Closed
Description
There's a qcg.pilotjob.api.manager.Manager.cancel()
method, but according to the documentation it always raises an exception. To avoid using up the rest of the core hours in the allocation on useless computation, I'd like to be able to shut down the other jobs by cancelling them if one of my jobs fails. I think this method would let me do that, could it be implemented? Ideally by first sending SIGTERM and then after 10s or so SIGKILL, but an immediate SIGKILL will work for the time being.
Metadata
Metadata
Assignees
Labels
No labels