[slurm-users] Include some cores of the head node to a partition

[Chris Samuel]

On Sunday, 29 April 2018 6:57:58 PM AEST Mahmood Naderan wrote:

> [root at rocks7 ~]#  scontrol show config | fgrep -i rocks7

Ah, I'd forgotten that wouldn't list the NodeName lines from your config file. 
Sorry.


> Chris,
> Regarding this section
> 
> NodeName=DEFAULT State=UNKNOWN
> NodeName=rocks7 NodeAddr=10.1.1.1

OK, so there you are not telling Slurm anything about your hardware. As you 
just want 20 cores for rocks7 then you probably want to add this to your 
NodeName line:

CPUs=20 Boards=1 SocketsPerBoard=4 CoresPerSocket=5 ThreadsPerCore=1

I've never tried to do this though so it might not be optimal.

> Should I add something like
> 
> PartitionName=OTHERS AllocNodes=rocks7 State=UP MaxCPUsPerNode=20
> 
> or even 32?

No, it looks like Slurm hasn't got the right config info in it for the node.

Try updating your config and restarting slurmctld and slurmd everywhere and see 
where that gets you.   If it's still drained but the hardware config looks good 
in Slurm then you can do "scontrol update node=rocks7 state=resume" to tell 
Slurm to try using it again.

All the best,
Chris
-- 
 Chris Samuel  :  http://www.csamuel.org/  :  Melbourne, VIC