Pods, ReplicaSets, and Deployments

Declare the desired state and let the loop converge

A pod is one or more containers scheduled together, sharing a network namespace. It is the smallest thing the scheduler places, and on its own it is fragile — nothing recreates a pod that dies.

A deployment is what you actually write. It owns a replica set, which owns the pods, and a controller continuously works to make reality match the spec you declared.

That reconciliation model explains most of Kubernetes behaviour. You do not tell it to start a container; you declare that three should exist, and something keeps checking.