🐛 Fix hang in poller task in cancel_on_disconnect

packages/service-library/src/servicelib/fastapi/requests_decorators.py

@@ -13,8 +13,7 @@
 class _HandlerWithRequestArg(Protocol):
     __name__: str
 
-    async def __call__(self, request: Request, *args: Any, **kwargs: Any) -> Any:
-        ...
+    async def __call__(self, request: Request, *args: Any, **kwargs: Any) -> Any: ...
 
 
 def _validate_signature(handler: _HandlerWithRequestArg):
@@ -36,13 +35,15 @@ def _validate_signature(handler: _HandlerWithRequestArg):
 _POLL_INTERVAL_S: float = 0.01
 
 
-async def _disconnect_poller(request: Request, result: Any):
+async def _disconnect_poller(close_event: asyncio.Event, request: Request, result: Any):
     """
     Poll for a disconnect.
     If the request disconnects, stop polling and return.
     """
     while not await request.is_disconnected():
         await asyncio.sleep(_POLL_INTERVAL_S)
+        if close_event.is_set():
+            break
     return result
 
 
@@ -59,8 +60,12 @@ async def wrapper(request: Request, *args, **kwargs):
 
         # Create two tasks:
         # one to poll the request and check if the client disconnected
+        # sometimes canceling a task doesn't cancel the task immediately. If the poller task is not "killed" immediately, the client doesn't
+        # get a response, and the request "hangs". For this reason, we use an event to signal the poller task to stop.
+        # See: https://github.com/ITISFoundation/osparc-issues/issues/1922
+        kill_poller_event = asyncio.Event()
         poller_task = asyncio.create_task(
-            _disconnect_poller(request, sentinel),
+            _disconnect_poller(kill_poller_event, request, sentinel),
             name=f"cancel_on_disconnect/poller/{handler.__name__}/{id(sentinel)}",
         )
         # , and another which is the request handler
@@ -72,6 +77,7 @@ async def wrapper(request: Request, *args, **kwargs):
         done, pending = await asyncio.wait(
             [poller_task, handler_task], return_when=asyncio.FIRST_COMPLETED
         )
+        kill_poller_event.set()
 
         # One has completed, cancel the other
         for t in pending:

services/api-server/src/simcore_service_api_server/api/routes/files.py

@@ -287,6 +287,7 @@ async def upload_files(files: list[UploadFile] = FileParam(...)):
     response_model=ClientFileUploadData,
     responses=_FILE_STATUS_CODES,
 )
+@cancel_on_disconnect
 async def get_upload_links(
     request: Request,
     client_file: UserFileToProgramJob | UserFile,
@@ -421,6 +422,7 @@ async def abort_multipart_upload(
     response_model=OutputFile,
     responses=_FILE_STATUS_CODES,
 )
+@cancel_on_disconnect
 async def complete_multipart_upload(
     request: Request,
     file_id: UUID,