[OpenSIPS-Users] uac_registrant memory problem

[Ovidiu Sas]

As soon as a few re-registrations are going out, should be enough.

Regards,
Ovidiu Sas


On Tue, Oct 15, 2013 at 3:26 PM, Jeff Pyle <jpyle at fidelityvoice.com> wrote:

> Ovidiu,
>
> I might not have been as explicit about uac_registrant on the devel list.
>  Then it was just a theory.  Now I have some evidence.
>
> This last time my test with 200 registrants began to show "no more pkg
> mem" without showing a huge real_used_size value.  It self-destructed after
> a bit less than 3 hours.  I'll test next with the memory debugs enabled as
> you instruct.
>
> To ensure I get the correct data for you...  If I estimate this
> configuration will have a problem in approximately 3 hours, would it make
> sense to grab the two SIGUSR1 log dumps at, say, 30 and 90 minutes?
>
>
> - Jeff
>
>
>
> On Tue, Oct 15, 2013 at 11:18 AM, Ovidiu Sas <osas at voipembedded.com>wrote:
>
>> Please try the following test in your lab:
>>  - compile opensips with memory debug enabled;
>>  - let opensips to act as a registrant only (create a simple config that
>> will only load uac_registrant and the relevant required modules);
>>  - provide me with the output of 'opensipsctl ps';
>>  - send a SIGUSR1 to the process that is leaking memory and capture the
>> logs;
>>  - wait for a few more re-registrations;
>>  - send a second SIGUSR1 to the process that is leaking memory and
>> capture the logs;
>>  - send me the logs.
>>
>> I will take a look and let you know my findings.
>>
>> In your e-mail sent to dev I missed your reference to the uac_registrant
>> module.
>> Sorry about that.
>>
>> Regards,
>> Ovidiu Sas
>>
>>
>> On Tue, Oct 15, 2013 at 10:52 AM, Jeff Pyle <jpyle at fidelityvoice.com>wrote:
>>
>>> Hello,
>>>
>>> This is on version 1.9 revision 7606eab from August 13 this year.
>>>
>>> I've been able to verify the more entries in the uac_registrant table,
>>> the quicker something like this happens:
>>>
>>> # opensipsctl fifo get_statistics all | grep real_used_size
>>> shmem:real_used_size = 3056920
>>> pkmem:0-real_used_size = 181472
>>> pkmem:1-real_used_size = 225304
>>> pkmem:2-real_used_size = 181456
>>> pkmem:3-real_used_size = 18446744073706908112
>>> pkmem:4-real_used_size = 982512
>>> pkmem:5-real_used_size = 189624
>>> pkmem:6-real_used_size = 189624
>>> pkmem:7-real_used_size = 189624
>>> pkmem:8-real_used_size = 189624
>>> pkmem:9-real_used_size = 181048
>>> pkmem:10-real_used_size = 189616
>>>
>>> It always seems to happen on pkmem:3.  Once this occurs I see lots of
>>> "no more pkg mem" errors in the log.
>>>
>>> It took about eight hours for this occur on a test configuration doing
>>> nothing but registering 100 users from the uac_registrant module with 1M of
>>> pkg mem.  I've increased the number of users to 200 in an effort to get it
>>> fail faster.
>>>
>>> In order to provide information to isolate and fix this problem, what
>>> kind of data to I need to gather and what do I do with it?  I tried to
>>> follow the instructions on the memory troubleshooting page<http://www.opensips.org/Documentation/TroubleShooting-OutOfMem>but I didn't get a response from the devel list.
>>>
>>>
>>> Regards,
>>> Jeff
>>>
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users at lists.opensips.org
>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>>
>>>
>>
>>
>> --
>> VoIP Embedded, Inc.
>> http://www.voipembedded.com
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>


-- 
VoIP Embedded, Inc.
http://www.voipembedded.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20131015/72586ab8/attachment.htm>