ansaurus

Question

java.lang.OutOfMemoryError: requested 32756 bytes for ChunkPool::allocate. Out of swap space?

Answer 1

+1 A:

Google turns up a bunch of responses to this. I started reading the first one, and while its conclusions (reduce the heap) are reasonable, it doesn't really talk about root causes. So ...

This is coming from native code, and indicates that the code is unable to allocate memory from the C heap. There are a couple of reasons this could be happening, one of which is that the virtual memory space is truly used op, the other is that the C heap is fragmented. If the HP has pmap, that's the first place to start (or find its equivalent and start there). If you see allocation blocks filling the virtual memory space in an unbroken block, then the solution is simple: reduce your heap.

However, if you see allocation blocks scattered throughout the virtual memory space, you're going to need some deeper digging. Sometimes this happens because you have Java objects that aren't reclaiming native resources. Sometimes it happens because of a quirk in the operating system (I've seen Solaris get confused with thread stacks).

kdgregory 2009-12-11 14:28:24

I don't see a man entry for pmap on my server, will see what i can find as an equivalent. the crash is not occuring at any predictable time, although it does happen at least once a day and sometimes twice. i was thinking of trying a new version of the HP JDK to see if that makes a difference.

marta 2009-12-11 14:59:28

heap seems ok in crash log? Heap PSYoungGen total 308544K, used 132524K [5deb0000, 72790000, 73400000) eden space 307328K, 43% used [5deb0000,6601b248,70ad0000) from space 1216K, 0% used [72660000,72660000,72790000) to space 13888K, 0% used [70c70000,70c70000,71a00000) PSOldGen total 174784K, used 90916K [33400000, 3deb0000, 5deb0000) object space 174784K, 52% used [33400000,38cc9350,3deb0000) PSPermGen total 226304K, used 113097K [23400000, 31100000, 33400000) object space 226304K, 49% used [23400000,2a272528,31100000)

marta 2009-12-11 15:01:02

sorry about that formaating, let me try again:Heap PSYoungGen total 308544K, used 132524K [5deb0000, 72790000, 73400000) eden space 307328K, 43% used [5deb0000,6601b248,70ad0000) from space 1216K, 0% used [72660000,72660000,72790000) to space 13888K, 0% used [70c70000,70c70000,71a00000) PSOldGen total 174784K, used 90916K [33400000, 3deb0000, 5deb0000) object space 174784K, 52% used [33400000,38cc9350,3deb0000) PSPermGen total 226304K, used 113097K [23400000, 31100000, 33400000) object space 226304K, 49% used [23400000,2a272528,31100000)

marta 2009-12-11 15:02:19

not sure what i am doing wrong. i've tried four spaces leading and trailing to get it to format right.

marta 2009-12-11 15:03:31

That particular kind of formatting only works in questions and answers, but not in comments, alas. If legibility is vital, you can edit your question to include the information and use the comment to point out the update.

Carl Smotricz 2009-12-11 15:18:17

thank you very much.

marta 2009-12-11 15:21:13

Answer 2

A:

So you're running out of memory. Compiling a JSP is a memory hog, so it's not too surprising that your container chose that point to blow up.

There are a few things that can be done to explore the problem, but I usually try the simplest remedy first: I give the program more memory.

This:

VM Arguments: jvm_args: -Xms256m -Xmx1024m -XX:PermSize=128m -XX:MaxPermSize=256m

indicates you are running with a pretty "standard" memory configuration. I would try increasing some of those numbers.

There's a (shell or .bat) script to start up WebLogic; I've forgotten its name or structure. But if you dig around in it, you will find Java being called with a set of options that are defined further up. You can shut down your server, back up this script file and edit it to fiddle the memory settings (-Xms, etc) and see where that gets you when you restart.

Carl Smotricz 2009-12-11 14:34:33

I've actually gone up and down with the Heap and Perm size parameters. Currently I an getting the error with the parms set like this:-Xms256m -Xmx1024m -XX:PermSize=128m -XX:MaxPermSize=512m I've also tried setting them with the same value both higher and lower (-Xms1024m -Xmx1024m, -Xms512m -Xmx512m.) Nothing makes a difference. Heap seems ok based on the log entries. I think if it were PermGen it would throw a different error but I've tried changing them too.

marta 2009-12-11 14:50:23

Thank you for this helpful information. Unfortunately, when this band-aid fix fails, I usually have to call in the troops (i.e some kind of support, since you're paying for a commercial operating system). I have a strong aversion to HP-UX: In my opinion, HP builds great hardware but should never have tried its hand at any software.

Carl Smotricz 2009-12-11 15:06:20

I found this in the readme for the newest version:In JDK 6.0.03, occasionally, a Java program can experience a memory leak in the C heap due to QXCR1000920722, resulting in an out-of-memory error.That sounds promising based on comments above.

marta 2009-12-11 15:22:37

Answer 3

+1 A:

Updating this for future searchers. Solution was to upgrade the JDK.

To recap, this was done due to random OOM JVM crashes. Per HP:

http://docs.hp.com/en/JDKJRE60RN/jdk%5Frnotes%5F6.0.05.html

In JDK 6.0.03, occasionally, a Java program can experience a memory leak in the C heap due to QXCR1000920722, resulting in an out-of-memory error. HP recommends that you upgrade from JDK 6.0.03 to JDK 6.0.04 (or greater) to prevent possible C heap memory leaks due to QXCR1000920722.

Thanks all.

marta 2009-12-16 15:43:21

ansaurus

tags:

views:

answers:

java.lang.OutOfMemoryError: requested 32756 bytes for ChunkPool::allocate. Out of swap space?

related questions