Add mlflow test to verify mlflow score script

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/context/Dockerfile

@@ -1,7 +1,7 @@
 FROM mcr.microsoft.com/azureml/inference-base-2004:{{latest-image-tag}}
 
 WORKDIR /
-ENV AZUREML_CONDA_ENVIRONMENT_PATH=/azureml-envs/minimal
+ENV AZUREML_CONDA_ENVIRONMENT_PATH=/azureml-envs/mlflow
 ENV AZUREML_CONDA_DEFAULT_ENVIRONMENT=$AZUREML_CONDA_ENVIRONMENT_PATH
 
 # Prepend path to AzureML conda environment
@@ -13,8 +13,6 @@ ENV AML_APP_ROOT="/var/mlflow_resources"
 ENV AZUREML_ENTRY_SCRIPT="mlflow_score_script.py"
 
 USER root
-# Copying of mlmonitoring will add once testing is completed.
-# COPY mlmonitoring /var/mlflow_resources/mlmonitoring
 
 # We'll copy the HF scripts as well to enable better handling for v2 packaging. This will not require changes to the
 # packages installed in the image, as the expectation is that these will all be brought along with the model.

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/context/conda_dependencies.yaml

@@ -6,4 +6,8 @@ dependencies:
 - python=3.9.13
 - pip
 - pip:
-  - azureml-inference-server-http=={{latest-pypi-version}}
+  - azureml-inference-server-http=={{latest-pypi-version}}
+  - azureml-ai-monitoring=={{latest-pypi-version}}
+  - numpy
+  - mlflow
+  - azureml-contrib-services

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/context/mlflow_score_script.py

@@ -17,7 +17,7 @@
 from mlflow.models import Model
 from mlflow.pyfunc import load_model
 from mlflow.pyfunc.scoring_server import _get_jsonable_obj
-from mlmonitoring import Collector
+from azureml.ai.monitoring import Collector
 from mlflow.types.utils import _infer_schema
 from mlflow.types.schema import Schema, ColSpec, DataType
 from mlflow.exceptions import MlflowException

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/minimal_inference_test.py → assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/mlflow_test.py

@@ -16,8 +16,8 @@
 STD_LOG = Path("artifacts/user_logs/std_log.txt")
 
 
-def test_minimal_cpu_inference():
-    """Tests a sample job using minimal 20.04 py39 cpu as the environment."""
+def test_mlflow_cpu_inference():
+    """Tests a sample job using mlflow 20.04 py39 cpu as the environment."""
     this_dir = Path(__file__).parent
 
     subscription_id = os.environ.get("subscription_id")
@@ -28,27 +28,30 @@ def test_minimal_cpu_inference():
         AzureCliCredential(), subscription_id, resource_group, workspace_name
     )
 
-    env_name = "minimal_cpu_inference"
+    env_name = "mlflow_py39_inference"
 
     env_docker_context = Environment(
         build=BuildContext(path=this_dir / BUILD_CONTEXT),
-        name="minimal_cpu_inference",
-        description="minimal 20.04 py39 cpu inference environment created from a Docker context.",
+        name="mlflow_py39_inference",
+        description="mlflow 20.04 py39 cpu inference environment created from a Docker context.",
     )
     ml_client.environments.create_or_update(env_docker_context)
 
     # create the command
     job = command(
         code=this_dir / JOB_SOURCE_CODE,  # local path where the code is stored
-        command="python main.py --score ${{inputs.score}}",
+        command="python main.py --model_dir ${{inputs.model_dir}} "
+        "--score ${{inputs.score}} --score_input ${{inputs.score_input}}",
         inputs=dict(
-            score="valid_score.py",
+            score="/var/mlflow_resources/mlflow_score_script.py",
+            score_input="sample_2_0_input.txt",
+            model_dir="mlflow_2_0_model_folder"
         ),
         environment=f"{env_name}@latest",
         compute=os.environ.get("cpu_cluster"),
-        display_name="minimal-cpu-inference-example",
-        description="A test run of the minimal 20.04 py39 cpu inference curated environment",
-        experiment_name="minimalCPUInferenceExperiment"
+        display_name="mlflow-py39-inference-example",
+        description="A test run of the mlflow 20.04 py39 cpu inference curated environment",
+        experiment_name="mlflow39InferenceExperiment"
     )
 
     returned_job = ml_client.create_or_update(job)

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/main.py

@@ -4,6 +4,7 @@
 """Validate minimal inference cpu environment by running azmlinfsrv."""
 
 # imports
+import json
 import os
 import subprocess
 import requests
@@ -15,21 +16,32 @@
 def main(args):
     """Start inference server and post scoring request."""
     # start the server
-    server_process = start_server("/var/tmp", ["--entry_script", args.score, "--port", "8081"])
+    server_process = start_server("/var/tmp",
+                                  ["--entry_script", args.score, "--port", "8081"],
+                                  args.model_dir)
 
     # score a request
-    req = score_with_post()
+    with open(args.score_input) as f:
+        payload_data = json.load(f)
+
+    headers = {"Content-Type": "application/json"}
+    res = score_with_post(headers=headers, data=payload_data)
     server_process.kill()
 
-    print(req)
+    print_file_contents("/var/tmp", "stderr.txt")
+    print_file_contents("/var/tmp", "stdout.txt")
+    print(res)
 
 
-def start_server(log_directory, args, timeout=timedelta(seconds=15)):
+def start_server(log_directory, args, model_dir, timeout=timedelta(seconds=60)):
     """Start inference server with options."""
     stderr_file = open(os.path.join(log_directory, "stderr.txt"), "w")
     stdout_file = open(os.path.join(log_directory, "stdout.txt"), "w")
 
     env = os.environ.copy()
+    env["AZUREML_MODEL_DIR"] = os.path.dirname(os.path.abspath(__file__))
+    env["MLFLOW_MODEL_FOLDER"] = model_dir
+    print(os.path.abspath(__file__))
     server_process = subprocess.Popen(["azmlinfsrv"] + args, stdout=stdout_file, stderr=stderr_file, env=env)
 
     max_time = datetime.now() + timeout
@@ -50,9 +62,6 @@ def start_server(log_directory, args, timeout=timedelta(seconds=15)):
         if status is not None:
             break
 
-    print(log_directory, "stderr.txt")
-    print(log_directory, "stdout.txt")
-
     return server_process
 
 
@@ -62,13 +71,27 @@ def score_with_post(headers=None, data=None):
     return requests.post(url=url, headers=headers, data=data)
 
 
+def print_file_contents(log_directory, file_name):
+    """Print out file contents."""
+    print(log_directory, file_name)
+    file_path = os.path.join(log_directory, file_name)
+    try:
+        with open(file_path, 'r') as file:
+            contents = file.read()
+            print(contents)
+    except FileNotFoundError:
+        print("file path is not valid.")
+
+
 def parse_args():
     """Parse input arguments."""
     # setup arg parser
     parser = argparse.ArgumentParser()
 
     # add arguments
     parser.add_argument("--score", type=str)
+    parser.add_argument("--model_dir", type=str)
+    parser.add_argument("--score_input", type=str)
 
     # parse args
     args = parser.parse_args()

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/mlflow_2_0_model_folder/MLmodel

@@ -0,0 +1,20 @@
+flavors:
+  python_function:
+    cloudpickle_version: 2.1.0
+    env:
+      conda: conda.yaml
+      virtualenv: python_env.yaml
+    loader_module: mlflow.pyfunc.model
+    python_model: python_model.pkl
+    python_version: 3.9.13
+mlflow_version: 2.0.1
+model_uuid: 687fb8fa7a044a1cb8ee79b5f76368f8
+saved_input_example_info:
+  artifact_path: input_example.json
+  pandas_orient: split
+  type: dataframe
+signature:
+  inputs: '[{"name": "a", "type": "double"}, {"name": "b", "type": "long"}, {"name":
+    "c", "type": "string"}]'
+  outputs: '[{"name": "a", "type": "double"}]'
+utc_time_created: '2022-11-18 22:14:59.029851'

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/mlflow_2_0_model_folder/conda.yaml

@@ -0,0 +1,9 @@
+channels:
+- conda-forge
+- anaconda
+dependencies:
+- python=3.9.13
+- pip
+- pip:
+  - mlflow
+  - cloudpickle==2.1.0

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/mlflow_2_0_model_folder/input_example.json

@@ -0,0 +1 @@
+{"columns": ["a", "b", "c"], "data": [[3.0, 1, "foo"]]}

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/mlflow_2_0_model_folder/python_env.yaml

@@ -0,0 +1,7 @@
+python: 3.9.13
+build_dependencies:
+- pip
+- setuptools==65.2.0
+- wheel==0.37.1
+dependencies:
+- -r requirements.txt

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/mlflow_2_0_model_folder/requirements.txt

@@ -0,0 +1,2 @@
+mlflow
+cloudpickle==2.1.0

assets/inference/environments/mlflow-ubuntu20.04-py39-cpu-inference/tests/src/sample_2_0_input.txt

@@ -0,0 +1 @@
+{"input_data":{"columns":["a", "b", "c"],"index":[0],"data":[[3.0, 1, "foo"]]}}