Kubernetes · Follow the flow

How Kubernetes starts an app

A Deployment request leads to Pod creation, node selection, and running containers.

How Kubernetes starts an appA Deployment request leads to Pod creation, node selection, and running containers.STEP 1You describe the appDeployment: three copies.Send it with kubectl.STEP 2API server + etcdCheck the request.Save accepted cluster state.STEP 3ControllersDeployment → ReplicaSet.ReplicaSet creates Pods.STEP 4SchedulerFind a suitable nodefor each waiting Pod.STEP 5Kubelet + runtimePrepare volumes and network.Pull images; start containers.STEP 6Running, then ReadyApplication starts.Readiness allows traffic.Read the numbered stages. Press Play to follow them.Arrows show the order of this explanation, not every network connection.

Play follows each stage once. Previous and Next let you study one stage. Keyboard: left/right to step; Space to play or pause; Home/End to jump. Swipe inside the diagram on phones.

Complete diagram shown. Enable playback to follow each stage.

These are ordered actions. Controllers, the scheduler, and kubelets communicate with the API server; this is not a packet path through those components. Storage requirements can influence scheduling. The API server writes cluster state to etcd.