AWS scale for big deployment

We are creating AWS auto-scaling for jitsi. However no documentation were found. We would like to have presence for conferencing in multiple regions as the ideal solution. But willing to start small and grow as we go. Any recommendation, suggestions, cloud formation templates, etc? We have just one “meet” server stood up but now would like to use conferencing in the autoscaling group but can’t find any details how to configure so and how to point “meet” to pick and choose a conferencing server.
This article gives somewhat basic info but no details whatsoever.

1 Like

Yes, this is the video tutorial which we have.
You may search the forum about haproxy to gather more information.
We can answer any particular questions you may have, you can also join the community call and ask some questions there. But there are no cloud formation or templates from us (the jitsi team), there were few community members opensourcing such stuff, probably also information you can find in the forum.

thanks for the quick response.
alright, let me start with dummy question. I have one jitsi server configured. how would i create a shard? no HA yet though. Just would like to have one jitsi and a few (at least two) jitsi-videobridge servers configured. If i understand correctly they need to “know” somehow each other. How would i achieve that? Should it be hardcodded or not and where that should be configured?

So, soon we will update stable, where the preferred method of installing jvb by default will be using mucs (brewery rooms) and not through component.
Install on a clean machine the unstable packages to see that.
If you take the jvb configs (change the host to connect to be the original server where you did the install, we call this signalling node) and multiply n times a new jvb machine with those settings you will have a shard with n bridges running.

What components should be installed on jvb? also, can i connect using IP or it should be DNS? or doesn’t matter?

jitsi-videobridge (soon it will be jitsi-videobridge2).

The only DNS (DNS entry and certificate for it) you need for the system is the domain you use to connect to the webserver.
All the rest are XMPP virtual hosts in prosody and you don’t need a DNS or certificate for those.
You can configure jvb to connect using the ip address.

thanks for the clarification.
i’m going to build another server for jvd only and will try to connect it to jitsi-meet

i have found this link:

and trying to understand how to make meet to communicate with jvb… are there any instructions? what and where changes should be made? links to configuration files?

I’m lost, can you give more details, what do you mean?

i have two servers ready.
one server - jitsi meet
second server - jvb
How would i connect them that they are talking to each other? What configurations files should be changed?

Make sure all jvbs you run use this method to connect to prosody and jicofo knows about the brewery room:

very confusing…
also, do you have any suggestions on how to configure on on AWS cloud?
The thing is that EC2 instance doesn’t have public IP nor interface on OS level configured but rather EIP attached.

i was able to create one EC2 instance and run conferencing successfully.
also i have route53 configured and pointed to jitsi, cert is self signed and working as expected.
the next step is to have JVB on this EC2 disabled and another EC2 instance configured with JVB only.
any suggestions?

trying to decouple jvb and jitsi meet - no luck…

I am trying to do the same thing right now, and I think for the current stable release, this video is how you should be doing it:

If I get it to work, will let you know. Certainly simpler methods with less config, as described above, which seems to be coming soon, would make it a lot easier … :slight_smile:


@sergiy I have managed it to work. You can contact me if you need help in configuration!

did u set this up with a elastic load balancer as well so the video bridge’s increase and decrease as necessary depending on usage?

Scaling videobridges up and down is done through Auto Scaling Group in AWS. I don’t think ELB has any role in scaling videobridges.

There are few posts in the forum covering this - search for “scaling” “networkthroughput”. Also, the video link posted above speaks to it.



@sergiy I have managed it to work. You can contact me if you need help in configuration!

Hi @hmharshit

How did you done in aws, all models (jitsi meet, videobridge xmpp) in single server. I am looking for jitsi meet front end support server and back end anther server or if you done any multiple domains or sub domains in single server. If you done can you please share those details.

Thanks in advance.

Hi @Bheema_Vatti

My configuration is
I have one main VM on which I’m running docker images of web, prosody,jicofo.

Then one auto scaling group with multiple JVBs communicating with the prosody.

Then another autoscaling group with multiple Jibris communicating with the same prosody server.

I’m not using any load balancer since I’ve only 1 shard. Let me know if you have any doubts.

Hope it helps!