Game freezes/crashes from error: "java.lang.IllegalStateException: Requested chunk unavailable during world generation" while data pack is installed.

I have the exact same issue and also make use of a lake feature in my datapack.
Hope there will come a fix to this!

I can confirm using a lake feature is the cause of crashing/freezing the game
When I removed the feature from my biome there was no issue
Removing in_square from my placed feature did not work and kept crashing/freezing my game.

Also this bug only happens when you refer to your own configured feature in your placed feature file! When referring to for example minecraft:lake_lava there is no problem
It happens for all my updated datapacks!

I haven't been looking into this too much since I was focused on other things, but the lake feature is an interesting discovery. I don't think some of my custom biomes that were giving me issues have any lake features besides the vanilla lava ones. Maybe there's a couple of broken types?

Yeah i haven't diagnosed the exact issue yet but I am having it too and it is affecting things that are either vanilla lava pools or are something else entirely.

How exactly do you find which feature (or even which chunk) is crashing you? Since it's a server crash there is no crash report anywhere for me

 get the same error when I empty the "lava" fluid tag and put lava in the "water" tag, something about lighting was mentioned in the console

Can still confirm in 1.21.3 with this simple chunk_crash_debug.zip pack. All it does is put a stone column in plains biomes which is fine, and a custom lake in the desert which causes the crash, even though there's nothing fancy going on with it. You just have to go near a desert.

I know it had an assignee but now it doesn't. Really hopping this can be looked at again though. I've spent months working on a pack and I've randomly been motivated to work on updating and maybe publishing it somewhere but am not willing to do that if it just seemingly randomly crashes. 

Can confirm in 1.21.4 still, unfortunately. Updated chunk_crash_debug.zip to 1.21.4Test.zip and the issue is still present. What's weird at least to me is the seed I tested 7575026638094145403 spawned me in plains. As expected they're fine even with the stone column test feature I added. However, running /locate biome desert and flying to said desert, I could actually load a few chunks of it even with the custom lake feature that seems to break the game. It still crashed after generating more chunks, but because some successfully generated and contained the lake feature this suggests something deeper than just some features being broken on a surface level.

I miss working on my big datapack 😞

Not sure if it's helpful but I've uploaded a log. I don't get much out of it but maybe the devs can. Again, I was able to see the desert start to load in with the custom, bugged lake feature for a few chunks and then this happened. Could it be tied to simulation distance somehow and not the render distance loading the chunks? I wouldn't think so but this issue is very odd to me. I also attempted to tweak the lake feature some by converting 

{
  "type": "minecraft:lake",
  "config": {
    "fluid": {
      "type": "minecraft:simple_state_provider",
      "state": {
        "Name": "minecraft:water"
      }
    },
    "barrier": {
      "type": "minecraft:simple_state_provider",
      "state": {
        "Name": "minecraft:stone"
      }
    }
  }
}

into

{
  "type": "minecraft:lake",
  "config": {
    "fluid": {
      "type": "minecraft:simple_state_provider",
      "state": {
        "Name": "minecraft:water",
        "Properties": {
          "level": "0"
        }
      }
    },
    "barrier": {
      "type": "minecraft:simple_state_provider",
      "state": {
        "Name": "minecraft:stone"
      }
    }
  }
}

to see if the addition of the "properties" fixed anything but it didn't seem to. I'll say an odd thing did occur though in the old version without the "properties" stuff where I teleported to the desert and the game refused to load any chunks but it didn't do the weird crash thing. That's the only time I've ever seen it do that. Before and after that one instance it just does this weird crash thing.

I also find it interesting that somehow the vanilla game must just be avoiding this issue as it seems to only be datapacks that suffer despite vanilla using the lake feature for lava lakes (and I don't believe this is likely the only bugged feature, just one I've confirmed). I'd imagine this would be impacting more people because worldgen mods and datapacks seem like a pretty common thing (perhaps mods aren't impacted since they work a bit differently? idk but even if it's just datapacks the point still stands) yet this only has 14 votes and 12 watchers so are those packs just by chance not using the broken features and/or trying to update to modern game versions to realize the issue even exists?

edit: can confirm that it seems removing the "in square" function form the placed_feature "feature2_placement" in that testing pack I uploaded does resolve the issue, but again, vanilla lava lakes use "in square" in their placement to my knowledge so what's the difference there and why do I sometimes get a few chunks to generate with the custom lake before the crash occurs?

MC-273228 says it's caused by water trying to be frozen with top_layer_freeze is causing the crash but that doesn't account for the cases of lava lakes also crashing, despite the fact that vanilla lava lakes generate just fine without crashing

My testing has been with water lakes, not lava so perhaps that is it. I do feel like in my big datapack not all the custom biomes that crashed had water features though.

I had this crash too and my pack did not have water lakes. it had custom lava lakes though

According to MC-273228 which seems to be a duplicate of this, it’s water lakes and something to do with water lakes and checking if water needs to freeze. It seems in migration some comments are missing here so I don’t remember if that was already discussed in this post. I’m also not certain that’s the only trigger though as I do believe I was able to get around this crash by removing the in-square function from the chunk-crash-debug.zip and I’m pretty sure a custom nether biome of mine also triggers it despite not having water.

I also somehow managed to trigger a different error, but only once so far but I think from this same issue. I think it’s a log for being disconnected from the server (also puts me in the same softlock mentioned above)

It’s not just lake features, nor is it in_square. I have a big pack from years ago that I’ve been bringing up to 1.21.5 and this issue remains my last obstacle (I think). I created a single biome world of one of my custom biomes and it was fine for a few minutes, then this issue occurred again. It has no custom lakes and I even removed the in_square function to try to see if that fixed it but then the world didn’t even load. Thought it might’ve been a custom delta feature but I got rid of it and it still happened. Maybe it’s the disk or patch features as I also have those.

The issue feels like it’s not a particular feature or a particular function and just custom features in general, but I also know that some custom biomes in another datapack of mine are just fine (though I guess those are just custom trees that are still similar to vanilla) + the debug zip can load the plains biome with just block columns fine. I’ve also been able to get the troublesome biomes to load for a few moments before so it’s some condition that randomly trips, but I don’t recall seeing anything in changelogs that would suggest what could now be causing this bug.

I know this datapack stuff is experimental and they have the warning that it may not always be supported/could break, but please Mojang. This doesn’t feel like “not supported” it feels like a bug that even vanilla features could randomly trip and we’re just getting lucky that they aren’t and it’s really a massive roadblock to making worldgen datapacks.

somehow triggered another disconnect but it doesn’t mean much to me.

Anyway, I’m really trying to narrow it down and I think deltas are fine, but the custom patch and disk features are not. I’m going to try removing in_square from their placement and see