Received: from localhost (maia-2.hub.org [200.46.204.187]) by postgresql.org (Postfix) with ESMTP id 8E3E79FBE8F for ; Thu, 14 Jun 2007 12:26:44 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.187]) (amavisd-maia, port 10024) with ESMTP id 59888-06 for ; Thu, 14 Jun 2007 12:26:32 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.5 Received: from mail01.enterprisedb.com (mail01.enterprisedb.com [63.246.7.168]) by postgresql.org (Postfix) with ESMTP id 61EA59FBE8B for ; Thu, 14 Jun 2007 12:26:32 -0300 (ADT) Thread-Index: AceumGvcJdCebG/WQReqGLciNZ6tpg== Received: from [192.168.0.4] ([85.211.203.94]) by mail01.enterprisedb.com with Microsoft SMTPSVC(6.0.3790.1830); Thu, 14 Jun 2007 11:26:45 -0400 Content-class: urn:content-classes:message X-MimeOLE: Produced By Microsoft MimeOLE V6.00.3790.2826 Importance: normal Subject: Re: [PG 8.1.0 / AIX 5.3] Vacuum processes freezing From: "Simon Riggs" To: =?iso-8859-1?Q?RESTOUX=2C_Lo=EFc?= Cc: In-Reply-To: References: Content-Type: text/plain; charset="utf-8" Date: Thu, 14 Jun 2007 16:23:23 +0100 Message-ID: <1181834603.5776.66.camel@silverbirch.site> MIME-Version: 1.0 X-Mailer: Evolution 2.6.0 Content-Transfer-Encoding: quoted-printable X-OriginalArrivalTime: 14 Jun 2007 15:26:46.0203 (UTC) FILETIME=[6B5038B0:01C7AE98] X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200706/254 X-Sequence-Number: 25287 On Wed, 2007-06-13 at 18:33 +0200, RESTOUX, Lo=C3=AFc wrote: > 2) I believed that the poor performances during the vacuum freeze were du= e to the obsolete data statistics. But after a full restart of the dabatase= , performances are good. Does PostgreSQL rebuild his statistics during star= tup ?=20 You probably don't need to run VACUUM FREEZE. VACUUM FREEZE will thrash the disks much more than normal VACUUM. We're improving that somewhat in 8.3, but the basic issue is that VACUUM FREEZE cleans out more dead rows and so will dirty more data blocks. Are you concurrently running DDL, Truncate or CLUSTER? That will interfere with the operation of VACUUM. --=20 Simon Riggs =20 EnterpriseDB http://www.enterprisedb.com