Flaky is a plugin for pytest that automatically reruns flaky tests.

Ideally, tests reliably pass or fail, but sometimes test fixtures must rely on components that aren't 100% reliable. With flaky, instead of removing those tests or marking them to @skip, they can be automatically retried.

For more information about flaky, see this presentation.

To mark a test as flaky, simply import flaky and decorate the test with @flaky:

from flaky import flaky

@flaky
def test_something_that_usually_passes(self):
    value_to_double = 21
    result = get_result_from_flaky_doubler(value_to_double)
    self.assertEqual(result, value_to_double * 2, 'Result doubled incorrectly.')

By default, flaky will retry a failing test once, but that behavior can be overridden by passing values to the flaky decorator. It accepts two parameters: max_runs, and min_passes; flaky will run tests up to max_runs times, until it has succeeded min_passes times. Once a test passes min_passes times, it's considered a success; once it has been run max_runs times without passing min_passes times, it's considered a failure.

@flaky(max_runs=3, min_passes=2)
def test_something_that_usually_passes(self):
    """This test must pass twice, and it can be run up to three times."""
    value_to_double = 21
    result = get_result_from_flaky_doubler(value_to_double)
    self.assertEqual(result, value_to_double * 2, 'Result doubled incorrectly.')

In addition to marking a single test flaky, entire test cases can be marked flaky:

@flaky
class TestMultipliers(TestCase):
    def test_flaky_doubler(self):
        value_to_double = 21
        result = get_result_from_flaky_doubler(value_to_double)
        self.assertEqual(result, value_to_double * 2, 'Result doubled incorrectly.')

    @flaky(max_runs=3)
    def test_flaky_tripler(self):
        value_to_triple = 14
        result = get_result_from_flaky_tripler(value_to_triple)
        self.assertEqual(result, value_to_triple * 3, 'Result tripled incorrectly.')

The @flaky class decorator will mark test_flaky_doubler as flaky, but it won't override the 3 max_runs for test_flaky_tripler (from the decorator on that test method).

When using pytest, @pytest.mark.flaky can be used in place of @flaky.

Depending on your tests, some failures are obviously not due to flakiness. Instead of rerunning after those failures, you can specify a filter function that can tell flaky to fail the test right away.

def is_not_crash(err, *args):
    return not issubclass(err[0], ProductCrashedError)

@flaky
def test_something():
    raise ProductCrashedError

@flaky(rerun_filter=is_not_crash)
def test_something_else():
    raise ProductCrashedError

Flaky will run test_something twice, but will only run test_something_else once.

It can also be used to incur a delay between test retries:

import time

def delay_rerun(*args):
    time.sleep(1)
    return True

@flaky(rerun_filter=delay_rerun)
def test_something_else():
    ...

With pytest, flaky will automatically run. It can, however be disabled via the command line:

pytest -p no:flaky

Pass --no-flaky-report to suppress the report at the end of the run detailing flaky test results.

Pass --no-success-flaky-report to suppress information about successful flaky tests.

Pass --force-flaky to treat all tests as flaky.

Pass --max-runs=MAX_RUNS and/or --min-passes=MIN_PASSES to control the behavior of flaky if --force-flaky is specified. Flaky decorators on individual tests will override these defaults.

Additional usage examples are in the code - see test/test_pytest/test_pytest_example.py

To install, simply:

pip install flaky

Flaky is tested with the following test runners and options:

• Py.test. Works with pytest-xdist but not with the --boxed option. Doctests cannot be marked flaky.

See CONTRIBUTING.rst.

Create a virtual environment and install packages -

mkvirtualenv flaky
pip install -r requirements-dev.txt

Run all tests using -

tox

The tox tests include code style checks via pycodestyle and pylint.

Copyright 2015 Box, Inc. All rights reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

   http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.