If you get the shm used size increasing, run following commands:

kamcmd cfg.set_now_int core memlog 1
kamcmd corex.shm_summary

Do it after some time of running, when you see that the increase has happened.

Then take the logs from syslog related to shm summary and send them over here.

On the other hand, I saw recently some article about latency spikes for tcp:

  - https://http2.cloudflare.com/the-story-of-one-latency-spike/

It is about http, but actually relates to tcp setting of os. Maybe you can verify your values in OS.

Cheers,
Daniel

On 20/11/15 12:40, Thibault Gueslin wrote:
I am adding the info in a mqueue.
In a single rtimer, I am consuming the mqueue and calling a lua script.

(I have followed the example of twitter).

I have increase the number of tcp_children and I got issue for memory allocation on allocations new tcp connection after a few hours.
Restart with SHMEM increased.
(but shame:used_size seems to increase)

Regards

2015-11-19 11:55 GMT+01:00 Daniel-Constantin Mierla <miconda@gmail.com>:
Are you doing the http requests for all sip requests? How many rtimer processes do you have consuming from mqueue?

What do you mean by "all tcp workers were busy"? They used lot of CPU?

Cheers,
Daniel


On 19/11/15 01:30, Thibault Gueslin wrote:
It seems to appear randomly.

I don't have database access (default config without db)
I have a mqueue with rtimer to do asynchronous http request.

I have seen also sometimes that all tcp worker were busy even on a test server with 4 clients.

Regards,

Thibault






Le 18 nov. 2015 à 13:28, Daniel-Constantin Mierla <miconda@gmail.com> a écrit :

Is it something repeating at constant interval of times, or are they random? Are you doing database queries in your config? Any of them via rtimer module?

Cheers,
Daniel


On 18/11/15 12:00, Thibault Gueslin wrote:
I am facing an issue with kamailio 4.3.3 that is very similar to

Had someone found the solution or faced similar issue ?

Sometimes kamailio is receiving the message (check with wireshark, logs in tcp) but the message is not handled nor relayed.
After a certain amount of time everything works fine.

I will do what as has been advised : enabling pike & check with gdb when it appears.

thibault



_______________________________________________
SIP Express Router (SER) and Kamailio (OpenSER) - sr-users mailing list
sr-users@lists.sip-router.org
http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-users


-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
Book: SIP Routing With Kamailio - http://www.asipto.com
Kamailio Advanced Training, Nov 30-Dec 2, Berlin - http://asipto.com/kat

-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
Book: SIP Routing With Kamailio - http://www.asipto.com
Kamailio Advanced Training, Nov 30-Dec 2, Berlin - http://asipto.com/kat


-- 
Daniel-Constantin Mierla
http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda
Book: SIP Routing With Kamailio - http://www.asipto.com
Kamailio Advanced Training, Nov 30-Dec 2, Berlin - http://asipto.com/kat