Reading Minecraft Server Logs and Crash Reports

A Minecraft server that will not start has almost always already told you why — in one line, buried in a few hundred lines of normal startup output. This guide covers where that output lives, how to read a stack trace well enough to name the mod or plugin at fault, how to find a bad one by bisection when the log does not name it, and the specific errors you are most likely to hit.

It applies to any Java Edition server on your Minecraft server — vanilla, Paper, Purpur, Fabric, Forge or NeoForge; Minecraft Server Software Compared covers the differences between them. For the separate problem of a server that starts fine but nobody can join, see Minecraft Players Can't Connect.

Where the Output Lives

There are four places to look, and they hold different things.

Two things about that table matter more than the rest:

• The Console tab only shows the current session. Restart the server and the evidence you needed scrolls away. logs/latest.log does not — open it in the File Manager, or pull it down over SFTP (How to upload files via SFTP covers the connection).
• Rotated logs are gzipped. If the failure was two restarts ago, the log you want is a .log.gz, not latest.log. Download it and open it locally.

A crash report is written only for a genuine crash, not for every error. When one is produced, the console says so:

[Server thread/ERROR]: This crash report has been saved to: ./crash-reports/crash-2026-08-19_04.11.32-server.txt

The file itself opens like this:

---- Minecraft Crash Report ----
// I let you down. Sorry :(

Time: 2026-08-19 04:11:32
Description: Exception in server tick loop

java.lang.NullPointerException: Cannot invoke "..." because "..." is null
	at ...

The comment line after the header is one of a rotating set of jokes Mojang ships with the game; ignore it. Description: is the field that matters — it names the phase the server died in. Exception in server tick loop, Exception ticking world and Watching Server (the watchdog) are the three you will see most.

How to Read a Stack Trace

A stack trace is one exception line followed by a stack of at lines. Read it in this order.

1. The top line: what actually went wrong.

java.lang.OutOfMemoryError: Java heap space

Everything before the colon is the exception type. Everything after it is the message. Together they are what you search for, and what you quote in a ticket.

2. The at lines: where it went wrong. Read them top-down and stop at the first line that is not net.minecraft, org.bukkit or java.base — that package name is usually the culprit.

java.lang.NullPointerException: Cannot read field "level" because "entity" is null
	at com.example.grieflog.EntityListener.onDeath(EntityListener.java:88)   <-- this one
	at org.bukkit.plugin.java.JavaPluginLoader$1.execute(JavaPluginLoader.java:306)
	at net.minecraft.server.level.ServerLevel.tick(ServerLevel.java:412)

com.example.grieflog is a third-party package, so GriefLog is the plugin to update or remove. PaperMC's own basic troubleshooting guide puts it plainly: in most cases the plugin whose name appears in the stack trace is the one causing the problem.

3. Caused by: — read to the bottom. When a trace has one or more Caused by: blocks, the last one is the real fault. The blocks above it are just wrappers.

java.lang.RuntimeException: Failed to load plugin
	at ...
Caused by: java.lang.NoClassDefFoundError: net/milkbowl/vault/economy/Economy

NoClassDefFoundError on a class from another project means a missing dependency — here, Vault.

Read from the bottom of the log, not the top

Startup output is long and mostly uninteresting. Open latest.log, jump to the end, and scroll up until you find the first ERROR or Caused by:. The last thing the server said before it stopped is nearly always the reason it stopped.

Startup Errors, in the Order You Hit Them

You need to agree to the EULA

[ServerMain/WARN]: Failed to load eula.txt
[ServerMain/INFO]: You need to agree to the EULA in order to run the server. Go to eula.txt for more info.

The server generates eula.txt on its very first run and exits:

#By changing the setting below to TRUE you are indicating your agreement to our EULA (https://aka.ms/MinecraftEULA).
#[Generation time]
eula=false

Change eula=false to eula=true in the File Manager and start again. The Failed to load eula.txt warning on a brand-new server is normal — the file did not exist yet.

FAILED TO BIND TO PORT!

[Server thread/WARN]: **** FAILED TO BIND TO PORT!
[Server thread/WARN]: The exception was: java.net.BindException: Address already in use
[Server thread/WARN]: Perhaps a server is already running on that port?

The server could not take the port it was told to listen on. On managed hosting there are two realistic causes:

• The old process is still running. A stop that did not complete leaves the port held. Use Restart, or stop and wait for the console to confirm it exited before starting again.
• server-ip or server-port was edited by hand. Both are managed by the panel. server-ip should stay empty, which means "listen on every interface"; the Minecraft Wiki names an address in server-ip that the machine does not own as the single most common cause of this error. server-port must stay on your panel allocation — the panel syncs it on boot, so a hand-edited value only breaks the next start.

UnsupportedClassVersionError — Wrong Java Version

This one has become common, because Minecraft moved its Java requirement in 2026 and a lot of servers were carried across without their runtime.

Exception in thread "ServerMain" java.lang.UnsupportedClassVersionError: net/minecraft/bundler/Main has been compiled by a more recent version of the Java Runtime (class file version 69.0), this version of the Java Runtime only recognizes class file versions up to 65.0

Read the two numbers. The first is what the jar needs; the second is what your runtime provides. Both are class file versions, not Java versions, and the conversion is simply Java version + 44:

So 69.0 needed, 65.0 available means a 26.x server jar on a Java 21 runtime. Mojang publishes the requirement per version in its own version manifest — the 26.2 entry carries "java_version": 25 — and PaperMC states the same minimum in its getting started guide. The Minecraft Wiki's server setup tutorial lists the same thresholds, and Updating Your Minecraft Server covers moving a live server across a Java boundary safely.

The fix is to raise the Java version on the Startup tab, not to downgrade Minecraft. Note that the reverse does not happen: newer Java runs older Minecraft happily, so there is no such thing as being "too new" for this error.

Plugins and mods hit this too

The class name at the front of the message tells you who compiled too new. net/minecraft/bundler/Main on a vanilla jar, or org/bukkit/craftbukkit/Main on Paper, is the server itself. Anything else — com/example/someplugin/Main — is a plugin or mod built for a newer Java than your server runs, and the fix is a build of that plugin for your version rather than a change on the Startup tab.

OutOfMemoryError

[Server thread/ERROR]: Encountered an unexpected exception
java.lang.OutOfMemoryError: Java heap space

The JVM ran out of heap. Three variants mean slightly different things:

For the first two, raise the Xmx heap value on the Startup tab. For the third, lower it: the heap plus the JVM's own off-heap overhead has to fit inside your plan's memory, so an Xmx set to the full plan size can starve everything else. Minecraft Server RAM and JVM Flags covers heap sizing properly.

A crash is not the only symptom. A server running slowly out of memory lags heavily and logs repeated GC pauses long before it finally dies, so an OOM crash after hours of complaints is normal.

Failed to load datapacks

[Server thread/ERROR]: Failed to load datapacks, can't proceed with server load. You can either fix your datapacks or reset to vanilla with --safeMode

A datapack in world/datapacks/ is malformed, or was built for a different pack format than your Minecraft version uses. Remove the pack you added most recently and restart.

Failed to load world data and Chunk Corruption

The blunt one, which stops the boot outright:

[Server thread/ERROR]: Failed to load world data. World files may be corrupted. Shutting down.

The subtler ones appear during play, chunk by chunk, and are worth catching early:

Genuine corruption is usually caused by a hard kill during a save, or by running out of disk. Restore from a backup — How to create a backup — rather than trying to repair region files by hand. sync-chunk-writes defaults to true precisely to prevent this class of damage, and is not something to switch off casually; Minecraft World Management covers world files in depth.

One special case is worth its own line, because it is not corruption at all:

java.lang.RuntimeException: Server attempted to load chunk saved with newer version of minecraft! 3955 > 3465

That is a downgrade. The world was opened by a newer Minecraft version than the server is now running, and the two numbers are world data versions, not Minecraft versions — 26.2 writes 4903. Minecraft does not support downgrading a world. Put the server back on the version that wrote it, or restore a backup taken before the upgrade; Updating Your Minecraft Server covers doing this in the right order.

Mixin Failures

Mixin is the bytecode-patching library that Fabric mods and many NeoForge mods use to modify the game. When two mods patch the same method incompatibly, or a mod was built for a different Minecraft version, Mixin refuses:

Mixin [examplemod.mixins.json:MinecraftServerMixin] from phase [DEFAULT] in config [examplemod.mixins.json] FAILED during APPLY
org.spongepowered.asm.mixin.injection.throwables.InvalidInjectionException: Critical injection failure: @Inject annotation on init could not find any targets matching 'tickServer' in net.minecraft.server.MinecraftServer

The pieces to read:

• The config name — examplemod.mixins.json — names the mod. That is your culprit, every time. On NeoForge the loader logs Failed to apply mixin. Mixin Class: <class>, ModID: <modid> alongside it, naming the mod ID directly.
• could not find any targets matching means the mod is patching a method that does not exist in this Minecraft version. It is built for a different version. Update it, or move the server back.
• failed injection check in the same family means the mod found fewer injection points than it required — usually another mod got there first.
• Mixin config X requires mixin subsystem version Y but Z was found means your loader is older than the mod expects. Update Fabric Loader or NeoForge.

Related failure types logged the same way are InvalidMixinException, MixinApplyError and MixinPrepareError; the diagnosis is identical — find the config name, find the mod.

Missing Dependencies

Fabric refuses to launch on an unsatisfied dependency and tells you exactly what to install:

Incompatible mods found!
	- mod 'Example Mod' (examplemod) 3.2.0 requires any version of fabric-api, which is missing!
A potential solution has been determined, this may resolve your problem:
	- Install Fabric API, any version.

requires ... which is missing! means install it. requires ... but only the wrong version is present means update it. Fabric prints the suggested fix under A potential solution has been determined — follow it literally.

The plugin equivalent on Paper and Spigot is shorter but means the same thing:

[Server thread/ERROR]: Could not load 'ExamplePlugin.jar' in folder 'plugins'
org.bukkit.plugin.UnknownDependencyException: Unknown/missing dependency plugins: [Vault]. Please download and install these plugins to run 'ExamplePlugin'.

The message names both the missing plugin and the one that wanted it. The common shared dependencies are Vault, ProtocolLib and PlaceholderAPI. See How to install plugins for Minecraft Java Edition.

Paper has one more of its own worth recognising, because the message is unusually clear about what to do:

[SimpleProviderStorage] Circular plugin loading detected!
[SimpleProviderStorage] Circular load order: PluginA -> PluginB -> PluginC -> PluginA

Two or more plugins declare each other as dependencies. Update the plugins named in that loop.

Crashes After the Server Is Running

The Watchdog

If a single tick takes longer than max-tick-time — 60,000 ms by default — the server assumes it has hung and kills itself:

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

Paper's watchdog trips much earlier and prints a thread dump instead of a crash, with a banner designed to stop people filing bug reports:

[Paper Watchdog Thread/ERROR]: --- DO NOT REPORT THIS TO PAPER - THIS IS NOT A BUG OR A CRASH ---
[Paper Watchdog Thread/ERROR]: The server has not responded for 10 seconds! Creating thread dump
[Paper Watchdog Thread/ERROR]: Server thread dump (Look for plugins here before reporting to Paper!):

A watchdog event is a lag diagnosis, not a crash diagnosis. Read the thread dump that follows it: the frames show where the main thread was stuck when it stopped responding, and that is your actual problem — a plugin doing something slow, a runaway machine in an automation modpack, or a chunk-generation storm. Diagnosing Minecraft Server Lag takes it from there.

Do not "fix" the watchdog by disabling it

Setting max-tick-time=-1 turns the watchdog off entirely. It stops the restarts; it does not stop the freezes, and it removes the thread dump that tells you what caused them. Treat it as a last resort while you investigate, not as a solution.

Everything Else

Exception in server tick loop, Exception ticking world and Uncaught exception in server thread all produce a crash report. Read it exactly as above: Description: for the phase, then the first non-vanilla package in the trace. A watchdog kill writes one too, with Description: Watching Server.

Bisection: Finding a Bad Mod or Plugin When the Log Does Not Name One

Sometimes nothing is named — the server dies without a useful trace, or the trace points only at vanilla code. Binary search finds the culprit in a handful of restarts instead of dozens. This is the method PaperMC recommends, and it works identically for mods/.

First, confirm it is a mod or plugin at all. Rename the whole folder — plugins to plugins-disabled, or mods to mods-disabled — and start. If the server comes up, you have your answer. If it does not, stop here: the problem is the server itself, its Java version, or the world.

Then halve it. With 32 plugins, five restarts is enough:

To disable a jar without deleting it, rename Example.jar to Example.jar-disabled, or move it to a temporary folder outside plugins/. The server ignores anything that is not a .jar.

Three things that make bisection go wrong:

• Keep dependency chains together. Moving Vault out while leaving three plugins that need it produces a new, unrelated failure and wastes a round.
• Change one thing per restart. If you also edit a config, you no longer know which change mattered.
• Take a backup first. A plugin that stores data may not like being removed and re-added.

Before you start, do the cheap check: make sure every plugin and mod is on a build made for your Minecraft version. Immediately after a version bump, "everything broke" is almost always a stack of plugins that have not been updated yet, not one bad actor.

What to Capture Before Asking for Help

Send these five and you skip a round trip. Send a screenshot of a screenshot and you will not.

1. logs/latest.log from the failed run, as a file — not a screenshot, and not the last twenty lines. Upload it to a paste service and send the link if it is large.
2. The crash report, if crash-reports/ gained a file.
3. Your server type and version — vanilla, Paper, Purpur, Fabric, Forge or NeoForge, plus the Minecraft version and the Java version on the Startup tab.
4. What changed. A mod added, a plugin updated, a version bumped, a config edited. "Nothing" is almost never true; a plugin auto-updater counts.
5. When it last worked, and whether it fails on every start or only sometimes.

Do not restart repeatedly before collecting the log

Every restart rotates latest.log into a .gz and starts a fresh one. Three hopeful restarts can push the useful log two files back. Copy it out first, then experiment.

What to Read Next

• Minecraft Players Can't Connect — for a server that starts fine but rejects joins
• Diagnosing Minecraft Server Lag — when the watchdog is firing but nothing crashes
• Minecraft Server RAM and JVM Flags — heap sizing, and why more is not always better
• Updating Your Minecraft Server — Java requirements, world data versions and safe upgrades
• Minecraft server.properties: The Complete Reference — every key named on this page
• Minecraft World Management — world files, backups of the world folder, and resets
• How to create a backup — take one before every bisection round
• How to upload files via SFTP — for pulling logs and crash reports off the server

Made with 💜 by GameServerKings