[slurm-users] Can Slurm schedule around resources shared between nodes?

[Conrad Herrmann]

I know that Slurm can schedule jobs to efficiently use resources that are
available on a single node by sharing those resources between multiple
jobs.  Examples include CPU codes, memory, and GPUs.

I'm trying to figure out if Slurm can schedule jobs around limited
resources that are shared by multiple nodes (or the whole partition)?
Examples might include network bandwidth or shared disk storage space.  In
both of these cases, I'm thinking the scheduler might delay jobs that would
consume too much of the resource.

Thanks in advance,
*Conrad Herrmann*
*Zoox*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.schedmd.com/pipermail/slurm-users/attachments/20190903/e4334ba2/attachment.htm>