This guide shows how to run Kubernetes in a Windows Host, using kind to run the control plane in Docker and WSL 2 to run Docker itself.

The whole point of doing that is to run Kubernetes:

1. In a Windows Host
2. Without using virtualization
3. Without using Docker Desktop's native Kubernetes feature (as it might break other workflows that do not relly on k8s).

scoop is a lightweight package manager for Windows. It installs applications in ~/.scoop and adds then to PATH. We will use it to install CLI tools.

Set-ExecutionPolicy RemoteSigned -scope CurrentUser
iwr -useb get.scoop.sh | iex
scoop --version

kind is a tool for running local Kubernetes clusters using Docker containers as “nodes”.

kubectl is a tool that allows runing commands against Kubernetes clusters.

We will install those tools using scoop.

scoop install kind kubectl
kind version
kubectl version --client

Docker Desktop an application created by Docker Inc.'s (the company that created Docker, the technology). It provides a user friendly graphical user interface to run and inspect Docker containers. It also creates and manages a virtual machine in which containers are executed. That allows Windows users to run Linux containers. The most recent versions of Docker for Desktop run containers using WSL 2, which is a Windows subsystem that ships a Linux kernel, allowing users to run Linux distros in Windows without the overhead of a virtual machine.

To install Docker Desktop, download it from the official web page and follow the interactive graphical installer (this will require privillege elevation).

Create a cluster by invoking kind, then check if kubectl is configured correctly:

kind create cluster
kubectl cluster-info --context kind-kind

If everything was setup correctly, you should be a message like so:

Kubernetes control plane is running at https://127.0.0.1:62109
CoreDNS is running at https://127.0.0.1:62109/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

That's it. You can now run commands against your cluster using kubectl or use graphical tools like Lens.