I have a virtual machine with 4 cores and 3 Gbyte RAM and a 1Gbit network
interface running Jitsi-meet.
When doing a conference with 2-3 participants the conference runs smoothly.
But with more participants some videostreams might grind to a halt and the
packet loss rises. The cores are almost not active at all, htop shows a
load of 7% max per core. And the memory load is about 800 Mbyte. Nagios
shows a network load of about 30 Mbit/s. This is of course momentary
measurements, but at least it gives an indication. The network is switched
with 100Mbit to the endpoints.
I suspected that the network is the bottleneck and setup a local test
environment which worked better, but still not ideal.
Are there any optimization tips and tricks that i can use to improve the