Received: from localhost (unknown [200.46.204.184]) by developer.postgresql.org (Postfix) with ESMTP id 3F5B32E0037 for ; Sun, 30 Mar 2008 12:40:40 -0300 (ADT) Received: from developer.postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.184]) (amavisd-maia, port 10024) with ESMTP id 10356-04 for ; Sun, 30 Mar 2008 12:40:39 -0300 (ADT) X-Greylist: from auto-whitelisted by SQLgrey-1.7.5 Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by developer.postgresql.org (Postfix) with ESMTP id 7329A2E0035 for ; Sun, 30 Mar 2008 12:40:39 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.14.2/8.14.2) with ESMTP id m2UFeZt9010440; Sun, 30 Mar 2008 11:40:35 -0400 (EDT) To: "Mikko Partio" cc: pgsql-admin@postgresql.org Subject: Re: Vacuum full crash In-reply-to: <2ca799770803300356r185b1adbv2ab17c7cc8f212a7@mail.gmail.com> References: <2ca799770803291340p75d8a4e2o7dfd6c61e73270f6@mail.gmail.com> <21772.1206824718@sss.pgh.pa.us> <2ca799770803300356r185b1adbv2ab17c7cc8f212a7@mail.gmail.com> Comments: In-reply-to "Mikko Partio" message dated "Sun, 30 Mar 2008 13:56:19 +0300" Date: Sun, 30 Mar 2008 11:40:35 -0400 Message-ID: <10439.1206891635@sss.pgh.pa.us> From: Tom Lane X-Archive-Number: 200803/256 X-Sequence-Number: 28739 "Mikko Partio" writes: > On Sun, Mar 30, 2008 at 12:05 AM, Tom Lane wrote: >> "Mikko Partio" writes: >>> 2008-03-29 22:25:15 EET [26841]: [3-1] PANIC: cannot abort transaction >>> 3778747509, it was already committed >> >> Yeah, this is a well-known problem with VACUUM FULL ... > Ok, that's good to know. Is there a bugfix on the way or is it more like a > 'feature'? It would be good to fix it I suppose, but what with VACUUM FULL being on the edge of deprecation anyway, it's hard to muster enthusiasm for doing a (probably) large additional amount of work on it, I think most hackers would prefer to spend time on making it unnecessary than making it more bulletproof. Note that there is (AFAIK) no risk of loss of committed data from this bug. The fact that the PANIC kills all your other sessions is annoying, but in the scenarios where this is likely to happen you're trying to stop them all anyway ... regards, tom lane