Jicofo-Jibri signaling - What changed?

A previously working single-server installation of Jitsi and Jibri stopped recording on recent updates (unstable). Not sure exactly when it first started (I wasn’t checking Jibri), but in the last few unstable versions, noticed recording won’t start.

Most times, there are no errors in Jibri. Occassionally, I see the following error:

XmppApi.handleStartJibriIq#226: Error starting Jibri service

However, Jibri service is active when I check the status.

In Jicofo, I see more clues:

Got Jibri status update: Jibri jibribrewery@internal.auth.dev.domain.com/jibri-nick has status off and failure reason error

And…

Failed to send start Jibri IQ: org.jitsi.jicofo.jibri.JibriSession$StartException$UnexpectedResponse: Unexpected response

Jibri authenticates and enters the brewery just fine, but for some reason, doesn’t start recording and after a while, Jicofo reports not finding it. I think there’s something going on with the control muc.

I’m absolutely certain Jibri will work if I reinstall it, but I’m curious as to what changed - if anything - in the signaling between Jicofo and Jibri. I’d like to know, even if just for academic reasons.

Anything in jibri logs? Can it be the chrome updates?

Nothing of significance really in Jibri logs. Sometimes, it shows “Error starting Jibri service”, other times, nothing at all. I have Chrome 97 running on the server now, if I recall correctly. Maybe I’ll circle through Chrome versions to see. But I feel pretty confident that if I reinstall, it will work, even with the Chrome 97 version I’m running.

What about after restarting jibri, do you experience the same?
As I suppose jicofo was updated, so was restarted …

Yeah, I restarted several times (Jibri, Jicofo, Prosody, JVB), cleared all logs to monitor behavior, no luck. Something seems to be stopping the control once it gets into the brewery. That’s why I’m wondering if there was a change in the latest versions of Jicofo and how it communicates with Jibri. :thinking:

Still battling to figure this out. Jicofo is logging these complaints:

Jicofo 2022-01-11 10:35:58.742 SEVERE: [18] [xmpp_connection=client] XmppProviderImpl.doConnect#225: Failed to connect/login: The following addresses failed: 'RFC 6120 A/AAAA Endpoint + [localhost:5222] (localhost/127.0.0.1:5222)' failed because: java.net.ConnectException: Connection refused
org.jivesoftware.smack.SmackException$EndpointConnectionException: The following addresses failed: 'RFC 6120 A/AAAA Endpoint + [localhost:5222] (localhost/127.0.0.1:5222)' failed because: java.net.ConnectException: Connection refused

Looks like Prosody is having problems establishing connection. JMS + Jibri are on the same server. But just in case, opened 5222/TCP on server and issue persists. Disabling prejoin screen doesn’t help (not that that would be the solution). I’m wondering if somehow prosody is expecting an explicit calling of the local IP instead of localhost. Seems like Smack is unable to connect to the xmpp service. Again, this just started fairly recently (last few versions of unstable).

hi Freddie. Wish you a nice day.
Is this issue still exist. And which version has it like jitsi 6433 contains the problem?

I run unstable versions in my dev environment (which is where I’m seeing this issue). I’m not sure anyone is experiencing this in the latest stable (at least, I haven’t seen anyone call it out).

Ok, thank you.

hi, Freddie. If you solve this issue please share the info if you have time :slightly_smiling_face:
In case, we face the issue in live that will be very helpful