Re: fatal error -- xfs_repair: duplicate inode range

[Steve Lord <lord@sgi.com>]

On Mon, 2002-03-18 at 11:36, Michael Best wrote:
> It appears that either the mailing list manager threw away my earlier
> posting to the mailing list or that there is no hope for recovery of my
> filesystem.

Or people took the weekend off.

You need a newer xfs_repair, and I suspect a newer kernel. You have
corruption in your filesystems, and older kernels can write over
block zero if they shutdown.

Steve

> 
> Any pointers would be appreciated.  ssh access to this machine is
> available.  
> 
> Trying to get a dump of the filesystem right now with dd.
> 
> -----
> 
> On the failed system and on this new system I am running the SGI_XFS
> 1.0.2 +
> Redhat 7.2 installer.
> 
> -----
> 
> I was called Sunday morning for a machine that had failed.
> 
> You could ping it from the network but nothing else.
> 
> I rebooted it, and got a XFS Page alloc kernel panic I believe. (unsure
> it's been hours since then).
> 
> I rebooted it in single user mode, read some logfiles and then rebooted
> it into multi user mode.
> 
> A short while later the machine appeared to fail again, when rebooted it
> could no longer load the initrd in order to mount the root (/)
> filesystem to boot.
> 
> I used a Mandrake 8.2rc1 disc to boot the system, after finding the
> mount command on the 1.0.2 installer disc was hopeless.
> 
> I ran XFS repair, which suggested something good had happened.  And then
> it suggested mounting it.  No luck.  It suggested I could use xfs_repair
> -L to try and mount it after that.
> 
> I tried that, no luck.  All further attempts produce output like this: 
> 
> fatal error -- xfs_repair:  duplicate inode range
> 
> Here are the logs from xfs_check and xfs_repair for this filesystem.
> 
> http://www.emergence.com/xfs_trouble/xfs_check_sda5.txt
> http://www.emergence.com/xfs_trouble/xfs_repair_sda5.txt
> 
> The machine was originally running a single Quantum Atlas V on Ultra 160
> Scsi chain.  I have moved this machine to two Maxtor 20G drives with
> Raid 1 mirroring.  The Scsi chain is still in the machine with the
> broken filesystem drive and I can make it available/prepare a filesystem
> dump (perhaps someone can explain how to do that).  Hopefully I have
> enough space on my drive for that.
> 
> I saw other people with this error in the mailing list archives but
> didn't see if there was a solution to this problem.
> 
> -- 
> Michael Best
> Systems Administrator           ph 780-413-6397 x230
> Emergence By Design            fax 780-433-7548
> #200, 11209 Jasper Avenue     toll 866-860-2666
> Edmonton, Alberta, T5K 0L5
-- 

Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@sgi.com