[OpenSIPS-Users] Dialog profiles replication issues

[Vlad Patrascu]

Hi Sammy,

I have pushed the fix for the crash, please check the latest commits.

Regards,

Vlad Patrascu
OpenSIPS Developer
http://www.opensips-solutions.com

On 1/7/20 10:40 PM, SamyGo wrote:
> Hi Vlad,
> Update: it wasn't the issue with dialog module or bin_proto. When I 
> used avp_db_query() inside async() that led to script processing to 
> halt, I've removed the async() and its working fine now, however the 
> crash for adding a new profile variable is still there and needs 
> resolution.
>
> Best Regards,
> Sammy
>
>
> On Tue, Jan 7, 2020 at 11:38 AM SamyGo <govoiper at gmail.com 
> <mailto:govoiper at gmail.com>> wrote:
>
>     Thanks Vald,
>     The problem I'm facing with profile sharing is strange, when I
>     make a call the whole flow freeze in that route with dialog
>     profile operations i.e get/set. I can wait for as many as 15+
>     seconds and nothing happens. CANCEL the call and resend INVITE and
>     it completes the route in normal way. Which is beyond my
>     understanding as why each alternate call is able to process the
>     route.
>
>     Here is the core dump trace: https://pastebin.com/XqiaMWnV
>
>     Best regards,
>     Sammy
>
>
>     On Tue, Jan 7, 2020 at 10:03 AM Vlad Patrascu <vladp at opensips.org
>     <mailto:vladp at opensips.org>> wrote:
>
>         Hi Sammy,
>
>         The behavior you described when replicating profiles is not
>         actually a loop but instead the normal operation. Dialog
>         profile counters are periodically broadcasted by all nodes in
>         the cluster to all other nodes. Also, an update is also
>         broadcasted when a counter decreases. As such there indeed
>         should not be any direct correlation between the number of SIP
>         invites and binary packets (containing profile counters) sent.
>         But if the SIP traffic contains BYEs or if you end dialogs you
>         might see a surge of binary packets.
>
>         As for the crash, can you post a full backtrace from gdb?
>
>         Regards,
>
>         Vlad Patrascu
>         OpenSIPS Developer
>         http://www.opensips-solutions.com
>
>         On 1/6/20 1:05 PM, Bogdan-Andrei Iancu wrote:
>>         reposting
>>         Bogdan-Andrei Iancu
>>
>>         OpenSIPS Founder and Developer
>>            https://www.opensips-solutions.com
>>         OpenSIPS Summit, Amsterdam, May 2020
>>            https://www.opensips.org/events/Summit-2020Amsterdam/
>>         OpenSIPS Bootcamp, Miami, March 2020
>>            https://opensips.org/training/OpenSIPS_Bootcamp_2020/
>>
>>         On 1/6/20 4:21 AM, SamyGo wrote:
>>>         Hi Bogdan,
>>>         Thanks for taking time out, here are the related config params.
>>>
>>>         listen=bin:OPENSIPS_IP:6677
>>>
>>>         ### dialog module for profiles/limits
>>>         loadmodule "dialog.so"
>>>         modparam("dialog", "db_mode", 0)
>>>         modparam("dialog", "profiles_with_value",
>>>         "userConcurrent/b;userInbound/b;userOutbound/b;userInternational/b;userTotalCalls/b;inboundDomain/b;outboundDomain/b;globalDomain/b;prefixx
>>>         /b;countryPrefix/b")
>>>         modparam("dialog", "profile_replication_cluster", 1)
>>>         modparam("dialog", "replicate_profiles_timer", 3000)
>>>         modparam("dialog", "replicate_profiles_check", 10)
>>>
>>>         ### Clustering Module
>>>         loadmodule "clusterer.so"
>>>         modparam("clusterer", "db_mode", 1)
>>>         modparam("clusterer", "my_node_id", 3)
>>>         modparam("clusterer", "sharing_tag", "gw1/1=active")
>>>         modparam("clusterer", "node_timeout", 20)
>>>         modparam("clusterer",
>>>         "db_url","DB_MODULE://DB_USER:DB_PASS@DB_HOST/DB_NAME")
>>>
>>>         Routing logic:
>>>         https://pastebin.com/6whjkhCs
>>>
>>>         Related Logs are here.
>>>         https://pastebin.com/a9a3iDQh
>>>
>>>          Wireshark graph showing storm of binary packets when SIP
>>>         invite initiates the related route.
>>>
>>>         *BONUS CRASH: *When in a cluster I add a new
>>>         profile_with_value , say someVariable/b and restart the
>>>         first box with this variable, the other boxes in cluster
>>>         crash by saying something along the lines of "unexpected
>>>         someVariable received"
>>>
>>>         I'm trying to get hold of the core files and as soon as
>>>         possible will share that here.
>>>
>>>
>>>         Regards,.
>>>         Sammy.
>>>
>>>
>>
>>
>>         _______________________________________________
>>         Users mailing list
>>         Users at lists.opensips.org  <mailto:Users at lists.opensips.org>
>>         http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>         _______________________________________________
>         Users mailing list
>         Users at lists.opensips.org <mailto: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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20200108/86aee397/attachment.html>