Minecraft Server Keeps Crashing — Complete Fix Guide (2026)

Your Minecraft server is crashing again and you don't know why. The first step is reading the crash log — but even then, error messages like java.lang.OutOfMemoryError or Watchdog thread killed can be confusing. This guide walks through the 8 most common crash causes and exactly how to fix each one.

Step 1: Read the Crash Log

Before applying any fix, identify the actual cause. Crash files are in two locations:

• /crash-reports/ — for hard crashes (contain a full stack trace)
• /logs/latest.log — for soft errors, plugin exceptions, and all server output

Open the most recent crash report. The critical section starts with -- Head -- and the first java.lang. line is usually the root cause. Paste it into Google — specific exceptions have well-known fixes.

Crash Type 1: OutOfMemoryError

Your server ran out of allocated RAM. The JVM ran out of heap space and crashed. Most common cause on free/cheap hosting with 1GB RAM.

Fix: Increase -Xmx Allocation

# Change your start command:
# Before (bad):
java -jar paper.jar

# After (correct) — use Aikar flags:
java -Xms2G -Xmx4G --add-modules=jdk.incubator.vector \
  -XX:+UseG1GC -XX:+ParallelRefProcEnabled \
  -XX:MaxGCPauseMillis=200 -jar paper.jar --nogui

Set -Xmx to 75% of available RAM. If your host gives you 2GB, use -Xmx1500M. Never set it above 80% — leave room for the OS.

If You Can't Increase RAM: Reduce Memory Usage

# server.properties
view-distance=6       # down from 10
simulation-distance=4 # down from 10

# paper-world-defaults.yml
entity-per-chunk-save-limit:
  experience_orb: 16
  snowball: 8
  arrow: 16

Crash Type 2: Watchdog Thread Crash

The Watchdog is a Paper/Spigot safety mechanism — it kills the server when the main thread is frozen for too long (default: 60 seconds). This means something caused the main thread to hang, not a normal crash.

Common Causes:

• Database query running synchronously on main thread (bad plugin code)
• Chunk loading taking too long (large redstone contraption / sudden explosion)
• Infinite loop in a plugin's tick method
• File system I/O hanging (especially on cheap VPS with slow disks)

Fix:

# After restart, wait 5 minutes, then run:
/spark profiler start
# Let it run for 2-3 minutes:
/spark profiler stop
# A web URL will appear in console showing exactly what's using CPU

Crash Type 3: Plugin Conflict / ClassNotFoundException

Two plugins are incompatible — usually because one plugin depends on an older API version of another. Very common when mixing plugins from different years.

Fix: Binary Search for Conflicting Plugin

1. Move half your plugins out of /plugins/ into a temp folder
2. Restart. If it no longer crashes, the conflict is in the half you moved
3. Move half of that set back. Repeat until you isolate the pair causing the conflict
4. Update the conflicting plugins to their latest versions

Crash Type 4: Corrupt Chunk

A region file in your world has been corrupted — usually from a hard disk failure, power loss during write, or host-side storage issue. Corrupted chunks cause a crash loop when players approach that area.

Fix: Delete the Corrupted Region File

# The error log shows coordinates like [18, -3] — these are CHUNK coordinates.
# Convert chunk to region file:
# region_x = floor(chunk_x / 32) = floor(18/32) = 0
# region_z = floor(chunk_z / 32) = floor(-3/32) = -1
# File to delete: world/region/r.0.-1.mca

# WARNING: This destroys all builds in that region (32x32 chunks = 512x512 blocks)
# Backup your world first before deleting!

After deleting the file, the chunk will regenerate as a fresh vanilla chunk when players approach that area.

Crash Type 5: Wrong Java Version

Minecraft 1.21 requires Java 21. If you're running Java 8 or Java 17 and try to launch Paper 1.21.x, it will crash immediately.

# Check your Java version:
java -version
# Must show: openjdk version "21.x.x"

# Install Java 21 from adoptium.net
# Then verify again before starting server

Crash Type 6: Timings Overload (Not a Crash — But Causes Lag Spikes)

This is a warning, but if it appears constantly, the server will eventually crash or become unplayable. The main thread can't process 20 ticks per second — it's falling behind.

Common causes and fixes:

• 🔥 Too many entities → install ClearLag, set entity limits in paper.yml
• 🔥 Redstone clock loops → find and disable them (/paper badentities)
• 🔥 Too many chunk loads → reduce view-distance to 6-8
• 🔥 Spawner farms → set mob spawn limits in bukkit.yml
• 🔥 Core plugin doing sync database calls → check Timings via /spark

Crash Type 7: Server JAR Corruption

Sometimes the Paper/Spigot JAR itself gets corrupted — especially on cheap shared hosting with unreliable storage. Signs: server starts then immediately crashes with a generic error, or Error: Unable to access jarfile paper.jar.

Fix:

Re-download the JAR from papermc.io and replace the old one. Also check that disk space isn't full (df -h on Linux) — a full disk causes identical symptoms.

Crash Type 8: Startup Script Errors

# Wrong: using -Xmx more than available RAM
java -Xmx16G -jar paper.jar  # on a 4GB server = instant crash

# Wrong: conflicting JVM flags
java -XX:+UseG1GC -XX:+UseZGC -jar paper.jar  # can't use two GC algorithms

# Wrong: old Java flags on Java 21
java -XX:+AggressiveOpts  # removed in Java 17+, crashes on 21