[SR-Users] parallel forking in combination with rtpengine

Hi everbody, I encountered an issue with my Kamailio in combination with Rtpengine. I wondered if anybody got a solution or a hint as this issue is only in combination with special parallel forked calls. I sent a call from my SIP gateway via a Kamailio to another SIP provider. The other SIP provider or the B-user seems to use a parallel forking as I receive 183 Session Progress and 180 Ringing with different to-tags. My kamailio is using a Rtpengine to relay the media from my SIP gateway to the other SIP provider. As Kamailio receives two different to-tags that include a SDP body rtpengine_manage makes Rtpengine to change RTP port. Some short example (cut down to the important messages): a-gateway kamailio b-provider | INVITE Port 16890 | INVITE Port 20400 | | ------------------------> | --------------------------> | | | | | 180 to-tag 1 / no SDP | 180 to-tag 1 / no SDP | | <------------------------ | <------------------------- | | | | | 183 to-tag 2 / Port 20408 | 183 to-tag 2 / Port 46298 | | <------------------------ | <------------------------- | | | | | | | | RTP | RTP | | Port 16890 <---- Port 20408 | Port 20400 <---- Port 46298 | | | | | | | | RTP | RTP | | Port 16890 ----> Port 20408 | Port 20400 ----> Port 46298 | | | | | | | | 180 to-tag 2 / Port 20408 | 180 to-tag 0 / Port 46298 | | <------------------------ | <------------------------- | | | | | | | | 200 OK to-tag 1 / Port 20408 | 200 OK to-tag 1 / Port 46298 | | <------------------------ | <------------------------- | | | | | | | | RTP | RTP | | Port 16890 <---- Port 20408 | Port 20400 <---- Port 46298 | | | | | | | | RTP | RTP | | Port 16890 ----> Port 20408 | Port 20418 ----> Port 46298 | Kamailio receives INVITE from a-gateway with rtp port 16890 and rtpengine rewrites it with port 20400 to b-provider. b-provider sends 180 ringing from to-tag 1 and no SDP which is just passed through kamailio to a-gateway. b-provider sends 183 progress from to-tag 2 and SDP with port 46298 to kamailio. rtpengines rewrites the sdp port to 20408 which is passed to a-gateway. b-provider start sending RTP via the negotiated ports which is passed correctly to a-gateway. b-provider sends 180 ringing from to-tag 2 with the same SDP. Rewritten and passed to a-gateway. b-provider sends 200 OK on Invite. This 200 Ok is from to-tag 1 but including the same SDP endpoint as to-tag 2 used before. Media from b-provider via rtpengine to a-gateway is passed the same way as before. Media from a-gateway to rtpengine is passed the same way as before. Media from rtpengine is passed with a different source port to b-provider. As this port was not announced in the callflow b-provider blocks the traffic. Any ideas how to fix this issue or force rtpengine to stay on the same port? Thanks Michael