[SR-Users] Evapi dispatcher is consuming all the private memory

Daniel-Constantin Mierla miconda at gmail.com
Thu Aug 15 08:36:30 CEST 2019


Hello,

can you try with master branch? The issue should be fixed by the commit:

  *
https://github.com/kamailio/kamailio/commit/d36a4b00c6becd42cfcec4fc7eb398371e1cc812

If works ok, then I will backport to stable branches.

Cheers,
Daniel

On 14.08.19 12:19, José Seabra wrote:
> Hello,
> Please find out the logs in the following link:
>
>   * https://pastebin.com/h2UNHwYC
>
> Let me know if you need something else.
>
> Cheers
> José Seabra
>
> Daniel-Constantin Mierla <miconda at gmail.com
> <mailto:miconda at gmail.com>> escreveu no dia quarta, 14/08/2019 à(s) 10:59:
>
>     Hello,
>
>     that's memory status, printing details for each allocated chunk. I
>     could spot a lot of chunks from the parser and get header field
>     functions, but to be sure there is nothing else, can you set
>     global parameter:
>
>     mem_summary=12
>
>     and then reproduce again. The logs printed in syslog should have
>     smaller size, giving the summary of allocated chunks - how many
>     times and the size for each location in the code.
>
>     Cheers,
>     Daniel
>
>     On 14.08.19 11:51, José Seabra wrote:
>>     Hello Daniel,
>>     Please find out the attached zip file with the output pkg_summary.
>>
>>     Let me know if you need something else.
>>
>>     Thank you for your support
>>
>>     Cheers
>>     José Seabra
>>
>>     Daniel-Constantin Mierla <miconda at gmail.com
>>     <mailto:miconda at gmail.com>> escreveu no dia quarta, 14/08/2019
>>     à(s) 10:03:
>>
>>         Hello,
>>
>>         can you enable pkg summary and generate it once the out of
>>         memory happens?
>>
>>         Set memlog to lower value than debug, when the memory
>>         happens, do:
>>
>>         https://www.kamailio.org/docs/modules/devel/modules/corex.html#corex.rpc.pkg_summary
>>
>>         Then send some traffic to that process and watch the syslog.
>>
>>         Alternative, enable memory summary via core parameter and
>>         stop kamailio when the errors happen, but you will get
>>         summary for all processes, so it might be harder to extract
>>         those log messages related the evapi process.
>>
>>         Cheers,
>>         Daniel
>>
>>         On 14.08.19 10:37, José Seabra wrote:
>>>         Hello there,
>>>         I'm using kamailio 5.2.1 with evapi module to send and
>>>         receive events from an external application.
>>>          
>>>         During my stress tests to kamailio i have been noticing that
>>>         at some point kamailio is getting out of memory:
>>>
>>>         edge-sip-proxy[20565]ERROR: ESP_LOG: <null>: <core>
>>>         [core/mem/q_malloc.c:291]: qm_find_free():
>>>         qm_find_free(0x7fb8730b6010, 64); Free fragment not found!
>>>
>>>         edge-sip-proxy[20565]: ERROR: ESP_LOG: <null>: <core>
>>>         [core/mem/q_malloc.c:425]: qm_malloc():
>>>         qm_malloc(0x7fb8730b6010, 64) called from core:
>>>         core/parser/msg_parser.c: parse_headers(325), module: core;
>>>         Free fragment not found!
>>>
>>>         edge-sip-proxy[20565]: ERROR: ESP_LOG: <null>: <core>
>>>         [core/parser/msg_parser.c:328]: parse_headers(): memory
>>>         allocation error
>>>
>>>         edge-sip-proxy[20565]: ERROR: ESP_LOG: <null>: pv
>>>         [pv_core.c:704]: pv_get_callid(): cannot parse Call-Id header
>>>
>>>         edge-sip-proxy[20565]: INFO: ESP_LOG: <null>:
>>>         [evapi:connection-received] EVAPI msg received ->
>>>         {"aor":"sip:FUZECI1-x1516 at uc.fuze.domain","contact":"<sip:FUZECI1-x1516 at 10.225.121.206:5060;transport=UDP>"}
>>>
>>>         edge-sip-proxy[20565]: ERROR: ESP_LOG: <null>: <core>
>>>         [core/mem/q_malloc.c:291]: qm_find_free():
>>>         qm_find_free(0x7fb8730b6010, 64); Free fragment not found!
>>>
>>>         edge-sip-proxy[20565]: ERROR: ESP_LOG: <null>: <core>
>>>         [core/mem/q_malloc.c:425]: qm_malloc():
>>>         qm_malloc(0x7fb8730b6010, 64) called from core:
>>>         core/parser/msg_parser.c: parse_headers(325), module: core;
>>>         Free fragment not found!
>>>
>>>
>>>         The process id 20565 is running out of memory as you can see
>>>         bellow:
>>>
>>>
>>>         output of kamcmd pkg.stats 
>>>
>>>
>>>         entry: 14
>>>
>>>         pid: 
>>>
>>>         rank: -2
>>>
>>>         used: 6058696
>>>
>>>         free: 824
>>>
>>>         real_used: 8387784
>>>
>>>         total_size: 8388608
>>>
>>>         total_frags: 30
>>>
>>>
>>>         This process ID is related with Evapi Dispatcher
>>>
>>>
>>>
>>>         kamcmd ps | grep 20565
>>>
>>>
>>>         20565EvAPI Dispatcher
>>>
>>>
>>>         After stop the sipp that was generating traffic to kamailio
>>>         the evapi process still have all the memory consumed
>>>
>>>         Can it be some memory leak in Evapi module?
>>>
>>>         Best Regards
>>>         José Seabra
>>>         -- 
>>>         Cumprimentos
>>>         José Seabra
>>>
>>>         _______________________________________________
>>>         Kamailio (SER) - Users Mailing List
>>>         sr-users at lists.kamailio.org <mailto:sr-users at lists.kamailio.org>
>>>         https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
>>
>>         -- 
>>         Daniel-Constantin Mierla -- www.asipto.com <http://www.asipto.com>
>>         www.twitter.com/miconda <http://www.twitter.com/miconda> -- www.linkedin.com/in/miconda <http://www.linkedin.com/in/miconda>
>>
>>
>>
>>     -- 
>>     Cumprimentos
>>     José Seabra
>
>     -- 
>     Daniel-Constantin Mierla -- www.asipto.com <http://www.asipto.com>
>     www.twitter.com/miconda <http://www.twitter.com/miconda> -- www.linkedin.com/in/miconda <http://www.linkedin.com/in/miconda>
>
>
>
> -- 
> Cumprimentos
> José Seabra

-- 
Daniel-Constantin Mierla -- www.asipto.com
www.twitter.com/miconda -- www.linkedin.com/in/miconda

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kamailio.org/pipermail/sr-users/attachments/20190815/844499d9/attachment.html>


More information about the sr-users mailing list