[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Location of Execution files



> I have just a  few simple questions. When A job is executing
> on a machine it creates the directory
> "c:\condor\execute\dir_<'a number'>" right? Is there a way of
> determining what <'a number'> is going to be or a way of
> controlling it?

It is always the process ID of the condor_starter process that gets
spawned to run the job. Condor places the path in your environment when
the job's Cmd is executed. I'm not at my pool right now but IIRC it's
_CONDOR_EXEC_DIR. Something close to that. So your job can read that env
var on startup.

As for controlling it: I seem to recall reading a snippet in a release
note for 7.2 or 7.3 about a new config variable that'll let you set a
per-slot execution directory. Or maybe that was just a wonderful dream I
had and I'm mistaken. If there isn't a config variable to do this
consider it my request. Sure would be nice to split execute directories
between multiple disks in my machines to reduce some I/O contention.

> Alternatively, is there a simple way anyone knows of
> manipulating in C++ a file I am sending across Condor using
> "transfer_input_files =" in the job submit file. If any
> questions, or further explanation are required, don't
> hesisate to email.

What are you trying to do? Edit a text file that contains C++ code and
then compile it as part of a job?

- Ian

Confidentiality Notice.
This message may contain information that is confidential or otherwise protected from disclosure. If you are not the intended recipient, you are hereby notified that any use, disclosure, dissemination, distribution,  or copying  of this message, or any attachments, is strictly prohibited.  If you have received this message in error, please advise the sender by reply e-mail, and delete the message and any attachments.  Thank you.