Re: [Condor-users] java machines with > 2gb

[Michael Rusch <mcrusch@xxxxxxxx>]

Thanks again for all of the replies.  Unfortunately, I'm still a no-go:

> Daniel Forrest wrote:
> > FWIW - on my system, java doesn't like -Xmx1g.  Try -Xmx1024m instead.

I just tried that, but it didn't work.

> From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-
> bounces@xxxxxxxxxxx] On Behalf Of Matthew Farrellee
> Or just don't pass anything at all and let the JVM pick its own default
> max heap size.

Also tried that.  Same problem

> -----Original Message-----
> From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-
> bounces@xxxxxxxxxxx] On Behalf Of Bryan S. Maher
> Sent: Tuesday, September 16, 2008 11:14 AM
> To: Condor-Users Mail List; Condor-Users Mail List
> Subject: Re: [Condor-users] java machines with > 2gb
> 
> Michael,
> 
> When I couldn't get Java to work on 32bit machines I resorted to the
> following. I wrote a small executable which logs it's command line
> arguments to a file.  I replaced the existing Java VM executable with my
> wrapper and captured the comand line used by Condor.  I then launched the
> VM from the shell using the caputered command line allowing me to observe
> the error output.  In my case it turned out the heap limit was even
> smaller than the workaround published in the manual.

This was clever.  I like it.  So, I did this (using a shell script), and
here's what I found.  It appears that condor_starter is not even running it
on start-up!  When I simply start condor cold, it never dumps anything to
the file.  However, when I run condor_starter -classad, then it does dump
stuff to the file, and when I run the command that it wanted to run, it runs
fine.

So, it appears that what is happening is that condor isn't even doing the
java test-drive when it starts up.

Michael