Multiple fixes related to SharedProcessPool & MultiProcessingStage

python/morpheus/morpheus/stages/general/multi_processing_stage.py

@@ -31,8 +31,27 @@
 
 
 class MultiProcessingBaseStage(SinglePortStage, typing.Generic[InputT, OutputT]):
+    """
+    Base class for all MultiProcessing stages that make use of the SharedProcessPool.
+
+    Parameters
+    ----------
+    c : `morpheus.config.Config`
+        Pipeline configuration instance.
+    process_pool_usage : float
+        The fraction of the process pool workers that this stage could use. Should be between 0 and 1.
+    max_in_flight_messages : int, default = None
+        The number of progress engines used by the stage. If None, it will be set to 1.5 times the total
+        number of process pool workers
+
+    Raises
+    ------
+    ValueError
+        If the process pool usage is not between 0 and 1.
+    """
 
     def __init__(self, *, c: Config, process_pool_usage: float, max_in_flight_messages: int = None):
+
         super().__init__(c=c)
 
         if not 0 <= process_pool_usage <= 1:
@@ -50,36 +69,43 @@ def __init__(self, *, c: Config, process_pool_usage: float, max_in_flight_messag
 
     def accepted_types(self) -> typing.Tuple:
         """
-        There are two approaches to inherit from this class:
-            - With generic types: MultiProcessingDerivedStage(MultiProcessingBaseStage[InputT, OutputT])
-            - With concrete types: MultiProcessingDerivedStage(MultiProcessingBaseStage[int, str])
+        Accepted input types for this stage are returned.
+
+        Raises
+        ------
+        RuntimeError
+            If the accepted types cannot be deduced from either __orig_class__ or __orig_bases__.
 
-        When inheriting with generic types, the derived class can be instantiated like this:
+        Returns
+        -------
+        typing.Tuple
+            Accepted input types.
+        """
 
-            stage = MultiProcessingDerivedStage[int, str]()
+        # There are two approaches to inherit from this class:
+        #     - With generic types: MultiProcessingDerivedStage(MultiProcessingBaseStage[InputT, OutputT])
+        #     - With concrete types: MultiProcessingDerivedStage(MultiProcessingBaseStage[int, str])
 
-        In this case, typing.Generic stores the stage type in stage.__orig_class__, the concrete types can be accessed
-        as below:
+        # When inheriting with generic types, the derived class can be instantiated like this:
 
-            input_type = typing.get_args(stage.__orig_class__)[0] # int
-            output_type = typing.get_args(stage.__orig_class__)[1] # str
+        #     stage = MultiProcessingDerivedStage[int, str]()
 
-        However, when instantiating a stage which inherits with concrete types:
+        # In this case, typing.Generic stores the stage type in stage.__orig_class__, the concrete types can be accessed
+        # as below:
 
-            stage = MultiProcessingDerivedStage()
+        #     input_type = typing.get_args(stage.__orig_class__)[0] # int
+        #     output_type = typing.get_args(stage.__orig_class__)[1] # str
 
-        The stage instance does not have __orig_class__ attribute (since it is not a generic type). Thus, the concrete
-        types need be retrieved from its base class (which is a generic type):
+        # However, when instantiating a stage which inherits with concrete types:
 
-            input_type = typing.get_args(stage.__orig_bases__[0])[0] # int
-            output_type = typing.get_args(stage.__orig_bases__[0])[1] # str
+        #     stage = MultiProcessingDerivedStage()
 
-        Raises:
-            RuntimeError: if the accepted cannot be deducted from either __orig_class__ or __orig_bases__
+        # The stage instance does not have __orig_class__ attribute (since it is not a generic type). Thus, the concrete
+        # types need be retrieved from its base class (which is a generic type):
+
+        #     input_type = typing.get_args(stage.__orig_bases__[0])[0] # int
+        #     output_type = typing.get_args(stage.__orig_bases__[0])[1] # str
 
-        Returns:
-            typing.Tuple: accepted input types
-        """
         if hasattr(self, "__orig_class__"):
             # inherited with generic types
             input_type = typing.get_args(self.__orig_class__)[0]  # pylint: disable=no-member
@@ -95,14 +121,20 @@ def accepted_types(self) -> typing.Tuple:
 
     def compute_schema(self, schema: StageSchema):
         """
-        See the comment on `accepted_types` for more information on accessing the input and output types.
+        Compute the output schema for the stage.
 
-        Args:
-            schema (StageSchema): StageSchema
+        Parameters
+        ----------
+        schema : StageSchema
+            The schema for the stage.
 
-        Raises:
-            RuntimeError: if the output type cannot be deducted from either __orig_class__ or __orig_bases__
+        Raises
+        ------
+        RuntimeError
+            If the output type cannot be deduced from either __orig_class__ or __orig_bases__.
         """
+
+        # See the comment on `accepted_types` for more information on accessing the input and output types.
         if hasattr(self, "__orig_class__"):
             # inherited with abstract types
             output_type = typing.get_args(self.__orig_class__)[1]  # pylint: disable=no-member
@@ -117,6 +149,7 @@ def compute_schema(self, schema: StageSchema):
         schema.output_schema.set_type(output_type)
 
     def supports_cpp_node(self):
+        """Whether this stage supports a C++ node."""
         return False
 
     @abstractmethod
@@ -162,6 +195,21 @@ def _get_func_signature(func: typing.Callable[[InputT], OutputT]) -> tuple[type,
 
 
 class MultiProcessingStage(MultiProcessingBaseStage[InputT, OutputT]):
+    """
+    A derived class of MultiProcessingBaseStage that allows the user to define a process function that will be executed
+    based on shared process pool.
+
+    Parameters
+    ----------
+    c : `morpheus.config.Config`
+        Pipeline configuration instance.
+    unique_name : str
+        A unique name for the stage.
+    process_fn:  typing.Callable[[InputT], OutputT]
+        The function that will be executed in the process pool.
+    max_in_flight_messages : int, default = None
+        The number of progress engines used by the stage.
+    """
 
     def __init__(self,
                  *,
@@ -178,6 +226,7 @@ def __init__(self,
 
     @property
     def name(self) -> str:
+        """Return the name of the stage."""
         return self._name
 
     def _on_data(self, data: InputT) -> OutputT:
@@ -192,6 +241,25 @@ def create(*,
                unique_name: str,
                process_fn: typing.Callable[[InputT], OutputT],
                process_pool_usage: float):
+        """
+        Create a MultiProcessingStage instance by deducing the input and output types from the process function.
+
+        Parameters
+        ----------
+        c : morpheus.config.Config
+            Pipeline configuration instance.
+        unique_name : str
+            A unique name for the stage.
+        process_fn : typing.Callable[[InputT], OutputT]
+            The function that will be executed in the process pool.
+        process_pool_usage : float
+            The fraction of the process pool workers that this stage could use. Should be between 0 and 1.
+
+        Returns
+        -------
+        MultiProcessingStage[InputT, OutputT]
+            A MultiProcessingStage instance with deduced input and output types.
+        """
 
         input_t, output_t = _get_func_signature(process_fn)
         return MultiProcessingStage[input_t, output_t](c=c,

python/morpheus/morpheus/utils/shared_process_pool.py

@@ -136,7 +136,7 @@ def _initialize(self):
         self._total_max_workers = math.floor(max(1, len(os.sched_getaffinity(0)) * cpu_usage))
         self._processes = []
 
-        self._context = mp.get_context("fork")
+        self._context = mp.get_context("forkserver")
         self._manager = self._context.Manager()
         self._task_queues = self._manager.dict()
         self._stage_semaphores = self._manager.dict()
@@ -196,8 +196,7 @@ def _worker(cancellation_token, task_queues, stage_semaphores):
                     continue
 
                 if task is None:
-                    logger.warning("SharedProcessPool._worker: Worker process %s has received a None task.",
-                                   os.getpid())
+                    logger.debug("SharedProcessPool._worker: Worker process %s has received a None task.", os.getpid())
                     semaphore.release()
                     continue
 
@@ -316,7 +315,7 @@ def start(self):
             If the SharedProcessPool is not shutdown.
         """
         if self._status == PoolStatus.RUNNING:
-            logger.warning("SharedProcessPool.start(): process pool is already running.")
+            logger.debug("SharedProcessPool.start(): process pool is already running.")
             return
 
         process_launcher = threading.Thread(target=self._launch_workers)
@@ -373,7 +372,7 @@ def stop(self):
         Stop receiving any new tasks.
         """
         if self._status not in (PoolStatus.RUNNING, PoolStatus.INITIALIZING):
-            logger.warning("SharedProcessPool.stop(): Cannot stop a SharedProcessPool that is not running.")
+            logger.debug("SharedProcessPool.stop(): Cannot stop a SharedProcessPool that is not running.")
             return
 
         # No new tasks will be accepted from this point
@@ -400,7 +399,7 @@ def join(self, timeout: float | None = None):
 
         if self._status != PoolStatus.STOPPED:
             if self._status == PoolStatus.SHUTDOWN:
-                logging.warning("SharedProcessPool.join(): process pool is already shut down.")
+                logger.debug("SharedProcessPool.join(): process pool is already shut down.")
                 return
 
             raise RuntimeError("Cannot join SharedProcessPool that is not stopped.")

tests/conftest.py

@@ -36,13 +36,15 @@
 from _utils.kafka import kafka_bootstrap_servers_fixture  # noqa: F401 pylint:disable=unused-import
 from _utils.kafka import kafka_consumer_fixture  # noqa: F401 pylint:disable=unused-import
 from _utils.kafka import kafka_topics_fixture  # noqa: F401 pylint:disable=unused-import
+from morpheus.utils.shared_process_pool import SharedProcessPool
 
 # Don't let pylint complain about pytest fixtures
 # pylint: disable=redefined-outer-name,unused-argument
 
 (PYTEST_KAFKA_AVAIL, PYTEST_KAFKA_ERROR) = _init_pytest_kafka()
 if PYTEST_KAFKA_AVAIL:
     # Pull out the fixtures into this namespace
+    # pylint: disable=ungrouped-imports
     from _utils.kafka import _kafka_consumer  # noqa: F401  pylint:disable=unused-import
     from _utils.kafka import kafka_server  # noqa: F401  pylint:disable=unused-import
     from _utils.kafka import zookeeper_proc  # noqa: F401  pylint:disable=unused-import
@@ -1150,3 +1152,24 @@ def mock_subscription_fixture():
     ms = mock.MagicMock()
     ms.is_subscribed.return_value = True
     return ms
+
+
+# ==== SharedProcessPool Fixtures ====
+# Any tests that use the SharedProcessPool should use this fixture
+@pytest.fixture(scope="module")
+def shared_process_pool_setup_and_teardown():
+    # Set lower CPU usage for unit test to avoid slowing down the test
+    os.environ["MORPHEUS_SHARED_PROCESS_POOL_CPU_USAGE"] = "0.1"
+
+    pool = SharedProcessPool()
+
+    # SharedProcessPool might be configured and used in other tests, stop and reset the pool before the test starts
+    pool.stop()
+    pool.join()
+    pool.reset()
+    yield pool
+
+    # Stop the pool after all tests are done
+    pool.stop()
+    pool.join()
+    os.environ.pop("MORPHEUS_SHARED_PROCESS_POOL_CPU_USAGE", None)

tests/test_multi_processing_stage.py

@@ -40,6 +40,11 @@
 from morpheus.stages.preprocess.deserialize_stage import DeserializeStage
 
 
+@pytest.fixture(scope="module", autouse=True)
+def setup_and_teardown(shared_process_pool_setup_and_teardown):  # pylint: disable=unused-argument
+    pass
+
+
 def _create_df(count: int) -> pd.DataFrame:
     return pd.DataFrame({"a": range(count)}, {"b": range(count)})
 

tests/utils/test_shared_process_pool.py

@@ -15,7 +15,6 @@
 
 import logging
 import multiprocessing as mp
-import os
 import threading
 from decimal import Decimal
 from fractions import Fraction
@@ -30,26 +29,12 @@
 # This test has issues with joining processes when testing with pytest `-s` option. Run pytest without `-s` flag
 
 
-@pytest.fixture(scope="session", autouse=True)
-def setup_and_teardown():
-    # Set lower CPU usage for unit test to avoid slowing down the test
-    os.environ["MORPHEUS_SHARED_PROCESS_POOL_CPU_USAGE"] = "0.1"
-
-    pool = SharedProcessPool()
-
-    # Since SharedProcessPool might be used in other tests, stop and reset the pool before the test starts
-    pool.stop()
-    pool.join()
-    pool.reset()
-    yield
-
-    # Stop the pool after all tests are done
-    pool.stop()
-    pool.join()
-    os.environ.pop("MORPHEUS_SHARED_PROCESS_POOL_CPU_USAGE", None)
+@pytest.fixture(scope="module", autouse=True)
+def setup_and_teardown(shared_process_pool_setup_and_teardown):  # pylint: disable=unused-argument
+    pass
 
 
-@pytest.fixture(name="shared_process_pool")
+@pytest.fixture(name="shared_process_pool", scope="function")
 def shared_process_pool_fixture():
 
     pool = SharedProcessPool()