I have a jitsi set (web,jicofo,prosody,jvb) in GKE and one additional jvb(only jvb without jicofo,web…) on local server, the jvb was set up in Docker. All work. The second jvb connects to jitsi set.
Then I adjusted OCTO, I followed this tutorial https://github.com/jitsi/jitsi-videobridge/blob/master/doc/octo.md. One moment which I didn’t understand it is about config.deploymentInfo.userRegion: is it a required customization or no? I didn’t use it.
I’ve tested it via SplitBridgeSelectionStrategy. Each new participant had different regions.
But when I use RegionBasedBridgeSelectionStrategy I see next:
- Each new room randomly chooses jvb;
- Each new room connects not always to clossest jvb;
- All participants connect to jvb which was initializer.
I seem now it works as load balancer but not detects region correct
- Can anybody help me. Maybe I have incorrect a structure for connecting beetween bridges and jitsi?
- Should I use any tool (e.g HAProxy) that routs traffic on bridge in depends region?
- Should I use a deploymentInfo section in config.js?
- If yes it necessary only for the first jitsi set or I should set up a full jitsi set with the second jvb?
- Also I’ve tested OCTO on meet.jit.si(https://webrtchacks.com/sfu-cascading/) from 2 loclations(Europe and USA) and I’ve noticed that all participants connect to one jvb who was initializer.
Below I’ve added screens my configuration and jicofo log with 3 participants (2 in Europe one in USA, start was in Europe)
jicofo.log (49.1 KB)