Deploy Teleport Agents with Terraform

An agent is a Teleport instance configured to run one or more Teleport services in order to proxy infrastructure resources. For a brief architectural overview of how agents run in a Teleport cluster, read the Introduction to Teleport Agents.

This guide shows you how to deploy a pool of Teleport agents running on virtual machines by declaring it as code using Terraform.

There are several methods you can use to join a Teleport agent to your cluster, which we discuss in the Joining Services to your Cluster guide. In this guide, we will use the join token method, where the operator stores a secure token on the Auth Service, and an agent presents the token in order to join a cluster.

No matter which join method you use, it will involve the following Terraform resources:

• Compute instances to run Teleport services
• A join token for each compute instance in the agent pool

Prerequisites

• A running Teleport cluster version 15.4.22 or above. If you want to get started with Teleport, sign up for a free trial or set up a demo environment.

• The tctl admin tool and tsh client tool.

  On Teleport Enterprise, you must use the Enterprise version of tctl, which you can download from your Teleport account workspace. Otherwise, visit Installation for instructions on downloading tctl and tsh for Teleport Community Edition.

• An AWS, Google Cloud, or Azure account with permissions to create virtual machine instances.

• Cloud infrastructure that enables virtual machine instances to connect to the Teleport Proxy Service. For example:

  • An AWS subnet with a public NAT gateway or NAT instance.
  • Google Cloud NAT
  • Azure NAT Gateway

  In minimum-security demo clusters, you can also configure the VM instances you deploy to have public IP addresses.

• Terraform v1.0.0 or higher.

• An identity file for the Teleport Terraform provider. Make sure you are familiar with how to set up the Teleport Terraform provider before following this guide.

• To check that you can connect to your Teleport cluster, sign in with tsh login, then verify that you can run tctl commands using your current credentials. tctl is supported on macOS and Linux machines.

  For example:

  tsh login --proxy=teleport.example.com --user=email@example.com
  tctl status
  Cluster  teleport.example.com
  Version  15.4.22
  CA pin   sha256:abdc1245efgh5678abdc1245efgh5678abdc1245efgh5678abdc1245efgh5678

  If you can connect to the cluster and run the tctl status command, you can use your current credentials to run subsequent tctl commands from your workstation. If you host your own Teleport cluster, you can also run tctl commands on the computer that hosts the Teleport Auth Service for full permissions.

Step 1/3. Import the Terraform module

1. Navigate to the directory where you plan to organize files for your root Terraform module.

2. Move the identity file for the Teleport Terraform provider into your project directory so the Terraform provider can access it. Name the file terraform-identity.

   If you don't have an identity file available, make sure you have followed the prerequisites for this guide.

3. Fetch the Teleport code repository and copy the example Terraform configuration for this project into your current working directory. Copy the appropriate child module for your cloud provider into a subdirectory called cloud and HCL configurations for Teleport resources into a subdirectory called teleport:

   • AWS
   • Google Cloud
   • Azure

   git clone --depth=1 https://github.com/gravitational/teleport teleport-clone
   cp -R teleport-clone/examples/agent-pool-terraform/teleport teleport
   cp -R teleport-clone/examples/agent-pool-terraform/aws cloud
   rm -rf teleport-clone

   git clone --depth=1 https://github.com/gravitational/teleport teleport-clone
   cp -R teleport-clone/examples/agent-pool-terraform/teleport teleport
   cp -R teleport-clone/examples/agent-pool-terraform/gcp cloud
   rm -rf teleport-clone

   git clone --depth=1 https://github.com/gravitational/teleport teleport-clone
   cp -R teleport-clone/examples/agent-pool-terraform/teleport teleport
   cp -R teleport-clone/examples/agent-pool-terraform/azure cloud
   rm -rf teleport-clone

4. Create a file called main.tf with the following content, which imports the agent-pool-terraform module for your cloud provider:

module "teleport" {
  source                = "./teleport"
  agent_count           = 2
  agent_roles           = ["Node"]
  proxy_service_address = "teleport.example.com:443"
  teleport_edition      = "oss"
  teleport_version      = "15.4.22"
}

module "agents" {
  region           = ""
  source           = "./cloud"
  subnet_id        = ""
  userdata_scripts = module.teleport.userdata_scripts
}

module "teleport" {
  source                = "./teleport"
  agent_count           = 2
  agent_roles           = ["Node"]
  proxy_service_address = "teleport.example.com:443"
  teleport_edition      = "oss"
  teleport_version      = "15.4.22"
}

module "agents" {
  gcp_zone         = "us-east1-b"
  google_project   = ""
  source           = "./cloud"
  subnet_id        = ""
  userdata_scripts = module.teleport.userdata_scripts
}

module "teleport" {
  source                = "./teleport"
  agent_count           = 2
  agent_roles           = ["Node"]
  proxy_service_address = "teleport.example.com:443"
  teleport_edition      = "oss"
  teleport_version      = "15.4.22"
}

module "agents" {
  azure_resource_group = ""
  public_key_path      = ""
  region               = "East US"
  source               = "./cloud"
  subnet_id            = ""
  userdata_scripts     = module.teleport.userdata_scripts
}

Edit the module "teleport" block in main.tf as follows:

1. Assign agent_count to 2 for high availability. As you scale your Teleport usage, you can increase this count to ease the load on each agent.

2. Modify the elements of the agent_roles list.

   The Teleport Auth Service associates a join token with one or more roles, identifying the Teleport services that are allowed to use the token. Modify agent_roles to include the token roles that correspond to the Teleport services you plan to run on your agent nodes:

   Role value	Service it enables
   App	Teleport Application Service
   Discovery	Teleport Discovery Service
   Db	Teleport Database Service
   Kube	Teleport Kubernetes Service
   Node	Teleport SSH Service

   The elements of the agent_roles list also determine which Teleport services the agents deployed by the agent-pool-terraform module will run. For example, if the value of agent_roles is ["Node", "Db"], the configuration file enables the Teleport SSH Service and Teleport Database Service.

   If the value of the agent_roles input variable includes the Node role, the configuration also adds the role: agent-pool label to the Teleport SSH Service on each instance. This makes it easier to access hosts in the agent pool later. The script makes Teleport the only option for accessing agent instances by disabling OpenSSH on startup and deleting any authorized public keys.

3. Assign proxy_service_address to the host and HTTPS port of your Teleport Proxy Service, e.g., mytenant.teleport.sh:443.

   Make sure to include the port.

4. Make sure teleport_edition matches your Teleport edition. Assign this to oss, cloud, or enterprise. The default is oss.

5. If needed, change the value of teleport_version to the version of Teleport you want to run on your agents. It must be either the same major version as your Teleport cluster or one major version behind.

Edit the module "agents" block in main.tf as follows:

1. If you are deploying your instance in a minimum-security demo environment and do not have a NAT gateway, NAT instance, or other method for connecting your instances to the Teleport Proxy Service, modify the module block to associate a public IP address with each agent instance:

   insecure_direct_access=true
   

2. Assign the remaining input variables depending on your cloud provider:

Step 2/3. Add provider configurations

In this step, you will configure the agent-pool-terraform module for your Teleport cluster and cloud provider.

In your project directory, create a file called providers.tf with the following content:

terraform {
  required_providers {
    aws = {
      source  = "hashicorp/aws"
      version = "~> 4.0"
    }

    teleport = {
      source  = "terraform.releases.teleport.dev/gravitational/teleport"
      version = "15.4.22"
    }
  }
}

provider "aws" {
  region = AWS_REGION
}

provider "teleport" {
  # Update addr to point to your Teleport Cloud tenant URL's host:port
  addr               = PROXY_SERVICE_ADDRESS
  identity_file_path = "terraform-identity"
}

terraform {
  required_providers {
    google = {
      source  = "hashicorp/google"
      version = "~> 5.5.0"
    }

    teleport = {
      source  = "terraform.releases.teleport.dev/gravitational/teleport"
      version = "15.4.22"
    }
  }
}

provider "google" {
  project = GOOGLE_CLOUD_PROJECT
  region  = GOOGLE_CLOUD_REGION
}

provider "teleport" {
  # Update addr to point to your Teleport Cloud tenant URL's host:port
  addr               = PROXY_SERVICE_ADDRESS
  identity_file_path = "terraform-identity"
}

terraform {
  required_providers {
    azurerm = {
      source  = "hashicorp/azurerm"
      version = "~> 3.0.0"
    }

    teleport = {
      source  = "terraform.releases.teleport.dev/gravitational/teleport"
      version = "15.4.22"
    }
  }
}

provider "teleport" {
  identity_file_path = "terraform-identity"
  # Update addr to point to your Teleport Cloud tenant URL's host:port
  addr               = PROXY_SERVICE_ADDRESS
}

provider "azurerm" {
  features {}
}

Step 3/3. Verify the deployment

Make sure your cloud provider credentials are available to Terraform using the standard approach for your organization.

Apply the Terraform configuration:

terraform init
terraform apply

Once the apply command completes, run the following command to verify that your agents have deployed successfully. This command, which assumes that the agents have the Node role, lists all Teleport SSH Service instances with the role=agent-pool label:

tsh ls role=agent-pool
Node Name                  Address    Labels-------------------------- ---------- ---------------ip-10-1-1-187.ec2.internal ⟵ Tunnel   role=agent-poolip-10-1-1-24.ec2.internal  ⟵ Tunnel   role=agent-pool

Next step: Enroll infrastructure resources

In this section, we describe the two ways to configure your agent pool to protect infrastructure resources with Teleport.

Define dynamic resources in Terraform

You can declare Terraform resources that enroll your infrastructure with Teleport. The Teleport Terraform provider currently supports the following:

To declare a dynamic resource with Terraform, add a configuration block similar to the ones below to a *.tf file in your agent-pool-terraform project directory.

The Teleport Terraform provider creates these on the Auth Service backend, and the relevant Teleport services query them in order to proxy user traffic. For a full list of supported resources and fields, see the Terraform provider reference.

resource "teleport_app" "example" {
  metadata = {
    name        = "example"
    description = "Test app"
    labels = {
      // Teleport adds this label by default, so add it here to
      // ensure a consistent state.
      "teleport.dev/origin" = "dynamic"
    }
  }

  spec = {
    uri = "localhost:3000"
  }
}

resource "teleport_database" "example" {
  metadata = {
    name        = "example"
    description = "Test database"
    labels = {
      // Teleport adds this label by default, so add it here to
      // ensure a consistent state.
      "teleport.dev/origin" = "dynamic"
    }
  }

  spec = {
    protocol = "postgres"
    uri      = "localhost"
  }
}

Configure Teleport services in the agent pool

Each Teleport service reads its local configuration file (/etc/teleport.yaml by default) to determine which infrastructure resources to proxy. You can edit this configuration file to enroll resources with Teleport.

In the setup we explored in this guide, you can edit the user data script for each instance to add configuration settings to, for example, the database_service or kubernetes_service sections.

To see how to configure each service, read its section of the documentation:

• SSH Service
• Database Service
• Kubernetes Service
• Windows Desktop Service
• Application Service

Further reading: How the module works

In this section, we explain the resources configured in the agent-pool-terraform module.

Join token

The agent-pool-terraform module deploys one virtual machine instance for each Teleport agent. Each agent joins the cluster using a token. We create each token using the teleport_provision_token Terraform resource, specifying the token's value with a random_string resource:

resource "random_string" "token" {
  count            = var.agent_count
  length           = 32
  override_special = "-.+"
}

resource "teleport_provision_token" "agent" {
  count   = var.agent_count
  version = "v2"
  spec = {
    roles = var.agent_roles
  }
  metadata = {
    name    = random_string.token[count.index].result
    expires = timeadd(timestamp(), "1h")
  }
}

When we apply the teleport_provision_token resources, the Teleport Terraform provider creates them on the Teleport Auth Service backend.

User data script

Each Teleport agent deployed by the agent-pool-terraform module loads a user data script that creates a Teleport configuration file for the agent. The services enabled by the configuration file depend on the value of the agent_roles input variable:

#!/bin/bash

curl https://cdn.teleport.dev/install-v${teleport_version}.sh | bash -s ${teleport_version} ${teleport_edition}

echo ${token} > /var/lib/teleport/token
cat<<EOF >/etc/teleport.yaml
version: v3
teleport:
  auth_token: /var/lib/teleport/token
  proxy_server: ${proxy_service_address}
%{ if contains(agent_roles, "App") }
app_service:
  enabled: false
  resources:
  - labels:
      "*": "*"
%{ endif }
auth_service:
  enabled: false
%{ if contains(agent_roles, "Db") }
db_service:
  enabled: false
  resources:
  - labels:
      "*": "*"
%{ endif }
%{ if contains(agent_roles, "Discovery") }
discovery_service:
  enabled: false
%{ endif }
%{ if contains(agent_roles, "Kube") }
kubernetes_service:
  enabled: false
  resources:
  - labels:
      "*": "*"
%{ endif }
proxy_service:
  enabled: false
%{ if contains(agent_roles, "Node") }
ssh_service:
  enabled: true
  labels:
    role: agent-pool
%{ else }
ssh_service:
  enabled: false
%{ endif }
EOF

systemctl restart teleport;

# Disable OpenSSH and any longstanding authorized keys.
systemctl disable --now ssh.service
find / -wholename "*/.ssh/authorized_keys" -delete

Virtual machine instances

Each cloud-specific child module of agent-pool-terraform declares resources to deploy a virtual machine instance on your cloud provider:

data "aws_ami" "amazon_linux_2023" {
  most_recent = true

  filter {
    name   = "description"
    values = ["Amazon Linux 2023 AMI*"]
  }

  filter {
    name   = "architecture"
    values = ["x86_64"]
  }

  filter {
    name   = "virtualization-type"
    values = ["hvm"]
  }

  filter {
    name   = "owner-alias"
    values = ["amazon"]
  }
}

resource "aws_instance" "teleport_agent" {
  count                       = length(var.userdata_scripts)
  ami                         = data.aws_ami.amazon_linux_2023.id
  instance_type               = "t3.small"
  subnet_id                   = var.subnet_id
  user_data                   = var.userdata_scripts[count.index]
  associate_public_ip_address = var.insecure_direct_access

  // Adheres to security best practices
  monitoring = true

  metadata_options {
    http_endpoint = "enabled"
    http_tokens   = "required"
  }

  root_block_device {
    encrypted = true
  }
}

locals {
  // Google Cloud provides public IP addresses to instances when the
  // network_interface block includes an empty access_config, so use a dynamic
  // block to enable a public IP based on the insecure_direct_access input.
  access_configs = var.insecure_direct_access ? [{}] : []
}

resource "google_compute_instance" "teleport_agent" {
  count = length(var.userdata_scripts)
  name  = "teleport-agent-${count.index}"
  zone  = var.gcp_zone

  boot_disk {
    initialize_params {
      image = "family/ubuntu-2204-lts"
    }
  }

  network_interface {
    subnetwork = var.subnet_id

    // If the user enables insecure direct access, allocate a public IP to the
    // instance.
    dynamic "access_config" {
      for_each = local.access_configs
      content {}
    }
  }

  machine_type = "e2-standard-2"

  metadata_startup_script = var.userdata_scripts[count.index]
}

locals {
  username = "admin_temp"
}

resource "azurerm_network_interface" "teleport_agent" {
  count               = length(var.userdata_scripts)
  name                = "teleport-agent-ni-${count.index}"
  location            = var.region
  resource_group_name = var.azure_resource_group
  ip_configuration {
    name                          = "teleport_agent_ip_config"
    subnet_id                     = var.subnet_id
    private_ip_address_allocation = "Dynamic"
    public_ip_address_id          = var.insecure_direct_access ? azurerm_public_ip.agent[count.index].id : ""
  }
}

resource "azurerm_public_ip" "agent" {
  count               = var.insecure_direct_access ? length(var.userdata_scripts) : 0
  name                = "agentIP-${count.index}"
  resource_group_name = var.azure_resource_group
  location            = var.region
  allocation_method   = "Static"
}

resource "azurerm_virtual_machine" "teleport_agent" {
  count = length(var.userdata_scripts)
  name  = "teleport-agent-${count.index}"

  location            = var.region
  resource_group_name = var.azure_resource_group

  network_interface_ids = [azurerm_network_interface.teleport_agent[count.index].id]
  os_profile_linux_config {
    disable_password_authentication = true
    ssh_keys {
      key_data = file(var.public_key_path)
      // The only allowed path. See:
      // https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/resources/virtual_machine
      path = "/home/${local.username}/.ssh/authorized_keys"
    }
  }

  os_profile {
    computer_name  = "teleport-agent-${count.index}"
    admin_username = local.username
    custom_data    = var.userdata_scripts[count.index]
  }

  vm_size = "Standard_B2s"

  storage_os_disk {
    name              = "teleport-agent-disk-${count.index}"
    create_option     = "FromImage"
    managed_disk_type = "Standard_LRS"
  }

  storage_image_reference {
    publisher = "Canonical"
    offer     = "0001-com-ubuntu-server-jammy"
    sku       = "22_04-lts"
    version   = "latest"
  }
}