MC-41874Extremely Slow Server Chunk Ticking -- Unknown Cause
Server chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.2: 19.94 ticks/sec
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.4: 11.03 ticks/sec
I suspect it is a side-effect of MC-12964 , in part, as well as the recent changes made to the rendering and network code in 1.7.
This is a Minecraft forum thread discussing the same effects: http://www.minecraftforum.net/topic/2218534-graphics-quality-and-performance-in-164-172-and-174/
Related: MC-43395
Environment
Windows 7 (64-bit), Java 1.7.0_45 (64-bit), 1 GB dedicated to Minecraft (910 available)
Attachments8
Created Issue:
Severe Singleplayer Memory Leak
When playing singleplayer worlds, especially loading new chunks, the memory usage spikes and refuses to go back down, even when pausing the game/exiting the world. Eventually the game runs out of memory and/or crashes. On two occasions, there was not enough memory left to save a crash report or display the out of memory screen. This seems to be caused by the ticking hogging the CPU.
Environment
Windows 7 (64-bit), Java 1.7.0_09 (32-bit), 512 MB dedicated to Minecraft (494 available)
Attachments9
2013-12-07_18.03.23.png
Extremely Slow Server Chunk Ticking(Memory Leak?)Extremely Slow Singleplayer Server Chunk Ticking
When playing singleplayer worlds, especially loading new chunks, the memory usage spikes and refuses to go back down, even when pausing the game/exiting the world. Eventually the game runs out of memory and/or crashes. On two occasions, there was not enough memory left to save a crash report or display the out of memory screen. This seems to becaused bythe tickinghogging theCPU.Server chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
New world, render distance 16 on 1.7.4: 1.31 t/sec
New world, render distance 16 on 1.7.2: 19.95 t/sec
Extremely SlowSingleplayerServer Chunk Ticking
Server chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
New world, render distance16 on1.7.4: 1.31 t/sec
New world, render distance16 on1.7.2: 19.95 t/secServer chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.2: 19.94 ticks/sec
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.4: 11.03 ticks/sec
Server chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.2: 19.94 ticks/sec
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.4: 11.03 ticks/secI suspect it is a side-effect of
MC-12964
Windows 7 (64-bit), Java 1.7.0_
09 (32-bit),512 MB dedicated to Minecraft (494available)Windows 7 (64-bit), Java 1.7.0_45 (64-bit), 1 GB dedicated to Minecraft (910 available)
Server chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.2: 19.94 ticks/sec
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.4: 11.03 ticks/secI suspect it is a side-effect of
MC-12964, in part, as well as the recent changes made to the rendering and network code in 1.7.
This is a Minecraft forum thread discussing the same effects: http://www.minecraftforum.net/topic/2218534-graphics-quality-and-performance-in-164-172-and-174/
Extremely Slow Server Chunk Ticking -- Unknown Cause
Server chunk ticking is extremely slow, especially when loading new chunks. This slow chunk ticking (around 1-3 ticks per second) occurs even with normal exploration. When in singleplayer, this can cause the "ticking" to use an abnormally large amount of CPU.
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.2: 19.94 ticks/sec
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.4: 11.03 ticks/secI suspect it is a side-effect of
MC-12964, in part, as well as the recent changes made to the rendering and network code in 1.7.
This is a Minecraft forum thread discussing the same effects: http://www.minecraftforum.net/topic/2218534-graphics-quality-and-performance-in-164-172-and-174/Related:
MC-43395
This might be an effect, but:
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.2: 19.94 ticks/sec
Seed 1, render distance 8, travel from spawn to x=2000, v1.7.4: 11.03 ticks/sec
The extra chunk ticking seems to have slowed down the server. You can see debug files, etc. at MC-41874
I have experienced similar issues with SMP in MC-41874. My comments can be found here:
https://mojang.atlassian.net/browse/MC-41874?focusedCommentId=125551&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-125551
I've had the same problems, and reported a bug before ( MC-41874 ); I've linked your issue in the description.
Your Java version is out of date! Please update your Java to the latest version (1.7.0_45): http://www.java.com/en/download/.
Please report if the issue persists with the latest Java version.
I will give you the benefit of the doubt and assume that this is an automated response. In this case, the bug persists ONLY with version 1.7.3. Version 1.7.2 has no problems in an identical environment. (see profile-results-1.7.2.txt)
He meant update to the latest Java version and see if you can reproduce, not the latest Minecraft version. Just so you know.
Anyway, by the look of things you've made a mistake with the allocated memory. Can you allocate any more (Edit Profile>JVM Arguments>change argument to the original 1 Gig or disable JVM arguments) or is this not possible because of your computer specs? It may be as simple as this, or I could be wrong.
I understood perfectly. I meant that the memory leak happens on 1.7.3, but not 1.7.2, so it cannot be the java, as 1.7.3 has no additional java requirements. And yes, I can allocate more ram, but I have a feeling that it won't make any difference.
I allocated 1G to Minecraft, and the lag spikes went down, but the dev console still spams errors everywhere, and chunk loading is still slow. In fact, increasing the ram made the problem WORSE. (2.17 ticks/sec as opposed to about 5 ticks/sec with 500M)
It's a requirement you have your java up to date before we proceed, sorry.
No longer.
With most recent 64-bit java, server chunk ticking is still around 13 tps, in 1.7.2 was 20. However, this will need more testing. Actually, do expect the report to be updated. Seems many people are experiencing this.
I've been having a substantial lag problem since I updated my vanilla server from 1.7.2 to 1.7.4. I'm using the 64-bit openjdk 7 jvm with CMS for my garbage collector and 2.5 gb heap.
At first I thought the problem only happened when 4-5 people were logged in at a time, but it doesn't matter so much. what does matter is how long the instance had been running and how far the heap has expanded.
Unfortunately I've only been reviewing the GC times since the problem started happening. what I have observed is a constant minor gc every few seconds for about two thousandths of a second but as the heap expands to the Xmx setting it takes significantly longer in the hundredths of a second.... sometimes more.
Restarting the service helps for 12-36 hours.
Let me know what other information could be helpful.
Before I moved GC logs to its own file, I was able to gleam this from the console's standard out after the issue had been observed:
Here is how we look shortly after the service was restarted:
The Can't keep up! messages are very consistent and correspond to the observed lag.
We are running a stock Debian 7 (wheezy) server with vanilla MC 1.7.4. The stock java version in this environment is:
This was not an issue in 1.7.2.
Perhaps duplicate of
MC-42088?I think that 42088 is an effect. However, if I am correct and this is caused by
MC-12964, then the chunk ticking is actually too fast, and it is clogging the server, especially with 32-bit java. (but even with 64-bit the ticks per second are cut in half)