[slurm-users] sbatch overallocation

[Diego Zuccato]

Il 10/10/20 18:53, Renfro, Michael ha scritto:

>   * Do you want to ensure that one job requesting 9 tasks (and 1 CPU per
>     task) can’t overstep its reservation and take resources away from
>     other jobs on those nodes? Cgroups [1] should be able to confine the
>     job to its 9 CPUs, and even if 8 processes get started at once in
>     the job, they’ll only drive up the nodes’ load average, and not
>     affect others’ performance.
IMHO cgroups is a must-have: each job is guaranteed to receive what it
asks for, and no more. If it tries to use more, it just self-contends
resources, w/o impacting other jobs.
Configuring it greatly reduced headaches on our cluster :)

-- 
Diego Zuccato
DIFA - Dip. di Fisica e Astronomia
Servizi Informatici
Alma Mater Studiorum - Università di Bologna
V.le Berti-Pichat 6/2 - 40127 Bologna - Italy
tel.: +39 051 20 95786