Ok, digged into 'switch_core_media.c' and apparently the 'remote_rtcp_port' is 0, so the function 'switch_rtp_activate_rtcp' is never called, which I assume means rtcp is never enabled.
From what I can see, it's set from channel variable "rtp_remote_audio_rtcp_port" and going backwards it looks like it comes from negotiating the sdp and the "a=rtcp:<port>" - so because the remote sip client never provides this, nothing is set.
I'm I wrong? (it's sooome time since I've done C code ;-)
Just to make sure I did a <param name="rtcp-video-interval-msec" value="passthru"/> in the sip_profile because I can see 'remote_port' branching happen later in the source and voilá:
2016-05-23 08:59:41.252648 [INFO] switch_core_media.c:6549 Activating RTCP PASSTHRU PORT 0
2016-05-23 08:59:41.252648 [DEBUG] switch_rtp.c:4189 RTCP passthru enabled. Remote Port: 29991
However I wish to monitor each leg separately, so this is not a viable option to me (and I'm no sure it works with remote port = 0 actually, haven't checked the code)
Long story short, couldn't/shouldn't FS just assume that 'remote_rtcp_port' is rtp_port +1 if 'a=rtcp' is missing in the in the SDP?
Hope someone can help ;-)
Regards,
Allan