[OpenSIPS-Users] cluster presence

Thu Aug 9 05:54:43 EDT 2018

Hi Volga,

The logs report a serious heavy execution of certain timer routines in 
OpenSIPS, like the presence cleanup takes more than 191 secs....most 
probably due long lasting DB queries ??? or because of heavy load in 
OpenSIPS that leads to starvation in handling the timer jobs - what is 
the internal load of OpenSIPS ? (use the 'load:' class of statistics to 
check it)

Regards,

Bogdan-Andrei Iancu

OpenSIPS Founder and Developer
   http://www.opensips-solutions.com
OpenSIPS Bootcamp 2018
   http://opensips.org/training/OpenSIPS_Bootcamp_2018/

On 07/01/2018 07:28 AM, volga629 at networklab.ca wrote:
> Hello Bogdan,
> I checked database connection and it looks normal ping less then sec 
> to database.
> I monitored load on database nodes and didn't notcied any extra load 
> on them.
> Right now I see 2 types of messages on all cluster nodes.
>
>
> Jun 30 22:56:45 aitossbc01 /usr/sbin/opensips[20245]: 
> WARNING:core:timer_ticker: timer task <presence-dbupdate> already 
> scheduled 100410 ms ago (now 700510 ms), skipping execution
> Jun 30 22:56:45 aitossbc01 /usr/sbin/opensips[20245]: 
> WARNING:core:timer_ticker: timer task <presence-pclean> already 
> scheduled 191700 ms ago (now 700510 ms), delaying execution
>
> and
>
> Jun 30 23:14:36 aitossbc02 /usr/sbin/opensips[9376]: 
> WARNING:core:utimer_ticker: utimer task <tm-utimer> already scheduled 
> 3880 ms ago (now 113380 ms), delaying execution
> Jun 30 23:14:36 aitossbc02 /usr/sbin/opensips[9376]: 
> WARNING:core:utimer_ticker: utimer task <tm-utimer> already scheduled 
> 5880 ms ago (now 113380 ms), delaying execution
>
> Also that not affecting call or call quality.
> Only filling up logs.
>
> volga629
>
> On Thu, Jun 14, 2018 at 6:48 AM, Bogdan-Andrei Iancu 
> <bogdan at opensips.org> wrote:
>> Hi Volga,
>>
>> How large is the presentity data set in your system ? I'm asking as 
>> the routine that seems to be slow queries the presentity table in 
>> order to get the expired presentities - this is done with or without 
>> clustering. Still, in clustering, all the nodes are doing this query, 
>> putting extra stress on the DB.
>>
>> Now, for each expire presentity, OpenSIPS has to send a NOTIFY to its 
>> subscribers - and here, having clustering enable, it is a difference. 
>> IF you have 3 nodes, so 3 shared tags, OpenSIPS will do 3 queries 
>> (one per tag) in order to fetch from active_watchers the subscribers 
>> with the tag looking for the presentity. So, who large is the 
>> subscriber's data set ?
>>
>> Do you notice any extra DB load when the cleanup timer kicks in ?
>>
>> Regards,
>>
>> Bogdan-Andrei Iancu
>>
>> OpenSIPS Founder and Developer
>>   http://www.opensips-solutions.com
>> OpenSIPS Summit 2018
>>   http://www.opensips.org/events/Summit-2018Amsterdam
>>
>> On 06/07/2018 03:43 PM, volga629 at networklab.ca wrote:
>>> Hello Bogdan-Andrei,
>>> Yes those messages start showing up when cluster enabled. Standalone 
>>> mode all works no issues.
>>> Right now in opensips cluster we have 2 active one backup. In PgSQL 
>>> 5 nodes 3 active 2 backup. In Mongodb cluster: 2 mongos 3 config 2 
>>> shred.
>>> We use postgres BDR cluster and MongoDB Cluster.
>>>
>>> Here are configuration
>>>
>>> #### Presence
>>> loadmodule "presence.so"
>>> loadmodule "presence_mwi.so"
>>> loadmodule "presence_xml.so"
>>> loadmodule "presence_dialoginfo.so"
>>> loadmodule "presence_callinfo.so"
>>> loadmodule "pua.so"
>>> loadmodule "pua_dialoginfo.so"
>>> loadmodule "xcap.so"
>>> modparam("presence|xcap|pua","db_url","postgres://URI/opensips_prod01")
>>> modparam("presence","server_address","sip:proxy at PUBLIC IP:5082")
>>> modparam("presence", "notify_offline_body", 1)
>>> modparam("presence", "fallback2db", 1)
>>> modparam("presence", "clean_period",  30)
>>> modparam("presence", "mix_dialog_presence", 1)
>>> modparam("presence", "cluster_id", 1)
>>> modparam("presence", "cluster_sharing_tags", "A=active")
>>> modparam("presence", "cluster_federation_mode", 1)
>>> modparam("presence", "cluster_pres_events" ,"presence , dialog;sla")
>>> modparam("presence_xml", "force_active", 1)
>>> modparam("presence_xml", "pidf_manipulation", 1)
>>> modparam("pua_dialoginfo", "presence_server", "sip:proxy at PUBLIC 
>>> IP:5082")
>>>
>>>
>>> volga629
>>>
>>>
>>> On Thu, Jun 7, 2018 at 7:32 AM, Bogdan-Andrei Iancu 
>>> <bogdan at opensips.org> wrote:
>>>> Hi Slava,
>>>>
>>>> What is the presence clustering configuration you have here ? also, 
>>>> what is the DB setup in regards to the cluster ?
>>>>
>>>> Also, did you start getting those errors only after enabling the 
>>>> clustering support? have you run an opensips individual presence 
>>>> node to see if you still get them ?
>>>>
>>>> Regards,
>>>>
>>>> Bogdan-Andrei Iancu
>>>>
>>>> OpenSIPS Founder and Developer
>>>>   http://www.opensips-solutions.com
>>>> OpenSIPS Summit 2018
>>>>   http://www.opensips.org/events/Summit-2018Amsterdam
>>>>
>>>> On 06/06/2018 05:20 PM, volga629 at networklab.ca wrote:
>>>>> Hello Everyone,
>>>>> I am trying put togher 3 nodes ( 2 active 1 backup) presence 
>>>>> cluster and log filled with messages regard cleanup timer.
>>>>> Any help thank you.
>>>>>
>>>>> opensips-2.4.1.b044f11ee-16.fc27.x86_64
>>>>>
>>>>>
>>>>> Jun  6 09:15:12 sbc01 /usr/sbin/opensips[4584]: 
>>>>> WARNING:core:timer_ticker: timer task <presence-pclean> 
>>>>> already scheduled for 5201330 ms (now 5653910 ms), it may 
>>>>> overlap..
>>>>> Jun  6 09:15:13 sbc01 /usr/sbin/opensips[4584]: 
>>>>> WARNING:core:timer_ticker: timer task <presence-pclean> 
>>>>> already scheduled for 5201330 ms (now 5654900 ms), it may 
>>>>> overlap..
>>>>>
>>>>>
>>>>> Slava.
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users at lists.opensips.org
>>>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>>>
>>>>
>>>
>>
>>
>