After 2.4.21 kernel release I wanted to try it out on some of the machines
here. For that I checked out CVS (SGI-XFS CVS-2003-06-16_05:00_UTC with
no debug enabled) compiled and booted. After 20 hours of uptime the
machine had to be rebooted hard (well ,that was actually a mistake but
thats not the point) by unplugging it. After power on, it booted until the
first "big" XFS filesystem recovery were it hanged (disk activity on that
fs stopped after some seconds). When I say big I mean a ~140 gb partition
(before that one, there was another XFS partitions which didnt had the
hanging problem but which is a lot smaller, 16gb, and also which has
internal journal different from the one were it hangs which has external
jurnal). I have tried many kernels from the lilo boot menu and no one
succeded except for the 2.4.9-34 kernel (contributed kernel from the 1.1
release dir). The kernels that I have tried and not succeded were
2.4.21-cvs (the version mentioned above) and 2.4.18-18SGI_XFS_1.2.0. I
also mention that all this kernels I have compiled them myself using gcc
Another strange thing is that in the kernel logs I couldnt find the boot
messages of the boot tries before that last one (2.4.9-34) that worked.
When 2.4.9-34 booted also gave me some interesting kernel logs:
Jun 17 15:41:08 s1 kernel: Starting XFS recovery on filesystem:
dac960(48,9) (dev: 8/6)
Jun 17 15:41:08 s1 kernel: xfs_inotobp: xfs_imap() returned an error 22
on dac960(48,9). Returning error.
Jun 17 15:41:08 s1 kernel: xfs_iunlink_remove: xfs_inotobp() returned an error
22 on dac960(48,9). Returning error.
Jun 17 15:41:08 s1 kernel: xfs_inactive: xfs_ifree() returned an error =
22 on dac960(48,9)
Jun 17 15:41:08 s1 kernel: xfs_force_shutdown(dac960(48,9),0x1) called
from line 1962 of file xfs_vnodeops.c. Return address = 0xc01cf242
Jun 17 15:41:08 s1 kernel: I/O Error Detected. Shutting down filesystem:
Jun 17 15:41:08 s1 kernel: Please umount the filesystem, and rectify the
Jun 17 15:41:08 s1 kernel: Ending XFS recovery on filesystem: dac960(48,9)
After 2.4.9-34 booted I rebooted the machine (software) and it booted
2.4.29-cvs just fine.
Because I never had problems with that machine before (hardware problems)
and because it did it only on XFS recovery I presume its a XFS bug ?
I would like to know what can I do to make sure it doesnt happen again.
Disclaimer: Any views or opinions presented within this e-mail are solely
those of the author and do not necessarily represent those of any company,
unless otherwise specifically stated.