Roastd / Minecraft

critical

The server stalled for too long and the watchdog stepped in

What it looks like in your log

The server has not responded for 60 seconds! Creating thread dump

What causes it

The main thread was stuck for longer than the watchdog allows, so nothing ticked while it waited.

How to fix it

Start the server again. If it repeats in the same place, read the thread dump under "Server thread dump": the first frame that is not Java, Minecraft or Bukkit code is the plugin or mod that was holding the tick.

Why it happens

The watchdog is a separate thread whose only job is to check that the server thread is still ticking. If one tick runs past max-tick-time (60 seconds by default), it writes a thread dump and kills the server, because a server that has not ticked in a minute is not coming back. The dump shows where the server thread was stuck.

The watchdog reports the hang. It never causes it. Raising max-tick-time or setting it to -1 stops the report, not the hang, and leaves you with a frozen server that nobody kills.

Fix it step by step

  1. 1Find the thread dumpBelow the "has not responded" line is "Server thread dump" and a stack trace. That stack is where the server thread was when the watchdog fired.
  2. 2Read from the top until it stops being MinecraftFrames in net.minecraft, org.bukkit or java.* are the game doing what it was asked. The first frame in a plugin or mod package is the code that asked. That is the culprit.
  3. 3If it is chunk generation, look at the worldA stack in world generation or structure placement means a chunk that cannot finish generating, usually a worldgen mod hitting a bug in one place. Note the coordinates from any nearby log lines and avoid or regenerate that area.
  4. 4If it is a plugin task, look at the configA plugin scanning the world, saving to a database on the main thread or waiting on a network call will hang the tick. Its config usually has an option to move that work off the tick or run it less often.
  5. 5Restart and watch for a repeatOne watchdog kill after a heavy operation can be a one-off. The same stack twice is not.

If that didn't work

If the dump shows the server thread in a sleep or wait with nothing above it, the machine itself paused: an oversold host, a VM migration or a disk that stopped responding. The server was fine; its host was not.

Lines reading "Seems like you crashed and the server got stuck" after this are the aftermath, not a second problem.

Not sure this is your problem? Paste your log and Roastd will tell you which of these it actually matched, and what else it found.

Check your own log
Read more: A single server tick took too long →

Errors that travel with this one