Hello all,
Yes, having multiple types of compute nodes in a cluster is definitely
supported. (You can also add different compute nodes to an existing
cluster, provided you edit the config file before adding.) I think the
config file posted by Champak solves the original question?
As to why jobs don't run on the front-end node: this is intentional.
The front-end node is running a number of services which are key to
the functioning of the whole cluster (SSH login, SLURM controller, NFS
server, etc) so you don't want it to be (possibly) taken down by a job
gone wild.
That said, if you really want jobs to run on the front-end node as
well, you can do it by adding the `slurm_worker` Ansible group to
`frontend_groups=` setting in the `[setup/...]` section. E.g.,
[setup/slurm-with-large-frontend]
frontend_groups=slurm_master,slurm_worker
compute_groups=slurm_worker
# ...everything else as usual...
Ciao,
R
--
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.
