From: Markus Trippelsdorf Subject: Re: Filessystem corruptions while using rsnapshot Date: Wed, 9 Sep 2009 23:42:23 +0200 Message-ID: <20090909214223.GB2011@phenom2.trippelsdorf.de> References: <20090909133026.GA1965@phenom2.trippelsdorf.de> <4AA7BE69.7070406@redhat.com> <20090909172015.GA1980@phenom2.trippelsdorf.de> <4AA7E411.7070005@redhat.com> <20090909212925.GA2011@phenom2.trippelsdorf.de> <4AA81FB1.7000000@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org To: Eric Sandeen Return-path: Received: from smtp-out-144.synserver.de ([212.40.180.144]:1057 "HELO smtp-out-144.synserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1752924AbZIIVmX (ORCPT ); Wed, 9 Sep 2009 17:42:23 -0400 Content-Disposition: inline In-Reply-To: <4AA81FB1.7000000@redhat.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Wed, Sep 09, 2009 at 04:35:45PM -0500, Eric Sandeen wrote: > Markus Trippelsdorf wrote: > > On Wed, Sep 09, 2009 at 12:21:21PM -0500, Eric Sandeen wrote: > >> Markus Trippelsdorf wrote: > >>> On Wed, Sep 09, 2009 at 09:40:41AM -0500, Eric Sandeen wrote: > >>>> Markus Trippelsdorf wrote: > >> ... > >> > >>>>> The bug is very simple to reproduce here. Just run rsnapshot several > >>>>> times per day and you will eventually hit the problem. > >>>> Could you provide a bzip2'd e2image -r of a corrupted filesystem for > >>>> analysis? > >>> OK I've uploaded the file (~80 MB): > >>> http://www.2shared.com/file/7681344/d19d7154/disc_e2image.html > >>> > >>> Hope it helps. > >> Thanks, can't guarantee that the post-mortem will lead to the anwer but > >> it's worth a look. > > > > I also took a look myself and it turned out that my maildir, which > > contains the LKML messages of the last three months, is the root of the > > problem. I can now reproduce the bug, by simply running (e.g.): > > > > cp -al /var/backup/hourly.3/localhost/home/markus/.maildir/lkml/. /var/tmp/debug > > > > where /var is mounted as: "/dev/sda1 on /var type ext4 (rw,noatime,commit=60)" > > and /var/backup/hourly.3/localhost/ is a previous backup produced by rsnapshot. > > From the original post it wasn't quite clear; do you hit runtime > corruption, indicated by the kernel logs when the fs is mounted, and a > subsequent fsck also finds errors? > > Just for completeness, if there are errors in the system logs, can you > post them here as well? There are no errors in the system logs. I only learned about this bug when a regular boot-time fsck was running that printed the problems mentionted above. -- Markus