X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (developer.postgresql.org [64.117.224.193]) by svr1.postgresql.org (Postfix) with ESMTP id 6ADEDB43E23 for ; Fri, 4 Jul 2003 17:36:56 +0000 (GMT) Received: from svr1.postgresql.org ([64.117.224.193]) by localhost (svr1.postgresql.org [64.117.224.193]) (amavisd-new, port 10024) with ESMTP id 65819-07 for ; Fri, 4 Jul 2003 14:36:46 -0300 (ADT) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by svr1.postgresql.org (Postfix) with ESMTP id 91992B43E21 for ; Fri, 4 Jul 2003 14:36:45 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.9/8.12.9) with ESMTP id h64HFIQQ027906; Fri, 4 Jul 2003 13:15:18 -0400 (EDT) To: josh@agliodbs.com Cc: Sean Chittenden , pgsql-performance@postgresql.org Subject: Re: Moving postgresql.conf tunables into 2003... In-reply-to: <200307031938.15357.josh@agliodbs.com> References: <20030703190502.GR72567@perrin.int.nxad.com> <200307031938.15357.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Thu, 03 Jul 2003 19:38:15 -0700" Date: Fri, 04 Jul 2003 13:15:17 -0400 Message-ID: <27905.1057338917@sss.pgh.pa.us> From: Tom Lane X-Archive-Number: 200307/83 X-Sequence-Number: 2376 Josh Berkus writes: > Tom Comments: >> I was arguing awhile back for bumping the default shared_buffers up, >> but the discussion trailed off with no real resolution. > I think we ran up against the still far-too-low SHMMAX settings in most > *nixes. We could raise this default once we can supply a script which will > help the user bump up the OS's memory settings at, say, initDB time. Actually, I think it would not be hard to get initdb to test whether larger shared-memory settings would work. We could do something like try -B of 64, 256, 512, 1024, and insert into postgresql.conf the largest value that works. I would want it to top out at a few thousand at most, because I don't think a default installation should try to commandeer the whole machine, but if we could get the typical installation to be running with even 1000 buffers rather than 64, we'd be *way* better off. (See "Postgres vs MySQL" thread nearby.) We could possibly also have initdb print some kind of message if it's forced to use an unreasonably small value for shared_buffers, so that people might have a clue that they need to do kernel reconfiguration. Comments? regards, tom lane