What is the control plane of Kubernetes responsible for, and what do worker nodes do?

The control plane manages the cluster: it accepts commands, schedules kube-scheduler processes, manages states via kube-controller-manager, and stores configuration in etcd. Worker nodes (workers) perform actual workloads (pods) under the management of kubelet and serve network requests through kube-proxy.

Control Plane:

• kube-apiserver: The API server of the cluster, a single entry point.

  # Example API request
  apiVersion: v1
  kind: Pod
  metadata:
    name: my-pod
  spec:
    containers:
    - name: my-container
      image: nginx
  

• etcd: A distributed, consistent key-value store for storing cluster state.
• kube-scheduler: Selects nodes for placing new pods.
• kube-controller-manager: Runs controllers that monitor the cluster state and make necessary changes.
  • Node Controller
  • Replication Controller, ReplicaSet, Deployment Controller
  • Endpoints Controller
  • Service Account & Token Controllers
• cloud-controller-manager (optional): Interacts with cloud providers to manage resources (load balancers, persistent volumes, etc.).

Worker Nodes:

• kubelet: An agent running on each node. Manages containers in pods, interacting with container runtimes (e.g., Docker, containerd).

  # Example command that kubelet might execute (simplified)
  docker run nginx
  

• kube-proxy: Maintains network rules on nodes, ensuring communication between pods, the cluster network, and the outside world.
• Container Runtime: Software responsible for running containers (Docker, containerd, CRI-O).