Has anything changed in the Linux xfs implementation between 2.6.12 and
2.6.16 in relation to log recovery on mount? With 2.6.12 and below a mount
of a dirty partition would undergo a recovery and then continue mounting
successfully. The same partition (dirty) booted with a 2.6.16 image compiled
today gives the error on mount:
XFS mounting filesystem hda7
Starting XFS recovery on filesystem: hda7 (logdev: internal)
XFS: xlog_recover_process_data: bad clientid
XFS: log mount/recovery failed: error 5
XFS: log mount failed
An xfs_repair fails with "journal contains valuable metadata, please mount
first". When I reboot with 2.6.12 kernel the partition mounts correctly.
It seems 2.6.16 is less tolerant to unclean partition shutdowns and
subsequence mounts. Is there a reason for this?
|