Chunk loading stops and game crashes

@unknown, is there anything in the launcher_log.txt ?

Repeated the error and checked the log, there's lots of lines like "game/net.minecraft.server.MinecraftServer (Server thread) warn Can't keep up! Did the system time change, or is the server overloaded? Running 5754ms behind, skipping 115 tick(s)" before the game completely froze.

^{(Attached the log file)}

@unknown, unrelated, but that resource pack is kinda broken, it throws a lot of errors in the launcher log:

[0608/110912:INFO:GameCallbacks.cpp(178)] game/uq (Client thread) info Reloading ResourceManager: Default, Better than Default 2.4_WIP17.zip
...
[0608/110915:INFO:GameCallbacks.cpp(178)] game/dhk (Client thread) warn In parent chain: minecraft:block/polished_granite2 -> minecraft:block/granite_smooth; unable to load model: 'minecraft:block/granite_smooth'
java.io.FileNotFoundException: minecraft:models/block/granite_smooth.json
	at ul.a(SourceFile:62)
	at uq.a(SourceFile:58)
	at dhk.c(SourceFile:398)
	at dhk.m(SourceFile:673)
	at dhk.l(SourceFile:647)
	at dhk.a(SourceFile:221)
	at dhl.a(SourceFile:23)
	at uq.a(SourceFile:117)
	at ciq.am(SourceFile:492)
	at ciq.a(SourceFile:364)
	at net.minecraft.client.main.Main.main(SourceFile:144)

The issue itself is most probably an out of memory thing, maybe caused by MC-129492 or similar

yeah, noticed the model file errors, already fixed them.
I tried to find a way to reproduce this bug, although it seems to happen quite randomly, not specific to certain seeds, coordinates or memory usage. In one case, I'm flying 6-7 minutes along oceans until the chunk generation freezes, in another case it happend only after 200 meters from the spawn point, with only a few new chunks generated. It always happens over oceans, that's the only really specific thing I noticed. Hard to pinpoint what triggers this ...

Tried to find this issue in 1.13-pre2, went well for several minutes, then chunk loading froze over a forest biome. First time it happend in a different biome, which would make it being a variant of MC-129492 even more plausible.

Affects 1.13-pre2

Still happens in 1.13-pre3. Just had this happen after only 5 seconds of flying in a fresh world. Sometimes it takes several minutes. Seems to be quite random.

I'm using the latest snapshot 1.13-pre3 and while flying in creative mode (SP) chunk loading/rendering suddenly completely stopped. I could still move around and place/break blocks but it seemed like the game time was frozen (no block updates). I was using a night vision potion while flying around and the potion timer was frozen at 0:00. My screen was flickering (most likely from the potion running out). When I tried to close the game my computer stopped responding (black screen) and I had to use task manager to get back to desktop.

Does anyone still have this issue or MC-129492 in the latest pre-release? Doesn't seem to happen on my side in 1.13-pre5 ...
Maybe it was fixed too along with the jre 8.0_51 update?

Reproduced in 1.13-pre5.

How to reproduce this issue with some reliability:

1. Start Minecraft with 2048M as the maximum RAM allocation.

2. Create a new Default world in creative mode.

3. Set view distance to 12.

4. Build the chunkloader machine described here: https://www.reddit.com/r/Minecraft/comments/8qfpk1/a_machine_for_pregenerating_chunks_in_vanilla/

5. For the chunkloader, start generating chunks at x=-4096, z=-4096 (command block 5), with 33 items in the two hoppers B and C. (Also swap the commands in command blocks 5 and 6, and add a 20-second delay with a line of repeaters before the setblock command that starts the machine).

6. For command blocks 1 and 2, have a relative jump of 256 blocks instead of 192.

7. Start the machine by pressing the button.

8. At some point, the symptoms described in this bug report and/or MC-129492 will occur and chunks will stop loading. This is when the server thread has crashed by running out of heap space.

When you reproduce the issue as described above, it may take a couple of hours before RAM is exhausted. For quicker reproduction, reduce the maximum memory to 1Gb.

The RAM allocation will creep up slowly as chunks are generated.

To check - if the same reproduction steps are run on a world with pre-generated 1.13 chunks, does it still run out of memory? If yes - it's a chunk loading issue. If no - it's a chunk generation issue. Also try on a world with chunks generated in 1.12.2 (to test the chunk conversion).

Attached is a chart "MC-130427.png" that shows the effect of teleporting around a world until the game exhausts RAM. The X axis is the number of teleports, the Y axis is the time taken to process the chunks. Three scenarios were examined: generating new chunks, loading existing chunks, converting chunks from a 1.12.2 world. In all tests, the world was allocated 1GB of RAM with a view distance of 13.

• New chunks - 142 teleports before crash

• Existing chunks - 163 teleports before crash

• Converting 1.12.2 - 71 teleports before test terminated

For the new chunks versus existing chunks, the behaviour was very similar. The garbage collection spikes were occurring at similar times, which suggests the resource leak has a cause independent of whether the chunk is freshly generated.

For converting 1.12.2 chunks, the behaviour was very different. RAM was exhausting much sooner.

This experiment suggests there are two resource leaks: (1) Chunk loading and unloading, (2) Converting chunks from 1.12.2. It suggests there are no resource leaks with chunk generation.