OpenGPU Network Provider-Suite

Requirements

• Docker version 20.10.0 or higher (including Docker Compose v2 support)
• Linux, macOS, or Windows host with Docker support

This guide explains two alternative ways to install and run the Provider Suite:

1. As a Docker Compose stack
2. As a single privileged container (Docker-in-Docker)

Recommended for:

• Individual providers who want an easier, modular way to manage and monitor each service (recommended for home setups or single-machine deployments).
• Users who want modularity and want to manage each service independently.
• Environments where running multiple containers is possible.

First, download the latest docker-compose.yml directly from the official repository:

curl -o docker-compose.yml https://raw.githubusercontent.com/OpenGPU-Network/provider-suite/refs/heads/main/docker-compose.yml

You must provide at least the following variables:

• PROVIDER_PRIVATE_KEY
• MASTER_ADDRESS

There are two ways to set these variables:

• A. Use a .env file: Place a file named .env in the same directory as your docker-compose.yml. Example:

  PROVIDER_PRIVATE_KEY=your_provider_private_key
  MASTER_ADDRESS=your_master_address

• B. Export environment variables manually:

  export PROVIDER_PRIVATE_KEY=your_provider_private_key
  export MASTER_ADDRESS=your_master_address

docker compose up -d

All services will start. View logs with:

docker compose logs -f

To gracefully stop all running containers in the stack (without removing them):

docker compose stop

To restart stopped services:

docker compose start

To fully restart (stop, remove, and recreate) all services:

docker compose down -v
# then
docker compose up -d

To remove all containers, networks, and volumes:

docker compose down -v

Recommended for:

• Cloud environments or platforms that require running everything inside a single container.
• Scenarios where Docker Compose is not supported.

In this setup, the stack runs inside a privileged container using the docker-dind base image.

Select the image suitable for your hardware:

docker run -d --name opengpu_network__provider_suite__cpu \
  --privileged \
  -e PROVIDER_PRIVATE_KEY='your_provider_private_key' \
  -e MASTER_ADDRESS='your_master_address' \
  opengpunetwork/provider-suite:cpu

docker run -d --name opengpu_network__provider_suite__cuda \
  --privileged --gpus all \
  -e PROVIDER_PRIVATE_KEY='your_provider_private_key' \
  -e MASTER_ADDRESS='your_master_address' \
  opengpunetwork/provider-suite:cuda

docker run -d --name opengpu_network__provider_suite__rocm \
  --privileged --device /dev/kfd --device /dev/dri \
  -e PROVIDER_PRIVATE_KEY='your_provider_private_key' \
  -e MASTER_ADDRESS='your_master_address' \
  opengpunetwork/provider-suite:rocm

Note: The --privileged flag is required for inner Docker engine access. For GPU support, ensure your host drivers and Docker runtime are properly configured.

Both deployment methods support graceful shutdown:

• For Compose:

  • Stop all services (keep containers and volumes):

    docker compose stop

  • Remove all (containers, networks, volumes):

    docker compose down

• For Single Container:

  docker stop <container_name>
  docker rm <container_name>

• Ensure your Docker version supports required features (Compose v2, --gpus all, etc.).
• If you encounter permission issues, check your Docker daemon and user permissions.

Never share your PROVIDER_PRIVATE_KEY or other secrets publicly. Use secure secret management solutions in production.