http://oss.sgi.com/bugzilla/show_bug.cgi?id=184
Summary: xfs_force_shutdown called on lvm-snapshot
Product: Linux XFS
Version: Current
Platform: All
OS/Version: Linux
Status: NEW
Severity: major
Priority: High
Component: XFS kernel code
AssignedTo: xfs-master@xxxxxxxxxxx
ReportedBy: sgi@xxxxxxxxxxxxxxxxx
I downloaded a kernel from cvs today at round about 12:00 +0200 o'clock. Then I
applied lvm and lvm-snapshot-patches from the 1.0.5 lvm release. I also applied
lm_sensor_patches, but I think they are not the probblem.
When I tried to backup my system from 3 snapshots I got the following
log-message:
Oct 18 15:52:51 boss kernel: Ending clean XFS mount for filesystem: lvm(58,4)
Oct 18 15:53:01 boss kernel: XFS mounting filesystem lvm(58,3)
Oct 18 15:53:01 boss kernel: Ending clean XFS mount for filesystem: lvm(58,3)
Oct 18 15:53:07 boss kernel: XFS mounting filesystem lvm(58,2)
Oct 18 15:53:08 boss kernel: Ending clean XFS mount for filesystem: lvm(58,2)
This was when I mounted my snapshots.
Oct 18 15:57:42 boss kernel: lvm - lvm_map: ll_rw_blk write for readonly LV
/dev/master/var-snap
Oct 18 15:57:42 boss last message repeated 7 times
Oct 18 15:57:42 boss kernel: I/O error in filesystem ("lvm(58,3)") meta-data dev
0x3a03 block 0x3013c4^I ("xlog_iodo
ne") error 5 buf count 32768
Oct 18 15:57:42 boss kernel: xfs_force_shutdown(lvm(58,3),0x2) called from line
949 of file xfs_log.c. Return addre
ss = 0xc01fbd9d
Oct 18 15:57:42 boss kernel: Log I/O Error Detected. Shutting down filesystem:
lvm(58,3)
Oct 18 15:57:42 boss kernel: Please umount the filesystem, and rectify the
problem(s)
And this was some minutes after, 58,3 is my /var-snapshot. Tar was backupping
this area at this moment.
I would suggest try to install a system on lvm with root-fs on lvm and /home and
/var on seperate partitions (the best would be debian-woody). Then create
snapshots and then try to backup them with tar or so. Trying to snapshot a
seperated lv would perhaps not be a good idea to reproduce it because perhaps
this bug depends on IO-activity on the source of the snapshot.
A backtrace on every
lvm - lvm_map: ll_rw_blk write for readonly LV /dev/master/var-snap
could be interesting too.
If you can reproduce this with lvm, It would be interesting to know if this
happens with evms-ro-snapshots too. This could help to fix the problem too.
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
|