[SR-Users] kamailio 4.4.1 DROUTING module memory leak ?
Daniel-Constantin Mierla
miconda at gmail.com
Sun Jan 14 11:52:13 CET 2018
Hello,
the patch didn't include any new file, there were changes only inside
the functions.
But yes, the source code tree was restructured in 5.0, files being
relocated to reflect better their purpose and what component they are
part of.
Cheers,
Daniel
On 14.01.18 08:39, Uri Shacked wrote:
> Hi,
>
> Looks like the patch solved the issue.
> Just one thing - i am using 4.4.1. and the is no
> "../../core/mem/shm_mem.h" in modules directory.
> So, I changed it to "../../mem/shm_mem.h"
>
> Thanks
>
> On Fri, Jan 12, 2018 at 10:52 AM, Daniel-Constantin Mierla
> <miconda at gmail.com <mailto:miconda at gmail.com>> wrote:
>
> I pushed an addition patch do drouting module. Can you try with it
> as well?
>
> Cheers,
> Daniel
>
>
> On 12.01.18 09:24, Uri Shacked wrote:
>> Daniel,
>> It seems to keep on growing after calls traffic increased.:-(
>> On production after dr_timer.c was changed and kamailio installed
>> again,
>> Thanks,
>> Uri
>>
>> On Fri, Jan 12, 2018 at 7:33 AM, Uri Shacked <ushacked at gmail.com
>> <mailto:ushacked at gmail.com>> wrote:
>>
>> Ok,
>> Installed on production at midnight.
>> After 7 hours with very small traffic the ac_get_maxval keeps
>> on growing.
>> At startup count = 6 size = 120 bytes
>> After 7 hours with no more than 1000 calls, cout = 3756 size
>> = 124144.
>>
>> I have a feeling is still leaks.
>>
>> בתאריך 11 בינו׳ 2018 20:02, "Uri Shacked"
>> <ushacked at gmail.com <mailto:ushacked at gmail.com>> כתב:
>>
>> installed.
>> hard to get enough traffic on test.
>> will deploy today, and have results on sunday.
>>
>> On Thu, Jan 11, 2018 at 4:08 PM, Daniel-Constantin Mierla
>> <miconda at gmail.com <mailto:miconda at gmail.com>> wrote:
>>
>>
>>
>> On 11.01.18 13:04, Uri Shacked wrote:
>>> Thanks.
>>> installed OK on my test environment.
>>
>> Only installed or also tested a bit and all looks
>> fine so far?
>>
>> Cheers,
>> Daniel
>>
>>
>>> will install in 2 days on my production and keep you
>>> posted.
>>>
>>>
>>> On Thu, Jan 11, 2018 at 1:21 PM, Daniel-Constantin
>>> Mierla <miconda at gmail.com
>>> <mailto:miconda at gmail.com>> wrote:
>>>
>>> Hello,
>>>
>>> can you try with the patch from this commit:
>>>
>>> -
>>> https://github.com/kamailio/kamailio/commit/e0f95ea7fc691f976564b07436848673c633195d
>>> <https://github.com/kamailio/kamailio/commit/e0f95ea7fc691f976564b07436848673c633195d>
>>>
>>> The memset() is optional, more for long term
>>> safety, the if block at the end of the function
>>> is relevant.
>>>
>>> In 4.4, the file to change is
>>> modules/drouting/dr_time.c instead of
>>> src/modules/drouting/dr_time.c.
>>>
>>> If all is fine, then I will backport it.
>>>
>>> Cheers,
>>> Daniel
>>>
>>>
>>> On 11.01.18 11:57, Uri Shacked wrote:
>>>> Hi,
>>>> I have some kind of memory growth (leak?).
>>>> kamailio 4.4.1.
>>>> started with 64M for shmem, had a crash after 5
>>>> days of traffic.
>>>> increased shmem to 128, but still memory grows
>>>> everyday.
>>>> when traffic load decrease, the memory growth
>>>> stops but memory stays on the same level. when
>>>> traffic increase again,used memory continue to
>>>> grows again.
>>>>
>>>> I started kamailio now with the "-x qm" option
>>>> to debug shmem.
>>>> every half hour i dump the status of it.
>>>>
>>>> there are several modules that the memory size
>>>> is increasing (some of them are obvious).
>>>>
>>>> but, one is very strange.... DROUTING
>>>>
>>>> I am using drouting module for each call on my
>>>> kamailio.
>>>> the DB tables are very small, and there is no
>>>> reloads.
>>>>
>>>> only one time used in my script :
>>>> ..
>>>> ...
>>>> subst_user('/(.*)/$avp(xxx)/');
>>>> if(!do_routing("$avp(yyyy)")){
>>>> xlog...somthing;
>>>> return(-1)
>>>> }
>>>> ..
>>>> ....
>>>>
>>>> the shmem that is rapidly growing and does not
>>>> make sense is:
>>>> "from drouting: dr_time.c: ac_get_maxval(219)"
>>>>
>>>> seems that when i used mem_join =1, the growth
>>>> was smaller, but still significant.
>>>> now i use mem_join = 0, it seems rapidly
>>>> increasing...
>>>>
>>>> I have more information from the logs, will
>>>> send it if necessary (it is just a lot...)
>>>>
>>>>
>>>> any ideas ?
>>>> cheers,
>>>> Uri
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> 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
>>>> <https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users>
>>>
>>> --
>>> Daniel-Constantin Mierla
>>> www.twitter.com/miconda
>>> <http://www.twitter.com/miconda> -- www.linkedin.com/in/miconda
>>> <http://www.linkedin.com/in/miconda>
>>> Kamailio Advanced Training - March 5-7, 2018, Berlin - www.asipto.com <http://www.asipto.com>
>>> Kamailio World Conference - May 14-16, 2018 - www.kamailioworld.com <http://www.kamailioworld.com>
>>>
>>>
>>
>> --
>> Daniel-Constantin Mierla
>> www.twitter.com/miconda <http://www.twitter.com/miconda> -- www.linkedin.com/in/miconda
>> <http://www.linkedin.com/in/miconda>
>> Kamailio Advanced Training - March 5-7, 2018, Berlin - www.asipto.com <http://www.asipto.com>
>> Kamailio World Conference - May 14-16, 2018 - www.kamailioworld.com <http://www.kamailioworld.com>
>>
>>
>>
>
> --
> Daniel-Constantin Mierla
> www.twitter.com/miconda <http://www.twitter.com/miconda> -- www.linkedin.com/in/miconda <http://www.linkedin.com/in/miconda>
> Kamailio Advanced Training - March 5-7, 2018, Berlin - www.asipto.com <http://www.asipto.com>
> Kamailio World Conference - May 14-16, 2018 - www.kamailioworld.com <http://www.kamailioworld.com>
>
>
--
Daniel-Constantin Mierla
www.twitter.com/miconda -- www.linkedin.com/in/miconda
Kamailio Advanced Training - March 5-7, 2018, Berlin - www.asipto.com
Kamailio World Conference - May 14-16, 2018 - www.kamailioworld.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kamailio.org/pipermail/sr-users/attachments/20180114/df096bd9/attachment.html>
More information about the sr-users
mailing list