Skip to main content

Kubernetes Cluster

Updated Apr 07, 2022 ·

Overview​

A Cluster is a collection of distributed physical or virtual servers or nodes, which is used to host and manage workloads.

It has two types nodes:

  • Master Node (Control Plane)

    • Manages the cluster, making global decisions.
    • Components:
      • kube-apiserver – Exposes Kubernetes API
      • kube-scheduler – Assigns workloads to nodes
      • kube-controller-manager – Keeps resources current
      • etcd – Stores cluster data and configurations

  • Worker Nodes (Data Plane)

    • Host application workloads.
    • Components running on all nodes (master and worker):
      • kubelet – Agent on each node, registers it with the cluster
      • kube-proxy – Network proxy ensuring workload reachability

Master Node (Control Plane)​

The master node manages core control functions of the cluster.

  • Primary access point for administration
  • Coordinates cluster operations
  • Manages monitoring and scheduling
  • Runs system Pods (API server, Cluster Store, Scheduler, Control Manager)
  • Forwards workloads to worker node Pods

API Server​

The API Server is the central hub for all Kubernetes communication and operations.

  • Acts as a REST API interface
  • Routes all configuration changes
  • Verifies operations and updates etcd

etcd (Cluster store)​

etcd is the persistent storage for the Kubernetes cluster.

  • Saves Kubernetes object states in key-value pairs
  • Monitors object state with watches on keys
  • Stateless services retrieve data via the API server

Scheduler​

The Scheduler assigns Pods to Nodes based on resource needs.

  • Monitors API server for unscheduled Pods
  • Matches Pods to Nodes based on resource constraints

Scheduling process:

  • Filtering – Finds Nodes with required resources
  • Scoring – Ranks Nodes to pick the best match for each Pod

Controller Manager​

The Controller Manager maintains system state via Controllers.

  • Runs controller loops to monitor
  • Update system state based on desired configurations

Controller types:

  • Node Controller – Manages node availability
  • Replication Controller – Maintains correct Pod counts
  • Endpoints Controller – Links services and Pods
  • Service Account and Token Controllers – Manages default accounts and tokens for namespaces

Cloud Controller Manager (for EKS Setup only)​

The Cloud Controller Manager manages cloud-based resources for AWS EKS.

  • Manages autoscaling and cloud integrations
  • Provisions EBS volumes and load balancers

Additionally, kubectl is essential for interacting with the API Server, though it is not part of the control plane itself.

Worker Node​

A worker node starts and maintains Pods, ensuring they run smoothly. It manages networking and operates within a cluster of multiple nodes.

Kubelet​

The Kubelet is the node agent that communicates with the Kubernetes API server.

  • Starts and stops Pods as directed by the API server
  • Monitors Pod and node status; reports back to API server
  • Liveness probes - for health check of Pods and applications

Kube-proxy​

The Kube-proxy is a network proxy that runs on each node to manage network traffic.

  • Provides rules for cluster IPs and Pod networking
  • Talks to the API server for networking information
  • Handles service abstraction and load balancing
  • Routes traffic to Pods

Container Runtime​

The Container Runtime is the environment where the container image runs.

  • Pulls images from the registry
  • Provides the execution environment
  • Default runtime is Docker, though others are supported

Scheduled/Add-Ons​

Add-ons are specialized Pods offering extra cluster services.

  • DNS Pod – Manages DNS and service discovery within the cluster

  • Ingress controllers and dashboards – Provide web-based administration


Feedback