tests.unit.pubsub_v1.subscriber.test_streaming_pull_manager: test_close_inactive_consumer failed

[flaky-bot]

This test failed!

To configure my behavior, see the Flaky Bot documentation.

If I'm commenting on this issue too often, add the flakybot: quiet label and
I will stop commenting.

---

commit: d40d027
buildURL: Build Status, Sponge
status: failed

Test output

def test_close_inactive_consumer():
        (
            manager,
            consumer,
            dispatcher,
            leaser,
            heartbeater,
            scheduler,
        ) = make_running_manager()
        consumer.is_active = False
    manager.close()

    consumer.stop.assert_not_called()
    leaser.stop.assert_called_once()


  dispatcher.stop.assert_called_once()


tests/unit/pubsub_v1/subscriber/test_streaming_pull_manager.py:589:

_mock_self = 
def assert_called_once(_mock_self):
    """assert that the mock was called only once.
    """
    self = _mock_self
    if not self.call_count == 1:
        msg = ("Expected '%s' to have been called once. Called %s times.%s"
               % (self._mock_name or 'mock',
                  self.call_count,
                  self._calls_repr()))


      raise AssertionError(msg)


E           AssertionError: Expected 'stop' to have been called once. Called 0 times.
.nox/unit-3-7/lib/python3.7/site-packages/mock/mock.py:891: AssertionError

[flaky-bot]

Looks like this issue is flaky. 😟

I'm going to leave this open and stop commenting.

A human should fix and close this.

---

When run at the same commit (d40d027), this test passed in one build (Build Status, Sponge) and failed in another build (Build Status, Sponge).

[plamut]

I found where the problem is. The streaming pull manager's close() is now non-blocking, shutdown is performed in a background thread.

The test calls manager.close() and then immediately proceeds to assertions, meaning that there's a race between the main test thread and the background shutdown thread. I was not able to reproduce the issue on my machine, but was able to trigger it if I added a sleep for a fraction of a second to the _shutdown() method.

I'll see if there's an easy way to somehow synchronize with the background thread, or we might need to refactor the test to invoke the _shutdown() method directly.