terraform-aws-eks-cluster

Terraform module to provision an EKS cluster on AWS.

This project is part of our comprehensive "SweetOps" approach towards DevOps.

It's 100% Open Source and licensed under the APACHE2.

We literally have hundreds of terraform modules that are Open Source and well-maintained. Check them out!

Introduction

The module provisions the following resources:

• EKS cluster of master nodes that can be used together with the terraform-aws-eks-workers, terraform-aws-eks-node-group and terraform-aws-eks-fargate-profile modules to create a full-blown cluster
• IAM Role to allow the cluster to access other AWS services
• Optionally, the module creates and automatically applies an authentication ConfigMap (aws-auth) to allow the worker nodes to join the cluster and to add additional users/roles/accounts. (This option is enabled by default, but has some caveats noted below. Set apply_config_map_aws_auth to false to avoid these issues.)

NOTE: Release 2.0.0 (previously released as version 0.45.0) contains some changes that could result in your existing EKS cluster being replaced (destroyed and recreated). To prevent this, follow the instructions in the v1 to v2 migration path.

NOTE: Every Terraform module that provisions an EKS cluster has faced the challenge that access to the cluster is partly controlled by a resource inside the cluster, a ConfigMap called aws-auth. You need to be able to access the cluster through the Kubernetes API to modify the ConfigMap, because there is no AWS API for it. This presents a problem: how do you authenticate to an API endpoint that you have not yet created?

We use the Terraform Kubernetes provider to access the cluster, and it uses the same underlying library that kubectl uses, so configuration is very similar. However, every kind of configuration we have tried has failed at some point.

• An authentication token can be retrieved using the aws_eks_cluster_auth data source. This works as long as the token does not expire while Terraform is running, and the token is refreshed during the "plan" phase before trying to refresh the state, and the token does not expire in the interval between "plan" and "apply". Unfortunately, failures of all these types have been seen. Nevertheless, this is the only method that is compatible with Terraform Cloud, so it is the default. It is the only method we fully support until AWS provides an API for managing aws-auth.
• After creating the EKS cluster, you can generate a KUBECONFIG file that configures access to it. This works most of the time, but if the file was present and used as part of the configuration to create the cluster, and then the file gets deleted (as would happen in a CI system like Terraform Cloud), Terraform would not cause the file to be regenerated in time to use it to refresh Terraform's state and the "plan" phase will fail. So any KUBECONFIG file has to be managed separately.
• An authentication token can be retrieved on demand by using the exec feature of the Kubernetes provider to call aws eks get-token. This requires that the aws CLI be installed and available to Terraform and that it has access to sufficient credentials to perform the authentication and is configured to use them. When those conditions are met, this is the most reliable method, and the one Cloud Posse prefers to use. However, since it has these requirements that are not always easily met, it is not the default method and it is not fully supported.

All of the above methods can face additional challenges when using terraform import to import resources into the Terraform state. The KUBECONFIG file method is the only sure way to import resources, due to Terraform limitations on providers. You will need to create the file, of course, but that is easily done with aws eks update-kubeconfig. Depending on the situation, you may also be able to import resources by setting -var apply_config_map_aws_auth=false during import.

At the moment, the exec option appears to be the most reliable method, so we recommend using it if possible, but because of the extra requirements it has, we use the data source as the default authentication method.

Additional Note: All of the above methods require network connectivity between the host running the terraform command and the EKS endpoint. If your EKS cluster does not have public access enabled, this means you need to take extra steps, such as using a VPN to provide access to the private endpoint, or running terraform on a host in the same VPC as the EKS cluster.

Failure during destroy: If the cluster is destroyed (via Terraform or otherwise) before the Terraform resource responsible for the aws-auth ConfigMap is destroyed, Terraform will get stuck trying to delete the ConfigMap, because it cannot contact the now destroyed cluster. This can show up as a connection refused error (usually to https://localhost/). The easiest ways to handle this is either to add -var apply_config_map_aws_auth=false to the destroy command or to remove the ConfigMap (...kubernetes_config_map.aws_auth[0]) from the Terraform state with terraform state rm.

NOTE: We give you the kubernetes_config_map_ignore_role_changes option and default it to true for the following reasons:

• We provision the EKS cluster
• Then we wait for the cluster to become available (see null_resource.wait_for_cluster in auth.tf
• Then we provision the Kubernetes Auth ConfigMap to map and add additional roles/users/accounts to Kubernetes groups
• That is all we do in this module, but after that, we expect you to use terraform-aws-eks-node-group to provision a managed Node Group
• Then EKS updates the Auth ConfigMap and adds worker roles to it (for the worker nodes to join the cluster)
• Since the ConfigMap is modified outside of Terraform state, Terraform wants to update it to to remove the worker roles EKS added
• If you update the ConfigMap without including the worker nodes that EKS added, you will disconnect them from the cluster

However, it is possible to get the worker node roles from the terraform-aws-eks-node-group via Terraform "remote state" and include them with any other roles you want to add (example code to be published later), so we make ignoring the role changes optional. (This is what we do for Cloud Posse clients.) If you do not ignore changes then you will have no problem with making future intentional changes.

The downside of having kubernetes_config_map_ignore_role_changes set to true is that if you later want to make changes, such as adding other IAM roles to Kubernetes groups, you cannot do so via Terraform, because the role changes are ignored. Because of Terraform restrictions, you cannot simply change kubernetes_config_map_ignore_role_changes from true to false, apply changes, and set it back to true again. Terraform does not allow the "ignore" settings to be changed on a resource, so kubernetes_config_map_ignore_role_changes is implemented as 2 different resources, one with ignore settings and one without. If you want to switch from ignoring to not ignoring, or vice versa, you must manually move the aws_auth resource in the terraform state. Change the setting of kubernetes_config_map_ignore_role_changes, run terraform plan, and you will see that an aws_auth resource is planned to be destroyed and another one is planned to be created. Use terraform state mv to move the destroyed resource to the created resource "address", something like

terraform state mv 'module.eks_cluster.kubernetes_config_map.aws_auth_ignore_changes[0]' 'module.eks_cluster.kubernetes_config_map.aws_auth[0]'

Then run terraform plan again and you should see only your desired changes made "in place". After applying your changes, if you want to set kubernetes_config_map_ignore_role_changes back to true, you will again need to use terraform state mv to move the auth-map back to its old "address".

Security & Compliance

Security scanning is graciously provided by Bridgecrew. Bridgecrew is the leading fully hosted, cloud-native solution providing continuous Terraform security and compliance.

Benchmark	Description
	Infrastructure Security Compliance
	Center for Internet Security, KUBERNETES Compliance
	Center for Internet Security, AWS Compliance
	Center for Internet Security, AZURE Compliance
	Payment Card Industry Data Security Standards Compliance
	National Institute of Standards and Technology Compliance
	Information Security Management System, ISO/IEC 27001 Compliance
	Service Organization Control 2 Compliance
	Center for Internet Security, GCP Compliance
	Health Insurance Portability and Accountability Compliance

Usage

IMPORTANT: We do not pin modules to versions in our examples because of the difficulty of keeping the versions in the documentation in sync with the latest released versions. We highly recommend that in your code you pin the version to the exact version you are using so that your infrastructure remains stable, and update versions in a systematic way so that they do not catch you by surprise.

For a complete example, see examples/complete.

For automated tests of the complete example using bats and Terratest (which tests and deploys the example on AWS), see test.

Other examples:

• terraform-aws-components/eks/cluster - Cloud Posse's service catalog of "root module" invocations for provisioning reference architectures

  provider "aws" {
    region = var.region
  }

  module "label" {
    source = "cloudposse/label/null"
    # Cloud Posse recommends pinning every module to a specific version
    # version  = "x.x.x"

    namespace  = var.namespace
    name       = var.name
    stage      = var.stage
    delimiter  = var.delimiter
    attributes = ["cluster"]
    tags       = var.tags
  }

  locals {
    # Prior to Kubernetes 1.19, the usage of the specific kubernetes.io/cluster/* resource tags below are required
    # for EKS and Kubernetes to discover and manage networking resources
    # https://www.terraform.io/docs/providers/aws/guides/eks-getting-started.html#base-vpc-networking
    tags = { "kubernetes.io/cluster/${module.label.id}" = "shared" }
  }

  module "vpc" {
    source = "cloudposse/vpc/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version     = "x.x.x"
    cidr_block = "172.16.0.0/16"

    tags    = local.tags
    context = module.label.context
  }

  module "subnets" {
    source = "cloudposse/dynamic-subnets/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version     = "x.x.x"

    availability_zones   = var.availability_zones
    vpc_id               = module.vpc.vpc_id
    igw_id               = module.vpc.igw_id
    cidr_block           = module.vpc.vpc_cidr_block
    nat_gateway_enabled  = true
    nat_instance_enabled = false

    tags    = local.tags
    context = module.label.context
  }

  module "eks_node_group" {
    source = "cloudposse/eks-node-group/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version     = "x.x.x"

    instance_types                     = [var.instance_type]
    subnet_ids                         = module.subnets.public_subnet_ids
    health_check_type                  = var.health_check_type
    min_size                           = var.min_size
    max_size                           = var.max_size
    cluster_name                       = module.eks_cluster.eks_cluster_id

    # Enable the Kubernetes cluster auto-scaler to find the auto-scaling group
    cluster_autoscaler_enabled = var.autoscaling_policies_enabled

    context = module.label.context

    # Ensure the cluster is fully created before trying to add the node group
    module_depends_on = module.eks_cluster.kubernetes_config_map_id
  }

  module "eks_cluster" {
    source = "cloudposse/eks-cluster/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version = "x.x.x"

    vpc_id     = module.vpc.vpc_id
    subnet_ids = module.subnets.public_subnet_ids

    kubernetes_version    = var.kubernetes_version
    oidc_provider_enabled = true

    addons = [
      // https://docs.aws.amazon.com/eks/latest/userguide/managing-vpc-cni.html#vpc-cni-latest-available-version
      {
        addon_name               = "vpc-cni"
        addon_version            = var.vpc_cni_version
        resolve_conflicts        = "NONE"
        service_account_role_arn = null
      },
      // https://docs.aws.amazon.com/eks/latest/userguide/managing-kube-proxy.html
      {
        addon_name               = "kube-proxy"
        addon_version            = var.kube_proxy_version
        resolve_conflicts        = "NONE"
        service_account_role_arn = null
      },
      // https://docs.aws.amazon.com/eks/latest/userguide/managing-coredns.html
      {
        addon_name               = "coredns"
        addon_version            = var.coredns_version
        resolve_conflicts        = "NONE"
        service_account_role_arn = null
      },
    ]
    addons_depends_on = [module.eks_node_group]

    context = module.label.context

    cluster_depends_on = [module.subnets]
  }

Module usage with two unmanaged worker groups:

  locals {
    # Unfortunately, the `aws_ami` data source attribute `most_recent` (https://github.com/cloudposse/terraform-aws-eks-workers/blob/34a43c25624a6efb3ba5d2770a601d7cb3c0d391/main.tf#L141)
    # does not work as you might expect. If you are not going to use a custom AMI you should
    # use the `eks_worker_ami_name_filter` variable to set the right kubernetes version for EKS workers,
    # otherwise the first version of Kubernetes supported by AWS (v1.11) for EKS workers will be selected, but
    # EKS control plane will ignore it to use one that matches the version specified by the `kubernetes_version` variable.
    eks_worker_ami_name_filter = "amazon-eks-node-${var.kubernetes_version}*"
  }

  module "eks_workers" {
    source = "cloudposse/eks-workers/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version     = "x.x.x"

    attributes                         = ["small"]
    instance_type                      = "t3.small"
    eks_worker_ami_name_filter         = local.eks_worker_ami_name_filter
    vpc_id                             = module.vpc.vpc_id
    subnet_ids                         = module.subnets.public_subnet_ids
    health_check_type                  = var.health_check_type
    min_size                           = var.min_size
    max_size                           = var.max_size
    wait_for_capacity_timeout          = var.wait_for_capacity_timeout
    cluster_name                       = module.label.id
    cluster_endpoint                   = module.eks_cluster.eks_cluster_endpoint
    cluster_certificate_authority_data = module.eks_cluster.eks_cluster_certificate_authority_data
    cluster_security_group_id          = module.eks_cluster.eks_cluster_managed_security_group_id

    # Auto-scaling policies and CloudWatch metric alarms
    autoscaling_policies_enabled           = var.autoscaling_policies_enabled
    cpu_utilization_high_threshold_percent = var.cpu_utilization_high_threshold_percent
    cpu_utilization_low_threshold_percent  = var.cpu_utilization_low_threshold_percent

    context = module.label.context
  }

  module "eks_workers_2" {
    source = "cloudposse/eks-workers/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version     = "x.x.x"

    attributes                         = ["medium"]
    instance_type                      = "t3.medium"
    eks_worker_ami_name_filter         = local.eks_worker_ami_name_filter
    vpc_id                             = module.vpc.vpc_id
    subnet_ids                         = module.subnets.public_subnet_ids
    health_check_type                  = var.health_check_type
    min_size                           = var.min_size
    max_size                           = var.max_size
    wait_for_capacity_timeout          = var.wait_for_capacity_timeout
    cluster_name                       = module.label.id
    cluster_endpoint                   = module.eks_cluster.eks_cluster_endpoint
    cluster_certificate_authority_data = module.eks_cluster.eks_cluster_certificate_authority_data
    cluster_security_group_id          = module.eks_cluster.eks_cluster_managed_security_group_id

    # Auto-scaling policies and CloudWatch metric alarms
    autoscaling_policies_enabled           = var.autoscaling_policies_enabled
    cpu_utilization_high_threshold_percent = var.cpu_utilization_high_threshold_percent
    cpu_utilization_low_threshold_percent  = var.cpu_utilization_low_threshold_percent

    context = module.label.context
  }

  module "eks_cluster" {
    source = "cloudposse/eks-cluster/aws"
    # Cloud Posse recommends pinning every module to a specific version
    # version     = "x.x.x"

    vpc_id     = module.vpc.vpc_id
    subnet_ids = module.subnets.public_subnet_ids

    kubernetes_version    = var.kubernetes_version
    oidc_provider_enabled = false

    workers_role_arns          = [module.eks_workers.workers_role_arn, module.eks_workers_2.workers_role_arn]
    allowed_security_group_ids = [module.eks_workers.security_group_id, module.eks_workers_2.security_group_id]

    context = module.label.context
  }

Makefile Targets

Available targets:

  help                                Help screen
  help/all                            Display help for all targets
  help/short                          This help short screen
  lint                                Lint terraform code

Requirements

Providers

Modules

Resources

Inputs

list(object({
    addon_name               = string
    addon_version            = optional(string, null)
    configuration_values     = optional(string, null)
    resolve_conflicts        = string
    service_account_role_arn = optional(string, null)
    create_timeout           = optional(string, null)
    update_timeout           = optional(string, null)
    delete_timeout           = optional(string, null)
  }))

[
  "cluster"
]

[
  "secrets"
]

{
  "additional_tag_map": {},
  "attributes": [],
  "delimiter": null,
  "descriptor_formats": {},
  "enabled": true,
  "environment": null,
  "id_length_limit": null,
  "label_key_case": null,
  "label_order": [],
  "label_value_case": null,
  "labels_as_tags": [
    "unset"
  ],
  "name": null,
  "namespace": null,
  "regex_replace_chars": null,
  "stage": null,
  "tags": {},
  "tenant": null
}

list(object({
    description              = string
    from_port                = number
    to_port                  = number
    protocol                 = string
    source_security_group_id = string
  }))

[
  "default"
]

[
  "/bin/sh",
  "-c"
]

list(object({
    rolearn  = string
    username = string
    groups   = list(string)
  }))

list(object({
    userarn  = string
    username = string
    groups   = list(string)
  }))

[
  "0.0.0.0/0"
]

Outputs

Share the Love

Like this project? Please give it a ★ on our GitHub! (it helps us a lot)

Are you using this project or any of our other projects? Consider leaving a testimonial. =)

Related Projects

Check out these related projects.

• terraform-aws-eks-workers - Terraform module to provision an AWS AutoScaling Group, IAM Role, and Security Group for EKS Workers
• terraform-aws-ec2-autoscale-group - Terraform module to provision Auto Scaling Group and Launch Template on AWS
• terraform-aws-ecs-container-definition - Terraform module to generate well-formed JSON documents (container definitions) that are passed to the aws_ecs_task_definition Terraform resource
• terraform-aws-ecs-alb-service-task - Terraform module which implements an ECS service which exposes a web service via ALB
• terraform-aws-ecs-web-app - Terraform module that implements a web app on ECS and supports autoscaling, CI/CD, monitoring, ALB integration, and much more
• terraform-aws-ecs-codepipeline - Terraform module for CI/CD with AWS Code Pipeline and Code Build for ECS
• terraform-aws-ecs-cloudwatch-autoscaling - Terraform module to autoscale ECS Service based on CloudWatch metrics
• terraform-aws-ecs-cloudwatch-sns-alarms - Terraform module to create CloudWatch Alarms on ECS Service level metrics
• terraform-aws-ec2-instance - Terraform module for providing a general purpose EC2 instance
• terraform-aws-ec2-instance-group - Terraform module for provisioning multiple general purpose EC2 hosts for stateful applications

Help

Got a question? We got answers.

File a GitHub issue, send us an email or join our Slack Community.

DevOps Accelerator for Startups

We are a DevOps Accelerator. We'll help you build your cloud infrastructure from the ground up so you can own it. Then we'll show you how to operate it and stick around for as long as you need us.

Work directly with our team of DevOps experts via email, slack, and video conferencing.

We deliver 10x the value for a fraction of the cost of a full-time engineer. Our track record is not even funny. If you want things done right and you need it done FAST, then we're your best bet.

• Reference Architecture. You'll get everything you need from the ground up built using 100% infrastructure as code.
• Release Engineering. You'll have end-to-end CI/CD with unlimited staging environments.
• Site Reliability Engineering. You'll have total visibility into your apps and microservices.
• Security Baseline. You'll have built-in governance with accountability and audit logs for all changes.
• GitOps. You'll be able to operate your infrastructure via Pull Requests.
• Training. You'll receive hands-on training so your team can operate what we build.
• Questions. You'll have a direct line of communication between our teams via a Shared Slack channel.
• Troubleshooting. You'll get help to triage when things aren't working.
• Code Reviews. You'll receive constructive feedback on Pull Requests.
• Bug Fixes. We'll rapidly work with you to fix any bugs in our projects.

Slack Community

Join our Open Source Community on Slack. It's FREE for everyone! Our "SweetOps" community is where you get to talk with others who share a similar vision for how to rollout and manage infrastructure. This is the best place to talk shop, ask questions, solicit feedback, and work together as a community to build totally sweet infrastructure.

Discourse Forums

Participate in our Discourse Forums. Here you'll find answers to commonly asked questions. Most questions will be related to the enormous number of projects we support on our GitHub. Come here to collaborate on answers, find solutions, and get ideas about the products and services we value. It only takes a minute to get started! Just sign in with SSO using your GitHub account.

Newsletter

Sign up for our newsletter that covers everything on our technology radar. Receive updates on what we're up to on GitHub as well as awesome new projects we discover.

Office Hours

Join us every Wednesday via Zoom for our weekly "Lunch & Learn" sessions. It's FREE for everyone!

Contributing

Bug Reports & Feature Requests

Please use the issue tracker to report any bugs or file feature requests.

Developing

If you are interested in being a contributor and want to get involved in developing this project or help out with our other projects, we would love to hear from you! Shoot us an email.

In general, PRs are welcome. We follow the typical "fork-and-pull" Git workflow.

1. Fork the repo on GitHub
2. Clone the project to your own machine
3. Commit changes to your own branch
4. Push your work back up to your fork
5. Submit a Pull Request so that we can review your changes

NOTE: Be sure to merge the latest changes from "upstream" before making a pull request!

Copyright

Copyright © 2017-2023 Cloud Posse, LLC

License

See LICENSE for full details.

Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements.  See the NOTICE file
distributed with this work for additional information
regarding copyright ownership.  The ASF licenses this file
to you under the Apache License, Version 2.0 (the
"License"); you may not use this file except in compliance
with the License.  You may obtain a copy of the License at

  https://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing,
software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
KIND, either express or implied.  See the License for the
specific language governing permissions and limitations
under the License.

Trademarks

All other trademarks referenced herein are the property of their respective owners.

About

This project is maintained and funded by Cloud Posse, LLC. Like it? Please let us know by leaving a testimonial!

We're a DevOps Professional Services company based in Los Angeles, CA. We ❤️ Open Source Software.

We offer paid support on all of our projects.

Check out our other projects, follow us on twitter, apply for a job, or hire us to help with your cloud strategy and implementation.

Contributors

Erik Osterman	Andriy Knysh	Igor Rodionov	Nuru	Oscar