[slurm-users] Requested partition configuration not available now

[Werner Saar]

Hi Mahmood,

this question is related to the slurm-roll.
The command rocks sync slurm has more tasks:

1. Rebuild of 411 is forced
2. on compute nodes, the command /etc/slurm/slurm-prep.sh start is executed
3. on compute nodes, slurmd is restarted
4. slurmctld is restarted.

Step 1 and 2 are required, to sync the config files.

Best regards

Werner



On 05/16/2018 09:12 AM, Mahmood Naderan wrote:
> Hi,
> After creating an account and a partition, I get an error that
> requested partition configuration not available now. Although I
> restarted the services on all nodes, I wonder why that happen?
>
> [root at rocks7 ~]# rocks run host compute-0-0 "systemctl restart slurmd"
> [root at rocks7 ~]# rocks run host compute-0-1 "systemctl restart slurmd"
> [root at rocks7 ~]# systemctl restart slurmd
> [root at rocks7 ~]# systemctl restart slurmctld
> [mahmood at rocks7 ~]$ sacctmgr list association
> format=partition,account,user | grep mahmood
>     emerald        em1    mahmood
>     diamond    monthly    mahmood
>        ruby         y8    mahmood
> [mahmood at rocks7 ~]$ srun -I -A monthly -p DIAMOND --mem=4GB --pty bash
> [mahmood at compute-0-1 ~]$ exit
> exit
> [mahmood at rocks7 ~]$ srun -I -A em1 -p EMERALD --mem=4GB --pty bash
> [mahmood at rocks7 ~]$ exit
> exit
> [mahmood at rocks7 ~]$ srun -I -A Y8 -p RUBY --mem=4GB --pty bash
> srun: error: Unable to allocate resources: Requested partition
> configuration not available now
> [mahmood at rocks7 ~]$ cat /etc/slurm/parts
> PartitionName=WHEEL RootOnly=yes Priority=1000 Nodes=ALL
> PartitionName=DIAMOND AllowAccounts=monthly Nodes=compute-0-[0-1]
> PartitionName=EMERALD AllowAccounts=em1,em4 Nodes=compute-0-[0-1],rocks7
> PartitionName=RUBY AllowAccounts=Y8 Nodes=compute-0-[0-1]
>
>
>
> Regards,
> Mahmood
>