Hello list

A few weeks ago I was pointed to this new Elastic Computing feature,
that I may use in my project, which consist on SLURM asking to launch
some VM nodes through the power saving features to process excess
workload.


My question is how SLURM decides how many extra VMs to request and
when does it asks for them?


I have a module that checks the CPU usage of the nodes. Currently I
use that one to migrate VMs from the overloaded nodes to the wasted
ones (balancing the load among them).

I'd like to use part of that balancing logic to tell how many more VMs
I want to allow running, so for example if I have 4 underutilized
nodes, I would allow to start up to 4 extra VMs to fill that idle
time, but if no such node exists then shutdown some (extra) VMs.

I'm looking forward for your comments.
Ismael


-- 
Do not let me induce you to satisfy my curiosity, from an expectation,
that I shall gratify yours. What I may judge proper to conceal, does
not concern myself alone.

Reply via email to