Create a Kubernetes Cluster with kubeadm and terraform on AWS

This repo contains terraform and ansible code required to spin up resources required to run your own kubernetes cluster using kubeadm and CRI-O container runtime. All nodes run ubuntu OS and are of type t2.micro by default.

Requirements

• terraform
• AWS CLI configured for a terraform user

Procedure

1. Initiate terraform

   terraform init

2. Make a Plan to view resources being provisioned

   terraform plan

3. Provision planned infrastructure (-auto-approve executes the command in non-interactive mode)

   terraform apply -auto-approve

   Terraform will write files containing the node ssh key and all the IPs of the nodes both public and private. You'll find them in the ansible folder

4. Run the playbook for the master node. Initialize kubeadm.

cd ansible
ansible-playbook -i hosts/kube-cluster-master-host master-playbook.yaml --private-key kube-cluster-key -u ubuntu

Copy the $MASTER_PRIVATE_IP, $TOKEN and $HASH from the result of the ansible playbook. (Do not include the <:port> for the master_private_ip)

5. Run the playbook to connect the worker nodes to the master node. Set the copied variables and pass them to the command below.

cd ansible
ansible-playbook -i hosts/kube-cluster-worker-host worker-playbook.yaml -e MASTER_PRIVATE_IP=10.16.1.126 -e TOKEN=bits3u.1pqsfufvnahyulk7 -e HASH=sha256:0391ba7564ef13cbe1aa59f80daf912860f1076d3a42c66af4da42686d48e715 --private-key kube-cluster-key -u ubuntu

Make sure to replace the variables in the above command with the copied results from step 4.

6. SSH into the master node and check if the nodes are up

cd ansible
ssh -i kube-cluster-key ubuntu@<MASTER_PUBLIC_IP>
kubectl get nodes

Result after connecting worker nodes