[slurm-users] User limits for multiple associated accounts

[Mahmood Naderan]

Excuse me... I see the output of squeue which says
170   IACTIVE     bash  mahmood PD       0:00      1 (AssocGrpMemLimit)

I don't understand why the memory limit is reach? I can not see the
memory usage of a running job from sacct commands. However, using
"top" on the compute node, I see 6 cores each uses 400MB. So it is
below 8G which defined for the user.


Regards,
Mahmood




On Fri, May 11, 2018 at 4:20 PM, Mahmood Naderan <mahmood.nt at gmail.com> wrote:
> Hi
> I have added a user to multiple partitions. That account name actually
> corresponds to a set of limitations which I define for a user.
>
> [root at rocks7 ~]# sacctmgr list association
> format=partition,account,user,grptres,maxwall
>  Partition    Account       User       GrpTRES     MaxWall
> ---------- ---------- ---------- ------------- -----------
>                  root
>                  root       root
>                   em1
>    iactive        em1    mahmood  cpu=6,mem=8G 30-00:00:00
>    plan1        em1    mahmood  cpu=6,mem=8G 30-00:00:00
>               monthly
>    plan2    monthly    mahmood cpu=32,mem=6+ 30-00:00:00
> [root at rocks7 ~]# squeue -j 167
>              JOBID PARTITION     NAME     USER ST       TIME  NODES
> NODELIST(REASON)
>                167   PLAN1     test  mahmood  R    5:58:41      1 compute-0-3
> [root at rocks7 ~]# squeue -j 167 -o %C
> CPUS
> 6
>
>
> As you see the user is running a job with the maximum core counts
> allowed. Now, if I run
>
> [mahmood at rocks7 Downloads]$ salloc -p IACTIVE -A em1
> salloc: Pending job allocation 170
> salloc: job 170 queued and waiting for resources
>
> Which is pending for resources. I want to be sure that the pending is
> REALLY related to reaching the maximum tres limits and NOT a
> configuration problem.
>
> Is that OK? Hope that I asked my question correctly ;)
>
>
> Regards,
> Mahmood