This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author aeros
Recipients aeros, asvetlov, vstinner, yselivanov
Date 2020-04-01.20:19:49
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1585772389.71.0.412184895317.issue40115@roundup.psfhosted.org>
In-reply-to
Content
I was able to find a fix! Specifically, I think the issue was a subtle problem with test_run_in_executor_cancel: it never properly shuts down the executor prior to closing the event loop. 

We do this in asyncio.run() (with the loop.shutdown_default_executor() that I added last year), but it should be called explicitly when using loop.close() directly. Otherwise, the resources of the executor may not be cleaned up properly. I think the issue was being covered up because of the usage of daemon threads in the ThreadPoolExecutor, but revealed itself when the workers were converted to non-daemon threads since they require proper cleanup.

The change is very minimal:

```
    def test_run_in_executor_cancel(self):
        called = False

        def patched_call_soon(*args):
            nonlocal called
            called = True

        def run():
            time.sleep(0.05)

        f2 = self.loop.run_in_executor(None, run)
        f2.cancel()
+      self.loop.run_until_complete(
+          self.loop.shutdown_default_executor())
        self.loop.close()
        self.loop.call_soon = patched_call_soon
        self.loop.call_soon_threadsafe = patched_call_soon
        time.sleep(0.4)
        self.assertFalse(called)
```

Before change:
```
[aeros:~/repos/aeros-cpython]$ ./python -m test --fail-env-changed -R 3:3 test_asyncio -m test.test_asyncio.test_events.EPollEventLoopTests.test_run_in_executor_cancel
0:00:00 load avg: 0.63 Run tests sequentially
0:00:00 load avg: 0.63 [1/1] test_asyncio
beginning 6 repetitions
123456
......
test_asyncio leaked [1, 1, 1] references, sum=3
test_asyncio leaked [2, 1, 1] memory blocks, sum=4
test_asyncio failed

== Tests result: FAILURE ==

1 test failed:
    test_asyncio

Total duration: 3.2 sec
Tests result: FAILURE
```

After change:
```
[aeros:~/repos/aeros-cpython]$ ./python -m test --fail-env-changed -R 3:3 test_asyncio -m test.test_asyncio.test_events.EPollEventLoopTests.test_run_in_executor_cancel
0:00:00 load avg: 0.24 Run tests sequentially
0:00:00 load avg: 0.24 [1/1] test_asyncio
beginning 6 repetitions
123456
......

== Tests result: SUCCESS ==

1 test OK.

Total duration: 3.5 sec
Tests result: SUCCESS
```

I'll also test the PR using the `test-with-buildbots` label to double check, it should be attached to this issue within the next couple of hours.
History
Date User Action Args
2020-04-01 20:19:49aerossetrecipients: + aeros, vstinner, asvetlov, yselivanov
2020-04-01 20:19:49aerossetmessageid: <1585772389.71.0.412184895317.issue40115@roundup.psfhosted.org>
2020-04-01 20:19:49aeroslinkissue40115 messages
2020-04-01 20:19:49aeroscreate