[slurm-users] CPU allocation within a node is not cyclic

[Ole Holm Nielsen]

On 06-10-2018 04:15, 崔灏 (CUI Hao) wrote:
> $ scontrol reconfigure
> slurm_reconfigure error: SelectType change requires restart of the
> slurmctld daemon to take effect
> 
> I'm afraid that restarting slurmctld will interrupt current tasks, so
> I'm still waiting for them to finish.

There should be no problem in restarting slurmctld.  It will only take a 
few seconds, and it will not disturb slurmd daemons on compute nodes or 
the running jobs, but you must check SlurmctldTimeout in slurm.conf by:

# scontrol show config | grep Timeout

Of course, you must first synchronize the slurm.conf file to all nodes 
in the cluster.

Perhaps my Wiki page about upgrading slurmctld will give you some 
further information:
https://wiki.fysik.dtu.dk/niflheim/Slurm_installation#upgrade-slurmctld

/Ole