Kubernetes Pod Scheduling Mental Model Drill
A 4-question reference set on how Kubernetes places pods: filter-then-score, requests vs limits, affinity and anti-affinity, and the taint/toleration machinery the kubelet uses to signal node health.
Question Bundle
Go
kubernetes-hpa
distributed-systems
backend
interview-prep
By CodeSnatch
May 5, 2026
·
Updated August 10, 2026
471 views
5
4.3 (12)
The Kubernetes scheduler picks a node for each Pending pod via filter-then-score. Walk through what each phase does and which built-in predicate causes the most surprise in production.
Examples
Example 1:
Input: A pod requests 4 CPU on a cluster where every node has 8 CPU but 6 already requested
Output: Filter phase rejects all nodes; pod stays Pending
Explanation: NodeResourcesFit checks requests against allocatable, not against actual usage.Example 2:
Input: A pod has no nodeSelector or affinity rules but a taint exists on every healthy node
Output: Filter phase removes every node; pod stays Pending
Explanation: TaintToleration is the predicate that catches most teams; tolerations must be on the pod, not the node.3 more questions and all solutions are locked.
Purchase this item to access all questions, code snippets, and solutions.