Closed PassiveLemon closed 9 months ago
Thanks for opening your first issue here! Be sure to follow the relevant issue templates, or risk having this issue marked as invalid.
I suspect it has to do with the Nvidia devices not being fully initialized at launch time but there's nothing in the logs, that I can see, that would indicate the problem.
The container works fine with Nvidia. It's likely due to something specific with nixos. You wouldn't normally have to mount all those Nvidia volumes as the Nvidia runtime handles that all for you.
You likely have to delay the docker Daemon from starting.
I just set the Docker systemd service to start after 15 seconds and it still occurs so I'm thinking its not Nvidia. I'll do some more testing and see what mounts I can remove. I remember I had problems getting hardware transcoding to work and setting those did fix it.
I noticed a large gap in time in these 2 lines:
[15:40:34] [INF] [10] Emby.Server.Implementations.Session.SessionManager: Reissuing access token: fb4a37497e5046b293a5b76f9c88070c
[15:41:30] [INF] [2] Main: Received a SIGTERM signal, shutting down
Could this indicate something?
Does the container start fine without Nvidia?
The solution was to remove those device mounts.
What I'm guess happened is that, at some point a while ago, I updated my system and didn't realize the Nvidia driver got updated in the process, which requires a reboot, so when transcoding stopped in the container, I added those mounts and restarted the system which made me think those mounts did something when it was just the restart.
Is there an existing issue for this?
Current Behavior
When first booting up my system, all my containers will start except for Jellyfin (its set to always restart). If I start it after the fact, it will launch fine.
Expected Behavior
Jellyfin launches upon first launch on a system boot.
Steps To Reproduce
Environment
The container is set up to support hardware accelerated transcoding.
CPU architecture
x86-64
Docker creation
Container logs