«A single server tick took 60.00 seconds»

Errors & Fixes Reviewed September 6, 2026 2 min read

The server runs a watchdog thread whose only job is to notice when the main thread has stopped responding. If a tick runs far past its budget, the watchdog concludes the server is hung and shuts it down deliberately — printing a stack trace of exactly what it was doing.

What you are seeing

  • A single server tick took 60.00 seconds (should be max 0.05)
  • Considering it to be crashed, server will forcibly shutdown.
  • A long thread dump immediately after, listing every thread.
  • The server was fine a moment earlier.

Why it happens

Something on the main thread stopped or slowed to the point of not returning. Common causes are a plugin waiting on a database or a web request without a timeout, an enormous world edit, a chunk generation storm, or a garbage collection pause on a server that is out of memory. The watchdog does not cause any of it — it reports it.

Reading the dump

  1. Find the thread named as the server thread

    The dump lists every thread; only one matters. It is usually labelled as the main or server thread and appears near the top.

  2. Read its stack from the top

    The top few lines are what it was doing at the moment it froze. A plugin package name there is your answer.

  3. If it is a database or HTTP call, that is the bug

    Blocking network calls on the main thread are the classic cause. That is a plugin issue to report, not a setting to change.

  4. If it is garbage collection, look at memory instead

    A server at its memory ceiling can pause for a minute while it tries to free space. The watchdog fires, but memory is the real problem.

Raising the watchdog timeout hides the crash, not the freeze

You can lengthen or disable the watchdog, and the server will then sit frozen instead of restarting. Players still cannot play, the world still is not saving, and now nothing writes a diagnostic. The only good reason to change it is to capture a longer trace while debugging.

The early-warning setting is the useful one

Paper can log a warning before the timeout, showing what a slow tick was doing without killing the server. Turning that on gives you the same information from ticks that recover, which is often where the pattern is visible first.

How to confirm it worked

The server runs through its usual peak with no watchdog warnings and no ticks above a second in the log.