You are correct. Today there is no way to change those after the slurmctld daemon starts, although that could _probably_ be permitted when a node first goes from state FUTURE to IDLE.

Quoting Jonathan Stoppani <[email protected]>:

The hostname and address can't be updated using scontrol update:

"Update of this parameter is not supported: NodeHostname=testing"

Are there other ways to update it?


P.S. Sorry, I actually wanted to reply to the list ;-)

On Jul 18, 2011, at 14:47 , [email protected] wrote:

The actual hostname and address and name that slurm uses can also be different.
See NodeName, NodeHostname, and NodeAddr in man slurm.conf.

Quoting Jonathan Stoppani <[email protected]>:

Thanks, this is a good pointer.

I think though that I will have to reconfigure the controller anyways, as I don't know the hostname of a given node until the allocation occurs.

On Jul 18, 2011, at 14:14 , [email protected] wrote:

There is a node state of "FUTURE" that may satisfy your needs. This is from "man slurm.conf":

FUTURE Indicates the node is defined for future use and need not exist when the SLURM daemons are started. These nodes can be made available for use simply by updating the node state using the scontrol command rather than restarting the slurmctld daemon. After these nodes are made available, change their State in the slurm.conf file. Until these nodes are made available, they will not be seen using any SLURM commands or Is nor will any attempt be made to contact them.


Quoting Jonathan Stoppani <[email protected]>:

Hi,

I'm developing a virtualization layer to dynamically feed virtual nodes to SLURM, based on user requests. These nodes can be virtual machines running on physical nodes of a cluster or come from cloud computing service providers or what else you can imagine....

Once the node (running slurmd) is up and running, I add the relevant configuration entry to the SLURM configuration file and issue an "scontrol reconfigure" command for the controller to pick it up.

Are there better ways to dynamically manage nodes? Are there reconfiguration consequences I should be aware of?

Thanks,
Jonathan
















Reply via email to