Terraform Automate EKS deployment

Deploy EKS with Terraform

Purpose

This repository provides a deployment of an EKS cluster and its associated infrastructure (VPC, SG, S3, etc.) in an automated way with GitHub Action.

It is possible to use this code to quickly deploy a cloud native environment (K8s) for demonstration purposes.

High Level Diagram

Sequence Diagram

Roll out Lab Deployment

Requirements

• GitHub Account to create private fork
• AWS Subscription with Programmatic Access (AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY )
• SSH Key

Configure GitHub Action Secrets

Open Settings > Security > Secrets and variables > Actions

• Secrets
  • AWS_ACCESS_KEY_ID
  • AWS_SECRET_ACCESS_KEY
  • SSH_PRIVATE_KEY (Private key yes, will be used by Ansible to connect to bastion instance to install tools)

Deploy Infrastructure Workflow

1. Open Actions > All Workflows > Deploy Infrastructure

2. Click on Run workflow , and set your configuration

   

• AWS region in which to deploy the infrastructure (Mandatory) : default = us-east-1

• Prefix name for resources to be created: s3 bucket, vpc, eks, ec2, dev, lab, etc : default = eks

  (all AWS resources created will be prefix, don't worry if you already have resources with this prefix, a random character string is also generated as a suffix, under the hood in the code, so each deployment will have a unique name, even if the same prefix is used several time)

• Number of worker nodes in eks_cluster : default = 3 (choice between 1 to 5)

• Worker nodes type in eks_cluster : default = t3.medium

• Public ssh key to connect to ec2-bastion host : Your SSH PUBLIC KEY

3. Click Run workflow, and deployment will start

4. You can monitor the deployment of all stages

   

NB: click on the stage to have details ie Ansible stage below

5. Retrieve the terraform outputs to connect to the instance by clicking on theDeploy infra - Terraform Apply job and scrolling to the bottom of this stage.

🎉 Congratulations your AWS Environment is now deployed, and your EKS cluster is running 🎉

Connect to your EKS and deploy application

The kubernetes APIs (kube-apiserver) are not publicly accessible, so you need to connect to the bastion servers via SSH (see terraform output above to retrieve the IP).

All tools are already deployed on the bastion instance by Ansible at boot time.

EKS public API server endpoint is disabled, only EKS private API server endpoint is enabled

It is only possible to connect to the Kubernetes cluster APIs via the ec2-bastion instance.

1. Connect to ec2-bastion with your ssh private key and user Ubuntu and IP retrieve from TF output (ec2instance_ip)

   ssh -i <ssh_private_key> ubuntu@${TF_output_ec2instance_ip}

2. Configure your AWS account (example you can use aws configure cli command)

3. Retrieve the name of your cluster (either from the terraform output or via aws cli)

   aws eks list-clusters

   

4. Update Kubeconfig

   aws eks update-kubeconfig --name ${eks_cluster_name}

   Example:

   

5. You can now manage your cluster and deploy applications (with kubectl by example)

   

6. Optional - Test - deploy application

   kubectl run weather-alpine --image=chrisley75/weather-alpine:0.2 -l app=weather-alpine --expose --port=8080
   kubectl patch svc weather-alpine -p '{"spec": {"type": "LoadBalancer"}}'
   kubectl get all
   NAME                 READY   STATUS    RESTARTS   AGE
   pod/weather-alpine   1/1     Running   0          96s
   
   NAME                     TYPE           CLUSTER-IP     EXTERNAL-IP                                                               PORT(S)          AGE
   service/kubernetes       ClusterIP      172.20.0.1     <none>                                                                    443/TCP          6h18m
   service/weather-alpine   LoadBalancer   172.20.69.39   a9e504e4734c44212bd738f8dc3a302e-1865608803.eu-west-3.elb.amazonaws.com   8080:31471/TCP   96s

🎉 Congratulations your EKS Cluster is fully functional. 🎉

Delete Infrastructure Workflow

1. Open Actions > All Workflows > Delete Infrastructure

2. Click on Run workflow , and **fill in the fields **

• AWS region in which the infrastructure to delete is deployed : default = us-east-1 (the region where is deployed your AWS infra)
• Prefix name for resources to be delete: s3 bucket, vpc, eks, ec2, etc. : default = eks (prefix set during build stage)

3. Click Run workflow, and delete will start