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

Re: [Condor-users] Howto make vanilla universive only run serial jobs and don't suspend any jobs on one-machine Condor?



dawnsong wrote:
> Hi, all,
> I just installed condor in my ubuntu 8.04 system with two CPUs (each has
> four cores) and 4G memory.
> I usually want to run vanilla jobs serially one by one or four by four. Now
> condor displays that my server could run 4 jobs to 6 jobs at the same
> time. But often, when I use this server to do some other things at the same
> time such as editing scripts, Condor suspends one or two jobs to make server
> run at least 4 jobs. Sometime, even I didn't interfere the server, when I
> submitted 10 or more jobs, Condor would suspend 4 or more jobs after they
> run for a while.
> 
> I guess condor could manage the system computing resource dynamically and
> assign jobs the resource. But once condor suspends some jobs, it would not
> restart or resume them again even the server stay idle for a whole day.
> 
> I tried condor_hold and condor_release but condor still can't resume or
> restart the jobs.
> 
> All I want is to make Condor run my jobs one by one or four by four. Or at
> least condor should not suspend any jobs since condor can't resume the
> vanilla jobs or restart the vanilla jobs. Any suggestions would be
> appreciated.
> 
> Besides, after the installation, I only changed the Part1 in global
> configuration file.
> 
> Thanks,
> 
> Xiaowei Song

You should change the SUSPEND policy in the configuration.

http://www.cs.wisc.edu/condor/manual/v7.3/3_5Policy_Configuration.html

If you just want your jobs to run once they've started, try looking for
SUSPEND = False.

Best,


matt