Re: [Condor-users] Java universe and memory (moved from devel to user)

[Jaime Frey <jfrey@xxxxxxxxxxx>]

On Mar 7, 2008, at 2:27 PM, Craig Bruce wrote:

> > the on_exit_remove or on_exit_hold can trap this and place it on hold
> > for you to deal with.
>
> I can't use either of these as the job never gets as far as exiting,  
> it just
> goes back to idle and will resubmit to get the same error, ad  
> infinitum.
>
> The exit code is 1, as an abnormal termination, so I tried this in
> on_exit_hold and periodic_hold, but the first doesn't run and second  
> runs
> before the exitcode is defined.
>
> Is there something like on_evict_hold? I couldn't find anything in the
> manual.


Condor evaluates on_exit_hold/remove when the job completes and is  
ready to leave the queue. Since Condor leaves the job in the queue on  
OutOfMemory, the on_exit expressions are evaluated.

Here's how you can use periodic_hold:
periodic_hold = NumJobStarts =!= Undefined && NumJobStarts > 2

The first half of the expression is required because NumJobStarts  
isn't defined in the job ad until it starts running for the first time.

This will catch jobs that re-execute other reasons as well, but it  
will stop infinite re-execution.

Thanks and regards,
Jaime Frey
UW-Madison Condor Team