So I’m trying to get the linuxserver.io/jellyfin image to run with hw transcoding support on microk8s. Looks as if it’s not possible to explicitly pass in a /dev on a node in k8s.
Microk8s have it’s own set of gpu daemonset you can enable with ‘microk8s enable gpu’. The pass in a gpu by using a resource/limit/nvidia.com/gpu: 1 setting. But alas,doesn’t seem tot do anything.
Anyone got experience in this, is there a workaround or is it only possible to run on docker with hw transcoding?