Received: from localhost (maia-2.hub.org [200.46.204.187]) by postgresql.org (Postfix) with ESMTP id B5FDF9FC0FA for ; Thu, 12 Apr 2007 01:52:26 -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 55441-01-2 for ; Thu, 12 Apr 2007 01:52:20 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.4 Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by postgresql.org (Postfix) with ESMTP id 7E8969FD380 for ; Thu, 12 Apr 2007 01:31:29 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.6/8.13.6) with ESMTP id l3C4VSHK008525; Thu, 12 Apr 2007 00:31:28 -0400 (EDT) To: Steve cc: pgsql-performance@postgresql.org Subject: Re: Question about memory allocations In-reply-to: References: <233A9987-ABA5-4491-A304-FDA1F8FCD962@gmail.com> <14019.1176155022@sss.pgh.pa.us> <22512E5C-A60F-43E7-A370-257F7AF956FD@gmail.com> <16374.1176171185@sss.pgh.pa.us> <25200.1176213242@sss.pgh.pa.us> <26925.1176221294@sss.pgh.pa.us> Comments: In-reply-to Steve message dated "Tue, 10 Apr 2007 15:28:48 -0400" Date: Thu, 12 Apr 2007 00:31:28 -0400 Message-ID: <8524.1176352288@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200704/247 X-Sequence-Number: 24124 Steve writes: > - What is temp_buffers used for exactly? Temporary tables. Pages of temp tables belonging to your own backend don't ever get loaded into the main shared-buffers arena, they are read into backend-local memory. temp_buffers is the max amount (per backend) of local memory to use for this purpose. > - Do full_page_writes and wal_buffers settings matter AT ALL for a machine > where fysnc = off ? Yes. > - What does wal_buffers mean and does increasing this value actually help > anything? It's the amount of space available to buffer WAL log data that's not been written to disk. If you have a lot of short transactions then there's not much benefit to increasing it (because the WAL will be getting forced to disk frequently anyway) but I've heard reports that for workloads involving long single transactions bumping it up to 64 or 100 or so helps. > - Any idea if this is a smart configuration for this machine? Um ... you didn't mention which PG version? > # This value is going to probably set off cries of using this as a set > # command instead of a big global value; No kidding. You do NOT want work_mem that high, at least not without an extremely predictable, simple workload. > wal_buffers = 512MB I haven't heard any reports that there's a point in values even as high as 1 meg for this. regards, tom lane