K8S bootstrap

Bootstrap a single-node Kubernetes cluster in a Vagrant Ubuntu VM, step-by-step, by manually configuring each component.

Motivation & credits

This repo is part of my "learning Kubernetes" experience and is heavily based on Joshua Sheppard's Kubernetes by Component series. Read more about my motivation and approach on this post.

Other references:

• Kelsey Hightower - Kubernetes The Hard Way
• Carson Anderson - Kubernetes Deconstructed (Kubecon 2017)

Remarks

The cluster resulting from the different steps in this repo is functional, although incomplete. Some remarks:

• All components run as Docker containers, except for Kubelet, which runs using systemd.
• TLS client authentication between components and the API Server is not configured.
• Authorization mode on the api-server is set to AlwaysAllow.
• RBAC is not configured.
• CoreDNS is not configured.

Prerequisites

• The cluster runs in a VM created/configured using Vagrant, so it needs to be installed on the host machine.
• Download the Kubernetes server binaries from the Kubernetes releases page into the current directory.
  • This is not part of the Vagrantfile to avoid the long download on each vagrant up.
  • Tested using v1.18.2. Adapt the URL accordingly.

  wget https://dl.k8s.io/v1.18.2/kubernetes-server-linux-amd64.tar.gz
  

• Some knowledge of Kubernetes is required. The different components are briefly introduced as they are used.

Run it!

Each step configures one or more components towards the full cluster configuration. There's a folder for each step, containing a short description and a Vagrantfile that configures a VM accordingly. The main learnings are highlighted on each step, but reading the full journey in Joshua Sheppard's posts may also help.

• Step 1 - Running Pods using kubelet
• Step 2 - Add the Kubernetes API Server and access it using kubectl
• Step 3 - Add kube-scheduler so that Pods are automatically assigned to the node
• Step 4 - Add kube-controller-manager and start using Deployments
• Step 5 - Add kube-proxy and access the cluster from the host machine