X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 64C453CC746 for ; Fri, 17 Dec 2004 19:47:10 +0000 (GMT) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 34928-03 for ; Fri, 17 Dec 2004 19:46:59 +0000 (GMT) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr1.postgresql.org (Postfix) with ESMTP id 0C8AD3CC616 for ; Fri, 17 Dec 2004 19:46:59 +0000 (GMT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id iBHJkvla022122; Fri, 17 Dec 2004 14:46:57 -0500 (EST) To: josh@agliodbs.com Cc: Pailloncy Jean-Gerard , pgsql-performance@postgresql.org Subject: Re: Error in VACUUM FULL VERBOSE ANALYZE (not enough memory) In-reply-to: <200412171047.49250.josh@agliodbs.com> References: <3D4B6E62-5055-11D9-96E2-000A95DE2550@ifrance.com> <07F4E72A-5059-11D9-96E2-000A95DE2550@ifrance.com> <200412171047.49250.josh@agliodbs.com> Comments: In-reply-to Josh Berkus message dated "Fri, 17 Dec 2004 10:47:49 -0800" Date: Fri, 17 Dec 2004 14:46:57 -0500 Message-ID: <22121.1103312817@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.004 tagged_above=0 required=5 tests=AWL X-Spam-Level: X-Archive-Number: 200412/223 X-Sequence-Number: 9598 Josh Berkus writes: > Jean-Gerard, >> When backend hits the tsearch2 index, SIZE/RES grows until it reachs >> 1GB, where I got the error. >> PID UID PRI NICE SIZE RES STATE WAIT TIME CPU COMMAND >> 20461 503 -5 0 765M 824M sleep biowai 4:26 33.20% postgres > OK, next thing to try is upgrading to 7.4.7. Since you have 7.4.2, this > should be a straightforward binary replacement. This looks like it must be a memory leak in the gist indexing code (either gist itself or tsearch2). I don't see any post-release fixes in the 7.4 branch that look like they fixed any such thing :-(, so it's probably still there in 7.4.7, and likely 8.0 too. Jean-Gerard, can you put together a self-contained test case? I suspect it need only look like "put some data in a table, make a tsearch2 index, delete half the rows in the table, VACUUM FULL". But I don't have time to try to cons up a test case right now, and especially not to figure out what to do to duplicate your problem if it doesn't happen on the first try. regards, tom lane