[DRBD-user] XFS filesystem locks on mounting with 0.8.0.5

Laurent Wacrenier lwa at noc.oxyd.net
Mon Aug 27 17:01:04 CEST 2007

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Hi,

We're experiencing a file system lock with DRBD and XFS.
When mouting the filesystem, the kernel logs
  Starting XFS recovery on filesystem: drbd0 (logdev: internal)
but nothing happens and the disk reports no IO.

The same behaviour occurs when we turn the secondary as primary and
try to mount the filesystem.

I've ran strace on "xfs_repair -L" and it freezes when zeroing the XFS
log file.

The partion is avaiable for reading: I've been able to read it
without problem. Also, the filesystem mounts with the "ro,norecovery"
mount options, but not with "ro" only.

I can also mount "ro" the disk device (/dev/sda3), and have been able
to mount "rw" a dump of the filesystem. It's only using the drdb layer
the bug occurs.

The volume itself has 350Gb but only 80Gb are used by now.

The bug started to occurs with Linux 2.6.22 and drdb 0.8.0.3. I upgraded
to 2.6.22.5 and 0.8.0.5 but nothing changed.

Does someone has a suggestion?

Thanks,

Laurent Wacrenier




More information about the drbd-user mailing list