TorteShui

I had the same issue as described before again and there seems to be a pattern. It might not be related to MC-98153. This time a whole village has been wiped out.

I have been working in the Nether and some nearby chunks with portals were loaded. I assume a piglin or it's zombified variant spawned close by and walked into the portal. The portals are in enclosed rooms with doors, but piglins can open doors and zombified ones can spawn in portals.

When (if that is what happened) they entered the portal, they caused the counterpart in the overworld to be loaded, but then probably instantly despawned because no player was around there. These regions were in villages and last loaded in 1.17.1 - the issue happened in 1.18.2. So upon loading this overworld region, the game attempts to convert the affected chunks, which you can see that the new caves are generated below zero (but only in a few chunks, not as much as you'd expect), and sometimes also a file in the entities/ folder if it didn't exist yet, otherwise it shows as changed.

My guess is that the chunks weren't loaded long enough for the conversion to be successfully completed (because the mob despawned), thus letting villagers (and potentially other mobs) disappear.

Not sure if this is a separate issue - if necessary, I can create an own bug report for this. I can also provide the world with and without the issue (I restored the affected entities/poi/region files from a backup), but the worlds are big and in principle this should happen to any 1.17.1 world with a portal in a village where a (hostile?) mob enters the portal in the nether in 1.18.2 without a player being in the overworld - probably not every time, though.

I noticed I had lost a villager in a village I haven't even been to since 1.17.1. Playing on a 1.18.2 vanilla server currently.

The region of the village was briefly loaded due to MC-98153 when going to or returning from the nether. The server created an entities file for this region and updated the poi and region files. This has happened with other regions as well, but this is the first time a villager vanished (replacing the affected files from the backup brought him back). There was nothing in the server or client logs.

Whether or not MC-98153 is the root cause is unclear, but it was at least the trigger in this case. Commented on there as well.

Please contact me if you need the world downloads (before and after it happened) - they're big, so I might have to strip them down first. Not sure if it would help, though, as I have done the same as today for quite some time and only today the issue happened, so it seems pretty sporadic.

It's not resolved yet, as of 1.18.2. I am constantly getting chunks loaded and the region/poi/entities files created/updated at the coordinates of the wrong dimension. Running on a vanilla server.

Easiest way to see the issue is just looking at assets\minecraft\textures\block\cactus_bottom.png in the client jar file.

The texture is different to the one before the texture update, that is correct. However, it is the same as the one in the New Default Betapack V2 (which is an older version of the updated textures) and still has the darker green outer ring which doesn't match the different side texture now (see the attached image). The inside is fine, just the ring should be updated in my opinion.

We've been plagued with this bug on our Realms server for a while. We had over 80 villagers in one village and the number of golems got that high as well, so I had to regularly purge them. After updating to 1.16.2, I did my last purge so far and the number of golems has been stable at 32 since then (with a villager count of 91). This is the only village that is pretty much permanently loaded (next to my base, used to be in spawn chunks before I moved spawn). I didn't check the other villages, but at least in the next village that is at least sometimes loaded, there doesn't seem to be a crazy amount of golems spawning anymore either.

So for me this bug seems to be fixed and it doesn't happen in all villages anymore. Providing the (or a) world where this still happens could thus be helpful...

(In case it matters: world was created in 1.13.2 and subsequently updated to 1.14, 1.14.4, 1.15.2 and 1.16.2, skipping the in-between versions.)

Apparently, 1.16.2 (or earlier) fixes existing villagers, too. I updated the other day and wanted to fix them manually with "data modify entity", but they all had their values changed from 16 to 48 already. Thank you 🙂

I think, MC now caches portal locations, and it seems that the cache did not get invalidated when destroying portal O1, so any attempt to go through N1 would always dump you at the old O1 location because it is still in the cache. If that's the case, it sounds like an actual bug and the ticket should be reopened.

I agree that it is for Mojang to decide and to find a correct fix.

Negative demand does something: it increases the necessary amount of trading before reaching positive demand again, so it does make sense to have it - but surely not below a certain level.

This patch would clamp it to zero, but negative values are possible - they just shouldn't go that much into the negative...

Will the fix for 20w19a and later 1.16 also fix already existing naturally generated villagers-,- i.e. is the follow range something in their NBT data or in the game code itself?

Edit: Ok, found in MC-166764 that it's in their NBT data, but my original question kind of remains: Will that value be updated when updating the world to 1.16?

It also seems to happen when you load one world, exit it and load the next world immediately after, without waiting (jokubas' comments), probably while the first world is still being saved or internal data structures not cleared yet. Also when players logged into the same world twice at the same time, at least once via singleplayer, e.g. MC-173080.

Affects 1.15.2 as well, had a glass block more than 10 blocks above the beacon and it only activated once I removed that block and placed the beacon again.

Sounds like duplicate of MC-173100 (which shouldn't be resolved as invalid).

I have the same issue in a microfarm where I press left and right mouse button continously - left to break crops, right to place new ones which are then bonemealed by dispensers. The block behind the crops gets destroyed after a while (you can see the damage accumulating over time), only when I release the left mouse button every once in a while and allow the right click to hit the block behind, it becomes undamaged. Realms world in vanilla 1.15.2.

Possibly related to MC-166980 beecause bees seem to beehave similar with beehives: bees can stay attached to a beehive they can't use anymore (potentially by breaking and replacing or just over time) and then get stuck wandering to the north-west and refuse to get into another beehive until the original one is broken. See this comment: https://bugs.mojang.com/browse/MC-166980?focusedCommentId=632081&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-632081

Villagers who lost their workstation also tend to wander to the edge of the village (many also accumulate in the north-west corner, but not all and it might be a coincidence), and combined with MC-2025 that makes them glitch through walls surrounding the village and then can't get back in, this makes it quite difficult to work with villagers (and bees).

I brought back most of the bees that flew north-west and put them into an enclosed area with empty beehives inside. They did not use the new beehives, only newly bred bees did. However, when I took the outside beehives and the bee nest (with silk touch, in case it matters) and put them back, they suddenly used the new beehives inside the enclosed area.

This seems to indicate that they still considered the outside beehives as theirs, but couldn't get in because they were full (not anymore, but they were probably full when they tried to get into them) and thus flew away. As soon as they noticed they were gone, they stopped seeing them as theirs and were free to find a new one.

So, when bees try to find a beehive, that code should probably be synchronized (mutex, critical section), so that it's impossible that 2 bees get assigned the last free spot in the beehive when actually only one can get it. (I don't think it's multi-threaded, but it can also happen in single-threaded code if all bees check for a free slot first before getting assigned to one.)

As a side note, this is a similar behavior like for villagers and beds/workstations as described in MC-150806, so these issues might be related and the faulty code could be in a parent class of bees and villagers, probably some general POI assignment code.

Older comments suggested that the villagers didn't die (no death message), but just vanished. Is that still the case or are the missing villagers now just dying? Not sure if there is a death message for unnamed villagers...

I've had no disappearances in any of my villages at all. They are all lit up and fenced in. Not all villagers sleep because some forgot their beds (and workstations) which is another known bug. Some of them roam around the edge of the village, searching for POIs, and I've had occurences of them glitching through a double fence-wall combination (see MC-2025), ending up outside the safe area (luckily none got killed recently). That might explain why villagers who don't sleep are affected.

Glitching through walls can also happen when getting out of bed (see MC-152945), maybe even glitching into walls and suffocating? That might explain why villagers who do sleep are affected. The ones in my villages sometimes sleep near walls, but not in tight spaces and often with glass (panes) at the walls. They also slide off, with their heads sometimes dangerously far inside the wall, but nothing happened (yet).

In the screenshot Maria Lemón sent, it seems as if the villages of one chunk all vanished, without any walls next to their beds (if I interpreted it correctly). This is not explained by my speculation above, so I also think it is actually several possible reasons. One question is what I asked first, whether they die or just vanish/despawn. Has anyone observed it happening in current versions?

It is not a bug for me, just surprising behavior. And I never said it was wrong in the code, I just stated how it is implemented. Whether it is a bug or WAI is for Mojang to decide. Many of the other tickets are actual bugs, like new portals being generated right next to an existing one, even being within range, happening after changes had been made for 1.15 (not necessarily related to this ticket, maybe just performance optimizations like storing portal coords so the search is faster, potentially allowing a bigger search radius). The issue here is debatable and since my opinion doesn't help the developers, I won't post any more comments, because this is not a discussion forum.

Yes, I have, like many others here. I just use Forge for that because it's easy, but there are other methods as well.

I don't know if it's a bug or intentional (probably, for performance reasons - checking 1024x256x1024 blocks is much heavier than 128x256x128). It is surprising behavior for sure, especially for new people, so I agree that it should be changed, whether it's a bug or not (unless there really is a useful usecase for this). Maybe the performance improvements Mojang did for 1.15 (maybe caching or keeping all portal coords stored) are a first step into that direction. Maybe it will be done in 1.16, the Nether Update - but that's just speculation and I think this is not the right place to discuss that since this is a bugtracker and not a discussion forum (and I don't want the Mods to smack us with a portal block or so ^^).