[OCTO minor issue] - Unable to see participant video from another jvb

Hello guys,

I have successfully configured and setup Dockerized OCTO using this PR - #804

Here, I have setup 3 JVB’s on a single machine, and I can see that all the participants are getting placed on different jvb’s in a round robin way.

The issue is, lets say I have 6 participants, they get placed on different jvb’s, as expected

Participant 1 - jvb1 , Participant 2 - jvb2 , Participant 3 - jvb3,
Participant 4 - jvb1 , Participant 5 - jvb2 , Participant 6 - jvb3

What happens is, only the participants from jvb1 ( here, Participant 1 and 4 ) can see each other’s video and hear audio. Same goes for jvb2 and jvb3.

Is there anything that I have missed ?

Can participants from different JVBs see blank tiles and missing video and audio or they don’t see any indicator of participants from other bridges?

@rn1984, yes, participants from different JVBs can see blank tiles and missing video and audio

I can also see information about their connection and the jvb they are connected to. just no audio and video

Means your JVBs register with Jicofo but can’t communicate with one another. It’s most times a firewall issue. Try to get all FWs down.

I am using centos, and my firewall is disabled too.

[root@k8s-master ~]# systemctl status firewalld
● firewalld.service - firewalld - dynamic firewall daemon
   Loaded: loaded (/usr/lib/systemd/system/firewalld.service; disabled; vendor preset: enabled)
   Active: inactive (dead)
     Docs: man:firewalld(1)

Is there anything more that I have missed ? Should I paste my config files ?

Sure let’s see the configs of Jicofo and JVB.
Sipcommunicator and config

@rn1984, here you go. Let me know if you need anything more.

I have copied this directly from every containers /config directory.

jicofo-sip-communicator.properties.txt (434 Bytes)

jvb1.conf.txt (1.4 KB)

jvb1-sip-communicator.properties.txt (135 Bytes)

jvb2.txt (1.4 KB)

jvb2-sip-communicator.properties.txt (135 Bytes)

@metadata @corby @emrah any help here ?

Thanks for the information Oursainsburys

Sorry for my late response.
Your files look like a mess to me. I suggest you follow my guide again and keep the structure of them.

I.e. JVB1.conf seems to have JMS values.

yes, jvb1 is the default JMS, which is inside default docker-compose file. I have copied the docker-compose to a new folder, with the env file, changed the jvb nickname, rest everything matches with the default jvb

Not familiar with this structure, I’ve been using the files as is from guides and it works.

You are referring to this post right - Octo Cascade Bridges - here's how! FULL GUIDE ? @rn1984

If yes, regular installation of docker works for me, but dockerized is a challenge

Did you ever solve your problem with the Docker setup? I am facing nearly the same problems…

Nope. But there will be a new PR soon. Check the conversation…

I SOLVED IT a minute ago, at least for know :smiley: … I know about the PR. My modification is based on this PR. I got the base Octo config working, but the bridges never really talked correctly to each other. It was a simple configuration issue:

You have to use for BIND_ADDRESS and the external IP (NOT DNS name!) of the JVB’s machine. It is working now for me!

@tusharsonawane Are you using docker-swarm now or still using docker-compose?

Is the participant from jvb1 able to hear, see the participant from jvb2 ?

@metadata still docker-compose.

I messed up in some configurations while using swarm

I am using docker-compose, because I wrote a complete script framework for us internally, that a) builds Jitsi images, b) pulls my own docker-jitsi-meet repo, because I integrated the Octo PR on my own, c) sets up all necessary Jitsi variables, d) injects old Web UI, because we have custom further modifications for it and the new UI simply is not working for us, e) configures Docker swarm network, f) sets up a complete Grafana instance for the whole setup and fires it up.

Everything is based on a single config file each instance. So I set up the config for main instance, copy some files over to the next JVB machine, which have been created during startup of main instance, modifiy only some address values and the JVB counter in the config file and fire it up again. On the second machine, only JVB instance will be created and some exporters for Grafana.

Then I am up and running. I cannot use Swarm, because of the additional config steps which are necessary on the main Grafana instance to pull data from the additional JVBs - at least the way I am doing it.