Received: from localhost (unknown [200.46.204.183]) by postgresql.org (Postfix) with ESMTP id 009CF6501D7 for ; Thu, 30 Oct 2008 12:58:46 -0300 (ADT) Received: from postgresql.org ([200.46.204.86]) by localhost (mx1.hub.org [200.46.204.183]) (amavisd-maia, port 10024) with ESMTP id 44519-02 for ; Thu, 30 Oct 2008 12:58:43 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.6 Received: from amidala.server.telavista.net (amidala.server.telavista.net [213.239.138.194]) by postgresql.org (Postfix) with ESMTP id 051CE6501E5 for ; Thu, 30 Oct 2008 12:58:42 -0300 (ADT) Received: from firewall.tecnocon.com (gateway.technocon.com [62.132.195.8]) by amidala.server.telavista.net (Postfix) with ESMTP id B3F266D78B7; Thu, 30 Oct 2008 16:58:32 +0100 (CET) Received: from [192.168.0.129] (unknown [192.168.0.129]) by firewall.tecnocon.com (Postfix) with ESMTP id 5513A225212; Thu, 30 Oct 2008 16:58:36 +0100 (CET) Message-ID: <4909D989.5020905@technocon.com> Date: Thu, 30 Oct 2008 16:58:01 +0100 From: Christiaan Willemsen User-Agent: Thunderbird 2.0.0.17 (Windows/20080914) MIME-Version: 1.0 To: "Joshua D. Drake" Cc: pgsql-performance@postgresql.org Subject: Re: Configuring for maximum memory usage References: <4909CF9C.80701@technocon.com> <4909D688.1090806@commandprompt.com> In-Reply-To: <4909D688.1090806@commandprompt.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-svmilter: B3F266D78B7: result=CLEAN, action=ACCEPT, virus=-1, spam=-1, rbl=0, nbl=0, pbl=0 X-Virus-Scanned: Maia Mailguard 1.0.1 X-Spam-Status: No, hits=0 tagged_above=0 required=5 tests=none X-Spam-Level: X-Archive-Number: 200810/121 X-Sequence-Number: 31702 Joshua D. Drake wrote: > > PostgreSQL is only going to use what it needs. It relies on the OS for > much of the caching etc... > So that would actually mean that I could raise the setting of the ARC cache to far more than 8 GB? As I said, our database is 250 GB, So I would expect that postgres needs more than it is using right now... Several tables have over 500 million records (obviously partitioned). At the moment we are doing queries over large datasets, So I would assume that postgress would need a bit more memory than this.. > > You are missing effective_cache_size. Try setting that to 32G. That one was set to 24 GB. But this setting only tells posgres how much caching it can expect from the OS? This is not actually memory that it will allocate, is it? > > You also didn't mention checkpoint_segments (which isn't memory but > still important) and default_statistics_target (which isn't memory but > still important). > is at the moment set to: checkpoint_segments = 40 default_statistics_target is set to default (I think that is 10) Thanks already, Christiaan