About

This repository is intended to host all the files necessary for deploying AWS resources and configuring them in order to contain a Kubernetes cluster (1 master & 2 worker nodes)

How to run

Just run the deploy-cluster.sh command and it should work. You need the following Environment variables set:

export AWS_ACCESS_KEY_ID=<AWS_ACCESS_KEY_ID>
export AWS_SECRET_ACCESS_KEY=<AWS_SECRET_ACCESS_KEY>
export CLOUDFLARE_TOKEN=<CLOUDFLARE_DNS_EDIT_TOKEN>

Basics

The following need to be covered:

Infrastructure deployment (Terraform)

Deploy custom VPC and subnet (include public IP and DNS entry)
Route tables and Internet Gateways
Security Groups (for control plane & worker nodes)
Deploy master, worker nodes and SSH public key for access (t3.medium should be minimum)

Configuration provisioning (Ansible)

Disable swap, configure hostname, set system clock
Deploy kubeadm
Create Kubernetes nodes and register cluster

Advanced networking & monitoring

Add demo nginx application
Add Prometheus & Grafana deployments
Add Ingress controller (using nginx-ingress-controller)
Add ALB resource for load balancing the web application (currently using a classic EC2 load balancer - ELB)
Add record for ALB in domain DNS server (managed externally)

Future improvements

Create script to execute ansible playbooks
Add script to execute all steps necessary to create a working cluster with Ingress (see deploy-cluster.sh)
Add ansible task to copy kube-config file from master node & edit for external use
Add persistent storage for pods
Add support for LoadBalancer resources (see the MetalLB project, then go to Installation page; in this case, nginx-ingress-controller can be switched to the cloud variant, but it should not interfere with the baremetal implementation which uses NodePort)
Create an ECR to store a custom image to show the running host
Migrate to Terragrunt for eliminating duplicate code