Navigating Kubernetes Locally: Tools and Techniques

If you're starting out in the world of Kubernetes (K8s)—whether you're a DevOps enthusiast, cloud learner, or developer—you're probably asking:

❓ How do I install Kubernetes locally?
❓ What is kubectl? Do I need Docker?
❓ What's minikube and kind?

This blog will help you clearly understand the ecosystem, tools, and setup flow.

---

📦 What is kubectl?

kubectl is the command-line tool for Kubernetes. It lets you:

• Deploy applications to a cluster

• Manage cluster resources

• Inspect logs, nodes, pods, and services

• Scale workloads and roll out updates

🧠 Think of it as:

Your remote control to interact with a Kubernetes cluster.

Example Commands:

kubectl get nodes         # List cluster nodes
kubectl get pods -A       # View all pods
kubectl apply -f app.yaml # Apply a configuration file
kubectl logs <pod-name>   # Check logs

BUT! kubectl does nothing on its own. It requires a cluster to talk to.

---

🔄 The Server-Client Relationship in Kubernetes

At its core, Kubernetes follows a client-server architecture, where:

• kubectl is the client

• Kubernetes API Server is the server

---

🧩 1. Who is the Client?

✅ kubectl
This is the CLI tool you use to send requests like:

• “Create a pod”

• “Show all deployments”

• “Scale this app”

• “Delete a service”

But kubectl doesn’t do the work—it simply tells the server what you want to happen.

---

🧠 2. Who is the Server?

✅ Kubernetes API Server
This is the brain of the cluster, running inside the control plane node (e.g., Minikube). It:

• Receives HTTP requests from kubectl or other clients

• Validates and processes them

• Stores data in etcd (the K8s database)

• Communicates with other components like the Scheduler, Controller Manager, and Kubelets to make it happen

---

🔁 The Communication Flow

You (CLI user)
     ↓
kubectl (client)
     ↓
Kubernetes API Server (server)
     ↓
Kubernetes internal components (Scheduler, Kubelets, etc.)

So, whenever you run:

kubectl get pods

The flow is:

1. kubectl formats that as an HTTP request

2. Sends it to the API server (usually at https://<cluster-ip>:6443)

   You can check your API server endpoint via:

    kubectl config view
   

   Or:

    kubectl cluster-info
   

   You’ll see output like:

    Kubernetes control plane is running at https://192.168.49.2:6443
   

3. The API server queries etcd or the cluster

4. Response is returned to you via kubectl

---

🏗️ What Do You Need to Run Kubernetes Locally?

To actually use Kubernetes commands via kubectl, you need an actual Kubernetes cluster. You can set it up in several ways:

---

🔧 Different Ways to Set Up Kubernetes

---

🌱 Recommended for Learning: minikube or kind

✅ Minikube

• Spins up a single-node cluster

• Works using drivers like Docker, VirtualBox, KVM, etc.

• Great for experimenting and running real apps locally

✅ kind (Kubernetes IN Docker)

• Runs clusters inside Docker containers

• Lightweight and fast

• Ideal for CI/CD pipelines or testing on Docker-enabled machines

---

🔄 Flow of Tools

1. You install and start a local cluster (via minikube or kind)
         ↓
2. This cluster creates a kubeconfig file (cluster access credentials)
         ↓
3. kubectl uses that config to connect and manage the cluster

So, the relationship looks like this:

minikube/kind → Kubernetes cluster ← kubectl

---

🔍 Example: You Start Minikube with Docker Installed

minikube start

Output:

* Automatically selected the docker driver
* Starting "minikube" primary control-plane node...

Minikube sees Docker is installed and uses it to spin up a container-based Kubernetes node.

---

❌ What Happens if Docker is Not Installed?

You’ll get something like:

😿  Exiting due to PROVIDER_NOT_FOUND: No compatible minikube driver found.

In this case, you either:

• Install Docker ✅

• Install another driver (like VirtualBox or Podman) ✅

• Or switch to another setup method like k3s

---

🎓 Final Thoughts

• Install kubectl to manage your cluster.

• Install minikube or kind to create a local cluster.

• Install Docker if you're using kind or want a faster minikube setup.

• Understand that kubectl needs a cluster to work with, and minikube/kind provides that.

Once set up, you're ready to start deploying apps and learning Kubernetes the real way.