[Condor-users] on_exit_remove in submit file

[<Greg.Hitchen@xxxxxxxx>]

Hi

 

We are running a windows pool in which a non-constant, random, small percentage

of execute nodes occasionally return the old -10737**** type errors. By using:

 

== 0)

in the submit file these are detected, requeued, and run OK somewhere else.

 

Just recently a user has also had issues with jobs returning immediately

and with the output file (specified in the output = statement) being 0 bytes in size

(it should contain a fair bit of data). If manually resubmitted, it will run OK.

This user has asked if there is a way to do something like:

 

== 0)  && (OutputFileSize != 0)

 

where obviously OutputFileSize is the size of the file specified in output =

 

While looking into this problem to figure out why this is happening for their jobs

I thought it might be useful if it was possible to do such a thing as specified in

the line above. I'm guessing it would need to be kludged in some way because

as far as I know their is no such classad, or maybe it's not possible at all.

 

I'd appreciate people's comments or suggestions on this.

 

Thanks

 

Cheers

 

Greg