Hello Koray,

Is your problem is that TCP worker processes of Kamailio are blocked because of the communication with rtpengine? Or that the rtpengine is blocked because of the UDP packets loss?

If the first, one possible workaround would be to increase the number of worker processes. This would distribute the problem over more processes and causes less blocking in the processing. Why a decrease of the timeout would not help for you? This way Kamailio would faster detect and try again during eventual packet loss.

Cheers,

Henning

--

Henning Westerholt - https://skalatan.de/blog/

Kamailio services - https://skalatan.de/services

From: sr-users <sr-users-bounces@lists.kamailio.org> On Behalf Of Koray Vatansever
Sent: Wednesday, July 10, 2019 8:31 AM
To: Kamailio (SER) - Users Mailing List <sr-users@lists.kamailio.org>
Subject: [SR-Users] Long distance rtpengine communication

Hi all,

We have kamailio servers which communicate with the geographically distributed rtpengine servers.

Sometimes udp packets are lost or delayed between kamailio and rtpengine.

And this causes call drops...

Since rtpengine communication is synchronized and tcp worker processes are blocked, I couldn't set higher timeout values and more retry count.

Are there any suggestions to overcome this problem?

Thanks,

Koray