Update `FakeChannel.await_result(...)` to drive async Rust (Tokio runtime/thread pool) by MadLittleMods · Pull Request #19879 · element-hq/synapse

MadLittleMods · 2026-06-25T00:58:34Z

Update FakeChannel.await_result(...) to drive async Rust (Tokio runtime/thread pool)

Split off from #19878 (needed to drive requests in tests that use async Rust)

Follow-up to #19871

Pull Request Checklist

Pull request is based on the develop branch
Pull request includes a changelog file. The entry should:
- Be a short description of your change which makes sense to users. "Fixed a bug that prevented receiving messages from other servers." instead of "Moved X method from EventStore to EventWorkerStore.".
- Use markdown where necessary, mostly for code blocks.
- End with either a period (.) or an exclamation mark (!).
- Start with a capital letter.
- Feel free to credit yourself, by adding a sentence "Contributed by @github_username." or "Contributed by [Your Name]." to the end of the entry.
Code style is correct (run the linters)

Split off from #19878

Same thing we did in #19871

…Case_new.test_wait_for_sync_token`

MadLittleMods · 2026-06-25T21:24:07Z

+        # `notifier.wait_for_stream_token(from_token)` only checks every 500ms so we
+        # need to match that in order to make sure we hit the wake-up for sure.
+        channel.await_result(timeout_ms=500)


Previously, this worked because await_result(...) would advance 0.1s past the specified timeout_ms and I guess that worked out to hit the 500ms notifier.wait_for_stream_token(from_token) check interval every time.

Now await_result(...) can timeout when we exactly hit the timeout_ms so we weren't hitting it.

I've updated the second wait to be 500ms to match notifier.wait_for_stream_token(from_token) so we know that we can always hit the interval regardless of the current Twisted clock time.

MadLittleMods · 2026-06-25T21:38:37Z

@@ -0,0 +1 @@
+Update `HomeserverTestCase.get_success(...)` and friends to drive async Rust (Tokio runtime/thread pool).


Same changelog as #19871 so they merge

We can consider FakeChannel.await_result(...) one of the "friends" of HomeserverTestCase.get_success(...)

MadLittleMods · 2026-06-25T21:39:02Z

+        # TODO: Why?
        self._reactor.run()


Prior art. Something to figure out in a future PR. Just noting that I don't know. The reactor should already be running and we can potentially remove it in another PR.

MadLittleMods · 2026-06-25T21:39:22Z

@@ -301,15 +302,85 @@ def transport(self) -> "FakeChannel":
    def await_result(self, timeout_ms: int = 1000) -> None:


Similar structure to what we've done in #19871

See #19879 See #19878 (comment)

…lt-rust-changes

Based one explanation from #19913 (comment)

``` Error: Traceback (most recent call last): File "/home/runner/work/synapse/synapse/tests/test_mau.py", line 74, in test_simple_deny_mau self.do_sync_for_user(token2) File "/home/runner/work/synapse/synapse/tests/test_mau.py", line 345, in do_sync_for_user raise HttpResponseException( synapse.api.errors.ProxiedRequestError: 403: Monthly Active User Limit Exceeded tests.test_mau.TestMauLimit.test_simple_deny_mau ```

See #19879 (comment)

MadLittleMods · 2026-07-06T23:32:14Z

            if user_id:
                timestamp = await self.store.user_last_seen_monthly_active(user_id)
-                if timestamp:
+                if timestamp is not None:


Fixing a flawed check (0 is falsey) that was exposed by #19879 (comment)

… doesn't care about expiration We don't need to assert exact values. These tests don't care about that

MadLittleMods · 2026-07-07T00:18:25Z

        self.assertEqual(channel.json_body["data"], "foo=bar")
        self.assertEqual(channel.json_body["sequence_token"], sequence_token)
-        self.assertEqual(channel.json_body["expires_in_ms"], expires_in_ms - 100)
+        self.assertGreater(channel.json_body["expires_in_ms"], 0)


Had to touch this because await_result(...) changed the minimum amount of time we advance by. These tests relied on that internal detail.

We don't need to test exact expires_in_ms values. We just care that it's > 0.

And we don't care about testing it on subsequent requests either. This test has nothing to do with expiration.

If we wanted to do this kind of thing, we should do something like the following:

duration_between_requests = Duration(milliseconds=100) # Advance by some known amount self.reactor.advance(duration_between_requests.as_secs()) # ... self.assertGreater(channel.json_body["expires_in_ms"], 0) self.assertLessEqual( channel.json_body["expires_in_ms"], prev_expires_in_ms - duration_between_requests.as_millis(), ) prev_expires_in_ms = channel.json_body["expires_in_ms"]

Related discussion: #19539 (comment)

…elimit` ``` [FAIL] Traceback (most recent call last): File "/home/runner/.cache/pypoetry/virtualenvs/matrix-synapse-pswDeSvb-py3.10/lib/python3.10/site-packages/parameterized/parameterized.py", line 620, in standalone_func return func(*(a + p.args), **p.kwargs, **kw) File "/home/runner/work/synapse/synapse/tests/rest/client/test_delayed_events.py", line 434, in test_send_delayed_event_ratelimit self.assertEqual(HTTPStatus.OK, channel.code, channel.result) File "/home/runner/.cache/pypoetry/virtualenvs/matrix-synapse-pswDeSvb-py3.10/lib/python3.10/site-packages/twisted/trial/_synctest.py", line 444, in assertEqual super().assertEqual(first, second, msg) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 845, in assertEqual assertion_func(first, second, msg=msg) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 838, in _baseAssertEqual raise self.failureException(msg) twisted.trial.unittest.FailTest: <HTTPStatus.OK: 200> != 429 : {'version': b'1.1', 'code': b'429', 'reason': b'Unknown Status', 'headers': Headers({b'Server': [b'1'], b'Date': [b'Tue, 07 Jul 2026 00:20:42 GMT'], b'Retry-After': [b'1'], b'Content-Type': [b'application/json'], b'Cache-Control': [b'no-cache, no-store, must-revalidate'], b'Access-Control-Allow-Origin': [b'*'], b'Access-Control-Allow-Methods': [b'GET, HEAD, POST, PUT, DELETE, OPTIONS'], b'Access-Control-Allow-Headers': [b'X-Requested-With, Content-Type, Authorization, Date'], b'Access-Control-Expose-Headers': [b'Synapse-Trace-Id, Server']}), 'body': b'{"errcode":"M_LIMIT_EXCEEDED","error":"Too Many Requests","retry_after_ms":396}', 'done': True} tests.rest.client.test_delayed_events.DelayedEventsTestCase.test_send_delayed_event_ratelimit_0 tests.rest.client.test_delayed_events.DelayedEventsTestCase.test_send_delayed_event_ratelimit_1 ```

…estCases` ``` [FAIL] Traceback (most recent call last): File "/home/runner/work/synapse/synapse/tests/federation/test_federation_sender.py", line 790, in test_prune_outbound_device_pokes1 self.assertGreaterEqual(mock_send_txn.call_count, 4) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 1250, in assertGreaterEqual self.fail(self._formatMessage(msg, standardMsg)) twisted.trial.unittest.FailTest: 2 not greater than or equal to 4 tests.federation.test_federation_sender.FederationSenderDevicesTestCases.test_prune_outbound_device_pokes1 =============================================================================== [FAIL] Traceback (most recent call last): File "/home/runner/work/synapse/synapse/tests/federation/test_federation_sender.py", line 847, in test_prune_outbound_device_pokes2 self.assertGreaterEqual(mock_send_txn.call_count, 3) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 1250, in assertGreaterEqual self.fail(self._formatMessage(msg, standardMsg)) twisted.trial.unittest.FailTest: 2 not greater than or equal to 3 tests.federation.test_federation_sender.FederationSenderDevicesTestCases.test_prune_outbound_device_pokes2 =============================================================================== [FAIL] Traceback (most recent call last): File "/home/runner/work/synapse/synapse/tests/federation/test_federation_sender.py", line 743, in test_unreachable_server self.assertGreaterEqual(mock_send_txn.call_count, 4) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 1250, in assertGreaterEqual self.fail(self._formatMessage(msg, standardMsg)) twisted.trial.unittest.FailTest: 2 not greater than or equal to 4 tests.federation.test_federation_sender.FederationSenderDevicesTestCases.test_unreachable_server =============================================================================== [FAIL] Traceback (most recent call last): File "/home/runner/work/synapse/synapse/tests/federation/test_federation_sender.py", line 594, in test_upload_signatures self.assertEqual(len(self.edus), 2) File "/home/runner/.cache/pypoetry/virtualenvs/matrix-synapse-pswDeSvb-py3.10/lib/python3.10/site-packages/twisted/trial/_synctest.py", line 444, in assertEqual super().assertEqual(first, second, msg) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 845, in assertEqual assertion_func(first, second, msg=msg) File "/opt/hostedtoolcache/Python/3.10.20/x64/lib/python3.10/unittest/case.py", line 838, in _baseAssertEqual raise self.failureException(msg) twisted.trial.unittest.FailTest: 1 != 2 tests.federation.test_federation_sender.FederationSenderDevicesTestCases.test_upload_signatures ```

MadLittleMods · 2026-07-07T08:00:36Z

        self.edus.extend(data["edus"])
        return {}

+    def wait_for_device_list_updates_to_be_sent(self) -> None:


Added wait_for_device_list_updates_to_be_sent to be more precise

MadLittleMods · 2026-07-07T08:03:36Z

+        # Wait some time in between each device list update as we want each of them to
+        # be attempted to be sent in their own transaction
+        self.reactor.advance(Duration(seconds=1).as_secs())


The main fixes to these tests was adding some time in between each device list update so they are sent in their own transaction. Tests assert self.assertGreaterEqual(mock_send_txn.call_count

Previously, we advanced 100ms on every login attempt because of await_result(...). await_result(...) now only advances 1ms and we need to space it out more.

MadLittleMods · 2026-07-07T08:04:46Z

+                # Note: The extra 1ms is from `make_request(...)` -> `await_result(...)`
+                #
+                # FIXME: This test shouldn't care about this internal detail (don't
+                # assert exact timing)
+                last_seen=123456001,


Something for a future PR

erikjohnston

Looks good!

MadLittleMods · 2026-07-07T09:07:34Z

Thanks for the review @erikjohnston 🐍

Try FakeChannel.await_result(...) changes on their own

f5710de

Split off from #19878

MadLittleMods added rust Z-Rust A-Testing labels Jun 25, 2026

MadLittleMods changed the title ~~Try FakeChannel.await_result(...) changes on their own~~ Try FakeChannel.await_result(...) changes to drive async Rust on their own Jun 25, 2026

Add changelog

bb9e266

MadLittleMods mentioned this pull request Jun 25, 2026

Rust database access via Python database connection pool v2 #19878

Merged

5 tasks

MadLittleMods added 2 commits June 24, 2026 21:13

Fix infinite loop getting the tests stuck in CI

0fb3722

Fix typo

c10f8cb

MadLittleMods mentioned this pull request Jun 25, 2026

Update HomeserverTestCase.get_success(...) and friends to drive async Rust (Tokio runtime/thread pool) #19871

Merged

5 tasks

MadLittleMods added 3 commits June 25, 2026 16:14

time.sleep(0.001) after a few loops

0281dd2

Same thing we did in #19871

Use 1s real-time instead of timeout_ms arg

842fd59

Fix `tests.rest.client.sliding_sync.test_sliding_sync.SlidingSyncTest…

3565a7f

…Case_new.test_wait_for_sync_token`

MadLittleMods commented Jun 25, 2026

View reviewed changes

MadLittleMods changed the title ~~Try FakeChannel.await_result(...) changes to drive async Rust on their own~~ Update FakeChannel.await_result(...) and friends to drive async Rust (Tokio runtime/thread pool) Jun 25, 2026

Update changelog

7c19a47

MadLittleMods commented Jun 25, 2026

View reviewed changes

MadLittleMods changed the title ~~Update FakeChannel.await_result(...) and friends to drive async Rust (Tokio runtime/thread pool)~~ Update FakeChannel.await_result(...) to drive async Rust (Tokio runtime/thread pool) Jun 25, 2026

MadLittleMods added a commit that referenced this pull request Jun 25, 2026

Backport FakeChannel.await_result(...) changes from #19879

1f08bc0

See #19879 See #19878 (comment)

MadLittleMods added 6 commits June 25, 2026 16:43

Merge branch 'develop' into madlittlemods/try-fake-channel-await-resu…

622ec3d

…lt-rust-changes

Merge branch 'develop' into madlittlemods/try-fake-channel-await-resu…

ade5ef6

…lt-rust-changes

Merge branch 'develop' into madlittlemods/try-fake-channel-await-resu…

60a1507

…lt-rust-changes

More clear explanation

81cbffa

Explain why we advance(0) right away

c349a33

Remove real-time timeout

b64e067

MadLittleMods mentioned this pull request Jul 6, 2026

Fix flaky 3PID inhibit error unit tests #19913

Merged

3 tasks

MadLittleMods added 2 commits July 6, 2026 17:33

Better reasoning

7a1bd83

Based one explanation from #19913 (comment)

Add more reasoning why

e9d6aa3

MadLittleMods commented Jul 6, 2026

View reviewed changes

Comment thread changelog.d/19879.misc

MadLittleMods commented Jul 6, 2026

View reviewed changes

Comment thread tests/server.py Outdated

Advance by CLOCK_SCHEDULE_EPSILON

5c4a3b3

See #19879 (comment)

MadLittleMods commented Jul 6, 2026

View reviewed changes

MadLittleMods added 5 commits July 6, 2026 18:54

Record start_time_seconds after previous one-off advance

21af1ca

We have to advance by 1ms

62bb99a

tests.rest.client.test_msc4388_rendezvous.RendezvousServletTestCase…

4107108

… doesn't care about expiration We don't need to assert exact values. These tests don't care about that

Fix lints

15dd368

Remove extra comment

609f7a0

MadLittleMods commented Jul 7, 2026

View reviewed changes

MadLittleMods added 5 commits July 6, 2026 19:28

No need to test specifics of sticky TTL in /sync

c08f96f

No need to test specifics of sticky TTL when sending events

a7d6999

Just use exact timing and move on

fbeb0cc

MadLittleMods commented Jul 7, 2026

View reviewed changes

MadLittleMods added 2 commits July 7, 2026 03:17

Fix possible typo

27b3366

Fix switch typo

d4f4ddd

MadLittleMods marked this pull request as ready for review July 7, 2026 08:34

MadLittleMods requested a review from a team as a code owner July 7, 2026 08:34

erikjohnston self-assigned this Jul 7, 2026

erikjohnston approved these changes Jul 7, 2026

View reviewed changes

MadLittleMods merged commit 026937b into develop Jul 7, 2026
45 of 47 checks passed

MadLittleMods deleted the madlittlemods/try-fake-channel-await-result-rust-changes branch July 7, 2026 09:07

anoadragon453 mentioned this pull request Jul 7, 2026

Revert "Fix flaky 3PID inhibit error unit tests" #19916

Open

		@@ -0,0 +1 @@
		Update `HomeserverTestCase.get_success(...)` and friends to drive async Rust (Tokio runtime/thread pool).

		@@ -301,15 +302,85 @@ def transport(self) -> "FakeChannel":
		def await_result(self, timeout_ms: int = 1000) -> None:

Uh oh!

Conversation

MadLittleMods commented Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Checklist

Uh oh!

MadLittleMods Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jun 25, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

MadLittleMods Jul 6, 2026

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jul 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jul 7, 2026

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jul 7, 2026

Choose a reason for hiding this comment

Uh oh!

MadLittleMods Jul 7, 2026

Choose a reason for hiding this comment

Uh oh!

erikjohnston left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

MadLittleMods commented Jul 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

MadLittleMods commented Jun 25, 2026 •

edited

Loading

MadLittleMods Jun 25, 2026 •

edited

Loading

MadLittleMods Jul 7, 2026 •

edited

Loading