Received: from localhost (maia-4.hub.org [200.46.204.183]) by postgresql.org (Postfix) with ESMTP id 7998E9FC0B2 for ; Thu, 7 Jun 2007 22:18:09 -0300 (ADT) Received: from postgresql.org ([200.46.204.71]) by localhost (mx1.hub.org [200.46.204.183]) (amavisd-maia, port 10024) with ESMTP id 09262-03 for ; Thu, 7 Jun 2007 22:18:05 -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 postgresql.org (Postfix) with ESMTP id 3B5129FBFF8 for ; Thu, 7 Jun 2007 22:18:06 -0300 (ADT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.14.1/8.14.1) with ESMTP id l581HfhT011459; Thu, 7 Jun 2007 21:17:41 -0400 (EDT) To: "George Pavlov" cc: "Ed L." , pgsql-general@postgresql.org Subject: Re: query log corrupted-looking entries In-reply-to: <8C5B026B51B6854CBE88121DBF097A86DEA716@ehost010-33.exch010.intermedia.net> References: <29860.1161202644@sss.pgh.pa.us> <8C5B026B51B6854CBE88121DBF097A86C39FC2@ehost010-33.exch010.intermedia.net> <200705291119.55567.pgsql@bluepolka.net> <8C5B026B51B6854CBE88121DBF097A86CDD9A4@ehost010-33.exch010.intermedia.net> <24195.1180733773@sss.pgh.pa.us> <8C5B026B51B6854CBE88121DBF097A86CDD9D2@ehost010-33.exch010.intermedia.net> <24885.1180736713@sss.pgh.pa.us> <8C5B026B51B6854CBE88121DBF097A86CDDA32@ehost010-33.exch010.intermedia.net> <8C5B026B51B6854CBE88121DBF097A86DEA716@ehost010-33.exch010.intermedia.net> Comments: In-reply-to "George Pavlov" message dated "Thu, 07 Jun 2007 17:57:03 -0700" Date: Thu, 07 Jun 2007 21:17:41 -0400 Message-ID: <11458.1181265461@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: Maia Mailguard 1.0.1 X-Archive-Number: 200706/559 X-Sequence-Number: 115382 "George Pavlov" writes: > I tried the patch and it has no effect whatsoever -- even with the > patch, under the correct load the corrupted entries are coming fast and > furious (I found a load profile on my app that reproduces these very > clearly). What are the total lengths of the log entries in which you see the failure? (The "length" here includes all the lines belonging to a single logical entry, eg, ERROR, DETAIL, HINT.) I believe that this shouldn't be happening except in the case in which the entry-interpolated-into exceeds PIPE_BUF bytes. I'm not entirely sure what PIPE_BUF is on Linux machines, but IIRC the Posix spec says it has to be at least 512, and on older Unix kernels it tends to be 8K. regards, tom lane