CMS: Call are dropped after 15 minutes

[Sebastien Denoncin]

Hi everybody,

I currently have a cluster of CMS consisting of 3 Vms.
I have the callbridge group is composed of CMS 1 and 2. CMS 3 being there for redundancy DB and xmpp.

After several tests, I noticed that some video calls are cut after 15 minutes.

See attachment. file call_dropped.png

For my next test I cut the CMS2. So all the calls come in on the CMS1.

In this case no drop.

If I turn off the CMS1 and put the CMS2 back on, it's the same -->  no drop.

In the scenario of droppedcalls, I notice that dropped calls are referenced as outgoing on the CMS interface

See attachment. file call_dropped01.png

Have you any idea? 

It appears that the problem occurs when moving from cms1 to cms2 or cms2 to cms1

thanks

[Gregory Brunn]

My only real experience with CMS dropping calls had to do with web-bridge. In early documentation it stated to basically have multiple webbridge and have round robin DNS records for them.  Calls would drop after some time this way. Only way to have redundant webbridge is truly with a load balance. 

How are you connecting to the CMS via a sip call into through CUCM or Expressway or via webrtc web client.

[Sebastien Denoncin]

Hi Gregory,

I'm connecting through Expressway. I didn't test this behavior through WebRTC.

Today I have a important video conf and to be sure that the call was not dropped , I shutdown the CMS2.

The meeting has not been dropped.

I think that it's when re-invite is sent, the CMS cannot find the remote end-point.

Thanks

Regards