Terraform Connector

The Terraform connector generates pyinfra inventory from Terraform output variables. This enables seamless integration between Terraform infrastructure provisioning and pyinfra configuration management.

Overview

The Terraform connector reads terraform output -json to generate hosts dynamically based on Terraform state.

# Use terraform output variable
pyinfra @terraform/server_group.value.server_ips deploy.py

# Use default pyinfra_inventory output
pyinfra @terraform deploy.py

The Terraform connector is in beta. Report issues on GitHub if you encounter problems.

Requirements

Terraform CLI installed and accessible
Terraform state with output variables
Current directory contains Terraform configuration

Terraform Output Format

The connector uses flattened JSON output from terraform output -json:

{
  "server_group": {
    "value": {
      "server_ips": [
        "1.2.3.4",
        "1.2.3.5",
        "1.2.3.6"
      ]
    }
  }
}

Access with:

pyinfra @terraform/server_group.value.server_ips deploy.py

Basic Usage

List of IPs

Simplest form - list of IP addresses or hostnames:

# terraform/main.tf
output "pyinfra_inventory" {
  value = [
    "192.168.1.10",
    "192.168.1.11",
    "192.168.1.12",
  ]
}

# Use default output name
pyinfra @terraform deploy.py

Named Output

Use specific output variable:

# terraform/main.tf
output "web_servers" {
  value = [
    aws_instance.web[0].public_ip,
    aws_instance.web[1].public_ip,
  ]
}

# Specify output path
pyinfra @terraform/web_servers.value deploy.py

Advanced Usage

Hosts with Data

Pass host-specific data:

# terraform/main.tf
output "pyinfra_inventory" {
  value = [
    {
      "ssh_hostname" = aws_instance.web[0].public_ip
      "ssh_user"     = "ubuntu"
      "app_role"     = "web"
    },
    {
      "ssh_hostname" = aws_instance.web[1].public_ip
      "ssh_user"     = "ubuntu"
      "app_role"     = "web"
    },
    {
      "ssh_hostname" = aws_instance.db.public_ip
      "ssh_user"     = "postgres"
      "app_role"     = "database"
    },
  ]
}

# deploy.py
from pyinfra import host
from pyinfra.operations import apt

if host.data.app_role == "web":
    apt.packages(
        name="Install nginx",
        packages=["nginx"],
        _sudo=True,
    )
elif host.data.app_role == "database":
    apt.packages(
        name="Install postgresql",
        packages=["postgresql"],
        _sudo=True,
    )

AWS Example

Complete AWS infrastructure with Terraform and pyinfra:

# terraform/main.tf
terraform {
  required_providers {
    aws = {
      source  = "hashicorp/aws"
      version = "~> 5.0"
    }
  }
}

provider "aws" {
  region = "us-east-1"
}

# Create EC2 instances
resource "aws_instance" "web" {
  count         = 3
  ami           = "ami-0c55b159cbfafe1f0"
  instance_type = "t2.micro"
  key_name      = "my-key"
  
  tags = {
    Name = "web-${count.index}"
    Role = "webserver"
  }
}

resource "aws_instance" "db" {
  ami           = "ami-0c55b159cbfafe1f0"
  instance_type = "t2.small"
  key_name      = "my-key"
  
  tags = {
    Name = "database"
    Role = "database"
  }
}

# Output for pyinfra
output "pyinfra_inventory" {
  value = concat(
    [
      for instance in aws_instance.web : {
        ssh_hostname = instance.public_ip
        ssh_user     = "ubuntu"
        ssh_key      = "~/.ssh/my-key.pem"
        role         = "webserver"
        instance_id  = instance.id
      }
    ],
    [{
      ssh_hostname = aws_instance.db.public_ip
      ssh_user     = "ubuntu"
      ssh_key      = "~/.ssh/my-key.pem"
      role         = "database"
      instance_id  = aws_instance.db.id
    }]
  )
}

# Apply terraform
cd terraform
terraform apply

# Deploy with pyinfra
cd ..
pyinfra @terraform deploy.py

Nested Outputs

Access nested output structures:

# terraform/main.tf
output "infrastructure" {
  value = {
    production = {
      web_servers = [
        aws_instance.prod_web[0].public_ip,
        aws_instance.prod_web[1].public_ip,
      ]
    }
    staging = {
      web_servers = [
        aws_instance.staging_web.public_ip,
      ]
    }
  }
}

# Deploy to production
pyinfra @terraform/infrastructure.value.production.web_servers deploy.py

# Deploy to staging
pyinfra @terraform/infrastructure.value.staging.web_servers deploy.py

Default Output Name

If no output name is specified, the connector looks for pyinfra_inventory.value:

# terraform/main.tf
output "pyinfra_inventory" {
  value = [
    "192.168.1.10",
    "192.168.1.11",
  ]
}

# These are equivalent
pyinfra @terraform deploy.py
pyinfra @terraform/pyinfra_inventory.value deploy.py

SSH Configuration

Set SSH connection details in Terraform outputs:

# terraform/outputs.tf
output "pyinfra_inventory" {
  value = [
    {
      ssh_hostname = aws_instance.server.public_ip
      ssh_user     = "ubuntu"
      ssh_key      = "~/.ssh/aws-key.pem"
      ssh_port     = 22
    },
  ]
}

Available SSH data keys:

ssh_hostname - Hostname or IP
ssh_user - SSH username
ssh_key - Path to private key
ssh_password - SSH password (not recommended)
ssh_port - SSH port (default: 22)
ssh_forward_agent - Enable agent forwarding

Workflow

Typical Terraform + pyinfra workflow:

# 1. Write Terraform configuration
cat > main.tf <<EOF
output "pyinfra_inventory" {
  value = ["192.168.1.10"]
}
EOF

# 2. Apply Terraform
terraform init
terraform apply

# 3. Verify output
terraform output -json

# 4. Deploy with pyinfra
pyinfra @terraform deploy.py

# 5. Make changes and re-apply
vi main.tf
terraform apply
pyinfra @terraform deploy.py

Multiple Environments

Manage multiple environments with Terraform workspaces:

# Create workspaces
terraform workspace new staging
terraform workspace new production

# Deploy to staging
terraform workspace select staging
terraform apply
pyinfra @terraform deploy_staging.py

# Deploy to production
terraform workspace select production
terraform apply
pyinfra @terraform deploy_production.py

Error Handling

The connector validates Terraform output:

# If output not found
pyinfra @terraform/missing_output deploy.py
# Error: Terraform output missing_output not found
# Available outputs:
#    - pyinfra_inventory.value
#    - web_servers.value

Testing

Test Terraform outputs before deploying:

# Dry run
pyinfra @terraform deploy.py --dry

# Check inventory
pyinfra @terraform --facts server.Hostname

Limitations

Requires Terraform CLI in PATH
Must run from directory with Terraform state
Output must be valid JSON
Re-reads output on each run (can be slow for large state)

Best Practices

Use descriptive output names

output "web_servers" { ... }
output "database_servers" { ... }

Include necessary SSH data

output "pyinfra_inventory" {
  value = [{
    ssh_hostname = ...
    ssh_user = ...
    ssh_key = ...
  }]
}

Add host data for conditional deploys

output "pyinfra_inventory" {
  value = [{
    ssh_hostname = ...
    role = "web"
    environment = "production"
  }]
}

Use Terraform outputs for dynamic data

output "pyinfra_inventory" {
  value = [{
    ssh_hostname = ...
    db_host = aws_db_instance.main.endpoint
    redis_host = aws_elasticache_cluster.main.cache_nodes[0].address
  }]
}

Complete Example

Here’s a complete example with Terraform and pyinfra:

# terraform/main.tf
output "pyinfra_inventory" {
  value = [
    {
      ssh_hostname = "web1.example.com"
      ssh_user = "ubuntu"
      ssh_key = "~/.ssh/id_rsa"
      role = "webserver"
      datacenter = "us-east-1"
    },
    {
      ssh_hostname = "web2.example.com"
      ssh_user = "ubuntu"
      ssh_key = "~/.ssh/id_rsa"
      role = "webserver"
      datacenter = "us-west-2"
    },
  ]
}

# deploy.py
from pyinfra import host
from pyinfra.operations import apt, files

# Install packages based on role
if host.data.role == "webserver":
    apt.packages(
        name="Install web server",
        packages=["nginx"],
        _sudo=True,
    )
    
    # Configure based on datacenter
    files.template(
        name="Configure nginx",
        src="templates/nginx.conf.j2",
        dest="/etc/nginx/nginx.conf",
        datacenter=host.data.datacenter,
        _sudo=True,
    )

# Deploy
cd terraform && terraform apply && cd ..
pyinfra @terraform deploy.py

Source Reference

Location: src/pyinfra/connectors/terraform.py:30

Key Methods

make_names_data() - Generate hosts from Terraform output (line 83)

Overview

Core API

Commands

Connectors

Facts

Terraform Connector

Overview

Requirements

Terraform Output Format

Basic Usage

List of IPs

Named Output

Advanced Usage

Hosts with Data

AWS Example

Nested Outputs

Default Output Name

SSH Configuration

Workflow

Multiple Environments

Error Handling

Testing

Limitations

Best Practices

Complete Example

Source Reference

Key Methods

Build docs developers (and LLMs) love

Overview

Core API

Commands

Connectors

Facts

Documentation Index

​Overview

​Requirements

​Terraform Output Format

​Basic Usage

​List of IPs

​Named Output

​Advanced Usage

​Hosts with Data

​AWS Example

​Nested Outputs

​Default Output Name

​SSH Configuration

​Workflow

​Multiple Environments

​Error Handling

​Testing

​Limitations

​Best Practices

​Complete Example

​Source Reference

​Key Methods

Build docs developers (and LLMs) love

Overview

Requirements

Terraform Output Format

Basic Usage

List of IPs

Named Output

Advanced Usage

Hosts with Data

AWS Example

Nested Outputs

Default Output Name

SSH Configuration

Workflow

Multiple Environments

Error Handling

Testing

Limitations

Best Practices

Complete Example

Source Reference

Key Methods