To schedule shutdown of HPC, one has to free nodes from jobs. If shutdown is close than it is easier to drain the nodes
scontrol update node=zeus[20-91,100-171,200-217,15] state=drain reason="Planned shutdown"
However there may be jobs still running at the shutdown time.
If the shutdown is in the future, when you are sure no jobs from the present are still be running, you can put a reservation in slurm for the required time and nodes.
To reserve the nodes to be free of jobs in required time, i.e. create reservation (duration is in min by default), e.g for 12 hours:
scontrol create reservation starttime=2009-02-06T16:00:00 duration=720 user=root flags=maint,ignore_jobs nodes=zeus[20-91,100-171,15,200-217]
or ...nodes=ALL
0 Comments.