[Cluster-devel] OOps mounting gfs file system with recent kerel 2.6.18 and cvs stable snapshot of cluster

[Cluster-devel] OOps mounting gfs file system with recent kerel 2.6.18 and cvs stable snapshot of cluster

[Guido Aulisi]

Hi I upgraded to the last 2.6.18 kernel, and to the last STABLE cluster cvs tag.
I get an OOPS when I try to mount a GFS FILE SYSTEM.
I'm still running the other node of the cluster with 2.6.17.1 and it can mount the file system.
Do I need to go back to 2.6.17?
Thanks.
Guido Aulis

This is the message:

Sep 23 09:17:16 enterprise kernel: Lock_DLM (built Sep 22 2006 15:12:31) installed
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: Joined cluster. Now mounting FS...
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: jid=1: Trying to acquire journal lock...
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: jid=1: Looking at journal...
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: jid=1: Done
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: Scanning for log elements...
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: Found 0 unlinked inodes
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: Found quota changes for 0 IDs
Sep 23 09:17:18 enterprise kernel: GFS: fsid=Trek:oraclefs.1: Done
Sep 23 09:17:18 enterprise kernel: BUG: unable to handle kernel NULL pointer dereference at virtual address 00000018
Sep 23 09:17:18 enterprise kernel:  printing eip:
Sep 23 09:17:18 enterprise kernel: c017d939
Sep 23 09:17:18 enterprise kernel: *pde = 35a71001
Sep 23 09:17:18 enterprise kernel: Oops: 0000 [#1]
Sep 23 09:17:18 enterprise kernel: SMP
Sep 23 09:17:18 enterprise kernel: Modules linked in: lock_dlm gfs lock_harness dlm cman dm_round_robin ipv6 dm_multipath button battery ac e100 mii e1000 bonding floppy qla2xxx scsi_transport_fc mptspi st aic7xxx scsi_transport_spi mptscsih mptbase serverworks ide_generic dm_snapshot dm_zero dm_mirror ide_disk ide_core
Sep 23 09:17:18 enterprise kernel: CPU:    2
Sep 23 09:17:18 enterprise kernel: EIP:    0060:[<c017d939>]    Tainted: GF     VLI
Sep 23 09:17:18 enterprise kernel: EFLAGS: 00010293   (2.6.18 #1)
Sep 23 09:17:18 enterprise kernel: EIP is at do_add_mount+0x63/0xf2
Sep 23 09:17:18 enterprise kernel: eax: 0000000c   ebx: f58abf38   ecx: 0000000b   edx: c511e4c0
Sep 23 09:17:18 enterprise kernel: esi: f5a3d800   edi: ffffffea   ebp: 00000000   esp: f58abf08
Sep 23 09:17:18 enterprise kernel: ds: 007b   es: 007b   ss: 0068
Sep 23 09:17:18 enterprise kernel: Process mount (pid: 11865, ti=f58aa000 task=f5186ab0 task.ti=f58aa000)
Sep 23 09:17:18 enterprise kernel: Stack: 0000000b c0355520 00000000 f58abf38 00000000 c017df49 00000000 f5406000
Sep 23 09:17:18 enterprise kernel:        f5bf6000 0000000b f5f66000 f5406000 f571a714 c511e4c0 f5186ab0 f58abfa4
Sep 23 09:17:18 enterprise kernel:        c0148c57 00000001 00000001 00000000 f58abfac 00000000 00000010 c034ee80
Sep 23 09:17:18 enterprise kernel: Call Trace:
Sep 23 09:17:18 enterprise kernel:  [<c017df49>] do_mount+0x1ea/0x1fa
Sep 23 09:17:18 enterprise kernel:  [<c0148c57>] __alloc_pages+0x4a/0x2b8
Sep 23 09:17:18 enterprise kernel:  [<c017dd14>] copy_mount_options+0x48/0x93
Sep 23 09:17:18 enterprise kernel:  [<c017e1ed>] sys_mount+0x79/0xb0
Sep 23 09:17:18 enterprise kernel:  [<c0103775>] sysenter_past_esp+0x56/0x79
Sep 23 09:17:18 enterprise kernel: Code: bf ea ff ff ff 25 00 e0 ff ff 8b 00 8b 80 58 04 00 00 39 42 64 75 7d 8b 46 14 66 bf f0 ff 39 42 14 74 6a 8b 46 10 bf ea ff ff ff <8b> 40 0c 0f b7 40 28 25 00 f0 00 00 3d 00 a0 00 00 74 56 8b 04
Sep 23 09:17:18 enterprise kernel: EIP: [<c017d939>] do_add_mount+0x63/0xf2 SS:ESP 0068:f58abf08

[Cluster-devel] OOps mounting gfs file system with recent kerel 2.6.18 and cvs stable snapshot of cluster

[David Teigland]

On Sat, Sep 23, 2006 at 09:40:17AM +0200, Guido Aulisi wrote:

> Hi I upgraded to the last 2.6.18 kernel, and to the last STABLE cluster
> cvs tag.  I get an OOPS when I try to mount a GFS FILE SYSTEM.  I'm
> still running the other node of the cluster with 2.6.17.1 and it can
> mount the file system.  Do I need to go back to 2.6.17?

Very odd, I haven't seen that before.  I suppose something could have
snuck in to break it; we hardly run 2.6.18 and STABLE at all.  All our
testing is either related to the RHEL4 branch (with RHEL kernels) or HEAD
(with 2.6.18 kernels).

Dave