home-cluster VS kube-plex

In using flux for gitops. Repo is at https://github.com/Truxnell/home-cluster

Example: https://github.com/munnerz/kube-plex

There are some projects to distribute transcoding jobs like the Unicorn Transcoder (last activity in 2021), Remote Transcoder (last activity in 2017) or the Kubernetes project (last activity in 2020)

not OP, but I have some experience with Kubernetes, plex is the last service that I haven't yet migrated to my K3s cluster. Generally speaking, Kubernetes doesn't mean HA, it does mean that if a node goes down, plex will ideally spin on on another (assuming everything is correctly configured) but an application needs to be designed specifically to be deployed in a HA way in order to scale horizontally like you are suggesting. You can see in the picture that only 1 Pod (a unit that holds a container) is actually running. There is/was a project to separate our transcode jobs into their own pods to allow for horizontal scaling of transcode jobs, but it looks like it hasn't been updated in years, sounds a bit hacky and I haven't had the confidence to try it just yet.

There is a project out there for Plex in Kubernetes btw!

-> https://github.com/munnerz/kube-plex

I've been there and it's not officially supported by Plex or even Jellyfin or Emby. There's kube-plex which is not maintained anymore, probably the most success you could have is with unicorn transcoder however it forces you to stay pinned to a Plex version until the maintainers push an update.

Theres kube-plex, which does this https://github.com/munnerz/kube-plex

Btw it looks like kube-plex (with pod transcoding) isn't dead yet https://github.com/munnerz/kube-plex/pull/97#issuecomment-620610658