Dear Ana:

> One thing I still do not have clear is how adding/removing nodes from
> the cluster affects SLURM (the queue, the state information, etc.) and
> jobs already running. Is it equivalent to "scontrol reconfigure" where
> you have to restart slurmctld every time you add or remove a node?

Yes, ElastiCluster writes the new config file and then restarts the
SLURM daemons.

> What if we want to do these changes of nodes frequently, how does this
> affect the users?

They should not notice, except for the occasional glitch while
`slurmctld` is restarted.  I'll admit that this has not received much
testing, though (OTOH, neither have I received any bug reports on this).

Note that, when scaling down a cluster, you should (1) set nodes you
want to remove in "DRAIN" state, (2) use `elasticluster remove-node` to
remove them.  (`elasticluster resize -r` removes the nodes immediately
starting with the highest-numbered ones, regardless of whether they are
running any jobs.)

Ciao,
R

-- 
Riccardo Murri, Schwerzenbacherstrasse 2, CH-8606 Nänikon, Switzerland

-- 
You received this message because you are subscribed to the Google Groups 
"elasticluster" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to