Hands-On Journey Experimenting with Kubernetes: FastAPI + React Deployment

Over the past few weeks, I’ve been diving deep into Kubernetes, exploring how to deploy and manage containerized applications. To really understand the mechanics, I decided to create a small but complete full-stack project: a FastAPI backend with a React frontend running on a local Kubernetes cluster using Kind (Kubernetes in Docker).

Why I Started Experimenting

Kubernetes is powerful but complex, and reading documentation can only take you so far. I wanted to:

• See how Deployments and Services interact in real-time


• Understand how frontend and backend communicate inside a cluster


• Experiment with replicas, scaling, and networking without affecting production

This project became my sandbox for testing Kubernetes concepts and workflows hands-on

How I Set Up the Project

I structured the project into three main areas:

1. Backend (FastAPI)

• Runs on Python 3.11, serving API requests on port 80


• Exposed internally with a ClusterIP service


• I’ve experimented with creating endpoints and testing them using pytest

2.Frontend (React + Nginx)

• React app is built statically and served by Nginx


• Exposed externally with a LoadBalancer service


• Configured SPA routing and CORS headers to communicate seamlessly with the backend

1. Kubernetes Manifests


2. Separate YAML files for backend and frontend


3. Each deployment has 2 replicas


4. Services use Kubernetes DNS for internal pod communication

My Hands-On Experiments

Here’s what I’ve been tinkering with:

• Creating a Kind Cluster: Spinning up a lightweight Kubernetes environment locally


• Loading Docker Images: Pre-loading backend and frontend images into the cluster with kind load docker-image


• Deploying and Updating: Iteratively modifying endpoints, rebuilding images, and redeploying
  -Service Discovery: Using Kubernetes DNS for frontend-backend communication instead of hardcoded IPs
  -Debugging: Inspecting logs, port-forwarding services, and fixing issues like ImagePullBackOff and misconfigured CORS

Through this, I’ve gained a deeper understanding of:

• Replicas and pod distribution


• ClusterIP vs LoadBalancer services


• Internal pod networking


• How environment variables manage API connections

Testing & Local Development

Backend

• Tested with pytest and FastAPI TestClient


• GET / returns a 200 status code with JSON


• Tests run locally with make test

Frontend

• Points to backend API via REACT_APP_API_URL


• Can be tested locally before deploying to Kubernetes

Experimenting with Kubernetes is all about deploying, breaking, fixing, and iterating. Creating a small project like this is the fastest way to understand how everything fits together.

Check out the repo:https://github.com/Copubah/simple-k8s-project