Skip to main content


Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Eclipse Projects » Memory Analyzer » ClassLoader analysis and IBM .phd files(ClassLoader analysis seems broken in MAT 1.0.1 for .phd files)
ClassLoader analysis and IBM .phd files [message #658673] Wed, 09 March 2011 12:55 Go to next message
Jon Seymour is currently offline Jon SeymourFriend
Messages: 5
Registered: March 2011
Junior Member
Hi,

I am trying to analyse a .phd file created by a 32 bit Windows IBM Java 5 JVM.

The ClassLoader analysis doesn't seem to work - every class seems to be owned by a single system classloader.

I did see a bug report

https://bugs.eclipse.org/bugs/show_bug.cgi?format=multiple&a mp;id=267592

that fixes a NullPointerException for this case, and presumably 1.0.1 already has that fix, but the analysis itself doesn't work.

Are there any plans to fix it or am I just not using the feature correctly?

jon.
Re: ClassLoader analysis and IBM .phd files [message #658722 is a reply to message #658673] Wed, 09 March 2011 15:49 Go to previous messageGo to next message
Andrew Johnson is currently offline Andrew JohnsonFriend
Messages: 141
Registered: July 2009
Senior Member
Which JVM version generated the dump?

Window > Heap Dump Details > JVM Version

Some older JVM versions don't generate class loader information in the PHD file, so MAT can't display useful class loader information.

Try putting the javacore.txt file in the same directory as the heapdump.phd file, delete the index files and reparse the dump. The javacore file does contain some class loader information and this might be enough for the DTFJ/PHD reader to find out the class loaders.
Re: ClassLoader analysis and IBM .phd files [message #658840 is a reply to message #658722] Thu, 10 March 2011 00:54 Go to previous message
Jon Seymour is currently offline Jon SeymourFriend
Messages: 5
Registered: March 2011
Junior Member
Andrew,

Thanks. This worked. I didn't have a javacore from exactly the same time as the heapdump, but I had one from a few minutes before. I renamed it to have the same name pattern as the .phd file, and this worked.

For the record the JVM I was using was:

J2RE 5.0 IBM J9 2.3 Windows XP x86-32 build j9vmwi3223ifx-20080811
VM build 20080809_21892_iHdsMr
JIT enabled - 20080620_1845_r8

jon.
Previous Topic:What's the best way to package and share OQL queries
Next Topic:Remainder sector is very big
Goto Forum:
  


Current Time: Fri Sep 21 00:02:57 GMT 2018

Powered by FUDForum. Page generated in 0.02661 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top