X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (neptune.hub.org [200.46.204.2]) by svr1.postgresql.org (Postfix) with ESMTP id D294ED1D660; Fri, 21 Nov 2003 21:34:17 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (neptune.hub.org [200.46.204.2]) (amavisd-new, port 10024) with ESMTP id 90740-10; Fri, 21 Nov 2003 17:33:48 -0400 (AST) Received: from davinci.ethosmedia.com (server228.ethosmedia.com [209.128.84.228]) by svr1.postgresql.org (Postfix) with ESMTP id 04671D1CA7F; Fri, 21 Nov 2003 17:33:46 -0400 (AST) Received: from [66.219.92.2] (HELO temoku) by davinci.ethosmedia.com (CommuniGate Pro SMTP 4.0.2) with ESMTP id 3956660; Fri, 21 Nov 2003 13:34:19 -0800 Content-Type: text/plain; charset="iso-8859-1" From: Josh Berkus Reply-To: josh@agliodbs.com Organization: Aglio Database Solutions To: "Matthew T. O'Connor" Subject: Re: [HACKERS] More detail on settings for pgavd? Date: Fri, 21 Nov 2003 13:23:18 -0800 User-Agent: KMail/1.4.3 Cc: Robert Treat , Shridhar Daithankar , pgsql-performance@postgresql.org, PostgreSQL-development References: <200311181558.45847.josh@agliodbs.com> <200311210909.00978.josh@agliodbs.com> <3FBE8289.1040009@zeut.net> In-Reply-To: <3FBE8289.1040009@zeut.net> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200311211323.19003.josh@agliodbs.com> X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200311/193 X-Sequence-Number: 4753 Matthew, > Actually, this might be a necessary addition as pg_autovacuum currently= =20 > suffers from the startup transients that the FSM used to suffer from,=20 > that is, it doesn't remember anything that happened the last time it=20 > ran. A pg_autovacuum database could also be used to store thresholds=20 > and counts from the last time it ran. I don't see how a seperate database is better than a table in the databases= .,=20 except that it means scanning only one table and not one per database. Fo= r=20 one thing, making it a seperate database could make it hard to back up and= =20 move your database+pg_avd config. But I don't feel strongly about it. > Where are you getting 13% from?=20 13% * 3/4 ~~ 10% And I think both of use agree that vacuuming tables with less than 10% chan= ges=20 is excessive and could lead to problems on its own, like overlapping vacuum= s. > Do you know of an easy way to get a=20 > count of the total pages used by a whole cluster? Select sum(relpages) from pg_class. > I do like the concept though as long as we find good values for=20 > min_fsm_percentage and min_autovac_scaling_factor. See above. I propose 0.13 and 0.1 > Which we already keep a copy of inside of pg_autovacuum, and update=20 > after we issue a vacuum. Even easier then. BTW, do we have any provisions to avoid overlapping vacuums? That is, to= =20 prevent a second vacuum on a table if an earlier one is still running? --=20 -Josh Berkus Aglio Database Solutions San Francisco