Deploy a Kubernetes Cluster on Ubuntu Server with Microk8s

Kubernetes is hard. There is no way around that. From start to finish, you’ll find so many roadblocks in your way that, at times, you might feel so inclined as to give up.

Don’t.

There’s actually a very easy way to deploy a Kubernetes cluster that can be used for development purposes or even production.

That method is thanks to Microk8s.

Microk8s was created by Canonical as a simplified means of deploying Kubernetes environments to either commodity hardware or third-party cloud hosts (such as Amazon Web Services, Azure, Rackspace, and Linode). Microk8s is, hands down, the easiest tool for getting up to speed with Kubernetes.

And it’s not even close.

Because it’s so simple, you might think Microk8s is limited to a single-node dev environment. In that thinking, you’d be wrong. The developers of Microk8s have actually made it possible to cluster instances together. Case in point, all it takes is three Microk8s nodes to add failover into the mix.

And that’s exactly what I’m going to do today… show you how to deploy a Kubernetes cluster with the help of Microk8s.

Requirements

I’m going to demonstrate on Ubuntu Server 22.04, but this process will work on any operating system that supports Snap packages. You’ll need three instances of Ubuntu Server that are on the same network and a user with sudo privileges. That’s it… let’s make some magic.

Installing Microk8s

The first thing you’ll do is install Microk8s on each node. So log in to your first node and issue the command:

sudo snap install microk8s --classic

Do the same thing on every one of your nodes. As well, you’ll need to add your user to the microk8s group with the command:

sudo usermod -aG microk8s $USER

Log out and log back in so the changes will take effect. Do this on each node.

Create a new directory with the command:

mkdir ~/.kube

Give that new directory the proper permissions with the following command:

sudo chown -f -R $USER ~/.kube

Huzzah!

Configure Hostnames and Hosts

Before you continue on, you’ll need to configure the hostnames and hosts file on each node. To set the hostname, you issue the command like so:

sudo hostnamectl set-hostname microk8scontroller

You can name the system anything you like (such as microk8scontroller, microk8snode1, microk8snode2, etc.).

Next, you’ll need to configure the hosts file on each machine. Open that file for editing with the command:

sudo nano /etc/hosts

At the bottom of that file, add lines such as (using your hostnames and IP addresses):

You’ll need to add the above to each node’s hosts file. Save and close the file.

At this point, if you could verify everything’s up and running so far by running the command on the controller node:

microk8s kubectl describe node

You should see a detailed description of the controller node.

How to Join Nodes

Unlike standard Kubernetes, this process is considerably easier. From the controller node, issue the command:

microk8s add-node

The output of the above will include the join command that will look something like this:

microk8s join 192.168.1.137:25000/d6c656de35d466889c92cebcaca0d843/30b781bb37e8

Head over to your first node and run that join command. When it completes, go to your next node and run the same command. Continue doing this until you’ve joined all of your nodes to the cluster.

You can then check to make sure the nodes have all joined with the command:

microk8s kubectl get nodes

You should see output that looks similar to this:

microk8scontroller   Ready    <none>   23d   v1.24.3-2+63243a96d1c393

microk8snode1        Ready    <none>   23d   v1.24.3-2+63243a96d1c393

microk8snode2        Ready    <none>   23d   v1.24.3-2+63243a96d1c393

You can also verify the status of the cluster with the command:

microk8s status

The output from the above command should look something like this:

microk8s is running

high-availability: yes

datastore master nodes: 192.168.1.100:19001

datastore standby nodes: none

It’s important to note that you will only see high-availability listed as yes if you have 3 or more nodes. If you have fewer than three nodes, high-availability will be listed as no. If you’re only using this as a development cluster (and don’t need HA immediately), you can get by with just two nodes. Otherwise, make sure to add more to enjoy that feature.

Deploying Hello World to Your Cluster

As a quick test, let’s deploy the Hello World app to the cluster. On the controller, run the following commands:

microk8s kubectl create deployment hello-world --image=tutum/hello-world:latest

microk8s kubectl expose deployment hello-world --type=NodePort --port=80 --name=hello-world-service

microk8s kubectl port-forward -n default service/hello-world-service 8080:80

The final command will not return your terminal to you, but will display that port forwarding is running with output like this:

Forwarding from 127.0.0.1:8080 -> 80

Forwarding from [::1]:8080 -> 80

Handling connection for 8080

Log in to your controller node from a second terminal and issue the command:

curl localhost:8080

And you’ll see output that includes <h1>Hello world!</h1>. Of course, if you’re working on a desktop you can open a browser and point it to this page and you’ll see the Hello World page printed.

And that, my new-to-Kube friends, is how you can quickly deploy a Kubernetes cluster with the help of Microk8s. This is a great method of getting a cluster up and running quickly, so you can spend more time developing and less time administering.