Slurm Power Saving Guide: SuspendProgram not called

Viewed 32

Following the tutorial https://slurm.schedmd.com/power_save.html I created two shell scripts /home/ubuntu/resume.sh and /home/ubuntu/suspend.sh as given by the guide. I then added

# Power Saving
SuspendTime = 5
SuspendProgram=/home/ubuntu/suspend.sh
ResumeProgram=/home/ubuntu/resume.sh

to my /etc/slurm/slurm.conf and called sudo scontrol reconfig, sudo /etc/init.d/slurmctld restart (which ran without errors). No job is running on the cluster, but nothing gets logged. So I threw another sudo scontrol reconfig in.. just in case, but still nothing gets logged.

What I've expected to happen: 0. 5 seconds pass... I. Log of the echo II. Shutdown of the nodes.

What happened: Nothing. Neither did any nodes shutdown nor has a /var/log/power_save.log been created.

I am unsure why the suspend script is not called properly.

Permissions

Since slurm needs permission to call the scripts, I executed chmod a+x for both files to make sure they can be executed. This didn't fix it, because the log tells you that no world permission is allowed. I was already able to work myself around this, but need a better general understanding before being able to answer this question myself fully since I am not entirely sure what did the trick.

0 Answers
Related