Lab: The Managed Kubernetes workflow simulated with k3d

What you will learn in this TP

• Create a local Kubernetes cluster
• Expose an application with a LoadBalancer
• Use a persistent volume

Introduction #

What you will learn in this section

• Create a local Kubernetes cluster
• Expose an application with a LoadBalancer
• Use a persistent volume

Context. Before ordering a Managed Kubernetes cluster, you validate your manifests and habits on a local cluster.

Prerequisites: Docker, the AWS CLI (for the S3 API), kubectl, k3d and Terraform 1.7 or later; everything is simulated locally, with no OVHcloud account  ·  Estimated time: 45 min

An exposed, persistent and scaled application, ready for MKS.

Hands-on #

Guided steps, with their solution.

1. 1. Create the cluster

   Why? k3d runs k3s in Docker; its built-in load balancer plays the role of the OVHcloud Load Balancer.

   Create a cluster with one server and two agents, publishing port 8080.

2. 2. Deploy and expose

   Why? On Managed Kubernetes, a LoadBalancer Service creates an OVHcloud Load Balancer with a public IP.

   Deploy three nginx replicas, expose them and test.

3. 3. Add a persistent volume

   Why? On OVHcloud, the csi-cinder storage class creates a Block Storage volume.

   Create a 1 Gi PVC and a pod that writes to it, delete the pod then check the data persists.

4. 4. Scale and clean up

   Why? On MKS, the node pool autoscaler adds nodes when pods no longer fit.

   Scale to 6 replicas, observe distribution across nodes then delete the cluster.

Difficulty level : ●●●○○ (3/5)