Received: from localhost (postgresql.org [64.49.215.8]) by postgresql.org (Postfix) with ESMTP id 66875475B33 for ; Fri, 24 Jan 2003 20:29:40 -0500 (EST) Received: from sss.pgh.pa.us (unknown [192.204.191.242]) by postgresql.org (Postfix) with ESMTP id 52CBB475461 for ; Fri, 24 Jan 2003 20:29:39 -0500 (EST) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.12.6/8.12.6) with ESMTP id h0P1Tf5u028261; Fri, 24 Jan 2003 20:29:41 -0500 (EST) To: Carlos Moreno , PgSQL Performance ML Subject: Re: Having trouble with backups (was: Re: Crash Recovery) In-reply-to: <28204.1043457542@sss.pgh.pa.us> References: <2147483647.1043361178@[10.0.1.133]> <1043416377.29437.70.camel@haggis> <3E3158DA.7080807@mochima.com> <22650.1043442101@sss.pgh.pa.us> <3E31DA80.7000000@mochima.com> <28204.1043457542@sss.pgh.pa.us> Comments: In-reply-to Tom Lane message dated "Fri, 24 Jan 2003 20:19:02 -0500" Date: Fri, 24 Jan 2003 20:29:41 -0500 Message-ID: <28260.1043458181@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by AMaViS new-20020517 X-Archive-Number: 200301/316 X-Sequence-Number: 963 I said: > Carlos Moreno writes: >> One quick question, Tom, before some general comments and >> reply to the other messages... Where would I specify any >> locks the software wants to do? > If you are not issuing any explicit "LOCK" SQL commands, then you can > disregard my theory. Actually, that's too simple. Are you creating and dropping tables, or issuing schema-change commands (such as ADD COLUMN or RENAME)? All of those things take exclusive locks on the tables they modify. Ordinary SELECT/INSERT/UPDATE/DELETE operations can run in parallel with pg_dump, but messing with the database structure is another story. I guess the real question here is whether your app is actually stopped dead (as it would be if waiting for a lock), or just slowed to a crawl (as a performance problem could do). I cannot tell if your "frozen" description is hyperbole or literal truth. One thing that might help diagnose it is to look at the output of ps auxww (or ps -ef on SysV-ish platforms) to see what all the backends are currently doing while the problem exists. regards, tom lane