Direct answer

Stop the server and make a full copy before repair attempts. The safest recovery is restoring a known-good backup. If no backup exists, use the crash/latest log to identify whether the failure is a player file, a specific chunk/region, a datapack/registry problem, or a plugin/mod issue before deleting any world data.

Rule out a software crash first

A plugin/mod stack trace can look like a “world crash” because it is triggered while loading a chunk or entity. Test the world on a copy with the relevant software corrected/removed according to its documentation before destroying region data.

Check storage health and free space

World saves need working disk space and reliable I/O. A full disk, abrupt power loss, host crash, or failed storage can damage writes. Free space and underlying disk health should be addressed before you trust a repaired world with new writes.

Region-file deletion is destructive

Do not delete random .mca files.

A region file contains many chunks. Removing one causes all chunks in that region to be lost/regenerated, including player builds. Only consider region-level replacement after identifying the exact damaged region and preserving the original file.

Recovery order

  1. Restore a pre-failure backup if available.
  2. If only one player is broken, isolate player data instead of the whole world.
  3. If the log identifies a datapack/registry problem, fix that first.
  4. If one chunk/region is proven corrupt and no backup exists, repair or replace only that data on a copy, then verify the result before production use.
Restore with a controlled comparison
  1. Stop the server before copying world files. Preserve the entire world, including dimension and entity/POI data, and record the first error and affected region coordinates.
  2. Try the latest known-good backup on a different server directory with the same version, mods and datapacks. If it loads, compare what changed between the snapshots.
  3. Avoid bulk deletion of .mca files or repeated start/stop cycles on the only copy. Escalate to region/NBT repair only after checking backups and compatible tools.

Official references

These references were checked while preparing this fix. Version-specific behavior can change, so use the project documentation for your exact release when a detail differs.

Was this article helpful?
Help us improve this article

Tell us what is missing, unclear, outdated, or incorrect. You can also suggest something we should add.

What could we improve?
Do not include passwords, server IPs, or other private information.0 / 1000