Iris MLflow Serve Testing Framework — Local Testing Guide

A comprehensive testing suite for validating MLflow model serving endpoints (mlflow serve) in iris classification scenarios. Ensures endpoint reliability and prediction consistency.

This approach is ideal for lightweight applications or for testing your model locally before moving it to a staging or production environment.

Authors

• Mehmet Acikgoz
• Maria Vechtomova

🚀 Quick Start

Prerequisites

• Python ≥3.11
• MLflow model artifact (pyfunc-lg-pipeline-model directory)

Setup & Execution

1. Install dependencies

   uv sync --all-extras
   

2. Launch MLflow Server

   uv run mlflow models serve -m pyfunc-lg-pipeline-model --port 5088
   

3. Run Test Suite (in new terminal)

   uv run pytest tests/
   

🌟 Key Features

Core Validation Suite

Feature	Verification Points
Service Health	✅ 200 status on /health endpoint
Server Responsiveness	✅ Successful /ping responses
Version Integrity	✅ MLflow 2.17.0 confirmation
Prediction Formats	✅ Supports both dataframe_split and dataframe_records

Advanced Testing Capabilities

• Class Validation: Ensures predictions {setosa, versicolor, virginica}
• Null Handling: Robust null value processing
• Batch Testing: CSV dataset validation (test_data/test_set.csv)
• Diagnostics: Detailed logging with loguru

🧪 Test Architecture

Endpoint Specifications

Endpoint	Method	Success Criteria
/health	GET	HTTP 200
/ping	GET	HTTP 200
/version	GET	Returns "2.17.0"
/invocations	POST	Valid prediction JSON

Test Types

1. Single Prediction

   • Validates response structure
   • Tests both DataFrame serialization formats

2. Batch Validation

   • Processes full CSV dataset
   • Verifies class membership for all predictions

📂 Project Structure

.
├── pyfunc-lg-pipeline-model/  # MLflow model artifacts
│   ├── MLmodel                # Model metadata
│   ├── code/                  # Custom code
│   ├── python_model.pkl       # Serialized model
│   └── ...                    # Additional model files
├── tests/
│   ├── test_data/             # Test datasets
│   │   └── test_set.csv       # Validation samples
│   └── test_mlflow_serve.py   # Core test suite
├── pyproject.toml             # Build configuration
└── README.md                  # This documentation

📦 Dependencies

Core Requirements

mlflow==2.17.0
virtualenv>=20.30.0

Testing Suite

loguru>=0.7.3
pandas>=2.2.3
pytest>=8.3.5
requests>=2.32.3

🔍 Troubleshooting

• Port Conflicts: Ensure port 5088 is available
• Model Path: Verify pyfunc-lg-pipeline-model directory exists
• Dependency Issues: Use uv sync --all-extras for clean installs