[SOLVED on 1137]Anyone else seeing memory leaks on 1133 "latest" version Forge?

[jabelar]

I updated to 1133 and am getting several indications of memory leaks.  JVM is regularly shutting down for being out of heap (even though I have run parameters set for mimimum of 4GB) and just now Windows warned it was getting dangerously low on memory (due to JVM process) and then promptly crashed all of Windows.

 

It could be something I did, although the mod I'm running is very basic and doesn't seem to have anything I would consider problematic.

 

It is probably me, but just wanted to see if anyone else saw similar issue?

[brandon3055]

Hmm i was just considering updating to 1133 10 seconds before reading this post.... oh what the heck im updating to 1133 will let you know if i encounter anything problematic.

 

Edit: My client has been running on 1133 for a few minutes now and i haven't noticed any increase in memory use.

[jabelar]

Hmm i was just considering updating to 1133 10 seconds before reading this post.... oh what the heck im updating to 1133 will let you know if i encounter anything problematic.

 

Edit: My client has been running on 1133 for a few minutes now and i haven't noticed any increase in memory use.

 

I should clarify that it doesn't happen to me every time.  Actually it seems to happen over several consecutive runs.  On average maybe on 10th run or so.  It is like the JVM doesn't exit properly after running or something -- I think I saw the task manager with four or five JVMs running.

 

I'll keep monitoring it to see if I can figure out any pattern to it.

[jabelar]

Okay, it seems like the JVM isn't exiting properly.  I'm getting multiple copies of JVM running.  Currently i see four running and they each take about 1.5GB.  Note that I only have one Minecraft instance running on task bar, I'm sure of that.  I'm also just quitting Minecraft using the standard "Quit Game" button.

 

I think (can you guys confirm), that there should be two JVM tasks showing in the task manager -- one for Eclipse and one for the running mod.  I can see that even after I exit Minecraft that a JVM task that says it is is the mod continues to run...

 

It still doesn't happen 100% of the time though.  I'll try to see if I can find a pattern.

 

Anyone have any ideas on why this might happen (JVM continues to run even though Minecraft is quit)?

[shadowmage4513]

Just a stab/guess -- but it is probably related to cpw's recent security manager additions that... 'catch' when you exit Minecraft.  I'd wager that there is still some work to do on the system.  Looks like it was added in @ version 1029.

 

I would say test to see that it happens with just forge/no mods installed, and submit a ticket with FML on the problem.

[jabelar]

Just a stab/guess -- but it is probably related to cpw's recent security manager additions that... 'catch' when you exit Minecraft.  I'd wager that there is still some work to do on the system.  Looks like it was added in @ version 1029.

 

I would say test to see that it happens with just forge/no mods installed, and submit a ticket with FML on the problem.

 

I was wondering the same thing (if it is related to security manager): when exiting the errors are now definitely different and related to the security manager stuff.  Quitting the game now shows following console errors:

[09:45:24] [Client thread/ERROR]: Unable to launch
java.lang.reflect.InvocationTargetException
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) ~[?:1.7.0_51]
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57) ~[?:1.7.0_51]
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) ~[?:1.7.0_51]
at java.lang.reflect.Method.invoke(Method.java:606) ~[?:1.7.0_51]
at net.minecraft.launchwrapper.Launch.launch(Launch.java:134) [launchwrapper-1.9.jar:?]
at net.minecraft.launchwrapper.Launch.main(Launch.java:28) [launchwrapper-1.9.jar:?]
Caused by: cpw.mods.fml.relauncher.FMLSecurityManager$ExitTrappedException
at cpw.mods.fml.relauncher.FMLSecurityManager.checkPermission(FMLSecurityManager.java:23) ~[forgeSrc-1.7.2-10.12.2.1133.jar:?]
at java.lang.SecurityManager.checkExit(SecurityManager.java:761) ~[?:1.7.0_51]
at java.lang.Runtime.exit(Runtime.java:107) ~[?:1.7.0_51]
at java.lang.System.exit(System.java:962) ~[?:1.7.0_51]
at net.minecraft.client.Minecraft.shutdownMinecraftApplet(Minecraft.java:878) ~[Minecraft.class:?]
at net.minecraft.client.Minecraft.run(Minecraft.java:948) ~[Minecraft.class:?]
at net.minecraft.client.main.Main.main(Main.java:112) ~[Main.class:?]
... 6 more

[Bektor]

Just a stab/guess -- but it is probably related to cpw's recent security manager additions that... 'catch' when you exit Minecraft.  I'd wager that there is still some work to do on the system.  Looks like it was added in @ version 1029.

 

I would say test to see that it happens with just forge/no mods installed, and submit a ticket with FML on the problem.

Memory leaks. That sounds like that: http://www.minecraftforge.net/forum/index.php/topic,20213.0.html

Well, its not exact the same problem, but there are memory lags to by generating a jungle world. (Here memory leaks on launching, there on generating a world with a jungle biome)

[jabelar]

I think it is narrowed down to JVM not shutting down, not a general leak per se.

[brandon3055]

Ok so i am seeing the same error in my console when i quit except it is followed by

Disconnected from the target VM, address: '127.0.0.1:50215', transport: 'socket'

Just a guess but maby my IDE (Intellij) is detecting that the process didn't shut down properly and is terminating it.

 

Edit: Oh it is happening to me you just have to actually load a world. Fortunately for me Intellij tracks each instants it launches separately so i can just shut it down from the console.

 

Unless you need to save your world i recommend terminating the instance from your IDE and that should completely stop it.

[jabelar]

Yeah, now that I know what is happening I can shut down the task in various ways -- I'm just using the task manager.

 

Anyway, I guess this is "solved" since it seems to be a actual issue with the latest release (as opposed to something I'm doing wrong) and it can be worked around.

[jabelar]

Looks like this issue is solved on release 1147.