Skip to main content

Deployment

Ragpi's Docker-based deployment method provides flexibility in where you can host the application. This guide walks through deploying Ragpi using Docker Compose, which can be used across various hosting environments including:

  • Local development environments
  • Virtual Private Servers (VPS) from providers like DigitalOcean, Linode, or Vultr
  • Cloud compute instances from AWS EC2, Google Compute Engine, or Azure Virtual Machines
  • Self-hosted servers or bare metal installations

The deployment process remains consistent across these environments, making it easy to migrate between platforms as your needs evolve. The only requirements are Docker, Docker Compose, and Git access on your chosen platform.

For production deployments, the bare minimum server requirements are:

  • 1 CPU core
  • 2GB RAM
  • 10GB disk space

These specifications provide enough resources to run all core services comfortably while leaving room for document processing and API operations. Adjust these specifications based on your expected workload and the size of your document collection.

Docker Compose Deployment​

Prerequisites​

  • A server or VPS with at least 1 CPU core, 2GB RAM, and 10GB disk space
  • Docker and Docker Compose installed on the server
  • Git installed on the server

Deployment Steps​

  1. Clone the repository:
git clone https://github.com/ragpi/ragpi.git
cd ragpi
  1. Set up environment variables:
cp .env.example .env

  1. Configure the .env file with your settings. Refer to the Configuration page for details on each environment variable.

  2. Start the core services using Docker Compose:

docker compose -f docker-compose.prod.yml up -d

Services Started​

The following services will be started:

Core Services:

  • Redis (port 6379)
  • PostgreSQL (port 5432)
  • API (port 8000)
  • Task Worker (Celery worker, no exposed ports)
important

By default, all services are configured to only listen on localhost (127.0.0.1) for security. To expose the API service to external traffic, you can either:

  • Add the DOCKER_API_PORT_MAPPING environment variable to the .env file with the value 0.0.0.0:8000:8000 to expose the API service on port 8000. (Not recommended for production)
  • Use a reverse proxy like Nginx, a tunneling service like Cloudflare Tunnel, or a similar method to securely expose the API service. (Recommended for production)

Discord Integration (Optional)​

To deploy with Discord integration, follow the main deployment steps 1-3 above, with these additional configurations:

  1. When configuring your .env file, include these additional Discord-specific variables:

    • DISCORD_TOKEN - Discord bot token for authenticating with Discord (Required)
    • DISCORD_CHANNEL_IDS - Comma-separated list of Discord channel IDs where the bot will listen to (Required)
    • RAGPI_API_KEY - API key to authenticate with to the Ragpi API. Set this to one of the values defined in the API_KEYS environment variable (Required only if API authentication is enabled)

    See Discord Configuration for a complete list of available options.

    Note: The RAGPI_BASE_URL environment variable is already set to the API service URL by default.

  2. For step 4, instead of the standard startup command, use the discord profile:

docker compose -f docker-compose.prod.yml --profile discord up -d

The Discord bot service will connect to your API service automatically using the configuration specified in the Docker Compose file.

Stopping Services​

To stop the core services, run:

docker compose -f docker-compose.prod.yml down

To stop the services with Discord integration, run:

docker compose -f docker-compose.prod.yml --profile discord down

Custom Deployment​

Ragpi supports deployment on Kubernetes or other container orchestration platforms using the ragpi/ragpi and ragpi/ragpi-discord Docker images. Configure components independently based on your infrastructure needs. The API image contains both the API service and worker components, which can be controlled via environment variables.

For reference on how to configure the services, see the service definitions in docker-compose.prod.yml in the repository root.