[SR-Users] async workers causing unexplained high load average

[Daniel-Constantin Mierla]

Hello,

back in the discussion after being traveling.

Interesting, because it shows like being the normal recvfrom() done on
the internal socket, expecting to be no load there. I have seen the
backtrace you sent in previous email and it is in the same recv function
from the system. The process is waiting for packets on the memory socket.

I will try to read more about and see if there is anything missing there.

Cheers,
Daniel

On 24/10/14 19:04, Alex Balashov wrote:
> I also find this in 'dmesg' about the async workers periodically:
>
> INFO: task kamailio:4480 blocked for more than 120 seconds.
>       Not tainted 2.6.32-431.29.2.el6.x86_64 #1
> "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> kamailio      D 0000000000000001     0  4480   4458 0x00000080
>  ffff880037b1db28 0000000000000082 ffff880037b1dbb8 ffffffff8112f183
>  ffff88000001dd80 0000000000000002 ffff880037b1dac8 ffffffff8109b39c
>  ffff8800a6069058 ffff880037b1dfd8 000000000000fbc8 ffff8800a6069058
> Call Trace:
>  [<ffffffff8112f183>] ? __alloc_pages_nodemask+0x113/0x8d0
>  [<ffffffff8109b39c>] ? remove_wait_queue+0x3c/0x50
>  [<ffffffff8152a5be>] __mutex_lock_slowpath+0x13e/0x180
>  [<ffffffff8152a45b>] mutex_lock+0x2b/0x50
>  [<ffffffff814f668a>] unix_dgram_recvmsg+0x7a/0x4d0
>  [<ffffffff8144a923>] sock_recvmsg+0x133/0x160
>  [<ffffffff8109afa0>] ? autoremove_wake_function+0x0/0x40
>  [<ffffffff8114b00a>] ? handle_mm_fault+0x22a/0x300
>  [<ffffffff8104a98c>] ? __do_page_fault+0x1ec/0x480
>  [<ffffffff8144aa9e>] sys_recvfrom+0xee/0x180
>  [<ffffffff810e1e07>] ? audit_syscall_entry+0x1d7/0x200
>  [<ffffffff8100b072>] system_call_fastpath+0x16/0x1b
>
>

-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda