[SR-Users] rtpengine failover in same SIP dialog with response Unknown call-ide

[Michal Popovic]

Hi,

we are facing the issue with automated failover in same SIP dialog in rtpengine module. This issue occurs after we upgraded version 4.4.7 to 5.3.2 and only for rtpengines that are in another datacenter with latency higher than about 100ms.

configuration
modparam("rtpengine", "rtpengine_sock", "1 == udp:10.0.0.1:2222 udp:10.0.0.2:2222")
modparam("rtpengine", "rtpengine_sock", "1 == udp:10.0.0.3:2222 udp:10.0.0.4:2222")

INVITE request from kamailio to 10.0.0.1:2222

3:ICE6:remove5:flagsl13:trust-address16:always-transcode17:asymmetric-codecs12:loop-protecte7:replacel6:origin18:session-connectione18:transport-protocol7:RTP/AVP8:rtcp-muxl5:demuxe7:call-id20:xxx:received-froml3:IP414:x.x.x.xe8:from-tag10:naluniuntp7:command5:offere

RESPONSE

6:result2:oke

ANSWER request is than send to 10.0.0.2:2222 with error

3:ICE5:force5:flagsl13:trust-address16:always-transcode17:asymmetric-codecs12:loop-protecte7:replacel6:origin18:session-connectione18:transport-protocol9:RTP/SAVPF8:rtcp-muxl6:accept5:offere7:call-id20:l63nam58lomphlpvli9o13:received-froml3:IP414:18.212.137.170e8:from-tag10:naluniuntp6:to-tag10:as297c65177:command6:answere

RESPONSE

d6:result5:error12:error-reason15:Unknown call-ide

Only possible fix is to have only one setid with only one server.

First RTPengine is not marked as failed in list of rtpengine servers after the first response.