A single server tick took 60.00 seconds: watchdog crashes

The watchdog stops a server that stopped responding. Setting max-tick-time to -1 silences it, which is right sometimes and a mistake other times. Here is which.

The full message reads something like:

[Server Watchdog/FATAL]: A single server tick took 60.00 seconds (should be max 0.05)
[Server Watchdog/FATAL]: Considering it to be crashed, server will forcibly shutdown

Everyone reads the watchdog line and concludes the watchdog is the problem. It is not. It is the smoke alarm.

#What the watchdog does

Minecraft aims for 20 ticks per second, so each tick should take at most 50 milliseconds. The watchdog runs on a separate thread and checks whether the main thread finished its tick within a timeout, which defaults to 60 seconds.

If not, it assumes the server has hung, prints a thread dump, and shuts down. The logic is sound: a server stuck in a tick is not going to unstick itself, and a shutdown that produces diagnostics beats a process that sits there indefinitely.

#Reading the thread dump

The watchdog prints every thread's stack. The one that matters is labelled Server thread, and the first ten or so lines of it show exactly what was executing when time ran out.

Look for a package name that is not net.minecraft. Something like com.someauthor.somemod.SomeClass.someMethod names both the mod and the operation. That is the answer, and it is usually within the first few frames.

If the whole stack is vanilla and mentions chunk generation or structure location, see the next section.

#The four common causes

1. World generation. Generating chunks in a heavy modpack legitimately takes a long time, especially with mods adding dimensions or large structures. First boot, a player flying into new territory, or a /tp to distant coordinates can all exceed 60 seconds.

This is the case where the watchdog is wrong, and where raising the limit is the correct fix.

2. A structure search that never terminates. /locate for a structure that does not generate in the current world, or a mod searching for something it will never find, walks outwards through chunks until something stops it. The watchdog is what stops it.

3. A mod deadlock. Two operations waiting on each other. The stack shows a thread parked or waiting rather than doing work, and it never resumes. This is a genuine bug in a mod and the fix is updating or removing it.

4. Disk stalls. If chunk writes hang, the tick waiting on them hangs too. Rare, and usually accompanied by other I/O errors in the log.

#Whether to raise max-tick-time

Setting max-tick-time=-1 in server.properties disables the watchdog.

Do it on a heavy modpack. All The Mods, RLCraft, FTB packs and anything with custom world generation will exceed 60 seconds during normal chunk generation. Being killed halfway through generating a chunk is considerably worse than waiting for it, and it can leave partially written region files behind.

Do not do it on a vanilla or lightly modded server. A 60-second tick there means something is genuinely stuck, and turning off the alarm leaves you with a server that hangs indefinitely instead of one that restarts and tells you why.

The setting is on the Settings tab, or in server.properties directly.

#After a watchdog crash

Before restarting, check whether the crash left damage behind. A server killed mid-chunk-write can produce corrupted region files, which show up later as chunk errors rather than immediately.

If the same watchdog crash repeats at the same point every time, it is not load. Something specific is hanging, and the thread dump names it.

Written and maintained by the Hostd engineering team. Last updated 2026-08-05. Notice a mistake? Tell us.