Although we can currently start / stop sack services on our multi-node test cluster, these services are distributed to whichever nodes are available, regardless of their label for ndslabs-role. This includes storage or etcd nodes, as well as compute nodes.

This prevents us from separating our allocated resources by function, and potentially causing resource contention on critical nodes under heavy load.

This ticket is complete when the NDS Labs stack services should instead only be scheduled on nodes where ndslabs-role is compute.