Ticket 6855

Summary:	sdiag manual page references MAX_AGENT_CNT, replace with MAX_SERVER_THREADS?
Product:	Slurm	Reporter:	Chris Samuel (NERSC) <csamuel>
Component:	Documentation	Assignee:	Alejandro Sanchez <alex>
Status:	RESOLVED FIXED	QA Contact:	Tim Wickberg <tim>
Severity:	4 - Minor Issue
Priority:	---
Version:	18.08.6
Hardware:	Linux
OS:	Linux
Site:	NERSC	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:	---
Machine Name:		CLE Version:
Version Fixed:	19.05.2 20.02.0pre1	Target Release:	---
DevPrio:	---	Emory-Cloud Sites:	---
Attachments:	Patch to fix reference to MAX_AGENT_CNT in sdiag man page

Description Chris Samuel (NERSC) 2019-04-12 17:15:30 MDT

Created attachment 9900 [details]
Patch to fix reference to MAX_AGENT_CNT in sdiag man page

Hi there,

When we were trying to chase down this mornings outage on Cori I noticed that the sdiag manual page says:

 Agent queue size
    [...] If this values is close to MAX_AGENT_CNT there could be some delays affecting jobs management.

It appears that commit 53534f4907c0333696d2a04046c52a92a5e39c40 removed MAX_AGENT_CNT and replaced its use with MAX_SERVER_THREAD back in 2015.

I'm guessing the gist of the paragraph is still correct, so I'll attach
a patch to just swap the preprocessor marco name over.

All the best,
Chris

Comment 2 Jason Booth 2019-05-01 16:38:51 MDT

Hi Chris,

 Thank you for the patch. We will review and let you know if anything else will be changed.

Comment 4 Alejandro Sanchez 2019-07-01 03:58:55 MDT

Hi Chris,

While MAX_AGENT_CNT is indeed deprecated and the man page needs to be amended, I think your contribution is not correct. I've triggered the review process for a different patch. We'll keep you updated.

Thanks.

Comment 5 Chris Samuel (NERSC) 2019-07-01 09:39:36 MDT

(In reply to Alejandro Sanchez from comment #4)

> Hi Chris,

Hiya!

> While MAX_AGENT_CNT is indeed deprecated and the man page needs to be
> amended, I think your contribution is not correct. I've triggered the review
> process for a different patch. We'll keep you updated.

Not a problem!  Thanks for this.

All the best,
Chris

Comment 8 Alejandro Sanchez 2019-07-26 09:33:25 MDT

Hi Chris,

sdiag docs have been clarified in the following commit:

https://github.com/SchedMD/slurm/commit/cbfb66807416df273f2fe7f43a23fb0511a8dae0

In 20.02 I've also exposed the number of agent threads to sdiag as a new stat.

I'm closing this bug. Plese, reopen if you have further questions. Thanks.

Comment 9 Chris Samuel (NERSC) 2019-07-26 13:07:25 MDT

Thanks Alejandro!