If I disable all external services, what would I need to go fiddling with to stop Jitsi from breaking?
Say as far as Jitsi is aware, every single client connecting to it is on the same subnet.
So I’m going to disable all external stun/turn services. Even DNS is on a private LAN with CA signed certs.
If there is documentation of this somewhere please point me in that direction. I’ve been messing with it for the better part of a day, and everything seems to keep breaking because I keep hitting places where Jitsi is expecting a WAN accessible architecture.
Disabling prosody config stun services has stopped the endless room reboot cycle loop, even though the connection does keep dropping on a steady intermittent every minute or so, it at least appears to be staying in the room somewhat… still something is very off. Multiple browsers on the same machine can’t join the same Jitsi room. It’s like every browser instance is a new unlinked session.
I’m still tweaking settings and going through the role, but I’m going off the base defaults setup in systemli’s ansible jitsi role for configuration reference.
Maybe upload js console logs with a particular problem. Guessing like that is hard.
Jvb announces its private and public address to all clients and if clients have direct access to the internal IP address and port 10000 for udp, everything should work. You can disable jvb from announcing its public address by commenting this line from its config: org.ice4j.ice.harvest.STUN_MAPPING_HARVESTER_ADDRESSES=meet-jit-si-turnrelay.jitsi.net:443
Except it never works. I assume this is something in my deployment, since I’m just wiping and re-standing up.
I can probably dig into this a little more, anything in particular I should be looking for?
the error seems pretty vague.
edit: Also, checking docs I only see a DNS need for meet.mydomain.com, do I need to add entries for all of the subdomains jitsi needs? It looks like there’s a few more.