celery_backend/config: Fix enabling of celery integration for sentry

About a year ago, we updated the way sentry is configured for celery
tasks by not relying anymore on the worker_init signal and the
SWH_SENTRY_DSN environment variable but rather by using a task_prerun
signal callback that sets sentry DSN through a dict mapping a task name
to its sentry DSN.

But as a result it broke our celery integration for sentry (that notably
adds task name and parameters in sentry events) as integrations must be
initialized in the worker_init signal callback to work properly. So ensure
they are by first setting a fake DSN in the worker_init signal callback and
update tests to check celery integration is enabled.

Related to swh/meta#4949.

swh/scheduler/celery_backend/config.py

@@ -6,6 +6,7 @@
 import functools
 import logging
 import os
+import sys
 from time import monotonic as _monotonic
 import traceback
 from typing import Any, Dict, Optional
@@ -176,9 +177,23 @@ def _init_sentry(sentry_dsn: str, main_package: Optional[str] = None):
 @worker_init.connect
 @_print_errors
 def on_worker_init(*args, **kwargs):
-    sentry_dsn = None  # will be set in `init_sentry` function
+    # use a fake sentry DSN to ensure celery integration for sentry is
+    # properly configured as it must happen in the worker_init signal
+    # callback, real sentry DSN is then setup in task_prerun signal
+    # callback (see celery_task_prerun function below)
+    sentry_dsn = "https://public@sentry.example.org/1"
     _init_sentry(sentry_dsn)
 
+    if "pytest" in sys.argv[0] or "PYTEST_XDIST_WORKER" in os.environ:
+        # when pytest collects tests, it breaks the proper configuration
+        # of the celery integration as a side effect, so we ensure that
+        # the celery.worker.consumer.build_tracer function gets overridden
+        # as it should have be
+        from celery.app import trace
+        from celery.worker.consumer import consumer
+
+        consumer.build_tracer = trace.build_tracer
+
 
 @Panel.register
 def monotonic(state):

swh/scheduler/tests/tasks.py

@@ -33,7 +33,7 @@ def multiping(self, n=10):
 
 
 @shared_task(name=TASK_ERROR)
-def not_implemented():
+def not_implemented(*args, **kwargs):
     raise NotImplementedError("Nope")
 
 

swh/scheduler/tests/test_celery_tasks.py

@@ -200,13 +200,25 @@ def test_run_ready_task_with_priority(
 
 
 def test_task_exception(
-    swh_scheduler_celery_app, swh_scheduler_celery_worker, swh_scheduler
+    swh_scheduler_celery_app,
+    swh_scheduler_celery_worker,
+    swh_scheduler,
+    mocker,
 ):
+    from sentry_sdk.integrations import celery
+
+    # using the sentry_events fixture does not work for celery tasks so
+    # we spy the capture_event method from the client of the sentry Hub
+    # used in celery integration instead
+    capture_event = mocker.spy(celery.Hub.current.client, "capture_event")
+
     task_type = swh_scheduler.get_task_type("swh-test-error")
     assert task_type
     assert task_type["backend_name"] == TASK_ERROR
 
-    swh_scheduler.create_tasks([create_task_dict("swh-test-error", "oneshot")])
+    swh_scheduler.create_tasks(
+        [create_task_dict("swh-test-error", "oneshot", "arg", kwarg="kwarg")]
+    )
 
     backend_tasks = run_ready_tasks(swh_scheduler, swh_scheduler_celery_app)
     assert len(backend_tasks) == 1
@@ -216,6 +228,16 @@ def test_task_exception(
     with pytest.raises(NotImplementedError):
         result.get()
 
+    # check celery integration for sentry is enabled
+    assert capture_event.mock_calls
+    sentry_event_extra = capture_event.mock_calls[0].kwargs["event"]["extra"]
+    assert "celery-job" in sentry_event_extra
+    assert sentry_event_extra["celery-job"] == {
+        "task_name": TASK_ERROR,
+        "args": ["arg"],
+        "kwargs": {"kwarg": "kwarg"},
+    }
+
 
 def test_statsd(swh_scheduler_celery_app, swh_scheduler_celery_worker, mocker):
     m = mocker.patch("swh.scheduler.task.Statsd._send_to_server")