2 minute read

Infrastructure code is considered 100% code but often good software principle, like testing, are not applied. Apart from checking the correctness of code, testing documents how code should behave, as well. In this post I will show how to test a Terraform module using the Python module tftest along with the testing framework pytest.

Terraform module

A Terraform module is just a folder containing some Terraform configuration files. The source code of the module is stored on Github. This module creates two AWS instances in two different zones a and b of the region selected by the user.

# omitted...

locals {
  zones = {
    main      = "${var.region}a"
    secondary = "${var.region}b"

resource "aws_instance" "ubuntu" {
  for_each = local.zones

  ami                         = data.aws_ami.ubuntu.id
  associate_public_ip_address = true
  availability_zone           = each.value
  instance_type               = var.instance_type

  tags = {
    Name = "instance-${each.key}"

The Terraform files that use the module are located under tests/fixtures/plan. The main.tf creates the two instances:

terraform {
  required_version = ">= 1.0"

module "test" {
  source        = "../../.."
  instance_type = var.instance_type
  region        = var.region

Test fixtures

The scope of tests is to check the system output based on input stimuli.

Each test is made of three basic steps:

  • arrange: prepare input data for the function
  • act: run the function under test
  • assert: verify that the test output is what we expected.

Data to be passed to tests are arranged using fixtures.

Fixtures are functions, which will run before each test function to which it is applied. Fixtures are used to feed some data to the tests such as database connections, URLs to test and some sort of input data. Therefore, instead of running the same code for every test, we can attach fixture functions to the tests and it will run and return the data to the test before executing each test.

In pytest, we can define the fixture functions in tests/conftest.py to make them accessible across multiple test files. The following fixture returns the path of the directory containing the Terraform configuration files:

def fixtures_dir():
  return os.path.join(os.path.dirname(os.path.abspath(__file__)), 'fixtures')

To generate the Terraform plan for tests, we use another fixture at tests/test_plan.py that calls the TerraformTest function and sets the tfvar file plan.auto.tfvars:

def plan(fixtures_dir):
  tf = tftest.TerraformTest('plan', fixtures_dir)
  return tf.plan(output=True)


Test the Terraform plan

The python module tftest is heavily inspired by two projects: Terratest for the lightweight approach to testing Terraform, and python-terraform for wrapping the Terraform command in Python.

For accessing variables from Terraform plan:


In my module, to test the variable values during plan:

def test_variables(plan):
  assert 'instance_type' in plan.variables
  assert plan.variables['instance_type'] == 't2.micro'

  assert 'region' in plan.variables
  assert plan.variables['region'] == 'us-east-1'

For accessing outputs from Terraform plan:


Since the outputs are not known during plan, we can just check that outputs are defined in plan:

def test_outputs(plan):
  assert 'main_arn' in plan.outputs
  assert 'main_ip_address' in plan.outputs
  assert 'secondary_arn' in plan.outputs
  assert 'secondary_ip_address' in plan.outputs

For accessing resources, e.g. "google_project_iam_member.test_root_resource":


For accessing modules like "module.test":


And for accessing module resources:


Below we check if the module is creating two "aws_instance" resources:

def test_modules(plan):
  mod = plan.modules['module.test']
  assert set(mod.resources.keys()) == set(


In this post I explained how to use the tftest python module to test the plan output of a terraform module.

Leave a comment