hi there,

I want to learn how you guys will do the configuration for such a case:

suppose I have a storage cluster running Ceph or Moosefs/Lizardfs, but the
CPU and memory usage on them are low. I want to scavenge the resources on
these resources and let users submit jobs to these nodes. Obviously I need
to limit the total amount of resources the batch jobs can consume, e.g, 70%
of the total CPU, memory and IO bandwidth, and so that Ceph can always have
30% of the resources.

I'm thinking of using cgroups plugin and list only 70% of the resource in
the node resource definition. Is that in the right direction?

Best Regard,
Chu

Reply via email to