I'm new to server administration and homelabbing, and I'm running a Minecraft Java server in an itzg/minecraft-server container on Ubuntu 24.04. The server worked reliably for about seven weeks, including recovering from an earlier power outage. After several new outages, the host and Docker still appear healthy, but the Minecraft container now starts for roughly 12 seconds and then restarts.
The container was created with:
`docker run -d -p 25565:25565 -v /home/jbarratt/data/minecraft:/data --name mc -e EULA=TRUE --restart always itzg/minecraft-server`
The logs show that the server is resolving the unpinned `LATEST` version to Minecraft 26.2, then attempting to upgrade the existing world `cubeworld`. It fails with:
`CowFSCreationException: Cannot build copy-on-write file system, missing write access for file: ./cubeworld/data/minecraft/.game_rules.dat.swp`
The EULA is already accepted, and a separate test container was able to generate and run a new world successfully. Docker and other containers also work normally. The container's health check fails because Minecraft never gets far enough to listen on port 25565. What is the likely cause, and how should I repair the world permissions or otherwise get the original server running again?
3 Answers
The EULA does not appear to be the problem because the command includes `-e EULA=TRUE` and the generated `eula.txt` contains `eula=true`. The container’s restart policy is only exposing the underlying Minecraft startup failure: the process exits, Docker restarts it, and the health check reports that nothing is listening. Use `docker logs mc` and inspect the mounted files with `ls -l` and `stat`; correct ownership only after stopping the container and making a backup.
Compose would make this easier to maintain, but it will not by itself fix the current failure. First stop the container, back up the world, verify the mount points and permissions, and confirm that the world is not being used by another container. Then either restore writable ownership for UID 1000 or temporarily roll back to the Minecraft version that was running before the outage. Once it works, keep regular backups and avoid tracking `LATEST` in production.
The important part of the log is the copy-on-write error, not the health-check failure. Minecraft is trying to upgrade the existing world, but the process cannot write to `.game_rules.dat.swp` inside `cubeworld/data/minecraft`. Check ownership and permissions throughout the mounted `/home/jbarratt/data/minecraft` tree, especially that world directory. The container normally runs as UID/GID 1000, so files owned by another user or marked read-only can prevent the upgrade.
Related Questions
Can't Load PhpMyadmin On After Server Update
Redirect www to non-www in Apache Conf
How To Check If Your SSL Cert Is SHA 1
Windows TrackPad Gestures