[Bug 11653] Oops resulting from NFSD/XFS interaction?

[bugme-daemon at bugzilla.kernel.org]

http://bugzilla.kernel.org/show_bug.cgi?id=11653


david at fromorbit.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|xfs-masters at oss.sgi.com     |david at fromorbit.com
             Status|NEW                         |ASSIGNED




------- Comment #4 from david at fromorbit.com  2008-09-27 18:26 -------
Joshua,

I might as well close this bug now, what you've done will have removed
any trace of the problem that caused the filesystem to shut down.

In future:

    - the NFS server holds references to the filesystem; you need to
      unexport it or shut down the NFS server to be able to unmount it.
    - xfs_check or xfs_repair can't correctly until the journal has been
      replayed. Hence after unmounting, you need to remount it to get
      journal replay to occur, then unmount it again and run xfs_check
    - xfs_repair -L is _dangerous_. It will cause transactions that can
      be safely replayed to be tossed away, and guarantees that xfs_repair
      finds inconsistencies in the filesystem. This typically hides whatever
      problem caused the shutdown - it is a repair method that should only
      be used when the journal is corrupted.

If you don't know how to handle the failure properly - ask us. We can
tell you exactly what you need to do to quickly recover the filesystem
and to do so in a manner that is also helpful to us in tracking down
the bug that caused the shutdown.

That being said, if it was a corrupted inode btree block, then we've probably
just fixed the last known occurrence of this and it should be available in
2.6.27-rc8....


-- 
Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
You are the assignee for the bug, or are watching the assignee.