Hello Jitsers. I have set up an environment where in an EC2 I have JICOFO, PROSODY and NGINX (JVB disabled) and two other EC2 with only JVB.
When performing some load tests, I noticed that JICOFO uses JVB1 until the service crashes.
Shouldn’t Jicofo select the idle JVB for newly created conferences? It seems to me that it is not taking into account the stress_level of the JVB at any time, and that is bad.
jitsi-meet installed version: 2.0.6865-2 (jicofo 1.0-846-1)
Two JVBs with version 2.1-681-g3544ed05-1
Welcome to the forum.
Did you configure the bridges to loadbalance?
Hi Freddie, yeap!
My log of prosody:
Jun 06 12:34:33 c2s556e16b18610 info Client connected
Jun 06 12:34:34 c2s556e16b18610 info Stream encrypted (TLSv1.3 with TLS_AES_256_GCM_SHA384)
Jun 06 12:34:34 c2s556e16b18610 info Authenticated as email@example.com
Jun 06 12:34:34 c2s556e16b742c0 info Client connected
Jun 06 12:34:34 c2s556e16b742c0 info Stream encrypted (TLSv1.3 with TLS_AES_256_GCM_SHA384)
Jun 06 12:34:34 c2s556e16b742c0 info Authenticated as firstname.lastname@example.org
Jun 06 12:34:37 c2s556e1642c4b0 info Client connected
Jun 06 12:34:37 c2s556e1642c4b0 info Stream encrypted (TLSv1.3 with TLS_AES_256_GCM_SHA384)
Jun 06 12:34:37 c2s556e1642c4b0 info Authenticated as email@example.com
This doesn’t show that you’ve configured your environment to loadbalance multiple bridges.
How did you configure it?
I created an EC2 template with JVB installation;
When uploading a JVB, in userData, a script is executed to replace/update the information of:
- jvb.conf file: prop
server-id in websockets;
- sip-communicator file: prop
systemctl restart jitsi-videobridge2…
I’m checking my script, maybe something is wrong!!!
Check your browser’s js console for errors.
Also, stop the local JVB and see if JMS can use any of the remote JVBs.
@Freddie It could have been something wrong with the script itself! I just uploaded two jitsi-torture and this time each jvb received a conference.