Part 3 of the series on the Kubernetes API delves into the complexities and dynamics of the Deployment resource, exploring how changes are rolled out over ReplicaSets, which manage the replication of Pods within a cluster. The post employs kubespy, a tool for observing Kubernetes resources in real-time, to elucidate the process of Deployment rollouts, including how the system adapts when a Pod crashes, is killed, or is relabeled. By using kubespy trace, readers can follow the lifecycle of a Deployment and understand the interplay between Deployment, ReplicaSet controllers, and the Kubelet when triggered by various events. The post also provides practical examples and encourages experimentation using the kubespy repository and outlines the effects of modifying Pod labels or deleting Pods. It highlights the potential of the Kubernetes API to provide valuable insights into cluster states and encourages readers to explore these functionalities using tools like kubespy and Pulumi CLI.