Multiple areas cannot enter the same conference room

I have two problems: 1. I use rapid installation and deployment of complete jitsi-meet in areas A and B. Now A creates a meeting room in its own area jitsi-meet. B can not enter the room created by A through the jitsi-meet in which B is located. A and B build Octo through, which seems to have no effect.
2. Can one jvb point to two xmpps and work at the same time?
Thank you very much for your help.

The idea with octo is not participants in different location to use different signalling nodes, they should use the same and report to jicofo its region they are in then jicofo will interconnect jvbs and will use the closest jvb to the client that requests it. But all participants for a conference will use the same signalling node (same jicofo).

Thank you very much for your reply. I see what you mean: one jicofo can have multiple jvbs under it, and report their location to jicofo through octo. It’s a great way for jicofo to use the latest jvb to the client requesting him. Now I need to deploy two or more jicofos to share the location information of jvb. When one jicofo goes down, another jicofo can enter the same room. Is there a better solution? I look forward to your reply. Thank you very much.

There is no such thing. So everything is divided in so called shards. A shard consist of a signalling node and several jvbs. The signalling node normally has a webserver, a xmpp server prosody and jicofo.

For failover there is a pool of haproxy server which get the http traffic and send it to closest signaling node, base on the ?room= query parameter, this way the same conference go at just one place. When a shard goes down, HAproxy can be instructed to not use that shard and will start using a new one and all clients will reload in a new room (same name but on different shard, different xmpp server).
Check the website and the news section there are few youtubue videos you can watch, that describe this.

Maybe it’s not clear what I’m describing. Sorry. Let’s say, for example, there are two signaling nodes under haproxy. Each signaling node has more than one jvbs. If one signaling node goes down, can you enter the previous room through another signaling node? Is there a way to achieve this? Sorry, I haven’t found the relevant information to fulfill this requirement. I want to achieve high availability. I look forward to your reply again.Thank you

How you can go in the same room, when this room was in the xmpp server and that server is down?

I described you above how is the availability handled. When a node goes down the participants in that room will detect that and will automatically reload and will enter all of them in a room with same name. This is how currently works.

Seeing this architecture information, I always thought that jvbs can share multiple xmpp information at the same time, which is what I always wanted to understand. Thank you for your help. I still have a question, that is, when a xvpp is below multiple jvb, each jvb belongs to region1, region2, region3… Then the layoutInfo parameter of config.js under my meet module needs to be defined, then the corresponding area can be The room is allocated? I see your information in the community, userRegion, region is fixed under the deploymentInfo parameter, which does not match the dynamic allocation to different region1, region2, is there a logic to determine whether to add? Thank you

What do you mean ‘whether to add’? If you want jicofo to choose closest jvb, it needs to know clients region.

This is my jvb configuration:*
org.jitsi.videobridge.PUBSUB_SERVICE=Domain Name

This is my jicofo configuration:
org.jitsi.focus.pubsub.ADDRESS=domain name
This is the configuration of my meet:config.js:
deploymentInfo: {
// shard: “shard1”,
// region: “europe”,
// userRegion: “asia”
Environment: ‘domain name’,
envType: ‘prod’,
releaseNumber: ‘45’,
Shard: “sharedStatsNode”,
Region: “cn-guangdong-1”,
userRegion: “cn-guangdong-1”
rttMonitor: {
Enabled: false,
initialDelay: 30000,
getStatsInterval: 10000,
analyticsInterval: 60000,
stunServers: {“cn-guangdong-1”: “Domain Name: 443”}

But every time I create a room, I still don’t distribute it to the specified room. Is there a problem with the deploymentInfo parameter setting?But every time I create a room, I still don’t distribute it to the specified room. Is there a problem with the deploymentInfo parameter setting? Cannot correctly judge jvb area

I can now confirm the region according to the browser, and can successfully call the config.js file corresponding to the meet, and then how to call the jvb of the backend according to the userRegion. I set this up and didn’t create a room in the corresponding jvb. Thank you

deploymentInfo: {
Environment: ‘domain name’,
envType: ‘prod’,
releaseNumber: ‘45’,
Shard: “sharedStatsNode”,
Region: “cn-guangdong-1”,
userRegion: “cn-guangdong-1”

Damencho, as far I can see you have signalling servers in every region.

How do you tell your haproxy instances to send requests for a specific room to the same signalling server? Do haproxy instances exchange with some kind of state information?

Yep, they are connected together and updating each other syncing their stick tables.

1 Like