Ticket 2047

Summary: Display job layout in sacct
Product: Slurm Reporter: Kilian Cavalotti <kilian>
Component: AccountingAssignee: Unassigned Developer <dev-unassigned>
Status: OPEN --- QA Contact:
Severity: 5 - Enhancement    
Priority: --- CC: jdesjard, sergey_meirovich, tim, vasev
Version: 14.11.8   
Hardware: Linux   
OS: Linux   
Site: Stanford Alineos Sites: ---
Atos/Eviden Sites: --- Confidential Site: ---
Coreweave sites: --- Cray Sites: ---
DS9 clusters: --- 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 Kilian Cavalotti 2015-10-21 06:29:02 MDT 
Hi,

I was wondering if there was a way to get the job layout (ie. which CPUs were assigned on which node) in sacct, the way you can display get it with "scontrol -d show job":

   Socks/Node=* NtasksPerN:B:S:C=0:0:*:* CoreSpec=*
     Nodes=sh-5-33 CPU_IDs=0-15 Mem=64000
     Nodes=sh-5-34 CPU_IDs=0-6,8-14 Mem=56000

I tried sacct --format=Layout, but it just seems to display the distribution method (like "Block" or "Cyclic").

Thanks!
--
Kilian
Comment 1 Pavel A Vasev 2019-02-11 23:45:10 MST 
Hi Kilian! Have you somehow achieved this?
Comment 2 Kilian Cavalotti 2019-02-12 08:45:07 MST 
Hi Pavel, 

(In reply to Pavel A Vasev from comment #1)
> Hi Kilian! Have you somehow achieved this?

I haven't looked in a while, but I don't think this has been implemented. 

Our current approach is to dump the contents of "scontrol -d show job" in a log file during PrologSlurmctld execution, but it's not really recommended, because it could put extra load on the Slurm controller.

Cheers,
-- 
Kilian
Comment 3 Pavel A Vasev 2019-02-12 09:46:56 MST 
Thank you very much for your answer, Kilian!
Comment 4 Jason Booth 2019-03-04 10:13:00 MST 
*** Ticket 6406 has been marked as a duplicate of this ticket. ***
Comment 5 Jason Booth 2019-03-04 14:40:16 MST 
*** Ticket 6626 has been marked as a duplicate of this ticket. ***