DevOps Challenge: Fix & Deploy Go App with Redis

Overview

• My task is to troubleshoot, fix, and deploy a Go web application that uses Redis for caching. There are some issues in the Dockerfile or the Go code. After fixing these issues, the app is to be deployed to a Kubernetes cluster with Redis.

Part 1: Fix the Dockerfile and Go Application

• Fixed bug in CMD ["bin/myapp"] to CMD ["app/myapp"]
• Changed PORT to 8080 to ensure clarity
• Set up a Redis container using docker-compose and passed its environment variables
• Optimized image size using multistage builds through golang:1.23-alpine and a scratch image, effectively reducing image size from 1.03GB to 5.71MB (~99.45% size reduction!!!)

Proof

docker-compose up --build

docker compose logs app

docker ps & curl http://localhost:8080

docker images (notice the reduction in image size!!)

Part 2: Deploy to Kubernetes

• Created Kubernetes YAML files and deployed Go (as stateless workload) and Redis (as stateful workload)
• Used separate namespaces: app & db
• Set up PVC for Redis persistent storage
• Set up clusterIP as none to make network ID stable
• Used configmap to manage variables
• Utilized one pod per each workload
• Uploaded local image to docker hub publicly gimmeursocks/docker-k8s-challenge_app
• Used nodeport due to local deployment using minikube
• Exposed Redis internally with stateful state with static hostname

Proof

kubectl apply -f namespaces.yaml

kubectl apply -f go-deployment.yaml
kubectl apply -f go-service-nodeport.yaml

kubectl apply -f redis-pvc.yaml
kubectl apply -f redis-statefulset.yaml
kubectl apply -f redis-service.yaml

kubectl get all -n app
kubectl get all -n db

docker tag & docker push

docker tag & docker push

curl result

Redis persistent storage

kubctl describe pod redis-0 -n db

Managing variables

One pod per each workload

Port exposure using nodeport

Internal communication between Redis

Pods and Services

Bonus

• Created Github Action workflow to automatically build and test the deployment of k8s
• Auto builds using docker compose and then pushes to docker hub
• It tests the deployment of k8s by installing and running minikube in the workflow
• Then waits for successful deployment to curl into it

Testing service URLs in the new deployment

Showing relevant debug information

Testing application through curl