[slurm-users] Slurm showing 100% utilization since last maintenance window
Prentice Bisbal
pbisbal at pppl.gov
Thu Jul 11 19:18:04 UTC 2019
I have a strange issue:
sreport is showing 100% utilization for our cluster every day since June
18. What is interesting about this is June 18th was our last maintenance
outage, when all the nodes were rebooted, including our slurm server
which runs both slurmdbd and slurmctld. Has anyone else seen this, or is
aware of this issue?
I can't remember if we updated the version of Slurm we're using at that
time. The version of slurm in use is right now is 18.08.7
Typically, our monthly usage varies between 55-65%, but because of this
error, June is at 87%, and we're on schedule for 100% usage for July.
Sinfo shows there are some idle nodes right now. It's pretty rare that
our cluster is actually at 100% utilization, so these numbers are
definitely not correct.
--
Prentice
More information about the slurm-users
mailing list