[BUG] Jellyfin not starting on Docker

[nxs-neo]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

Jellyfin container is stuck on starting, but does not actually start.
There are no error messages in the terminal. I guess it started after the latest update.

Expected Behavior

The container starts and web ui is accessible.

Steps To Reproduce

1. run docker compose up -d

Environment

- OS: Arch Linux
- How docker service was installed: distro package manager from cachy os repos

CPU architecture

x86-64

Docker creation

jellyfin:
    image: lscr.io/linuxserver/jellyfin:latest
    container_name: jellyfin
    runtime: nvidia
    gpus: all
    environment:
      - PUID=1000
      - PGID=984
      - TZ=Etc/UTC
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=all
    volumes:
      - ./jellyfin:/config
      - /run/media/user/backup-disk/servarr/:/data
    ports:
      - 8096:8096
      - 8920:8920
      - 7359:7359/udp
      - 1900:1900/udp
    restart: unless-stopped

Container logs

"docker logs jellyfin" hangs and does not output anything.

[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]

Without logs theres not much we can do. It does sound like the docker install might be broken, or in need of a reboot. Check the usual suspects like dmesg ang the docker daemon logs

[Obstidian]

I can confirm this issue is occuring on my unraid Docker after update too.
from the appdata/jellyfin/log folder, it seems that the container didn´t make any logs. The terminal log shows nothing too.
the only output that I get is

Edit: I turned off my VM that I passed trough my GPU, and the container started. @nxs-neo maybe if you have some VM with gpu passtrough, try shutting it down.

[nxs-neo]

No I'm not running a vm I'm running it through docker installed directly on my system. I tried turning off the nvidia stuff but it didn't work.

[hsteinmetz]

Can confirm, the container is stuck at starting/created and can't be removed or restarted. I've tried reinstalling docker, rebuilding the container and pruning everything but it still doesn't work. Every other container and image works flawlessly so it's definitely an issue with this image

[Roxedus]

The fact that there is no logs only means that docker doesn't even get to the image part of its run sequence.
I also cannot reproduce these symptoms. Looking at system and service logs is the next logical step in troubleshooting.

[j0nnymoe]

Just upgraded my instance with zero issues on unraid.

Also smoke tests work fine: https://ci-tests.linuxserver.io/linuxserver/jellyfin/latest/index.html

[Protarios]

Have the same issues. Not running anymore on Unraid and no logs.
Every other docker container is running fine.

[Koko2110]

Chiming in to say I am having this same behavior as well, also with no logs to provide.

I was passing an Nvidia GPU to the Jellyfin docker instance, but even commenting those lines out now doesn't help.

I suppose I will look into downgrading the image for now.

Update: Very confused. I downgraded to some previously known well working versions of the Jellyfin image and the result is the same. I even downgraded pacman's docker-compose, as I noticed there have been updates to it over the past few days, but no improvement. Did some sort of kernel update or a kernel module we have in common get updated recently that broke this?

It is probably worth pointing out that the container is created properly, the "Starting" part is where we get stuck. This, coupled with my playing around with different versions of the image, leads me to believe that it is in fact not the image that is problematic. Can't even say it's an Arch thing, as some of you are on Unraid.

[nxs-neo]

Can confirm, the container is stuck at starting/created and can't be removed or restarted. I've tried reinstalling docker, rebuilding the container and pruning everything but it still doesn't work. Every other container and image works flawlessly so it's definitely an issue with this image

My exact issue. What distro, kernel wtc are you running?

[hsteinmetz]

My exact issue. What distro, kernel wtc are you running?

I'm on Arch with kernel release 6.14.6-arch1-1. Docker is the newest from the repos, so 28.1.1 and and compose is v2.36.2.

Update: Very confused. I downgraded to some previously known well working versions of the Jellyfin image and the result is the same.

That's interesting, I downgraded the image to 10.10.6 and everything works just like before.

[Koko2110]

That's interesting, I downgraded the image to 10.10.6 and everything works just like before.

I hadn't downgraded that far back at the time of writing, but even now trying 10.10.6ubu2404-ls57 doesn't help. Which version exactly did you go back to? Did you do anything other than setting the image in the yml?

I am also on Arch with the exact same kernel, docker and docker compose versions.

[hsteinmetz]

Which version exactly did you go back to? Did you do anything other than setting the image in the yml?

Build version of the image is 10.10.6ubu2404-ls57, so the exact same as you have. The problem (at least for me) seems to be the nvidia container runtime. I had it configured just like OP and as soon as I disabled it, everything (even the latest image version) worked again.