Skip to content

README_Docker.md

Latest commit

 

History

History
210 lines (147 loc) · 4.47 KB

README_Docker.md

File metadata and controls

210 lines (147 loc) · 4.47 KB

Docker Setup for RAG QA System

This guide explains how to dockerize and run the RAG QA System using Docker.

Prerequisites

  • Docker installed on your system
  • Docker Compose (optional, for easier management)
  • The T5 model downloaded to /home/fg12/.cache/huggingface/hub/models--t5-small

Quick Start

1. Build and Run with Script

# Make the build script executable (if not already done)
chmod +x build_docker.sh

# Run the build script
./build_docker.sh

This script will:

  • Copy the T5 model from your local cache to the project
  • Build the Docker image
  • Provide instructions for running the container

2. Run the Container

After building, you can run the container in several ways:

Using Docker directly:

docker run -p 8501:8501 rag-qa-system

Using Docker Compose:

docker-compose up

Using Docker Compose in detached mode:

docker-compose up -d

3. Access the Application

Once running, the Streamlit application will be available at:

Manual Build Process

If you prefer to build manually:

1. Copy the Model

# Create models directory
mkdir -p models

# Copy the T5 model
cp -r /home/fg12/.cache/huggingface/hub/models--t5-small models/

2. Build the Image

docker build -t rag-qa-system .

3. Run the Container

docker run -p 8501:8501 rag-qa-system

Docker Compose Features

The docker-compose.yml file includes:

  • Port mapping: Maps container port 8501 to host port 8501
  • Volume mounting: Mounts sample_data.json for easy updates
  • Health checks: Monitors application health
  • Restart policy: Automatically restarts on failure

Configuration

Environment Variables

You can customize the container behavior with environment variables:

docker run -p 8501:8501 \
  -e PYTHONUNBUFFERED=1 \
  -e STREAMLIT_SERVER_PORT=8501 \
  rag-qa-system

Volume Mounts

For development, you can mount additional volumes:

docker run -p 8501:8501 \
  -v $(pwd)/sample_data.json:/app/sample_data.json:ro \
  -v $(pwd)/custom_data:/app/custom_data:ro \
  rag-qa-system

Troubleshooting

Model Loading Issues

If the model fails to load:

  1. Check model path: Ensure the model is copied to models/ directory
  2. Verify model structure: The model should be in models/models--t5-small/snapshots/...
  3. Check permissions: Ensure the model files are readable

Port Conflicts

If port 8501 is already in use:

# Use a different port
docker run -p 8502:8501 rag-qa-system

Memory Issues

For large models, you might need to increase Docker memory limits:

docker run -p 8501:8501 \
  --memory=4g \
  --memory-swap=4g \
  rag-qa-system

GPU Support

To enable GPU support (if available):

docker run -p 8501:8501 \
  --gpus all \
  rag-qa-system

Development

Rebuilding the Image

After making changes to the code:

# Rebuild the image
docker build -t rag-qa-system .

# Or force rebuild without cache
docker build --no-cache -t rag-qa-system .

Viewing Logs

# View container logs
docker logs <container_id>

# Follow logs in real-time
docker logs -f <container_id>

Accessing the Container

# Execute commands inside the container
docker exec -it <container_id> /bin/bash

# Run Python commands
docker exec -it <container_id> uv run python -c "print('Hello from container')"

Production Deployment

For production deployment:

  1. Use a production base image: Consider using python:3.11-slim or python:3.11-alpine
  2. Add security: Run as non-root user (already implemented)
  3. Optimize size: Use multi-stage builds to reduce image size
  4. Add monitoring: Include health checks and logging
  5. Use secrets: Store sensitive data in Docker secrets or environment variables

File Structure

rag_project/
├── Dockerfile              # Docker configuration
├── docker-compose.yml      # Docker Compose configuration
├── .dockerignore          # Files to exclude from build
├── build_docker.sh        # Build script
├── models/                # Local model storage
│   └── models--t5-small/  # T5 model files
├── rag_project/           # Application code
├── streamlit_app.py       # Streamlit application
├── sample_data.json       # Sample Q&A data
└── pyproject.toml         # Project dependencies