Building a Kubernetes Cluster with external or stacked etcd configuration for an on-prem "home lab"

I decided to put together a bunch of instructions on building a Kubernetes Cluster. Having built (and broke) many Kubernetes clusters over the past few years, I thought it would be useful to actually document the steps so that anyone wanting a simple set of instructions to follow can use these to quickly spin up a on-prem Kubernetes Cluster.

These instructions are not intended for Production Grade Kubernetes Clusters, as there are many examples of Enterprise / Production Grade Kubernetes Cluster providers e.g. Red Hat OpenShift, IBM Cloud Private, but they will help you build up a Kubernetes Cluster with an external or stacked etcd configuration for use within a non-production environment or as I've done, in my own "home lab" environment.

Audience

Any one with the passion (or just a bunch of free time) to build their own Kubernetes Cluster.

Index

1. Planning
   • Architecture Overview
   • Stacked or External etcd Configuration
2. Load Balancer for Kubernetes API-Server
   • HAProxy and Heartbeat
3. Kubernetes
   • Pre-requisites
   • etcd
   1. Masters
      1. External or Stacked Etcd
         • Configure Master Node with External etcd
         • Configure Master Node with Stacked etcd
      • CNI Plugins
      • Storage Provisioner
      • Storage Provisioner - Ceph
   2. Create an Highly Available Kube-APIServer
      • Join Additional Masters
   3. Workers
      • Add Worker Nodes
   • Redeploy CoreDNS
   • MetalLB - Load Balancer
   • Helm
   • Metric Server
   • Ingress Controller
   • Sealed Secrets
   • Kubernetes Dashboard
   4. Optional
      • DNS Auto-Scaling
      • Internal Docker Registory