Ticket 15143 - configuration hash
Summary: configuration hash
Status: RESOLVED INFOGIVEN
Alias: None
Product: Slurm
Classification: Unclassified
Component: Configuration (show other tickets)
Version: 22.05.3
Hardware: Linux Linux
: 4 - Minor Issue
Assignee: Carlos Tripiana Montes
QA Contact:
URL:
Depends on:
Blocks:
 
Reported: 2022-10-11 02:31 MDT by Yann
Modified: 2022-10-18 07:06 MDT (History)
1 user (show)

See Also:
Site: Université de Genève
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: ---

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this ticket.
Description Yann 2022-10-11 02:31:58 MDT 
Hi,

I was wondering what are the files included in the conf hash check?

I guess the file acct_gather.conf isn't included as it seems impossible to have the same file for every node as the file depends for example on ipmi sensors. Or is there a way to list all the nodes in this file so we can ensure the file is identical on the whole cluster as well, like we do for the gres.conf file.

Best

Yann
Comment 1 Carlos Tripiana Montes 2022-10-11 03:53:49 MDT 
Hi Yann,

> I was wondering what are the files included in the conf hash check?

Only slurm.conf.

> I guess the file acct_gather.conf isn't included as it seems impossible to
> have the same file for every node as the file depends for example on ipmi
> sensors. Or is there a way to list all the nodes in this file so we can
> ensure the file is identical on the whole cluster as well, like we do for
> the gres.conf file.

acct_gather.conf is allowed to be different for each node, to adapt it to the underlying HW, as you have mentioned.

We don't have the need to specify all nodes in a single acct_gather.conf as slurmctld doesn't really need to know the exact underlying plugin the node is using to grab the metrics. gres.conf needs to be a general one with all nodes in it because slurmctld needs to know all the cluster GRES, so it can properly schedule the jobs.

I'm directly marking the bug as info given but if you need further assistance please reopen it and tell us your thoughts.

Regards,
Carlos.
Comment 2 Yann 2022-10-18 07:06:40 MDT 
Thanks for the information

Best

Yann