[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [HTCondor-users] Possible for user to limit number of jobs per physical machine?
- Date: Fri, 11 Sep 2020 08:34:42 +0200 (CEST)
- From: "Beyer, Christoph" <christoph.beyer@xxxxxxx>
- Subject: Re: [HTCondor-users] Possible for user to limit number of jobs per physical machine?
OK, found your first post now :)
could you not just put a number field in the job like 'max_jobs_per_node=<number>'
And then put into the start expression on the workernode max_jobs_per_node > number_of_running_jobs_user (you need the running jobs per user fields in the host classadd maybe) ... ?
Hope I got you right this time ;)
Building 02b, Room 009
----- UrsprÃngliche Mail -----
Von: "Carsten Aulbert" <carsten.aulbert@xxxxxxxxxx>
An: "htcondor-users" <htcondor-users@xxxxxxxxxxx>, "Christoph Beyer" <christoph.beyer@xxxxxxx>
Gesendet: Freitag, 11. September 2020 08:23:25
Betreff: Re: [HTCondor-users] Possible for user to limit number of jobs per physical machine?
Good morning Christoph,
On 9/11/20 8:11 AM, Beyer, Christoph wrote:
> I think one way would be to use startd_cron to check periodically how many jobs of each user are currently running on the node.
> Export the values in the node-classadd for ex 'running_jobs_user_xy' and alter the start configuration of your slot accordingly to not start any jobs of user xy if running_jobs_user_xy > number ...
> Couple of lines in the config but not very complicated ...
yeah, but that information is already present in the main slot machine
class ad ("ChildRemoteUser") without the delay of startd_cron. The
question remains if that were be a target-able attribute from the users'
submit files honored/used both by the negotiator and starter.
I will keep testing :)
Dr. Carsten Aulbert, Max Planck Institute for Gravitational Physics,
CallinstraÃe 38, 30167 Hannover, Germany
Phone: +49 511 762 17185