X-Original-To: pgsql-performance-postgresql.org@postgresql.org Received: from localhost (mx1.hub.org [200.46.208.251]) by postgresql.org (Postfix) with ESMTP id 9A5CC9FA322 for ; Wed, 21 Jun 2006 20:09:04 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.208.251]) (amavisd-new, port 10024) with ESMTP id 63498-09 for ; Wed, 21 Jun 2006 20:09:00 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey- Received: from q1exch01.Q1LABS.INC (Q1EXCH01.Q1Labs.com [205.174.165.4]) by postgresql.org (Postfix) with ESMTP id E880D9FA3AF for ; Wed, 21 Jun 2006 20:08:59 -0300 (ADT) Received: from q1exch02.Q1LABS.INC ([10.100.50.8]) by q1exch01.Q1LABS.INC with Microsoft SMTPSVC(6.0.3790.1830); Wed, 21 Jun 2006 20:08:56 -0300 X-PMWin-Version: 2.5.0e, Antispam-Engine: 2.3.0.1, Antivirus-Engine: 2.34.3 thread-index: AcaVh6tppk5hP6meQOGJsnpT2etgGw== Received: from [10.100.100.123] ([10.100.100.123]) by q1exch02.Q1LABS.INC with Microsoft SMTPSVC(6.0.3790.1830); Wed, 21 Jun 2006 20:08:55 -0300 From: "jody brownell" To: "Jim C. Nasby" Content-Class: urn:content-classes:message Importance: normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.2663 Subject: Re: Help tuning autovacuum - seeing lots of relationbloat Date: Wed, 21 Jun 2006 20:08:55 -0300 User-Agent: KMail/1.8.3 Cc: "Csaba Nagy" , "postgres performance list" References: <200606211052.43227.jody.brownell@q1labs.com> <200606211641.45674.jody.brownell@q1labs.com> <20060621221150.GN93655@pervasive.com> In-Reply-To: <20060621221150.GN93655@pervasive.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-ID: <200606212008.56137.jody.brownell@q1labs.com> X-OriginalArrivalTime: 21 Jun 2006 23:08:55.0559 (UTC) FILETIME=[AB69A970:01C69587] X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200606/427 X-Sequence-Number: 19784 Well, for one we did introduce a TX leak which was preventing autovac from running. I guess that was _the_ issue. I have since fixed it and an now testing.... looks much better, nothing concerning.... (fingers crossed until morning :)). debug logs are full of vac/anal of the tables... so, for now I am back on track moving forward... Now that auto vac is actually running, the box is feeling slightly more sluggish. BTW - As soon as we deliver to QA, I will post the test case for the memory leak I was seeing the other day. (I have not forgotten, I am just swamped) Thanks for the help all. Much appreciated. Cheers. On Wednesday 21 June 2006 19:11, Jim C. Nasby wrote: > On Wed, Jun 21, 2006 at 04:41:45PM -0300, jody brownell wrote: > > BTW, in production with a similar load - autovacuum with default out of the box > > settings seems to work quite well.... > > > > I double checked this earlier today. > > So what's different between production and the machine with the problem? > > The issue with autovac is that it will only vacuum one table at a time, > so if it's off vacuuming some other table for a long period of time it > won't be touching this table, which will be a problem. Now, if that's > actually what's happening...