Question

A platform engineer is explaining to a new hire how the kube-scheduler decides where to place a newly created Pod that has no nodeName set. The Pod requests 500m CPU, and only three of the cluster's ten nodes have enough allocatable CPU remaining. How does the scheduler ultimately choose one node for the Pod?

Accepted Answer

It first filters out nodes that cannot run the Pod, then scores the feasible nodes and binds the Pod to the highest-scoring one

Answer

It picks the first node returned by the API server in alphabetical order, ignoring resource availability

Answer

It randomly selects any of the ten nodes and lets the kubelet reject the Pod if resources are insufficient

Answer

It always chooses the node with the most total CPU capacity regardless of how much is already in use