Using EKS and Gitlab CI to deploy applications

Setup EKS using Terraform and Ansible. Deploy applications on EKS using Gitlab CI and Helm.


The main goal is to have a production-ready environment, showcasing AWS architecture, Terraform, Ansible, Kubernetes (EKS), Gitlab CI, DockerHub and Helm.

Key aspects:

  • All resources are managed in code. Even the bootstrap of the project .

  • HA where applicable (EKS Load Balancer)

  • The application is deployed from code.

General overview

The process described on this tutorial show how to:

  1. Terraform: leverage an S3 bucket to store states (here);

  2. Terraform: leverage a DynamoDB table to store locks (here);

  3. Terraform: leverage a Virtual Machine (here);

  4. Ansible: setup VM swap, hostname and packages. Also, create a Docker container which runs the EKS setup process, setup kubeconfig and Helm (here).

  5. Gitlab CI: run a pipeline to build (here) a Docker image and push to DockerHub (here);

  6. Gitlab CI: deploy the Docker image on K8s using Helm (here and here).

Stack overview

  • AWS

  • Terraform

  • Ansible

  • Gitlab CI

  • Dockerhub

  • EKS

  • K8s

  • Helm

Before you begin

AWS account

Make sure you have an AWS account. AWS will host the Kubernetes cluster (EKS) and a small Virtual Machine.

This tutorial does not fit the AWS free tier. Make sure you delete all resources to avoid charges.

Gitlab account

Make sure you have a Gitlab account to commit your code and use pipelines.


The stack bootstrap is done using Docker, so make sure you have Docker installed on your workstation.

Provisioning the infrastructure


You need to create an IAM user which will be used with Terraform.

Login to you AWS console, go to Services, IAM.

Go to Users, Add user.

Add a user called iac (stands for Infra as Code) with Programmatic Access.

Attach AdministratorAccess and click on Next: Tags button.

You can use restricted policies if you want. In this example we are using AdministratorAccess to keep it simple.

Optionally add tags then click on Next: Review.

Review and create user.

Click on Show in Secret access key section. Copy and save in a safe place your Access key ID and Secret access key.

The credentials are displayed just once. Save them now or you will have to create a new user.

From now on, the Access key ID and Secret access key will be referenced in this tutorial as iac IAM user credentials.

Terraform apply

Clone or fork eintopf repository (Terraform and Ansible scripts):

Setup AWS credentials

Follow the instructions here:

Use your iac IAM user credentials.

Run Terraform scripts

Once your have your AWS credentials setup, run the Terraform scripts.

To do so, follow the instructions here:

At this point, if you go to you AWS console, you should have:

  • An S3 bucket;

  • A DynamoDB table;

  • A Virtual Machine.

Setup the infrastructure

eintopf also has Ansible playbooks and roles do setup the infrastructure. The playbook playbooks/mgmt.yaml setup a Virtual Machine with a Docker container. This container is responsible for setting up the EKS cluster.

First, make sure you have ansible-vault configured:

Then, setup the infrastructure:

At this point, if you go to you AWS console, you should have:

  • An EKS cluster.

Also, the Virtual Machine called mgmt hosts a Docker container called devops. Inside this container you will find you Kubernetes cluster kubeconfig file, which is used with kubectl and also must be configured on your Gitlab CI variables to run the deployment test. More info here:

You can copy the kubeconfig file content and paste and save it in a safe place, specially if you want o use kubectl from your workstation and not form the VM.

To convert the kubeconfig content in the format to be used on Gitlab CI variable, run:

cat /root/.kube/config | base64

Test a deployment

Clone or fork the following repo:

Setup the required variables to run the pipeline:

Push a change and watch the deployment logs.

Last updated