Below is what I envision finally. We use AWS.
- A farm of Jitsi servers to as a cluster which auto scales/ unscales ( i hear OCTO for this)
- A farm of Jibri servers which auto scales and unscales. When jitsi conference is created a corresponding JIBRI to be created. If user chooses not to use the recording/ cutom RTMP then Jibri to be shutoff.
- At anyinstant have atleast 2 Jitsi, 2Jibris running. So that new conference is fast.
What we had tried so far.
Like many others, we installed the Jitsi, Jibri via the command line. Configured prosody etc. But the problem with this approach is its very very time-consuming. When we tried using the QA AMI images on prod, it was impossible. We had to recreate the images and hence went into several issues eating a lot of time.
Now we are starting off with docker. The point is Jitsi, Jibri we are able to get them running on 1 linux box. Not auto scalable like i thik of
Kubernetes is the best for our problem (I believe) but people talk about Kubernetes on the forum but nobody provides a tutorial to setup things and get going.
I am ok to go with AWS Auto scaling. But sofar Docker based installation is the easiest and quckest.
So the question to the community:
- Can somebody provide step by step process for auto scaling in kubernetis. OR
- Auto scale where i install jitsi and jibri on docker on 2 different machines on docker ( normal installation we are unable to do it) and then auto scale using AWS auto-scaling
- I saw the Kubernetes given in the docker project. But it works only on 1 node.
or any other quickest way to install and get the autoscaling done. I am new to docker, kubernetes so kindly consider giving links which gives details.