[slurm-dev] Re: slurm-dev Re: knl_generic plugin on non-KNL node

[Victor Gamayunov]

Hi Pierre-Marie,

On Thu, Aug 10, 2017 at 5:27 PM, Le Biot, Pierre-Marie
 wrote:
> >Is there a way do selectively disable the plugin on non-KNL nodes?
>
> non-KNL nodes will be assigned dummy features (xyz is mine) but won't be 
> rebooted when a constraint is specified by a job
>
> scontrol show node xx
> [...]
>AvailableFeatures=a2a,hemi,quad,snc2,snc4,cache,flat,hybrid,auto,xyz
>ActiveFeatures=a2a,cache,xyz
> [...]
>
> Best Regards,
> Pierre-Marie Le Biot

When I do this, therebooting stops, but the plugin seems to remove all
custom features specified in slurm.conf (NodeName= Feature=xyz),
and leaves only KNL features:

scontrol show node 
   AvailableFeatures=cache,hybrid,flat,auto,a2a,snc2,snc4,hemi,quad
   ActiveFeatures=cache,quad

How do you specify "xyz"?

Thanks
Victor

[slurm-dev] Re: knl_generic plugin on non-KNL node

[Victor Gamayunov]

Hi Gilles,

On Fri, Jul 7, 2017 at 1:16 AM Gilles Gouaillardet 
wrote:

> in your slurm.conf, you should have a line like this one
> NodeName=n[1-4] Feature=knl Sockets=1 CoresPerSocket=68 State=UNKNOWN
> at first, make sure your regular Xeon nodes do *not* have the 'knl' feature
>

I do have that, but it doesn't seem to make any difference.

i guess an other option is not to have the
> NodeFeaturesPlugins=knl_generic
> line on your regular Xeon nodes
> (note that unless you specify an option, you will get some warnings
> since all your slurm.conf are not identical)
>

I was thinking about doing that, but I didn't like the idea of having
different conf files. I'll try this anyway.

Am I right thinking that the decision to reboot the node is made based
on the bitmask which is modified by the plugin?

Thanks,
Victor

[slurm-dev] knl_generic plugin on non-KNL node

[Victor Gamayunov]

Hi,

i have a cluster with a mix of regular Xeon and KNL nodes. I use
knl_generic to switch KNL modes which works very well.
However, there is a side effect on non-KNL nodes: every time I allocate a
non-KNL node and specify a constraint (which has nothing to do with KNL),
the node is rebooted every time.

Is there a way do selectively disable the plugin on non-KNL nodes?

I looked at the code but couldn't quite figure out what forces the reboot.

Thanks
Victor