Integration Tests

Integration tests

The tests on the tests/integration runs using pytest and are meant to be tests that interact with the product in a higher level via API, web UI or using user CLI such as pulp-cli

Important

What to do on integration tests

• call API endpoints using api_client provided by fixtures
• call API endpoints using any other api client if needed
• run pulp-cli commands

What not to do on integration tests

• DO NOT import objects from galaxy_ng project
• DO NOT import objects from any of pulp_ projects
• DO NOT import objects any other testing on /tests/
• DO NOT run django management commands or any other python level script
• DO NOT make direct access to the database or running hub filesystem

In other words, integration tests are to run at the same level as a user or external system consuming the project via its public APIs.

Setting local environment to run integration tests

1. Run your system normally, e.g: ./compose up
2. With system running set the testing admin token with make docker/loadtoken and make docker/load_test_data that will set auth token to a fixed known value for tests and initialize test data.
3. export environment variable export HUB_LOCAL=1
4. Run integration tests using pytest via RUN_INTEGRATION help script.
   • This command can accept any argument accepted by pytest
   • Running all standalone tests: dev/common/RUN_INTEGRATION.sh -sv --log-cli-level=DEBUG
   • Running only specific set od tests: dev/common/RUN_INTEGRATION.sh -sv --log-cli-level=DEBUG "-m standalone_only" -k mytest

Writing integration tests

1. Based on the feature/bugfix you are working on find an appropriate module under tests/integration or create a new one if needed.
2. Write pytest style test cases
3. Use the appropriate markers found on tests/integration/conftest.py and if needed add new markers to the list
4. Reuse fixtures available on tests/integration/conftest.py and if needed add new fixtures
5. Reuse utilities from tests/integration/utils.py and add more utilities if needed

Example

tests/integration/test_namespace_foo_exists.py

import pytest
from ansible.errors import AnsibleError

from ..utils import get_client
from ..utils import generate_unused_namespace
from ..utils import get_all_namespaces

@pytest.mark.namespace
@pytest.mark.cloud_only
def test_namespace_foo_exists(ansible_config):
    """Ensure `foo` is an existing namespace"""
    config = ansible_config("ansible_partner")
    api_client = get_client(config, request_token=True, require_auth=True)

    try:
        resp = api_client('/api/automation-hub/v3/namespaces/foo/')
    except AnsibleError as e:
        if e.http_code == 404:
            pytest.fail(f"Namespace foo doesn't exist")
        else:
            raise e
    assert resp["name"] == "foo"

Important notes

• Tests aimed to run only on specific development mode must have either @pytest.mark.cloud_only or @pytest.mark.standalone_only