XFS hang during xfs_fsr run

[Michael Weissenbacher]

> If you've got the inode numbers, then your running with the verbose
> flag set? Do you still have the logs for those inodes that it hung
> on?
Yes I am running xfs_fsr with the -v flag. But I'm not 100% sure if the 
log was truncated since it resides in /var which locked up. Here are the 
last log entries before the oopses:
ino=134269708  (this inode was /var/log/xfs_fsr.log)
(hang)
...
ino=277040401  (this inode was /var/spool/imap/x/user/xxxx/cyrus.cache)
(hang)

I can pin down those things at the moment:
- Both times it was a "hot" file - in one case a logfile, in the other 
case a database file.
- Usually it should say "file busy" and continue but sometimes it 
doesn't and just hangs the filesystem + oopses the kernel.
- It happens randomly, if i rerun xfs_fsr after the hang it usually goes 
over the "problem" file without a hickup.
- On /var/log/xfs_fsr.log it hung even though the (+f) chattr was set.

> xfs_fsr doesn't do directory traversals to find files for defrag -
> it uses more efficient bulkstat+open-by-handle method to visit every
> inode in the filesystem once. As a result, it will still open inodes
> that have the nodefrag flag set on them, but will then ignore them once
> it finds the flag is set.
Yes, usually it says "marked as don't defrag, ignoring" but this one 
time it hung.

> A trace would tell us which one it was....
Will see that i can get another one.

cheers,
Michael