This tutorial shows you how to start using kluctl.

Before you begin

A few things must be prepared before you actually begin.

Get a Kubernetes cluster

The first step is of course: You need a kubernetes cluster. It doesn’t really matter where this cluster is hosted, if it’s a local (e.g. kind) cluster, managed cluster, or a self-hosted cluster, kops or kubespray based, AWS, GCE, Azure, … and so on. Kluctl is completely independent of how Kubernetes is deployed and where it is hosted.

There is however a minimum Kubernetes version that must be met: 1.20.0. This is due to the heavy use of server-side apply which was not stable enough in older versions of Kubernetes.

Prepare your kubeconfig

Your local kubeconfig should be configured to have access to the target Kubernetes cluster via a dedicated context. The context name should match with the name that you want to use for the cluster from now on. Let’s assume the name is, then you’d have to ensure that the kubeconfig context is correctly pointing and authorized for this cluster.

See Configure Access to Multiple Clusters for documentation on how to manage multiple clusters with a single kubeconfig. Depending on the Kubernetes provisioning/deployment tooling you used, you might also be able to directly export the context into your local kubeconfig. For example, kops is able to export and merge the kubeconfig for a given cluster.


  • Checkout one of the example Kluctl projects
  • Deploy to your local cluster
  • Change something and re-deploy

Install Kluctl

The kluctl command-line interface (CLI) is required to perform deployments. Read the installation instructions to figure out how to install it.

Use Kluctl with a plain Kustomize deployment

The simplest way to test out Kluctl is to use an existing Kustomize deployment and just test out the CLI. For example, try it with the podtato-head project:

  $ git clone
$ cd podtato-head/delivery/kustomize/base
$ kluctl deploy

Then try to modify something inside the Kustomize deployment and retry the kluctl deploy call.

Try out the Kluctl examples

For more advanced examples, check out the Kluctl example projects. Clone the example project found at

  git clone

Choose one of the examples

You can choose whatever example you like from the cloned repository. We will however continue this guide by referring to the simple-helm example found in that repository. Change the current directory:

  cd kluctl-examples/simple-helm

Create your local cluster

Create a local cluster with kind:

  kind create cluster

This will update your kubeconfig to contain a context with the name kind-kind. By default, all examples will use the currently active context.

Deploy the example

Now run the following command to deploy the example:

  kluctl deploy -t simple-helm

Kluctl will perform a diff first and then ask for your confirmation to deploy it. In this case, you should only see some objects being newly deployed.

  kubectl -n simple-helm get pod

Change something and re-deploy

Now change something inside the deployment project. You could for example add replicaCount: 2 to deployment/nginx/helm-values.yml. After you have saved your changes, run the deploy command again:

  kluctl deploy -t simple-helm

This time it should show your modifications in the diff. Confirm that you want to perform the deployment and then verify it:

  kubectl -n simple-helm get pod

You should need 2 instances of the nginx POD running now.

Where to continue?

Continue by reading through the recipes and tutorials. Also, consult the reference documentation for details about specifics.