> I've been using XFS on numerous systems for about 2 years now, and I've
> recently been bitten by a problem I've seen once or twice before, and
> was wondering if anyone out there knows what the real cause of this is.
> Machine is a fairly heavily used server, hardware RAID-5 on 7 disks,
> running Debian Woody with a vanilla kernel from kernel.org, patched with
> the appropriate xfs-all (not the split) patch set. Currently:-
> SGI XFS snapshot-2.4.22-2003-10-10_04:57_UTC with no debug enabled
> The most heavily used filesystem is on /var, where there is a Cyrus imap
> spool, resulting in lots of small files, and a lot of file activity.
> There is a bug in the version of Cyrus I'm running which periodically
> causes imapd child processes to sig11 and die. This may be entirely
> unrelated, but could conceivably have some bearing on the situation.
I've been running cyrus-imapd servers on XFS for years now without any
problems related to XFS. I also had the sig11 and die problem but it never
affected XFS, and I think it really should not.
> Last week, I get this:-
> xfs_inotobp: xfs_imap() returned an error 22 on sd (8,8). Returning error.
> xfs_iunlink_remove: xfs_inotobp() returned an error 22 on sd (8,8).
> Returning an error.
> xfs_inactive: 0xfs_ifree() error 22 on sd (8,8)
> xfs_force_shutdown: (sd(8,8)0x1) called from line 1873 of file
> Return address = 0x01ef8ba
> File system sd (8,8): I/O error detected.
I'm not an expert for those error messages but I guess it unfortunately a
hardware error, isn't it? Did you check dmesg output when this happened?
> Shutting down file system: sd (8,8)
> Please umount the fs, & rectify the problem(s)
> Fixed by taking the machine single user, running xfs_repair over /var,
> and then remounting the file-system.
> This has happened to this machine twice in a period of about 14 months,
> and while it hasn't caused me serious trouble, it's evident from the XFS
> FAQ that it shouldn't be happening, and I'd like to know why it is, and
> if there's anything I can do about it. There's a bug report at
> http://oss.sgi.com/bugzilla/show_bug.cgi?id=274 which looks as though it
> may be the same thing I'm seeing.
> I'd be interested to hear from anyone who can shed any more light on the