Jitsi problem with more than 2 participants in a call

[Pasha_Laptev]

Hello guys!
Need your quick help, we are using jitsi and like life, but after 24.11.20 we faced a problem if we have more than 2 guys in a call - jitsi is not working at all!

Could you please help us!

We only have this console log with a strange error:

[saghul]

Looks like the WS connection towards your JVB could not be established. Can you check the network tab there in devtools to see the error? Also check the JVB logs.

[zhanggui201]

I deployed using dockers and encountered this problem. Did you solve it here?

[Freddie]

Check to make sure you have port 10000 open and properly forwarded. Also, make sure your firewall is not blocking any of the required ports. If those are fine, then check the Advanced configuration section of the Quick Install Guide.

[saghul]

Yes this should be working fine in the docker setup. Note that you need to set the PUBLIC_URL environment variable.

[Pasha_Laptev]

Hello, again guys!
Firewall is turned off, all the ports are open but I have this error log:

jitsi is still not working on our side, guys need your help as it is very permanent and we need to make meetings via your miracle service

Maybe lets have a look in real time on skype, or something else, any ideas ?

Thanks in advance!

[Arthur_Morales]

I had this problem with JVB running in Docker as well. It took me a few days debugging my setup. I am using Kubernetes though. I think a few posts here helped me a lot especially this one:

Latest unstable JVB2 is broken with three more participants

These are the correct props to set. What about port forwarding, is it working? You can test it using nc.

What seems to be happening for you is the same thing that happened to me. When you have 2 participants Jitsi uses P2P connection between the participants. When a third person joins it’ll involve your JVB server to serve the video call.

I fixed my problem by injecting manually into sip-communicator.properties the variables NAT_HARVESTER_LOCAL_ADDRESS and NAT_HARVESTER_PUBLIC_ADDRESS as such:

export LOCAL_CLUSTERIP_ADDRESS=$(dig +short @10.96.0.10 ${HOSTNAME}.jitsi.svc.cluster.local | tail -n1)

    echo  "org.ice4j.ice.harvest.NAT_HARVESTER_LOCAL_ADDRESS=${LOCAL_CLUSTERIP_ADDRESS}" >> /defaults/sip-communicator.properties
    echo  "org.ice4j.ice.harvest.NAT_HARVESTER_PUBLIC_ADDRESS=${DOCKER_HOST_ADDRESS}" >> /defaults/sip-communicator.properties

What I am doing here is reading the cluster ip service from kubernetes and pushing into NAT_HARVESTER_LOCAL_ADDRESS

[Pasha_Laptev]

First of all we use docker-compose to run all jitsi containers.
I’ve found that we have a file docker-jitsi-meet/jvb/rootfs/defaults/sip-communicator.properties

So this is data in this file
{{ if .Env.DOCKER_HOST_ADDRESS }}
org.ice4j.ice.harvest.NAT_HARVESTER_LOCAL_ADDRESS={{ .Env.LOCAL_ADDRESS }}
org.ice4j.ice.harvest.NAT_HARVESTER_PUBLIC_ADDRESS={{ .Env.DOCKER_HOST_ADDRESS }}
{{ end }}

So I’ve added
DOCKER_HOST_ADDRESS={{our external IP that I use to connect to the server}}
LOCAL_ADDRESS=172.17.0.1 (i’ve found this ip in ifconfig in docker section)

But we still have the same issue.

[saghul]

Can you check the network tab in Chrome DevTools to see the error when the WS connection is attempted?

[Pasha_Laptev]

The network is good, you can view these screenshots:
Any ideas guys?

[Pasha_Laptev]

[Pasha_Laptev]

[Pasha_Laptev]

[saghul]

You seem to be using datachannels, so yeah, no WS there. Is port 10000 open on 178.128.207.120 ? That is the ICE candidate for the server.