Experiment: pytest-timeout with the Sentry test isolation

.github/workflows/ci.yml

@@ -96,7 +96,8 @@ jobs:
   # Build and test the project
   build-lint-test:
     strategy:
-      fail-fast: true
+      # Let every matrix job finish so one flaky job does not hide the others.
+      fail-fast: false
       matrix:
         python: ["3.10", "3.14"]
         os: [ubuntu-latest, macos-intel, macos-arm, windows-latest]

bedrock_agentcore/strands_agent/pyproject.toml

@@ -21,6 +21,7 @@ dev = [
     "poethepoet>=0.36.0",
     "pytest>=7.1.2,<8",
     "pytest-asyncio>=0.18.3,<0.19",
+    "pytest-timeout>=2.2,<3",
 ]
 
 [build-system]
@@ -47,6 +48,8 @@ lint-types = "uv run --all-groups mypy --check-untyped-defs --namespace-packages
 test = "uv run pytest"
 
 [tool.pytest.ini_options]
+# Fail a test that hangs, with a stack dump, instead of stalling the job.
+timeout = 300
 asyncio_mode = "auto"
 
 [tool.ruff]

lambda_worker/pyproject.toml

@@ -15,6 +15,7 @@ dev = [
     "poethepoet>=0.36.0",
     "pytest>=7.1.2,<8",
     "pytest-asyncio>=0.18.3,<0.19",
+    "pytest-timeout>=2.2,<3",
 ]
 
 [build-system]
@@ -42,6 +43,8 @@ test = "uv run pytest"
 
 [tool.pytest.ini_options]
 asyncio_mode = "auto"
+# Fail a test that hangs, with a stack dump, instead of stalling the job.
+timeout = 300
 
 [tool.ruff]
 target-version = "py310"

openai_agents/mcp_v2/pyproject.toml

@@ -13,6 +13,7 @@ dev = [
     "mypy>=1.4.1,<2",
     "poethepoet>=0.36.0",
     "pytest>=9.0.3,<10",
+    "pytest-timeout>=2.2,<3",
 ]
 
 [tool.poe.tasks]
@@ -35,5 +36,7 @@ target-version = "py310"
 ignore_missing_imports = true
 
 [tool.pytest.ini_options]
+# Fail a test that hangs, with a stack dump, instead of stalling the job.
+timeout = 300
 pythonpath = ["."]
 testpaths = ["tests"]

pyproject.toml

@@ -20,6 +20,7 @@ dev = [
     "mypy>=1.4.1,<2",
     "pytest>=7.1.2,<8",
     "pytest-asyncio>=0.18.3,<0.19",
+    "pytest-timeout>=2.2,<3",
     "fakeredis>=2,<3",
     "frozenlist>=1.4.0,<2",
     "pyright>=1.1.394",
@@ -135,6 +136,8 @@ test = "uv run --all-groups --no-group cloud-export-to-parquet pytest"
 
 [tool.pytest.ini_options]
 asyncio_mode = "auto"
+# Fail a test that hangs, with a stack dump, instead of stalling the job.
+timeout = 300
 log_cli = true
 log_cli_level = "INFO"
 log_cli_format = "%(asctime)s [%(levelname)8s] %(message)s (%(filename)s:%(lineno)s)"

tests/sentry/test_interceptor.py

@@ -1,5 +1,7 @@
+import json
 import sys
 import unittest.mock
+import uuid
 from collections import abc
 
 import pytest
@@ -34,22 +36,26 @@ def transport() -> FakeSentryTransport:
 
 
 @pytest.fixture(autouse=True)
-def sentry_init(transport: FakeSentryTransport) -> None:
-    """Initialize Sentry for testing."""
-    sentry_sdk.init(
-        transport=transport,
-        integrations=[
-            AsyncioIntegration(),
-        ],
-    )
+def sentry_init(transport: FakeSentryTransport) -> abc.Iterator[None]:
+    """Give each test its own Sentry client and empty scopes, so nothing an
+    earlier test captured or left as a breadcrumb leaks into this one."""
+    for scope in (
+        sentry_sdk.get_global_scope(),
+        sentry_sdk.get_isolation_scope(),
+        sentry_sdk.get_current_scope(),
+    ):
+        scope.clear()
+    sentry_sdk.init(transport=transport, integrations=[AsyncioIntegration()])
+    yield
+    sentry_sdk.get_client().close()
 
 
 @pytest.fixture
 async def worker(client: Client) -> abc.AsyncIterator[Worker]:
     """Fixture to provide a worker for testing."""
     async with Worker(
         client,
-        task_queue="sentry-task-queue",
+        task_queue=f"sentry-{uuid.uuid4()}",
         workflows=[SentryExampleWorkflow],
         activities=[broken_activity, working_activity],
         interceptors=[SentryInterceptor()],
@@ -71,7 +77,7 @@ async def test_sentry_interceptor_reports_no_errors_when_workflow_succeeds(
         await client.execute_workflow(
             SentryExampleWorkflow.run,
             SentryExampleWorkflowInput(option="working"),
-            id="sentry-workflow-id",
+            id=f"sentry-workflow-{uuid.uuid4()}",
             task_queue=worker.task_queue,
         )
     except Exception:
@@ -85,12 +91,13 @@ async def test_sentry_interceptor_captures_errors(
     client: Client, worker: Worker, transport: FakeSentryTransport
 ) -> None:
     """Test that errors are captured with correct Sentry metadata."""
+    workflow_id = f"sentry-workflow-{uuid.uuid4()}"
     # WHEN
     try:
         await client.execute_workflow(
             SentryExampleWorkflow.run,
             SentryExampleWorkflowInput(option="broken"),
-            id="sentry-workflow-id",
+            id=workflow_id,
             task_queue=worker.task_queue,
         )
         pytest.fail("Workflow should raise an exception")
@@ -99,7 +106,28 @@ async def test_sentry_interceptor_captures_errors(
 
     # THEN
     # there should be two events: one for the failed activity and one for the failed workflow
-    assert len(transport.events) == 2, "Two events should be captured"
+    assert len(transport.events) == 2, (
+        "Two events should be captured; got: "
+        + json.dumps(
+            [
+                {
+                    k: dict(e).get(k)
+                    for k in (
+                        "level",
+                        "logger",
+                        "message",
+                        "logentry",
+                        "exception",
+                        "transaction",
+                        "tags",
+                        "extra",
+                    )
+                }
+                for e in transport.events
+            ],
+            default=str,
+        )
+    )
 
     # Check the first event - should be the activity exception
     # --------------------------------------------------------
@@ -114,10 +142,10 @@ async def test_sentry_interceptor_captures_errors(
         "temporal.execution_type": "activity",
         "module": "sentry.activity.broken_activity",
         "temporal.workflow.type": "SentryExampleWorkflow",
-        "temporal.workflow.id": "sentry-workflow-id",
+        "temporal.workflow.id": workflow_id,
         "temporal.activity.id": "1",
         "temporal.activity.type": "broken_activity",
-        "temporal.activity.task_queue": "sentry-task-queue",
+        "temporal.activity.task_queue": worker.task_queue,
         "temporal.workflow.namespace": "default",
         "temporal.workflow.run_id": unittest.mock.ANY,
     }
@@ -146,8 +174,8 @@ async def test_sentry_interceptor_captures_errors(
         "temporal.execution_type": "workflow",
         "module": "sentry.workflow.SentryExampleWorkflow.run",
         "temporal.workflow.type": "SentryExampleWorkflow",
-        "temporal.workflow.id": "sentry-workflow-id",
-        "temporal.workflow.task_queue": "sentry-task-queue",
+        "temporal.workflow.id": workflow_id,
+        "temporal.workflow.task_queue": worker.task_queue,
         "temporal.workflow.namespace": "default",
         "temporal.workflow.run_id": unittest.mock.ANY,
     }