Skip to content
This documentation applies to Codacy Self-hosted v8.0.0

For the latest updates and improvements, see the latest Cloud documentation instead.

Creating an Amazon EKS cluster#

Follow the instructions below to set up an Amazon EKS cluster from scratch, including the necessary underlying infrastructure, using Terraform.

The following diagram is a non-exhaustive overview of what you can expect to have deployed in your AWS account by using this quickstart guide.

Codacy Amazon EKS quickstart

1. Prepare your environment#

Prepare your environment to set up the Amazon EKS cluster:

  1. Make sure that you have the following tools installed on your machine:

  2. Set up the AWS CLI credentials for your AWS account using the AWS CLI and Terraform documentation as reference.

    Note that, as stated on the Terraform documentation, if your .aws/credentials are more complex you might need to set AWS_SDK_LOAD_CONFIG=1 for Terraform to work correctly:

    export AWS_SDK_LOAD_CONFIG=1
    
  3. Clone the Codacy chart repository and change to the directory that includes the provided Terraform configuration files:

    git clone https://github.com/codacy/chart.git
    cd chart/docs/infrastructure/EKS/
    

    This folder includes the following infrastructure stacks:

    • backend: Optional S3 bucket for storing the Terraform state and a DynamoDB table for state locking
    • main: Amazon EKS cluster, including the setup of all network and node infrastructure to go from zero to a fully functional cluster

    You must have administration privileges on AWS to deploy (and eventually destroy) this infrastructure. The policy file aws-terraform-minimum-admin-policy.json lists the minimum privileges that are required.

2. Set up the Terraform state storage backend#

The backend stores the current and historical state of your infrastructure.

Although using the backend is optional, we recommend that you deploy it, particularly if you're planning to use these Terraform templates to make modifications to the cluster in the future:

  1. Initialize Terraform and deploy the infrastructure described in the backend/ directory, then follow Terraform's instructions:

    cd backend/
    terraform init && terraform apply
    

    This creates an Amazon S3 bucket with a unique name to save the infrastructure state.

  2. Take note of the value of state_bucket_name in the output of the command.

  3. Edit the main/config.tf file and follow the instructions included in the comments to set the name of the Amazon S3 bucket created above and enable the use of the backend in those infrastructure stacks.

3. Create a vanilla Amazon EKS cluster#

Create a cluster that includes all the required network and node setup:

  1. Initialize Terraform and deploy the infrastructure described in the main/ directory, then follow Terraform's instructions:

    cd ../main/
    terraform init && terraform apply
    

    This process takes around 10 minutes.

  2. Consider if you want to tailor the cluster to your needs by customizing the cluster configuration.

    The cluster configuration (such as the type and number of nodes, network CIDRs, etc.) is exposed as variables in the main/variables.tf file.

    To customize the defaults of that file we recommend that you use a variable definitions file and set the variables in a file named terraform.tfvars in the directory main/. The following is an example terraform.tfvars:

    some_key = "a_string_value"
    another_key = 3
    someting_else = true
    

    Subsequently running terraform apply loads the variables in the terraform.tfvars file by default:

    terraform apply
    
  3. Set up the kubeconfig file that stores the information needed by kubectl to connect to the new cluster by default:

    aws eks update-kubeconfig --name codacy-cluster --alias codacy-cluster
    
  4. Get information about the pods in the cluster to test that the cluster was created and that kubectl can successfully connect to the cluster:

    kubectl get pods -A
    

4. Prepare to set up the Ingress Controller#

Prepare your infrastructure for the Ingress Controller setup, which is performed later during the installation process:

  1. Make sure that your network resources are correctly tagged, and create the following required tags if they are missing:

    Resource Type Key = Value
    VPC kubernetes.io/cluster/codacy-cluster = shared
    Subnet (public) kubernetes.io/cluster/codacy-cluster = shared
    kubernetes.io/role/elb = 1
    Subnet (private) kubernetes.io/cluster/codacy-cluster = shared
    kubernetes.io/role/internal-elb = 1

    For more information refer to the AWS documentation.

  2. Add the following chart repositories to Helm:

    helm repo add stable https://charts.helm.sh/stable
    helm repo update
    

5. Install the NGINX Ingress Controller#

Install the NGINX Ingress Controller:

  1. Download the configuration file values-nginx.yaml for the NGINX Ingress Controller.

    If you wish to use a private load balancer or restrict the IP range for the provisioned load balancer edit the file and enable the required annotation and/or the corresponding setting where indicated.

  2. Install the NGINX Ingress Controller:

    kubectl create namespace codacy
    helm upgrade --install --namespace codacy --version 1.39.0 codacy-nginx-ingress stable/nginx-ingress -f values-nginx.yaml
    

Uninstalling the Amazon EKS cluster#

Warning

If you proceed beyond this point you'll permanently delete and break things.

  1. Delete the Kubernetes cluster.

    Run the following command in the main/ directory:

    terraform destroy
    

    This process takes around 10 minutes.

  2. Remove the Terraform backend.

    If you created the Terraform backend with the provided stack you can now safely delete it.

    The backend is purposely created with extra settings to prevent its accidental destruction. To destroy it cleanly you must first disable these settings by editing the file backend/state_and_lock.tf and following the instructions included in the comments.

    Afterwards, run the following command in the backend/ directory:

    terraform apply && terraform destroy
    

    Note that you first have to run terraform apply to update the settings, and only then will terraform destroy be able to destroy the backend.

Share your feedback 📢

Did this page help you?

Thanks for the feedback! Is there anything else you'd like to tell us about this page?

We're sorry to hear that. Please let us know what we can improve:

Alternatively, you can create a more detailed issue on our GitHub repository.

Thanks for helping improve the Codacy documentation.

If you have a question or need help please contact support@codacy.com.

Last modified April 20, 2022