X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [64.117.224.130]) by svr1.postgresql.org (Postfix) with ESMTP id DFE8ED1B535 for ; Mon, 15 Sep 2003 22:49:41 +0000 (GMT) Received: from svr1.postgresql.org ([64.117.224.193]) by localhost (neptune.hub.org [64.117.224.130]) (amavisd-new, port 10024) with ESMTP id 53755-04 for ; Mon, 15 Sep 2003 19:49:11 -0300 (ADT) Received: from megazone.bigpanda.com (megazone.bigpanda.com [64.147.171.210]) by svr1.postgresql.org (Postfix) with ESMTP id 4B8B1D1B52B for ; Mon, 15 Sep 2003 19:49:08 -0300 (ADT) Received: by megazone.bigpanda.com (Postfix, from userid 1001) id 820D33521D; Mon, 15 Sep 2003 15:49:10 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by megazone.bigpanda.com (Postfix) with ESMTP id 80F1F3520B; Mon, 15 Sep 2003 15:49:10 -0700 (PDT) Date: Mon, 15 Sep 2003 15:49:10 -0700 (PDT) From: Stephan Szabo To: Joseph Bove Cc: pgsql-performance@postgresql.org Subject: Re: Inconsistent performance In-Reply-To: <5.1.0.14.2.20030915182422.00a8a2d8@mail.vetstar.com> Message-ID: <20030915154131.I76677@megazone.bigpanda.com> References: <5.1.0.14.2.20030915182422.00a8a2d8@mail.vetstar.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200309/224 X-Sequence-Number: 3590 On Mon, 15 Sep 2003, Joseph Bove wrote: > Stephan, > > Actually, it's inconsistent with the exact same command. I've now > replicated the problem by doing the following command: > > select count (*) from table; > > The table in question has 88899 rows. > > The response time is anywhere from 1 second to 12 seconds. Different > response times can occur in the same minute of testing! Well, that's really only got one valid plan right now (seqscan and aggregate). It'd be mildly interesting to see what explain analyze says in slow and fast states, although I'd be willing to bet that it's just going to effectively show that the seqscan is taking more or less time. I think we're going to need to see the configuration settings for the server and possibly some info on how big the table is (say relpages for the pg_class row associated with the table after a vacuum full).