Nova contains a suite of unit tests, in the nova/tests directory.
Any proposed code change will be automatically rejected by the OpenStack Jenkins server  if the change causes unit test failures.
Run the unit tests by doing:
The run_tests.sh script supports several flags. You can view a list of flags by doing:
This will show the following help information:
Usage: ./run_tests.sh [OPTION]... Run Nova's test suite(s) -V, --virtual-env Always use virtualenv. Install automatically if not present -N, --no-virtual-env Don't use virtualenv. Run tests in local environment -s, --no-site-packages Isolate the virtualenv from the global Python environment -f, --force Force a clean re-build of the virtual environment. Useful when dependencies have been added. -u, --update Update the virtual environment with any newer package versions -p, --pep8 Just run PEP8 and HACKING compliance check -P, --no-pep8 Don't run static code checks -c, --coverage Generate coverage report -d, --debug Run tests with testtools instead of testr. This allows you to use the debugger. -h, --help Print this usage message --hide-elapsed Don't print the elapsed time for each test along with slow test list --virtual-env-path <path> Location of the virtualenv directory Default: $(pwd) --virtual-env-name <name> Name of the virtualenv directory Default: .venv --tools-path <dir> Location of the tools directory Default: $(pwd) Note: with no options specified, the script will try to run the tests in a virtual environment, If no virtualenv is found, the script will ask if you would like to create one. If you prefer to run tests NOT in a virtual environment, simply pass the -N option.
Because run_tests.sh is a wrapper around testrepository, it also accepts the same flags as testr. See the testr user manual for details about these additional flags.
Instead of running all tests, you can specify an individual directory, file, class, or method that contains test code.
To run the tests in the nova/tests/scheduler directory:
To run the tests in the nova/tests/virt/libvirt/test_libvirt.py file:
To run the tests in the CacheConcurrencyTestCase class in nova/tests/virt/libvirt/test_libvirt.py:
To run the ValidateIntegerTestCase.test_invalid_inputs test method in nova/tests/test_utils.py:
By default, the tests use the Python packages installed inside a virtualenv . (This is equivalent to using the -V, --virtualenv flag). If the virtualenv does not exist, it will be created the first time the tests are run.
If you wish to recreate the virtualenv, call run_tests.sh with the flag:
Recreating the virtualenv is useful if the package dependencies have changed since the virtualenv was last created. If the requirements.txt or tools/install_venv.py files have changed, it’s a good idea to recreate the virtualenv.
By default, the unit tests will see both the packages in the virtualenv and the packages that have been installed in the Python global environment. In some cases, the packages in the Python global environment may cause a conflict with the packages in the virtualenv. If this occurs, you can isolate the virtualenv from the global environment by using the flag:
-s, --no-site packages
If you do not wish to use a virtualenv at all, use the flag:
Running Tests from Shared Folders
If you are running the unit tests from a shared folder, you may see tests start to fail or stop completely as a result of Python lockfile issues . You can get around this by manually setting or updating the following line in nova/tests/conf_fixture.py:
Note that you may use any location (not just /tmp!) as long as it is not a shared folder.
|||See Continuous Integration with Jenkins.|
|||See Setting Up a Development Environment for more details about the use of virtualenv.|
|||There is an effort underway to use a fake DB implementation for the unit tests. See https://lists.launchpad.net/openstack/msg05604.html|
|||See Vish’s comment in this bug report: https://bugs.launchpad.net/nova/+bug/882933|