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

[condor-users] Confirm my understanding of the spool directory



>From reading the code, I think I understand how condor treats the spool directory, but I would like for someone to confirm this.  When a job is created, two directories corresponding to it are created on the Central Manager under "spool," a main directory and a tmp directory.  If the tmp directory already exists, it is emptied.  If the main directory already exists, any files in it are submitted along with the job's input files.

When the job is preempted, any intermediate files are returned to the spool tmp directory.  If the tmp directory contains a .ccommit.con file, its contents are placed in the main directory.  The tmp directory is then emptied.

When the job is matched with a node, any files in the main directory are sent to the grid node along with any input files.

Is this correct?

Why would my simulation not be creating a .ccommit file?  I've noticed that intermediate files are returned to my central manager, but no .ccommit.con file exists, so they're just erased when the job is matched with another node.

Thanks,
David
Condor Support Information:
http://www.cs.wisc.edu/condor/condor-support/
To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
unsubscribe condor-users <your_email_address>