[26.2-31] Block ESP causes shutdown watchdog crash — reproduced with only Meteor installed

[sludgepudge]

Describe the bug

Whilst testing Meteor out as a server owner to better understand it's features, I discovered that it's causing the shutdown to hang when using Block ESP.

Activating Block ESP and subsequently quitting Minecraft can leave a background worker alive, preventing the Java process from exiting. After approximately 15 seconds, Minecraft generates:

java.lang.Error: Watchdog (Client shutdown from post-main)

Reproduced on 4 October 2026 in a vanilla Minecraft 26.2 Fabric instance with Meteor 26.2-32 as the only installed mod, alongside its bundled dependencies. Using Prism Launcher 11.1.1.

Additional evidence from an earlier session on the same Meteor build but a different instance of Minecraft:
A Java Flight Recorder capture identifies Block ESP starting the exact worker that subsequently survives shutdown. The recording and crash report match Java thread ID 6207.

Relevant creation stack:

ThreadPoolExecutor.addWorker(...)
ThreadPoolExecutor.execute(...)
AbstractExecutorService.submit(...)
Executors$DelegatedExecutorService.submit(...)
BlockESP.searchChunk(ChunkAccess):162
BlockESP.onActivate():101
Module.toggle():99
Modules.onAction(...):297
Modules.onKey(...):270

The JFR was captured in my normal Fabulously Optimized instance; the subsequent minimal-instance reproduction confirms that the other mods in that pack are not required for the shutdown failure.

Expected behaviour:
Block ESP’s worker should not prevent Minecraft’s Java process from exiting.

Versions:

• Minecraft: 26.2
• Meteor Client: 26.2-32
• Fabric Loader: 0.19.5
• Java: Microsoft OpenJDK 25.0.1, 64-bit
• OS: Windows 11
• Launcher: Prism Launcher 11.1.1

Evidence available:

• Minimal-instance launcher log, including crash report dated 2026-10-04 19:29:42.
• Focused JFR excerpt matching the Block ESP worker to the earlier shutdown crash.

The same Block ESP thread-ownership evidence was also captured previously on all versions between Meteor 26.2-25 and 26.2-31 inclusively.

Steps to reproduce

1. Launch Minecraft 26.2 with Fabric Loader 0.19.5 and Meteor 26.2-32.
2. Enter a world.
3. Activate Block ESP.
4. Quit Minecraft normally.
5. Wait approximately 15 seconds for the shutdown watchdog crash.

Meteor Version

26.2-32

Minecraft Version

26.2

Operating System

Windows

Before submitting a bug report

• This bug wasn't already reported (I have searched bug reports on GitHub).

• This is a valid bug (I am able to reproduce this on the latest dev build).

[crosby-moe]

interesting, although the threads can simply be marked as daemon i do wonder what is causing the thread to hang for over 15 seconds,

[sludgepudge]

The worker appears to be idle rather than hanging on an operation. At shutdown its stack is ThreadPoolExecutor.getTask() -> LinkedBlockingQueue.take(), and the JFR creation event ties that exact thread ID to BlockESP.searchChunk().

My understanding is that the 15 seconds is the client shutdown watchdog’s timeout: the non-daemon executor worker remains alive waiting for another task, so the JVM can't finish exiting. Marking it daemon should prevent that; explicitly shutting down the executor on client exit would also address it.

[Frko5000]

i think have a fix for this i didnt test it yet tho

[crosby-moe]

oh dear, i assumed it would have just done that itself

[Frko5000]

Buh i didnt have time