From: djani22@dynamicweb.hu
Message-ID: <014d01c59c28$744dae40$0400a8c0@LocalHost>
To: <linux-xfs@oss.sgi.com>
Subject: Fw: XFS repair problem.
Date: Mon, 8 Aug 2005 16:49:54 +0200
MIME-Version: 1.0
Content-Type: text/plain;
	charset="UTF-8"
Content-Transfer-Encoding: 7bit
Sender: linux-xfs-bounce@oss.sgi.com
Errors-to: linux-xfs-bounce@oss.sgi.com
Precedence: bulk
Status: O
Content-Length: 3548
Lines: 140


----- Original Message -----
From: <djani22@dynamicweb.hu>
To: "evilninja" <evilninja@gmx.net>
Sent: Monday, August 08, 2005 3:55 PM
Subject: Re: XFS repair problem.


> ----- Original Message -----
> From: "evilninja" <evilninja@gmx.net>
> To: <linux-xfs@oss.sgi.com>
> Cc: <djani22@dynamicweb.hu>
> Sent: Monday, August 08, 2005 2:09 PM
> Subject: Re: XFS repair problem.
>
>
> > djani22@dynamicweb.hu schrieb:
> > > Hello all!
> > >
> > > I have a "big" filesystem (8TB), and a BIG problem!
> > > I can't fix it!
> > >
> > > [root@dy-base xfsprogs-2.5.6]# xfs_check /dev/md31
> > -------------------------^
> >
> > please use current version of xfsprogs:
> >
> > ftp://oss.sgi.com/projects/xfs/cmd_tars/xfsprogs-2.6.36.src.tar.gz
>
> OK, I try it!
>
> >
> > > xfs_check: out of memory
> > > [root@dy-base xfsprogs-2.5.6]# free
> > >              total       used       free     shared    buffers
> cached
> > > Mem:       4149468    4015820     133648          0         16
> 3718360
> > > -/+ buffers/cache:     297444    3852024
> > > Swap:            0          0          0
> >
> > does it help, if you enable some swapspace?
>
> I can't.
> No raw disks in this system.
> Only the swap over the network is the possible way, but I dont think
so....
>
> >
> > > But, in this time, the xfs_repair is fails ALWAYS too!
> > > I have tryed  it 23 times!
> > >
> > > The different results:
> > >
> > > 1. Simple segfault.
> >
> > did you check for bad memory? an overnight run of memtest86+ could
tell...
>
> No, but this is a 14 day system, and before I start it, I have tested it,
> witth this test. :-)
>
> In the system is 4GB DDR2 -ECC ram.
> I don't think, it is the problem.... :-)
>
> And additionaly, only the xfs_repair generate segfaults...
>
>
> >
> > > The data on fs and the service is VERY important for me.
> >
> > hm, how did you *backup* 8TB of data anyway?
>
> Yes, you have right!
> I don't use backup.
>
> This is a free web storage.
> The service is more importanat, than data... :-/
> But I dont want to loose ~3000 webpages....
>
> >
> > > Dual xeon, 4G ram.
> > > The xfs is on RAID 0, 32k chunks
> >
> > any chance you can run top(1) next to xfs_repair to see how many memory
is
> > used?
>
> Max 945MB
> top line:
> 17595 root      18   0  945M 941M   764 R    31.4 23.2   0:44   3
xfs_repair
>
> The new version:
>
> 1. time
>
> Phase 6 - check inode connectivity...
>         - resetting contents of realtime bitmap and summary inodes
>         - ensuring existence of lost+found directory
>         - traversing filesystem starting at / ...
>         - traversal finished ...
>         - traversing all unattached subtrees ...
> rebuilding directory inode 4211081386
> xfs_repair: phase6.c:2603: longform_dir2_rebuild_finish: Assertion `error
==
> 0' failed.
>
> 2-3-4-5 times the end of  log is the same:
> rebuilding directory inode 4211081386
> xfs_repair: phase6.c:2603: longform_dir2_rebuild_finish: Assertion `error
==
> 0'failed.
>
> No segfault in this version.
>
> But the xfs_check's problem still here....
> [root@dy-base /]# /usr/local/bin/xfs_check /dev/md31
> xfs_check: out of memory
> top:
> Mem:  4149468k av,  192640k used, 3956828k free,       0k shrd,       0k
> buff
>         74608k active,              64068k inactive
> Swap:       0k av,       0k used,       0k free                  127160k
> cached
>
>
> How can I siply delete that inode? (4211081386)
>
> Thanks for all help!
>
> >
> > thanks,
> > Christian.
> > --
> > BOFH excuse #194:
> >
> > We only support a 1200 bps connection.
>