In multiplayer, a block change next to an unloaded chunk that changes the opacity level of the block causes a massive client freeze, about 1-2 seconds for a single block.
A light level change next to an unloaded chunk also causes a freeze, but a much shorter one. Around 10-20 blocks have to change their light level for the freeze to be noticeable.
A block change that does not affect the light level or opacity level does not appear to cause any freeze.
These freezes only seem to happen when connected to a server, not in single player. I believe these are the causes of most common freezing in custom PvP maps, which are often slightly larger than the view distance and have many block changes happening everywhere, all the time.
The following video demonstrates the freeze from a single block opacity change. Notice that the light level of the block is zero, and the block is completely enclosed in other opaque blocks:
Linked issues
is duplicated by 1
Comments 5
This is the same as https://bugs.mojang.com/browse/MC-12799 and the fix (obviously) solved that one aswell.
no, it's not, that one is about an fps drop for so long, you can't move, this one is for about an sec
@@unknown Well I dont know, but I used a setup similar to the setup in this report and I got FPS drop for longer periods of time. At about 35 second in to the video (in this report) you can see that the laggspikes come several times after eachother. My assumption is that he got a better computer than I do, and theirfor the lagspikes are shorter. (= my pauses between the lagspikes gets shorter and theirfor act as a long lagspike with just some breaks in it now and then).
Anyway.. The part that I posted about in the comments of that post behaves in the exact same way as this one with the only difference that I get got longer lagspikes. Also... My replication setup that I used to replicate it before is fixed in this snapshot and I cant reproduce the problem I posted about anymore.
Basically what I am saying is that the length of a laggspike depends a bit of your computer, so if the only difference between two reports are the length of laggspikes, doesnt it seem like its the same issue?
Nathan Adams
I can see why this used to happen... The client was probably looking for all the loaded chunks to find the one next to it, so it could determine if the side of the block would be invisible or not...