Direct answer

A watchdog crash means the main server thread stopped making progress for far too long. The default max-tick-time is 60000 ms, and Paper may print watchdog thread dumps earlier while the server is stalled. Read the first useful frames of the server-thread stack and reproduce the stall with spark if possible.

Do not disable the watchdog as the first fix

Disabling the watchdog hides the safety stop, not the stall.

Setting max-tick-time=-1 can leave a deadlocked or permanently stuck server process alive. It does not make the underlying operation faster.

Read the watchdog dump

Paper’s troubleshooting guide recommends checking the top of the server thread stack because it shows what the main thread was doing when the dump was captured. Plugin package names, chunk/world calls, file I/O, or repeated entity logic can point toward the next test.

Profile before the next crash

If the stall is repeatable but not instantly fatal, run spark during the problem. If it happens when one player joins, teleports, generates chunks, or triggers a machine/farm, include that action in the reproduction.

Take a backup before changing world files

If the stack points at world/chunk data, stop the server and back it up before repair attempts. Never modify active world files while the server is still running.

Official references

These references were checked while preparing this fix. Version-specific behavior can change, so use the project documentation for your exact release when a detail differs.

Was this article helpful?
Help us improve this article

Tell us what is missing, unclear, outdated, or incorrect. You can also suggest something we should add.

What could we improve?
Do not include passwords, server IPs, or other private information.0 / 1000