1/3 Hands on Kubernetes with Minikube

"Kubernetes is a powerful orchestration system but its full production setup usually involves multiple nodes, external load balancers and distributed storage. To simplify learning and experimentation, Minikube can be used to provide a single node Kubernetes cluster that runs locally while still reflecting the same architecture used in production. In this environment, the control plane and worker components run on the same node, allowing us to study how Kubernetes functions internally without the complexity of multiple servers."

"In my setup, I used the docker image to run minikube in my Windows host and once installed, a cluster can be launched with: minikube start This creates a virtualized node with its own internal IP, usually around 192.168.49.x. In my setup, the Minikube node is running on 192.168.49.2. The container runtime in Minikube is Docker 28.1.1, which executes the containerized workloads inside each Pod."

"The API server pod kube-apiserver-minikube runs on the node at 192.168.49.2:6443. The API Server is the central component that exposes the Kubernetes API, processes all requests, and manages the state of the cluster by coordinating with other components. This is the entry point to the cluster for cluster admins. Any request made via kubectl or the dashboard first reaches this API server."

Minikube provides a single-node Kubernetes cluster that runs locally and mirrors production architecture. It virtualizes a node with an internal IP, commonly 192.168.49.x, and uses a container runtime (e.g., Docker 28.1.1) to execute containers within Pods. The kube-system namespace contains control plane pods including kube-apiserver-minikube and etcd-minikube. The API server listens on 192.168.49.2:6443 and serves as the cluster entry point for kubectl and dashboards. The etcd datastore runs on 192.168.49.2:2379 and persists all cluster resources. Namespaces isolate pods, services, and resources to organize the cluster. Running control plane and worker components on the same node simplifies learning, experimentation, and internal investigation without multi-server complexity.

Read at Medium