A Minecraft server calculates the world in ticks, 20 times per second. If a tick takes too long or the memory runs out, the server writes warnings to the log or, in the worst case, aborts. Each section below explains one of these messages: how to recognize it, how serious it is and what you can do. First come messages after which the server crashes or shuts down, then warnings where it keeps running.

You can find the full message in the Console or under Files in logs/latest.log, and crash reports in the crash-reports/ folder. Always read the lines before it too: they often show what slowed the server down. Before changing configuration or world files, create a backup under Backups.

Server stops responding: Watchdog abort

What the message looks like:

The server has not responded for 25 seconds! Creating thread dump
A single server tick took <Seconds> seconds

What is behind it: The watchdog is a guard that checks whether the server is still working. If a single tick hangs for too long, the server is considered frozen and the watchdog shuts it down. Until then, Paper writes intermediate messages such as "has not responded for 25 seconds" along with a thread dump, a snapshot of what the server is currently doing. Common causes are memory pressure or a heavy plugin or mod.

How to fix it:

  1. In logs/latest.log, read the lines before the first watchdog message. If another error or Stopping server already appears there, that error is the actual cause and the watchdog is only the consequence.
  2. In the thread dump, look at the top lines under Server thread. A plugin or mod name there shows what the server was last busy with.
  3. Check the memory. If OutOfMemoryError also appears in the log, it is not enough (sections further below).
  4. Update or remove the plugin you found under Plugins or the mod under Mods and restart the server.

In detail: Properly fixing "Can't keep up" and watchdog timeouts

Memory exhausted: Java heap space

What the message looks like:

java.lang.OutOfMemoryError: Java heap space
Exception in thread "<Thread>" java.lang.OutOfMemoryError: Java heap space

What is behind it: Java manages the server's memory in an area called the heap. It has filled up: there was not enough memory for the current load. As a result, the server usually crashes or becomes unstable. Typical memory hogs are many loaded chunks, large modpacks, many mobs and items, and plugins with large amounts of data.

How to fix it:

  1. Restart the server so that it runs stably again.
  2. Determine how much memory your setup really needs. In detail: Planning Minecraft RAM requirements
  3. If your package is not enough, choose a plan with more RAM under Settings & Plan → Plan.
  4. Until then, reduce the load, for example the view distance view-distance in server.properties.

Good to know: The thread in which the message appears is not automatically the culprit. A thread is a parallel strand of work, and this one just happened to be the first to ask for memory when none was left.

Server crashed: Crash report saved

What the message looks like:

This crash report has been saved to: crash-reports/crash-<Date>_<Time of day>-server.txt

What is behind it: The server was shut down by a hard error and wrote a crash report. The message itself only says where this report is located. The actual cause, such as a mod, a plugin, the world or missing memory, is in the file.

How to fix it:

  1. Under Files, open the crash-reports/ folder and the named file in it. Its name contains the date and time of the crash.
  2. Read the first section: after Description: there is a short description, followed by the error line and the stack trace, meaning the list of program locations that led to the error.
  3. The mod or plugin name mentioned there is the culprit. Update it under Mods or Plugins, or remove it.
  4. If OutOfMemoryError appears there, the memory is not enough. Choose more RAM under Settings & Plan → Plan.
  5. Start the server and watch the Console until it is fully running.

Good to know: Crash reports stay around. Delete reports you have read so that you immediately find the new one next time. For mod servers: Mod errors on Forge, NeoForge and Fabric servers

Error in the main game loop

What the message looks like:

Exception in server tick loop
Caused by: java.lang.<Error type>: <Details>

What is behind it: The main loop calculates the world tick by tick. An exception, meaning a program error, has propagated into this loop and disrupted it. The server then usually crashes and writes a crash report. The actual cause is almost always in the Caused by line that follows directly.

How to fix it:

  1. Look for the message in logs/latest.log or in the newest report under crash-reports/.
  2. Read the Caused by line directly below it. If there are several, the lowest one usually leads to the root of the problem.
  3. The lines with at … below it name packages and files. From these you can identify the plugin or mod involved.
  4. Update or remove it under Plugins or Mods and restart the server.

Good to know: This message is not a lag problem but a specific error. More memory only helps if the Caused by line names an OutOfMemoryError.

In detail: Properly fixing "Can't keep up" and watchdog timeouts

Memory at its limit: GC overhead limit exceeded

What the message looks like:

java.lang.OutOfMemoryError: GC overhead limit exceeded

What is behind it: Java automatically cleans up memory that is no longer needed. This is called garbage collection, or GC for short. This message means: Java spends almost all its time cleaning up and hardly recovers any memory in the process. The heap, meaning Java's memory, is practically full. The server barely responds or crashes.

How to fix it:

  1. Restart the server.
  2. Determine how much memory your setup needs. In detail: Planning Minecraft RAM requirements
  3. If your package is not enough, choose more RAM under Settings & Plan → Plan.
  4. Reduce the chunk and entity load: lower the view distance view-distance in server.properties, shrink large animal farms and clean up items lying around.

Good to know: A restart only buys you some breathing room. If usage rises steadily to the limit after every restart even though the load stays the same, a plugin or mod may also be holding on to memory.

No new thread possible: unable to create native thread

What the message looks like:

java.lang.OutOfMemoryError: unable to create native thread

What is behind it: A thread is a parallel strand of work. Java tried to start another one and could not. Despite OutOfMemoryError in the name, this is not proof of a full Java heap. Usually too many threads are running at the same time, a process limit has been reached or memory outside the heap is missing. The server can then no longer start new tasks and becomes unstable. This line does not reveal which plugin or mod is creating the threads.

How to fix it:

  1. Do not simply book more RAM across the board. More memory for Java does not automatically solve this message.
  2. In logs/latest.log, read the lines before the message: what was running at the time of the error, which plugins were reporting frequently?
  3. Prove the culprit at runtime. With spark profiler start --thread * in the Console you can see all running threads. Their names often reveal the plugin or mod.
  4. Only once a culprit is proven do you update it under Plugins or Mods, check its settings or remove it.

If you get stuck, open a ticket and include the time of the message. On spark: spark: Finding and understanding lag

Server can't keep up: "Can't keep up"

What the message looks like:

Can't keep up! Is the server overloaded? Running 2014ms or 40 ticks behind

What is behind it: Minecraft runs 20 ticks per second, so a tick may take 50 milliseconds. If ticks take longer, a backlog builds up, which the message states in milliseconds and ticks. The server keeps running, but noticeably slower: mined blocks reappear, mobs stutter. The cause is usually memory or CPU pressure. Occasional messages at startup or when generating new areas are normal.

How to fix it:

  1. Measure the load with spark tps in the Console. What matters is MSPT, the milliseconds per tick: up to 50 ms everything is healthy.
  2. Find the culprit with spark profiler start --timeout 120 and open the link the server outputs afterwards.
  3. Reduce the load: lower the view distance view-distance and the simulation distance simulation-distance in server.properties, pregenerate the world, tone down large farms.
  4. Check whether the memory is sufficient (section "Memory exhausted: Java heap space").

Good to know: spark is already included on current Paper servers; on other servers you install it as a plugin or mod.

In detail: Properly fixing "Can't keep up" and watchdog timeouts

Paper watchdog diagnostics: Server briefly hung

What the message looks like:

--- DO NOT REPORT THIS TO PAPER - THIS IS NOT A BUG OR A CRASH ---
Server thread dump (Look for plugins here before reporting to Paper!):

What is behind it: If the server thread hangs for a few seconds, Paper writes a diagnostic with a thread dump, meaning a snapshot of what the server is currently working on. As the first line says: this is not a bug in Paper and not a crash, but a clear sign of lag. If the server recovers, it stays a one-off hang. If it lasts until the limit, the watchdog abort from the first section follows.

How to fix it:

  1. In the dump, read the top lines under Server thread dump. If a plugin or mod name appears there, it is a strong hint at the culprit.
  2. Update this plugin under Plugins or remove it as a test.
  3. If the dumps pile up, check the memory, your plugins and the view distance view-distance in server.properties.
  4. Isolated dumps, for example at startup or when saving a large world, you can simply observe instead of intervening right away.

In detail: Properly fixing "Can't keep up" and watchdog timeouts

Server hangs while loading a chunk

What the message looks like:

[ChunkTaskScheduler] Chunk wait task info below:
[ChunkTaskScheduler] Chunk holder: NewChunkHolder{world=<World>, chunkX=<x>, chunkZ=<z>, …, currentGenStatus=minecraft:initialize_light, requestedGenStatus=minecraft:full, …}

What is behind it: Paper waited longer than expected for a chunk to load or generate, meaning a world section of 16 × 16 blocks, and wrote a diagnostic about it. The server stalls at that moment. currentGenStatus shows how far the chunk has gotten, requestedGenStatus the target. Usually overload is behind it: world generation, many players far apart or slow plugins. Less often a region is damaged.

How to fix it:

  1. Lower the view distance view-distance in server.properties.
  2. Pregenerate the world so that chunks are not first created while playing: Chunky: Pregenerating a Minecraft world
  3. Check plugins with spark profiler start --timeout 120 in the Console.
  4. If the diagnostic always names the same world and the same chunk coordinates, the region is damaged. Stop the server and restore a backup under Backups.

Good to know: If conspicuously large coordinates appear there, someone was traveling very far out. A world border with worldborder set <Diameter> limits that.

Paper watchdog notice block: Server did not respond

What the message looks like:

The server has stopped responding! This is (probably) not a Paper bug.
If you see a plugin in the Server thread dump below, then please report it to that author
If this is the case, consider increasing timeout-time in spigot.yml but note that this will replace the crash with LARGE lag spikes

What is behind it: These lines are a fixed notice text that Paper writes before the thread dump. The server's main thread has not responded for longer than the watchdog limit. The text itself names typical reasons: an overloaded plugin, very large world operations all at once, or HTTP and database accesses that block the server thread.

How to fix it:

  1. Read the thread dump below the block. If a plugin appears there, update, configure or remove it under Plugins.
  2. If you see world saving or editing there, split large actions, for example with WorldEdit, into smaller steps.
  3. If you see HTTP or MySQL accesses, a plugin is usually waiting for an external service or a database. Check its connection details.
  4. Increase timeout-time in spigot.yml only deliberately. As the notice says, you are trading the crash for long hangs.
  5. For repeated hangs, reduce the plugin load and check the memory.

In detail: Properly fixing "Can't keep up" and watchdog timeouts

Paper chunk task fails with BootstrapMethodError

What the message looks like:

[ca.spottedleaf.concurrentutil.executor.thread.BalancedPrioritisedThreadPool] Exception thrown from thread <Thread>
java.lang.BootstrapMethodError: bootstrap method initialization exception

What is behind it: Paper loads and generates chunks in its own threads, meaning parallel strands of work alongside the main game. In one of them, Java could not prepare a call site in the program. Possible reasons are a bug in Paper, an incompatible extension or an unsupported Java version. The message alone does not pinpoint any of these causes.

How to fix it:

  1. Update Paper under Version to a current stable build of your Minecraft version.
  2. Use a Java version that this build supports: Which Java version does your Minecraft server need?
  3. Read the full stack trace in logs/latest.log. If a Caused by line names a plugin, update it under Plugins.
  4. Do not swap server libraries by hand. That is more likely to cause new errors.
  5. If the error occurs repeatedly after the update, report the full stack trace to the Paper project.

Your message isn't listed?

Read the entire message in logs/latest.log and above all the lines before it: overload almost always has a trigger that comes before the first warning. In the stack trace or thread dump, package names, meaning the folder names separated by dots such as com.<Developer>.<Plugin>, and file names such as <Plugin>.jar reveal which plugin or mod is involved. Before changing configuration files, create a backup under Backups. If you get stuck, open a ticket and include the time of the message.