In my last blog post, we outlined the different methods of creating and maintaining a Kubernetes cluster, before deciding on Kops. In this blog post, we’ll actually create the cluster using Kops. I’ll provide source code and instructions, so by the end of this post, you can have your own Kubernetes cluster!

This tutorial is strongly based on Kops AWS tutorial, although its even simplifier because I’ve written some generic terraform configurations which simplify initial AWS configuration.

Note, following this tutorial creates AWS resources that cost ~$100 a month.

Step 0: Prerequirements

This tutorial assumes that you have an AWS account in which we can launch our Kubernetes cluster. Additionally, it assumes that you’ve installed kops, kubectl, and terraform.

It also assumes you’ve cloned my personal-k8s project onto your local machine.

Step 1: Decision Time and AWS setup

via GIPHY

Before we can use Kops with AWS, we need to create the AWS resources Kops needs to function. And before we can create the AWS resources Kops needs to function, we need to make some decisions. Specifically, what is the domain in which we’ll host all k8s DNS, and in which s3 bucket should kops store its state.

With the answers to those questions in mind, you can follow the instructions section of personal-k8s/bootstrap. The instructions indicate when you need to parameterize the existing code with your preferred values.

Completing these instructions will create all the AWS resources that Kops needs, meaning that we’re now ready to create our Kubernetes cluster!

Step 2: Create Your Cluster

via GIPHY

We are now ready to create the Kubernetes cluster. Remember, before we undertake any operation involving Kops, we want to run source /PATH/TO/personal-k8s/bootstrap/env.sh, which will populate useful environment variables.

Now, we can run the following command to perform the first step in creating our Kubernetes cluster. Note, this command will create a cluster configuration, but not yet generate any AWS resources.

kops create cluster --name=$NAME --state=$KOPS_STATE_STORE --zones=$AZ --ssh-public-key PATH_TO_YOUR_PUBLIC_KEY

This initial cluster configuration presumes a number of sensible defaults. You can examine them all with the following command:

kops edit cluster --name=$NAME

Kops documentation provides instructions on modifying these default values, but for our initial use case they should work perfectly. One important note is that by default, Kops creates machines/DNS records that are publicly accessible. Kubernetes has additional security mechanisms preventing unwanted access, so I’m comfortable with the machines being initially publicly accessible. You will have to make the decision of whether that’s something with which you are comfortable. If not, Kops does support using private DNS records and hosting machines in a private subnet. I have not yet had the chance to experiment with them, but hope to in the future.

If you’re happy with your cluster’s configuration values, then we can instruct Kops to create the physical resources.

kops update cluster ${NAME} --yes

Kops will perform a whole bunch of operations to create the Kubernetes cluster on AWS. After these operations are complete, Kops will update your ~/.kube/config file with credentials for accessing the cluster. As a result, if you run kubectl get nodes, you should see your cluster’s nodes. Try launching a simple deployment to verify everything is working as expected.

kubectl run test-deployment --image=nginx --replicas=1

Running kubectl get pods should return a test-deployment-SOME_UID pod with a status of Running.

Step 3: Update Your Cluster

Kops provides easy tooling for updating your cluster. You can run kops edit cluster ${NAME} to edit the cluster configuration, followed by kops update cluster ${NAME} to preview the changes, and kops update cluster ${NAME} --yes to apply them. See Kops documentation for more details.

Kops additionally provides the following useful command to ensure your cluster is working as expected.

kops validate cluster

Step Hopefully Never: Deleting Your Cluster

Hopefully you will continue to use this Kubernetes cluster forever, but should you need to delete all of the AWS resources Kops created, Kops provides tooling to do so.

First, run the following command to preview specifically what Kops will delete.

kops delete cluster --name ${NAME}

If you are comfortable deleting everything output in the previous command, you can run the following command to delete your Kubernetes cluster.

kops delete cluster --name ${NAME} --yes

The personal-k8s/bootstrap documentation provides instructions for deleting all of the AWS resources you created to support Kops.

Wrapping Up

via GIPHY

Congrats, you did! You now have a Kubernetes cluster of your own running on AWS. If you’re looking for an application to start running on this cluster, a static website (like a blog) is a great first choice… and coincidentally, my next blog post will examine exactly that :)