Skip to content

Exposing Your App with Ingress

Your ClusterIP Service is only reachable from inside the cluster. To let real users reach your app through a normal URL, you need an Ingress - a rule that tells the cluster's ingress controller how to route outside HTTP traffic to your Service.

This cluster runs Traefik as its ingress controller (you saw it back in section 1). An Ingress resource is how you ask Traefik: "send requests for this hostname to this Service."

graph LR
    A[Browser<br/>app.dockerlabs.xyz] --> B[Traefik<br/>ingress controller]
    B --> C[Service: web]
    C --> D[Pod]
    C --> E[Pod]

🌐 Create the Ingress

Open k8s/ingress.yaml. It routes the host app.dockerlabs.xyz to your web Service on port 80.

  1. Apply the Ingress:

    kubectl apply -f k8s/ingress.yaml

  2. Confirm it was created and picked up the host:

    kubectl get ingress web

  3. Give Traefik a moment, then request your app from the command line:

    curl http://app.dockerlabs.xyz/api

    You should get JSON back showing "version":"2.0" and a pod name - your request traveled from outside the cluster, through Traefik, to one of your pods. 🎉

👀 See it in the browser

Now open it the way a real user would. In the Labspace, open your app in the Your App tab (or browse to http://app.dockerlabs.xyz).

You'll see the "Hello from Kubernetes!" page showing v2.0 and the pod serving you. Refresh a few times - the pod name changes as Traefik and your Service load-balance across replicas. You've now got a complete path from browser → Ingress → Service → Pods. 🚀

🏁 The fundamentals, done

Look back at everything you built and ran on a real Kubernetes cluster:

  • 📦 Built an image and ran it in a Pod
  • 🚀 Used a Deployment to keep replicas healthy and self-healing
  • 🔌 Gave them a stable address and load balancing with a Service
  • 📈 Scaled up and down, did a zero-downtime rolling update, and rolled back
  • 🌐 Exposed the app to the world through an Ingress

That's the core of how applications run on Kubernetes everywhere - including the cluster built into Docker Desktop.

You wrote every manifest by hand to learn what each one does. But once you know Kubernetes, wouldn't it be nice to go straight from a compose.yaml you already have to ready-to-apply manifests? That's exactly what the next bonus section shows - and you'll even run an LLM on your cluster. 🤖➡️