[Gluster-users] process hung tasks in 3.4.1

[Vijay Bellur]

On 12/06/2013 08:06 PM, Peter Mistich wrote:
> Hello,
>
> I was wondering if anyone has seen this issue in glusterfs 3.4.1
>
> I've seen a few post about it in 3.2.0.
> http://gluster.org/pipermail/gluster-users/2011-June/007984.html
>
> not sure if the same problem
>
>
> Here is info on the system
>
>
> client:
> Red Hat Enterprise Linux Server release 6.4 (Santiago)
> 2.6.32-358.23.2.el6.x86_64
>
> glusterfs-3.4.1-3.el6.x86_64
> glusterfs-cli-3.4.1-3.el6.x86_64
> glusterfs-libs-3.4.1-3.el6.x86_64
> glusterfs-fuse-3.4.1-3.el6.x86_64
>
>
> servers:
> Red Hat Enterprise Linux Server release 6.4 (Santiago)
> 2.6.32-431.el6.x86_64
>
>
> dmesg:
> INFO: task python:31598 blocked for more than 120 seconds.
> "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> python        D 0000000000000002     0 31598      1 0x00000080
>   ffff880811cb7e38 0000000000000082 0000000000000000 ffff8803e0d05398
>   ffff8803e0d054f0 0000000300000001 00000000000000d0 ffff880724f67c40
>   ffff88081334c638 ffff880811cb7fd8 000000000000fb88 ffff88081334c638
> Call Trace:
>   [<ffffffffa0177075>] fuse_request_send+0xe5/0x290 [fuse]
>   [<ffffffff81096da0>] ? autoremove_wake_function+0x0/0x40
>   [<ffffffffa017cb36>] fuse_flush+0x106/0x140 [fuse]
>   [<ffffffff8117e0bc>] filp_close+0x3c/0x90
>   [<ffffffff8117e1b5>] sys_close+0xa5/0x100
>   [<ffffffff8100b072>] system_call_fastpath+0x16/0x1b
>
>

This refers to a flush/close call taking more than 120 seconds. Though 
uncommon for a close system call to consume more time, it can happen if 
the network latencies and/or load on the servers is high. Does this task 
hang forever?

-Vijay