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

Re: [HTCondor-users] Experimenting with preemption

On 10/02/2018 11:28 AM, Michael Pelletier wrote:
Hi everyone,

I'm starting to do some testing with preemptable jobs, after years of sidestepping it due to the limitations of our usual applications, and I was wondering if there's a way to manually trigger a preemption event for a job to verify the mechanisms for a given application. The condor_hold command hard-kills everything, and condor_vacate works on machines rather than jobs from what I can tell. It looks so far like I'd have to send a test job to a particular machine and make sure it's the only job there, then condor_vacate it, in order to test that I have the right kill_sig and what have you for the job, but I was hoping to find something that would work on a job level directly.

Are you looking for condor_vacate_job?

Also, condor_vacate takes a slot name or a machine name, so you can

condor_vacate slot3@some_machine_name