Testing AWX

Comprehensive guide to testing AWX including unit tests, functional tests, and collection tests

AWX has a comprehensive test suite covering unit tests, functional tests, and integration tests. This guide covers how to run and write tests for AWX.

Test Framework

AWX uses pytest as its primary test framework. The pytest configuration is defined in pytest.ini at the repository root.

Pytest Configuration

Key settings from pytest.ini:

[pytest]
DJANGO_SETTINGS_MODULE = awx.main.tests.settings_for_test
python_files = *.py
addopts = --reuse-db --nomigrations --tb=native

markers =
    ac: access control test
    survey: tests related to survey feature
    inventory_import: tests of code used by inventory import command
    job_permissions:
    activity_stream_access:
    job_runtime_vars:

Running Tests

Basic Test Commands

From inside the AWX development container:

# Run all tests
(container)/awx_devel$ make test

# Run specific test directory
(container)/awx_devel$ make test TEST_DIRS="awx/main/tests/unit"

# Run tests in parallel
(container)/awx_devel$ make test PARALLEL_TESTS="-n auto"

# Run with coverage
(container)/awx_devel$ make test COVERAGE_ARGS="--cov --cov-report=xml"

Using pytest Directly

# Run all unit tests
pytest awx/main/tests/unit

# Run specific test file
pytest awx/main/tests/unit/test_tasks.py

# Run specific test function
pytest awx/main/tests/unit/test_tasks.py::test_job_launch

# Run tests matching a pattern
pytest -k "test_inventory"

# Run with verbose output
pytest -v awx/main/tests/unit

# Run with even more detail
pytest -vv awx/main/tests/unit

Test Organization

Test Directory Structure

awx/main/tests/
├── unit/              # Unit tests
│   ├── api/          # API endpoint tests
│   ├── models/       # Model tests
│   ├── tasks/        # Task tests
│   ├── utils/        # Utility function tests
│   └── ...
├── functional/        # Functional/integration tests
│   ├── api/
│   ├── models/
│   └── ...
└── data/             # Test fixtures and data
    ├── inventory/
    ├── playbooks/
    └── ...

Test Types

Unit Tests: Test individual functions and classes in isolation

• Location: awx/main/tests/unit/
• Fast, isolated, use mocking

Functional Tests: Test API endpoints and workflows

• Location: awx/main/tests/functional/
• Integration-style tests with database

Collection Tests: Test the AWX Ansible collection

• Location: awx_collection/test/awx/
• Uses ansible-test framework

Writing Unit Tests

Basic Test Structure

import pytest
from awx.main.models import Job

class TestJobModel:
    def test_job_creation(self):
        """Test that a job can be created"""
        job = Job.objects.create(
            name="Test Job",
            job_type="run"
        )
        assert job.name == "Test Job"
        assert job.job_type == "run"

    def test_job_status_transition(self):
        """Test job status transitions"""
        job = Job.objects.create(status="pending")
        job.status = "running"
        job.save()
        assert job.status == "running"

Using Fixtures

Pytest fixtures provide reusable test data:

import pytest
from awx.main.models import Organization, Inventory

@pytest.fixture
def organization():
    """Create a test organization"""
    return Organization.objects.create(name="Test Org")

@pytest.fixture
def inventory(organization):
    """Create a test inventory"""
    return Inventory.objects.create(
        name="Test Inventory",
        organization=organization
    )

def test_inventory_belongs_to_org(inventory, organization):
    """Test inventory-organization relationship"""
    assert inventory.organization == organization
    assert inventory in organization.inventories.all()

Using Markers

Markers categorize and filter tests:

import pytest

@pytest.mark.ac
def test_access_control():
    """Test access control logic"""
    pass

@pytest.mark.survey
def test_survey_validation():
    """Test survey validation"""
    pass

# Run only access control tests
# pytest -m ac

Mocking

Use unittest.mock for isolating dependencies:

from unittest.mock import patch, Mock
import pytest

@patch('awx.main.tasks.jobs.ansible_runner')
def test_job_execution(mock_runner):
    """Test job execution with mocked runner"""
    mock_runner.run.return_value = Mock(
        status="successful",
        rc=0
    )
    
    # Test job execution logic
    job = Job.objects.create()
    result = job.run()
    
    assert result.status == "successful"
    mock_runner.run.assert_called_once()

Functional Tests

API Testing

Functional tests often test API endpoints:

import pytest
from rest_framework.test import APIClient

@pytest.mark.django_db
class TestJobTemplateAPI:
    def test_create_job_template(self, admin_user, organization):
        """Test creating a job template via API"""
        client = APIClient()
        client.force_authenticate(user=admin_user)
        
        response = client.post('/api/v2/job_templates/', {
            'name': 'Test Template',
            'organization': organization.id,
            'project': project.id,
            'playbook': 'test.yml'
        })
        
        assert response.status_code == 201
        assert response.data['name'] == 'Test Template'
    
    def test_launch_job_template(self, admin_user, job_template):
        """Test launching a job template"""
        client = APIClient()
        client.force_authenticate(user=admin_user)
        
        response = client.post(
            f'/api/v2/job_templates/{job_template.id}/launch/',
            {}
        )
        
        assert response.status_code == 201
        assert 'job' in response.data

Collection Tests

AWX Collection Testing

The AWX Ansible collection has its own test suite:

# Install the collection
make install_collection

# Run collection unit tests
make test_collection

# Run specific collection tests
cd awx_collection
ansible-test units --docker

# Run integration tests
ansible-test integration --docker

Collection Test Structure

awx_collection/
├── test/
│   └── awx/
│       ├── unit/           # Unit tests
│       └── integration/    # Integration tests
└── plugins/
    ├── modules/            # Ansible modules
    └── inventory/          # Inventory plugins

Writing Collection Tests

# awx_collection/test/awx/unit/module_utils/test_controller_api.py
import pytest
from ansible_collections.awx.awx.plugins.module_utils.controller_api import ControllerAPIModule

class TestControllerAPI:
    def test_api_authentication(self):
        """Test API authentication"""
        module = ControllerAPIModule(
            argument_spec={},
            controller_host='https://awx.example.com',
            controller_username='admin',
            controller_password='password'
        )
        assert module.authenticated

Code Coverage

Generating Coverage Reports

# Run tests with coverage
make test COVERAGE_ARGS="--cov --cov-report=html --cov-report=xml"

# View HTML coverage report
# Open htmlcov/index.html in a browser

# Coverage is also reported in codecov.io for PRs

Coverage Configuration

.coveragerc file configures coverage settings:

[run]
source = awx
omit = 
    */tests/*
    */migrations/*

[report]
exclude_lines =
    pragma: no cover
    def __repr__
    raise AssertionError
    raise NotImplementedError

Testing Best Practices

1. Test Isolation

Each test should be independent:

# Good: Test creates its own data
def test_user_creation():
    user = User.objects.create(username="testuser")
    assert user.username == "testuser"

# Bad: Test depends on database state
def test_user_exists():
    user = User.objects.get(username="existing_user")
    assert user is not None

2. Use Fixtures for Setup

@pytest.fixture
def job_template(organization, project, inventory):
    """Provide a configured job template"""
    return JobTemplate.objects.create(
        name="Test Template",
        organization=organization,
        project=project,
        inventory=inventory,
        playbook="test.yml"
    )

3. Test Edge Cases

def test_job_with_empty_inventory():
    """Test job behavior with empty inventory"""
    job = Job.objects.create(inventory=empty_inventory)
    with pytest.raises(ValidationError):
        job.validate_hosts()

def test_job_with_invalid_playbook():
    """Test job with non-existent playbook"""
    job = Job.objects.create(playbook="nonexistent.yml")
    with pytest.raises(PlaybookNotFound):
        job.run()

4. Clear Test Names

# Good: Descriptive test names
def test_job_template_requires_inventory():
    pass

def test_job_template_launch_creates_job():
    pass

# Bad: Unclear test names
def test_jt_1():
    pass

def test_launch():
    pass

5. Mock External Dependencies

@patch('awx.main.tasks.jobs.requests.get')
def test_project_update_with_git(mock_get):
    """Test project update without making real Git calls"""
    mock_get.return_value = Mock(status_code=200)
    
    project = Project.objects.create(scm_type='git')
    project.update()
    
    mock_get.assert_called_once()

Running Tests in CI

AWX uses GitHub Actions for continuous integration.

Local CI Testing

Run the same checks that CI runs:

# Code formatting
make black

# Run linters
black --check awx/

# Run tests with coverage
make test COVERAGE_ARGS="--cov --cov-report=xml --junitxml=reports/junit.xml"

# Run collection sanity tests
cd awx_collection
ansible-test sanity --docker

Debugging Tests

Using pdb

def test_complex_logic():
    job = Job.objects.create()
    
    # Set breakpoint
    import pdb; pdb.set_trace()
    
    result = job.run()
    assert result.status == "successful"

Verbose Output

# Show print statements
pytest -s awx/main/tests/unit/test_tasks.py

# Show test names as they run
pytest -v awx/main/tests/unit/

# Show full diff on assertion failures
pytest -vv awx/main/tests/unit/

Capture Logs

# Show log output
pytest --log-cli-level=DEBUG awx/main/tests/unit/

# Capture logs to file
pytest --log-file=test.log awx/main/tests/unit/

Database Testing

Pytest automatically manages test databases:

@pytest.mark.django_db
def test_with_database():
    """Test that requires database access"""
    user = User.objects.create(username="test")
    assert User.objects.filter(username="test").exists()

Database Reuse

The --reuse-db flag (in pytest.ini) speeds up test runs by reusing the test database between runs.

Performance Testing

Test Execution Time

# Show slowest tests
pytest --durations=10 awx/main/tests/unit/

# Show all test durations
pytest --durations=0 awx/main/tests/unit/

Profiling Tests

# Profile test execution
pytest --profile awx/main/tests/unit/

# Profile and save results
pytest --profile-svg awx/main/tests/unit/

Next Steps

• Architecture Overview
• Task Manager System
• Contributing Guidelines