Ticket 586

Summary: GrpSubmit exceeded with job arrays
Product: Slurm Reporter: Ryan Cox <ryan_cox>
Component: LimitsAssignee: Moe Jette <jette>
Status: RESOLVED FIXED QA Contact:
Severity: 4 - Minor Issue    
Priority: --- CC: da
Version: 2.6.x   
Hardware: Linux   
OS: Linux   
Site: BYU - Brigham Young University 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: 2.6.6
Target Release: --- DevPrio: ---
Emory-Cloud Sites: ---
Attachments: grpsubmit exceeded

Description Ryan Cox 2014-02-07 02:43:23 MST 
Created attachment 627 [details]
grpsubmit exceeded

It seems that users can get around GrpSubmit limits due to job arrays.  I attached a file showing a failed attempt to exceed the job limit first without job arrays then a successful attempt with job arrays.  We're on 2.6.4 8fba61066b30a4874632c59b770b9ed46fb5adf1.  We have seen other accounts do this as well, not just staff.
Comment 1 Moe Jette 2014-02-07 04:00:55 MST 
I have been able to reproduce this problem. There are use cases that definitely fail.
Comment 2 Moe Jette 2014-02-07 04:18:51 MST 
Fixed with one line change. This will be in version 2.6.6 when released, probably within a day or two. The commit with the fix is shown below. Append ".patch" to the URL to generate a patch file.

https://github.com/SchedMD/slurm/commit/9469053d55233c208e58dedff5a9fae5ce29f0a2
Comment 3 Ryan Cox 2014-02-07 04:22:04 MST 
Thanks for the quick fix.  We'll upgrade when 2.6.6 is released.