[BUG] Docker won't start anymore because of the Nvidia GPU

[Crylia]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

Docker worked fine with my 1660 for a while but now it suddenly won't start anymore

Expected Behavior

Docker to function normal

Steps To Reproduce

Install docker
Install nvidia driver
confirm both working
add --gpu-all in the docker params
see it not working anymore

Environment

- OS: Unraid 6.12.8
- How docker service was installed: Via unraid

CPU architecture

x86-64

Docker creation

docker run
  -d
  --name='jellyfin'
  --net='bridge'
  -e TZ="Europe/Berlin"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Bocchi"
  -e HOST_CONTAINERNAME="jellyfin"
  -e 'JELLYFIN_PublishedServerUrl'='192.168.0.26'
  -e 'PUID'='99'
  -e 'PGID'='100'
  -e 'UMASK'='022'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:8096]'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/linuxserver/docker-templates/master/linuxserver.io/img/jellyfin-logo.png'
  -p '8096:8096/tcp'
  -p '8920:8920/tcp'
  -p '7359:7359/udp'
  -p '1900:1900/udp'
  -v '/mnt/user/Jellyfin/':'/data/tvshows':'rw'
  -v '/mnt/user/Jellyfin/':'/data/movies':'rw'
  -v '/mnt/user/Jellyfin/':'/media':'rw'
  -v '/mnt/user/jellyfin_data/':'/config':'rw'
  --gpus all 'lscr.io/linuxserver/jellyfin:latest'

15bb8699383217bc43a142677de7a0cabb4245a7db5bac531731083e485918ff
docker: 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: open failed: /proc/sys/kernel/overflowuid: permission denied: unknown.

The command failed.

Container logs

None, won't start

[github-actions]

Thanks for opening your first issue here! Be sure to follow the relevant issue templates, or risk having this issue marked as invalid.

[Roxedus]

This looks very much with a issue for the nvidia toolkit, and probably not a container issue.

[j0nnymoe]

Last I checked, --gpus all isn't the way you do this. You need to set the nvidia runtime + the ENV's required.

[Crylia]

Last I checked, --gpus all isn't the way you do this. You need to set the nvidia runtime + the ENV's required.

This is how I had it originally, I removed it because it wouldn't work and since the exact same issue came up with --gpus all I figured it would be sufficient

[Crylia]

This looks very much with a issue for the nvidia toolkit, and probably not a container issue.

So this mean's I'm out of luck?

[j0nnymoe]

Try reinstalling the plugin? Try an older driver version?

[Crylia]

Try reinstalling the plugin? Try an older driver version?

I tried another jellyfin docker (bbergle-jellyfin) and it works fine.

Maybe its an ubuntu related problem? Since the other docker is using debian

[j0nnymoe]

No issues here using jellyfin with Nvidia on unraid.

[Crylia]

Probably something todo with my specific setup then, I'd close it for now as switching the docker fixed it for me.