Ticket 16641

Summary:	[NNL] power management plugin for intel
Product:	Slurm	Reporter:	Eric Hausgaard <eric.hausgaard>
Component:	Configuration	Assignee:	Ben Roberts <ben>
Status:	RESOLVED INFOGIVEN	QA Contact:
Severity:	4 - Minor Issue
Priority:	---
Version:	23.02.1
Hardware:	Linux
OS:	Linux
Site:	UNNPP	Alineos Sites:	---
Atos/Eviden Sites:	---	Confidential Site:	---
Coreweave sites:	---	Cray Sites:	---
DS9 clusters:	---	HPCnow Sites:	---
HPE Sites:	---	IBM Sites:	---
NOAA SIte:	---	OCF Sites:	---
Recursion Pharma Sites:	---	SFW Sites:	---
SNIC sites:	---	Linux Distro:	RHEL
Machine Name:	NNL	CLE Version:
Version Fixed:		Target Release:	---
DevPrio:	---	Emory-Cloud Sites:	---

Description Eric Hausgaard 2023-05-03 08:54:16 MDT

I reviewed https://slurm.schedmd.com/power_mgmt.html and it indicates there is only a plugin for cray systems.  is there a power plugin for intel based CPUs and IPMI?

Comment 1 Ben Roberts 2023-05-03 09:30:15 MDT

Hi Eric,

We do have the ability to power up and down nodes with power management scripts at different times.  I think the page of the documentation that would be more useful to you is the Slurm Power Saving Guide:
https://slurm.schedmd.com/power_save.html

The SuspendProgram and ResumeProgram allow you to write your own scripts that can make calls to IPMI to control the nodes.  This page should give you a good starting point, but if you have any questions about anything please let me know.

Thanks,
Ben

Comment 2 Eric Hausgaard 2023-05-03 10:56:05 MDT

I think I was looking for this instead:
https://slurm.schedmd.com/slurm.conf.html#OPT_AcctGatherEnergyType

Comment 3 Ben Roberts 2023-05-03 13:28:29 MDT

Ah, sorry I misunderstood what you were looking for.  I thought you were looking for a way to power nodes on and off rather than collecting statistics about their power usage.  I'm glad you found what you wanted.  Let me know if there's anything else I can do to help or if this ticket is ok to close.

Thanks,
Ben

Comment 4 Eric Hausgaard 2023-05-04 05:15:13 MDT

I was looking for power statistics,  but having the ability to shut down a node with IPMI would be useful. Not that I think I'd be able to use it in production.

Comment 5 Ben Roberts 2023-05-04 08:59:19 MDT

Hi Eric,

Powering on and off nodes is an option if you ever think it might be feasible in your environment.  I'll go ahead and close this ticket for now.  Let us know if there is anything else we can do to help.

Thanks,
Ben