X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id D8CF8D1B174 for ; Mon, 28 Jun 2004 02:19:49 -0300 (ADT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 53343-07 for ; Mon, 28 Jun 2004 05:19:48 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id F2E85D1B170 for ; Mon, 28 Jun 2004 02:19:46 -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 i5S5JgNb017885; Mon, 28 Jun 2004 01:19:42 -0400 (EDT) To: Frank Knobbe Cc: Christopher Kings-Lynne , pgsql-performance@postgresql.org Subject: Re: postgres 7.4 at 100% In-reply-to: <1088397641.1105.10.camel@localhost> References: <40DBEBDF.3090605@archonet.com> <87llia78dj.fsf@asmodeus.mcnaught.org> <40DD8425.3080108@commandprompt.com> <40DE5C25.6070805@familyhealth.com.au> <1088397641.1105.10.camel@localhost> Comments: In-reply-to Frank Knobbe message dated "Sun, 27 Jun 2004 23:46:58 -0500" Date: Mon, 28 Jun 2004 01:19:42 -0400 Message-ID: <17884.1088399982@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200406/271 X-Sequence-Number: 7329 Frank Knobbe writes: > On Sun, 2004-06-27 at 00:33, Christopher Kings-Lynne wrote: >>> I upped effective_cache to 16000 KB and I could only up the >>> shared_buffers to 3000. Anything more and postgres would not start. >> You need to greatly incrase the shared memory max setting on your >> machine so that you can use at the very least, 10000 shared buffers. > Doug said the same, yet the PG Tuning article recommends not make this > too large as it is just temporary used by the query queue or so. The original report was that the guy had it set to 40 (!?), which is clearly far below the minimum reasonable value. But I'd not expect a huge difference between 3000 and 10000 --- in my experience, 1000 is enough to get you over the "knee" of the performance curve and into the domain of marginal improvements. So while he surely should not go back to 40, it seems there's another factor involved here that we've not recognized yet. regards, tom lane