I have a single node system containing 2 sockets, 6 cores each. I want to be able to limit the number of simultaneous jobs (of type VCS) to 7 and allow 5 additional non VCS jobs to run.
I want to be able to SUSPEND lower priority jobs with higher priority jobs
(of type VCS).
I want to be able to run batch (sbatch) jobs and interactive (srun) jobs
simultaneously and still obey the limit of 7 and priority.
Attach is the config I’m trying.
If I put 7 jobs in partition vcsSLow
sbatch -p vcsSLow –L vcsSLow ./runit 300
the 8th job will stay in the PENDING state.
If I then put a job in the vcsS partition:
sbatch -p vcsS –L vcsS ./runit 300
It SUSPENDS one of the 7 running jobs and runs.
So far this is the desired behavior.
If I try to use srun as:
srun -p vcsS –L vcsS ./runit 300
I get and error message
srun: error: Task launch for 442.0 failed on node svthwVCS: Invalid job
credential
srun: error: Application launch failed: Invalid job credential
srun: Job step aborted: Waiting up to 2 seconds for job step to finish.
srun: error: Timed out waiting for job step to complete
How can I get sbatch jobs and srun jobs to execute simultaneously?
Thank you,
Donald W Monroe
SimpliVity Inc
8 Technology Dr
Westborough, MA 01581
508-433-0411
www.simpliVity.com <http://www.simplivity.com/>
config
Description: Binary data
