• Infrastructure as Code (IaC): using Terraform, you get an infrastructure that’s smooth and efficient
• State management: Terraform saves the current infrastructure state, so you can review further changes without applying them. Also, state can be stored remotely, so you can work on the infrastructure in a team
• Scalability and flexibility: the infrastructure built based on this boilerplate can be expanded and updated anytime
• Comprehensiveness: you get scaling and monitoring instruments along with the basic infrastructure. You don’t need to manually modify anything in the infrastructure; you can simply make changes in Terraform as needed and deploy them to AWS and Kubernetes
• Control over resources: the IaC approach makes the infrastructure more observable and prevents waste of resources
• Clear documentation: your Terraform code effectively becomes your project documentation. It means that you can add new members to the team, and it won’t take them too much time to figure out how the infrastructure works

• Safe and polished: we’ve used these solutions in our own large-scale, high-load projects. We’ve been perfecting this infrastructure building process for months, making sure that it results in a system that is safe to use, secure, and reliable
• Saves time: you can spend weeks doing your own research and making the unavoidable mistakes to build an infrastructure like this. Instead, you can rely on this boilerplate and create the infrastructure you need within a day
• It’s free: we’re happy to share the results of our work

This repository contains terraform modules and configuration of the Mad Devs team for the rapid deployment of a Kubernetes cluster, supporting services, and the underlying infrastructure in the AWS.

In our company’s work, we have tried many infrastructure solutions and services and traveled the path from on-premise hardware to serverless. As of today, Kubernetes has become our standard platform for deploying applications, and AWS has become the main cloud.

It is worth noting here that although 90% of our and our clients’ projects are hosted on AWS and AWS EKS is used as the Kubernetes platform, we do not insist, do not drag everything to Kubernetes, and do not force anyone to be hosted on AWS. Kubernetes is offered only after the collection and analysis of service architecture requirements.

And then, when choosing Kubernetes, it makes almost no difference to applications how the cluster itself is created—manually, through kops or using managed services from cloud providers—in essence, the Kubernetes platform is the same everywhere. So the choice of a particular provider is then made based on additional requirements, expertise, etc.

We know that the current implementation is far from being perfect. For example, we deploy services to the cluster using terraform: it is rather clumsy and against the Kuber approaches, but it is convenient for bootstrap because, by using state and interpolation, we convey proper IDs, ARNs, and other attributes to resources and names or secrets to templates and generate values ​​from them for the required charts all within terraform.

There are more specific drawbacks: the data "template_file" resources that we used for most templates are extremely inconvenient for development and debugging, especially if there are 500+ line rolls like terraform/layer2-k8s/templates/elk-values.yaml. Also, despite helm3 got rid of the tiller, a large number of helm releases still at some point leads to plan hanging. Partially, but not always, it can be solved by terraform apply -target, but for the consistency of the state, it is desirable to execute plan and apply on the entire configuration. If you are going to use this boilerplate, it is advisable to split the terraform/layer2-k8s layer into several ones, taking out large and complex releases into separate modules.

You may reasonably question the number of .tf files. This monolith certainly should be refactored and split into many micro-modules adopting terragrunt approach. This is exactly what we will do in the near future, solving along the way the problems described above.

You can find more about this project in Anton Babenko stream:

• Boilerplate for a basic AWS infrastructure with EKS cluster
  • Advantages of this boilerplate
  • Why you should use this boilerplate
  • Description
  • Table of contents
  • Architecture diagram
  • Current infrastructure cost
  • EKS Upgrading
  • Namespace structure in the K8S cluster
  • Useful tools
  • Useful VSCode extensions
  • AWS account
    • IAM settings
    • Setting up awscli
  • How to use this repo
    • Getting ready
      • S3 state backend
      • Inputs
      • Secrets
      • Domain and SSL
    • Working with terraform
      • init
      • plan
      • apply
    • terragrunt
      • Apply infrastructure by layers with terragrunt
      • Target apply by terragrunt
      • Destroy infrastructure by terragrunt
  • What to do after deployment
  • Update terraform version
  • Updated terraform providers
    • examples
  • TFSEC
  • Contributing

This diagram describes the default infrastructure:

• We use three availability Zones
• VPC
  • Three public subnets for resources that can be accessible from the Internet
    • Elastic load balancing - entry point to the k8s cluster
    • Internet gateway - entry point to the created VPC
    • Single Nat Gateway - service for organizing access for instances from private networks to public ones.
  • Three private subnets with Internet access via Nat Gateway
  • Three intra subnets without Internet access
  • Three private subnets for RDS
  • Route tables for private networks
  • Route tables for public networks
• Autoscaling groups
  • On-demand - a group with 1-5 on-demand instances for resources with continuous uptime requirements
  • Spot - a group with 1-6 spot instances for resources where interruption of work is not critical
  • CI - a group with 0-3 spot instances created based on gitlab-runner requests; located in the public network
• EKS control plane - nodes of the k8s clusters’ control plane
• Route53 - DNS management service
• Cloudwatch - service for obtaining the metrics about resources’ state of operation in the AWS cloud
• AWS Certificate manager - service for AWS certificate management
• SSM parameter store - service for storing, retrieving, and controlling configuration values
• S3 bucket - this bucket is used to store terraform state
• Elastic container registry - service for storing docker images

The cost is indicated without counting the amount of traffic for Nat Gateway Load Balancer and S3

To upgrade k8s cluster to a new version, please use official guide and check changelog/breaking changes. Starting from v1.18 EKS supports K8S add-ons. We use them to update things like vpc-cni, kube-proxy, coredns. To get the latest add-ons versions, run:

aws eks describe-addon-versions --kubernetes-version 1.21 --query 'addons[].[addonName, addonVersions[0].addonVersion]'

where 1.21 - is a k8s version on which we are updating. DO NOT FORGET!!! to update cluster-autoscaler too. It's version must be the same as the cluster version. Also IT'S VERY RECOMMENDED to check that deployed objects have actual apiVersions that won't be deleted after upgrading. There is a tool pluto that can help to do it.

Switch to the correct cluster
Run `pluto detect-helm -o markdown --target-versions k8s=v1.22.0`, where `k8s=v1.22.0` is a k8s version we want to update to.

This diagram shows the namespaces used in the cluster and the services deployed there

• tfenv - tool for managing different versions of terraform; the required version can be specified directly as an argument or via .terraform-version
• tgenv - tool for managing different versions of terragrunt; the required version can be specified directly as an argument or via .terragrunt-version
• terraform - terraform itself, our main development tool: tfenv install
• awscli - console utility to work with AWS API
• kubectl - conssole utility to work with Kubernetes API
• kubectx + kubens - power tools for kubectl help you switch between Kubernetes clusters and namespaces
• helm - tool to create application packages and deploy them into k8s
• helmfile - "docker compose" for helm
• terragrunt - small terraform wrapper providing DRY approach in some cases: tgenv install
• awsudo - simple console utility that allows running awscli commands assuming specific roles
• aws-vault - tool for securely managing AWS keys and running console commands
• aws-mfa - utility for automating the reception of temporary STS tockens when MFA is enabled
• vscode - our main IDE

Optionally, a pre-commit hook can be set up and configured for terraform: pre-commit-terraform, this will allow formatting and validating code at the commit stage

• editorconfig
• terraform
• drawio
• yaml
• embrace
• js-beautify
• docker
• git-extension-pack
• githistory
• kubernetes-tools
• markdown-preview-enhanced
• markdownlint
• file-tree-generator
• gotemplate-syntax

We will not go deep into security settings since everyone has different requirements. However, there are the simplest and most basic steps worth following to move on. If you have everything in place, feel free to skip this section.

It is highly recommended not to use a root account to work with AWS. Make an extra effort of creating users with required/limited rights.

So, you have created an account, passed confirmation, perhaps even created Access Keys for the console. In any case, go to your account security settings and be sure to follow these steps:

• Set a strong password
• Activate MFA for the root account
• Delete and do not create access keys of the root account

Further in the IAM console:

• In the Policies menu, create MFASecurity policy that prohibits users from using services without activating MFA

• In the Roles menu, create new role administrator. Select Another AWS Account - and enter your account number in the Account ID field. Check the Require MFA checkbox. In the next Permissions window, attach the AdministratorAccess policy to it.

• In the Policies menu, create assumeAdminRole policy:

  {
    "Version": "2012-10-17",
    "Statement": {
        "Effect": "Allow",
        "Action": "sts:AssumeRole",
        "Resource": "arn:aws:iam::<your-account-id>:role/administrator"
    }
  }

• In the Groups menu, create the admin group; in the next window, attach assumeAdminRole and MFASecurity policy to it. Finish creating the group.

• In the Users menu, create a user to work with AWS by selecting both checkboxes in Select AWS access type. In the next window, add the user to the admin group. Finish and download CSV with credentials.

In this doc, we haven't considered a more secure and correct method of user management that uses external Identity providers. These include G-suite, Okta, and others

• Terraform can work with environment variables for AWS access key ID and a secret access key or AWS profile; in this example, we will create an aws profile:

  $ aws configure --profile maddevs
  AWS Access Key ID [None]: *****************
  AWS Secret Access Key [None]: *********************
  Default region name [None]: us-east-1
  Default output format [None]: json

  $ export AWS_PROFILE=maddevs

• Go here to learn how to get temporary session tokens and assume role

• Alternatively, to use your awscli, terraform and other CLI utils with MFA and roles, you can use aws-mfa, aws-vault and awsudo

S3 is used as a backend for storing terraform state and for exchanging data between layers. You can manually create s3 bucket and then put backend setting into backend.tf file in each layer. Alternatively you can run from terraform/ directory:

$ export TF_REMOTE_STATE_BUCKET=my-new-state-bucket
$ terragrunt run-all init

File terraform/layer1-aws/demo.tfvars.example contains example values. Copy this file to terraform/layer1-aws/terraform.tfvars and set you values:

$ cp terraform/layer1-aws/demo.tfvars.example terraform/layer1-aws/terraform.tfvars

You can find all possible variables in each layer's Readme.

In the root of layer2-k8s is the aws-sm-secrets.tf where several local variables expect AWS Secrets Manager secret with the pattern /${local.name_wo_region}/infra/layer2-k8s. These secrets are used for authentication with Kibana and Grafana using GitLab and register gitlab runner.

{
  "kibana_gitlab_client_id": "access key token",
  "kibana_gitlab_client_secret": "secret key token",
  "kibana_gitlab_group": "gitlab group",
  "grafana_gitlab_client_id": "access key token",
  "grafana_gitlab_client_secret": "secret key token",
  "gitlab_registration_token": "gitlab-runner token",
  "grafana_gitlab_group": "gitlab group",
  "alertmanager_slack_url": "slack url",
  "alertmanager_slack_channel": "slack channel"
}

Set proper secrets; you can set empty/mock values. If you won't use these secrets, delete this .tf file from the layer2-k8s root.

You will need to purchase or use an already purchased domain in Route53. The domain name and zone ID will need to be set in the domain_name and zone_id variables in layer1.

By default, the variable create_acm_certificate is set to false. Which instructs terraform to search ARN of an existing ACM certificate. Set to true if you want terraform to create a new ACM SSL certificate.

The terraform init command is used to initialize the state and its backend, downloads providers, plugins, and modules. This is the first command to be executed in layer1 and layer2:

$ terraform init

Correct output:

* provider.aws: version = "~> 2.10"
* provider.local: version = "~> 1.2"
* provider.null: version = "~> 2.1"
* provider.random: version = "~> 2.1"
* provider.template: version = "~> 2.1"

Terraform has been successfully initialized!

The terraform plan command reads terraform state and configuration files and displays a list of changes and actions that need to be performed to bring the state in line with the configuration. It's a convenient way to test changes before applying them. When used with the -out parameter, it saves a batch of changes to a specified file that can later be used with terraform apply. Call example:

$ terraform plan
# ~600 rows skipped
Plan: 82 to add, 0 to change, 0 to destroy.

------------------------------------------------------------------------

Note: You didn't specify an "-out" parameter to save this plan, so Terraform
can't guarantee that exactly these actions will be performed if
"terraform apply" is subsequently run.

The terraform apply command scans .tf in the current directory and brings the state to the configuration described in them by making changes in the infrastructure. By default, plan with a continuation dialog is performed before applying. Optionally, you can specify a saved plan file as input:

$ terraform apply
# ~600 rows skipped
Plan: 82 to add, 0 to change, 0 to destroy.

Do you want to perform these actions?
  Terraform will perform the actions described above.
  Only 'yes' will be accepted to approve.

  Enter a value: yes

Apply complete! Resources: 82 added, 0 changed, 0 destroyed.

We do not always need to re-read and compare the entire state if small changes have been added that do not affect the entire infrastructure. For this, you can use targeted apply; for example:

$ terraform apply -target helm_release.kibana

Details can be found here

The first time, the apply command must be executed in the layers in order: first layer1, then layer2. Infrastructure destroy should be done in the reverse order.

• Terragrunt version: 0.29.2

Terragrunt version pinned in terragrunt.hcl file.

We've also used terragrunt to simplify s3 bucket creation and terraform backend configuration. All you need to do is to set s3 bucket name in the TF_REMOTE_STATE_BUCKET env variable and run terragrunt command in the terraform/ directory:

$ export TF_REMOTE_STATE_BUCKET=my-new-state-bucket
$ terragrunt run-all init
$ terragrunt run-all apply

By running this terragrunt will create s3 bucket, configure terraform backend and then will run terraform init and terraform apply in layer-1 and layer-2 sequentially.

Go to layer folder terraform/layer1-aws/ or terraform/layer2-k8s/ and run this command:

terragrunt apply

The layer2-k8s has a dependence on layer1-aws.

Go to layer folder terraform/layer1-aws/ or terraform/layer2-k8s/ and run this command:

terragrunt apply -target=module.eks

The -target is formed from the following parts resource type and resource name. For example: -target=module.eks, -target=helm_release.loki_stack

To destroy both layers, run this command from terraform/ folder:

terragrant run-all destroy

To destroy layer2-k8s, run this command from terraform/layare2-k8s folder:

terragrunt destroy

The layer2-k8s has dependence from layer1-aws and when you destroy layer1-aws, layer2-k8s destroyed automatically.

After applying this configuration, you will get the infrastructure described and outlined at the beginning of the document. In AWS and within the EKS cluster, the basic resources and services necessary for the operation of the EKS k8s cluster will be created.

You can get access to the cluster using this command:

aws eks update-kubeconfig --name maddevs-demo-use1 --region us-east-1

Change terraform version in this files

terraform/.terraform-version - the main terraform version for tfenv tool

.github/workflows/terraform-ci.yml - the terraform version for github actions need for terraform-validate and terraform-format.

Terraform version in each layer.

terraform/layer1-aws/main.tf
terraform/layer2-k8s/main.tf

Change terraform providers version in this files

terraform/layer1-aws/main.tf
terraform/layer2-k8s/main.tf

When we changed terraform provider versions, we need to update terraform state. For update terraform state in layers we need to run this command:

terragrunt run-all init -upgrade

Or in each layer run command:

terragrunt init -upgrade

Each layer has an examples/ directory that contains working examples that expand the basic configuration. The files’ names and contents are in accordance with our coding conventions, so no additional description is required. If you need to use something, just move it from this folder to the root of the layer.

This will allow you to expand your basic functionality by launching a monitoring system based on ELK or Prometheus Stack, etc.

We use GitHub Actions and tfsec to check our terraform code using static analysis to spot potential security issues. However, we needed to skip some checks. The list of those checks is below:

If you're interested in contributing to the project:

• Start by reading the Contributing guide
• Explore current issues.