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

Re: [HTCondor-users] rejecting jobs



On 2/7/2018 4:00 AM, Alessandra Forti wrote:
Hi,

is there are way to rejct jobs in HTCondor in a way that they don't end up hanging afterwards in the queue?


What do you mean by reject?

What specifically do you want to happen?

Currently by default, jobs leave the queue when either
  a) the job exits of its own accord (i.e. it completes), OR
  b) the job is removed (via condor_rm or periodic_remove)

Under what other situations do you want the job to leave the queue?
Perhaps you want a job to only be started once, and never restarted again (even if the job is preempted, or the execute machine is shut down in the middle of the job, or execute node loses power, etc) ? If so, take a look at this HOWTO recipe:

  http://wiki.htcondor.org/index.cgi/wiki?p=HowToAvoidJobRestarts

Hope this helps,
Todd

I tried to add a rule to some machines START macro, but I end up with many idle jobs and cleaning them up afterwards with some *REMOVE macro is not ideal. It'd be better if the batch system rejected the jobs with an error. I know this is close to the concept of queue, and indeed it is what i'm trying to do.

cheers
alessandra