Azure OpenAI Chat Application

This application is a Flask-based web service that integrates with Azure OpenAI to provide AI consulting services. It offers expert guidance on Azure architecture, cost optimization, security, and best practices.

• Prerequisites
• Project Structure
• Configuration
• Local Development
• Deployment
• CI/CD Pipeline
• Monitoring and Maintenance
• Troubleshooting

• Python 3.10 or higher
• Azure subscription
• Azure OpenAI service instance
• GitHub account
• Azure Web App Service

AzureSmartBot/
├── .github/
│   └── workflows/
│       └── main_webappservice.yml
├── templates/
│   └── index.html
├── .env.example
├── .gitignore
├── app.py
├── requirements.txt
└── README.md

Create a .env file for local development:

AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint
AZURE_OPENAI_API_KEY=your_azure_openai_api_key
AZURE_OPENAI_DEPLOYMENT_NAME=your_deployment_name
PORT=8000

Add these settings in Azure Portal > App Service > Configuration > Application settings:

• AZURE_OPENAI_ENDPOINT
• AZURE_OPENAI_API_KEY
• AZURE_OPENAI_DEPLOYMENT_NAME
• WEBSITES_CONTAINER_START_TIME_LIMIT=600
• PORT=8000

1. Clone the repository:

git clone https://github.com/Curious4Tech/AzureSmartBot.git
cd AzureSmartBot

2. Create and activate virtual environment:

python -m venv venv
source venv/bin/activate  # For Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Set up environment variables:

cp .env.example .env
# Edit .env with your values

5. Run the application:

python app.py

The application will be available at http://localhost:8000

1. Create an Azure Web App Service
2. Configure application settings in Azure Portal
3. Deploy using VS Code Azure Tools or Azure CLI

Set in Azure Portal > Configuration > General settings:

gunicorn --workers=2 --timeout=120 --access-logfile="-" --error-logfile="-" --bind=0.0.0.0:8000 app:app

Use the azure portal to configure it by authenticating to your github account and choosing your repo, then Azure will provide a default workflow that will is enough for this demo but you can also costumize your workflows.

1. In Azure Portal:

   • Go to your Web App
   • Download publish profile
   • Copy the content

2. In GitHub:

   • Go to repository settings
   • Add new secret AZURE_WEBAPP_PUBLISH_PROFILE
   • Paste the publish profile content

3. Update workflow file:

   • Replace your-webapp-name with your actual Azure Web App name

The application includes a health check endpoint at /health that verifies:

• Application status
• Azure OpenAI connection

• Application logs available in Azure Portal
• Structured logging format:

%(asctime)s - %(name)s - %(levelname)s - %(message)s

1. Enable Application Insights in Azure Portal
2. Monitor:
   • Response times
   • Failed requests
   • Server errors
   • Container logs

1. Container Startup Issues:

Solution: Check startup command and environment variables in Azure Portal

2. OpenAI Client Errors:

Solution: Verify Azure OpenAI credentials and endpoint configuration

3. Connection Timeouts:

Solution: Adjust timeout settings in startup command or check network configuration

Main application file containing:

• Flask application setup
• Azure OpenAI client initialization
• API endpoints
• Error handling

# Key components
- Flask application initialization
- Azure OpenAI client setup
- System message configuration
- API endpoints for chat and health check

flask>=2.0.0
python-dotenv>=1.0.0
openai>=1.0.0
gunicorn>=20.1.0
httpx>=0.24.0

1. Environment Variables:

   • Never commit sensitive data
   • Use Azure Key Vault for production secrets

2. API Security:

   • Rate limiting implemented
   • Input validation on all endpoints
   • Error handling to prevent information disclosure

3. Deployment Security:

   • HTTPS enabled
   • Secure configuration in Azure App Service

For issues or questions:

1. Check Azure App Service logs
2. Review GitHub Actions workflow runs
3. Contact system administrator