Ocfs2 users - Jan 2007 - ocfs2_truncate_file kernel BUG in Fedora Core 5 2.6.16-1.2133_FC5smp update kernel

dear all,
don't know if it can help, anyway:

faulty node OS: Fedora Core release 5
KERNEL: Linux rack9 2.6.16-1.2133_FC5smp #1 SMP Tue Jun 6 01:52:09 EDT 2006
i686 i686 i386 GNU/Linux
CLUSTER: 11 Linux kernels, mixed environment FC4,FC5,FC6
SAN: FC Infortrend storage, QLogic16 port  FC switch, FC adapter LSI FC929X



(2600,0):dlm_send_remote_convert_request:393 ERROR: status = -107
(2600,0):dlm_wait_for_node_death:285 41AE1AA4C5534E50A93784D2AD94A94D:
waiting 5000ms for notification of death of no
de 3
(2600,0):dlm_send_remote_convert_request:393 ERROR: status = -107
(2600,0):dlm_wait_for_node_death:285 41AE1AA4C5534E50A93784D2AD94A94D:
waiting 5000ms for notification of death of no
de 3
(3294,0):ocfs2_dlm_eviction_cb:119 device (8,1): dlm has evicted node 3
(3294,1):o2net_set_nn_state:415 connected to node theboss (num 3) at
10.0.2.20:7777
(3294,1):__dlm_print_nodes:377 Nodes in my domain
("41AE1AA4C5534E50A93784D2AD94A94D"):
(3294,1):__dlm_print_nodes:381  node 1
(3294,1):__dlm_print_nodes:381  node 2
(3294,1):__dlm_print_nodes:381  node 3
(3294,1):__dlm_print_nodes:381  node 4
(3294,1):__dlm_print_nodes:381  node 5
(3294,1):__dlm_print_nodes:381  node 6
(3294,1):__dlm_print_nodes:381  node 7
(3294,1):__dlm_print_nodes:381  node 8
(3294,1):__dlm_print_nodes:381  node 9
(3294,1):__dlm_print_nodes:381  node 10
(3294,1):__dlm_print_nodes:381  node 11
...
(18315,0):ocfs2_truncate_file:240 ERROR: bug expression:
le64_to_cpu(fe->i_size) != i_size_read(inode)
(18315,0):ocfs2_truncate_file:240 ERROR: Inode 1029752381, inode i_size 586 !=
di i_size = 408, i_flags = 0x1
------------[ cut here ]------------
kernel BUG at fs/ocfs2/file.c:240!
invalid opcode: 0000 [#1]
SMP
last sysfs file: /o2cb/interface_revision
Modules linked in: nfs ocfs2 ocfs2_dlmfs ocfs2_dlm ocfs2_nodemanager
configfs nfsd exportfs lockd nfs_acl ipv6 autofs4 hidp rfcomm l2cap
bluetooth sunrpc video button battery ac lp parport_pc parport floppy nvram
sd_mod sg uhci_hcd mptfc mptscsih mptbase scsi_transport_fc scsi_mod tg3
e100 mii i2c_i801 i2c_core e7xxx_edac edac_mc dm_snapshot dm_zero dm_mirror
dm_mod ext3 jbd
CPU:    0
EIP:    0060:[<f8cde912>]    Not tainted VLI
EFLAGS: 00010286   (2.6.16-1.2133_FC5smp #1)
EIP is at ocfs2_setattr+0x6c3/0x125a [ocfs2]
eax: 00000073   ebx: 00000000   ecx: eb7fadec   edx: f8d02e22
esi: 0000024a   edi: 00000000   ebp: cd84d000   esp: eb7fae18
ds: 007b   es: 007b   ss: 0068
Process cp (pid: 18315, threadinfo=eb7fa000 task=f7d1b9b0)
Stack: <0>00000000 eb7fae9c f5103800 00000000 00000000 00000001 dce9a3a8
dce9a558
       00000198 00000000 f5103800 f62fa7f0 00000198 00000000 00000000
00000000
       00000000 f62fa7f0 dce9a3a8 eb7fae9c dce9a3a8 00000068 c017b355
eaa11490
Call Trace:
 [<c017b355>] notify_change+0x14a/0x2f0     [<c0162b81>]
do_truncate+0x52/0x68
 [<c0170b38>] may_open+0x1b0/0x1fd     [<c01729c6>]
open_namei+0x225/0x54c
 [<c01628a7>] do_filp_open+0x1c/0x31     [<c01628fa>]
do_sys_open+0x3e/0xaa
 [<c0162993>] sys_open+0x16/0x18     [<c0103db9>]
syscall_call+0x7/0xb
Code: fc ff ff ff b1 d8 fc ff ff 68 f0 00 00 00 68 56 e5 cf f8 ff 70 10 8b
00 ff b0 9c 00 00 00 68 22 2e d0 f8 e8 ca 56 44 c7 83 c4 30 <0f> 0b f0 00
20
29 d0 f8 8b 55 20 8b 4d 24 39 4c 24 3c 0f 82 ca
Continuing in 1 seconds.


-- 
davide.rossetti@gmail.com ICQ:290677265 SKYPE:d.rossetti
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://oss.oracle.com/pipermail/ocfs2-users/attachments/20070130/1d68d387/attachment.html

Hi,

On Tue, Jan 30, 2007 at 04:11:35PM +0100, davide rossetti
wrote:
> (18315,0):ocfs2_truncate_file:240 ERROR: bug expression:
le64_to_cpu(fe->
> i_size) != i_size_read(inode)
> (18315,0):ocfs2_truncate_file:240 ERROR: Inode 1029752381, inode i_size =
586 !
> = di i_size = 408, i_flags = 0x1
This is almost certainly one of the dlm lvb bugs which was fixed since then.
Is there any chance you can try a 2.6.19 kernel?

You should also run fsck.ocfs2 -f against the volume, just to be safe.
	--Mark

--
Mark Fasheh
Senior Software Developer, Oracle
mark.fasheh@oracle.com