Frequent disconnection issues in an environment of 40 to 80 users

Hi support team,

First of all, thank you for your work on this software, which is particularly useful in staying connected with each other during this difficult time.

We run into problems on a regular basis and have no idea what to do. Occasionally, Jitsi stops responding to queries and all users are logged out because Jicofo can no longer speak to the bridges.

Here is what we see in the logs during these disconnections:

Jicofo logs :

Jicofo 2020-12-17 16:25:57.865 SEVERE: [3699] org.jitsi.jicofo.AbstractChannelAllocator.log() jvbbrewery@internal.auth.visio.prive.domain.our/2760437a-7299-476b-910b-6ba5e919b87a - failed to allocate channels, will consider the bridge faulty: Timed out waiting for a response.
Jicofo 2020-12-17 16:25:57.866 SEVERE: [3699] org.jitsi.jicofo.JitsiMeetConferenceImpl.log() One of our bridges failed: jvbbrewery@internal.auth.visio.prive.domain.our/2760437a-7299-476b-910b-6ba5e919b87a
Jicofo 2020-12-17 16:25:57.881 WARNING: [3699] org.jitsi.jicofo.AbstractParticipant.log() Canceling ParticipantChannelAllocator[BridgeSession[id=61574_35fc45, bridge=Bridge[jid=jvbbrewery@internal.auth.visio.prive.domain.our/2760437a-7299-476b-910b-6ba5e919b87a, relayId=null, region=null, stress=0.07]]@130839928, Participant[resopb@conference.visio.prive.domain.our/e399608e]@527340874]@1582538609
Jicofo 2020-12-17 16:25:57.882 WARNING: [3699] org.jitsi.jicofo.bridge.BridgeSelectionStrategy.log() Failed to select initial bridge for participantRegion=null
Jicofo 2020-12-17 16:25:57.882 SEVERE: [3699] org.jitsi.jicofo.JitsiMeetConferenceImpl.log() Can not invite participant – no bridge available.

JVB logs :

2020-12-17 16:28:55.779 WARNING: [8918] CountingErrorHandler.packetHandlingFailed#62: Failed to handle packet:
2020-12-17 16:28:55.782 SEVERE: [8598] [confId=bf2167e48718ed91 epId=cb1b0fdd local_ufrag=5i2ji1epoksbpp gid=198439 stats_id=Magdalen-SJz conf_name=po@conference.visio.prive.domain.our] IceTransport.send#225: Error sending packet

It’s a fresh install, so we’re using the latest software versions on Debian.

Any idea what could go wrong?

Best regards and thank you.

Any help ?

Can you describe how you exposed your server?
Is your server under NAT?

You give us some more information

This guide: Self-Hosting Guide - Debian/Ubuntu server · Jitsi Meet Handbook
Section: Advanced configuration and Systemd/Limits

Did you read it?

Bye Bye…

you did not even describe the hardware/bandwidth, is your system properly dimensioned ?

Thank you for your kind answers.

The server used is a “PowerEdge R640 Server”
Is :

  • Core: 80
  • RAM: 200GB
  • Bandwidth: 1GB

I followed the installation guide meticulously.

systemd is configured with these limitations:
sysctl -a | egrep -i “fs.file-max”
fs.file-max = 9223372036854775807

DefaultLimitNOFILE = 130000
DefaultLimitNPROC = 130000
DefaultTasksMax = 130000

Ah, and the server is not running over NAT

cpu and ram are certainly well dimensioned; even generously overdimensioned, for this kind of load it’s more often suggested to have 4 Cpu and 8 Gb ram.
1Gbits/s bandwidth seems miserly in comparison but should be enough I think unless all users have 4K screens, usually for 1080 an average bandwith use per client is given at 2.5 Mbits/s so with 80 clients it should be good. This error sending packet is concerning though; are you sure you don’t have hardware errors on network connection ? did you do some network monitoring ?