Hi!, thank you for your amazing project, it’s really awesome what it can do!.
Now I’m trying to improve an existing installation of jitsi-meet,
Original it was a all-in-one installation, Prososy, Jicofo, Video-Bridge and JIBRI in one big server on AWS.
Now I extracted VideoBridge and JIBRI to other instances, now the infra is like this:
4 x c5.4xlarge for Video bridge instances
*Conferences per VideoBridge: *
*Participants per VideoBridge: *
CPU per Video bridge
When there is more than 30 users per conference, lot of them experiments degradation on audio (normaly no use of video)
I made some tunnings like limit number os video senders, disable stadistics, limit to 480 the resolution, set the memory os java process to 16gb (VIDEOBRIDGE_MAX_MEMORY=16192m
), tunnings to the Linux and ethernet, etc. but still experiments problems when more then 30 users per conference.
Maybe I’m forgetting something, or there is another way to distribute load between VideoBridge instances, maybe reduce the amount of cpu and memory per instance and duplicate the number of VMs… please… any idea?
Thank you again!!!