Jellyfin no longer works with NVIDIA after KVM install

Hi all,

I had my Jellyfin setup up and running well, utilising my P2000 to transcode when needed.

I installed KVM and then noticed that my docker container for Jellyfin will no longer start.

This is my compose:

---
version: "2.1"
services:
  jellyfin:
    image: lscr.io/linuxserver/jellyfin:latest
    container_name: jellyfin
    environment:
      - PUID=1000
      - PGID=100
      - TZ=Europe/London
      - JELLYFIN_PublishedServerUrl=10.0.0.10
      - NVIDIA_VISIBLE_DEVICES=all
    volumes:
      - /appdata/jellyfin:/config
      - /srv/mergerfs/data/media:/media
      - /transcoding:/transcoding
    ports:
      - 8096:8096
    restart: unless-stopped
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]

This was working well, and I followed the guide to get my NVIDIA drivers installed fine.

But like I say since I installed KVM it no longer starts, this is the error from Portainer:

failed to deploy a stack: Network jellyfin_default Creating Network jellyfin_default Created Container jellyfin Creating Container jellyfin Created Container jellyfin Starting Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy' nvidia-container-cli: initialization error: nvml error: driver not loaded: unknown

I'm not sure if the KVM is a red-herring or not as I didnt notice JF being down right away, it was a few days later.

Or an update that may have run.

Also not sure if just a good ol fashioned reboot of OMV would solve this or not?

I did restart docker and this did not help.

Quote from ryecoaaron

I can't think of anything that installing the kvm plugin would cause this. Kind of looks like your nvidia driver is not loaded. maybe rebooting would help.

Just rebooted to update to new OMV version, no joy, still same error message

Ok so fixed it:

Went back to the guide and re-ran:

sudo m-a prepare

This worked and all good now. Thanks

Quote from ryecoaaron

What is the output of: lsmod | grep nvidia

root@omv:~# lsmod | grep nvidia

nvidia_modeset 1204224 0

video 65536 1 nvidia_modeset

nvidia_uvm 1216512 2

nvidia 35532800 362 nvidia_uvm,nvidia_modeset

drm 614400 7 drm_kms_helper,drm_vram_helper,ast,nvidia,drm_ttm_helper,ttm

But yeah got it resolved via re-running: sudo m-a prepare

Restarted Docker and all good again.

Quote from ryecoaaron

good to hear.

Thank you very much tho, appreciate all you do for us