[Linux-cluster] Kernel oops when GFS2 used as localfs

Sridhar Ramaswamy (srramasw) srramasw at cisco.com
Wed Mar 14 00:25:14 UTC 2007


I hit various kernel oops while running some metadata intensive test on
GFS2 filesystem. This volume is intended to be used as local fileystem.
So it was created using lock_nolock. Also no cman, fence and any such
things are running in the system.
 
RHEL4 distro
Linux kernel 2.6.20.1 (from kernel.org)
cluster-2.00.00 (from tarball)
 
David.T earlier suggested to move to newer kernel (2.6.21-rc), which I'm
planning to do. Meanwhile I want to poll the alias on such issues in
GFS2. Ofcourse I'm all ears on any other ideas to resolve this!
 
thanks,
Sridhar
 
$time bonnie++ -u 99 -f -x 3 -d /mnt/gfs2/bonnie -s 0 -n 30:4096:4096:30
 
(1) 
 
Mar 13 15:25:56 cfs1 kernel: ------------[ cut here ]------------
Mar 13 15:25:56 cfs1 kernel: kernel BUG at fs/gfs2/meta_io.c:474!
Mar 13 15:25:56 cfs1 kernel: invalid opcode: 0000 [#1]
Mar 13 15:25:56 cfs1 kernel: SMP
Mar 13 15:25:56 cfs1 kernel: Modules linked in: lock_nolock gfs2
reiserfs nfsd exportfs nfs lockd nfs_acl ipv6 parport_pc lp parport
autofs4 sunrpc dm_mirror dm_mod button battery ac uhci_hcd ehci_hcd
intel_rng rng_core i2c_i801 i2c_core e1000 e100 mii floppy ext3 jbd
Mar 13 15:25:56 cfs1 kernel: CPU:    1
Mar 13 15:25:56 cfs1 kernel: EIP:    0060:[<e0c88da5>]    Not tainted
VLI
Mar 13 15:25:56 cfs1 kernel: EFLAGS: 00010246   (2.6.20.1 #1)
Mar 13 15:25:56 cfs1 kernel: EIP is at
gfs2_meta_indirect_buffer+0x4c/0x278 [gfs2]
Mar 13 15:25:56 cfs1 kernel: eax: 00000000   ebx: 00012bf5   ecx:
ce5a6dd4   edx: dc5eae00
Mar 13 15:25:56 cfs1 kernel: esi: 00000000   edi: 00000000   ebp:
dc5ea9a8   esp: ce5a6d58
Mar 13 15:25:56 cfs1 kernel: ds: 007b   es: 007b   ss: 0068
Mar 13 15:25:56 cfs1 kernel: Process bonnie++ (pid: 5509, ti=ce5a6000
task=dc320570 task.ti=ce5a6000)
Mar 13 15:25:56 cfs1 kernel: Stack: c156d274 ce5a6dd4 c016eb91 ce5a6dd4
00000000 dc5eae00 00000000 d6f34000
Mar 13 15:25:56 cfs1 kernel:        00000000 00000000 dc5ea9a8 d57794a8
ce5a6e08 e0c83ad3 00012bf5 00000000
Mar 13 15:25:56 cfs1 kernel:        00000000 ce5a6da0 00000000 00000000
dc5ea9a8 d57794a8 e0c84cc9 ce5a6dd4
Mar 13 15:25:56 cfs1 kernel: Call Trace:
Mar 13 15:25:56 cfs1 kernel:  [<c016eb91>] iget5_locked+0x3d/0x67
Mar 13 15:25:56 cfs1 kernel:  [<e0c83ad3>] gfs2_inode_refresh+0x34/0xfe
[gfs2]
Mar 13 15:25:56 cfs1 kernel:  [<e0c84cc9>] gfs2_createi+0x12c/0x191
[gfs2]
Mar 13 15:25:56 cfs1 kernel:  [<e0c8da3c>] gfs2_create+0x5c/0x103 [gfs2]
Mar 13 15:25:56 cfs1 kernel:  [<e0c84be7>] gfs2_createi+0x4a/0x191
[gfs2]
Mar 13 15:25:56 cfs1 kernel:  [<e0c820c4>] gfs2_glock_nq_num+0x3f/0x64
[gfs2]
Mar 13 15:25:56 cfs1 kernel:  [<c016552e>] vfs_create+0xc3/0x126
Mar 13 15:25:56 cfs1 kernel:  [<c01657f2>] open_namei_create+0x47/0x88
Mar 13 15:25:56 cfs1 kernel:  [<c016597d>] open_namei+0x14a/0x539
Mar 13 15:25:56 cfs1 kernel:  [<c015d27b>] do_filp_open+0x25/0x39
Mar 13 15:25:56 cfs1 kernel:  [<c01d200a>] strncpy_from_user+0x3c/0x5b
Mar 13 15:25:56 cfs1 kernel:  [<c015d431>] get_unused_fd+0xa8/0xb1
Mar 13 15:25:56 cfs1 kernel:  [<c015d509>] do_sys_open+0x42/0xbe
Mar 13 15:25:56 cfs1 kernel:  [<c015d59f>] sys_open+0x1a/0x1c
Mar 13 15:25:56 cfs1 kernel:  [<c015d5dd>] sys_creat+0x1f/0x23
Mar 13 15:25:56 cfs1 kernel:  [<c0103410>] sysenter_past_esp+0x5d/0x81
Mar 13 15:25:56 cfs1 kernel:  =======================
Mar 13 15:25:56 cfs1 kernel: Code: 80 a8 01 00 00 89 44 24 1c 8b 85 f0
01 00 00 c7 44 24 20 00 00 00 00 89 54 24 14 85 c0 89 44 24 18 c7 44 24
10 00 00 00 00 75 04 <0f> 0b eb fe 83 7c 24 1c 00 75 04 0f 0b eb fe 8d
85 24 04 00 00
Mar 13 15:25:57 cfs1 kernel: EIP: [<e0c88da5>]
gfs2_meta_indirect_buffer+0x4c/0x278 [gfs2] SS:ESP 0068:ce5a6d58
 
 
(2) 
 
Mar 13 17:00:30 cfs1 kernel: Call Trace:
Mar 13 17:00:30 cfs1 kernel:  [<e0badde1>] gfs2_unlink+0x53/0xe0 [gfs2]
Mar 13 17:00:30 cfs1 kernel:  [<e0baddc8>] gfs2_unlink+0x3a/0xe0 [gfs2]
Mar 13 17:00:30 cfs1 kernel:  [<e0badde1>] gfs2_unlink+0x53/0xe0 [gfs2]
Mar 13 17:00:30 cfs1 kernel:  [<c0166572>] vfs_unlink+0xa1/0xc5
Mar 13 17:00:30 cfs1 kernel:  [<c016662b>] do_unlinkat+0x95/0xf5
Mar 13 17:00:30 cfs1 kernel:  [<c01187de>] scheduler_tick+0x8f/0x95
Mar 13 17:00:30 cfs1 kernel:  [<c0103410>] sysenter_past_esp+0x5d/0x81
Mar 13 17:00:30 cfs1 kernel:  =======================

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20070313/f64de53d/attachment.htm>


More information about the Linux-cluster mailing list