Yusuf Goolamabbas wrote:
Hi, Was reading Christophs and Nathan's email on lkml about using v2 XFS
log format when using software raid5 and then Nathan's message about
creating the filesystem with -ssize=4096
I've got a software MD RAID5 that was not created with that option. I
could've sworn that I read that mkfs.xfs would query the block size if
it was being created on a MD device and adjust accordingly.
In any case the logs fill up with this seems to confirm a block mismatch
(512 vs 4096).
May 25 11:17:13 [kernel] raid5: switching cache buffer size, 512 --> 4096
May 25 11:17:13 [kernel] raid5: switching cache buffer size, 4096 --> 512
May 25 11:17:14 [kernel] raid5: switching cache buffer size, 512 --> 4096
May 25 11:17:14 [kernel] raid5: switching cache buffer size, 4096 --> 512
May 25 11:17:14 [kernel] raid5: switching cache buffer size, 512 --> 4096
May 25 11:17:14 [kernel] raid5: switching cache buffer size, 4096 --> 512
May 25 11:17:14 [kernel] raid5: switching cache buffer size, 512 --> 4096
May 25 11:17:15 [kernel] raid5: switching cache buffer size, 4096 --> 512
May 25 11:17:15 [kernel] raid5: switching cache buffer size, 512 --> 4096
May 25 11:17:15 [kernel] raid5: switching cache buffer size, 4096 --> 512
----->%-----
What is the best course of action to fix this problem? Backup data,
reformat partition and restore data? Should I adjust the MD device to
match XFS or adjust XFS to match MD?
TIA!
ds
|