Ride Sharing sample k8s microservice project

Checkout the https://www.selfmadeengineer.com/ where a full course was built around this project.

Services

• api-gateway: API Gateway
• driver-service: Driver Service

Intro

This is a golang microservices sample project using Kubernetes for both local development and for production, making you more confident on developing new microservices and deploying them.

Requirements

To run this project locally all you need is Tilt and Minikube

Additionally, the /web folder is a NextJS web app, for that you need NodeJS (v20.12.0).

The project also offers a skaffold.yaml file which is obsolete, it's still in the project for demo purposes of Tilt vs Skaffold. Use it if you know what you're doing.

Run

tilt up

Monitor

kubectl get pods

or

minikube dashboard

Deployment (Google Cloud example)

It's advisable to first run the steps manually and then build a proper CI/CD flow according to your infrastructure.

0. Enviroments

REGION: europe-west1 # change according to your location
PROJECT_ID: <your-gcp-project-id>

1. Build Docker Images

Build all docker images and tag them accordingly to push to Artifact Registry.

docker build -t {REGION}-docker.pkg.dev/{PROJECT_ID}/ride-sharing/api-gateway:latest --platform linux/amd64 -f infra/production/docker/api-gateway.Dockerfile .
docker push {REGION}-docker.pkg.dev/{PROJECT_ID}/ride-sharing/driver-service:latest -platform linux/amd64  .

2. Create a Artifact Registry repository

Go to Google Cloud > Artifact Registry and manually create a docker repository to host your project images.

3. Push the Docker images to artifact registry

Docker push the images. If you get errors pushing:

1. Make sure to gcloud login, select the right project or even gcloud init.
2. Configure artifact on your docker config gcloud auth configure-docker {REGION}-docker.pkg.dev Docs

4. Create a Google Kubernetes Cluster

You can either run a gcloud command to start a GKE cluster or manually create a cluster on the UI (recommended).

5. Update manifests files

Connect to your remote cluster and apply the kubernetes manifests.

gcloud container clusters get-credentials ride-sharing-cluster --region {REGION}--project {PROJECT_ID} && \
kubectl apply -f infra/production/k8s

If you need to redeploy you can use the same command above or just kubectl apply -f infra/production/k8s Sometimes pods might need to be deleted for new ones to be deployed.

kubectl get pods
kubectl delete pod <pod-name>

# or for all deployments
kubectl rollout restart deployment

6. Enjoy!

Get the External IP from the api-gateway
kubectl get services

cURL for the /servies or establish a websocket connection to /ws/drivers!

curl http://{EXTERNAL_IP}:8081/services

Go back to locally developing your project by changing kubernetes context

kubectl config get-contexts

# For Docker Desktop
kubectl config use-context docker-desktop

# OR for Minikube
kubectl config use-context minikube

Explanation

This is a basic example of how to structure a microservice architecture with Kubernetes.

You can try to run the tilt up command and then try to access the api-gateway service at http://localhost:8081.

You will see that the websocket request is being routed to one of the driver-service pods.

You can try to scale the driver-service to 3 replicas and then try to access the driver-service service again.

You will see that the request is being routed to one of the driver-service pods in round-robin fashion (check k8s config).

This is because the driver-service service is configured to use the round-robin algorithm to distribute traffic.

This is pretty neat and it's a good way to start with Kubernetes and explore the features it offers.