Received: from localhost (maia-2.hub.org [200.46.204.187]) by postgresql.org (Postfix) with ESMTP id B10059FB743 for ; Tue, 17 Apr 2007 10:12:49 -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 46590-10 for ; Tue, 17 Apr 2007 10:12:37 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.4 Received: from out4.smtp.messagingengine.com (out4.smtp.messagingengine.com [66.111.4.28]) by postgresql.org (Postfix) with ESMTP id 436019FB378 for ; Tue, 17 Apr 2007 10:12:42 -0300 (ADT) Received: from compute1.internal (compute1.internal [10.202.2.41]) by out1.messagingengine.com (Postfix) with ESMTP id E5EBC214212; Tue, 17 Apr 2007 09:12:36 -0400 (EDT) Received: from heartbeat1.messagingengine.com ([10.202.2.160]) by compute1.internal (MEProxy); Tue, 17 Apr 2007 09:12:42 -0400 X-Sasl-enc: Sq+uxNxIYAWHqQ2PCKd4xx7vOyXZVJ/pfnhbmL+7zJDZ 1176815562 Received: from [192.168.4.67] (pool-72-88-194-98.nwrknj.fios.verizon.net [72.88.194.98]) by mail.messagingengine.com (Postfix) with ESMTP id 520421DBCF; Tue, 17 Apr 2007 09:12:42 -0400 (EDT) Message-ID: <4624C7E5.7060506@zeut.net> Date: Tue, 17 Apr 2007 09:13:09 -0400 From: Matthew O'Connor User-Agent: Thunderbird 2.0.0.0 (Windows/20070326) MIME-Version: 1.0 To: "Kieran Cooper, Lyris UK" CC: pgsql-admin@postgresql.org Subject: Re: Vacuum stops with misleading max_fsm_pages error References: <001901c78071$9a9a1f00$0b02a8c0@KieranX15> <4623F54E.3030900@zeut.net> <00ed01c780c5$91dd2390$0b02a8c0@KieranX15> In-Reply-To: <00ed01c780c5$91dd2390$0b02a8c0@KieranX15> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200704/151 X-Sequence-Number: 25215 Kieran Cooper, Lyris UK wrote: >> What is it about autovacuum that isn't working for you? > > Hi Matthew. I have to admit that I haven't tested full yet, but here's > the scenario: > When Lyris sends a mailing there is an intense amount of database > activity - particularly on two tables. When I had auto vacuum set up, > the database just wasn't responding fast enough, so the mailing speed > dropped as Lyris dynamically adjusted based on the speed of db response. > I guess what I need to do is tweak the intervals and the sleep time so > that vacuum backs off when the database is really busy - do you have any > thoughts on what settings I should try in order to acheive that? The sleep intervals are one thing, but what you probably really want to play with are they vacuum cost delay and limit settings. This is throttle down how much IO autovacuum can consume. http://www.postgresql.org/docs/8.2/interactive/runtime-config-autovacuum.html