Corrupted chunks and region files: reading the errors
Region file truncated, chunk too big, failed to save chunk: what each one means, what causes them, and the order to try fixes in.
The short version
Chunk corruption shows up as oversized chunk, truncated region file, or failed to save chunk errors, and usually follows an unclean shutdown, a full disk, or a mod removed mid-game. Restoring a backup is the reliable fix. Region editors can excise the damaged chunk if the backup is too old to use.
World corruption is rarer than the internet suggests and more alarming when it happens. The errors are specific enough to act on.
#The errors and what they mean
Region file ... truncated or ... is truncated: expected X but read Y. A region file is shorter than its own header says it should be. Classic result of a write interrupted part-way: unclean shutdown, or the disk filled mid-save.
Chunk too big or oversized chunk. A chunk's serialised data exceeds what the format expects. Usually enormous quantities of entities or block entities in one chunk, which is as often a lag symptom as a corruption one.
Failed to save chunk. The write did not complete. Check disk space first; this is no space left on device wearing a different hat surprisingly often.
Exception generating new chunk or a crash naming ChunkHolder or ProtoChunk. Something went wrong loading or generating a specific chunk. Frequently a mod expecting data that is not there.
#What causes it
Unclean shutdowns. A server killed mid-write, whether by a crash, a watchdog kill or the process being terminated, can leave a region file half-written. This is the most common cause by a wide margin.
A full disk. Writes fail part-way and leave exactly the damage described above.
Removing a mod that added blocks. The world still contains those blocks and the server has no definition to load them with. Behaviour ranges from silently replacing them with air to refusing to load the chunk.
Changing Minecraft version in the wrong direction. Chunks written by a newer version can be unreadable to an older one. See changing the Minecraft version.
#Fixing it, in order
1. Restore a backup. The reliable answer. A backup from before the corruption gives you a known-good world, and everything else on this list is a compromise. If you are not sure when it started, restore to before the last unclean shutdown.
2. Identify the chunk. If a backup is too old, get precise. The crash usually names chunk coordinates or a region filename such as r.-1.4.mca. A crash that only happens when someone visits one area is telling you exactly where to look.
3. Delete the damaged region. MCA Selector opens the world, shows the region grid, and deletes selected regions. The server regenerates them fresh on next load.
This works, and it means the area regenerates as new terrain. Anything built there is gone, and the boundary between old and new generation will be visible. Take a backup before doing it, because it is itself destructive.
4. Excise the entities instead. If the error is an oversized chunk, the chunk is usually fine and merely overloaded. Clearing entities from it, or letting it load once with a much reduced view distance, can settle it without deleting anything.
#Preventing it
Stop the server properly. Use the dashboard's stop and restart controls rather than killing the process. A clean stop flushes chunks to disk; anything else is a gamble against the timing of the next write.
Watch disk space. See above. A disk that fills during a save is a reliable way to damage a world.
Take a backup before removing mods. Particularly anything that places blocks or adds dimensions. Removing a world-generation mod from a world it has generated is not a reversible operation.
Set max-tick-time appropriately on heavy packs. A watchdog kill during chunk generation is an unclean shutdown by another name.
#If it keeps coming back
Corruption that recurs after a clean restore is not corruption; it is something actively causing it. Look for a mod that crashes repeatedly at the same point, a disk that is filling and being cleared, or a restart schedule that is not stopping the server cleanly.
Fixing the world without fixing the cause buys you a few days.
#Frequently asked questions
What causes corrupted chunks in Minecraft?
Unclean shutdowns during a chunk write, a full disk interrupting saves, a mod that added blocks being removed while those blocks existed in the world, and hardware faults. The first two are by far the most common, and both are avoidable.
How do I fix a corrupted Minecraft world?
Restore a backup from before the corruption appeared: it is the only fix that reliably returns a clean world. If the backup is too old, tools such as MCA Selector can delete the damaged region so the server regenerates it, which loses whatever was built there.
My server crashes when a player enters one area. Is that corruption?
Very likely. A crash tied to a location rather than an action points at a specific chunk the server cannot load. The crash usually names chunk coordinates, which you can convert to a region file and deal with directly.
Written and maintained by the Hostd engineering team. Last updated 2026-08-05. Notice a mistake? Tell us.