10.9.2 docker crash

[toytown]

Just the default "bridge" mode. To be fair its only crashed twice in the last 8 months so it's not that big of a deal, just hoping the logs might provide some info for the developers so they can resolve it.

[toytown]

So, same thing happened on 10.10.5 last night, crashing the docker.

There was no transcodes happening at the time, server had tons of free ram/disk space. There was a single client logged on who had got half way through watching a TV show and then paused it, it was in this paused state for a few hours. The TV show had fully transcoded, so that part can be removed from the equation. The logs seem to indicate its the same problem as what happened with my original post.

Restarting the docker brings back jellyfin perfectly with no problems.

https://text.is/3Z37R

[TheDreadPirate]

In your case it looks like your server lost network connectivity. Can you share your docker compose? And whether you are using a reverse proxy.

[toytown]

Servers local network is fine, all the other services running at the time reported no errors and no logs on the router to say if the port had any issues or was dropped.

The docker (official) is created by the command line using "docker run -d -p 8096:8096 --name jelly --privileged --group-add="106" --device /dev/dri/renderD128dev/dri/renderD128 --device /dev/dri/card0dev/dri/card0" and then a bunch of mount points for storage, so i don't think this is an issue either. No proxy server is being used.

If the server lost connection to the client (uk>usa) then that could give the "no route to host" issue, but it shouldn't be allowed to bring down jellyfin.

[TheDreadPirate]

Can you try switching to host networking?

[toytown]

I'll give it a shot next time i update the docker. The big problem with troubleshooting this, is it only happens once every 3-4 months.

[toytown]

Is your container using bridge or host networking?

Sorry for the EPIC late reply, the docker is using "bridge" networking.   Jellyfin crashed again yesterday with the same "no route to host", every other service on the server is working fine (including all networking).  What i can tell, is that the user (remote) simply powered off his Nvidia Shield (ungraceful shutdown of jellyfin client i guess) and that pretty much killed the server whilst watching a TV Show.   Docker is 10.10.7 and client is 0.18.11.

It happens so infrequently, hopefully fixed in the upcoming release!

Code:

[2025-08-08 22:08:17.082 +00:00] [WRN] [83] Emby.Server.Implementations.HttpServer.WebSocketConnection: WS "192.168.1.1" error receiving data: "The remote party closed the WebSocket connection without completing the close handshake."
[2025-08-08 22:08:17.084 +00:00] [INF] [83] Emby.Server.Implementations.HttpServer.WebSocketManager: WS "192.168.1.1" closed
[2025-08-08 22:08:17.082 +00:00] [ERR] [24] Microsoft.AspNetCore.Server.Kestrel: Unexpected exception in TimingPipeFlusher.FlushAsync.
System.Net.Sockets.SocketException (113): No route to host