changes to get pre-checks passing

Author: Bryannah Hernandez

src/sagemaker/serve/builder/model_builder.py

@@ -437,7 +437,8 @@ def _prepare_for_mode(
             return None
 
         raise ValueError(
-            "Please specify mode in: %s, %s" % (Mode.LOCAL_CONTAINER, Mode.SAGEMAKER_ENDPOINT, Mode.IN_PROCESS)
+            "Please specify mode in: %s, %s, %s"
+            % (Mode.LOCAL_CONTAINER, Mode.SAGEMAKER_ENDPOINT, Mode.IN_PROCESS)
         )
 
     def _get_client_translators(self):
@@ -616,10 +617,12 @@ def _overwrite_mode_in_deploy(self, overwrite_mode: str):
             s3_upload_path, env_vars_sagemaker = self._prepare_for_mode()
             self.pysdk_model.model_data = s3_upload_path
             self.pysdk_model.env.update(env_vars_sagemaker)
-
         elif overwrite_mode == Mode.LOCAL_CONTAINER:
             self.mode = self.pysdk_model.mode = Mode.LOCAL_CONTAINER
             self._prepare_for_mode()
+        elif overwrite_mode == Mode.IN_PROCESS:
+            self.mode = self.pysdk_model.mode = Mode.IN_PROCESS
+            self._prepare_for_mode()
         else:
             raise ValueError("Mode %s is not supported!" % overwrite_mode)
 

src/sagemaker/serve/builder/transformers_builder.py

@@ -19,8 +19,6 @@
 from pathlib import Path
 from packaging.version import Version
 
-from pathlib import Path
-
 from sagemaker.model import Model
 from sagemaker import image_uris
 from sagemaker.serve.utils.local_hardware import (
@@ -231,7 +229,7 @@ def _transformers_model_builder_deploy_wrapper(self, *args, **kwargs) -> Type[Pr
                 jumpstart=False,
             )
             return predictor
-        
+
         if self.mode == Mode.IN_PROCESS:
             timeout = kwargs.get("model_data_download_timeout")
 
@@ -361,6 +359,7 @@ def _auto_detect_container(self):
             raise ValueError(
                 "Cannot detect and set image_uri. Please pass model or inference spec."
             )
+
     def _build_for_transformers(self):
         """Method that triggers model build
 

src/sagemaker/serve/mode/in_process_mode.py

@@ -4,10 +4,7 @@
 from pathlib import Path
 import logging
 from typing import Dict, Type
-import base64
 import time
-import subprocess
-import docker
 
 from sagemaker.base_predictor import PredictorBase
 from sagemaker.serve.spec.inference_spec import InferenceSpec
@@ -108,4 +105,3 @@ def create_server(
 
         if not healthy:
             raise LocalDeepPingException(_PING_HEALTH_CHECK_FAIL_MSG)
-        

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

@@ -75,6 +75,7 @@ def prepare_mms_js_resources(
 
     return _copy_jumpstart_artifacts(model_data, js_id, code_dir)
 
+
 def prepare_for_mms(
     model_path: str,
     shared_libs: List[str],

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

@@ -32,6 +32,17 @@ def _start_serving(
         env_vars: dict,
     ):
         """Placeholder docstring"""
+        env = {
+            "SAGEMAKER_SUBMIT_DIRECTORY": "/opt/ml/model/code",
+            "SAGEMAKER_PROGRAM": "inference.py",
+            "SAGEMAKER_SERVE_SECRET_KEY": secret_key,
+            "LOCAL_PYTHON": platform.python_version(),
+        }
+        if env_vars:
+            env_vars.update(env)
+        else:
+            env_vars = env
+
         self.container = client.containers.run(
             image,
             "serve",
@@ -44,31 +55,31 @@ def _start_serving(
                     "mode": "rw",
                 },
             },
-            environment={
-                "SAGEMAKER_SUBMIT_DIRECTORY": "/opt/ml/model/code",
-                "SAGEMAKER_PROGRAM": "inference.py",
-                "SAGEMAKER_SERVE_SECRET_KEY": secret_key,
-                "LOCAL_PYTHON": platform.python_version(),
-                **env_vars,
-            },
+            environment=env_vars,
         )
 
     def _invoke_multi_model_server_serving(self, request: object, content_type: str, accept: str):
         """Placeholder docstring"""
+        logger.info(content_type)
+        logger.info(accept)
+
         try:
             response = requests.post(
-                f"http://{0.0.0.0}:8080/invocations",
+                "http://0.0.0.0:8080/invocations",
                 data=request,
                 headers={"Content-Type": content_type, "Accept": accept},
                 timeout=600,
             )
             response.raise_for_status()
+
+            logger.info(response.content)
+
             return response.content
         except Exception as e:
             raise Exception("Unable to send request to the local container server") from e
 
         return (True, response)
-    
+
     def _multi_model_server_deep_ping(self, predictor: PredictorBase):
         """Placeholder docstring"""
         response = None
@@ -84,7 +95,7 @@ def _multi_model_server_deep_ping(self, predictor: PredictorBase):
         #     return False, response
 
         return (True, response)
-    
+
 
 class LocalMultiModelServer:
     """Local Multi Model server instance"""
@@ -202,7 +213,6 @@ def _upload_server_artifacts(
                     "S3Uri": model_data_url + "/",
                 }
             }
-
             if model_data_url
             else None
         )

src/sagemaker/serve/utils/exceptions.py

@@ -22,7 +22,7 @@ class LocalDeepPingException(ModelBuilderException):
 
     def __init__(self, message):
         super().__init__(message=message)
-        
+
 
 class InProcessDeepPingException(ModelBuilderException):
     """Raise when in process model serving does not pass the deep ping check"""

tests/unit/sagemaker/serve/model_server/multi_model_server/test_multi_model_server_prepare.py

@@ -105,4 +105,3 @@ def test_create_dir_structure_invalid_path(self, mock_path):
             _create_dir_structure(mock_model_path)
 
         self.assertEquals("model_dir is not a valid directory", str(context.exception))
-