Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 78.68%. Comparing base (2f60642) to head (f486566).
⚠️ Report is 1 commits behind head on main.

@@           Coverage Diff           @@
##             main    #9959   +/-   ##
=======================================
  Coverage   78.67%   78.68%           
=======================================
  Files         153      153           
  Lines       19304    19311    +7     
  Branches     2212     2214    +2     
=======================================
+ Hits        15188    15195    +7     
  Misses       3817     3817           
  Partials      299      299           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

can you point to relevant issue which might be related to this fix?

No, but I can create one if it would be useful for you.

It seems that ExceptionGroup was adding in 3.11. I'm not sure we need to keep the sub-exceptions, I'll just include their messages.

can you also check why python 3.9 and 3.10 builds are failing with following tests?

=================================== FAILURES ===================================
___ test_DelayedDelivery.test_bind_queues_raises_exception_group_on_failures ___

self = <t.unit.worker.test_native_delayed_delivery.test_DelayedDelivery object at 0x7f99dee542b0>
mock_bind =

  @patch('celery.worker.consumer.delayed_delivery.bind_queue_to_native_delayed_delivery_exchange')
  def test_bind_queues_raises_exception_group_on_failures(self, mock_bind):
      """Test that ExceptionGroup is raised with all binding failures."""
      consumer_mock = MagicMock()
      consumer_mock.app.conf.broker_native_delayed_delivery_queue_type = \
          'classic'
      consumer_mock.app.conf.broker_url = 'amqp://'
  
      # Create three queues
      queue1 = Queue('queue1', exchange=Exchange('exchange1', type='topic'))
      queue2 = Queue('queue2', exchange=Exchange('exchange2', type='topic'))
      queue3 = Queue('queue3', exchange=Exchange('exchange3', type='topic'))
  
      consumer_mock.app.amqp.queues = {
          'queue1': queue1,
          'queue2': queue2,
          'queue3': queue3,
      }
  
      # Make queue1 and queue3 fail with different errors
      def bind_side_effect(connection, queue):
          if queue.name == 'queue1':
              raise ValueError("Queue1 binding failed")
          elif queue.name == 'queue3':
              raise RuntimeError("Queue3 binding failed")
  
      mock_bind.side_effect = bind_side_effect
  
      delayed_delivery = DelayedDelivery(consumer_mock)
  
      # Should raise ExceptionGroup containing both exceptions

  with pytest.raises(ExceptionGroup) as exc_info:

E NameError: name 'ExceptionGroup' is not defined

t/unit/worker/test_native_delayed_delivery.py:386: NameError
=============================== warnings summary ===============================
/home/runner/_work/celery/celery/celery/worker/consumer/consumer.py:517: CPendingDeprecationWarning: The broker_connection_retry configuration setting will no longer determine
whether broker connection retries are made during startup in Celery 6.0 and above.
If you wish to refrain from retrying connections on startup,
you should set broker_connection_retry_on_startup to False instead.
warnings.warn(

t/unit/worker/test_request.py::test_Request::test_from_message_invalid_kwargs
/home/runner/_work/celery/celery/celery/app/trace.py:705: RuntimeWarning: Exception raised outside body: InvalidTaskError('Task keyword arguments is not a mapping'):
Traceback (most recent call last):
File "/home/runner/_work/celery/celery/celery/app/trace.py", line 429, in trace_task
kwargs.items
AttributeError: 'str' object has no attribute 'items'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/runner/_work/celery/celery/celery/app/trace.py", line 431, in trace_task
    raise InvalidTaskError(
celery.exceptions.InvalidTaskError: Task keyword arguments is not a mapping

  warn(RuntimeWarning(

can you also check why python 3.9 and 3.10 builds are failing?

Yup! I'm already on it. It's because ExceptionGroup was adding in Python 3.11. I forgot to check this and was running locally with the latest version of Python. PEP654

@auvipy I've created an issue here: #9960

@auvipy:

can you also explore possibility to add integration test for this?

I've added an integration test which is able to reproduce part of the issue described in #9960 (I hope this works in the CI 🤞, it's working locally). This test starts the worker with an extra queue in task_queues and checks the bindings using the RabbitMQ management API. This requires updating the RabbitMQ image from latest to management (which is the same as latest but with the management plugin enabled by default).

I could in theory test delayed delivery in addition to the bindings, but as I see many of the integration tests are marked as flaky: I think keeping the scope of the test to the bindings created is probably wise.

@auvipy:

can you please look into this failing test?

It's really strange, as it was passing in this prior commit, with the diff between the two being enabling the rabbitmq management plugin, the exception message for the ExceptionGroup, and the addition of an integration test. The error message is also awful strange: Exception("<class 't.integration.tasks.UnpickleableException'>(['foo'])") was raised, while we're expecting celery.utils.serialization.UnpickleableExceptionWrapper which has the base class of Exception and wraps a t.integration.tasks.UnpickleableException ... they look very similar.

The thing is, I get the same error on main despite having passed previously. Other PRs such as #9962 are also failing with the same error. I think it may be due to a difference in the available package versions, as the package versions aren't pinned to an exact version, but rather with an interval allowing higher versions to be installed.

If I compare the pip freeze from the passing run and the pip freeze from the failing run, I see this:

--- passing.txt	2025-10-24 10:26:21.889142303 +0200
+++ failing.txt	2025-10-24 10:26:25.857341881 +0200
@@ -12,2 +12,2 @@
-boto3==1.40.55
-botocore==1.40.55
+boto3==1.40.56
+botocore==1.40.56
@@ -17,2 +17,2 @@
-cassandra-driver==3.29.2
-celery @ file:///home/runner/_work/celery/celery/.tox/.tmp/package/1/celery-5.6.0b2-0.editable-py3-none-any.whl#sha256=14d56489ca97ae53d64c3e6288c7c7e2cdcf46b09fa188d32a8a82647ffe644b
+cassandra-driver==3.29.3
+celery @ file:///home/runner/_work/celery/celery/.tox/.tmp/package/1/celery-5.6.0b2-0.editable-py3-none-any.whl#sha256=444fa6a249a4a54018ea7342007f53c7d5a42de8238e819449cd1bbc246e2eb2
@@ -46 +46 @@
-geomet==0.2.1.post1
+geomet==1.1.0
@@ -68 +68 @@
-kombu @ git+https://github.com/celery/kombu.git@1bef09386e936be9757e713682bdf38bc1c0b737
+kombu @ git+https://github.com/celery/kombu.git@b8cf1f68a9af3d129e57be52ce48d3f85ecc9a80
@@ -145 +145 @@
-tblib==3.1.0
+tblib==3.2.0

The tblib (Serialization library for Exceptions and Tracebacks) looks suspect. If I pin tblib==3.1.0 in requirements/extras/tblib.txt on main, the tests pass again. I think if you re-run the integration tests on main, we'll install tblib 3.2.0 and the tests will fail. I'm not really familiar with this library nor this code written by @ask 15 years ago, so I'm not sure what is really to be fixed here. Maybe we should replace the requirement with tblib>=1.5.0,tblib<3.2.0;python_version>='3.8.0' in requirements/extras/tblib.txt.

In any case, this failure doesn't seem to be related to my PR and we should probably fix it in a different PR.

yes the failures doesn't seems to be related to your pr