[OpenSIPS-Users] Memory leak
Duane Larson
duane.larson at gmail.com
Fri Jan 21 21:50:05 CET 2011
Thanks Anca and Bogdan. Both of you are right in a way. SIPP is, after a
while, leaving a lot of dialogs in "state 3". It could be because I am
using the same server as the SIPP UAS and UAC. I will need to split the
SIPP processes up to see if that changes things.
I am not sure if all the Dialogs in "state 3" are because of the stress test
load I am putting on OpenSIPS or because of a combination of things
(internet and SIPP test machine). Either way I want to be sure to implement
a way to delete these stale dialogs. I saw this post
http://opensips-open-sip-server.1449251.n2.nabble.com/destroy-dialog-on-transaction-timeout-and-load-balancer-td5832178.html
Sounds like I should try to implement option 2 for dialogs that are (state
!= 4 && lifetime > 5 minute).
On a side note, last night SIPP left 501 dialogs in "state 3" on OpenSIPS.
I waited 12 hours which is the default dialog timeout for those dialogs to
be removed from the dialog table. Once they were removed I didn't see free
memory of the machine go up any. And when I did a "top" command I saw that
the 4 opensips child processes still had 6.4% of memory usage. Should the
processes have freed up that memory? Or is that considered allocated memory
that Opensips can use?
On Fri, Jan 21, 2011 at 6:19 AM, Bogdan-Andrei Iancu <bogdan at opensips.org>wrote:
> Hi Duane,
>
> actually, checking the dump you sent, I see no trace of a leak - the pkg
> dump shows only script and DB conn mem, while the shm is empty.
>
> So, in your case, you may have a memory overload because because of runtime
> issues (but not a leak). If dialogs are not removed by BYEs, I guess you end
> up with TONS of dialogs in memory...
> dialog stuff is not releated to USRLOC at all....maybe the BYEs are not
> going trough loose_route() in script and not pushed for dialog matching
>
>
> Regards,
> Bogdan
>
> Duane Larson wrote:
>
>> After looking a little more do you think this would cause the memory
>> issue...
>>
>> Looks like my SIPP test starts the call, does whatever I want and sends
>> the BYE, but the dialog is still in the database after the call has ended
>> between the UAS and UAC. I am currently using USRLOC db_mode = 2, so I am
>> wondering if the dialogs are all just staying in memory. Not sure why my
>> SIPP xmls or OpenSIPS config is not allowing the dialog to end in OpenSIPS
>>
>> On Thu, Jan 20, 2011 at 2:07 PM, Duane Larson <duane.larson at gmail.com<mailto:
>> duane.larson at gmail.com>> wrote:
>>
>> I do have some memory stuff in syslog. I have posted the output here
>> http://paste.ubuntu.com/556285/
>>
>>
>>
>>
>> On Thu, Jan 20, 2011 at 4:13 AM, Bogdan-Andrei Iancu
>> <bogdan at opensips.org <mailto:bogdan at opensips.org>> wrote:
>>
>> Hi Duane,
>>
>> with
>>
>> mem_log=10
>> mem_dump=1
>> debug=3
>>
>> do you see at runtime any logs related to memory ?? (alloc,
>> dealloc logs, no error) . Normally you shouldn't see
>> any....only when a me dump is done (at shutdown for example).
>>
>> Regards,
>> Bogdan
>>
>> Duane Larson wrote:
>>
>> I am trying to stress test my opensips config with SIPP
>> but was seeing that after running it for a while I would
>> start getting failed calls and eventually opensips would
>> die. I recompiled opensips by following
>>
>> http://www.opensips.org/Resources/DocsTsMem
>>
>> I also added the following to the config
>> mem_log=10
>> mem_dump=1
>> debug=3
>>
>>
>> Now when I try and start OpenSIPS it dumps a TON of stuff
>> to syslog. I do a ps -ef and there is one opensips
>> process running for a while and then it dies. Before I
>> started opensips my memory was at about
>> 1500000k
>> then when I started opensips it started going down to
>> 1400000k
>>
>> When the single opensips process finally dies the memory
>> never comes back. I am not sure what I need to do next in
>> order to find out where the issue is.
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org <mailto:Users at lists.opensips.org>
>>
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>>
>> -- Bogdan-Andrei Iancu
>> OpenSIPS Event - expo, conf, social, bootcamp
>> 2 - 4 February 2011, ITExpo, Miami, USA
>> OpenSIPS solutions and "know-how"
>>
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org <mailto:Users at lists.opensips.org>
>>
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>>
>>
>> -- --
>> *--*--*--*--*--*
>> Duane
>> *--*--*--*--*--*
>> --
>>
>>
>>
>>
>> --
>> --
>> *--*--*--*--*--*
>> Duane
>> *--*--*--*--*--*
>> --
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>
>
> --
> Bogdan-Andrei Iancu
> OpenSIPS Event - expo, conf, social, bootcamp
> 2 - 4 February 2011, ITExpo, Miami, USA
> OpenSIPS solutions and "know-how"
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
--
--
*--*--*--*--*--*
Duane
*--*--*--*--*--*
--
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20110121/5de9c3bc/attachment-0001.htm>
More information about the Users
mailing list