Skip to content

fix(taskbroker): Handle signals cooperatively to prevent semaphore errors #763

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
sentry wants to merge 1 commit into from
Closed

fix(taskbroker): Handle signals cooperatively to prevent semaphore errors #763

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 12 additions & 6 deletions clients/python/src/taskbroker_client/worker/worker.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -581,6 +581,7 @@ def __init__(
self._metrics = app.metrics

self._grpc_sync_event = self._mp_context.Event()
self._shutdown_requested = threading.Event()

self._gettask_backoff_seconds = 0
self._setstatus_backoff_seconds = 0
Expand All @@ -595,20 +596,23 @@ def start(self) -> int:
self.worker_pool.start_result_thread()
self.worker_pool.start_spawn_children_thread()

# Convert signals into KeyboardInterrupt.
# Running shutdown() within the signal handler can lead to deadlocks
# Set a threading flag and wake up any blocked multiprocessing.Event.wait() call
# rather than raising KeyboardInterrupt. Raising KeyboardInterrupt inside
# multiprocessing.Event.wait() can corrupt the internal semaphore state, causing
# "ValueError: semaphore or lock released too many times".
def signal_handler(*args: Any) -> None:
raise KeyboardInterrupt()
self._shutdown_requested.set()
self._grpc_sync_event.set()

signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)

try:
while True:
while not self._shutdown_requested.is_set():
self.run_once()
except KeyboardInterrupt:
finally:
self.shutdown()
raise
return 0

def run_once(self) -> None:
"""Access point for tests to run a single worker loop"""
Expand Down Expand Up @@ -691,6 +695,8 @@ def _send_update_task(

def fetch_task(self) -> InflightTaskActivation | None:
self._grpc_sync_event.wait(self._gettask_backoff_seconds)
if self._shutdown_requested.is_set():
return None
try:
activation = self.client.get_task(self._namespace)
except grpc.RpcError as e:
Expand Down
Loading