Ticket 3628

Summary:	Method to power cycle KNLs for memory mode changes
Product:	Slurm	Reporter:	John Westlund <john>
Component:	KNL	Assignee:	Jacob Jenson <jacob>
Status:	RESOLVED TIMEDOUT	QA Contact:
Severity:	3 - Medium Impact
Priority:	---
Version:	16.05.8
Hardware:	Linux
OS:	Linux
Site:	Intel CRT	Slinky Site:	---
Alineos Sites:	---	Atos/Eviden Sites:	---
Confidential Site:	---	Coreweave sites:	---
Cray Sites:	---	DS9 clusters:	---
Google sites:	---	HPCnow Sites:	---
HPE Sites:	---	IBM Sites:	---
NOAA SIte:	---	NoveTech Sites:	---
Nvidia HWinf-CS Sites:	---	OCF Sites:	---
Recursion Pharma Sites:	---	SFW Sites:	---
SNIC sites:	---	Tzag Elita Sites:	---
Linux Distro:	---	Machine Name:
CLE Version:		Version Fixed:
Target Release:	---	DevPrio:	---
Emory-Cloud Sites:	---

Description John Westlund 2017-03-28 15:20:22 MDT

I have a customer that is trying to change memory modes on KNL hardware that appears to not support changing memory modes via reboot -- the memory mode can only be changed via a power cycle.

All documentation I see references running reboot on the compute node as part of the KNL Slurm configuration. I'm asking them to try running an ipmi power chassis reset locally on the client as the reboot command -- I'm not sure if that will work.

Is there a method of issuing a power control command on the system hosting the slurmctld when the user requests to change memory modes?

Thanks

Comment 1 Tim Wickberg 2017-03-28 15:40:36 MDT

Downgrading the severity here - please see https://www.schedmd.com/support.php for details on severity levels.

Unfortunately I need to hand you off to Jacob for a bit, as I don't believe we have any support contracts in place at present; until he gives me the go ahead I can't directly engage on questions.

- Tim

Comment 2 John Westlund 2017-03-28 15:45:19 MDT

Thanks Tim. I'll share the severity level definitions with the people on this end requesting it be Sev1

Comment 5 Jacob Jenson 2017-12-12 10:36:48 MST

Once a support contract is in place for this system then the Slurm support team can engage.