Ticking entity: what the crash means, and how to fix it

1What the crash says

Twenty times a second, the server runs every entity in the parts of the world that are active, around players and at spawn: it moves mobs and minecarts, and counts down dropped items. When the code that runs one of them throws an error, Minecraft stops the whole server and writes a crash report in the server's crash-reports folder. The line under Description is the error itself.

Ticking block entity is the same crash for a block with data and code of its own: a hopper, a furnace, a spawner, or a machine or pipe from a mod.

2Which entity, and where

Further down, the report names what failed and where it is:

• Entity Type says what it is. The part before the colon is what adds it: minecraft for the game, or a mod's ID.
• Exact location is where it is, as x, y and z.
• in 0,-4,0 is its chunk, by the first and last numbers: x 0 and z 0. A chunk is 16 by 16 blocks, so that's the block x and z divided by 16, rounded down.
• Region: (0,0; …) names the file it's saved in: entities/r.0.0.mca, in the world folder. A region is 32 by 32 chunks.
• The lines under the error name the code that failed. In a real crash, look for a mod's or a plugin's name there; ours named the game's own minecart code, where we'd planted the error.

For a block, the section is called Block entity being ticked, and it gives the block's Name, like minecraft:hopper, and its Block location. A block's data is saved with the land around it, in region/r.0.0.mca.

3What each server did

On 29 Sep we made one minecart throw an error each time the server ran it, and one hopper the same, as a broken mod would. A small Java agent changed the game's code as it loaded, on Minecraft 1.21.1's own server, Fabric, Paper and NeoForge:

Each server that stopped had saved the world first, with the entity still in it, so the crash came back:

• At spawn, Minecraft crashed in the same second it said Done, at every start.
• 2,000 blocks away, it ran normally with a player at spawn. When the player teleported next to the minecart, it crashed within a second, and the player saw Server closed.
• With nobody on, it stopped running entities shortly after it started. A broken minecart we added 10 seconds after a start crashed it at once; one added after 18 seconds did nothing until the next start.

4How to fix it

Take the first fix that fits your server. Before you change the world's files, stop the server and copy the world folder.

• Restore a backup from before the crash. The world comes back as it was then, without the broken entity; what changed since is lost.
• Paper: it's already gone. Paper removed our minecart by itself and logged its world and position. If it happens again, the lines under that log line name the code.
• Forge or NeoForge: let it remove the entity, once. Set removeErroringEntities, or removeErroringBlockEntities for a block, to true and start the server. NeoForge 21.1 keeps them in config/neoforge-server.toml, Forge 1.20.1 in world/serverconfig/forge-server.toml; both start off. Then turn it off again: the file itself warns BE WARNED THIS COULD SCREW UP EVERYTHING, and while it's on, whatever fails next is deleted instead of stopping the server.
• Fabric, Forge or NeoForge: Neruina. This mod, downloaded 9.7 million times on Modrinth, stops the entity that fails and keeps the server running. /neruina kill @e[type=minecart,limit=1] then removed ours. It doesn't remove anything by itself: once we took Neruina off, the hopper it had stopped crashed the server again.
• Otherwise, delete the entities file. With the server stopped, delete the file the Region line names. Ours then started: the blocks and a chest's diamonds stayed, and every entity in that 512 by 512 block area was gone, a cow we'd named too.
• Or only that chunk, with MCA Selector. This free world editor deleted the chunk from the report, which came back as new land: what was built in it went, and a cow one chunk over stayed. For a block, it replaced only the hopper with air, and the chest beside it kept its diamonds.
• Then the cause. If the lines under the error name a mod's code, the next entity of that kind can fail the same way: update the mod, or remove it.

# In the server's folder, with the server stopped
cp -r world world-backup
rm world/entities/r.0.0.mca   # the numbers from the Region line
# Or only chunk 0,0, with MCA Selector 2.9
java -jar mcaselector-2.9.jar --mode delete --world world --query "xPos = 0 AND zPos = 0"

# In the server's folder, with the server stopped
cp -r world world-backup
# Every hopper in chunk 0,0 becomes air
java -jar mcaselector-2.9.jar --mode change --world world --query "xPos = 0 AND zPos = 0" --fields 'ReplaceBlocks = "minecraft:hopper=minecraft:air"'

5On a Playkeeper server

• It starts a crashed server again by itself, at once after the first crash and 30 seconds after the second. After 3 crashes in 15 minutes it stops and says so, so an entity at spawn can't keep it crashing.
• The Overview shows the last errors, with the crash report's Description line. Playkeeper doesn't name this cause itself, so the fix it offers is starting again; use the fixes above instead.
• The Files tab deletes world/entities/r.0.0.mca once the server has stopped. While it runs, the world can't change.
• The World tab restores a backup, and keeps the world it replaces as a rollback archive.