runtime: 814.601 ms
(6 rows) 

Time: 817.095 ms
athenapost=> select count(*), crm_id from ticket_crm_map group by crm_id;
count | crm_id
 -------+--------
2554 |     63
129 |     25
 17 |     24
110 |     23
 74 |     22
 69 |     21
  2 |     20
 53 |     82
 10 |     17
 16 |     81
46637 |     16
 14 |     80
  2 |     15
1062 |     79
 87 |     78
 93 |     77
 60 |     44
363 |     76
225 |     10
  4 |     74
 83 |      9
 27 |     73
182 |      8
2275 |      7
 15 |     71
554 |      6
 44 |     70
631 |      5
 37 |      4
190 |      3
112 |      2
93889 |      1
(32 rows) 

Time: 436.697 ms 


From pgsql-performance-owner@postgresql.org  Mon Apr 26 16:46:47 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id AFF98D1B55E
	for ;
	Mon, 26 Apr 2004 14:21:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 40197-10
	for ;
	Mon, 26 Apr 2004 14:20:59 -0300 (ADT)
Received: from mail.astudios.com (ns1.astudios.com [66.49.77.3])
	by svr1.postgresql.org (Postfix) with ESMTP id 2457ED1B47F
	for ;
	Mon, 26 Apr 2004 14:20:58 -0300 (ADT)
thread-index: AcQrs1J3TY2Q3ehWSlu3D9wgKvl1cw==
Content-Class: urn:content-classes:message
Importance: normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.132
Received: from [10.4.2.156] ([66.194.26.150]) by mail.astudios.com with
	Microsoft SMTPSVC(6.0.3790.0); Mon, 26 Apr 2004 13:24:25 -0400
From: "Chris Hoover" 
To: "Tom Lane" 
Subject: Re: Help with performance problems
Date: Mon, 26 Apr 2004 13:20:56 -0400
User-Agent: KMail/1.6.1
Cc: 
References: <200404230931.17850.revoohc@sermonaudio.com>
	<6865.1082775504@sss.pgh.pa.us>
In-Reply-To: <6865.1082775504@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Disposition: inline
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Message-ID: <200404261320.56455.revoohc@sermonaudio.com>
X-OriginalArrivalTime: 26 Apr 2004 17:24:25.0406 (UTC)
	FILETIME=[525B1DE0:01C42BB3]
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/445
X-Sequence-Number: 6745

Ok, I was able to run a vacuumdb -f -v on my largest db over the weekend.=
=20=20
However, I am having trouble reading the results of the table portion.  Her=
e=20
area a couple of tables, what should I be looking at.  First table is the k=
ey=20
table to the db, and the second is the largest table in the db.=20=20

Thanks Chris

INFO:  --Relation public.clmhdr--
INFO:  Pages 32191: Changed 0, reaped 5357, Empty 0, New 0; Tup 339351: Vac=
=20
48358, Keep/VTL 0/0, UnUsed 129, MinLen 560, MaxLen 696; Re-using: Free/Av
ail. Space 42011004/32546120; EndEmpty/Avail. Pages 0/5310.
        CPU 0.53s/0.09u sec elapsed 0.61 sec.
INFO:  Index clmhdr_pkey: Pages 1429; Tuples 339351: Deleted 48358.
        CPU 0.06s/0.28u sec elapsed 4.54 sec.
INFO:  Index clmhdr_hdr_user_id_idx: Pages 1711; Tuples 339351: Deleted 483=
58.
        CPU 0.09s/0.31u sec elapsed 2.40 sec.
INFO:  Index clmhdr_hdr_clm_status_idx: Pages 1237; Tuples 339351: Deleted=
=20
48358.
        CPU 0.03s/0.26u sec elapsed 1.66 sec.
INFO:  Index clmhdr_hdr_create_dt_idx: Pages 1475; Tuples 339351: Deleted=
=20
48358.
        CPU 0.05s/0.24u sec elapsed 1.96 sec.
INFO:  Index clmhdr_inv_idx: Pages 1429; Tuples 339351: Deleted 48358.
        CPU 0.08s/0.22u sec elapsed 1.20 sec.
INFO:  Index clmhdr_userid_status_idx: Pages 2161; Tuples 339351: Deleted=
=20
48358.
        CPU 0.05s/0.18u sec elapsed 3.02 sec.
INFO:  Rel clmhdr: Pages: 32191 --> 28247; Tuple(s) moved: 8257.
        CPU 0.37s/1.81u sec elapsed 16.24 sec.
INFO:  Index clmhdr_pkey: Pages 1429; Tuples 339351: Deleted 8257.
        CPU 0.00s/0.03u sec elapsed 0.03 sec.
INFO:  Index clmhdr_hdr_user_id_idx: Pages 1743; Tuples 339351: Deleted 825=
7.
        CPU 0.00s/0.05u sec elapsed 0.04 sec.
INFO:  Index clmhdr_hdr_clm_status_idx: Pages 1265; Tuples 339351: Deleted=
=20
8257.
        CPU 0.00s/0.04u sec elapsed 0.03 sec.
INFO:  Index clmhdr_hdr_create_dt_idx: Pages 1503; Tuples 339351: Deleted=
=20
8257.
        CPU 0.00s/0.04u sec elapsed 0.12 sec.
INFO:  Index clmhdr_inv_idx: Pages 1429; Tuples 339351: Deleted 8257.
        CPU 0.00s/0.04u sec elapsed 0.03 sec.
INFO:  Index clmhdr_userid_status_idx: Pages 2203; Tuples 339351: Deleted=
=20
8257.
        CPU 0.01s/0.03u sec elapsed 0.04 sec.

INFO:  --Relation public.sent837--
INFO:  Pages 463552: Changed 0, reaped 6690, Empty 0, New 0; Tup 27431539: =
Vac=20
204348, Keep/VTL 0/0, UnUsed 2801, MinLen 107, MaxLen 347; Re-using:=20
Free/Avail. Space 54541468/34925860; EndEmpty/Avail. Pages 0/70583.
        CPU 10.68s/2.18u sec elapsed 188.32 sec.
INFO:  Index sent837_pkey: Pages 124424; Tuples 27431539: Deleted 204348.
        CPU 4.24s/3.45u sec elapsed 144.79 sec.
INFO:  Rel sent837: Pages: 463552 --> 459954; Tuple(s) moved: 91775.
        CPU 1.12s/9.36u sec elapsed 20.13 sec.
INFO:  Index sent837_pkey: Pages 124424; Tuples 27431539: Deleted 91775.
        CPU 3.51s/2.03u sec elapsed 6.13 sec.


From pgsql-performance-owner@postgresql.org  Mon Apr 26 19:46:42 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E71C9D1B50D
	for ;
	Mon, 26 Apr 2004 15:58:16 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 97239-04
	for ;
	Mon, 26 Apr 2004 15:58:11 -0300 (ADT)
Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210])
	by svr1.postgresql.org (Postfix) with ESMTP id E5821D1B4F6
	for ;
	Mon, 26 Apr 2004 15:58:08 -0300 (ADT)
Received: by megazone.bigpanda.com (Postfix, from userid 1001)
	id CA06335778; Mon, 26 Apr 2004 11:58:09 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1])
	by megazone.bigpanda.com (Postfix) with ESMTP
	id C862335754; Mon, 26 Apr 2004 11:58:09 -0700 (PDT)
Date: Mon, 26 Apr 2004 11:58:09 -0700 (PDT)
From: Stephan Szabo 
To: brad-pgperf@duttonbros.com
Cc: pgsql-performance@postgresql.org
Subject: Re: index usage
In-Reply-To: <20040423222121.12061.qmail@uno.mnl.com>
Message-ID: <20040426115215.R11044@megazone.bigpanda.com>
References: <20040423222121.12061.qmail@uno.mnl.com>
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/450
X-Sequence-Number: 6750


On Fri, 23 Apr 2004 brad-pgperf@duttonbros.com wrote:

> I have a query which I think should be using an index all of the time but
> postgres only uses the index part of the time.  The index
> (ticket_crm_map_crm_id_suppid) has the where clause column (crm_id) listed
> first followed by the selected column (support_person_id).  Wouldn't the
> most efficient plan be to scan the index each time because the only columns
> needed are in the index?  Below is the table, 2 queries showing the

Not necessarily.  The rows in the actual file still need to be checked to
see if they're visible to the select and if it's expected that the entire
file (or a reasonable % of the pages anyway) will need to be loaded using
the index isn't necessarily a win.

> athenapost=> explain analyze select distinct support_person_id from
> ticket_crm_map where crm_id = 1;
>                                                           QUERY PLAN
>  ----------------------------------------------------------------------------
>  -----------------------------------------------------
> Unique  (cost=10911.12..11349.26 rows=32 width=4) (actual
> time=659.102..791.517 rows=24 loops=1)
>   ->  Sort  (cost=10911.12..11130.19 rows=87628 width=4) (actual
> time=659.090..713.285 rows=93889 loops=1)
>         Sort Key: support_person_id
>         ->  Seq Scan on ticket_crm_map  (cost=0.00..3717.25 rows=87628
> width=4) (actual time=0.027..359.299 rows=93889 loops=1)
>               Filter: (crm_id = 1)
> Total runtime: 814.601 ms

How far off is this from the index scan version in time?  Try doing
set enable_seqscan=off; and then explain analyzing again.
It's possible that you may wish to lower random_page_cost to change the
estimated effect of how much more expensive random reads are compared to
sequential ones.

From pgsql-performance-owner@postgresql.org  Mon Apr 26 19:25:12 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A0747D1B437
	for ;
	Mon, 26 Apr 2004 16:16:30 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 06849-01
	for ;
	Mon, 26 Apr 2004 16:16:29 -0300 (ADT)
Received: from uno.mnl.com (uno.mnl.com [63.97.246.49])
	by svr1.postgresql.org (Postfix) with SMTP id A0112D1B520
	for ;
	Mon, 26 Apr 2004 16:16:27 -0300 (ADT)
Received: (qmail 47297 invoked by uid 1000); 26 Apr 2004 19:16:28 -0000
Message-ID: <20040426191628.47296.qmail@uno.mnl.com>
References: <20040423222121.12061.qmail@uno.mnl.com>
	<20040426115215.R11044@megazone.bigpanda.com>
In-Reply-To: <20040426115215.R11044@megazone.bigpanda.com>
From: brad-pgperf@duttonbros.com
To: Stephan Szabo ,
	pgsql-performance@postgresql.org
Subject: Re: index usage
Date: Mon, 26 Apr 2004 12:16:28 -0700
Mime-Version: 1.0
Content-Type: text/plain; format=flowed; charset="utf-8"
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0 tests=NO_REAL_NAME
X-Spam-Level: 
X-Archive-Number: 200404/449
X-Sequence-Number: 6749


When checking an index in postgres the original table has to be checked for 
each result to find if the index entry is still valid?  In which case you 
can't blindly scan the whole index and assume the data is good. I was used 
to Oracle behavior where the index is up to date so it can do the scan 
without hitting the original table. 

Does this sound correct to anyone? 

Thanks,
Brad 


Stephan Szabo writes:
> On Fri, 23 Apr 2004 brad-pgperf@duttonbros.com wrote: 
> 
>> I have a query which I think should be using an index all of the time but
>> postgres only uses the index part of the time.  The index
>> (ticket_crm_map_crm_id_suppid) has the where clause column (crm_id) listed
>> first followed by the selected column (support_person_id).  Wouldn't the
>> most efficient plan be to scan the index each time because the only columns
>> needed are in the index?  Below is the table, 2 queries showing the
> 
> Not necessarily.  The rows in the actual file still need to be checked to
> see if they're visible to the select and if it's expected that the entire
> file (or a reasonable % of the pages anyway) will need to be loaded using
> the index isn't necessarily a win. 
> 
 


From pgsql-performance-owner@postgresql.org  Mon Apr 26 19:24:22 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1651AD1B7EF
	for ;
	Mon, 26 Apr 2004 16:20:46 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 07896-01
	for ;
	Mon, 26 Apr 2004 16:20:45 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 72A0ED1B587
	for ;
	Mon, 26 Apr 2004 16:20:43 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4950733; Mon, 26 Apr 2004 12:22:10 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: "Magnus Naeslund(t)" , Tom Lane 
Subject: Re: Wierd context-switching issue on Xeon
Date: Mon, 26 Apr 2004 12:20:58 -0700
User-Agent: KMail/1.5.4
Cc: Joe Conway , "scott.marlowe" ,
	Bruce Momjian , lutzeb@aeccom.com,
	pgsql-performance@postgresql.org, Neil Conway 
References: 
	<1931.1082422389@sss.pgh.pa.us> <4085A895.6080403@fbab.net>
In-Reply-To: <4085A895.6080403@fbab.net>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404261220.58257.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/448
X-Sequence-Number: 6748

Magus,

> It would be interesting to see what a locking implementation ala FUTEX 
> style would give on an 2.6 kernel, as i understood it that would work 
> cross process with some work.

I'mm working on testing a FUTEX patch, but am having some trouble with it.  
Will let you know the results ....

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-admin-owner@postgresql.org  Mon Apr 26 20:00:10 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C5263D1B478
	for ;
	Mon, 26 Apr 2004 20:00:06 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 02804-07
	for ;
	Mon, 26 Apr 2004 20:00:05 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 2FE40D1B46C
	for ; Mon, 26 Apr 2004 20:00:02 -0300 (ADT)
Received: (qmail 10274 invoked from network); 26 Apr 2004 23:00:00 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 26 Apr 2004 23:00:00 -0000
Subject: Re: [JDBC] is a good practice to create an index on the oid?
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Edoardo Ceccarelli 
Cc: "pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
In-Reply-To: <408D3AE8.5020006@axa.it>
References: <408D3AE8.5020006@axa.it>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1083020647.13952.164.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Mon, 26 Apr 2004 19:04:07 -0400
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-Spam-Status: No, hits=3.0 tagged_above=0.0 required=5.0
	tests=SUSPICIOUS_RECIPS
X-Spam-Level: ***
X-Archive-Number: 200404/314
X-Sequence-Number: 13287

Yes, you can create an index on the oid, but unless you are selecting on
it, it is of little use.

you would have to do select * from foo where oid=? to get any value out
of the index.

Dave
On Mon, 2004-04-26 at 12:38, Edoardo Ceccarelli wrote:
> I am using the oid of the table as the main key and I've found that is 
> not indexed (maybe because I have declared another primary key in the table)
> 
> it is a good practice to create an index like this on the oid of a table?
> CREATE INDEX idoid annuncio400 USING btree (oid);
> 
> 
> does it work as a normal index?
> 
> Thank you
> Edoardo
> 
> ---------------------------(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
> 
> 
> 
> !DSPAM:408d7c38183971270217895!
> 
> 
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Mon Apr 26 21:03:13 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 82C54D1B48B
	for ;
	Mon, 26 Apr 2004 21:03:11 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 21012-04
	for ;
	Mon, 26 Apr 2004 21:03:11 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 3E0F1D1B4D7
	for ;
	Mon, 26 Apr 2004 21:03:08 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4952398; Mon, 26 Apr 2004 17:04:37 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: pg@fastcrypt.com, Tom Lane 
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Mon, 26 Apr 2004 17:03:25 -0700
User-Agent: KMail/1.5.4
Cc: pgsql-performance@postgresql.org
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<18322.1082601330@sss.pgh.pa.us>
	<1082603201.1556.272.camel@localhost.localdomain>
In-Reply-To: <1082603201.1556.272.camel@localhost.localdomain>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404261703.25631.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/452
X-Sequence-Number: 6752

Dave,

> Yeah, I did some more testing myself, and actually get better numbers
> with increasing spins per delay to 1000, but my suspicion is that it is
> highly dependent on finding the right delay for the processor you are
> on.

Well, it certainly didn't help here:

procs                      memory      swap          io     system         cpu
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 2  0      0 14870744 123872 1129912    0    0     0     0 1027 187341 48 27 
26  0
 2  0      0 14869912 123872 1129912    0    0     0    48 1030 126490 65 18 
16  0
 2  0      0 14867032 123872 1129912    0    0     0     0 1021 106046 72 16 
12  0
 2  0      0 14869912 123872 1129912    0    0     0     0 1025 90256 76 14 10  
0
 2  0      0 14870424 123872 1129912    0    0     0     0 1022 135249 63 22 
16  0
 2  0      0 14872664 123872 1129912    0    0     0     0 1023 131111 63 20 
17  0
 1  0      0 14871128 123872 1129912    0    0     0    48 1024 155728 57 22 
20  0
 2  0      0 14871128 123872 1129912    0    0     0     0 1028 189655 49 29 
22  0
 2  0      0 14871064 123872 1129912    0    0     0     0 1018 190744 48 29 
23  0
 2  0      0 14871064 123872 1129912    0    0     0     0 1027 186812 51 26 
23  0


-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Mon Apr 26 21:11:57 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6E399D1B47F
	for ;
	Mon, 26 Apr 2004 21:11:55 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 23925-06
	for ;
	Mon, 26 Apr 2004 21:11:56 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 831C8D1B475
	for ;
	Mon, 26 Apr 2004 21:11:52 -0300 (ADT)
Received: (qmail 28807 invoked from network); 27 Apr 2004 00:11:53 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 27 Apr 2004 00:11:53 -0000
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Josh Berkus 
Cc: Tom Lane , pgsql-performance@postgresql.org
In-Reply-To: <200404261703.25631.josh@agliodbs.com>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<18322.1082601330@sss.pgh.pa.us>
	<1082603201.1556.272.camel@localhost.localdomain>
	<200404261703.25631.josh@agliodbs.com>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1083024960.14759.166.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Mon, 26 Apr 2004 20:16:00 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/453
X-Sequence-Number: 6753

Are you testing this with Tom's code, you need to do a baseline
measurement with 10 and then increase it, you will still get lots of cs,
but it will be less.

Dave
On Mon, 2004-04-26 at 20:03, Josh Berkus wrote:
> Dave,
> 
> > Yeah, I did some more testing myself, and actually get better numbers
> > with increasing spins per delay to 1000, but my suspicion is that it is
> > highly dependent on finding the right delay for the processor you are
> > on.
> 
> Well, it certainly didn't help here:
> 
> procs                      memory      swap          io     system         cpu
>  r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
>  2  0      0 14870744 123872 1129912    0    0     0     0 1027 187341 48 27 
> 26  0
>  2  0      0 14869912 123872 1129912    0    0     0    48 1030 126490 65 18 
> 16  0
>  2  0      0 14867032 123872 1129912    0    0     0     0 1021 106046 72 16 
> 12  0
>  2  0      0 14869912 123872 1129912    0    0     0     0 1025 90256 76 14 10  
> 0
>  2  0      0 14870424 123872 1129912    0    0     0     0 1022 135249 63 22 
> 16  0
>  2  0      0 14872664 123872 1129912    0    0     0     0 1023 131111 63 20 
> 17  0
>  1  0      0 14871128 123872 1129912    0    0     0    48 1024 155728 57 22 
> 20  0
>  2  0      0 14871128 123872 1129912    0    0     0     0 1028 189655 49 29 
> 22  0
>  2  0      0 14871064 123872 1129912    0    0     0     0 1018 190744 48 29 
> 23  0
>  2  0      0 14871064 123872 1129912    0    0     0     0 1027 186812 51 26 
> 23  0
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-admin-owner@postgresql.org  Mon Apr 26 23:43:17 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id F367FD1B47C; Mon, 26 Apr 2004 23:43:07 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 69259-02; Mon, 26 Apr 2004 23:43:10 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP
	id CDDC7D1B44C; Mon, 26 Apr 2004 23:43:04 -0300 (ADT)
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
	i3R2gtWL087170; Tue, 27 Apr 2004 10:42:59 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <408DC9E1.8020704@familyhealth.com.au>
Date: Tue, 27 Apr 2004 10:48:01 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Edoardo Ceccarelli 
Cc: pgsql-jdbc@postgresql.org, pgsql-performance@postgresql.org,
	pgsql-admin@postgresql.org
Subject: Re: [PERFORM] is a good practice to create an index on the oid?
References: <408D3AE8.5020006@axa.it>
In-Reply-To: <408D3AE8.5020006@axa.it>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/315
X-Sequence-Number: 13288

> I am using the oid of the table as the main key and I've found that is 
> not indexed (maybe because I have declared another primary key in the 
> table)
> 
> it is a good practice to create an index like this on the oid of a table?
> CREATE INDEX idoid annuncio400 USING btree (oid);

Yes it is - in fact you really should add a unique index, not just a 
normal index, as you want to enforce uniqueness of the oid column.  It 
is theoretically possible to end up with duplicate oids in wraparound 
situations.

Even better though is to not use oids at all, of course...

Chris


From pgsql-admin-owner@postgresql.org  Tue Apr 27 08:55:42 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 047B1D1B8BD
	for ;
	Tue, 27 Apr 2004 08:55:37 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 41113-09
	for ;
	Tue, 27 Apr 2004 08:55:37 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 7F492D1B554
	for ; Tue, 27 Apr 2004 08:55:33 -0300 (ADT)
Received: (qmail 12966 invoked from network); 27 Apr 2004 11:55:32 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 27 Apr 2004 11:55:32 -0000
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Christopher Kings-Lynne 
Cc: Edoardo Ceccarelli ,
	"pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
In-Reply-To: <408DC9E1.8020704@familyhealth.com.au>
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1083067187.14759.178.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Tue, 27 Apr 2004 07:59:47 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/317
X-Sequence-Number: 13290

AFAIK, oids aren't used for anything internally, so duplicates don't
really matter. Besides, what would you do about duplicate oid's ?

The best suggestion is of course his last, don't use them.


On Mon, 2004-04-26 at 22:48, Christopher Kings-Lynne wrote:
> > I am using the oid of the table as the main key and I've found that is 
> > not indexed (maybe because I have declared another primary key in the 
> > table)
> > 
> > it is a good practice to create an index like this on the oid of a table?
> > CREATE INDEX idoid annuncio400 USING btree (oid);
> 
> Yes it is - in fact you really should add a unique index, not just a 
> normal index, as you want to enforce uniqueness of the oid column.  It 
> is theoretically possible to end up with duplicate oids in wraparound 
> situations.
> 
> Even better though is to not use oids at all, of course...
> 
> Chris
> 
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
> 
>                http://www.postgresql.org/docs/faqs/FAQ.html
> 
> 
> 
> !DSPAM:408dcc51235334924183622!
> 
> 
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Tue Apr 27 09:09:51 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id AF00FD1B590
	for ;
	Tue, 27 Apr 2004 09:09:50 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 49620-10
	for ;
	Tue, 27 Apr 2004 09:09:50 -0300 (ADT)
Received: from email03.aon.at (WARSL402PIP6.highway.telekom.at [195.3.96.93])
	by svr1.postgresql.org (Postfix) with SMTP id E6175D1B570
	for ;
	Tue, 27 Apr 2004 09:09:46 -0300 (ADT)
Received: (qmail 175214 invoked from network); 27 Apr 2004 12:09:48 -0000
Received: from m168p020.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.10.244]) (envelope-sender )
	by qmail3rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 27 Apr 2004 12:09:48 -0000
From: Manfred Koizar 
To: Qing Zhao 
Cc: =?ISO-8859-1?Q?Fr=E9d=E9ric_Robinet?= ,
	
Subject: Re: Shared buffers, Sort memory, Effective Cache Size
Date: Tue, 27 Apr 2004 14:12:20 +0200
Message-ID: 
References: <009901c427bd$c4752410$c601a8c0@MIMS>
	<8937BC90-93B5-11D8-AC12-000A95AB8896@supplyfx.com>
In-Reply-To: <8937BC90-93B5-11D8-AC12-000A95AB8896@supplyfx.com>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/456
X-Sequence-Number: 6756

On Wed, 21 Apr 2004 10:01:30 -0700, Qing Zhao 
wrote:
>I have recently configured my PG7.3 on a G5 (8GB RAM) with
>shmmax set to 512MB and shared_buffer=50000, sort_mem=4096
>and effective cache size = 10000.  It seems working great so far but
>I am wondering if I should make effctive cache size larger myself.

Yes, much larger!  And while you are at it make shared_buffers smaller.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Tue Apr 27 09:54:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1B9E2D1B503
	for ;
	Tue, 27 Apr 2004 09:54:17 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 72453-02
	for ;
	Tue, 27 Apr 2004 09:54:17 -0300 (ADT)
Received: from 192.168.0.11 (fw.i-trade.hu [212.92.1.62])
	by svr1.postgresql.org (Postfix) with SMTP id 98C48D1B4E6
	for ;
	Tue, 27 Apr 2004 09:54:12 -0300 (ADT)
Received: (qmail 22728 invoked from network); 27 Apr 2004 12:34:16 -0000
Received: from unknown (HELO atesz) (212.92.7.130)
	by 0 with SMTP; 27 Apr 2004 12:34:15 -0000
From: "Atesz" 
To: 
Subject: Re: planner/optimizer question
Date: Tue, 27 Apr 2004 14:54:27 +0200
Message-ID: <002901c42c56$c8e077a0$0b02010a@atesz>
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
Importance: Normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106
In-Reply-To: <20040426171103.45600.qmail@uno.mnl.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/457
X-Sequence-Number: 6757

Hi,

You should try the next queries:

select support_person_id from ticket_crm_map where crm_id = 7 GROUP BY
support_person_id;
select support_person_id from ticket_crm_map where crm_id = 1 GROUP BY
support_person_id;

It can use the 'ticket_crm_map_crm_id_suppid' index. 

Generally the Postgres use an k-column index if columns of your
conditions are prefix of the index column.
For example:
CREATE INDEX test_idx on test(col1,col2,col3,col4);
SELECT * FROM test WHERE col1=3 AND col2=13;  -- This can use the index.

But the next queries cannot use the index:
SELECT * FROM test WHERE col1=3 AND col3=13;.
SELECT * FROM test WHERE col2=3;

If you have problem with seq_scan or sort, you can disable globally and
locally: 
SET enable_seqscan=0;
SET enable_sort = 0;

Regards, Antal Attila



From pgsql-admin-owner@postgresql.org  Tue Apr 27 12:01:29 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 20B36D1B4DF; Tue, 27 Apr 2004 12:01:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 27869-02; Tue, 27 Apr 2004 12:01:23 -0300 (ADT)
Received: from vscan02.westnet.com.au (vscan02.westnet.com.au [203.10.1.132])
	by svr1.postgresql.org (Postfix) with ESMTP
	id BD55AD1B4B3; Tue, 27 Apr 2004 12:01:22 -0300 (ADT)
Received: from localhost (localhost.localdomain [127.0.0.1])
	by localhost (Postfix) with ESMTP id 1F05C100F39;
	Tue, 27 Apr 2004 23:01:18 +0800 (WST)
Received: from familyhealth.com.au (dsl-202-72-133-22.wa.westnet.com.au
	[202.72.133.22])
	by vscan02.westnet.com.au (Postfix) with ESMTP id 90034100DE2;
	Tue, 27 Apr 2004 23:01:17 +0800 (WST)
Message-ID: <408E75C2.5060703@familyhealth.com.au>
Date: Tue, 27 Apr 2004 23:01:22 +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: pg@fastcrypt.com
Cc: Edoardo Ceccarelli ,
	"pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>
	<1083067187.14759.178.camel@localhost.localdomain>
In-Reply-To: <1083067187.14759.178.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-Spam-Status: No, hits=3.0 tagged_above=0.0 required=5.0
	tests=SUSPICIOUS_RECIPS
X-Spam-Level: ***
X-Archive-Number: 200404/320
X-Sequence-Number: 13293

> AFAIK, oids aren't used for anything internally, so duplicates don't
> really matter. Besides, what would you do about duplicate oid's ?

If he's using them _externally_, then he does have to worry about 
duplicates.

Chris


From pgsql-admin-owner@postgresql.org  Tue Apr 27 12:14:28 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6A042D1B4FD
	for ;
	Tue, 27 Apr 2004 12:14:26 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 32819-04
	for ;
	Tue, 27 Apr 2004 12:14:23 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 50717D1B475
	for ; Tue, 27 Apr 2004 12:14:23 -0300 (ADT)
Received: (qmail 17086 invoked from network); 27 Apr 2004 15:14:18 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 27 Apr 2004 15:14:18 -0000
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Christopher Kings-Lynne 
Cc: Edoardo Ceccarelli ,
	"pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
In-Reply-To: <408E75C2.5060703@familyhealth.com.au>
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>
	<1083067187.14759.178.camel@localhost.localdomain>
	<408E75C2.5060703@familyhealth.com.au>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1083079114.14759.204.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Tue, 27 Apr 2004 11:18:34 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/321
X-Sequence-Number: 13294

Edoardo, 

Are you using them for referential integrity? If so you would be wise to
use sequences instead. 

Christopher: yes you are correct, I wasn't sure if that is what he was
doing.

Dave
On Tue, 2004-04-27 at 11:01, Christopher Kings-Lynne wrote:
> > AFAIK, oids aren't used for anything internally, so duplicates don't
> > really matter. Besides, what would you do about duplicate oid's ?
> 
> If he's using them _externally_, then he does have to worry about 
> duplicates.
> 
> Chris
> 
> 
> 
> !DSPAM:408e75e0137721921318500!
> 
> 
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Tue Apr 27 14:32:49 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 3C6C8D1B577
	for ;
	Tue, 27 Apr 2004 14:32:41 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 92717-06
	for ;
	Tue, 27 Apr 2004 14:32:39 -0300 (ADT)
Received: from mail.informatik.hu-berlin.de (mail.informatik.hu-berlin.de
	[141.20.20.50])
	by svr1.postgresql.org (Postfix) with ESMTP id 5438FD1B570
	for ;
	Tue, 27 Apr 2004 14:32:37 -0300 (ADT)
Received: from informatik.hu-berlin.de (karotte [141.20.27.143])
	(authenticated bits=0) by mail.informatik.hu-berlin.de
	(8.12.10/8.12.10/INF-2.0-MA-SOLARIS-2.8) with ESMTP id i3RHWaID026870
	(version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO)
	for ;
	Tue, 27 Apr 2004 19:32:37 +0200 (MEST)
Message-ID: <408E9934.8060509@informatik.hu-berlin.de>
Date: Tue, 27 Apr 2004 19:32:36 +0200
From: Silke Trissl 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; de-AT; rv:1.4) Gecko/20030624
X-Accept-Language: de-at, de, en-us, en
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Join problem
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/460
X-Sequence-Number: 6760

Hi,

I came across a very intriguing thing:

I had to join two tables and in both tables I wanted to restrict the 
result set by some (text/varchar) attributes.

Here is an example:

Table "item" # 147 000 entries

     Column     |         Type          | Modifiers
---------------+-----------------------+------------
  id            | integer               | not null
  description   | text                  |
  comment       | text                  | not null
  order_id      | integer               |


Table "orders" # 210 000 entries
       Column      |          Type          | Modifiers
-----------------+------------------------+-----------
  order_id        | integer                |
  order_name      | character varying(255) |


The tables have 147 000 and 210 000 entries, respectively.

First I tried the following query, which took ages:

(Query 1)
EXPLAIN ANALYZE
SELECT item.id
FROM item, orders
WHERE orders.order_name ~* 'Smit'
   AND item.description ~* 'CD'
   and orders.order_id = item.order_id;



I found out, that the change of the operator from '~*' to '=' for the 
item.description brought a great boost in performance (425 secs to 1 
sec!), but not in cost (Query plans at the end).

(Query 2)
  EXPLAIN ANALYZE
SELECT item.id
FROM item, orders
WHERE orders.order_name ~* 'Smit'
   AND item.description = 'CD'
   and orders.order_id = item.order_id;


The main difference was that Query 2 used the Hash join instead of the 
Nested Loop, so I disabled the option 'NESTED LOOP' and got for Query 1 
a similar time as for Query 2.


Can anyone tell me, why in one case the Hash join and in the other the 
much worse Nested Loop is prefered?
And my second question is, is there any possibility to execute the first 
query without disabling the Nested Loop first, but get the good 
performance of the Hash join?


Many thanks in advance for your help or suggestions

Silke


QUERY PLANS:

#####################################

Query 1:
   QUERY PLAN
------------------------------------------------------------------------------------------------------------------ 

  Nested Loop  (cost=0.00..28836.75 rows=1 width=4) (actual 
time=65350.780..452130.702 rows=6 loops=1)
    Join Filter: ("inner".order_id = "outer".order_id)
    ->  Seq Scan on item  (cost=0.00..28814.24 rows=1 width=8) (actual 
time=33.180..1365.190 rows=716 loops=1)
          Filter: (description ~* 'CD'::text)
    ->  Seq Scan on orders  (cost=0.00..22.50 rows=1 width=4) (actual 
time=21.644..629.500 rows=18 loops=716)
          Filter: ((order_name)::text ~* 'Smith'::text)
  Total runtime: 452130.782 ms
###########################################################################

Query 2:
   QUERY PLAN
------------------------------------------------------------------------------------------------------------------ 

   Hash Join  (cost=22.50..28840.44 rows=4 width=4) (actual 
time=1187.798..1187.798 rows=0 loops=1)
    Hash Cond: ("outer".order_id = "inner".order_id)
    ->  Seq Scan on item  (cost=0.00..28814.24 rows=733 width=8) (actual 
time=542.737..542.737 rows=0 loops=1)
          Filter: (description = 'CD'::text)
    ->  Hash  (cost=22.50..22.50 rows=1 width=4) (actual 
time=645.042..645.042 rows=0 loops=1)
          ->  Seq Scan on orders  (cost=0.00..22.50 rows=1 width=4) 
(actual time=22.373..644.996 rows=18 loops=1)
                Filter: ((order_name)::text ~* 'Smith'::text)
  Total runtime: 1187.865 ms
############################################################################


Query 1 with  'set enable_nestloop to false'

    QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------
  Hash Join  (cost=22.50..28836.75 rows=1 width=4) (actual 
time=1068.593..2003.330 rows=6 loops=1)
    Hash Cond: ("outer".item_id = "inner".item_id)
    ->  Seq Scan on item  (cost=0.00..28814.24 rows=1 width=8) (actual 
time=33.347..1357.073 rows=716 loops=1)
          Filter: (description ~* 'CD'::text)
    ->  Hash  (cost=22.50..22.50 rows=1 width=4) (actual 
time=645.287..645.287 rows=0 loops=1)
          ->  Seq Scan on orders  (cost=0.00..22.50 rows=1 width=4) 
(actual time=22.212..645.239 rows=18 loops=1)
                Filter: ((order_name)::text ~* 'CD'::text)
  Total runtime: 2003.409 ms


From pgsql-performance-owner@postgresql.org  Tue Apr 27 15:03:10 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9315CD1B432
	for ;
	Tue, 27 Apr 2004 15:03:09 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 08121-04
	for ;
	Tue, 27 Apr 2004 15:03:07 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id 1FAABD1B553
	for ;
	Tue, 27 Apr 2004 15:03:06 -0300 (ADT)
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 i3RI2gBh007433;
	Tue, 27 Apr 2004 12:02:42 -0600 (MDT)
Date: Tue, 27 Apr 2004 12:02:45 -0600 (MDT)
From: "scott.marlowe" 
To: Rob Fielding 
Cc: 
Subject: Re: OT: Help with performance problems
In-Reply-To: <408CD006.3090004@dsvr.net>
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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/461
X-Sequence-Number: 6761

On Mon, 26 Apr 2004, Rob Fielding wrote:

> scott.marlowe wrote:
> > On Fri, 23 Apr 2004, Chris Hoover wrote:
> > 
> > 
> >>DB's on Powervaults 220S using raid 5 (over 6 disks)
> > 
> > 
> > What controller is this, the adaptec?  We've found it to be slower than 
> > the LSI megaraid based controller, but YMMV.
> 
> Wow, really? You got any more details of the chipset, mobo and kernel 
> driver ?

We're running on a Dell 2650, the controller is the U320 LSI megaraid 2 
channel (they only make the one that I know of right now).  Don't know my 
mobo chipset offhand, but might be able to find out what one dell includes 
on the 2650.  The kernel driver is the latest megaraid2 driver as of about 
Feb this year.

> I've been taken to my wits end wrestling with an LSI MegaRAID 320-1 
> controller on a supermicro board all weekend. I just couldn't get 
> anything more than 10MB/sec out of it with megaraid driver v1 OR v2 in 
> Linux 2.4.26, nor the version in 2.6.6-rc2. After 2 days of humming the 
> Adaptec mantra I gave in and switched the array straight onto the 
> onboard Adaptec 160 controller (same cable and everything). Software 
> RAID 5 gets me over 40MB sec for a nominal cpu hit - more than 4 times 
> what I could get out of the MegaRAID controller :( Even the 2nd SCSI-2 
> channel gets 40MB/sec max (pg_xlog :)
> 
> And HOW LONG does it take to detect drives during POST....ohhhh never 
> mind ... I really just wanna rant :) There should be a free counseling 
> service for enraged sysops.

I wonder if your controller is broken or something?  Or maybe on a PCI 
slow that has to share IRQs or something.  I've had great luck with 
SuperMicro mobos in the past (we're talking dual PPro 200 mobos, so 
seriously, IN THE PAST here... )  Hell, my Dual PPro 200 with an old 
MegaRAID 428 got 18 Megs a second cfer rate no problem.

Have you tried that lsi card in another machine / mobo combo?  Can you 
disable the onboard adaptec?  We have on our Dell 2650s, the only active 
controllers are the onboard IDE and the add in LSI-320-2 controller.

We're running ours with 128 Meg cache (I think could be 64) set to write 
back.  I think our throughput on a RAID-1 pair was somewhere around 40+ 
megs a second reads with bonnie++  With RAID-5 it was not really much 
faster at reads (something like 60 megs a second) but was much more 
scalable under heavy parellel read/write access for PostgreSQL.

Have you updated the BIOS on the mobo to see if that helps?  I'm just 
throwing darts at the wall here.


From pgsql-performance-owner@postgresql.org  Tue Apr 27 15:05:23 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A81EBD1B44C
	for ;
	Tue, 27 Apr 2004 15:05:20 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 03257-10
	for ;
	Tue, 27 Apr 2004 15:05:19 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id A650ED1B432
	for ;
	Tue, 27 Apr 2004 15:05:17 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4957685; Tue, 27 Apr 2004 11:06:43 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: pg@fastcrypt.com
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Tue, 27 Apr 2004 11:05:00 -0700
User-Agent: KMail/1.4.3
Cc: Tom Lane , pgsql-performance@postgresql.org
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404261703.25631.josh@agliodbs.com>
	<1083024960.14759.166.camel@localhost.localdomain>
In-Reply-To: <1083024960.14759.166.camel@localhost.localdomain>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404271105.00395.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/462
X-Sequence-Number: 6762

Dave,

> Are you testing this with Tom's code, you need to do a baseline
> measurement with 10 and then increase it, you will still get lots of cs,
> but it will be less.

No, that was just a test of 1000 straight up.    Tom outlined a method, but I 
didn't see any code that would help me find a better level, other than just 
trying each +100 increase one at a time.   This would take days of testing 
...
-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Tue Apr 27 15:24:11 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D5E7DD1B4B3
	for ;
	Tue, 27 Apr 2004 15:23:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 17850-02
	for ;
	Tue, 27 Apr 2004 15:23:20 -0300 (ADT)
Received: from net2.micro-automation.com (net2.micro-automation.com
	[64.7.141.29])
	by svr1.postgresql.org (Postfix) with SMTP id 6C4C8D1B575
	for ;
	Tue, 27 Apr 2004 15:23:17 -0300 (ADT)
Received: (qmail 21878 invoked from network); 27 Apr 2004 18:23:15 -0000
Received: from dcdsl.ebox.com (HELO ?192.168.1.36?) (davec@64.7.143.116)
	by net2.micro-automation.com with SMTP; 27 Apr 2004 18:23:15 -0000
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
From: Dave Cramer 
Reply-To: pg@fastcrypt.com
To: Josh Berkus 
Cc: Tom Lane , pgsql-performance@postgresql.org
In-Reply-To: <200404271105.00395.josh@agliodbs.com>
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404261703.25631.josh@agliodbs.com>
	<1083024960.14759.166.camel@localhost.localdomain>
	<200404271105.00395.josh@agliodbs.com>
Content-Type: text/plain
Organization: Cramer Consulting
Message-Id: <1083090452.14759.224.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.6 
Date: Tue, 27 Apr 2004 14:27:33 -0400
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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/463
X-Sequence-Number: 6763

Josh,

I think you can safely increase by orders of magnitude here, instead of
by +100, my wild ass guess is that the sweet spot is the spin time
should be approximately the time it takes to consume the resource. So if
you have a really fast machine then the spin count should be higher. 

Also you have to take into consideration your memory bus speed, with the
pause instruction inserted in the loop the timing is now dependent on
memory speed.

But... you need a baseline first.

Dave
On Tue, 2004-04-27 at 14:05, Josh Berkus wrote:
> Dave,
> 
> > Are you testing this with Tom's code, you need to do a baseline
> > measurement with 10 and then increase it, you will still get lots of cs,
> > but it will be less.
> 
> No, that was just a test of 1000 straight up.    Tom outlined a method, but I 
> didn't see any code that would help me find a better level, other than just 
> trying each +100 increase one at a time.   This would take days of testing 
> ...
-- 
Dave Cramer
519 939 0336
ICQ # 14675561


From pgsql-performance-owner@postgresql.org  Tue Apr 27 18:02:55 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 06FC2D1B454
	for ;
	Tue, 27 Apr 2004 18:02:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 28052-03
	for ;
	Tue, 27 Apr 2004 18:02:54 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 8BB9DD1B45E
	for ;
	Tue, 27 Apr 2004 18:02:51 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4958932; Tue, 27 Apr 2004 14:04:18 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: pg@fastcrypt.com
Subject: Re: Wierd context-switching issue on Xeon patch for 7.4.1
Date: Tue, 27 Apr 2004 14:03:13 -0700
User-Agent: KMail/1.5.4
Cc: pgsql-performance@postgresql.org
References: 
 <4BAFBB6B9CC46F41B2AD7D9F4BBAF78509814A@vt-pe2550-001.vantage.vantage.com>
	<200404271105.00395.josh@agliodbs.com>
	<1083090452.14759.224.camel@localhost.localdomain>
In-Reply-To: <1083090452.14759.224.camel@localhost.localdomain>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404271403.13543.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/464
X-Sequence-Number: 6764

Dave,

> But... you need a baseline first.

A baseline on CS?   I have that ....

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Tue Apr 27 18:27:28 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 19887D1B4A3
	for ;
	Tue, 27 Apr 2004 18:27:28 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 31555-10
	for ;
	Tue, 27 Apr 2004 18:27:27 -0300 (ADT)
Received: from fep4.012.net.il (fep4.012.net.il [212.117.129.203])
	by svr1.postgresql.org (Postfix) with ESMTP id 49DC4D1B48A
	for ;
	Tue, 27 Apr 2004 18:27:24 -0300 (ADT)
Received: from 80.178.84.42.forward.012.net.il ([80.178.84.42])
	by fep4.012.net.il with ESMTP
	id <20040427212718.QOXC29794.fep4@80.178.84.42.forward.012.net.il>
	for ;
	Wed, 28 Apr 2004 00:27:18 +0300
Date: Wed, 28 Apr 2004 00:27:40 +0300
From: Vitaly Belman 
Reply-To: Vitaly Belman 
X-Priority: 3 (Normal)
Message-ID: <282909734.20040428002740@012.net.il>
To: pgsql-performance@postgresql.org
Subject: Simply join in PostrgeSQL takes too long
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=0.8 tagged_above=0.0 required=5.0
	tests=PRIORITY_NO_NAME
X-Spam-Level: 
X-Archive-Number: 200404/465
X-Sequence-Number: 6765

Hello pgsql-performance,

  I discussed the whole subject for some time in DevShed and didn't
  achieve much (as for results). I wonder if any of you guys can help
  out:

  http://forums.devshed.com/t136202/s.html

Regards,
 Vitaly Belman
 
 ICQ: 1912453
 AIM: VitalyB1984
 MSN: tmdagent@hotmail.com
 Yahoo!: VitalyBe


From pgsql-admin-owner@postgresql.org  Tue Apr 27 18:43:40 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 6A64FD1B50C; Tue, 27 Apr 2004 18:43:38 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36987-10; Tue, 27 Apr 2004 18:43:38 -0300 (ADT)
Received: from FE-mail04.sfg.albacom.net (FE-mail04.albacom.net
	[213.217.149.84]) by svr1.postgresql.org (Postfix) with ESMTP
	id 641CFD1B4F6; Tue, 27 Apr 2004 18:43:35 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail04.sfg.albacom.net (7.0.009)
	id 408E270C00015408; Tue, 27 Apr 2004 23:42:57 +0200
Message-ID: <408ED3E2.802@axa.it>
Date: Tue, 27 Apr 2004 23:42:58 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: pg@fastcrypt.com
Cc: Christopher Kings-Lynne ,
	"pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>	
	<1083067187.14759.178.camel@localhost.localdomain>	
	<408E75C2.5060703@familyhealth.com.au>
	<1083079114.14759.204.camel@localhost.localdomain>
In-Reply-To: <1083079114.14759.204.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-Spam-Status: No, hits=3.0 tagged_above=0.0 required=5.0
	tests=SUSPICIOUS_RECIPS
X-Spam-Level: ***
X-Archive-Number: 200404/327
X-Sequence-Number: 13300

I am going to use them as primary key of the table, so I'll surely need 
them unique :)
thank you for you help
Edoardo

Dave Cramer ha scritto:

>Edoardo, 
>
>Are you using them for referential integrity? If so you would be wise to
>use sequences instead. 
>
>Christopher: yes you are correct, I wasn't sure if that is what he was
>doing.
>
>Dave
>On Tue, 2004-04-27 at 11:01, Christopher Kings-Lynne wrote:
>  
>
>>>AFAIK, oids aren't used for anything internally, so duplicates don't
>>>really matter. Besides, what would you do about duplicate oid's ?
>>>      
>>>
>>If he's using them _externally_, then he does have to worry about 
>>duplicates.
>>
>>Chris
>>
>>
>>
>>!DSPAM:408e75e0137721921318500!
>>
>>
>>    
>>

From pgsql-performance-owner@postgresql.org  Tue Apr 27 18:56:00 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6097CD1B8B7
	for ;
	Tue, 27 Apr 2004 18:55:58 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 47269-02
	for ;
	Tue, 27 Apr 2004 18:55:58 -0300 (ADT)
Received: from 192.168.0.11 (fw.i-trade.hu [212.92.1.62])
	by svr1.postgresql.org (Postfix) with SMTP id EB9D3D1B8A5
	for ;
	Tue, 27 Apr 2004 18:55:54 -0300 (ADT)
Received: (qmail 32297 invoked from network); 27 Apr 2004 21:35:54 -0000
Received: from unknown (HELO atesz) (212.92.7.130)
	by 0 with SMTP; 27 Apr 2004 21:35:54 -0000
From: "Atesz" 
To: "'Vitaly Belman'" , 
Subject: Re: Simply join in PostrgeSQL takes too long
Date: Tue, 27 Apr 2004 23:56:07 +0200
Message-ID: <002a01c42ca2$7492ac30$0b02010a@atesz>
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
Importance: Normal
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106
In-Reply-To: <282909734.20040428002740@012.net.il>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/467
X-Sequence-Number: 6767

Hi,

You can try some variation:

SELECT 
  book_id
FROM  
  bookgenres, genre_children
WHERE
   bookgenres.genre_id = genre_children.genre_child_id AND  
   genre_children.genre_id = 1
GROUP BY book_id
LIMIT 10

The next works if the 'genre_child_id' is UNIQUE on the 'genre_children'
table.

SELECT 
  book_id
FROM  
  bookgenres
WHERE
   bookgenres.genre_id = (SELECT genre_child_id FROM genre_children
WHERE genre_id = 1)
GROUP BY book_id
LIMIT 10

You may need some index. Try these with EXPLAIN!
CREATE INDEX bookgenres_genre_id_book_id ON bookgenres(genre_id,
book_id);  or
CREATE INDEX bookgenres_book_id_genre_id ON bookgenres(book_id,
genre_id);
CREATE INDEX genre_children_genre_id ON genre_children(genre_id);

Regards, Antal Attila



From pgsql-performance-owner@postgresql.org  Tue Apr 27 19:00:15 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id EE54AD1BCA7
	for ;
	Tue, 27 Apr 2004 19:00:13 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 49648-02
	for ;
	Tue, 27 Apr 2004 19:00:08 -0300 (ADT)
Received: from obie.mrkdns.com (obie.mrkdns.com [69.93.72.82])
	by svr1.postgresql.org (Postfix) with ESMTP id 278A4D1BC54
	for ;
	Tue, 27 Apr 2004 19:00:05 -0300 (ADT)
Received: from 82-33-1-22.cable.ubr13.newt.blueyonder.co.uk ([82.33.1.22]
	helo=chuckie.co.uk) by obie.mrkdns.com with asmtp (Exim 4.30)
	id 1BIacd-0002my-H7; Tue, 27 Apr 2004 23:00:11 +0100
Message-ID: <408ED7FC.2010705@chuckie.co.uk>
Date: Tue, 27 Apr 2004 23:00:28 +0100
From: Nick Barr 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Vitaly Belman 
Cc: pgsql-performance@postgresql.org
Subject: Re: Simply join in PostrgeSQL takes too long
References: <282909734.20040428002740@012.net.il>
In-Reply-To: <282909734.20040428002740@012.net.il>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-AntiAbuse: This header was added to track abuse,
	please include it with any abuse report
X-AntiAbuse: Primary Hostname - obie.mrkdns.com
X-AntiAbuse: Original Domain - postgresql.org
X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12]
X-AntiAbuse: Sender Address Domain - chuckie.co.uk
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/468
X-Sequence-Number: 6768

Vitaly Belman wrote:
> Hello pgsql-performance,
> 
>   I discussed the whole subject for some time in DevShed and didn't
>   achieve much (as for results). I wonder if any of you guys can help
>   out:
> 
>   http://forums.devshed.com/t136202/s.html
> 

So cutting and pasting:

----- SCHEMA -----
CREATE TABLE bv_bookgenres (
     book_id INT NOT NULL,
     genre_id INT NOT NULL
);
CREATE TABLE bv_genre_children (
     genre_id INT,
     genre_child_id INT
);
-------------------

----- QUERY -----
select DISTINCT
       book_id
from
   bookgenres,
   genre_children
WHERE
      bookgenres.genre_id = genre_children.genre_child_id AND
      genre_children.genre_id = 1
LIMIT 10
-----------------

----- EXPLAIN ANALYZE -----
QUERY PLAN
Limit  (cost=6503.51..6503.70 rows=10 width=4) (actual 
time=703.000..703.000 rows=10 loops=1)
   ->  Unique  (cost=6503.51..6738.20 rows=12210 width=4) (actual 
time=703.000..703.000 rows=10 loops=1)
         ->  Sort  (cost=6503.51..6620.85 rows=46937 width=4) (actual 
time=703.000..703.000 rows=24 loops=1)
               Sort Key: bv_bookgenres.book_id
               ->  Merge Join  (cost=582.45..2861.57 rows=46937 width=4) 
(actual time=46.000..501.000 rows=45082 loops=1)
                     Merge Cond: ("outer".genre_id = "inner".genre_child_id)
                     ->  Index Scan using genre_id on bv_bookgenres 
(cost=0.00..1462.84 rows=45082 width=8) (actual time=0.000..158.000 
rows=45082 loops=1)
                     ->  Sort  (cost=582.45..598.09 rows=6256 width=2) 
(actual time=46.000..77.000 rows=49815 loops=1)
                           Sort Key: bv_genre_children.genre_child_id
                           ->  Index Scan using genre_id2 on 
bv_genre_children  (cost=0.00..187.98 rows=6256 width=2) (actual 
time=0.000..31.000 rows=6379 loops=1)
                                 Index Cond: (genre_id = 1)
Total runtime: 703.000 ms
-------------------------------

----- CONF SETTINGS -----
shared_buffers = 1000		# min 16, at least max_connections*2, 8KB each
sort_mem = 10000
#work_mem = 1024		# min 64, size in KB
#maintenance_work_mem = 16384	# min 1024, size in KB
#max_stack_depth = 2048		# min 100, size in KB
-------------------------

Have you VACUUM ANALYZED recently. If not do that then rerun the EXPLAIN 
ANALYZE.

You might wanna bump shared_buffers. You have 512MB RAM right? You 
probably want to bump shared_buffers to 10000, restart PG then run a 
VACUUM ANALYZE. Then rerun the EXPLAIN ANALYZE.

If that doesnt help try doing a

ALTER TABLE bv_genre_children ALTER COLUMN genre_child_id SET STATISTICS 
100;

followed by a:

VACUUM ANALYZE bv_genre_children;

You might also want to be tweaking the effective_cache_size parameter in 
  postgresql.conf, but I am unsure how this would work on Windows. Does 
Windows have a kernel disk cache anyone?




HTH

Nick






From pgsql-performance-owner@postgresql.org  Tue Apr 27 19:01:55 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C2DDAD1B4DF
	for ;
	Tue, 27 Apr 2004 19:01:53 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 48376-05
	for ;
	Tue, 27 Apr 2004 19:01:47 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9FB2AD1BC54
	for ;
	Tue, 27 Apr 2004 19:01:43 -0300 (ADT)
Received: from [134.22.68.249] (dyn-68-249.tor.dsl.tht.net [134.22.68.249])
	by tht.net (Postfix) with ESMTP
	id 99FD276A7A; Tue, 27 Apr 2004 18:01:50 -0400 (EDT)
Subject: Re: Simply join in PostrgeSQL takes too long
From: Rod Taylor 
To: Vitaly Belman 
Cc: Postgresql Performance 
In-Reply-To: <282909734.20040428002740@012.net.il>
References: <282909734.20040428002740@012.net.il>
Content-Type: text/plain
Message-Id: <1083103293.30065.100.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Tue, 27 Apr 2004 18:01:34 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/469
X-Sequence-Number: 6769

On Tue, 2004-04-27 at 17:27, Vitaly Belman wrote:
> Hello pgsql-performance,
> 
>   I discussed the whole subject for some time in DevShed and didn't
>   achieve much (as for results). I wonder if any of you guys can help
>   out:
> 
>   http://forums.devshed.com/t136202/s.html

You're taking the wrong approach. Rather than using a select query to
ensure that the book_id is distinct, add a constraint to the table so
that is guaranteed.

        CREATE UNIQUE INDEX bv_bookgeneres_unq ON bv_bookgenres(book_id,
        genre_id);
        
Now you can do a simple join (Drop the DISTINCT keyword) and achieve the
same results.

The point is that a book cannot be of a certain genre more than once.

Without the distinct, this should take a matter of a few milliseconds to
execute.



From pgsql-performance-owner@postgresql.org  Tue Apr 27 19:36:58 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F2742D1BB4C
	for ;
	Tue, 27 Apr 2004 19:36:55 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 60874-02
	for ;
	Tue, 27 Apr 2004 19:36:55 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id B8C45D1BB45
	for ;
	Tue, 27 Apr 2004 19:36:52 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4959495; Tue, 27 Apr 2004 15:38:21 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: Vitaly Belman ,
	pgsql-performance@postgresql.org
Subject: Re: Simply join in PostrgeSQL takes too long
Date: Tue, 27 Apr 2004 15:37:09 -0700
User-Agent: KMail/1.5.4
References: <282909734.20040428002740@012.net.il>
In-Reply-To: <282909734.20040428002740@012.net.il>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404271537.09547.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/470
X-Sequence-Number: 6770

Vitaly,

I'm afraid that your helper on DevShed is right; 7.5 for Windows is still in 
development, we've not even *started* to check it for performance yet.  

Since the Merge Join is taking 90% of your query time, I might suggest 
increasing shared_buffers and sort_mem to see if that helps. 

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-admin-owner@postgresql.org  Tue Apr 27 22:25:23 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 17793D1B498; Tue, 27 Apr 2004 22:25:13 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 06476-06; Tue, 27 Apr 2004 22:25:13 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 005F4D1C942; Tue, 27 Apr 2004 22:25:08 -0300 (ADT)
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
	i3S1OxWL006768; Wed, 28 Apr 2004 09:24:59 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <408F0944.5060608@familyhealth.com.au>
Date: Wed, 28 Apr 2004 09:30:44 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Edoardo Ceccarelli 
Cc: pg@fastcrypt.com, "pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>	
	<1083067187.14759.178.camel@localhost.localdomain>	
	<408E75C2.5060703@familyhealth.com.au>
	<1083079114.14759.204.camel@localhost.localdomain>
	<408ED3E2.802@axa.it>
In-Reply-To: <408ED3E2.802@axa.it>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=3.0 tagged_above=0.0 required=5.0
	tests=SUSPICIOUS_RECIPS
X-Spam-Level: ***
X-Archive-Number: 200404/329
X-Sequence-Number: 13302

> I am going to use them as primary key of the table, so I'll surely need 
> them unique :)

Eduoardo, I REALLY suggest you don't use them at all.  You should make a 
primary key like this:

CREATE TABLE blah (
   id SERIAL PRIMARY KEY,
   ...
);

Also note that by default, OIDs are NOT dumped by pg_dump.  You will 
need to add extra switches to your pg_dump backup to ensure that they are.

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 28 01:28:19 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 76797D1B8B6
	for ;
	Wed, 28 Apr 2004 01:28:18 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 65937-01
	for ;
	Wed, 28 Apr 2004 01:28:17 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 43616D1B582
	for ;
	Wed, 28 Apr 2004 01:28:16 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3S4RxpG023033;
	Wed, 28 Apr 2004 00:28:00 -0400 (EDT)
To: brad-pgperf@duttonbros.com
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
In-reply-to: <20040426171103.45600.qmail@uno.mnl.com> 
References: <20040426171103.45600.qmail@uno.mnl.com>
Comments: In-reply-to brad-pgperf@duttonbros.com
	message dated "Mon, 26 Apr 2004 10:11:03 -0700"
Date: Wed, 28 Apr 2004 00:27:59 -0400
Message-ID: <23032.1083126479@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/472
X-Sequence-Number: 6772

brad-pgperf@duttonbros.com writes:
> ... Wouldn't the most efficient plan be to scan the index regardless
> of crm_id because the only columns needed are in the index?

No.  People coming from other databases often have the misconception
that queries can be answered by looking only at an index.  That is never
true in Postgres because row validity info is only stored in the table;
so we must always visit the table entry to make sure the row is still
valid/visible for the current query.

Accordingly, columns added to the index that aren't constrained by the
WHERE clause are not very useful ...

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 28 03:35:30 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 51047D1B455
	for ;
	Wed, 28 Apr 2004 03:35:29 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 03735-04
	for ;
	Wed, 28 Apr 2004 03:35:26 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 6B51AD1B454
	for ;
	Wed, 28 Apr 2004 03:35:24 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 72
	for ; Wed, 28 Apr 2004 07:34:44 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Wed, 28 Apr 2004 07:35:41 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <408F5ECD.29849.CA91084@localhost>
In-reply-to: <23032.1083126479@sss.pgh.pa.us>
References: <20040426171103.45600.qmail@uno.mnl.com> 
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/473
X-Sequence-Number: 6773

I know you will shoot me down, but...

Why is there an entry in the index for a row if the row is not valid? 
Wouldn't it be better for the index entry validity to track the row validity. 
If a particular data value for a query (join, where etc.) can be satisfied 
by the index entry itself this would be a big performance gain.

Cheers,
Gary.

On 28 Apr 2004 at 0:27, Tom Lane wrote:

> brad-pgperf@duttonbros.com writes:
> > ... Wouldn't the most efficient plan be to scan the index regardless
> > of crm_id because the only columns needed are in the index?
> 
> No.  People coming from other databases often have the misconception
> that queries can be answered by looking only at an index.  That is never
> true in Postgres because row validity info is only stored in the table;
> so we must always visit the table entry to make sure the row is still
> valid/visible for the current query.
> 
> Accordingly, columns added to the index that aren't constrained by the
> WHERE clause are not very useful ...
> 
> 			regards, tom lane
> 
> ---------------------------(end of broadcast)---------------------------
> TIP 8: explain analyze is your friend



From pgsql-performance-owner@postgresql.org  Wed Apr 28 03:58:33 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 37846D1B454
	for ;
	Wed, 28 Apr 2004 03:58:32 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 13158-01
	for ;
	Wed, 28 Apr 2004 03:58:30 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP id B051FD1B453
	for ;
	Wed, 28 Apr 2004 03:58:27 -0300 (ADT)
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
	i3S6wLWL037650; Wed, 28 Apr 2004 14:58:22 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <408F576E.1080504@familyhealth.com.au>
Date: Wed, 28 Apr 2004 15:04:14 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Gary Doades 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question
References: <20040426171103.45600.qmail@uno.mnl.com>
	<408F5ECD.29849.CA91084@localhost>
In-Reply-To: <408F5ECD.29849.CA91084@localhost>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/474
X-Sequence-Number: 6774

> Why is there an entry in the index for a row if the row is not valid? 
> Wouldn't it be better for the index entry validity to track the row validity. 
> If a particular data value for a query (join, where etc.) can be satisfied 
> by the index entry itself this would be a big performance gain.

For SELECTs, yes - but for INSERT, UPDATE and DELETE it would be a big 
performance loss.

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 28 04:07:45 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A2D41D1B8E4
	for ;
	Wed, 28 Apr 2004 04:07:43 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 07145-10
	for ;
	Wed, 28 Apr 2004 04:07:44 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id B4D51D1B8D1
	for ;
	Wed, 28 Apr 2004 04:07:41 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 80
	for ; Wed, 28 Apr 2004 08:07:05 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Wed, 28 Apr 2004 08:08:03 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question
Message-ID: <408F6663.22977.CC6AF7D@localhost>
In-reply-to: <408F576E.1080504@familyhealth.com.au>
References: <408F5ECD.29849.CA91084@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/475
X-Sequence-Number: 6775

I can understand the performance loss on non-selects for keeping the 
index validity state tracking the row validity, but would that outweigh the 
performance gains on selects? Depends on your mix of selects to non 
selects I guess, but other database systems seem to imply that keeping 
the index on track is worth it overall.

Cheers,
Gary.

On 28 Apr 2004 at 15:04, Christopher Kings-Lynne wrote:

> > Why is there an entry in the index for a row if the row is not valid? 
> > Wouldn't it be better for the index entry validity to track the row validity. 
> > If a particular data value for a query (join, where etc.) can be satisfied 
> > by the index entry itself this would be a big performance gain.
> 
> For SELECTs, yes - but for INSERT, UPDATE and DELETE it would be a big 
> performance loss.
> 
> Chris
> 



From pgsql-performance-owner@postgresql.org  Wed Apr 28 04:18:32 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id BBB88D1B454
	for ;
	Wed, 28 Apr 2004 04:17:59 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 20161-07
	for ;
	Wed, 28 Apr 2004 04:17:59 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP id C8722D1B4E0
	for ;
	Wed, 28 Apr 2004 04:17:56 -0300 (ADT)
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
	i3S7HrWL039443; Wed, 28 Apr 2004 15:17:53 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <408F5C04.7020101@familyhealth.com.au>
Date: Wed, 28 Apr 2004 15:23:48 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Gary Doades 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question
References: <408F5ECD.29849.CA91084@localhost>
	<408F6663.22977.CC6AF7D@localhost>
In-Reply-To: <408F6663.22977.CC6AF7D@localhost>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/476
X-Sequence-Number: 6776

> I can understand the performance loss on non-selects for keeping the 
> index validity state tracking the row validity, but would that outweigh the 
> performance gains on selects? Depends on your mix of selects to non 
> selects I guess, but other database systems seem to imply that keeping 
> the index on track is worth it overall.

Yes, some sort of flag on index creation would be sweet :)

Chris


From pgsql-performance-owner@postgresql.org  Wed Apr 28 04:32:33 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1D7F1D1B445
	for ;
	Wed, 28 Apr 2004 04:32:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 22187-07
	for ;
	Wed, 28 Apr 2004 04:32:29 -0300 (ADT)
Received: from email05.aon.at (WARSL402PIP4.highway.telekom.at [195.3.96.79])
	by svr1.postgresql.org (Postfix) with SMTP id E2126D1B4E0
	for ;
	Wed, 28 Apr 2004 04:32:26 -0300 (ADT)
Received: (qmail 556408 invoked from network); 28 Apr 2004 07:32:27 -0000
Received: from m150p026.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.186]) (envelope-sender )
	by qmail5rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 28 Apr 2004 07:32:27 -0000
From: Manfred Koizar 
To: "Gary Doades" 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
Date: Wed, 28 Apr 2004 09:34:59 +0200
Message-ID: 
References: <20040426171103.45600.qmail@uno.mnl.com>
	<23032.1083126479@sss.pgh.pa.us> <408F5ECD.29849.CA91084@localhost>
In-Reply-To: <408F5ECD.29849.CA91084@localhost>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/477
X-Sequence-Number: 6777

On Wed, 28 Apr 2004 07:35:41 +0100, "Gary Doades" 
wrote:
>Why is there an entry in the index for a row if the row is not valid? 

Because whether a row is seen as valid or not lies in the eye of the
transaction looking at it.  Full visibility information is stored in the
heap tuple header.  The developers' consensus is that this overhead
should not be in every index tuple.

Servus
 Manfred

From pgsql-admin-owner@postgresql.org  Wed Apr 28 05:13:59 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id ABBFCD1B8D1; Wed, 28 Apr 2004 05:13:51 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 40407-07; Wed, 28 Apr 2004 05:13:49 -0300 (ADT)
Received: from FE-mail04.sfg.albacom.net (FE-mail04.albacom.net
	[213.217.149.84]) by svr1.postgresql.org (Postfix) with ESMTP
	id CC622D1B8EF; Wed, 28 Apr 2004 05:13:47 -0300 (ADT)
Received: from axa.it (217.221.80.72) by FE-mail04.sfg.albacom.net (7.0.009)
	id 408E270C0002A496; Wed, 28 Apr 2004 10:13:12 +0200
Message-ID: <408F679A.9030608@axa.it>
Date: Wed, 28 Apr 2004 10:13:14 +0200
From: Edoardo Ceccarelli 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: it, en, en-us
MIME-Version: 1.0
To: Christopher Kings-Lynne 
Cc: pg@fastcrypt.com, "pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>	
	<1083067187.14759.178.camel@localhost.localdomain>	
	<408E75C2.5060703@familyhealth.com.au>
	<1083079114.14759.204.camel@localhost.localdomain>
	<408ED3E2.802@axa.it> <408F0944.5060608@familyhealth.com.au>
In-Reply-To: <408F0944.5060608@familyhealth.com.au>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=3.0 tagged_above=0.0 required=5.0
	tests=SUSPICIOUS_RECIPS
X-Spam-Level: ***
X-Archive-Number: 200404/330
X-Sequence-Number: 13303

do you mean that, declaring an index serial, I'd never have to deal with 
incrementing its primary key? good to know!
anyway  in this particular situation I don't need such accurate 
behaviour: this table is filled up with a lot of data twice per week and 
it's used only to answer queries.
I could drop it whenever I want :)

Thanks again,
eddy

Christopher Kings-Lynne ha scritto:

>> I am going to use them as primary key of the table, so I'll surely 
>> need them unique :)
>
>
> Eduoardo, I REALLY suggest you don't use them at all.  You should make 
> a primary key like this:
>
> CREATE TABLE blah (
>   id SERIAL PRIMARY KEY,
>   ...
> );
>
> Also note that by default, OIDs are NOT dumped by pg_dump.  You will 
> need to add extra switches to your pg_dump backup to ensure that they 
> are.
>
> Chris
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
>
>
>

From pgsql-performance-owner@postgresql.org  Wed Apr 28 05:12:38 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9670BD1B432
	for ;
	Wed, 28 Apr 2004 05:12:37 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 37012-07
	for ;
	Wed, 28 Apr 2004 05:12:36 -0300 (ADT)
Received: from 192.168.0.11 (fw.i-trade.hu [212.92.1.62])
	by svr1.postgresql.org (Postfix) with SMTP id 96F6DD1B8E4
	for ;
	Wed, 28 Apr 2004 05:12:33 -0300 (ADT)
Received: (qmail 27852 invoked from network); 28 Apr 2004 07:52:31 -0000
Received: from unknown (HELO freemail.hu) (212.92.7.130)
	by 0 with SMTP; 28 Apr 2004 07:52:31 -0000
Message-ID: <408F6817.1060708@freemail.hu>
Date: Wed, 28 Apr 2004 10:15:19 +0200
From: =?ISO-8859-2?Q?Suller_Andr=E1s?= 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: hu, en-us
MIME-Version: 1.0
To: Silke Trissl 
Cc: pgsql-performance@postgresql.org
Subject: Re: Join problem
References: <408E9934.8060509@informatik.hu-berlin.de>
In-Reply-To: <408E9934.8060509@informatik.hu-berlin.de>
Content-Type: text/plain; charset=ISO-8859-2; format=flowed
Content-Transfer-Encoding: 8bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/478
X-Sequence-Number: 6778

these two queries are not equal. Query1 returns 6 rows, query2 returns 0 
rows, because '~*' and '=' operators are not same. BTW when you use '=', 
it could use index on "item.description".
On query1, "Seq Scan on item" estimates 1 row, on query2 it estimates 
733 rows. IMHO that's why query1 uses nested loop, query2 uses hash join.

bye,
Suller Andras

Silke Trissl �rta:

> Hi,
>
> Query 1:
>   QUERY PLAN
> ------------------------------------------------------------------------------------------------------------------ 
>
>  Nested Loop  (cost=0.00..28836.75 rows=1 width=4) (actual 
> time=65350.780..452130.702 rows=6 loops=1)
>    Join Filter: ("inner".order_id = "outer".order_id)
>    ->  Seq Scan on item  (cost=0.00..28814.24 rows=1 width=8) (actual 
> time=33.180..1365.190 rows=716 loops=1)
>          Filter: (description ~* 'CD'::text)
>    ->  Seq Scan on orders  (cost=0.00..22.50 rows=1 width=4) (actual 
> time=21.644..629.500 rows=18 loops=716)
>          Filter: ((order_name)::text ~* 'Smith'::text)
>  Total runtime: 452130.782 ms
> ########################################################################### 
>
>
> Query 2:
>   QUERY PLAN
> ------------------------------------------------------------------------------------------------------------------ 
>
>   Hash Join  (cost=22.50..28840.44 rows=4 width=4) (actual 
> time=1187.798..1187.798 rows=0 loops=1)
>    Hash Cond: ("outer".order_id = "inner".order_id)
>    ->  Seq Scan on item  (cost=0.00..28814.24 rows=733 width=8) 
> (actual time=542.737..542.737 rows=0 loops=1)
>          Filter: (description = 'CD'::text)
>    ->  Hash  (cost=22.50..22.50 rows=1 width=4) (actual 
> time=645.042..645.042 rows=0 loops=1)
>          ->  Seq Scan on orders  (cost=0.00..22.50 rows=1 width=4) 
> (actual time=22.373..644.996 rows=18 loops=1)
>                Filter: ((order_name)::text ~* 'Smith'::text)
>  Total runtime: 1187.865 ms
> ############################################################################ 
>
>
>
> Query 1 with  'set enable_nestloop to false'
>
>    QUERY PLAN
> ----------------------------------------------------------------------------------------------------------------------- 
>
>  Hash Join  (cost=22.50..28836.75 rows=1 width=4) (actual 
> time=1068.593..2003.330 rows=6 loops=1)
>    Hash Cond: ("outer".item_id = "inner".item_id)
>    ->  Seq Scan on item  (cost=0.00..28814.24 rows=1 width=8) (actual 
> time=33.347..1357.073 rows=716 loops=1)
>          Filter: (description ~* 'CD'::text)
>    ->  Hash  (cost=22.50..22.50 rows=1 width=4) (actual 
> time=645.287..645.287 rows=0 loops=1)
>          ->  Seq Scan on orders  (cost=0.00..22.50 rows=1 width=4) 
> (actual time=22.212..645.239 rows=18 loops=1)
>                Filter: ((order_name)::text ~* 'CD'::text)
>  Total runtime: 2003.409 ms




From pgsql-performance-owner@postgresql.org  Wed Apr 28 05:24:04 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6832FD1B560
	for ;
	Wed, 28 Apr 2004 05:22:11 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36576-08
	for ;
	Wed, 28 Apr 2004 05:22:11 -0300 (ADT)
Received: from email07.aon.at (WARSL402PIP8.highway.telekom.at [195.3.96.97])
	by svr1.postgresql.org (Postfix) with SMTP id A7EC9D1B4E0
	for ;
	Wed, 28 Apr 2004 05:22:08 -0300 (ADT)
Received: (qmail 633540 invoked from network); 28 Apr 2004 08:22:09 -0000
Received: from m150p026.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.186]) (envelope-sender )
	by 172.18.5.236 (qmail-ldap-1.03) with SMTP
	for ; 28 Apr 2004 08:22:09 -0000
From: Manfred Koizar 
To: Rod Taylor 
Cc: Vitaly Belman ,
	Postgresql Performance 
Subject: Re: Simply join in PostrgeSQL takes too long
Date: Wed, 28 Apr 2004 10:24:41 +0200
Message-ID: 
References: <282909734.20040428002740@012.net.il>
	<1083103293.30065.100.camel@jester>
In-Reply-To: <1083103293.30065.100.camel@jester>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/480
X-Sequence-Number: 6780

On Tue, 27 Apr 2004 18:01:34 -0400, Rod Taylor  wrote:
>On Tue, 2004-04-27 at 17:27, Vitaly Belman wrote:
>> Hello pgsql-performance,
>> 
>>   I discussed the whole subject for some time in DevShed and didn't
>>   achieve much (as for results). I wonder if any of you guys can help
>>   out:
>> 
>>   http://forums.devshed.com/t136202/s.html

>The point is that a book cannot be of a certain genre more than once.

Rod, he has a hierarchy of genres.  Genre 1 has 6379 child genres and a
book can be in more than one of these.

Vitaly, though LIMIT makes this look like a small query, DISTINCT
requires the whole result set to be retrieved.  0.7 seconds doesn't look
so bad for several thousand rows.  Did you try with other genre_ids?

Maybe a merge join is not the best choice.  Set enable_mergejoin to
false and see whether you get a (hopefully faster) hash join, assuming
that sort_mem is large enough to keep the hash table in memory.

If you send me your table contents I'll try it on Linux.

Servus
 Manfred

From pgsql-admin-owner@postgresql.org  Wed Apr 28 05:27:27 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id B3898D1B445; Wed, 28 Apr 2004 05:19:44 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 43428-01; Wed, 28 Apr 2004 05:19:43 -0300 (ADT)
Received: from houston.familyhealth.com.au (fhnet.arach.net.au
 [203.22.197.21])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 59FEBD1B478; Wed, 28 Apr 2004 05:19:40 -0300 (ADT)
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
	i3S8JTWL045110; Wed, 28 Apr 2004 16:19:29 +0800 (WST)
	(envelope-from chriskl@familyhealth.com.au)
Message-ID: <408F6A76.9070908@familyhealth.com.au>
Date: Wed, 28 Apr 2004 16:25:26 +0800
From: Christopher Kings-Lynne 
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Edoardo Ceccarelli 
Cc: pg@fastcrypt.com, "pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>	
	<1083067187.14759.178.camel@localhost.localdomain>	
	<408E75C2.5060703@familyhealth.com.au>
	<1083079114.14759.204.camel@localhost.localdomain>
	<408ED3E2.802@axa.it> <408F0944.5060608@familyhealth.com.au>
	<408F679A.9030608@axa.it>
In-Reply-To: <408F679A.9030608@axa.it>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=3.0 tagged_above=0.0 required=5.0
	tests=SUSPICIOUS_RECIPS
X-Spam-Level: ***
X-Archive-Number: 200404/331
X-Sequence-Number: 13304

> do you mean that, declaring an index serial, I'd never have to deal with 
> incrementing its primary key? good to know!

Yep. You can use 'DEFAULT' as the value, eg:

INSERT INTO blah (DEFAULT, ...);

> anyway  in this particular situation I don't need such accurate 
> behaviour: this table is filled up with a lot of data twice per week and 
> it's used only to answer queries.
> I could drop it whenever I want :)

Sure.

Chris


From pgsql-admin-owner@postgresql.org  Wed Apr 28 05:38:59 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B55F5D1B8BD
	for ;
	Wed, 28 Apr 2004 05:28:08 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 43467-06
	for ;
	Wed, 28 Apr 2004 05:28:08 -0300 (ADT)
Received: from wolff.to (wolff.to [66.93.249.74])
	by svr1.postgresql.org (Postfix) with SMTP id 79120D1B8B1
	for ; Wed, 28 Apr 2004 05:28:06 -0300 (ADT)
Received: (qmail 8577 invoked by uid 500); 28 Apr 2004 08:32:23 -0000
Date: Wed, 28 Apr 2004 03:32:23 -0500
From: Bruno Wolff III 
To: Edoardo Ceccarelli 
Cc: Christopher Kings-Lynne , pg@fastcrypt.com,
	"pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
Subject: Re: [JDBC] [PERFORM] is a good practice to create an index on the
Message-ID: <20040428083223.GA8384@wolff.to>
Mail-Followup-To: Edoardo Ceccarelli ,
	Christopher Kings-Lynne ,
	pg@fastcrypt.com,
	"pgsql-jdbc@postgresql.org" ,
	pgsql-performance@postgresql.org, pgsql-admin@postgresql.org
References: <408D3AE8.5020006@axa.it> <408DC9E1.8020704@familyhealth.com.au>
	<1083067187.14759.178.camel@localhost.localdomain>
	<408E75C2.5060703@familyhealth.com.au>
	<1083079114.14759.204.camel@localhost.localdomain>
	<408ED3E2.802@axa.it> <408F0944.5060608@familyhealth.com.au>
	<408F679A.9030608@axa.it>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <408F679A.9030608@axa.it>
User-Agent: Mutt/1.5.6i
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/332
X-Sequence-Number: 13305

On Wed, Apr 28, 2004 at 10:13:14 +0200,
  Edoardo Ceccarelli  wrote:
> do you mean that, declaring an index serial, I'd never have to deal with 
> incrementing its primary key? good to know!

That isn't what is happening. Serial is a special type. It is int plus
a default rule linked to a sequence. No index is created by default
for the serial type. Declaring a column as a primary key will however
create a unique index on that column.

Also note that you should only assume that the serial values are unique.
(This assumes that you don't use setval and that you don't roll a sequence
over.) Within a single session you can assume the sequence values will
be monotonicly increasing. The values that end up in your table can have
gaps. Typically this happens when a transaction rolls back after obtaining
a new value from a sequence. It can also happen if you grab sequence
values in larger blocks (which might be more efficient if a session normally
acquires mulitple values from a particular sequence) than the default 1.

> anyway  in this particular situation I don't need such accurate 
> behaviour: this table is filled up with a lot of data twice per week and 
> it's used only to answer queries.
> I could drop it whenever I want :)

You really don't want to use oids.

From pgsql-performance-owner@postgresql.org  Wed Apr 28 09:24:01 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id F0034D1B8A5
	for ;
	Wed, 28 Apr 2004 09:23:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 29669-02
	for ;
	Wed, 28 Apr 2004 09:23:56 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 22076D1B58D
	for ;
	Wed, 28 Apr 2004 09:23:52 -0300 (ADT)
Received: from [134.22.68.249] (dyn-68-249.tor.dsl.tht.net [134.22.68.249])
	by tht.net (Postfix) with ESMTP
	id B276B76A96; Wed, 28 Apr 2004 08:23:55 -0400 (EDT)
Subject: Re: Simply join in PostrgeSQL takes too long
From: Rod Taylor 
To: Manfred Koizar 
Cc: Vitaly Belman ,
	Postgresql Performance 
In-Reply-To: 
References: <282909734.20040428002740@012.net.il>
	<1083103293.30065.100.camel@jester>
	
Content-Type: text/plain
Message-Id: <1083155014.30065.189.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Wed, 28 Apr 2004 08:23:35 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/483
X-Sequence-Number: 6783

> Rod, he has a hierarchy of genres.  Genre 1 has 6379 child genres and a
> book can be in more than one of these.

     bookgenres.genre_id = genre_children.genre_child_id AND
     genre_children.genre_id = 1

I see, sorry. I didn't notice the genre_child_id in the where clause.
First glance had them all as genre_id.

When I run into this I usually create a 3rd table managed by triggers
that would relate the book to all genre entries. Insert takes a little
longer, but the selects can still be very quick.

The below plpgsql forces the kind of algorithm we wish the planner could
choose. It should be fairly quick irregardless of dataset.


CREATE OR REPLACE FUNCTION book_results(numeric) RETURNS SETOF numeric
AS
'
DECLARE
  v_genre ALIAS FOR $1;
  v_limit integer = 10;
  t_rows RECORD;
  v_transmitted integer = 0;

  v_transmitted_values numeric[] = ARRAY[1];

BEGIN
  FOR t_rows IN SELECT book_id
                  FROM bv_bookgenres AS b
                  JOIN bv_genre_children AS g ON (b.genre_id =
g.genre_child_id)
                 WHERE g.genre_id = v_genre
                  LOOP

    -- If this is a new value, transmit it to the end user
    IF NOT t_rows.book_id = ANY(v_transmitted_values) THEN
      v_transmitted_values := array_append(v_transmitted_values,
t_rows.book_id);
      v_transmitted := v_transmitted + 1;
      RETURN NEXT t_rows.book_id;
    END IF;

    EXIT WHEN v_transmitted >= v_limit;
  END LOOP;

  RETURN;
END;
' LANGUAGE plpgsql;

EXPLAIN ANALYZE SELECT * FROM book_results(1);
SELECT * FROM book_results(1);


From pgsql-performance-owner@postgresql.org  Wed Apr 28 09:52:24 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id BB399D1B519
	for ;
	Wed, 28 Apr 2004 09:52:21 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 39914-05
	for ;
	Wed, 28 Apr 2004 09:52:23 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 8D9D5D1B503
	for ;
	Wed, 28 Apr 2004 09:52:19 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3SCqLDj027628;
	Wed, 28 Apr 2004 08:52:21 -0400 (EDT)
To: Silke Trissl 
Cc: pgsql-performance@postgresql.org
Subject: Re: Join problem 
In-reply-to: <408E9934.8060509@informatik.hu-berlin.de> 
References: <408E9934.8060509@informatik.hu-berlin.de>
Comments: In-reply-to Silke Trissl 
	message dated "Tue, 27 Apr 2004 19:32:36 +0200"
Date: Wed, 28 Apr 2004 08:52:20 -0400
Message-ID: <27627.1083156740@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/484
X-Sequence-Number: 6784

Silke Trissl  writes:
> I found out, that the change of the operator from '~*' to '=' for the 
> item.description brought a great boost in performance (425 secs to 1 
> sec!), but not in cost (Query plans at the end).

The main problem seems to be bad estimation of the number of rows
extracted from the item table.  Have you ANALYZEd that table lately?

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed Apr 28 10:05:18 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 21FF7D1B8AA
	for ;
	Wed, 28 Apr 2004 10:05:06 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 42082-07
	for ;
	Wed, 28 Apr 2004 10:05:08 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 1EA39D1B893
	for ;
	Wed, 28 Apr 2004 10:05:04 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3SD54l0027807;
	Wed, 28 Apr 2004 09:05:04 -0400 (EDT)
To: Manfred Koizar 
Cc: "Gary Doades" ,
	pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
In-reply-to:  
References: <20040426171103.45600.qmail@uno.mnl.com>
	<23032.1083126479@sss.pgh.pa.us> <408F5ECD.29849.CA91084@localhost>
	
Comments: In-reply-to Manfred Koizar 
	message dated "Wed, 28 Apr 2004 09:34:59 +0200"
Date: Wed, 28 Apr 2004 09:05:04 -0400
Message-ID: <27806.1083157504@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/485
X-Sequence-Number: 6785

Manfred Koizar  writes:
> On Wed, 28 Apr 2004 07:35:41 +0100, "Gary Doades" 
> wrote:
>> Why is there an entry in the index for a row if the row is not valid? 

> Because whether a row is seen as valid or not lies in the eye of the
> transaction looking at it.  Full visibility information is stored in the
> heap tuple header.  The developers' consensus is that this overhead
> should not be in every index tuple.

Storing that information would at least double the overhead space used
for each index tuple.  The resulting index bloat would significantly
slow index operations by requiring more I/O.  So it's far from clear
that this would be a win, even for those who care only about select
speed.

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Wed May  5 13:52:35 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6FB57D1B50F
	for ;
	Wed, 28 Apr 2004 10:29:20 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 51147-10
	for ;
	Wed, 28 Apr 2004 10:29:21 -0300 (ADT)
Received: from mail.socialserve.com (office.socialserve.com [208.60.89.34])
	by svr1.postgresql.org (Postfix) with ESMTP id 07786D1B892
	for ;
	Wed, 28 Apr 2004 10:29:18 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by mail.socialserve.com (Postfix) with ESMTP
	id A629F975F; Wed, 28 Apr 2004 09:29:23 -0400 (EDT)
Received: from mail.socialserve.com ([127.0.0.1])
	by localhost (ms2 [127.0.0.1]) (amavisd-new, port 10024) with ESMTP
	id 04220-08; Wed, 28 Apr 2004 09:29:19 -0400 (EDT)
Received: from [10.1.1.94] (dynamic-94.socialserve.com [10.1.1.94])
	(using TLSv1 with cipher RC4-SHA (128/128 bits))
	(No client certificate requested)
	by mail.socialserve.com (Postfix) with ESMTP
	id B8479975E; Wed, 28 Apr 2004 09:29:19 -0400 (EDT)
Mime-Version: 1.0 (Apple Message framework v613)
Content-Type: text/plain; charset=US-ASCII; format=flowed
Message-Id: <0B7369F9-9918-11D8-A107-000A9566A412@socialserve.com>
Content-Transfer-Encoding: 7bit
Cc: pgsql-performance@postgresql.org
From: James Robinson 
Subject: History of oids in postgres?
Date: Wed, 28 Apr 2004 09:29:15 -0400
To: bruno@wolff.to
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at socialserve.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200405/33
X-Sequence-Number: 6864

Bruno et al,

	Any self-repsecting lurker would know that oids as row identifiers are 
depreciated in postgres. Can anyone provide a brief history regarding 
the reasoning behind using them as row identifiers in the first place? 
I see a discussion of their use as various primary keys in he system 
catalog in the oid-datatype doc page, but not regarding their history 
as 'user-space' row ids.

Thanks,
James

----
James Robinson
Socialserve.com


From pgsql-performance-owner@postgresql.org  Wed Apr 28 12:41:27 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id B6566D1BB45
	for ;
	Wed, 28 Apr 2004 12:41:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 10653-07
	for ;
	Wed, 28 Apr 2004 12:41:24 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id 80FABD1B91B
	for ;
	Wed, 28 Apr 2004 12:41:23 -0300 (ADT)
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 i3SFeBBh025640;
	Wed, 28 Apr 2004 09:40:11 -0600 (MDT)
Date: Wed, 28 Apr 2004 09:40:08 -0600 (MDT)
From: "scott.marlowe" 
To: Stephan Szabo 
Cc: , 
Subject: Re: index usage
In-Reply-To: <20040426115215.R11044@megazone.bigpanda.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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/486
X-Sequence-Number: 6786

On Mon, 26 Apr 2004, Stephan Szabo wrote:

> 
> On Fri, 23 Apr 2004 brad-pgperf@duttonbros.com wrote:
> 
> > I have a query which I think should be using an index all of the time but
> > postgres only uses the index part of the time.  The index
> > (ticket_crm_map_crm_id_suppid) has the where clause column (crm_id) listed
> > first followed by the selected column (support_person_id).  Wouldn't the
> > most efficient plan be to scan the index each time because the only columns
> > needed are in the index?  Below is the table, 2 queries showing the
> 
> Not necessarily.  The rows in the actual file still need to be checked to
> see if they're visible to the select and if it's expected that the entire
> file (or a reasonable % of the pages anyway) will need to be loaded using
> the index isn't necessarily a win.

While those of us familiar with PostgreSQL are well aware of the fact that 
indexes can't be used directly to garner information, but only as a lookup 
to a tuple in the table, it seems this misconception is quite common among 
those coming to postgreSQL from other databases.

Is there any information that directly reflects this issue in the docs?  
There are tons of hints that it works this way in how they're written, but 
nothing that just comes out and says that with pgsql's mvcc 
implementation, an index scan still has to hit the pages that contain the 
tuples, so often in pgsql a seq scan is a win where in other databases and 
index scan would have been a win?

If not, where would I add it if I were going to write something up for the 
docs?  Just wondering...


From pgsql-performance-owner@postgresql.org  Wed Apr 28 13:42:08 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id EE9B7D1B4D7
	for ;
	Wed, 28 Apr 2004 13:42:06 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 25269-10
	for ;
	Wed, 28 Apr 2004 13:42:05 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id AE70CD1B4A3
	for ;
	Wed, 28 Apr 2004 13:42:04 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3SGfeNQ000425;
	Wed, 28 Apr 2004 12:41:41 -0400 (EDT)
To: "scott.marlowe" 
Cc: Stephan Szabo ,
	brad-pgperf@duttonbros.com, pgsql-performance@postgresql.org
Subject: Re: index usage 
In-reply-to:  
References: 
Comments: In-reply-to "scott.marlowe" 
	message dated "Wed, 28 Apr 2004 09:40:08 -0600"
Date: Wed, 28 Apr 2004 12:41:40 -0400
Message-ID: <424.1083170500@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/487
X-Sequence-Number: 6787

"scott.marlowe"  writes:
> There are tons of hints that it works this way in how they're written, but 
> nothing that just comes out and says that with pgsql's mvcc 
> implementation, an index scan still has to hit the pages that contain the 
> tuples, so often in pgsql a seq scan is a win where in other databases and 
> index scan would have been a win?

> If not, where would I add it if I were going to write something up for the 
> docs?  Just wondering...

AFAIR the only place in the docs that mentions seqscan or indexscan at
all is the discussion of EXPLAIN in "Performance Tips".  Perhaps a
suitably-enlarged version of that section could cover this.

			regards, tom lane

From pgsql-admin-owner@postgresql.org  Wed Apr 28 15:12:26 2004
X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 4041ED1B8EE; Wed, 28 Apr 2004 15:12:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 65938-10; Wed, 28 Apr 2004 15:12:24 -0300 (ADT)
Received: from mail01.stbernard.com (mail01.stbernard.com [199.245.188.4])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 70446D1B8E3; Wed, 28 Apr 2004 15:12:22 -0300 (ADT)
X-MimeOLE: Produced By Microsoft Exchange V6.0.6375.0
Content-Class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Subject: 7.4.2 out of memory
Date: Wed, 28 Apr 2004 11:12:17 -0700
Message-ID: 
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [ADMIN] [JDBC] [PERFORM] is a good practice to create an index
	on the
Thread-Index: AcQs/Igo8CGZODf/T/GyXkvW/iN6swATt3PQ
From: "Jie Liang" 
To: 
Cc: , 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/335
X-Sequence-Number: 13308

All,
After I upgraded postgres from 7.3.4 to 7.4.2, one of my program got follow=
ing error:
DRROR:	out of memory
DETAIL:	Fail on request of size 92.

any idea??
does memory management have big difference between 7.3.4 and 7.4.2???
this program using a chunk of share memory and a lot of temp tables.


Thanks.



Jie Liang

From pgsql-performance-owner@postgresql.org  Wed Apr 28 16:57:44 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 42940D1B8E1
	for ;
	Wed, 28 Apr 2004 16:57:42 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 44732-07
	for ;
	Wed, 28 Apr 2004 16:57:40 -0300 (ADT)
Received: from mail1.ihs.com (mail1.ihs.com [170.207.70.222])
	by svr1.postgresql.org (Postfix) with ESMTP id F2034D1B8D2
	for ;
	Wed, 28 Apr 2004 16:57:37 -0300 (ADT)
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 i3SJvOBh024619;
	Wed, 28 Apr 2004 13:57:25 -0600 (MDT)
Date: Wed, 28 Apr 2004 13:57:21 -0600 (MDT)
From: "scott.marlowe" 
To: Jie Liang 
Cc: 
Subject: Re: [ADMIN] 7.4.2 out of memory
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-IHS-MailScanner-Envelope-Sender: scott.marlowe@ihs.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/488
X-Sequence-Number: 6788

On Wed, 28 Apr 2004, Jie Liang wrote:

> All,
> After I upgraded postgres from 7.3.4 to 7.4.2, one of my program got following error:
> DRROR:	out of memory
> DETAIL:	Fail on request of size 92.
> 
> any idea??
> does memory management have big difference between 7.3.4 and 7.4.2???
> this program using a chunk of share memory and a lot of temp tables.

More than likely this is a hash aggregate problem (or can they spill to 
disk in 7.4.2 yet?  I don't think they can, but maybe we should ask Tom.

Try setting this before running the query and see what happens:

set enable_hashagg = false;




From pgsql-performance-owner@postgresql.org  Wed May  5 13:54:21 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 2C1BED1DF65
	for ;
	Wed, 28 Apr 2004 17:41:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 61532-08
	for ;
	Wed, 28 Apr 2004 17:41:24 -0300 (ADT)
Received: from mail01.stbernard.com (mail01.stbernard.com [199.245.188.4])
	by svr1.postgresql.org (Postfix) with ESMTP id 028B5D1BC54
	for ;
	Wed, 28 Apr 2004 17:41:22 -0300 (ADT)
X-MimeOLE: Produced By Microsoft Exchange V6.0.6375.0
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: [ADMIN] 7.4.2 out of memory
Date: Wed, 28 Apr 2004 13:41:21 -0700
Message-ID: 
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [ADMIN] 7.4.2 out of memory
Thread-Index: AcQtWxiGD9sHnFvTSvup5/T5vwgEyAABdS4g
From: "Jie Liang" 
To: "scott.marlowe" 
Cc: 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200405/34
X-Sequence-Number: 6865

Sccot,

Thank you very much, I think taht you are right about this.
I tested a single query, there is no problem. I'll do a full test with my p=
rogram.

Jie Liang

-----Original Message-----
From: scott.marlowe [mailto:scott.marlowe@ihs.com]
Sent: Wednesday, April 28, 2004 12:57 PM
To: Jie Liang
Cc: pgsql-performance@postgresql.org
Subject: Re: [ADMIN] 7.4.2 out of memory


On Wed, 28 Apr 2004, Jie Liang wrote:

> All,
> After I upgraded postgres from 7.3.4 to 7.4.2, one of my program got foll=
owing error:
> DRROR:	out of memory
> DETAIL:	Fail on request of size 92.
>=20
> any idea??
> does memory management have big difference between 7.3.4 and 7.4.2???
> this program using a chunk of share memory and a lot of temp tables.

More than likely this is a hash aggregate problem (or can they spill to=20
disk in 7.4.2 yet?  I don't think they can, but maybe we should ask Tom.

Try setting this before running the query and see what happens:

set enable_hashagg =3D false;




From pgsql-performance-owner@postgresql.org  Wed May  5 13:59:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 27EBCD1B8BD; Wed, 28 Apr 2004 19:02:04 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 91649-09; Wed, 28 Apr 2004 19:02:04 -0300 (ADT)
Received: from mail01.stbernard.com (mail01.stbernard.com [199.245.188.4])
	by svr1.postgresql.org (Postfix) with ESMTP
	id 66F42D1B8B5; Wed, 28 Apr 2004 19:02:01 -0300 (ADT)
X-MimeOLE: Produced By Microsoft Exchange V6.0.6375.0
Content-Class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Subject: LIKE and INDEX
Date: Wed, 28 Apr 2004 15:02:04 -0700
Message-ID: 
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [ADMIN] 7.4.2 out of memory
Thread-Index: AcQtWxiGD9sHnFvTSvup5/T5vwgEyAAD7HNA
From: "Jie Liang" 
To: 
Cc: 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200405/37
X-Sequence-Number: 6868

All,
This is old topic, when I use:
select url from urlinfo where url like 'http://www.lycos.de%';
it uses the index, good!

but if I use:
select url from urlinfo where url like 'http://%.lycos.de';
it won't use index at all, NOT good!
is there any way I can force secon query use index???

Thanks.

Jie Liang

                                           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
---------------------------------------------------------------------------=
---------------------
 Index Scan using urlinfo_ukey on urlinfo  (cost=3D0.00..6.01 rows=3D1 widt=
h=3D33)
   Index Cond: ((url >=3D 'http://www.lycos.de/'::text) AND (url < 'http://=
www.lycos.de0'::text))
   Filter: (url ~ '^http://www\\.lycos\\.de/.*$'::text)
(3 rows)

                         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
-------------------------------------------------------------
 Seq Scan on urlinfo  (cost=3D0.00..100440.48 rows=3D4 width=3D33)
   Filter: (url ~ '^http://.*\\.lycos\\.de$'::text)
(2 rows)

From pgsql-performance-owner@postgresql.org  Wed Apr 28 23:41:52 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 76E21D1DF5C
	for ;
	Wed, 28 Apr 2004 23:41:48 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 36471-09
	for ;
	Wed, 28 Apr 2004 23:41:27 -0300 (ADT)
Received: from srvr3.iniquinet.com (srvr2.iniquinet.com [64.240.87.12])
	by svr1.postgresql.org (Postfix) with SMTP id 4FC30D1EF8B
	for ;
	Wed, 28 Apr 2004 21:59:28 -0300 (ADT)
Received: (qmail 28632 invoked by uid 104); 29 Apr 2004 00:59:31 -0000
Received: from Robert_Creager@LogicalChaos.org by srvr3.iniquinet.com by uid
	101 with qmail-scanner-1.15
	(clamscan: 0.65. spamassassin: 2.55.  Clear:SA:0(-6.5/6.0):.
	Processed in 32.720762 secs); 29 Apr 2004 00:59:31 -0000
Received: from unknown (HELO thunder.mshome.net) (216.58.165.126)
	by srvr3.iniquinet.com with SMTP; 29 Apr 2004 00:58:58 -0000
Received: from localhost.localdomain (thunder.mshome.net [192.168.0.250])
	by thunder.mshome.net (Postfix) with ESMTP id EBB3CE805E;
	Wed, 28 Apr 2004 18:58:57 -0600 (MDT)
Received: from logicalchaos.org (thunder.mshome.net [192.168.0.250])
	by thunder.mshome.net (Postfix) with SMTP id 20001E805E;
	Wed, 28 Apr 2004 18:58:55 -0600 (MDT)
Date: Wed, 28 Apr 2004 18:57:53 -0600
From: Robert Creager 
To: Josh Berkus 
Cc: pg@fastcrypt.com, =?ISO-8859-1?Q?Dirk=5FLutzeb=E4ck?= ,
	ohp@pyrenet.fr, Tom Lane ,
	Joe Conway , "scott.marlowe" ,
	Bruce Momjian ,
	pgsql-performance@postgresql.org, Neil Conway 
Subject: Re: Wierd context-switching issue on Xeon
Message-Id: <20040428185753.56614b2c@thunder.mshome.net>
In-Reply-To: <200404211029.43675.josh@agliodbs.com>
References: 
	<408664CF.9040507@aeccom.com>
	<1082559931.1557.235.camel@localhost.localdomain>
	<200404211029.43675.josh@agliodbs.com>
Organization: Starlight Vision, LLC.
X-Mailer: Sylpheed version 0.9.9claws (GTK+ 1.2.10; i586-mandrake-linux-gnu)
Mime-Version: 1.0
Content-Type: multipart/signed; protocol="application/pgp-signature";
	micalg="pgp-sha1";
	boundary="Signature=_Wed__28_Apr_2004_18_57_53_-0600_NKZvs8lnAAF58pWn"
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No,
 hits=2.0 tagged_above=0.0 required=5.0 tests=DRASTIC_REDUCED
X-Spam-Level: *
X-Archive-Number: 200404/489
X-Sequence-Number: 6789

--Signature=_Wed__28_Apr_2004_18_57_53_-0600_NKZvs8lnAAF58pWn
Content-Type: text/plain; charset=US-ASCII
Content-Disposition: inline
Content-Transfer-Encoding: 7bit

When grilled further on (Wed, 21 Apr 2004 10:29:43 -0700),
Josh Berkus  confessed:

> Dave,
> 
> > After some testing if you use the current head code for s_lock.c which
> > has some mods in it to alleviate this situation, and change
> > SPINS_PER_DELAY to 10 you can drastically reduce the cs with tom's test.
> > I am seeing a slight degradation in throughput using pgbench -c 10 -t
> > 1000 but it might be liveable, considering the alternative is unbearable
> > in some situations.
> >
> > Can anyone else replicate my results?
> 
> Can you produce a patch against 7.4.1?   I'd like to test your fix against a 
> real-world database.

I would like to see the same, as I have a system that exhibits the same behavior
on a production db that's running 7.4.1.

Cheers,
Rob


-- 
 18:55:22 up  1:40,  4 users,  load average: 2.00, 2.04, 2.00
Linux 2.6.5-01 #7 SMP Fri Apr 16 22:45:31 MDT 2004

--Signature=_Wed__28_Apr_2004_18_57_53_-0600_NKZvs8lnAAF58pWn
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iEYEARECAAYFAkCQUxEACgkQLQ/DKuwDYzkGIACfVeE6q/BGr/ZwShuQ8pM0ch2F
YlcAoIQSiwr7bp0G1OuazNm2OkqES15D
=XbD0
-----END PGP SIGNATURE-----

--Signature=_Wed__28_Apr_2004_18_57_53_-0600_NKZvs8lnAAF58pWn--


From pgsql-performance-owner@postgresql.org  Thu Apr 29 10:20:43 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 80E25D1E3E5
	for ;
	Thu, 29 Apr 2004 10:20:39 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 71725-01
	for ;
	Thu, 29 Apr 2004 10:20:21 -0300 (ADT)
Received: from server.pyrenet.fr (server.pyrenet.fr [194.250.190.1])
	by svr1.postgresql.org (Postfix) with ESMTP id B193CD1E28E
	for ;
	Thu, 29 Apr 2004 10:20:17 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by localhost.pyrenet.fr (Postfix) with ESMTP
	id 6612D2B8CA; Thu, 29 Apr 2004 15:20:21 +0200 (MET DST)
Received: from server.pyrenet.fr ([127.0.0.1])
	by localhost (server [127.0.0.1]) (amavisd-new, port 10024) with ESMTP
	id 11398-01; Thu, 29 Apr 2004 15:20:18 +0200 (MET DST)
Received: from server.pyrenet.fr (server.pyrenet.fr [194.250.190.1])
	by server.pyrenet.fr (Postfix) with ESMTP
	id 24CE12B8A2; Thu, 29 Apr 2004 15:20:18 +0200 (MET DST)
Date: Thu, 29 Apr 2004 15:20:18 +0200 (MET DST)
From: ohp@pyrenet.fr
Reply-To: ohp@pyrenet.fr
To: Robert Creager 
Cc: Josh Berkus , pg@fastcrypt.com,
	=?ISO-8859-1?Q?Dirk=5FLutzeb=E4ck?= ,
	Tom Lane , Joe Conway ,
	"scott.marlowe" ,
	Bruce Momjian ,
	pgsql-performance@postgresql.org, Neil Conway 
Subject: Re: Wierd context-switching issue on Xeon
In-Reply-To: <20040428185753.56614b2c@thunder.mshome.net>
Message-ID: 
References: 
	<408664CF.9040507@aeccom.com>
	<1082559931.1557.235.camel@localhost.localdomain>
	<200404211029.43675.josh@agliodbs.com>
	<20040428185753.56614b2c@thunder.mshome.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=iso-8859-1
Content-Transfer-Encoding: 8BIT
X-Virus-Scanned: by amavisd-new at pyrenet.fr
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=2.3 tagged_above=0.0 required=5.0
	tests=DRASTIC_REDUCED, NO_REAL_NAME
X-Spam-Level: **
X-Archive-Number: 200404/490
X-Sequence-Number: 6790

Hi

I'd LOVE to contribute on this but I don't have vmstat and I'm not running
linux.

How can I help?
Regards
On Wed, 28 Apr 2004, Robert Creager wrote:

> Date: Wed, 28 Apr 2004 18:57:53 -0600
> From: Robert Creager 
> To: Josh Berkus 
> Cc: pg@fastcrypt.com, Dirk_Lutzeb�ck , ohp@pyrenet.fr,
>      Tom Lane , Joe Conway ,
>      scott.marlowe ,
>      Bruce Momjian , pgsql-performance@postgresql.org,
>      Neil Conway 
> Subject: Re: [PERFORM] Wierd context-switching issue on Xeon
>
> When grilled further on (Wed, 21 Apr 2004 10:29:43 -0700),
> Josh Berkus  confessed:
>
> > Dave,
> >
> > > After some testing if you use the current head code for s_lock.c which
> > > has some mods in it to alleviate this situation, and change
> > > SPINS_PER_DELAY to 10 you can drastically reduce the cs with tom's test.
> > > I am seeing a slight degradation in throughput using pgbench -c 10 -t
> > > 1000 but it might be liveable, considering the alternative is unbearable
> > > in some situations.
> > >
> > > Can anyone else replicate my results?
> >
> > Can you produce a patch against 7.4.1?   I'd like to test your fix against a
> > real-world database.
>
> I would like to see the same, as I have a system that exhibits the same behavior
> on a production db that's running 7.4.1.
>
> Cheers,
> Rob
>
>
>

-- 
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
------------------------------------------------------------------------------
Make your life a dream, make your dream a reality. (St Exupery)

From pgsql-performance-owner@postgresql.org  Thu Apr 29 13:41:30 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id BF58CD1E410
	for ;
	Thu, 29 Apr 2004 13:41:27 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 62612-09
	for ;
	Thu, 29 Apr 2004 13:41:07 -0300 (ADT)
Received: from lin-mgregory11.caset.buffalo.edu
	(lin-mgregory11.caset.buffalo.edu [128.205.152.74])
	by svr1.postgresql.org (Postfix) with ESMTP id 4CDAAD1CAA7
	for ;
	Thu, 29 Apr 2004 13:41:06 -0300 (ADT)
Received: by lin-mgregory11.caset.buffalo.edu (Postfix, from userid 129391)
	id 4C7F21D499; Thu, 29 Apr 2004 12:41:06 -0400 (EDT)
Date: Thu, 29 Apr 2004 12:41:06 -0400
From: Roelant Ossewaarde 
To: pgsql-performance@postgresql.org
Subject: Use arrays or not?
Message-ID: <20040429164059.GC24558@belboek.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.4i
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/491
X-Sequence-Number: 6791

Hi,

I am building an application using postgresql to store XML-records. There is a
debate within the group of developers about the best way to store our data. I
hope you can help us make a decision.

The data consists of XML-records, with a lot of XML-fields. I want to store
the XML as it is, so taking the information from the XML-records and then
storing it in a different-from-XML-format is not an option.

Each XML-record describes data about one book. If an update of bookdata comes,
the XML itself is not changed, but a new XML-record is stored with the updated
data. Via a complex scheme of combining a base record and its updates, the
final dataset is produced that is used in the application.

There are different XML-formats that need to be combined. Right now, we can
handle three different XML-formats, each with its own structure (but all
describing book-data).

Searching is done via a simple table lookup on three different fields: title,
author and subject. The data for these fields is extracted from the
database. Each book has a unique identifier (EAN13, derivative of ISBN).

Here is one way to organize the database:
table title:
TITLE | EAN13, indexing on TITLE

table author:
AUTHOR | EAN13, indexing on AUTHOR

table subject:
SUBJECT | EAN13, indexing on SUBJECT.

Finally:
table record:
EAN13 | ARRAY OF XML-records.

It's the last table that I am most curious (and worried) about, the question
being mainly what the optimal way of structuring that table is. Option 1 is
the given option: adding/deleting an XML-record for the same book requires
adding/deleting it to/from the array of XML-records.

Option 2 would be something like this:
EAN13 | XML-record 
where, if a book has several records describing it, there are multiple entries
of the EAN13|XML-record - pair. Adding an XML-record for the same book,
requires adding a new entry to the table as a whole.

So, option 1-tables look like this:
EAN13 | ARRAY OF XML-records
0001 | {..., ..., ...}
0002 | {..., ..., ...}

Option-2 tables look like this:
EAN13 | ARRAY OF XML-records
0001 | ...
0001 | ...
0002 | ...
0002 | ...

We can't decide which one is best. These are some issues we can think of:

Indexing: For option 1, the EAN13-index remains unique, even if you have
multiple XML-records; for option 2 it does not, since multiple XML-records are
stored as multiple tuples. On the other hand, an additional internal index can
be used to link the several tuples of option 2 to the information in the
`lookup'-tables (author, title, keyword). Does any of these two options
increase query efficiency, ie. speed?

Database growth: On average, the information about a book is updated three
times per year. In option 1, this means that the length of the table does not
increase, but the width does. If we choose option 2, if we have three updates
per book each year, the length of the table triples, but the width does
not. What is more costly to store for postgres, long arrays or long tables?

Integrity: Option 1 means that our software needs to keep track of all the
bookkeeping for arrays, since such support is quite rudimentary in
postgres. For example, it is hard to take out a record from the middle of an
array. Also, a multidimensional array, which contains for each record the
record itself and its type, is even harder to maintain. Option 2 has a simpler
datatype, so integrity can be easier inforced using the standard
postgres-machinery of variable-types etc.

Arrays are non-standard SQL, and I hear that PHP-support for postgres & arrays
is rudimentary. So that might be an argument to avoid using them, and go for
option 2. From the standpoint of performance (or wisdom), can you help me
decide what I should choose? Or is there maybe an even better way to structure
my data?

Thanks for any contribution!

Roelant.

From pgsql-performance-owner@postgresql.org  Thu Apr 29 13:54:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 8BFBED1CA32
	for ;
	Thu, 29 Apr 2004 13:54:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 67275-10
	for ;
	Thu, 29 Apr 2004 13:54:04 -0300 (ADT)
Received: from email10.aon.at (WARSL402PIP8.highway.telekom.at [195.3.96.97])
	by svr1.postgresql.org (Postfix) with SMTP id C7E44D1E3DD
	for ;
	Thu, 29 Apr 2004 13:54:02 -0300 (ADT)
Received: (qmail 229764 invoked from network); 29 Apr 2004 16:54:02 -0000
Received: from m148p023.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.119]) (envelope-sender )
	by qmail1rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 29 Apr 2004 16:54:02 -0000
From: Manfred Koizar 
To: sailesh@cs.berkeley.edu
Cc: Tom Lane , Robert Treat ,
	pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] Number of pages in a random sample
Date: Thu, 29 Apr 2004 18:56:36 +0200
Message-ID: <0nc290hvvr6tcv45enh93afcgbg8vuk6j9@email.aon.at>
References: <1081879359.25537.526.camel@camel>
 <25556.1081883922@sss.pgh.pa.us>
	
	<22642.1082074729@sss.pgh.pa.us>
	
	<29060.1082126089@sss.pgh.pa.us>
	
	<25762.1082390410@sss.pgh.pa.us>
	
	
In-Reply-To: 
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/492
X-Sequence-Number: 6792

On Mon, 26 Apr 2004 08:08:16 -0700, Sailesh Krishnamurthy
 wrote:
> "A Bi-Level Bernoulli Scheme for Database Sampling"
> Peter Haas, Christian Koenig (SIGMOD 2004) 

Does this apply to our problem?  AFAIK with Bernoulli sampling you don't
know the sample size in advance.

Anyway, thanks for the hint.  Unfortunately I couldn't find the
document.  Do you have a link?

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Thu Apr 29 14:01:00 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 45001D1CA32
	for ;
	Thu, 29 Apr 2004 14:00:55 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 76003-06
	for ;
	Thu, 29 Apr 2004 14:00:30 -0300 (ADT)
Received: from email07.aon.at (WARSL402PIP8.highway.telekom.at [195.3.96.97])
	by svr1.postgresql.org (Postfix) with SMTP id E924FD1E3DD
	for ;
	Thu, 29 Apr 2004 14:00:28 -0300 (ADT)
Received: (qmail 250966 invoked from network); 29 Apr 2004 17:00:28 -0000
Received: from m148p023.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.119]) (envelope-sender )
	by 172.18.5.236 (qmail-ldap-1.03) with SMTP
	for ; 29 Apr 2004 17:00:28 -0000
From: Manfred Koizar 
To: Tom Lane 
Cc: "Gary Doades" ,
	pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
Date: Thu, 29 Apr 2004 19:03:03 +0200
Message-ID: 
References: <20040426171103.45600.qmail@uno.mnl.com>
	<23032.1083126479@sss.pgh.pa.us> <408F5ECD.29849.CA91084@localhost>
	
	<27806.1083157504@sss.pgh.pa.us>
In-Reply-To: <27806.1083157504@sss.pgh.pa.us>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/493
X-Sequence-Number: 6793

On Wed, 28 Apr 2004 09:05:04 -0400, Tom Lane  wrote:
>> [ ... visibility information in index tuples ... ]

>Storing that information would at least double the overhead space used
>for each index tuple.  The resulting index bloat would significantly
>slow index operations by requiring more I/O.  So it's far from clear
>that this would be a win, even for those who care only about select
>speed.

While the storage overhead could be reduced to 1 bit (not a joke) we'd
still have the I/O overhead of locating and updating index tuples for
every heap tuple deleted/updated.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Thu Apr 29 14:11:38 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 4C4EBD1DF5D
	for ;
	Thu, 29 Apr 2004 14:11:36 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 81803-04
	for ;
	Thu, 29 Apr 2004 14:11:16 -0300 (ADT)
Received: from email06.aon.at (WARSL402PIP3.highway.telekom.at [195.3.96.75])
	by svr1.postgresql.org (Postfix) with SMTP id C584FD1E5E9
	for ;
	Thu, 29 Apr 2004 14:11:14 -0300 (ADT)
Received: (qmail 525140 invoked from network); 29 Apr 2004 17:11:14 -0000
Received: from m148p023.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.119]) (envelope-sender )
	by qmail6rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 29 Apr 2004 17:11:14 -0000
From: Manfred Koizar 
To: Rod Taylor 
Cc: Vitaly Belman ,
	Postgresql Performance 
Subject: Re: Simply join in PostrgeSQL takes too long
Date: Thu, 29 Apr 2004 19:13:49 +0200
Message-ID: 
References: <282909734.20040428002740@012.net.il>
	<1083103293.30065.100.camel@jester>
	
	<1083155014.30065.189.camel@jester>
In-Reply-To: <1083155014.30065.189.camel@jester>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/494
X-Sequence-Number: 6794

On Wed, 28 Apr 2004 08:23:35 -0400, Rod Taylor  wrote:
>The below plpgsql forces the kind of algorithm we wish the planner could
>choose. It should be fairly quick irregardless of dataset.

That reminds me of hash aggregation.  So here's another idea for Vitaly:

	SELECT book_id
	  FROM ...
	 WHERE ...
	 GROUP BY book_id
	 LIMIT ...

Servus
 Manfred

From pgsql-sql-owner@postgresql.org  Thu Apr 29 14:24:55 2004
X-Original-To: pgsql-sql-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 0B7A4D1DF5D
	for ;
	Thu, 29 Apr 2004 14:24:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 83223-06
	for ;
	Thu, 29 Apr 2004 14:24:33 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id B4301D1E181
	for ; Thu, 29 Apr 2004 14:24:31 -0300 (ADT)
Received: from [63.195.55.98] (HELO spooky)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4973814; Thu, 29 Apr 2004 10:25:57 -0700
Content-Type: text/plain;
  charset="iso-8859-1"
From: Josh Berkus 
Organization: Aglio Database Solutions
To: Roelant Ossewaarde 
Subject: Re: Use arrays or not?
Date: Thu, 29 Apr 2004 10:23:58 -0700
User-Agent: KMail/1.4.3
References: <20040429164059.GC24558@belboek.com>
In-Reply-To: <20040429164059.GC24558@belboek.com>
Cc: pgsql-sql@postgresql.org
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <200404291023.58712.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/318
X-Sequence-Number: 17693

Roelant,

Yours is not a performance question, so I'm crossing it over to SQL for advice 
on database design.

> I am building an application using postgresql to store XML-records. There
> is a debate within the group of developers about the best way to store our
> data. I hope you can help us make a decision.
>
> The data consists of XML-records, with a lot of XML-fields. I want to store
> the XML as it is, so taking the information from the XML-records and then
> storing it in a different-from-XML-format is not an option.
>
> Each XML-record describes data about one book. If an update of bookdata
> comes, the XML itself is not changed, but a new XML-record is stored with
> the updated data. Via a complex scheme of combining a base record and its
> updates, the final dataset is produced that is used in the application.
>
> There are different XML-formats that need to be combined. Right now, we can
> handle three different XML-formats, each with its own structure (but all
> describing book-data).
>
> Searching is done via a simple table lookup on three different fields:
> title, author and subject. The data for these fields is extracted from the
> database. Each book has a unique identifier (EAN13, derivative of ISBN).
>
> Here is one way to organize the database:
> table title:
> TITLE | EAN13, indexing on TITLE
>
> table author:
> AUTHOR | EAN13, indexing on AUTHOR
>
> table subject:
> SUBJECT | EAN13, indexing on SUBJECT.

This is a *very* strange way of setting up your database.  Are you new to 
Relational Databases and SQL?   If so, I'd recommend starting with a book on 
relational database design.

Either that, or you're a victim of UML design.   

If only one author, title and subject are allowed per book, you should have:

table books
	EAN13 | TITLE  | AUTHOR | SUBJECT

> Finally:
> table record:
> EAN13 | ARRAY OF XML-records.
>
> It's the last table that I am most curious (and worried) about, the
> question being mainly what the optimal way of structuring that table is.
> Option 1 is the given option: adding/deleting an XML-record for the same
> book requires adding/deleting it to/from the array of XML-records.
>
> Option 2 would be something like this:
> EAN13 | XML-record
> where, if a book has several records describing it, there are multiple
> entries of the EAN13|XML-record - pair. Adding an XML-record for the same
> book, requires adding a new entry to the table as a whole.

In my mind, there is no question that this is the best way to do things.  It 
is a normalized data structure, as opposed to the arrays, which are now.

>
> So, option 1-tables look like this:
> EAN13 | ARRAY OF XML-records
> 0001 | {..., ..., ...}
> 0002 | {..., ..., ...}
>
> Option-2 tables look like this:
> EAN13 | ARRAY OF XML-records
> 0001 | ...
> 0001 | ...
> 0002 | ...
> 0002 | ...
>
> We can't decide which one is best. These are some issues we can think of:
>
> Indexing: For option 1, the EAN13-index remains unique, even if you have
> multiple XML-records; for option 2 it does not, since multiple XML-records
> are stored as multiple tuples. On the other hand, an additional internal
> index can be used to link the several tuples of option 2 to the information
> in the `lookup'-tables (author, title, keyword). Does any of these two
> options increase query efficiency, ie. speed?
>
> Database growth: On average, the information about a book is updated three
> times per year. In option 1, this means that the length of the table does
> not increase, but the width does. If we choose option 2, if we have three
> updates per book each year, the length of the table triples, but the width
> does not. What is more costly to store for postgres, long arrays or long
> tables?
>
> Integrity: Option 1 means that our software needs to keep track of all the
> bookkeeping for arrays, since such support is quite rudimentary in
> postgres. For example, it is hard to take out a record from the middle of
> an array. Also, a multidimensional array, which contains for each record
> the record itself and its type, is even harder to maintain. Option 2 has a
> simpler datatype, so integrity can be easier inforced using the standard
> postgres-machinery of variable-types etc.
>
> Arrays are non-standard SQL, and I hear that PHP-support for postgres &
> arrays is rudimentary. So that might be an argument to avoid using them,
> and go for option 2. From the standpoint of performance (or wisdom), can
> you help me decide what I should choose? Or is there maybe an even better
> way to structure my data?
>
> Thanks for any contribution!
>
> Roelant.


-- 
Josh Berkus
Aglio Database Solutions
San Francisco

From pgsql-performance-owner@postgresql.org  Thu Apr 29 14:37:49 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 15C70D1E60F
	for ;
	Thu, 29 Apr 2004 14:37:46 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 89915-06
	for ;
	Thu, 29 Apr 2004 14:37:25 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 9C508D1E616
	for ;
	Thu, 29 Apr 2004 14:37:24 -0300 (ADT)
Received: from [134.22.68.249] (dyn-68-249.tor.dsl.tht.net [134.22.68.249])
	by tht.net (Postfix) with ESMTP
	id C252276AF3; Thu, 29 Apr 2004 13:37:27 -0400 (EDT)
Subject: Re: Simply join in PostrgeSQL takes too long
From: Rod Taylor 
To: Manfred Koizar 
Cc: Vitaly Belman ,
	Postgresql Performance 
In-Reply-To: 
References: <282909734.20040428002740@012.net.il>
	<1083103293.30065.100.camel@jester>
	
	<1083155014.30065.189.camel@jester>
	
Content-Type: text/plain
Message-Id: <1083260206.30065.300.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Thu, 29 Apr 2004 13:36:47 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/495
X-Sequence-Number: 6795

On Thu, 2004-04-29 at 13:13, Manfred Koizar wrote:
> On Wed, 28 Apr 2004 08:23:35 -0400, Rod Taylor  wrote:
> >The below plpgsql forces the kind of algorithm we wish the planner could
> >choose. It should be fairly quick irregardless of dataset.
> 
> That reminds me of hash aggregation.  So here's another idea for Vitaly:

The reason for the function is that the sort routines (hash aggregation
included) will not stop in mid-sort, although I believe that feature is
on the TODO list.

I believe Vitaly will achieve 10ms or less query times using that
function.



From pgsql-performance-owner@postgresql.org  Thu Apr 29 14:57:04 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 2B9B6D1E60F
	for ;
	Thu, 29 Apr 2004 14:57:03 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 96471-08
	for ;
	Thu, 29 Apr 2004 14:56:42 -0300 (ADT)
Received: from email08.aon.at (WARSL402PIP7.highway.telekom.at [195.3.96.94])
	by svr1.postgresql.org (Postfix) with SMTP id 34706D1E6B3
	for ;
	Thu, 29 Apr 2004 14:56:41 -0300 (ADT)
Received: (qmail 357716 invoked from network); 29 Apr 2004 17:56:38 -0000
Received: from m148p023.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.8.119]) (envelope-sender )
	by 172.18.5.237 (qmail-ldap-1.03) with SMTP
	for ; 29 Apr 2004 17:56:38 -0000
From: Manfred Koizar 
To: Rod Taylor 
Cc: Vitaly Belman ,
	Postgresql Performance 
Subject: Re: Simply join in PostrgeSQL takes too long
Date: Thu, 29 Apr 2004 19:59:13 +0200
Message-ID: 
References: <282909734.20040428002740@012.net.il>
	<1083103293.30065.100.camel@jester>
	
	<1083155014.30065.189.camel@jester>
	
	<1083260206.30065.300.camel@jester>
In-Reply-To: <1083260206.30065.300.camel@jester>
X-Mailer: Forte Agent 1.93/32.576 English (American)
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/496
X-Sequence-Number: 6796

On Thu, 29 Apr 2004 13:36:47 -0400, Rod Taylor  wrote:
>The reason for the function is that the sort routines (hash aggregation
>included) will not stop in mid-sort

Good point.

Servus
 Manfred

From pgsql-performance-owner@postgresql.org  Thu Apr 29 15:13:53 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 76078D1E752
	for ;
	Thu, 29 Apr 2004 15:13:49 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 10306-01
	for ;
	Thu, 29 Apr 2004 15:13:25 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id E1FC2D1CA40
	for ;
	Thu, 29 Apr 2004 15:13:22 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 211
	for ; Thu, 29 Apr 2004 19:12:49 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Thu, 29 Apr 2004 19:13:51 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <409153EF.24647.144E9BF7@localhost>
In-reply-to: 
References: <27806.1083157504@sss.pgh.pa.us>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/497
X-Sequence-Number: 6797

On 29 Apr 2004 at 19:03, Manfred Koizar wrote:
 
> While the storage overhead could be reduced to 1 bit (not a joke) we'd
> still have the I/O overhead of locating and updating index tuples for
> every heap tuple deleted/updated.

But this is what a lot of DBMSs do and seem to do well enough. I can see that the 
MVCC system gives additional problems, but maybe it shouldn't be dismissed so lightly.

Coming from a MS SQLServer platform I have spent a lot of time optimising SQL in 
PostgreSQL to be comparable to SQLServer. For the most part I have done this, but 
some things are just slower in PostgreSQL.

Recently  I have been looking at raw performance (CPU, IO) rather than the plans. I 
have some test queries that (as far as I can determine) use the same access plans on 
PostgreSQL and SQLServer. Getting to the detail, an index scan of an index on a 
integer column (222512 rows) takes 60ms on SQLServer and 540ms on PostgreSQL.
A full seq table scan on the same table without the index on the other hand takes 370ms 
in SQLServer and 420ms in PostgreSQL.

I know that the platforms are different (windows 2000 vs Linux 2.6.3), but the statement 
was executed several times to make sure the index and data was in cache (no disk io) 
on both systems. Same data, Same CPU, Same disks, Same memory, Same 
motherboards.

The only thing I can think of is the way that the index scan is performed on each 
platform, SQLServer can use the data directly from the index. This makes the biggest 
difference in multi join statements where several of the intermediate tables do not need 
to be accessed at all, the data is contained in the join indexes. This results in almost an 
order of magnitude performance difference for the same data.

I would be nice to get a feel for how much performance loss would be incurred in 
maintaining the index flags against possible performance gains for getting the data back 
out again.

Regards,
Gary.


From pgsql-performance-owner@postgresql.org  Thu Apr 29 15:21:57 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E18E3D1CA43
	for ;
	Thu, 29 Apr 2004 15:21:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 05967-06
	for ;
	Thu, 29 Apr 2004 15:21:34 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id ECC56D1E600
	for ;
	Thu, 29 Apr 2004 15:21:32 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4974375; Thu, 29 Apr 2004 11:22:58 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: Robert Creager 
Subject: Re: Wierd context-switching issue on Xeon
Date: Thu, 29 Apr 2004 11:21:51 -0700
User-Agent: KMail/1.5.4
Cc: pg@fastcrypt.com, =?iso-8859-1?q?Dirk=5FLutzeb=E4ck?= ,
	ohp@pyrenet.fr, Tom Lane ,
	Joe Conway , "scott.marlowe" ,
	Bruce Momjian ,
	pgsql-performance@postgresql.org, Neil Conway 
References: 
	<200404211029.43675.josh@agliodbs.com>
	<20040428185753.56614b2c@thunder.mshome.net>
In-Reply-To: <20040428185753.56614b2c@thunder.mshome.net>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404291121.51247.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/498
X-Sequence-Number: 6798

Rob,

> I would like to see the same, as I have a system that exhibits the same 
behavior
> on a production db that's running 7.4.1.

If you checked the thread follow-ups,  you'd see that *decreasing* 
spins_per_delay was not beneficial.   Instead, try increasing them, one step 
at a time:

(take baseline measurement at 100)
250
500
1000
1500
2000
3000
5000

... until you find an optimal level.   Then report the results to us!

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Thu Apr 29 16:13:14 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id AEF33D1E930
	for ;
	Thu, 29 Apr 2004 16:13:12 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 26966-08
	for ;
	Thu, 29 Apr 2004 16:12:53 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 212B7D1E5F7
	for ;
	Thu, 29 Apr 2004 16:12:52 -0300 (ADT)
Received: from [134.22.68.249] (dyn-68-249.tor.dsl.tht.net [134.22.68.249])
	by tht.net (Postfix) with ESMTP
	id E5F9276B03; Thu, 29 Apr 2004 15:12:56 -0400 (EDT)
Subject: Re: planner/optimizer question
From: Rod Taylor 
To: Gary Doades 
Cc: Postgresql Performance 
In-Reply-To: <409153EF.24647.144E9BF7@localhost>
References: <27806.1083157504@sss.pgh.pa.us>
	<409153EF.24647.144E9BF7@localhost>
Content-Type: text/plain
Message-Id: <1083265934.30065.359.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Thu, 29 Apr 2004 15:12:15 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/499
X-Sequence-Number: 6799

> I would be nice to get a feel for how much performance loss would be incurred in 
> maintaining the index flags against possible performance gains for getting the data back 
> out again.

I guess the real question is, why maintain index flags and not simply
drop the index entry altogether?

A more interesting case would be to have the backend process record
index tuples that it would invalidate (if committed), then on commit
send that list to a garbage collection process.

It's still vacuum -- just the reaction time for it would be much
quicker.



From pgsql-performance-owner@postgresql.org  Thu Apr 29 16:23:11 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D368ED1E9A7
	for ;
	Thu, 29 Apr 2004 16:23:10 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 27231-09
	for ;
	Thu, 29 Apr 2004 16:22:51 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 92DEED1E5DC
	for ;
	Thu, 29 Apr 2004 16:22:48 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 226
	for ; Thu, 29 Apr 2004 20:22:17 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Thu, 29 Apr 2004 20:23:19 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question
Message-ID: <40916437.19555.148E3602@localhost>
In-reply-to: <1083265934.30065.359.camel@jester>
References: <409153EF.24647.144E9BF7@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/500
X-Sequence-Number: 6800

> 
> I guess the real question is, why maintain index flags and not simply
> drop the index entry altogether?
> 
> A more interesting case would be to have the backend process record
> index tuples that it would invalidate (if committed), then on commit
> send that list to a garbage collection process.
> 
> It's still vacuum -- just the reaction time for it would be much
> quicker.
> 
This was my original question.

I guess the problem is with MVCC. The row may have gone from your 
current view of the table but not from someone elses. I don't (yet) 
understand the way it works to say for sure, but I still think it is worth 
pursuing further for someone who does know the deep stuff. They seem 
to have concluded that it is not worth it however.

Cheers,
Gary.




From pgsql-performance-owner@postgresql.org  Thu Apr 29 17:14:42 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 17971D1EB6B
	for ;
	Thu, 29 Apr 2004 17:12:54 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 60864-02
	for ;
	Thu, 29 Apr 2004 17:12:33 -0300 (ADT)
Received: from www.totalcardinc.com (unknown [64.33.232.186])
	by svr1.postgresql.org (Postfix) with ESMTP id 0F569D1EB30
	for ;
	Thu, 29 Apr 2004 17:12:31 -0300 (ADT)
Received: from RSCHWARZW2K ([10.250.0.37]) (authenticated bits=0)
	by www.totalcardinc.com (8.12.8/8.12.8) with ESMTP id i3TKCTiK000820;
	Thu, 29 Apr 2004 15:12:29 -0500
From: "Rosser Schwarz" 
To: "'Gary Doades'" ,
	
Subject: Re: planner/optimizer question 
Date: Thu, 29 Apr 2004 15:12:29 -0500
Message-ID: <003101c42e26$4c384e90$2500fa0a@CardServices.TCI.com>
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 10.0.4510
In-Reply-To: <409153EF.24647.144E9BF7@localhost>
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/501
X-Sequence-Number: 6801

while you weren't looking, Gary Doades wrote:

> Recently  I have been looking at raw performance (CPU, IO) 
> rather than the plans. I have some test queries that (as far
> as I can determine) use the same access plans on PostgreSQL
> and SQLServer. Getting to the detail, an index scan of an
> index on a integer column (222512 rows) takes 60ms on
> SQLServer and  540ms on PostgreSQL.

After a recent power outage, I had the opportunity to watch both
PostgreSQL and MS SQL come back from forced shutdowns (clean,
though there were active connections, in one case a bulk insert).
PostgreSQL was available and responsive as soon as the postmaster
had started.  MS SQL, on the other hand, took the better part of
an hour to become remotely usable again -- on a radically faster
machine (Dell 6650, versus the 6450 we run PostgreSQL on).

Digging a bit, I noted that once MS SQL was up again, it was
using nearly 2GB main memory even when more or less idle.  From
this, and having observed the performance differences between
the two, I'm left with little alternative but to surmise that
part of MS SQL's noted performance advantage [1] is due to its
forcibly storing its indices in main memory.  Its startup lag
(during which it was utterly unusable; even SELECTs blocked)
could be accounted for by reindexing the tables. [2]

Granted, this is only a hypothesis, is rather unverifyable, and
probably belongs more on ADVOCACY than it does PERFORM, but it
seemed relevant.

It's also entirely possible your indices are using inaccurate
statistical information.  Have you ANALYZEd recently?

/rls

[1] Again, at least in our case, the comparison is entirely
    invalid, as MS SQL gets a hell of a lot more machine than
    PostgreSQL.  Even so, for day-to-day work and queries, even
    our DBA, an until-recently fervent MS SQL advocate can't
    fault PostgreSQL's SELECT, INSERT or DELETE performance.
    We still can't get UPDATEs (at least bulk such) to pass
    muster.

[2] This is further supported by having observed MS SQL run a
    "recovery process" on databases that were entirely unused,
    even for SELECT queries, at the time of the outage.  The
    only thing it might conceivably need to recover on them
    is in-memory indices that were lost when power was lost.

--
Rosser Schwarz
Total Card, Inc.


From pgsql-performance-owner@postgresql.org  Thu Apr 29 17:26:15 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 3524CD1E959
	for ;
	Thu, 29 Apr 2004 17:26:14 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 64003-05
	for ;
	Thu, 29 Apr 2004 17:25:54 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 2F97BD1E968
	for ;
	Thu, 29 Apr 2004 17:25:52 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 236
	for ; Thu, 29 Apr 2004 21:25:16 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Thu, 29 Apr 2004 21:26:18 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <409172FA.1509.14C7E00E@localhost>
In-reply-to: <003101c42e26$4c384e90$2500fa0a@CardServices.TCI.com>
References: <409153EF.24647.144E9BF7@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/502
X-Sequence-Number: 6802


> It's also entirely possible your indices are using inaccurate
> statistical information.  Have you ANALYZEd recently?
> 

In this example the statistics don't matter. The plans used were the same for 
MSSQL and Postgres. I was trying to eliminate the difference in plans 
between the two, which obviously does make a difference, sometimes in 
MSSQL favour and sometimes the other way round. Both systems, having 
decided to do the same index scan, took noticably different times. The 
Postgres database was fully vacuumed and analysed anyway.

I agree about MSSQL recovery time. it sucks. This is why they are making a 
big point about the improved recovery time in "yukon". Although the recovery 
time is important, I see this as an exception, whereas at the moment I am 
interested in the everyday.

Cheers,
Gary.


From pgsql-sql-owner@postgresql.org  Tue May  4 09:55:29 2004
X-Original-To: pgsql-sql-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 83E85D1E785
	for ;
	Thu, 29 Apr 2004 17:27:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 61085-09
	for ;
	Thu, 29 Apr 2004 17:26:41 -0300 (ADT)
Received: from lin-mgregory11.caset.buffalo.edu
	(lin-mgregory11.caset.buffalo.edu [128.205.152.74])
	by svr1.postgresql.org (Postfix) with ESMTP id 1C6D3D1ED90
	for ; Thu, 29 Apr 2004 17:26:39 -0300 (ADT)
Received: by lin-mgregory11.caset.buffalo.edu (Postfix, from userid 129391)
	id CF44F1D49B; Thu, 29 Apr 2004 16:26:40 -0400 (EDT)
Date: Thu, 29 Apr 2004 16:26:40 -0400
From: Roelant Ossewaarde 
To: Richard Huxton 
Cc: Josh Berkus ,
	Roelant Ossewaarde , pgsql-sql@postgresql.org
Subject: Re: Use arrays or not?
Message-ID: <20040429202631.GB25328@belboek.com>
References: <20040429164059.GC24558@belboek.com>
	<200404291023.58712.josh@agliodbs.com> <40914CC2.6050909@archonet>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <40914CC2.6050909@archonet>
User-Agent: Mutt/1.4i
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200405/23
X-Sequence-Number: 17737

Hi,

Thanks for your fast response. But I'm afraid I stated some things unclear.

> >>The data consists of XML-records, with a lot of XML-fields. I want to 
> >>store
> >>the XML as it is, so taking the information from the XML-records and then
> >>storing it in a different-from-XML-format is not an option.
> 
> Actually, your data consists of data. The XML is just scaffolding to 
> enable you to locate and understand your data. Since you are storing it 
> in a relational database, that will use relational scaffolding for its 
> organisation instead. At least partially, you will *have* to parse the 
> values out and organise them differently.
I do, and I know. But I want to store the XML-records as they are. So given an
author, title and an XML-record that are related to that author and title, how
to store that. That's the question. I have good reasons to store the
XML-records as they are, without further parsing them.

> >>Each XML-record describes data about one book. If an update of bookdata
> >>comes, the XML itself is not changed, but a new XML-record is stored with
> >>the updated data. Via a complex scheme of combining a base record and its
> >>updates, the final dataset is produced that is used in the application.
> >>
> >>Searching is done via a simple table lookup on three different fields:
> >>title, author and subject. The data for these fields is extracted from the
> >>database. Each book has a unique identifier (EAN13, derivative of ISBN).
> >>
> >>Here is one way to organize the database:
> >>table title:
> >>TITLE | EAN13, indexing on TITLE
> >>
> >>table author:
> >>AUTHOR | EAN13, indexing on AUTHOR
> >>
> >>table subject:
> >>SUBJECT | EAN13, indexing on SUBJECT.
> >
> >
> >This is a *very* strange way of setting up your database.  Are you new to 
> >Relational Databases and SQL?   If so, I'd recommend starting with a book 
> >on relational database design.
> I agree with Josh - think about a book.

Thank your for the recommendations. But the above thing is just background
information, it will not be stored as such. The important question for me is
the question whether to use arrays or not. With index in the above examples I
do not mean the actual postgres-index, I mean that those are the fields that
are used in searching. One never searches on an EAN13-number, only on author,
title and subject. And one never, by the way, searches for a specific
XML-record, only the total of the stored XML-records per book should be retrieved.

> 
> >If only one author, title and subject are allowed per book, you should 
> >have:
> >
> >table books
> >	EAN13 | TITLE  | AUTHOR | SUBJECT
> 
> If, on the other hand you can have multiple authors (likely) you'll want 
>  something like:
> 
> CREATE TABLE author (
>   ean13        varchar(13), -- Guessing ean13 format
>   author_num   int4,
>   author_name  text,
>   PRIMARY KEY (ean13, author_num)
> );
> 
> Then you can have rows like:
> 
> ('my-ean-number-here', 1, 'Aaron Aardvark')
> ('my-ean-number-here', 2, 'Betty Bee')
> etc.

Yes, I have such a thing. There can be multiple titles, multiple authors and
multiple keywords per book. 

> 
> 
> >>Finally:
> >>table record:
> >>EAN13 | ARRAY OF XML-records.
> >>
> >>It's the last table that I am most curious (and worried) about, the
> >>question being mainly what the optimal way of structuring that table is.
> >>Option 1 is the given option: adding/deleting an XML-record for the same
> >>book requires adding/deleting it to/from the array of XML-records.
> >>
> >>Option 2 would be something like this:
> >>EAN13 | XML-record
> >>where, if a book has several records describing it, there are multiple
> >>entries of the EAN13|XML-record - pair. Adding an XML-record for the same
> >>book, requires adding a new entry to the table as a whole.
> >
> >
> >In my mind, there is no question that this is the best way to do things.  
> >It is a normalized data structure, as opposed to the arrays, which are now.
> 
> Although your option 2 doesn't go quite far enough. You'll also want to 
> know what order these come in. So, assuming you can't have two updates 
> at the same time:
> 
> CREATE TABLE book_history (
>   ean13  varchar(13), -- Guessing ean13 format
>   ts     timestamp with time zone,
>   xml    text,
>   PRIMARY KEY (ean13, ts)
> );

The order is not important; the interpretation of the XML-records is done by
an external module. The order is determined upon the content of the
XML-records, because they can come from different sources and can be combined
in different ways, depending on the application processing the
XML-records. Order is not determined at the moment that the records are
stored, but at the moment the records are interpreted.

> As for your other concerns:
> >>Indexing:
> >>Database growth:
> >>Integrity:
> Just worry about the integrity - if you keep the design simple, 
> PostgreSQL will manage quite large growth on quite small hardware.

What would be a situation in which one should use arrays then?


> Now... I don't think you want to do what you're trying to do. Don't take 
> this personally, but unless you're extremely pushed for time and 
> resources this is almost certainly a bad design choice.

> 1. Wrong tool for the job
> Basically you're taking a relational database and treating it like a 
> filesystem. All you need for what you're doing is a directory-tree to 
> represent the ean13 structure and one file per xml-record. Index the 
> author/title fields with dbm/SQLite. You could write the whole thing in 
> a day - simple, efficient, leverages existing unix tools.

That is correct. I think nothing would beat a dbm-style solution qua
performance, and I'm still considering using that. The added value of a system
like postgresql is the client/server-interface and the omnipresent support of
programming languages, and not in the least the familiarity of most people
with mysql/postgresql in comparison to dbm.

> 2. Wrong job for the tool
> How do I find out which publisher produced the most books in 2003 (I'm 
> assuming this is in your XML somewhere)? Which book is available in the 
> most languages?
> How many updates were applied last month? How many different books did 
> they affect? Why do the numbers not match - which books had multiple 
> changes?
> The first set of questions need you to write code, the second set don't. 
> Why? Because the second set rely on information stored simply and 
> explicitly in the database (book_history as it happens).

We know what set of questions will be asked: we only need to access through
author, title and subject keywords. Another reason maybe to choose a
non-relational model.

> 3. The medium isn't the message
> You don't want to open up your XML records to store them in the 
> database, but I assume you have to in your PHP code, or you can't 
> process individual values. As it stands you're having to extract certain 
> information when an XML update arrives anyway. If the title of a book is 
> amended, then you'll need to remember to update the book_title table. If 
> it's simple to extract more, why not do so? If some of it is fiddly to 
> represent in an SQL database then at least extract everything that is 
> convenient.

There are good reasons for that. The XMLs not necessarily contain the input to
author/title/subject-tables (or columns). I just want to store the XML,
interpret it later. (really, it makes sense!)

> Oh - and I would probably store a "current snapshot" of the book's 
> record separately too. Saves your application having to recalculate it 
> every time it's needed.

I *want* to recalculate it every time it's needed. Because the different
XML-records can be combined in several ways, depending on the application that
is accessing the database-client.

So, let me rephrase my questions:
1. When and why would anyone use arrays?
2. When designing the database, is it really true that there is no performance
difference between a table of which the number of tuples grow by a factor of,
say 10, and a table of which the size of the tuples grow by a factor of, say
10?

Thanks, 

Roelant.

From pgsql-sql-owner@postgresql.org  Thu Apr 29 17:52:22 2004
X-Original-To: pgsql-sql-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DBFA2D1ED9A
	for ;
	Thu, 29 Apr 2004 17:47:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 69735-09
	for ;
	Thu, 29 Apr 2004 17:47:12 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 169C8D1E964
	for ; Thu, 29 Apr 2004 17:47:10 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4975391; Thu, 29 Apr 2004 13:48:24 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: Roelant Ossewaarde ,
	Richard Huxton 
Subject: Re: Use arrays or not?
Date: Thu, 29 Apr 2004 13:47:16 -0700
User-Agent: KMail/1.5.4
Cc: pgsql-sql@postgresql.org
References: <20040429164059.GC24558@belboek.com> <40914CC2.6050909@archonet>
	<20040429202631.GB25328@belboek.com>
In-Reply-To: <20040429202631.GB25328@belboek.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404291347.16488.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/324
X-Sequence-Number: 17699

Roelant,

> So, let me rephrase my questions:
> 1. When and why would anyone use arrays?

When the data itself is an ordered set of items which is indivisible and lacks 
meaning outside the ordered set.   For example, a set of ordered pairs of 
molecules in a gene snippet.  Or a mathematical matrix.

> 2. When designing the database, is it really true that there is no 
performance
> difference between a table of which the number of tuples grow by a factor 
of,
> say 10, and a table of which the size of the tuples grow by a factor of, say
> 10?

Nobody's tested anything.   I would *tend* to think that PostgreSQL would 
handle more-of-less-wide-rows somewhat better, but that's just a guess.   

Hmmm ... not completely a guess.   Postgres, by default, compresses fields 
over 8K in size (see TOAST in the docs).  This makes those fields somewhat 
slower to update.   So if 1 XML rec < 8k but 4 XML rec > 8k, there could be a 
small-but-noticeable performance loss from going to "broad" rows.

If I had your application, I would not go for the array approach, jjust to 
avoid maintainence headaches.    For example, what happens when the books 
start having a variable number of XML records?   Normalized designs are 
almost always easier to deal with from a perspective of long-term 
maintainence.

The arrays, as far as I can tell, gain you nothing in ethier performance or 
convenience.

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Thu Apr 29 18:07:14 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 45002D1EE8E
	for ;
	Thu, 29 Apr 2004 17:54:40 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 74407-06
	for ;
	Thu, 29 Apr 2004 17:54:20 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 11CE4D1EDB9
	for ;
	Thu, 29 Apr 2004 17:54:18 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4975456; Thu, 29 Apr 2004 13:55:47 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: "Gary Doades" ,
	pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question
Date: Thu, 29 Apr 2004 13:54:33 -0700
User-Agent: KMail/1.5.4
References: <409153EF.24647.144E9BF7@localhost>
	<409172FA.1509.14C7E00E@localhost>
In-Reply-To: <409172FA.1509.14C7E00E@localhost>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404291354.33304.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/503
X-Sequence-Number: 6803

Gary,

> In this example the statistics don't matter. The plans used were the same 
for 
> MSSQL and Postgres. I was trying to eliminate the difference in plans 
> between the two, which obviously does make a difference, sometimes in 
> MSSQL favour and sometimes the other way round. Both systems, having 
> decided to do the same index scan, took noticably different times. The 
> Postgres database was fully vacuumed and analysed anyway.

It's also quite possble the MSSQL simply has more efficient index scanning 
implementation that we do.    They've certainly had incentive; their storage 
system sucks big time for random lookups and they need those fast indexes.  
(just try to build a 1GB adjacency list tree on SQL Server.   I dare ya).

Certainly the fact that MSSQL is essentially a single-user database makes 
things easier for them.    They don't have to maintain multiple copies of the 
index tuples in memory.    I think that may be our main performance loss.

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Thu Apr 29 18:08:54 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E9C41D1EE95
	for ;
	Thu, 29 Apr 2004 18:08:52 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 79425-06
	for ;
	Thu, 29 Apr 2004 18:08:34 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id E9EB7D1EDA0
	for ;
	Thu, 29 Apr 2004 18:08:30 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 243
	for ; Thu, 29 Apr 2004 22:08:01 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Thu, 29 Apr 2004 22:08:28 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question
Message-ID: <40917CDC.3078.14EE7897@localhost>
In-reply-to: <20040429203519.GI26634@it.is.rice.edu>
References: <409172FA.1509.14C7E00E@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/504
X-Sequence-Number: 6804

On 29 Apr 2004 at 15:35, Kenneth Marshall wrote:

> Did you try to cluster based on the index?
> 
> --Ken

Yes, This speeds up the index scan a little (12%). This to me just 
reinforces the overhead that subsequently having to go and fetch the 
data tuple actually has on the performance.

Cheers,
Gary.


From pgsql-performance-owner@postgresql.org  Thu Apr 29 18:09:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 36EE5D1EEAC
	for ;
	Thu, 29 Apr 2004 18:09:24 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 84785-01
	for ;
	Thu, 29 Apr 2004 18:09:04 -0300 (ADT)
Received: from fep12.012.net.il (fep12.012.net.il [212.117.129.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 56FC9D1EDB9
	for ;
	Thu, 29 Apr 2004 18:09:01 -0300 (ADT)
Received: from 80.178.84.42.forward.012.net.il ([80.178.84.42])
	by fep12.012.net.il with ESMTP
	id <20040429210945.DJKT26710.fep12@80.178.84.42.forward.012.net.il>;
	Fri, 30 Apr 2004 00:09:45 +0300
Date: Fri, 30 Apr 2004 00:09:36 +0300
From: Vitaly Belman 
Reply-To: Vitaly Belman 
X-Priority: 3 (Normal)
Message-ID: <147254626500.20040430000936@012.net.il>
To: Manfred Koizar 
Cc: Rod Taylor ,
	Postgresql Performance 
Subject: Re: Simply join in PostrgeSQL takes too long
In-Reply-To: 
References: <282909734.20040428002740@012.net.il>
	<1083103293.30065.100.camel@jester>
	
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=0.9 tagged_above=0.0 required=5.0
	tests=PRIORITY_NO_NAME, RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/505
X-Sequence-Number: 6805

Hello Manfred,

I thank everyone for helping me on this - So many tips.

I am in the middle of going through them all, till now disabling the
enable_mergejoin really helped.

Also, I agree that the design might be flawed (I could use triggers
and stuff like that) but for now I am just comparing how my project
will run on PostgreSQL (Considering migration from MySQL).

I'll be reporting back on how the other stuff helped.

Regards,
 Vitaly Belman
 
 ICQ: 1912453
 AIM: VitalyB1984
 MSN: tmdagent@hotmail.com
 Yahoo!: VitalyBe

Wednesday, April 28, 2004, 11:24:41 AM, you wrote:

MK> On Tue, 27 Apr 2004 18:01:34 -0400, Rod Taylor  wrote:
>>On Tue, 2004-04-27 at 17:27, Vitaly Belman wrote:
>>> Hello pgsql-performance,
>>> 
>>>   I discussed the whole subject for some time in DevShed and didn't
>>>   achieve much (as for results). I wonder if any of you guys can help
>>>   out:
>>> 
>>>   http://forums.devshed.com/t136202/s.html

>>The point is that a book cannot be of a certain genre more than once.

MK> Rod, he has a hierarchy of genres.  Genre 1 has 6379 child genres and a
MK> book can be in more than one of these.

MK> Vitaly, though LIMIT makes this look like a small query, DISTINCT
MK> requires the whole result set to be retrieved.  0.7 seconds doesn't look
MK> so bad for several thousand rows.  Did you try with other genre_ids?

MK> Maybe a merge join is not the best choice.  Set enable_mergejoin to
MK> false and see whether you get a (hopefully faster) hash join, assuming
MK> that sort_mem is large enough to keep the hash table in memory.

MK> If you send me your table contents I'll try it on Linux.

MK> Servus
MK>  Manfred


From pgsql-performance-owner@postgresql.org  Thu Apr 29 18:27:08 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1E45AD1EF10
	for ;
	Thu, 29 Apr 2004 18:15:02 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 84879-02
	for ;
	Thu, 29 Apr 2004 18:14:42 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id D9B4AD1EEC5
	for ;
	Thu, 29 Apr 2004 18:14:39 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 247
	for ; Thu, 29 Apr 2004 22:14:10 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Thu, 29 Apr 2004 22:15:12 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question
Message-ID: <40917E70.21362.14F4A330@localhost>
In-reply-to: <200404291354.33304.josh@agliodbs.com>
References: <409172FA.1509.14C7E00E@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/506
X-Sequence-Number: 6806

On 29 Apr 2004 at 13:54, Josh Berkus wrote:

> Gary,
> 
> 
> It's also quite possble the MSSQL simply has more efficient index scanning 
> implementation that we do.    They've certainly had incentive; their storage 
> system sucks big time for random lookups and they need those fast indexes.  
> (just try to build a 1GB adjacency list tree on SQL Server.   I dare ya).
> 
> Certainly the fact that MSSQL is essentially a single-user database makes 
> things easier for them.    They don't have to maintain multiple copies of the 
> index tuples in memory.    I think that may be our main performance loss.
> 

Possibly, but MSSQL certainly uses data from indexes and cuts out the 
subsequent (possibly random seek) data fetch. This is also why the 
"Index Tuning Wizard" often recommends multi column compound 
indexes in some cases. I've tried these recommendations on occasions 
and they certainly speed up the selects significantly. If anyhing the index 
scan on the new compound index must be slower then the original single 
column index and yet it still gets the data faster.

This indicates to me that it is not the scan (or IO) performance that is 
making the difference, but not having to go get the data row.

Cheers,
Gary.


From pgsql-performance-owner@postgresql.org  Thu Apr 29 18:55:06 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A2CFFD1EB57
	for ;
	Thu, 29 Apr 2004 18:54:52 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 95162-09
	for ;
	Thu, 29 Apr 2004 18:54:34 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 9EA52D1B582
	for ;
	Thu, 29 Apr 2004 18:54:31 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3TLsWkY017176;
	Thu, 29 Apr 2004 17:54:32 -0400 (EDT)
To: "Gary Doades" 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
In-reply-to: <409172FA.1509.14C7E00E@localhost> 
References: <409153EF.24647.144E9BF7@localhost>
	<409172FA.1509.14C7E00E@localhost>
Comments: In-reply-to "Gary Doades" 
	message dated "Thu, 29 Apr 2004 21:26:18 +0100"
Date: Thu, 29 Apr 2004 17:54:31 -0400
Message-ID: <17175.1083275671@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/507
X-Sequence-Number: 6807

"Gary Doades"  writes:
> In this example the statistics don't matter.

Don't they?

A prior poster mentioned that he thought MSSQL tries to keep all its
indexes in memory.  I wonder whether you are giving Postgres a fair
chance to do the same.  What postgresql.conf settings are you using?

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 29 19:13:50 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6D64CD1EB2E
	for ;
	Thu, 29 Apr 2004 19:10:14 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 07043-01
	for ;
	Thu, 29 Apr 2004 19:09:51 -0300 (ADT)
Received: from davinci.ethosmedia.com (server228.ethosmedia.com
	[209.128.84.228])
	by svr1.postgresql.org (Postfix) with ESMTP id 5D9A8D1EB33
	for ;
	Thu, 29 Apr 2004 19:09:48 -0300 (ADT)
Received: from [64.81.245.111] (HELO 192.168.1.102)
	by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.1.8)
	with ESMTP id 4976073; Thu, 29 Apr 2004 15:11:18 -0700
From: Josh Berkus 
Reply-To: josh@agliodbs.com
Organization: Aglio Database Solutions
To: Vitaly Belman ,
	Manfred Koizar 
Subject: Re: Simply join in PostrgeSQL takes too long
Date: Thu, 29 Apr 2004 15:10:08 -0700
User-Agent: KMail/1.5.4
Cc: Rod Taylor ,
	Postgresql Performance 
References: <282909734.20040428002740@012.net.il>
	
	<147254626500.20040430000936@012.net.il>
In-Reply-To: <147254626500.20040430000936@012.net.il>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200404291510.08757.josh@agliodbs.com>
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.3 tagged_above=0.0 required=5.0
	tests=RCVD_NUMERIC_HELO
X-Spam-Level: 
X-Archive-Number: 200404/508
X-Sequence-Number: 6808

Vitaly,
 
> I am in the middle of going through them all, till now disabling the
> enable_mergejoin really helped.

In that case, your random_page_cost is probably too low.   Check the ratio of 
per-tuple times on index vs. seqscan seeks.

-- 
-Josh Berkus
 Aglio Database Solutions
 San Francisco


From pgsql-performance-owner@postgresql.org  Thu Apr 29 19:32:36 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1C650D1EDA2
	for ;
	Thu, 29 Apr 2004 19:30:51 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 14173-02
	for ;
	Thu, 29 Apr 2004 19:30:31 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id AE3B2D1EDC0
	for ;
	Thu, 29 Apr 2004 19:30:27 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 264
	for ; Thu, 29 Apr 2004 23:29:58 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Thu, 29 Apr 2004 23:31:00 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <40919034.20317.153A0B6F@localhost>
In-reply-to: <17175.1083275671@sss.pgh.pa.us>
References: <409172FA.1509.14C7E00E@localhost> 
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: Multipart/Alternative; boundary="Alt-Boundary-23407.356125551"
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=1.3 tagged_above=0.0 required=5.0 tests=HTML_40_50,
	HTML_FONTCOLOR_RED, HTML_FONTCOLOR_UNSAFE, HTML_MESSAGE,
	HTML_TITLE_EMPTY, LINES_OF_YELLING, LINES_OF_YELLING_2
X-Spam-Level: *
X-Archive-Number: 200404/509
X-Sequence-Number: 6809

--Alt-Boundary-23407.356125551
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body

On 29 Apr 2004 at 17:54, Tom Lane wrote:

> "Gary Doades"  writes:
> > In this example the statistics don't matter.
> 
> Don't they?
> 
> A prior poster mentioned that he thought MSSQL tries to keep all its
> indexes in memory.  I wonder whether you are giving Postgres a fair
> chance to do the same.  What postgresql.conf settings are you using?
> 
> 			regards, tom lane

As far as I understand it the statistics only contribute to determining the query plan. 
Once the access methods are determined, the stats don't matter during the running of 
the query.

I believe I have given Postgres exactly the same chance. The data is small enough to fit 
into RAM (all the tables in the query add up to around 50meg) and I executed the query 
several times to get a consistent figure for the explain analyze. 

Having picked out an index scan as being the highest time user I concentrated on that in 
this case and compared the same index scan on MSSQL. At least MSSQL reported it as 
an index scan on the same index for the same number of rows.

There was nothing wrong with the query plan that Postgres used. As far as I could see it 
was probably the best one to use, it just physically took longer than the same access 
plan on MSSQL.

The query and plan are included below, the main thing I was looking at was the index 
scan on staff_booking_pkey being 676ms long.

The only postgresql.conf parameters changed from the default are:

shared_buffers = 3000	
sort_mem = 4096
effective_cache_size = 15000
default_statistics_target = 100

There was no disk IO (above the small background IO) during the final run of the query 
as reported by vmstat (Task Mangler on Windows).

SELECT B.CONTRACT_ID,SUM(R.DURATION+1)/60.0 AS SUMDUR FROM 
SEARCH_REQT_RESULT TSR
JOIN STAFF_BOOKING B ON (B.STAFF_ID = TSR.STAFF_ID)
JOIN ORDER_REQT R ON (R.REQT_ID = B.REQT_ID)
JOIN BOOKING_PLAN BP ON (BP.BOOKING_ID = B.BOOKING_ID) AND 
BP.BOOKING_DATE BETWEEN '2004-04-12' AND '2004-04-18' AND 
TSR.SEARCH_ID = 8 GROUP BY B.CONTRACT_ID

QUERY PLAN
HashAggregate  (cost=11205.80..11209.81 rows=401 width=6) (actual 
time=1179.729..1179.980 rows=50 loops=1)
  ->  Nested Loop  (cost=326.47..11203.79 rows=401 width=6) (actual 
time=39.700..1177.149 rows=652 loops=1)
        ->  Hash Join  (cost=326.47..9990.37 rows=401 width=8) (actual 
time=39.537..1154.807 rows=652 loops=1)
              Hash Cond: ("outer".staff_id = "inner".staff_id)
              ->  Merge Join  (cost=320.39..9885.06 rows=3809 width=12) (actual 
time=38.316..1143.953 rows=4079 loops=1)
                    Merge Cond: ("outer".booking_id = "inner".booking_id)
                    ->  Index Scan using staff_booking_pkey on staff_booking b  
(cost=0.00..8951.94 rows=222612 width=16) (actual time=0.218..676.219 rows=222609 
loops=1)
                    ->  Sort  (cost=320.39..329.91 rows=3808 width=4) (actual 
time=26.225..32.754 rows=4079 loops=1)
                          Sort Key: bp.booking_id
                          ->  Index Scan using booking_plan_idx2 on booking_plan bp  
(cost=0.00..93.92 rows=3808 width=4) (actual time=0.223..14.186 rows=4079 loops=1)
                                Index Cond: ((booking_date >= '2004-04-12'::date) AND 
(booking_date <= '2004-04-18'::date))
              ->  Hash  (cost=5.59..5.59 rows=193 width=4) (actual time=1.139..1.139 
rows=0 loops=1)
                    ->  Index Scan using fk_idx_search_reqt_result on search_reqt_result tsr  
(cost=0.00..5.59 rows=193 width=4) (actual time=0.213..0.764 rows=192 loops=1)
                          Index Cond: (search_id = 8)
        ->  Index Scan using order_reqt_pkey on order_reqt r  (cost=0.00..3.01 rows=1 
width=6) (actual time=0.023..0.025 rows=1 loops=652)
              Index Cond: (r.reqt_id = "outer".reqt_id)
Total runtime: 1181.239 ms


Cheers,
Gary.

--Alt-Boundary-23407.356125551
Content-type: text/html; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body







On 29 Apr 2004 at 17:54, Tom Lane wrote:







> "Gary Doades" <gpd@gpdnet.co.uk> writes:


> > In this example the statistics don't matter.


> 


> Don't they?


> 


> A prior poster mentioned that he thought MSSQL tries to keep all its


> indexes in memory.  I wonder whether you are giving Postgres a fair


> chance to do the same.  What postgresql.conf settings are you using?


> 


>                                  regards, 
tom lane





As far as I understand it the statistics only contribute to determining the query plan. 
Once the access methods are determined, the stats don't matter during the running of 
the query.





I believe I have given Postgres exactly the same chance. The data is small enough to fit 
into RAM (all the tables in the query add up to around 50meg) and I executed the query 
several times to get a consistent figure for the explain analyze. 







Having picked out an index scan as being the highest time user I concentrated on that in 
this case and compared the same index scan on MSSQL. At least MSSQL reported it as 
an index scan on the same index for the same number of rows.







There was nothing wrong with the query plan that Postgres used. As far as I could see it 
was probably the best one to use, it just physically took longer than the same access 
plan on MSSQL.





The query and plan are included below, the main thing I was looking at was the index 
scan on staff_booking_pkey being 676ms long.





The only postgresql.conf parameters changed from the default are:





shared_buffers = 3000   


sort_mem = 4096


effective_cache_size = 15000


default_statistics_target = 100





There was no disk IO (above the small background IO) during the final run of the query 
as reported by vmstat (Task Mangler on Windows).







SELECT B.CONTRACT_ID,SUM(R.DURATION+1)/60.0 AS SUMDUR FROM 
SEARCH_REQT_RESULT TSR


JOIN STAFF_BOOKING B ON (B.STAFF_ID = TSR.STAFF_ID)


JOIN ORDER_REQT R ON (R.REQT_ID = B.REQT_ID)


JOIN BOOKING_PLAN BP ON (BP.BOOKING_ID = B.BOOKING_ID) AND 
BP.BOOKING_DATE BETWEEN '2004-04-12' AND '2004-04-18' AND 
TSR.SEARCH_ID = 8 GROUP BY B.CONTRACT_ID







QUERY PLAN


HashAggregate  (cost=11205.80..11209.81 rows=401 width=6) (actual 
time=1179.729..1179.980 rows=50 loops=1)


  ->  Nested Loop  (cost=326.47..11203.79 rows=401 width=6) (actual 
time=39.700..1177.149 rows=652 loops=1)


        ->  Hash Join  (cost=326.47..9990.37 rows=401 width=8) 
(actual 
time=39.537..1154.807 rows=652 loops=1)


              Hash Cond: ("outer".staff_id 
= "inner".staff_id)


              ->  Merge Join  
(cost=320.39..9885.06 rows=3809 width=12) (actual 
time=38.316..1143.953 rows=4079 loops=1)


                    
Merge Cond: ("outer".booking_id = "inner".booking_id)


                    
->  Index Scan using staff_booking_pkey on staff_booking b 
(cost=0.00..8951.94 rows=222612 width=16) (actual time=0.218..676.219 rows=222609 
loops=1)


                    
->  Sort  (cost=320.39..329.91 rows=3808 width=4) (actual 
time=26.225..32.754 rows=4079 loops=1)


                          
Sort Key: bp.booking_id


                          
->  Index Scan using booking_plan_idx2 on booking_plan bp 
(cost=0.00..93.92 rows=3808 width=4) (actual time=0.223..14.186 rows=4079 loops=1)


                                
Index Cond: ((booking_date >= '2004-04-12'::date) AND 
(booking_date <= '2004-04-18'::date))


              ->  Hash  (cost=5.59..5.59 
rows=193 width=4) (actual time=1.139..1.139 
rows=0 loops=1)


                    
->  Index Scan using fk_idx_search_reqt_result on search_reqt_result tsr 
(cost=0.00..5.59 rows=193 width=4) (actual time=0.213..0.764 rows=192 loops=1)


                          
Index Cond: (search_id = 8)


        ->  Index Scan using order_reqt_pkey on order_reqt r  
(cost=0.00..3.01 rows=1 
width=6) (actual time=0.023..0.025 rows=1 loops=652)


              Index Cond: (r.reqt_id = "outer".reqt_id)


Total runtime: 1181.239 ms












Cheers,


Gary.




--Alt-Boundary-23407.356125551--

From pgsql-performance-owner@postgresql.org  Thu Apr 29 20:01:02 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 2A5E1D1E7C0
	for ;
	Thu, 29 Apr 2004 20:01:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 17063-10
	for ;
	Thu, 29 Apr 2004 20:00:42 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 351EAD1E321
	for ;
	Thu, 29 Apr 2004 20:00:38 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 268
	for ; Fri, 30 Apr 2004 00:00:08 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Fri, 30 Apr 2004 00:01:10 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <40919746.25550.1555A9CC@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/510
X-Sequence-Number: 6810

> 
> Having picked out an index scan as being the highest time user I 
> concentrated on that in  this case and compared the same index scan on 
> MSSQL. At least MSSQL reported it as  an index scan on the same index 
> for the same number of rows. 
> 

I should have also pointed out that MSSQL reported that same index scan as taking 65% of the overall query time.
It was just "faster". The overall query took 103ms in MSSQL.

Gary.


From pgsql-performance-owner@postgresql.org  Thu Apr 29 20:15:31 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 51058D1E155
	for ;
	Thu, 29 Apr 2004 20:15:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 26790-05
	for ;
	Thu, 29 Apr 2004 20:15:07 -0300 (ADT)
Received: from news.hub.org (news.hub.org [200.46.204.72])
	by svr1.postgresql.org (Postfix) with ESMTP id DD98ED1E607
	for ;
	Thu, 29 Apr 2004 20:15:04 -0300 (ADT)
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 i3TNF3T3050408
	for ; Thu, 29 Apr 2004 23:15:04 GMT
	(envelope-from news@news.hub.org)
Received: (from news@localhost)
	by news.hub.org (8.12.9/8.12.9/Submit) id i3TN8MwP045235
	for pgsql-performance@postgresql.org; Thu, 29 Apr 2004 23:08:22 GMT
From: Joseph Shraibman 
X-Newsgroups: comp.databases.postgresql.performance
Subject: Insert only tables and vacuum performance
Date: Thu, 29 Apr 2004 19:08:23 -0400
Organization: Hub.Org Networking Services
Lines: 7
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 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
To: pgsql-performance@postgresql.org
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/512
X-Sequence-Number: 6812

I have a table that is never updated, only INSERTED into. Is there a way 
I can prevent vacuum wasting time on this table besides vacuuming each 
table in the db by itself and omitting this table?

How feasable would it be to have a marker somewhere in pg that is 
"updated since last vacuum" that would be cleared when vacuum runs, and 
if set vacuum will ignore that table?

From pgsql-performance-owner@postgresql.org  Thu Apr 29 20:15:29 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 4C982D1C50D
	for ;
	Thu, 29 Apr 2004 20:15:25 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 29982-01
	for ;
	Thu, 29 Apr 2004 20:15:07 -0300 (ADT)
Received: from news.hub.org (news.hub.org [200.46.204.72])
	by svr1.postgresql.org (Postfix) with ESMTP id B0AFDD1E5FC
	for ;
	Thu, 29 Apr 2004 20:15:04 -0300 (ADT)
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 i3TNF3T1050408
	for ; Thu, 29 Apr 2004 23:15:04 GMT
	(envelope-from news@news.hub.org)
Received: (from news@localhost)
	by news.hub.org (8.12.9/8.12.9/Submit) id i3TN954J045921
	for pgsql-performance@postgresql.org; Thu, 29 Apr 2004 23:09:05 GMT
From: Joseph Shraibman 
X-Newsgroups: comp.databases.postgresql.performance
Subject: analyzer/planner and clustered rows
Date: Thu, 29 Apr 2004 19:09:09 -0400
Organization: Hub.Org Networking Services
Lines: 4
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 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
To: pgsql-performance@postgresql.org
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/511
X-Sequence-Number: 6811

How does the analyzer/planner deal with rows clustered together?  Does 
it just assume that if this col is clustered on then the actual data 
will be clustered?  What if the data in the table happens to be close 
together because it was inserted together originally?

From pgsql-performance-owner@postgresql.org  Thu Apr 29 20:18:15 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 5D5F9D1E73F
	for ;
	Thu, 29 Apr 2004 20:18:11 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 29982-02
	for ;
	Thu, 29 Apr 2004 20:17:53 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 5FE8AD1B43A
	for ;
	Thu, 29 Apr 2004 20:17:50 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3TNHoSJ017899;
	Thu, 29 Apr 2004 19:17:50 -0400 (EDT)
To: josh@agliodbs.com
Cc: "Gary Doades" ,
	pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
In-reply-to: <200404291354.33304.josh@agliodbs.com> 
References: <409153EF.24647.144E9BF7@localhost>
	<409172FA.1509.14C7E00E@localhost>
	<200404291354.33304.josh@agliodbs.com>
Comments: In-reply-to Josh Berkus 
	message dated "Thu, 29 Apr 2004 13:54:33 -0700"
Date: Thu, 29 Apr 2004 19:17:49 -0400
Message-ID: <17898.1083280669@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/513
X-Sequence-Number: 6813

Josh Berkus  writes:
> Certainly the fact that MSSQL is essentially a single-user database makes 
> things easier for them.

Our recent testing (cf the "Xeon" thread) says that the interlocking we
do to make the world safe for multiple backends has a fairly high cost
(at least on some hardware) compared to the rest of the work in
scenarios where you are doing zero-I/O scans of data already in memory.
Especially so for index scans.  I'm not sure this completely explains
the differential that Gary is complaining about, but it could be part of
it.  Is it really true that MSSQL doesn't support concurrent operations?

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Thu Apr 29 20:45:28 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 10706D1E2A0
	for ;
	Thu, 29 Apr 2004 20:45:24 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 33783-10
	for ;
	Thu, 29 Apr 2004 20:45:05 -0300 (ADT)
Received: from news.hub.org (news.hub.org [200.46.204.72])
	by svr1.postgresql.org (Postfix) with ESMTP id D449BD1E31C
	for ;
	Thu, 29 Apr 2004 20:45:02 -0300 (ADT)
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 i3TNj2Sx097842
	for ; Thu, 29 Apr 2004 23:45:02 GMT
	(envelope-from news@news.hub.org)
Received: (from news@localhost)
	by news.hub.org (8.12.9/8.12.9/Submit) id i3TNJ7rt054355
	for pgsql-performance@postgresql.org; Thu, 29 Apr 2004 23:19:07 GMT
From: Joseph Shraibman 
X-Newsgroups: comp.databases.postgresql.performance
Subject: why can't 2 indexes be used at once?
Date: Thu, 29 Apr 2004 19:19:09 -0400
Organization: Hub.Org Networking Services
Lines: 12
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 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
To: pgsql-performance@postgresql.org
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/515
X-Sequence-Number: 6815

Lets say I have two columns, A and B.  They are each indexed seperately. 
  If I do a query like:
SELECT * FROM table WHERE A = 1 AND B = 2;
postgres can only use one index.

I assume that postgres uses the index data to narrow down pages in the 
table to visit when doing its search. Then it goes through and filters 
on the second condition.

My question: why can't it go through the first index, get a list of 
pages in the table, then go through the second index, union the result 
with the results from first index, and then go into the table?

From pgsql-performance-owner@postgresql.org  Thu Apr 29 20:45:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E325AD1CAA7
	for ;
	Thu, 29 Apr 2004 20:45:23 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 39487-02
	for ;
	Thu, 29 Apr 2004 20:45:05 -0300 (ADT)
Received: from news.hub.org (news.hub.org [200.46.204.72])
	by svr1.postgresql.org (Postfix) with ESMTP id D4F8DD1E350
	for ;
	Thu, 29 Apr 2004 20:45:02 -0300 (ADT)
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 i3TNj2T1097842
	for ; Thu, 29 Apr 2004 23:45:02 GMT
	(envelope-from news@news.hub.org)
Received: (from news@localhost)
	by news.hub.org (8.12.9/8.12.9/Submit) id i3TNObQK063246
	for pgsql-performance@postgresql.org; Thu, 29 Apr 2004 23:24:37 GMT
From: Joseph Shraibman 
X-Newsgroups: comp.databases.postgresql.performance
Subject: Re: Insert only tables and vacuum performance
Date: Thu, 29 Apr 2004 19:24:40 -0400
Organization: Hub.Org Networking Services
Lines: 13
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 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
In-Reply-To: 
To: pgsql-performance@postgresql.org
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/514
X-Sequence-Number: 6814

Joseph Shraibman wrote:
> I have a table that is never updated, only INSERTED into. Is there a way 
> I can prevent vacuum wasting time on this table besides vacuuming each 
> table in the db by itself and omitting this table?
> 
> How feasable would it be to have a marker somewhere in pg that is 
> "updated since last vacuum" that would be cleared when vacuum runs, and 
> if set vacuum will ignore that table?

Or even better an offset into the datatable for the earliest deleted 
row, so if you have a table where you update the row shortly after 
insert and then never touch it vacuum can skip most of the table 
(inserts are done at the end of the table, right?)

From pgsql-performance-owner@postgresql.org  Thu Apr 29 21:45:23 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 17F9ED1DF83
	for ;
	Thu, 29 Apr 2004 21:45:21 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 54734-07
	for ;
	Thu, 29 Apr 2004 21:45:02 -0300 (ADT)
Received: from tht.net (vista.tht.net [216.126.88.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 59532D1B4E0
	for ;
	Thu, 29 Apr 2004 21:44:58 -0300 (ADT)
Received: from [134.22.68.249] (dyn-68-249.tor.dsl.tht.net [134.22.68.249])
	by tht.net (Postfix) with ESMTP
	id DB30276A69; Thu, 29 Apr 2004 20:45:03 -0400 (EDT)
Subject: Re: Insert only tables and vacuum performance
From: Rod Taylor 
To: Joseph Shraibman 
Cc: Postgresql Performance 
In-Reply-To: 
References:   
Content-Type: text/plain
Message-Id: <1083285855.30065.462.camel@jester>
Mime-Version: 1.0
X-Mailer: Ximian Evolution 1.4.5 
Date: Thu, 29 Apr 2004 20:44:16 -0400
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=
X-Spam-Level: 
X-Archive-Number: 200404/516
X-Sequence-Number: 6816

> Or even better an offset into the datatable for the earliest deleted 
> row, so if you have a table where you update the row shortly after 
> insert and then never touch it vacuum can skip most of the table 
> (inserts are done at the end of the table, right?)

Inserts are done at the end of the table as a last resort. But anyway,
how do you handle a rolled back insert?



From pgsql-performance-owner@postgresql.org  Thu Apr 29 21:53:46 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 94F36D1DF83
	for ;
	Thu, 29 Apr 2004 21:53:44 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 57460-07
	for ;
	Thu, 29 Apr 2004 21:53:22 -0300 (ADT)
Received: from tupari.net (node-40242ac2.lga.onnet.us.uu.net [64.36.42.194])
	by svr1.postgresql.org (Postfix) with ESMTP id A7AA0D1CAA9
	for ;
	Thu, 29 Apr 2004 21:53:18 -0300 (ADT)
Received: from selectacast.net ([192.168.0.2])
	by tupari.net (8.12.8/8.12.8) with ESMTP id i3U0rLCE028271;
	Thu, 29 Apr 2004 20:53:22 -0400
Message-ID: <4091A381.3020307@selectacast.net>
Date: Thu, 29 Apr 2004 20:53:21 -0400
From: Joseph Shraibman 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Rod Taylor 
Cc: Postgresql Performance 
Subject: Re: Insert only tables and vacuum performance
References:  
	<1083285855.30065.462.camel@jester>
In-Reply-To: <1083285855.30065.462.camel@jester>
Content-Type: text/plain; charset=us-ascii; format=flowed
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=
X-Spam-Level: 
X-Archive-Number: 200404/517
X-Sequence-Number: 6817

Rod Taylor wrote:
>>Or even better an offset into the datatable for the earliest deleted 
>>row, so if you have a table where you update the row shortly after 
>>insert and then never touch it vacuum can skip most of the table 
>>(inserts are done at the end of the table, right?)
> 
> 
> Inserts are done at the end of the table as a last resort.

But if most of the table is never updated then the inserts would tend to 
be at the end, right?

> But anyway,
> how do you handle a rolled back insert?
> 
It is considered like a deleted row to be vacuumed.

From pgsql-performance-owner@postgresql.org  Fri Apr 30 00:58:12 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id E1396D1CA4D
	for ;
	Fri, 30 Apr 2004 00:58:10 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 14374-03
	for ;
	Fri, 30 Apr 2004 00:57:50 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 2C5A6D1E5F1
	for ;
	Fri, 30 Apr 2004 00:57:50 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3U3vni0020260;
	Thu, 29 Apr 2004 23:57:50 -0400 (EDT)
To: Joseph Shraibman 
Cc: pgsql-performance@postgresql.org
Subject: Re: why can't 2 indexes be used at once? 
In-reply-to:  
References: 
Comments: In-reply-to Joseph Shraibman 
	message dated "Thu, 29 Apr 2004 19:19:09 -0400"
Date: Thu, 29 Apr 2004 23:57:49 -0400
Message-ID: <20259.1083297469@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/518
X-Sequence-Number: 6818

Joseph Shraibman  writes:
> My question: why can't it go through the first index, get a list of 
> pages in the table, then go through the second index, union the result 
> with the results from first index, and then go into the table?

See TODO list ...

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Fri Apr 30 01:30:49 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 1E662D1E5EE
	for ;
	Fri, 30 Apr 2004 01:30:48 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 21470-07
	for ;
	Fri, 30 Apr 2004 01:30:28 -0300 (ADT)
Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130])
	by svr1.postgresql.org (Postfix) with ESMTP id 2B52ED1DC9D
	for ;
	Fri, 30 Apr 2004 01:30:27 -0300 (ADT)
Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1])
	by sss.pgh.pa.us (8.12.11/8.12.11) with ESMTP id i3U4URcg020620;
	Fri, 30 Apr 2004 00:30:27 -0400 (EDT)
To: Joseph Shraibman 
Cc: pgsql-performance@postgresql.org
Subject: Re: Insert only tables and vacuum performance 
In-reply-to:  
References: 
Comments: In-reply-to Joseph Shraibman 
	message dated "Thu, 29 Apr 2004 19:08:23 -0400"
Date: Fri, 30 Apr 2004 00:30:26 -0400
Message-ID: <20619.1083299426@sss.pgh.pa.us>
From: Tom Lane 
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/519
X-Sequence-Number: 6819

Joseph Shraibman  writes:
> I have a table that is never updated, only INSERTED into. Is there a way 
> I can prevent vacuum wasting time on this table

What makes you think vacuum is wasting much time on this table?  AFAICS
it will only update any unfixed hint bits ...

			regards, tom lane

From pgsql-performance-owner@postgresql.org  Fri Apr 30 02:27:21 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D9C6DD1DAC4
	for ;
	Fri, 30 Apr 2004 02:27:20 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 38484-07
	for ;
	Fri, 30 Apr 2004 02:27:01 -0300 (ADT)
Received: from zigo.dhs.org (as2-4-3.an.g.bonet.se [194.236.34.191])
	by svr1.postgresql.org (Postfix) with ESMTP id F0DDFD1DF8B
	for ;
	Fri, 30 Apr 2004 02:26:59 -0300 (ADT)
Received: from zigo.zigo.dhs.org (zigo.zigo.dhs.org [192.168.0.1])
	by zigo.dhs.org (Postfix) with ESMTP
	id 6CCB780DB; Fri, 30 Apr 2004 07:26:59 +0200 (CEST)
Date: Fri, 30 Apr 2004 07:26:59 +0200 (CEST)
From: Dennis Bjorklund 
To: Gary Doades 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
In-Reply-To: <40919746.25550.1555A9CC@localhost>
Message-ID: 
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=ISO-8859-1
Content-Transfer-Encoding: 8BIT
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/520
X-Sequence-Number: 6820

On Fri, 30 Apr 2004, Gary Doades wrote:

> I should have also pointed out that MSSQL reported that same index scan
> as taking 65% of the overall query time. It was just "faster". The
> overall query took 103ms in MSSQL.

Are your results based on a single client accessing the database and no 
concurrent updates?

Would adding more clients, and maybe having some client that
updates/inserts into the tables, still make mssql faster then pg? Maybe
it's so simple as pg being optimized for more concurrent users then mssql?

I'm just asking, I don't know much about the inner workings of 
mssql.

-- 
/Dennis Bj�rklund


From pgsql-sql-owner@postgresql.org  Fri Apr 30 03:45:58 2004
X-Original-To: pgsql-sql-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A57FED1DAC4
	for ;
	Fri, 30 Apr 2004 03:45:57 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 57814-09
	for ;
	Fri, 30 Apr 2004 03:45:37 -0300 (ADT)
Received: from matrix.gatewaynet.com (matrix.gatewaynet.com [217.19.69.50])
	by svr1.postgresql.org (Postfix) with ESMTP id CCCE6D1CA39
	for ; Fri, 30 Apr 2004 03:45:34 -0300 (ADT)
Received: from matrix.gatewaynet.com (localhost.localdomain [127.0.0.1])
	by matrix.gatewaynet.com (8.12.8/8.12.8) with ESMTP id i3U5ngFr002040;
	Fri, 30 Apr 2004 08:49:42 +0300
Received: from localhost (achill@localhost)
	by matrix.gatewaynet.com (8.12.8/8.12.8/Submit) with ESMTP id
	i3U5nfJV002036; Fri, 30 Apr 2004 08:49:41 +0300
Date: Fri, 30 Apr 2004 08:49:41 +0300 (EEST)
From: Achilleus Mantzios 
To: Josh Berkus 
Cc: Roelant Ossewaarde , 
Subject: Re: Use arrays or not?
In-Reply-To: <200404291023.58712.josh@agliodbs.com>
Message-ID: 
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/331
X-Sequence-Number: 17706


Just some comments from my experience:

PgSQL Arrays are mainly for modeling arrays not relations.
Arrays can be very useful if they are not ever gonna be treated
as relations and if performance is an issue (e.g. dynamic graphs, on the 
fly statistics etc..).

Also (besides other solutions) int[] arrays is a handy way of implementing
tree structures in a database.

For your case as you set it the XML arrays will make your life difficult.

O kyrios Josh Berkus egrapse stis Apr 29, 2004 :

> Roelant,
> 
> Yours is not a performance question, so I'm crossing it over to SQL for advice 
> on database design.
> 
> > I am building an application using postgresql to store XML-records. There
> > is a debate within the group of developers about the best way to store our
> > data. I hope you can help us make a decision.
> >
> > The data consists of XML-records, with a lot of XML-fields. I want to store
> > the XML as it is, so taking the information from the XML-records and then
> > storing it in a different-from-XML-format is not an option.
> >
> > Each XML-record describes data about one book. If an update of bookdata
> > comes, the XML itself is not changed, but a new XML-record is stored with
> > the updated data. Via a complex scheme of combining a base record and its
> > updates, the final dataset is produced that is used in the application.
> >
> > There are different XML-formats that need to be combined. Right now, we can
> > handle three different XML-formats, each with its own structure (but all
> > describing book-data).
> >
> > Searching is done via a simple table lookup on three different fields:
> > title, author and subject. The data for these fields is extracted from the
> > database. Each book has a unique identifier (EAN13, derivative of ISBN).
> >
> > Here is one way to organize the database:
> > table title:
> > TITLE | EAN13, indexing on TITLE
> >
> > table author:
> > AUTHOR | EAN13, indexing on AUTHOR
> >
> > table subject:
> > SUBJECT | EAN13, indexing on SUBJECT.
> 
> This is a *very* strange way of setting up your database.  Are you new to 
> Relational Databases and SQL?   If so, I'd recommend starting with a book on 
> relational database design.
> 
> Either that, or you're a victim of UML design.   
> 
> If only one author, title and subject are allowed per book, you should have:
> 
> table books
> 	EAN13 | TITLE  | AUTHOR | SUBJECT
> 
> > Finally:
> > table record:
> > EAN13 | ARRAY OF XML-records.
> >
> > It's the last table that I am most curious (and worried) about, the
> > question being mainly what the optimal way of structuring that table is.
> > Option 1 is the given option: adding/deleting an XML-record for the same
> > book requires adding/deleting it to/from the array of XML-records.
> >
> > Option 2 would be something like this:
> > EAN13 | XML-record
> > where, if a book has several records describing it, there are multiple
> > entries of the EAN13|XML-record - pair. Adding an XML-record for the same
> > book, requires adding a new entry to the table as a whole.
> 
> In my mind, there is no question that this is the best way to do things.  It 
> is a normalized data structure, as opposed to the arrays, which are now.
> 
> >
> > So, option 1-tables look like this:
> > EAN13 | ARRAY OF XML-records
> > 0001 | {..., ..., ...}
> > 0002 | {..., ..., ...}
> >
> > Option-2 tables look like this:
> > EAN13 | ARRAY OF XML-records
> > 0001 | ...
> > 0001 | ...
> > 0002 | ...
> > 0002 | ...
> >
> > We can't decide which one is best. These are some issues we can think of:
> >
> > Indexing: For option 1, the EAN13-index remains unique, even if you have
> > multiple XML-records; for option 2 it does not, since multiple XML-records
> > are stored as multiple tuples. On the other hand, an additional internal
> > index can be used to link the several tuples of option 2 to the information
> > in the `lookup'-tables (author, title, keyword). Does any of these two
> > options increase query efficiency, ie. speed?
> >
> > Database growth: On average, the information about a book is updated three
> > times per year. In option 1, this means that the length of the table does
> > not increase, but the width does. If we choose option 2, if we have three
> > updates per book each year, the length of the table triples, but the width
> > does not. What is more costly to store for postgres, long arrays or long
> > tables?
> >
> > Integrity: Option 1 means that our software needs to keep track of all the
> > bookkeeping for arrays, since such support is quite rudimentary in
> > postgres. For example, it is hard to take out a record from the middle of
> > an array. Also, a multidimensional array, which contains for each record
> > the record itself and its type, is even harder to maintain. Option 2 has a
> > simpler datatype, so integrity can be easier inforced using the standard
> > postgres-machinery of variable-types etc.
> >
> > Arrays are non-standard SQL, and I hear that PHP-support for postgres &
> > arrays is rudimentary. So that might be an argument to avoid using them,
> > and go for option 2. From the standpoint of performance (or wisdom), can
> > you help me decide what I should choose? Or is there maybe an even better
> > way to structure my data?
> >
> > Thanks for any contribution!
> >
> > Roelant.
> 
> 
> 

-- 
-Achilleus


From pgsql-performance-owner@postgresql.org  Fri Apr 30 02:58:08 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 79E21D1DC64
	for ;
	Fri, 30 Apr 2004 02:58:07 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 50644-01
	for ;
	Fri, 30 Apr 2004 02:57:46 -0300 (ADT)
Received: from tupari.net (node-40242ac2.lga.onnet.us.uu.net [64.36.42.194])
	by svr1.postgresql.org (Postfix) with ESMTP id 3C839D1E60E
	for ;
	Fri, 30 Apr 2004 02:57:45 -0300 (ADT)
Received: from selectacast.net ([192.168.0.2])
	by tupari.net (8.12.8/8.12.8) with ESMTP id i3U5viCE028467;
	Fri, 30 Apr 2004 01:57:45 -0400
Message-ID: <4091EAD8.4020503@selectacast.net>
Date: Fri, 30 Apr 2004 01:57:44 -0400
From: Joseph Shraibman 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040421
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: Tom Lane , pgsql-performance@postgresql.org
Subject: Re: Insert only tables and vacuum performance
References:  <20619.1083299426@sss.pgh.pa.us>
In-Reply-To: <20619.1083299426@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-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/521
X-Sequence-Number: 6821

Tom Lane wrote:
> Joseph Shraibman  writes:
> 
>>I have a table that is never updated, only INSERTED into. Is there a way 
>>I can prevent vacuum wasting time on this table
> 
> 
> What makes you think vacuum is wasting much time on this table?  AFAICS
> it will only update any unfixed hint bits ...
> 
> 			regards, tom lane

INFO:  "elog": found 0 removable, 12869411 nonremovable row versions in 
196195 pages
DETAIL:  0 dead row versions cannot be removed yet.
There were 5 unused item pointers.
0 pages are entirely empty.
CPU 31.61s/4.53u sec elapsed 1096.83 sec.

It took 1096.83 seconds, and what did it accomplish?  And what are hint 
bits?

From pgsql-performance-owner@postgresql.org  Fri Apr 30 03:34:47 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id D79ABD1CAB1
	for ;
	Fri, 30 Apr 2004 03:34:45 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 55501-06
	for ;
	Fri, 30 Apr 2004 03:34:26 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id EBC7DD1CA4F
	for ;
	Fri, 30 Apr 2004 03:34:23 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 326
	for ; Fri, 30 Apr 2004 07:33:52 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Fri, 30 Apr 2004 07:33:11 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <40920137.13806.16F37CC5@localhost>
In-reply-to: <17898.1083280669@sss.pgh.pa.us>
References: <200404291354.33304.josh@agliodbs.com> 
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/522
X-Sequence-Number: 6822

On 29 Apr 2004 at 19:17, Tom Lane wrote:

> Josh Berkus  writes:
> > Certainly the fact that MSSQL is essentially a single-user database makes 
> > things easier for them.
> 
> Our recent testing (cf the "Xeon" thread) says that the interlocking we
> do to make the world safe for multiple backends has a fairly high cost
> (at least on some hardware) compared to the rest of the work in
> scenarios where you are doing zero-I/O scans of data already in memory.
> Especially so for index scans.  I'm not sure this completely explains
> the differential that Gary is complaining about, but it could be part of
> it.  Is it really true that MSSQL doesn't support concurrent operations?
> 
> 			regards, tom lane

As far as I am aware SQLSever supports concurrent operations. It 
certainly creates more threads for each connection. None of my 
observations of the system under load (50 ish concurrent users, 150 ish 
connections) suggest that it is serializing queries.

These tests are currentl on single processor Athlon XP 2000+ systems.

Regards,
Gary.

From pgsql-performance-owner@postgresql.org  Fri Apr 30 04:01:15 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 4E382D1E5C5
	for ;
	Fri, 30 Apr 2004 04:01:14 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 66568-02
	for ;
	Fri, 30 Apr 2004 04:00:55 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 964D7D1E3DF
	for ;
	Fri, 30 Apr 2004 04:00:52 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 330
	for ; Fri, 30 Apr 2004 08:00:22 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Fri, 30 Apr 2004 08:01:26 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <409207D6.17204.170D5A52@localhost>
In-reply-to: 
References: <40919746.25550.1555A9CC@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=ISO-8859-1
Content-transfer-encoding: Quoted-printable
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/523
X-Sequence-Number: 6823

On 30 Apr 2004 at 7:26, Dennis Bjorklund wrote:

> On Fri, 30 Apr 2004, Gary Doades wrote:
>
> > I should have also pointed out that MSSQL reported that same index scan
> > as taking 65% of the overall query time. It was just "faster". The
> > overall query took 103ms in MSSQL.
>
> Are your results based on a single client accessing the database and no
> concurrent updates?
>
> Would adding more clients, and maybe having some client that
> updates/inserts into the tables, still make mssql faster then pg? Maybe
> it's so simple as pg being optimized for more concurrent users then mssql?
>
> I'm just asking, I don't know much about the inner workings of
> mssql.
>
> --
> /Dennis Bj=F6rklund
>

At the moment it is difficult to set up many clients for testing concurrent
stuff. In the past I have had several SQLServer clients under test,
mainly select queries. MSSQL can certainly execute queries while other
queries are still running in the background.

Our production app is fairly well biased towards selects. Currently it is
about 70% selects, 20% inserts, 6% deletes and 4% updates. Very few
updates are more than one row based on the primary key. Over 90% of
the time spend running SQL is in select queries.

My limited concurrent testing on Postgres gives very good performance
on updates, inserts, deletes, but it is suffering on the selects in certain
areas which why I have been concentrating my efforts on that area.

Having got similar (or the same) access plans in both Postgres and
MSSQL I was getting down to the next level of checking what was going
on when executing the already planned query.

I do have another database system I could try. Sybase SQLAnywhere.
This is not the original Sybase Entrerprise which has the same roots as
MSSQL. In the past my testing suggested that SQLAnywhere
performance was as godd or better than MSSQL. I mey try to set it up
with the same data in these tests for a more detailed comparison.

Regards,
Gary.


From pgsql-performance-owner@postgresql.org  Fri Apr 30 04:31:00 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id C4345D1E6B4
	for ;
	Fri, 30 Apr 2004 04:30:59 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 74171-05
	for ;
	Fri, 30 Apr 2004 04:30:39 -0300 (ADT)
Received: from email10.aon.at (WARSL402PIP8.highway.telekom.at [195.3.96.97])
	by svr1.postgresql.org (Postfix) with SMTP id 9A8E9D1E5D3
	for ;
	Fri, 30 Apr 2004 04:30:37 -0300 (ADT)
Received: (qmail 227216 invoked from network); 30 Apr 2004 07:30:38 -0000
Received: from m154p027.dipool.highway.telekom.at (HELO PASCAL) ([62.46.9.59])
	(envelope-sender )
	by qmail1rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 30 Apr 2004 07:30:38 -0000
From: Manfred Koizar 
To: Joseph Shraibman 
Cc: pgsql-performance@postgresql.org
Subject: Re: analyzer/planner and clustered rows
Date: Fri, 30 Apr 2004 09:33:13 +0200
Message-ID: 
References: 
In-Reply-To: 
X-Mailer: Forte Agent 2.0/32.652
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=4.5 tagged_above=0.0 required=5.0 tests=RCVD_IN_DSBL,
	RCVD_IN_SORBS, RCVD_IN_SORBS_HTTP, RCVD_IN_SORBS_MISC,
	RCVD_IN_SORBS_SOCKS
X-Spam-Level: ****
X-Archive-Number: 200404/524
X-Sequence-Number: 6824

On Thu, 29 Apr 2004 19:09:09 -0400, Joseph Shraibman
 wrote:
>How does the analyzer/planner deal with rows clustered together?

There's a correlation value per column.  Just try

	SELECT attname, correlation
	  FROM pg_stats
	 WHERE tablename = '...';

if you are interested.  It indicates how well the hypothetical order of
tuples if sorted by that column corresponds to the physical order.  +1.0
is perfect correlation, 0.0 is totally chaotic, -1.0 means reverse
order.  The optimizer is more willing to choose an index scan if
correlation for the first index column is near +/-1.

>  What if the data in the table happens to be close 
>together because it was inserted together originally?

Having equal values close to each other is not enough, the values should
be increasing, too.  Compare

	5 5 5 4 4 4 7 7 7 2 2 2 6 6 6 3 3 3 8 8 8   low correlation
and
	2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 8 8 8   correlation = 1.0


From pgsql-performance-owner@postgresql.org  Fri Apr 30 09:32:50 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id DBD13D1E74F
	for ;
	Fri, 30 Apr 2004 09:32:48 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 65805-10
	for ;
	Fri, 30 Apr 2004 09:32:30 -0300 (ADT)
Received: from jefftrout.com (h00a0cc4084e5.ne.client2.attbi.com
	[24.128.241.68])
	by svr1.postgresql.org (Postfix) with SMTP id 6BEBED1D070
	for ;
	Fri, 30 Apr 2004 09:32:26 -0300 (ADT)
Received: (qmail 35520 invoked from network); 30 Apr 2004 12:32:19 -0000
Received: from localhost (HELO ?10.124.7.125?) (threshar@127.0.0.1)
	by localhost with SMTP; 30 Apr 2004 12:32:19 -0000
In-Reply-To: <409207D6.17204.170D5A52@localhost>
References: <40919746.25550.1555A9CC@localhost>
	<409207D6.17204.170D5A52@localhost>
Mime-Version: 1.0 (Apple Message framework v613)
Content-Type: text/plain; charset=US-ASCII; format=flowed
Message-Id: <69EF9EF2-9AA2-11D8-A4E1-000D9366F0C4@torgo.978.org>
Content-Transfer-Encoding: 7bit
Cc: pgsql-performance@postgresql.org
From: Jeff 
Subject: Re: planner/optimizer question 
Date: Fri, 30 Apr 2004 08:32:16 -0400
To: "Gary Doades" 
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=2.6 tagged_above=0.0 required=5.0
	tests=RCVD_IN_DYNABLOCK, RCVD_IN_SORBS
X-Spam-Level: **
X-Archive-Number: 200404/525
X-Sequence-Number: 6825


On Apr 30, 2004, at 3:01 AM, Gary Doades wrote:

[ pg query plan, etc ]

I wonder if other parts of the plan are affecting the speed.

I've recently run into a case where a merge join plan was chosen for 
this query, which took 11 seconds to execute.  Forcing it to pick a 
nested loop join dropped it to 3.  (Updating my 
default_statistics_target to 500 caused the planner to choose nested 
loop join)

So, is the plan really the same?

  A better comparision query may be a simple "select a from mytable 
where a between foo and bar"  to get an index scan.  In that case its a 
straight up, vanilla index scan.  Nothing else getting in the way.

--
Jeff Trout 
http://www.jefftrout.com/
http://www.stuarthamm.net/


From pgsql-performance-owner@postgresql.org  Tue May  4 01:24:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 8D10CD1E153
	for ;
	Fri, 30 Apr 2004 09:46:21 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 73309-05
	for ;
	Fri, 30 Apr 2004 09:46:03 -0300 (ADT)
Received: from mwinf0102.wanadoo.fr (smtp1.wanadoo.fr [193.252.22.30])
	by svr1.postgresql.org (Postfix) with ESMTP id CEFDCD1E74D
	for ;
	Fri, 30 Apr 2004 09:45:58 -0300 (ADT)
Received: from [192.168.0.2] (AMontsouris-108-1-33-27.w81-49.abo.wanadoo.fr
	[81.49.55.27])
	by mwinf0102.wanadoo.fr (SMTP Server) with ESMTP id 8E8DC1BE7A1E
	for ;
	Fri, 30 Apr 2004 14:45:58 +0200 (CEST)
Mime-Version: 1.0 (Apple Message framework v613)
In-Reply-To: <40615443.4030003@paradise.net.nz>
References: <200403231248.34425.darcy@wavefire.com>
	<40615443.4030003@paradise.net.nz>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Message-Id: <52623D23-9AA4-11D8-9268-000A95DE2550@gdr-isis.enst.fr>
Content-Transfer-Encoding: quoted-printable
From: =?ISO-8859-1?Q?Pailloncy_Jean-G=E9rard?= 
Subject: Re: Fwd: FreeBSD, PostgreSQL, semwait and sbwait!
Date: Fri, 30 Apr 2004 14:45:55 +0200
To: pgsql-performance@postgresql.org
X-Mailer: Apple Mail (2.613)
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200405/19
X-Sequence-Number: 6850

Hello,

>> We're having a substantial problem with our FreeBSD 5.2 database=20
>> server
>> running PostgreSQL - it's getting a lot of traffic (figure about 3,000
>> queries per second), but queries are slow, and it's seemingly waiting=20
>> on
>> other things than CPU time
>>
> Could this be a 5.2 performance issue ?
>
> In spite of certain areas where the 5.x series performance is known to=20
> be much better than 4.x (e.g networking), this may not be manifested=20
> in practice for a complete application.
> (e.g. I am still running 4.9 as it outperformed 5.1 vastly for a ~100=20
> database sessions running queries - note that I have not tried out=20
> 5.2, so am happy to be corrected on this)
I found the same problem.

I use OpenBSD 3.3,
On Pentium 2,4 GHz with 1 Gb RAM, RAID 10.
With PostgreSQL 7.4.1 with 32 Kb bock's size (to match ffs and raid=20
block's size)
With pg_autovacuum daemon from Pg 7.5.

I run a web indexer.
sd0 raid-1 with system pg-log and indexer-log
sd1 raid-10 with pg-data and indexer-data
The sd1 disk achives between 10 and 40 Mb/s on normal operation.

When I get semwait in top, system waits ;-)
Not much disk activity.
Not much log in pg or indexer.
Just wait....

What can I do ?

 > sudo top  -s1 -S -I
load averages:  4.45,  4.45,  3.86=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
       11:25:52
97 processes:  1 running, 96 idle
CPU states:  2.3% user,  0.0% nice,  3.8% system,  0.8% interrupt,=20
93.1% idle
Memory: Real: 473M/803M act/tot  Free: 201M  Swap: 0K/3953M used/tot

   PID USERNAME PRI NICE  SIZE   RES STATE WAIT     TIME    CPU COMMAND
  2143 postgres  -5    0 4008K   37M sleep biowai   1:02  1.81% postgres
28662 postgres  14    0 4060K   37M sleep semwai   0:59  1.17% postgres
25794 postgres  14    0 4072K   37M sleep semwai   1:30  0.93% postgres
23271 postgres  -5    0 4060K   37M sleep biowai   1:13  0.29% postgres
14619 root      28    0  276K  844K run   -        0:01  0.00% top

 > vmstat -w1 sd0 sd1
  r b w    avm    fre   flt  re  pi  po  fr  sr sd0 sd1   in    sy   cs=20
us sy id
  0 4 0 527412  36288  1850   0   0   0   0   0  26  72  368  8190  588=20=
=20
0  4 96
  0 4 0 527420  36288  1856   0   0   0   0   0   0  86  356  8653  620=20=
=20
2  2 97
  0 4 0 527432  36280  1853   0   0   0   0   0   0  54  321  8318  458=20=
=20
1  3 96
  0 4 0 527436  36248  1864   0   0   0   0   0   0  77  358  8417  539=20=
=20
1  2 97
  0 4 0 522828  40932  2133   0   0   0   0   0   7  70  412 15665  724=20=
=20
2  3 95
  0 4 0 522896  40872  1891   0   0   0   0   0  15  72  340  9656  727=20=
=20
3  5 92
  0 4 0 522900  40872  1841   0   0   0   0   0   0  69  322  8308  536=20=
=20
1  2 98
  0 4 0 522920  40860  1846   0   0   0   0   0   1  69  327  8023  520=20=
=20
2  2 97
  0 4 0 522944  40848  1849   0   0   0   0   0   4  76  336  8035  567=20=
=20
1  2 97
  0 4 0 522960  40848  1843   0   0   0   0   0   0  77  331 14669  587=20=
=20
3  2 95
  0 4 0 522976  40836  1848   0   0   0   0   0   4  81  339  8384  581=20=
=20
1  2 97
  0 4 0 522980  40836  1841   0   0   0   0   0   3  65  320  8068  502=20=
=20
1  4 95
  0 4 0 523000  40824  1848   0   0   0   0   0  14  74  341  8226  564=20=
=20
3  2 95
  0 4 0 523020  40812  1844   0   0   0   0   0   0  67  317  7606  530=20=
=20
2  1 97
  1 4 0 523052  40796  1661   0   0   0   0   0   0  68  315 11603  493=20=
=20
2  2 97
  1 4 0 523056  40800   233   0   0   0   0   0  12  87  341 12550  609=20=
=20
2  2 96
  0 4 0 523076  40788  1845   0   0   0   0   0   0  82  334 12457  626=20=
=20
2  2 96
  0 4 0 523100  40776  1851   0   0   0   0   0   0  91  345 10914  623=20=
=20
2  3 95
  0 4 0 523120  40764  1845   0   0   0   0   0   0  92  343 19213  596=20=
=20
1  5 95
  0 4 0 523136  40752  1845   0   0   0   0   0   0  97  349  8659  605=20=
=20
2  2 96
  0 4 0 523144  40748  4501   0   0   0   0   0  32  78  385 15632  934=20
25 12 64
  0 4 0 523168  40728  1853   0   0   0   0   0   3  74  335  3965  531=20=
=20
0  2 98

 > ps -Upostgresql -Ostart | grep -v idle
   PID STARTED  TT   STAT      TIME COMMAND
  8267 10:53AM  ??  Is      0:00.28 /usr/local/bin/pg_autovacuum -D -L=20
/var/pgsql/autovacuum
23271 10:54AM  ??  I       1:13.56 postmaster: dps dps 127.0.0.1 SELECT=20
(postgres)
28662 10:55AM  ??  I       0:59.98 postmaster: dps dps 127.0.0.1 SELECT=20
(postgres)
25794 10:56AM  ??  D       1:30.48 postmaster: dps dps 127.0.0.1 SELECT=20
(postgres)
  2143 11:02AM  ??  D       1:02.06 postmaster: dps dps 127.0.0.1 DELETE=20
(postgres)
25904 10:52AM  C0- I       0:00.07 /usr/local/bin/postmaster -D=20
/var/pgsql (postgres)
10908 10:52AM  C0- I       0:05.96 postmaster: stats collector process=20=
=20
   (postgres)
  7045 10:52AM  C0- I       0:05.19 postmaster: stats buffer process=20=20=
=20=20
(postgres)

 > grep -v -E '^#' /var/pgsql/postgresql.conf
tcpip_socket =3D true
max_connections =3D 100
shared_buffers =3D 1024           # 32KB
max_fsm_pages =3D 1000000         # min max_fsm_relations*16, 6 bytes each
max_fsm_relations =3D 200         # min 100, ~50 bytes each
wal_buffers =3D 32                # min 4, 8KB each
checkpoint_segments =3D 16        # in logfile segments, min 1, 16MB each
commit_delay =3D 100              # range 0-100000, in microseconds
effective_cache_size =3D 4096     # 32KB each
random_page_cost =3D 3
default_statistics_target =3D 200 # range 1-1000
client_min_messages =3D notice    # Values, in order of decreasing detail:
log_min_messages =3D log          # Values, in order of decreasing detail:
log_min_duration_statement =3D 20000 # Log all statements whose
log_timestamp =3D true
stats_start_collector =3D true
stats_command_string =3D true
stats_block_level =3D true
stats_row_level =3D true
stats_reset_on_server_start =3D true
lc_messages =3D 'C'               # locale for system error message=20
strings
lc_monetary =3D 'C'               # locale for monetary formatting
lc_numeric =3D 'C'                # locale for number formatting
lc_time =3D 'C'                   # locale for time formatting
explain_pretty_print =3D true

 > sysctl -a | grep seminfo
kern.seminfo.semmni =3D 256
kern.seminfo.semmns =3D 2048
kern.seminfo.semmnu =3D 30
kern.seminfo.semmsl =3D 60
kern.seminfo.semopm =3D 100
kern.seminfo.semume =3D 10
kern.seminfo.semusz =3D 100
kern.seminfo.semvmx =3D 32767
kern.seminfo.semaem =3D 16384

 > systat
vmstat
     7 users    Load  3.48  3.64  3.56                  Fri Apr 30=20
14:42:18 2004

           memory totals (in KB)              PAGING   SWAPPING=20=20=20=20=
=20
Interrupts
          real   virtual    free              in  out   in  out      361=20
total
Active 514768    527436   36280      ops                            100=20
clock
All    992496   1005164 4071736      pages                          128=20
rtc
                                                                      45=20
fxp0
Proc:r  d  s  w    Csw   Trp   Sys   Int   Sof  Flt     6 forks      88=20
twe0
         4 26       580  1848  8395   361   249 1856     6 fkppw
                                                           fksvm
    3.0% Sys   1.1% User   0.0% Nice  95.9% Idle           pwait
|    |    |    |    |    |    |    |    |    |    |       relck
=3D>                                                        rlkok
                                                           noram
Namei         Sys-cache    Proc-cache    No-cache      80 ndcpy
     Calls     hits    %    hits     %    miss   %      54 fltcp
       812      806   99       5     1       1   0     208 zfod
                                                        95 cow
Discs  cd0  sd0  sd1  sd2  fd0                        128 fmin
seeks         6   82                                  170 ftarg
xfers         6   82                                60208 itarg
Kbyte        47 2554                                  226 wired
   sec            1.0                                      pdfre

 > tail -f /var/pgsql/log
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000CA"
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000CB"
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000CC"
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000BF"
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000C0"
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000C1"
2004-04-30 11:35:03 LOG:  recycled transaction log file=20
"000000C8000000C2"
2004-04-30 11:36:46 LOG:  duration: 28284.360 ms  statement: SELECT=20
rec_id,url FROM url WHERE status > 300 AND status<>304 AND=20
(referrer=3D'28615' OR referrer=3D'0') AND bad_since_time<1083317778
2004-04-30 11:36:46 LOG:  duration: 24918.201 ms  statement: SELECT=20
rec_id,url FROM url WHERE status > 300 AND status<>304 AND=20
(referrer=3D'122879' OR referrer=3D'0') AND bad_since_time<1083317781
2004-04-30 11:36:46 LOG:  duration: 21173.427 ms  statement: SELECT=20
rec_id,url FROM url WHERE status > 300 AND status<>304 AND=20
(referrer=3D'586182' OR referrer=3D'0') AND bad_since_time<1083317785

 From PhpPgAdmin: Table url: Info
Row Performance
Sequential      Index   Rows
Scan    Read    Scan    Fetch     Insert  Udate  Delete
1       414173  85711   10963854  20431   8707   594

I/O Performance
Heap                   Index                TOAST           TOAST Index
Disk    Cache   %      Disk   Cache   %     Disk Cache %    Disk Cache %
3298907 7790769 (70%)  200782 1274898 (86%) 0    0     (0%) 0    0=20=20=20=
=20=20
(0%)

Index Row Performance
Index                   Scan    Read     Fetch
url_bad_since_time      0       0        0
url_crc                 2924    131566   131566
url_hops                0       0        0
url_last_mod_time       0       0        0
url_next_index_time     5       5120     5120
url_pkey                9187    8980     8980
url_referrer            4431    10753641 10753641
url_seed                0       0        0
url_serverid            0       0        0
url_siteid              0       0        0
url_status              0       0        0
url_url                 69164   64547    64547

Index I/O Performance
Index                   Disk    Cache   %
url_bad_since_time      7169    80280   (92%)
url_crc                 9106    19200   (68%)
url_hops                9071    109864  (92%)
url_last_mod_time       5836    27887   (83%)
url_next_index_time     12004   109887  (90%)
url_pkey                7501    52825   (88%)
url_referrer            58765   97634   (62%)
url_seed                30293   88712   (75%)
url_serverid            8647    110078  (93%)
url_siteid              8888    109864  (93%)
url_status              7448    111250  (94%)
url_url                 36054   357417  (91%)


Cordialement,
Jean-G=E9rard Pailloncy


From pgsql-performance-owner@postgresql.org  Fri Apr 30 14:46:48 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id CA57ED1E8CB
	for ;
	Fri, 30 Apr 2004 14:46:45 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 85733-05
	for ;
	Fri, 30 Apr 2004 14:46:24 -0300 (ADT)
Received: from mailhost2.tudelft.nl (mailhost2.tudelft.nl [130.161.180.2])
	by svr1.postgresql.org (Postfix) with ESMTP id A0534D1BABD
	for ;
	Fri, 30 Apr 2004 14:46:22 -0300 (ADT)
Received: from localhost (localhost [127.0.0.1])
	by rav.antivirus (Postfix) with ESMTP id C640D17D93
	for ;
	Fri, 30 Apr 2004 19:46:22 +0200 (MEST)
Received: from listserv.tudelft.nl (listserv.tudelft.nl [130.161.180.33])
	by mailhost2.tudelft.nl (Postfix) with ESMTP id A8C7017D8F
	for ;
	Fri, 30 Apr 2004 19:46:22 +0200 (MEST)
Received: from oli.tudelft.nl (jochemd.tnw-s.tudelft.nl [145.94.90.156])
	by listserv.tudelft.nl (8.12.10/8.12.8) with ESMTP id i3UHkMF7011958
	for ;
	Fri, 30 Apr 2004 19:46:22 +0200 (MEST)
Message-ID: <409290F0.10509@oli.tudelft.nl>
Date: Fri, 30 Apr 2004 19:46:24 +0200
From: Jochem van Dieten 
Organization: OnLine Internet
User-Agent: Mozilla Thunderbird 0.5 (Windows/20040207)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question
References: 
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-new at tudelft.nl
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/526
X-Sequence-Number: 6826

Manfred Koizar wrote:
> On Wed, 28 Apr 2004 09:05:04 -0400, Tom Lane  wrote:
>>> [ ... visibility information in index tuples ... ]
>> 
>> Storing that information would at least double the overhead space used
>> for each index tuple.  The resulting index bloat would significantly
>> slow index operations by requiring more I/O.  So it's far from clear
>> that this would be a win, even for those who care only about select
>> speed.
> 
> While the storage overhead could be reduced to 1 bit (not a joke)

You mean adding an isLossy bit and only where it is set the head 
tuple has to be checked for visibility, if it is not set the head 
tuple does not have to be checked?


> we'd
> still have the I/O overhead of locating and updating index tuples for
> every heap tuple deleted/updated.

Would there be additional I/O for the additional bit in the index 
tuple (I am unable to find the layout of index tuple headers in 
the docs)?

Jochem

-- 
I don't get it
immigrants don't work
and steal our jobs
     - Loesje



From pgsql-performance-owner@postgresql.org  Fri Apr 30 15:29:32 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 7583ED1E6C8
	for ;
	Fri, 30 Apr 2004 15:29:31 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 99270-09
	for ;
	Fri, 30 Apr 2004 15:29:12 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id A13C5D1E796
	for ;
	Fri, 30 Apr 2004 15:29:09 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 422
	for ; Fri, 30 Apr 2004 19:28:39 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Fri, 30 Apr 2004 19:29:44 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question 
Message-ID: <4092A928.27720.19838422@localhost>
In-reply-to: <69EF9EF2-9AA2-11D8-A4E1-000D9366F0C4@torgo.978.org>
References: <409207D6.17204.170D5A52@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/527
X-Sequence-Number: 6827

On 30 Apr 2004 at 8:32, Jeff wrote:
> 
>   A better comparision query may be a simple "select a from mytable 
> where a between foo and bar"  to get an index scan.  In that case its a 
> straight up, vanilla index scan.  Nothing else getting in the way.
> 

Yes, you're right and I have done this just to prove to myself that it is the index scan that 
is the bottleneck. I have some complex SQL that executes very quickly with Postgres, 
similar to MSSQL, but the index scans in most of those only touch a few rows for a few 
loops. It seems to be a problem when the index scan is scanning very many rows and 
for each of these it has to go to the table just to find out if the index it just looked at is 
still valid.

Gary.


From pgsql-performance-owner@postgresql.org  Fri Apr 30 15:59:26 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 38A4DD1CAA2
	for ;
	Fri, 30 Apr 2004 15:59:23 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 17934-01
	for ;
	Fri, 30 Apr 2004 15:59:03 -0300 (ADT)
Received: from gpdnet.co.uk (gpdnet.plus.com [212.56.100.243])
	by svr1.postgresql.org (Postfix) with ESMTP id 9BEB3D1E728
	for ;
	Fri, 30 Apr 2004 15:59:01 -0300 (ADT)
Received: from gary (unverified [192.168.1.2])
	by gpdnet.co.uk (SurgeMail 1.8g3) with ESMTP id 424
	for ; Fri, 30 Apr 2004 19:58:33 +0100
From: "Gary Doades" 
To: pgsql-performance@postgresql.org
Date: Fri, 30 Apr 2004 19:59:38 +0100
MIME-Version: 1.0
Subject: Re: planner/optimizer question
Message-ID: <4092B02A.15469.199EE3DE@localhost>
In-reply-to: <409264C4.8060506@speedfc.com>
References: <40920137.13806.16F37CC5@localhost>
X-mailer: Pegasus Mail for Windows (v4.12a)
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body
X-Server: High Performance Mail Server - http://surgemail.com
X-Virus-Scanned: by amavisd-new at postgresql.org
X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/528
X-Sequence-Number: 6828

On 30 Apr 2004 at 9:37, Kevin Barnard wrote:

> 
> I was always under the impression that MSSQL used leaf and row level locking and therefore 
> was not a concurrent, in the same sense that postgres is, database. It would still allow for 
> concurrent connections and such but updates will get blocked/ delayed. I might be wrong.
> 

Ultimately you may be right. I don't know enough about SQLServer 
internals to say either way. Anyway, most of our system is in selects for 
70% of the time. I could try and set up a test for this when I get a bit 
more time.

Unfortunately I suspect that this topic won't get taken much further. In 
order to test this it would mean modifying quite a bit of code. Whether 
putting additional info in the index header and not visiting the data row 
if all the required data is in the index would be beneficial would require 
quite a bit of work by someone who knows more than I do. I reckon that 
no-one has the time to do this at the moment.

Regards,
Gary.


From pgsql-performance-owner@postgresql.org  Fri Apr 30 16:49:03 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 6BEFFD1E796
	for ;
	Fri, 30 Apr 2004 16:49:00 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 24604-09
	for ;
	Fri, 30 Apr 2004 16:48:41 -0300 (ADT)
Received: from mail63.csoft.net (unknown [63.111.22.74])
	by svr1.postgresql.org (Postfix) with SMTP id 39478D1E835
	for ;
	Fri, 30 Apr 2004 16:48:39 -0300 (ADT)
Received: (qmail 2506 invoked by uid 1112); 30 Apr 2004 19:48:32 -0000
Date: Fri, 30 Apr 2004 14:48:32 -0500 (EST)
From: Kris Jurka 
X-X-Sender: books@leary.csoft.net
To: Gary Doades 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question 
In-Reply-To: <4092A928.27720.19838422@localhost>
Message-ID: 
References: <409207D6.17204.170D5A52@localhost>
	<4092A928.27720.19838422@localhost>
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=0.0 tagged_above=0.0 required=5.0 tests=
X-Spam-Level: 
X-Archive-Number: 200404/529
X-Sequence-Number: 6829



On Fri, 30 Apr 2004, Gary Doades wrote:

> Yes, you're right and I have done this just to prove to myself that it
> is the index scan that is the bottleneck. I have some complex SQL that
> executes very quickly with Postgres, similar to MSSQL, but the index
> scans in most of those only touch a few rows for a few loops. It seems
> to be a problem when the index scan is scanning very many rows and for
> each of these it has to go to the table just to find out if the index it
> just looked at is still valid.
> 

Another way to speed this up is the TODO item: "Use bitmaps to fetch 
heap pages in sequential order"  For an indexscan that fetches a number 
of rows those rows may be anywhere in the base table so as each index 
entry is found it fetches the corresponding table row from the heap.  This 
is not ideal because you can be jumping around in the heap and end up 
fetching the same page multiple times because table rows are in the same 
page, but were found in different places in the index.

Kris Jurka

From pgsql-performance-owner@postgresql.org  Fri Apr 30 18:34:52 2004
X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org
Received: from localhost (unknown [200.46.204.2])
	by svr1.postgresql.org (Postfix) with ESMTP id 199E6D1E70E
	for ;
	Fri, 30 Apr 2004 18:34:51 -0300 (ADT)
Received: from svr1.postgresql.org ([200.46.204.71])
	by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024)
	with ESMTP id 64926-08
	for ;
	Fri, 30 Apr 2004 18:34:27 -0300 (ADT)
Received: from email05.aon.at (WARSL402PIP4.highway.telekom.at [195.3.96.79])
	by svr1.postgresql.org (Postfix) with SMTP id 2229CD1B4F7
	for ;
	Fri, 30 Apr 2004 18:34:23 -0300 (ADT)
Received: (qmail 561446 invoked from network); 30 Apr 2004 21:34:22 -0000
Received: from m161p030.dipool.highway.telekom.at (HELO PASCAL)
	([62.46.10.30]) (envelope-sender )
	by qmail5rs.highway.telekom.at (qmail-ldap-1.03) with SMTP
	for ; 30 Apr 2004 21:34:22 -0000
From: Manfred Koizar 
To: Jochem van Dieten 
Cc: pgsql-performance@postgresql.org
Subject: Re: planner/optimizer question
Date: Fri, 30 Apr 2004 23:36:58 +0200
Message-ID: 
References: 
	<409290F0.10509@oli.tudelft.nl>
In-Reply-To: <409290F0.10509@oli.tudelft.nl>
X-Mailer: Forte Agent 2.0/32.652
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=0.1 tagged_above=0.0 required=5.0 tests=RCVD_IN_SORBS
X-Spam-Level: 
X-Archive-Number: 200404/530
X-Sequence-Number: 6830

On Fri, 30 Apr 2004 19:46:24 +0200, Jochem van Dieten
 wrote:
>> While the storage overhead could be reduced to 1 bit (not a joke)
>
>You mean adding an isLossy bit and only where it is set the head 
>tuple has to be checked for visibility, if it is not set the head 
>tuple does not have to be checked?

Yes, something like this.  Actually I imagined it the other way round: a
visible-to-all flag similar to the existing dead-to-all flag (search for
LP_DELETE and ItemIdDeleted in nbtree.c).

>> we'd
>> still have the I/O overhead of locating and updating index tuples for
>> every heap tuple deleted/updated.
>
>Would there be additional I/O for the additional bit in the index 
>tuple (I am unable to find the layout of index tuple headers in 
>the docs)?

Yes, the visible-to-all flag would be set as a by-product of an index
scan, if the heap tuple is found to be visible to all active
transactions.  This update is non-critical and, I think, not very
expensive.

Deleting (and hence updating) a tuple is more critical, regarding both
consistency and performance.  We'd have to locate all index entries
pointing to the heap tuple and set their visible-to-all flags to false.