For some visitors port 10000 fails


We have a setup with jitsi on a Ubuntu server and have configured so participants whose network is not allowing port 10000 is handled.
Works almost all the time, but for some participants (always the same networks) there video is not working.

When I block port 10000 on my own local network I get this in jvb-log and video still work:
(only a part of the log…)

No socket found for>
No socket found for>
Pair succeeded: our-public-ip:10000/udp/srflx → our-public-ip:46066/udp/prflx

jvb.log from when we have a participant that doesn’t get video:
No socket found for>
No socket found for>

For some reason that is above my brain level at the moment, for participants on networks blocking port 10000 that ends up with working video we get reference to the 192.168.* addresses.
But for participants where video fails 172.* and 10.* addresses is used/referenced.

If anyone can point me in the correct way it would be highly appreciated!

First, you need to test using 3 participants (for example by opening 3 browser tabs to simulate 3 participants). Otherwise P2P mode may prevent to analyse the issue correctly.

For participants who cannot access UDP/10000, you need a TURNS server which serves through TCP/443.