So this is likely a result of my naivety but forwarding 5060 to my PBX seems to have resolved the issue. I guess I should have seen that with the many my-public-ip:5060 entries in the logs, but I thought 5060 was used just for registering. I had forwarded a different port to 5060 on my PBX and with everything working except the outbound call duration, I assumed it was ok.
I realize security through obscurity isn’t really security, but I was really hoping to not have to forward the default ports to get things to work.
Regardless, as of now it seems to have resolved my dropped call issue. Go figure.
Thank you to everyone in this thread for all your help. I cannot express in words just how much it’s appreciated.
That is often an artifact of a “sip helper” being active on your router (it work for simple solitary extensions but not for servers). Very simple answer to that , just don’t use 5060 (or anything between 5000 and 5999) for your SIP extensions, such a recipe is often promoted and trivial to implement , but for some unknown reason very rarely attempted, go figure . . .
Yes, and have the clients register on ip.addre.ss:nnnn if a separate port setting is unavailable and where nnnn is > 1024 and less than 50000(roughly), adjust your firewall as appropriate
Presumably I will need to update my local clients to match the new port as well, yes? I believe I read that Asterisk cannot listen on multiple ports for SIP traffic.
Nice! I would not have guessed that to be the issue based on the traffic we were seeing, but I guess it comes down to fully understanding the environment. Glad you got it figured out!