ocfs2 error in /var/log/messages

hi there,

we are facing the situation that our shared ocfs2 filesystem sometimes crashes on one of the 6 nodes. all processes using files on that fileshare are hanging. the system itself is useable. any ideas how to resolve that issue?! see the following code snippet from the log:

Jul 28 08:13:21 linux kernel: <3>(11902,5):ocfs2_meta_lock_update:1448 ERROR: bug expression:
inode->i_generation != le32_to_cpu(fe->i_generation)
Jul 28 08:13:21 linux kernel: (11902,5):ocfs2_meta_lock_update:1448 ERROR: Invalid dinode 82200258 disk generation: 3296561854 inode ->i_generation: 3296549019
Jul 28 08:13:21 linux kernel: ------------ cut here ------------
Jul 28 08:13:21 linux kernel: kernel BUG at fs/ocfs2 dlmglue.c:1448!
Jul 28 08:13:21 linux kernel: invalid operand: 0000 #4]
Jul 28 08:13:21 linux kernel: SMP
Jul 28 08:13:21 linux kernel: CPU: 5
Jul 28 08:13:21 linux kernel: EIP: 0060:<fa43afc0>] Tainted: PX
Jul 28 08:13:21 linux kernel: EFLAGS: 00010282 (2.6.5-7.312-bigsmp SLES9_SP4_BRANCH-200806061544330200)
Jul 28 08:13:21 linux kernel: EIP is at
ocfs2_meta_lock_full+0x3e0/0x1180 [ocfs2]
Jul 28 08:13:21 linux kernel: eax: 00000084 ebx: e1336680 ecx:
c03bbef4 edx: 0000f624
Jul 28 08:13:21 linux kernel: esi: fa46fdf6 edi: 000005a8 ebp:
d1568100 esp: d95ffdd8
Jul 28 08:13:21 linux kernel: ds: 007b es: 007b ss: 0068
Jul 28 08:13:21 linux kernel: Process java (pid: 11902,
threadinfo=d95fe000 task=e1336680)
Jul 28 08:13:21 linux kernel: Stack: fa474470 00002e7e 00000005
fa46fdf6 000005a8 04e646c2 00000000 c47d8abe
Jul 28 08:13:21 linux kernel: c47d589b d156830c d6868118
d6ae2000 d1568108 f54f686c f54f686c d1568108
Jul 28 08:13:21 linux kernel: 00000001 00000000 00000000
d156830c ef2da2b8 00000000 00000000 da46f034
Jul 28 08:13:21 linux kernel: Call Trace:
Jul 28 08:13:21 linux kernel: <c0185aff>] real_lookup+0xbf/0xf0
Jul 28 08:13:21 linux kernel: <fa445118>]
ocfs2_inode_revalidate+0x128/0x2f0 [ocfs2]
Jul 28 08:13:21 linux kernel: <c019154f>] dput+0x6f/0x1b0
Jul 28 08:13:21 linux kernel: <fa43fa69>] ocfs2_getattr+0x49/0x1f0 [ocfs2]
Jul 28 08:13:21 linux kernel: <fa43fa20>] ocfs2_getattr+0x0/0x1f0 [ocfs2]
Jul 28 08:13:21 linux kernel: <c0180a64>] vfs_getattr_it+0x54/0x120
Jul 28 08:13:21 linux kernel: <c01898b9>] __user_walk_it+0x59/0x70
Jul 28 08:13:21 linux kernel: <c018131b>] vfs_lstat+0x8b/0x140
Jul 28 08:13:21 linux kernel: <c01813df>] sys_lstat64+0xf/0x30
Jul 28 08:13:21 linux kernel: <c01091b9>] sysenter_past_esp+0x52/0x71
Jul 28 08:13:21 linux kernel:
Jul 28 08:13:21 linux kernel: Code: 0f 0b a8 05 b8 f8 46 fa 8b 4c 24 2c 8b 51 4c 8b 41 48 89 d3

thanks for answers, helge.

You should probably go ahead and report this bug on Novell’s bugzilla, it looks bad.

I have to admit I have an interest in seeing this bug fixed as I will be upgrading some OCFS2 systems later in the year.