suggestions and test added

Author: Bryannah Hernandez

src/sagemaker/serve/builder/model_builder.py

@@ -915,8 +915,12 @@ def _build_validations(self):
         """Validations needed for model server overrides, or auto-detection or fallback"""
         if self.mode == Mode.IN_PROCESS and self.model_server is not ModelServer.MMS:
             raise ValueError(
-                "IN_PROCESS mode is not supported yet for model server. It is "
-                "supported for MMS/Transformers server in beta release."
+                "IN_PROCESS mode is only supported for MMS/Transformers server in beta release."
+            )
+        
+        if self.mode == Mode.IN_PROCESS and self.model_server == ModelServer.MMS:
+            raise Exception(
+                "IN_PROCESS mode is supported for MMS/Transformers server in beta release."
             )
 
         if self.inference_spec and self.model:

src/sagemaker/serve/builder/transformers_builder.py

@@ -47,6 +47,7 @@
 
 logger = logging.getLogger(__name__)
 DEFAULT_TIMEOUT = 1800
+LOCAL_MODES = [Mode.LOCAL_CONTAINER, Mode.IN_PROCESS]
 
 
 """Retrieves images for different libraries - Pytorch, TensorFlow from HuggingFace hub
@@ -164,7 +165,7 @@ def _get_hf_metadata_create_model(self) -> Type[Model]:
                 vpc_config=self.vpc_config,
             )
 
-        if self.mode == Mode.LOCAL_CONTAINER or self.mode == Mode.IN_PROCESS:
+        if self.mode in LOCAL_MODES:
             self.image_uri = pysdk_model.serving_image_uri(
                 self.sagemaker_session.boto_region_name, "local"
             )
@@ -293,7 +294,7 @@ def _build_transformers_env(self):
 
         self.pysdk_model = self._create_transformers_model()
 
-        if self.mode == Mode.LOCAL_CONTAINER or self.mode == Mode.IN_PROCESS:
+        if self.mode in LOCAL_MODES:
             self._prepare_for_mode()
 
         return self.pysdk_model

src/sagemaker/serve/mode/in_process_mode.py

@@ -16,10 +16,8 @@
 
 logger = logging.getLogger(__name__)
 
-_PING_HEALTH_CHECK_INTERVAL_SEC = 5
-
 _PING_HEALTH_CHECK_FAIL_MSG = (
-    "Container did not pass the ping health check. "
+    "Ping health check did not pass. "
     + "Please increase container_timeout_seconds or review your inference code."
 )
 
@@ -55,7 +53,7 @@ def __init__(
         self._invoke_serving = None
 
     def load(self, model_path: str = None):
-        """Placeholder docstring"""
+        """Loads model path, checks that path exists"""
         path = Path(model_path if model_path else self.model_path)
         if not path.exists():
             raise Exception("model_path does not exist")
@@ -65,7 +63,7 @@ def load(self, model_path: str = None):
         return self.inference_spec.load(str(path))
 
     def prepare(self):
-        """Placeholder docstring"""
+        """Prepares the server"""
 
     def create_server(
         self,
@@ -75,9 +73,7 @@ def create_server(
         env_vars: Dict[str, str] = None,
         model_path: str = None,
     ):
-        """Placeholder docstring"""
-
-        # self._pull_image(image=image)
+        """Creating the server and checking ping health."""
 
         # self.destroy_server()
 

src/sagemaker/serve/model_server/multi_model_server/server.py

@@ -31,7 +31,7 @@ def _start_serving(
         secret_key: str,
         env_vars: dict,
     ):
-        """Placeholder docstring"""
+        """Initializes the start of the server"""
         env = {
             "SAGEMAKER_SUBMIT_DIRECTORY": "/opt/ml/model/code",
             "SAGEMAKER_PROGRAM": "inference.py",
@@ -59,7 +59,7 @@ def _start_serving(
         )
 
     def _invoke_multi_model_server_serving(self, request: object, content_type: str, accept: str):
-        """Placeholder docstring"""
+        """Invokes the MMS server by sending POST request"""
         logger.info(content_type)
         logger.info(accept)
 
@@ -81,9 +81,8 @@ def _invoke_multi_model_server_serving(self, request: object, content_type: str,
         return (True, response)
 
     def _multi_model_server_deep_ping(self, predictor: PredictorBase):
-        """Placeholder docstring"""
+        """Sends a deep ping to ensure prediction"""
         response = None
-        logger.debug("AM I HERE? PING PING")
 
         # try:
         #     response = predictor.predict(self.schema_builder.sample_input)
@@ -108,7 +107,7 @@ def _start_serving(
         secret_key: str,
         env_vars: dict,
     ):
-        """Placeholder docstring"""
+        """Initializes the start of the server"""
         env = {
             "SAGEMAKER_SUBMIT_DIRECTORY": "/opt/ml/model/code",
             "SAGEMAKER_PROGRAM": "inference.py",
@@ -136,7 +135,7 @@ def _start_serving(
         )
 
     def _invoke_multi_model_server_serving(self, request: object, content_type: str, accept: str):
-        """Placeholder docstring"""
+        """Invokes MMS server by hitting the docker host"""
         try:
             response = requests.post(
                 f"http://{get_docker_host()}:8080/invocations",
@@ -150,7 +149,7 @@ def _invoke_multi_model_server_serving(self, request: object, content_type: str,
             raise Exception("Unable to send request to the local container server") from e
 
     def _multi_model_server_deep_ping(self, predictor: PredictorBase):
-        """Placeholder docstring"""
+        """Deep ping in order to ensure prediction"""
         response = None
         try:
             response = predictor.predict(self.schema_builder.sample_input)

tests/unit/sagemaker/serve/builder/test_model_builder.py

@@ -77,8 +77,19 @@ def test_validation_in_progress_mode_not_supported(self, mock_serveSettings):
         )
         self.assertRaisesRegex(
             Exception,
-            "IN_PROCESS mode is not supported yet for model server. It is "
-            "supported for MMS/Transformers server in beta release.",
+            "IN_PROCESS mode is only supported for MMS/Transformers server in beta release.",
+            builder.build,
+            Mode.IN_PROCESS,
+            mock_role_arn,
+            mock_session,
+        )
+
+    patch("sagemaker.serve.builder.model_builder._ServeSettings")
+    def test_validation_in_progress_mode_supported(self, mock_serveSettings):
+        builder = ModelBuilder(model_server=ModelServer.MMS)
+        self.assertRaisesRegex(
+            Exception,
+            "IN_PROCESS mode is supported for MMS/Transformers server in beta release.",
             builder.build,
             Mode.IN_PROCESS,
             mock_role_arn,