mojira.dev

nejento

Assigned

No issues.

Reported

MC-232009 Minecraft does not resolve SRV records correctly Fixed MC-182412 Dark Oak teleports to ground Duplicate MC-153145 Disappearing texture of boats and stuttering walking Duplicate MC-152300 Raining through glass Duplicate MC-133461 Maps in item frames are going dark when a some part is out of vision Duplicate MC-133460 Rain particles on bottom of water block Duplicate MC-101759 /stopsound command doesn't mute specific sounds in master source Fixed

Comments

The CPU usage in 1.14.2 is still incredibly high. Only one player in the game and the server suddenly used something over 1 GB of RAM and CPU usage was still too high.
When discovering the map and generating new chunks and also moving on already discovered chunks the performance was bad and the console was spamming information about player moving too quickly.

Here is the debug log of a clean vanilla 1.14.2 server and latest.log of the same run:

[media][media]


Run was done on Linux running Java 8u191 on 64-bit architecture.

As gamehosting side we still have to consider this version unoptimized as it is still too demanding, especially since the version 1.13 was perfectly stable and not resource-hungry as this new version. We still don't understand what happened.

Here's something I'd like to add. I've done some testing on an unofficial Spigot server and done some timings tests there because of a bit better profiler. I completely understand that their unofficial server is indeed unofficial and brings some optimizations like offloading some tasks to different threads or remove some things from ticking. However the reason why I'd like to share results of my tests of running a server on some settings on different Java version with various launch parametres is that when you run some plugins that hook onto events that use the most of the CPU times, they might be able to help the team see what events are causing the issues.
I completely understand that you might want to ignore those results and please if you consider them useless, feel free to ignore them or remove. I personally believe that any kind of result of some testing might be able to help fix these pesky issues that are really not helping 1.14 be the greatest update.

The testing was done on Linux server running a few more servers at once (shared hosting server) running on Core i7 CPU with 5 GB of RAM. The testing included first walking slowly around spawn area, then flying in creative quite a big distance away in one direction then teleporting to another world where I flew quickly from spawn. Then teleporting to another world where a huge amount of player skulls with skins are available for creative building.
Map is force upgraded from previous version 1.13.2 to 1.14 and then force upgraded to 1.14.1.
There are some additional plugins which manipulate with multiple worlds (Multiverse-Core), multiple inventories management plugin (Multiverse-Inventories) while this plugin hooks to Multiverse-Core, plugin for creating NPCs (Citizens 2) while no NPC was created during that time, plugin hooking to multiple events (CraftBook) and protocol-altering plugin that works as a library (ProtocolLib).
From the testing it is visible that only multiworld inventory management plugin Multiverse-Inventories took the most CPU time while travelling between worlds which either could mean that there might be an issue with unloading chunks or loading them.

Timings (every run had set parameters -Xms128M and -Xmx5G):

For comparison (same maps, same methodology, very similar plugins): Testing on Oracle Java 8u191 ran with parameters -XX:+UseParNewGC and -XX:MaxGCPauseMillis=20: https://timings.spigotmc.org/?url=fatohucoqa

Finally again, if you consider these timings results and testing not helpful (because it ran on unofficial modified server even though it is based on original vanilla server) feel free to ingore it or mods delete it. I personally think that every little bit might help solve this very annoying issue.

Hello,
I can confirm this bug is real. I installed both Code Connection from Windows Store and even from Minecraft: Education Edition website and neither of them can connect to Minecraft: Windows 10 Edition (Version 1.7).

The connection reports as connecting via websockets and then immediately closes. Chat reports "Connection closed". This connection close makes Code Connection blink as it refreshes its content.

I attached two crash reports with very similar (probably the same) problem. (

[media]

,

[media]

)

I can also add to that that it happened for example when a Creeper was exploding underground. This is a case of running a multiplayer server on Java 10 

 

I also added 

[media]

which seems as a different error caused by same thing.

Can confirm. I also can confirm, that when you join a world this speed "buff" can happen as well and that is not dependant on Beacons.

I also can add that when on a multiplayer vanilla server, this speed "buff" happens only client-side and server speed is not affected. Other players can see your speed movement. Even though hand animation is affected, the actual speed of eating something is not affected. The same for mining.

During this weird speed movement the server doesn't report "moved too quickly" and doesn't kick for flying or other kind of cheat.

Your movement is sped up as well as falling, climbing ladders, hand animation and speed of jumps.

This probably isn't some Java thing, since the original reporter stated he was using Java 8 and I'm running Minecraft both client and server on Java 10.

 

(side note: this probably is duplicate to MC-134833)