I am running Asterisk 11.2.1 and am using it to make SIP calls between clients on both LAN and WAN. I have set up the firewall with the appropriate port forwarding and NAT policies, and am able to make calls almost all of the time.
However, I've had instances of one-way audio while load-testing my system with calls. From reading the documentation about rtp.conf, I know that for each RTP port, an RTCP port = RTP port + 1 is used. I have defined a wide enough range (rtpstart and rtpend) for the number of clients I have, and I know my firewall has the exact same range forwarded.
Is it possible that, under a loaded scenario where unused RTP ports are scarce, that Asterisk could choose rtpend as the RTP port, and rtpend + 1 (outside firewall range) as the RTCP port? Do I have to allow for this possibility on my firewall? I want to know for certain whether this is the case or not so I can move further in my root cause analysis.
Thank you!