eclipse MAT in headless mode

[Bernd Eckenfels]

Hello,

I think Xmx for the headless mode is defined in the start script (or
the command line), not in the .ini file. (but could be wrong dont have
a current version here)

You can either  use the created reports only (they are HTML inside ZIP
or something like that), or you can after the headless parsing is
finished start the Gui and just open the heap with "open". It will
detect that the file is already parsed and use the existing index
files. (For this to work they have to be in the same dir as the hprof
file).

The headless mode does take a bit less memory than the Gui mode (but
not much, so typically it does not make sense to use the headless mode
on the workstation where you want to use the gui).

For the Xmx, you can go risky and specify RAMsize or even a few
percentage more. Chances are the passing does not use all of it (but
completes before OOMError).

Gruss
Bernd

Am Tue, 18 Mar 2014 14:33:31
+0100 schrieb Luca Maderna <lmaderna at gmail.com>:

> Thanks Bernd,
> 
> Unfortunately I can't work on production remote machine.
> So I launched on my pc
> 
> ParseHeapDump.bat <dumpFile> org.eclipse.mat.api:suspects
> org.eclipse.mat.api:top_components #
> 
> after few minutes goes in OutOfMemory, so now I'm retrying setting in
> MemoryAnalyzer.ini a larger memory setting.
> 
> Anyway, when ParserHeapDump finish correctly how can I use these index
> files?
> I've only to open my *.hprof file and MAT automatically use index and
> report files?
> 
> thank you very much
> luca
> 
> 
> 2014-03-13 21:36 GMT+01:00 Bernd Eckenfels <bernd-2014 at eckenfels.net>:
> 
> > Hello,
> >
> > copy your heapdump into an otherwise empty directory (if it is gz
> > uncompress it, in case of IBM you need to run jextract on it (but
> > not uncompress it afterwards).
> >
> > Then you unpack MAT on your machine, you will find a batch file and
> > a shell script ParseHeapDump.sh in the start directory, use it to
> > process your heapdump. You need to set your PATH variable to
> > incluse an aproperiate java executable first.
> >
> > It will populate the directory with all kind of index
> > files. The whole directory can then be copied to a smaller desktop
> > machine.
> >
> > Instead of only parsing you can also generate some reports. This FAQ
> > entry here is about something else, but it lists the report names:
> >
> >
> > http://wiki.eclipse.org/MemoryAnalyzer/FAQ#How_to_run_on_64bit_VM_while_the_native_SWT_are_32bit
> >
> > Bernd
> >
> >  Am Thu, 13 Mar 2014
> > 12:22:18 +0100 schrieb Luca Maderna <lmaderna at gmail.com>:
> >
> > > Hi.
> > > Could you suggest me how run MAT in headless mode?
> > >
> > > I've to analyze large dump files (about 8GB) and I've some problem
> > > open it in my pc's.
> > >
> > > thanks
> > > luca
> > >
> > _______________________________________________
> > hotspot-gc-use mailing list
> > hotspot-gc-use at openjdk.java.net
> > http://mail.openjdk.java.net/mailman/listinfo/hotspot-gc-use
> >
>