k8s-learning

A repository created to learn kubernetes concepts.

Pre-requisites

• Install docker, kubernetes
• Install minikube
• Enable the following addons on minikube
  • minikube addons enable ingress
  • minikube addons enable registry

Components in the project

The project contains the following components:

• Frontend - Written in javascript and requires an nginx server
• Backend - Written in go

The frontend and backend are packaged into docker images that are then used in the kubernetes deployment

K8s Cluster architecture

The K8s cluster has the following configuration

• Single node cluster
• 3 replicas of frontend
• 3 replicas of backend
• 1 LoadBalancer service each for frontend and backend
• 1 nginx Ingress controller configured with dns configured for test-app.local domain

Detailed description of each component

Frontend

• Deployment configured to have 3 replicas with labels app: test-react-app and tier: frontend
• Container image: localhost:5000/testapp:latest
• Image created using the frontend Dockerfile
• nginx server + react app

Backend

• Deployment configured to have 3 replicas with labels app: test-backend-app and tier: backend
• Container image: localhost:5000/testbackend:latest
• Image created using the backend Dockerfile
• Go + Gin

How to generate images

• Enable the registry addon for minikube by running minikube addons enable registry

• Run docker build for either the frontend or backend by running:

  cd testapp
  docker build -t testapp .
  docker tag testapp localhost:5000/testapp

  cd backend
  docker build -t testbackend .
  docker tag testbackend localhost:5000/testbackend

• Redirect port 5000 to point to the container registry in the minikube cluster by running the following script in a different terminal

  scripts/registry-port-forward.sh

  Then, run the following scripts to push the docker images to the container registry

  docker push localhost:5000/testapp
  docker push localhost:5000/testbackend

Setup hosts

The host test-app.local needs to be added to the system's hosts file.

Get the IP address of the minikube host:

> minikube ip
192.168.58.2

Use this IP to bind it to the test-app.local domain

vi /etc/hosts

Add the following line to the hosts file (use the minikube host IP address)

192.168.58.2	test-app.local

Run the K8s deployment

At this point, the K8s deployment can be run by running the following:

cd deployment
kubectl apply -f test-deployment.yaml

• This creates 6 pods for frontend and backend
• test-app.local/* will always point to one of the frontend pods depending on availability and load
• test-app.local/backend/* will always point to one of the backend pods depending on availability and load - in these requests, the /backend/ is stripped off and the rest of the path is passed on to the pods

The frontend screenshot at http://test-app.local/

The backend screenshot at http://test-app.local/backend/albums