PerlandFailure catalogue

The server printed a problem the diagnosis does not recognise yet

Errorunclassified_exceptionunclassified_errorunclassified_warning

An exception, error line or warning that no rule of Perland's diagnosis knows yet; it is shown with its log excerpt so you can read what the server printed.

What the log shows

Written example4 lines
[16:20:03] [Server thread/ERROR]: [TreasureMaps] Could not restore treasure map 12 from maps.yml
java.lang.IllegalStateException: Map slot 3 is already taken by another chest
	at com.example.treasuremaps.MapRegistry.claim(MapRegistry.java:88) ~[TreasureMaps-2.4.1.jar:?]
	at com.example.treasuremaps.MapRegistry.load(MapRegistry.java:52) ~[TreasureMaps-2.4.1.jar:?]

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

What it means

The diagnosis reads a server’s last run and matches every warning and error line against a table of known failures. Each known failure has its own page in this catalogue. When no rule matches, the line is still shown, under one of three codes:

  • unclassified_exception: an error that came with a Java stack trace (an exception).
  • unclassified_error: an error or fatal line without a stack trace.
  • unclassified_warning: a warning line.

The diagnosis does not know what these mean, so it says nothing about the cause. It shows what the server printed: an excerpt of the log, the exception and the first lines of its stack trace, and, when it can tell, the plugin or mod the trace points to. An unrecognised error is not necessarily serious, and an unrecognised fatal line may well be the reason the server stopped.

Why it happens

  • The failure is rare or new. A plugin or mod that few servers run, a new version of the server software, or a message that changed wording since the rule was written.
  • It is specific to one plugin or mod. Plugins print their own errors in their own words, and most of them have no rule.
  • It is a consequence of another issue listed above it in the same run: later errors often follow from an earlier one.

How to fix

  1. Read the issue’s excerpt and exception. The first lines usually say what went wrong in plain words, and the stack trace names the plugin or mod involved (its package or its jar file).
  2. Look at the issues listed above it. If a known issue comes first, fix that one and check whether this one goes away.
  3. Search for the exception’s message together with the plugin’s or mod’s name. Its issue tracker, wiki or support channel is the best place to ask, with the excerpt.
  4. Make sure the plugin or mod is a version made for your server software and Minecraft version.

In Perland

The issue appears in the server’s Diagnosis tab with its excerpt, and members who can change the server can choose “Explain with AI” to ask Claude what it likely means and what to try (a suggestion that may be wrong, limited per day by the team’s plan). Engines that opt in send unrecognised issues to Perland’s diagnosis hub so new rules and pages get written: an observation names no team, server or account, and its log lines have IP addresses, UUIDs, URL queries and secrets replaced, though a name written in a log line itself stays as written. Those samples stay private to Perland: only aggregate counts, such as the ones on these pages, are ever published. A self-hosted Perland engine opts in with DIAGNOSIS_HUB_ENABLED=true and DIAGNOSIS_HUB_URL=https://hub.perland.online, and needs a database and PERLAND_ENCRYPTION_KEY.

References