We upgraded kernel on all system of a cluster and have some troubles
with Apache 2. The architecture is composed by 4 NFS clients and 1 NFS
Server, all based on 3.8.8 kernel version ( before, it was 2.6.32 on all
systems and was running with no problem )
Sometimes and after having reboot on kernel 3.8.8, NFS Clients display
this trace in the dmesg :
[12100.793549] INFO: task apache2:8695 blocked for more than 120 seconds.
[12100.793577] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[12100.793621] apache2 D 0000000000000002 0 8695 7799
[12100.793623] ffff880215309d98 0000000000000086 ffff880215308010
[12100.793627] ffff880215086300 0000000000013940 ffff880215309fd8
[12100.793630] ffff880215309fd8 0000000000013940 ffff8802365418c0
[12100.793633] Call Trace:
[12100.793636] [<ffffffff815681f6>] schedule+0x64/0x66
[12100.793640] [<ffffffff8156843e>] schedule_preempt_disabled+0xe/0x10
[12100.793643] [<ffffffff81567190>] __mutex_lock_common+0x11d/0x18b
[12100.793646] [<ffffffff8113dc27>] ? filename_lookup+0x74/0x84
[12100.793649] [<ffffffff81567219>] __mutex_lock_slowpath+0x1b/0x1d
[12100.793652] [<ffffffff81566fd5>] mutex_lock+0x1b/0x2c
[12100.793655] [<ffffffff8113e76d>] do_unlinkat+0x92/0x231
[12100.793659] [<ffffffff81131d7e>] ? fsnotify_access+0x5d/0x65
[12100.793662] [<ffffffff81132f23>] ? sys_read+0x81/0x8e
[12100.793665] [<ffffffff8113e922>] sys_unlink+0x16/0x18
[12100.793668] [<ffffffff81570259>] system_call_fastpath+0x16/0x1b
The NFS Server exports a filesystem formatted in ReiserFS ( we mean it
was the problem ) and we migrate to ext4, the problem is always present.
I have one other architecture with a NFS Server based on a 2.6.24 Kernel
and some clients with a 3.8.8 Kernel, the problem doesn't appear.
We made some tests with a 3.2 kernel and it was the same problem.
Can you give me some help ?
Thanks a lot.
Lilian RIGARD - Devclic
G?rant - CEO & CTO
E-mail : [email protected] <mailto:[email protected]>
Web : http://www.devclic.fr