Eventually - Taking Kubernetes for a Spin

With more than two years now since the announcement: Kubernetes is Now Available In Docker Desktop Stable Channel It’s high time to take Kubernetes for a spin.

Kubernetes (K8s) - Production-Grade Container Orchestration - Automated container deployment, scaling, and management

It takes two major components for this experiment:

Prerequisites:

Kubernetes - e.g. as part of Docker Desktop
Install and Set Up kubectl a K8s CLI tool

Docker Desktop - The fastest way to containerize applications on your desktop

Please follow the instructions for your operating system in use.

Check the installation with docker --version.

The kubectl command-line tool lets you control Kubernetes clusters.

In case you are using a Mac: brew install kubectl otherwise please check Install and Set Up kubectl

Check the installation with kubectl version --client.

With cluster-info we get the first insights:

$ kubectl cluster-info
Kubernetes master is running at https://kubernetes.docker.internal:6443
KubeDNS is running at https://kubernetes.docker.internal:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

Looks good master is running...

Tune into the playground cluster

Usually, you want access to additional clusters, e.g. development, canary, … you name it.

Inspired by Configure Access to Multiple Clusters we tune into a K8s playground (running unsecured on localhost:8080) besides our local K8s cluster.

Tune into the playground cluster with the configuration option set-cluster...

kubectl config --kubeconfig=config-playground set-cluster playground --server=http://localhost:8080 --insecure-skip-tls-verify

...finishing touches to the sandbox context with set-context

kubectl config --kubeconfig=config-playground set-context sandbox --cluster=playground --namespace=default --user=developer

View the result with kubectl config --kubeconfig=config-playground view.

Switch the context with kubectl config --kubeconfig=config-playground use-context sandbox and finally check your current context with kubectl config current-context and you are ready to go!

Note: You can use the environment variable KUBECONFIG to avoid the cumbersome --kubeconfig parameter.

You should see something similar to:

$ export KUBECONFIG=config-playground
$ kubectl config view --minify
apiVersion: v1
clusters:
- cluster:
    insecure-skip-tls-verify: true
    server: http://localhost:8080
  name: playground
contexts:
- context:
    cluster: playground
    namespace: default
    user: developer
  name: sandbox
current-context: sandbox
kind: Config
preferences: {}
users:
- name: developer
  user: {}

The first pod - Dashboard

Let's continue with two more tools in the K8s ecosystem: The Web UI / Dashboard and Helm.

Deployment with `kubectl`

We'll deploy the Web UI (Dashboard) via kubectl first:

kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/v2.0.4/aio/deploy/recommended.yaml

Note: As you might have noticed, we used a deployment snippet from the internet.

You might want to check the latest version (and the content) directly at the GitHub project kubernetes/dashboard...

Get the list of all pods with the namespace kubernetes-dashboard:

$ kubectl get pods --namespace kubernetes-dashboard
NAME                                         READY   STATUS    RESTARTS   AGE
dashboard-metrics-scraper-6b4884c9d5-v5qlf   1/1     Running   0          111m
kubernetes-dashboard-7d8574ffd9-v6crt        1/1     Running   0          111m

No ports are publically accessible by default. Run kubectl proxy to expose the dashboard.

Add ServiceAccount and ClusterRoleBinding with the shell script create-dashboard-user.sh. Grab the token with describe secret...for more details check the script mentioned above.

Visit the proxy URL http://localhost:8001/api/v1/namespaces/kubernetes-dashboard/services/https:kubernetes-dashboard:/proxy/. and enter the token...that's it!

Deployment with Helm

Helm - The package manager for Kubernetes

Instead of using kubectl with YAML, you can use Helm with a bunch of Kubernetes packages.

On a Mac, use brew install helm to install Helm, otherwise check Installing Helm.

Verify the installation with helm version.

Artifact HUB - Find, install and publish Kubernetes packages

Installing the dashboard (using a so-called chart) with helm install would look something like:

$ helm install local-dashboard kubernetes-dashboard/kubernetes-dashboard
...
Get the Kubernetes Dashboard URL by running:
  export POD_NAME=$(kubectl get pods -n default -l "app.kubernetes.io/name=kubernetes-dashboard,app.kubernetes.io/instance=local-dashboard" -o jsonpath="{.items[0].metadata.name}")
  echo https://127.0.0.1:8443/
  kubectl -n default port-forward $POD_NAME 8443:8443

Note: At the time of writing, I had certificate issues when accessing the dashboard locally.

Please check https://artifacthub.io/packages/helm/k8s-dashboard/kubernetes-dashboard for more details about this chart.

A simple helm list generates what you would expect … a list of deployments.

To round up this section run:

$ helm delete local-dashboard
release "local-dashboard" uninstalled

Cleanup

Run kubectl delete pod <NAME> will result in spawning new pods...

Extract the NODE running the Kubernetes dashboard pods run on with kubectl and awk...

$ kubectl get pods --namespace kubernetes-dashboard -o wide | awk '{print $7,$1}'
NODE NAME
docker-desktop dashboard-metrics-scraper-6b4884c9d5-v5qlf
docker-desktop kubernetes-dashboard-7d8574ffd9-9db86

Run delete deployments to properly delete the dashboard.

$ kubectl delete deployments --namespace kubernetes-dashboard kubernetes-dashboard
deployment.apps "kubernetes-dashboard" deleted

Tidy up the playground with the deletion of the service account and the role binding...

kubectl -n kubernetes-dashboard delete serviceaccount admin-user
kubectl -n kubernetes-dashboard delete clusterrolebinding admin-user

The cluster is a good as new...have fun on your K8s journey! 🎉

Bonus - Troubleshooting with Octant

Octant - Visualize your Kubernetes workloads

Run the command brew install octant in case you are using a Mac. Check the releases from the Github project vmware-tanzu/octant otherwise.

Visit the Octant Overview with the browser of your choice to gain insights and start troubleshooting your K8s cluster(s).

k9s - Kubernetes Manager for Console Power Users

k9s - Kubernetes CLI To Manage Your Clusters In Style!

The documentation start with Who Let The Pods Out?. A proof that technical documentations can be funny to read…

Bonus - Katacoda

Visit the Katacoda Kubernetes Playground for additional hands-on experiments...

Resources

Photo by Matt Artz from Unsplash