PerlandFailure catalogue

The server started with autopause on and its watchdog still active

Warningautopause_watchdog_enabled

The itzg/minecraft-server image warned at startup that the server's watchdog is on while autopause is enabled, so the server may be stopped when it resumes from a pause.

What the log shows

Written example4 lines
2026-11-05T09:12:41.118Z [init] [WARN] The server.properties for the server doesn't have the Server Watchdog (effectively) disabled.
2026-11-05T09:12:41.118Z [init] [WARN]          Autopause functionality resuming the process might trigger the Watchdog and restart the server completely.
2026-11-05T09:12:41.118Z [init] [WARN]          Set the MAX_TICK_TIME env variable (or max-tick-time property) to a high value (or disable the Watchdog with value -1 for versions 1.8.1+).
2026-11-05T09:12:41.204Z [init] Autopause functionality enabled

Your log will have other names, versions and times. Search a line of it from all failures.

What it means

Autopause in the itzg/minecraft-server Docker image freezes the server’s Java process while nobody is online. When it resumes, the tick that was running before the pause seems to Java to have lasted the whole pause. If the server’s watchdog is on, it takes that for a hung server and stops it.

Before the server starts, the image checks max-tick-time in server.properties (or the MAX_TICK_TIME setting). When the watchdog is not turned off, it prints this warning. Nothing has gone wrong yet: the server runs normally, but the first long pause can end in autopause_watchdog.

Why it happens

  • max-tick-time keeps its default (60000 milliseconds) or another value instead of -1.
  • MAX_TICK_TIME is set to a value other than -1. Newer versions of the image write -1 themselves and warn only when the setting asks for something else.
  • The image’s own -1 did not reach server.properties. In some versions, the value autopause chooses is not written into the file, so the check still sees the default.

How to fix

  1. Set MAX_TICK_TIME=-1 (or max-tick-time=-1 in server.properties), then restart the server. On versions older than 1.8.1, use a very large number instead of -1.
  2. After the restart, check that the warning is gone and that server.properties holds max-tick-time=-1.
  3. On Paper, Spigot and their forks, also turn their own watchdog off with -Ddisable.watchdog=true (with the itzg image, disable.watchdog:true in JVM_DD_OPTS).
  4. If the watchdog must stay on, turn autopause off with ENABLE_AUTOPAUSE=false.

In Perland

The issue appears in the server’s Diagnosis tab as a warning. On servers Perland pauses while idle, Perland writes MAX_TICK_TIME as -1 and turns Paper’s watchdog off before every start, so a restart at a time that suits you makes the warning go away. If the server’s settings set MAX_TICK_TIME to another value, Perland keeps it and the warning stays; remove that value to let Perland’s take over.

References

Log lines this page covers

  • The server.properties for the server doesn't have the Server Watchdog (effectively) disabled.
  • Autopause functionality resuming the process might trigger the Watchdog and restart the server completely.
  • MAX_TICK_TIME is non-default, for autopause to work properly, this check should be disabled