Hi everyone,
what is the best way to get information on the resource usage of every node in a cluster?
Using ray status only gives an overall information on the cluster, but no information on how many resources are on each node and how many used on each node.
This information is really useful for getting information on why a job is stuck (in case of multiple resources/custom Resources).
A summed up view does not help here, because if the one resource is available on node x and the second resource is available on node x the sum shows both resources available but a task requireing both resources will never be able to start.