ReplicaSets: Desired State and Self-Healing Mechanics

Delete a pod from a Deployment and it reappears within seconds. The Deployment doesn't do this directly. The ReplicaSet beneath it does. Every Deployment creates and manages ReplicaSets, and every ReplicaSet runs a continuous reconciliation loop that compares what you asked for against what exists.

The CKA tests your understanding of these workload primitives. ReplicaSets are the engine that Deployments depend on. Understanding how the controller works explains why pods come back after deletion, how scale-down picks which pods to kill, what happens when labels drift, and why stale ReplicaSets accumulate after rollouts.