Skip to content

Configuration

Learn how to configure and publish sources on the OpenGPU network. This guide covers the configuration parameters and publishing process that apply to both pre-built environments and custom sources.

📋 Prerequisites

Before configuring any source, ensure you have:

  • ✅ OpenGPU SDK installed (Installation Guide)
  • ✅ Wallet configured with private key in environment variables
  • ✅ Chain configuration set up (testnet or mainnet)
  • ✅ Sufficient OGPU for gas fees in your wallet

Environment Setup

# Set your private key
export CLIENT_PRIVATE_KEY="your_private_key_here"

# Verify installation
python -c "import ogpu.client; print('OpenGPU SDK ready!')"

🔧 SourceInfo Configuration

Both pre-built and custom sources use the same SourceInfo structure:

import ogpu.client
from web3 import Web3

source_info = ogpu.client.SourceInfo(
    name="my-service",                              # Human-readable name
    description="Service description",              # Detailed description
    logoUrl="https://example.com/logo.png",        # Optional logo URL
    imageEnvs=ogpu.client.ImageEnvironments(       # Container environments
        cpu="https://public-url/docker-compose-cpu.yml",
        nvidia="https://public-url/docker-compose-gpu.yml"
    ),
    minPayment=Web3.to_wei(0.01, "ether"),         # Minimum payment
    maxExpiryDuration=3600,                        # Maximum execution time (seconds)
    deliveryMethod=ogpu.client.DeliveryMethod.FIRST_RESPONSE,  # Result delivery method
    minAvailableLockup=Web3.to_wei(0, "ether")    # Optional: required provider stake
)

📊 Configuration Parameters

Parameter Type Description Example
name str Human-readable source name "sentiment-analyzer"
description str Detailed source description "AI sentiment analysis service"
logoUrl str URL to source logo/icon "https://example.com/logo.png"
imageEnvs ImageEnvironments Public URLs to Docker Compose files See examples below
minPayment int Minimum payment in wei Web3.to_wei(0.01, "ether")
maxExpiryDuration int Maximum execution time in seconds 3600 (1 hour)
deliveryMethod DeliveryMethod How results are delivered FIRST_RESPONSE or MANUAL_CONFIRMATION
minAvailableLockup int Required provider stake in wei (optional) Web3.to_wei(0.1, "ether")

🖥️ Image Environments

The ImageEnvironments specifies public URLs to Docker Compose files for different hardware:

Single Environment (CPU Only)

imageEnvs = ogpu.client.ImageEnvironments(
    cpu="https://cipfs.ogpuscan.io/ipfs/QmYourCPUCompose"
)

Multi-Environment with GPU Support

imageEnvs = ogpu.client.ImageEnvironments(
    cpu="https://cipfs.ogpuscan.io/ipfs/QmYourCPUCompose",
    nvidia="https://cipfs.ogpuscan.io/ipfs/QmYourGPUCompose",
    amd="https://cipfs.ogpuscan.io/ipfs/QmYourAMDCompose"
)

Public URLs Required

Each URL must point to a publicly accessible Docker Compose file that defines: - The container image to use - Environment variables and configuration
- Port mappings (must expose port 5555) - Volume mounts if needed

Environment Selection Guidelines

  • CPU: Always include for maximum compatibility
  • NVIDIA: Include for CUDA-accelerated workloads
  • AMD: Include for ROCm-accelerated workloads

💰 Payment Configuration

Configure economic parameters for your source:

Basic Payment Setup

# Basic payment setup
minPayment = Web3.to_wei(0.01, "ether")      # 0.01 OGPU minimum

# Alternative units
minPayment = Web3.to_wei(10, "gwei")         # 10 Gwei  
minPayment = 1000000000000000000              # 1 OGPU in wei (direct)

Provider Stake Requirements

# Require provider stake (optional)
minAvailableLockup = Web3.to_wei(0.1, "ether")  # Providers must stake 0.1 OGPU

📡 Delivery Methods

Choose how task results are handled:

Automatic Acceptance

# First valid response is accepted automatically
deliveryMethod = ogpu.client.DeliveryMethod.FIRST_RESPONSE

Use when: - Results are deterministic - You trust the provider network - You want faster task completion

Manual Confirmation

# Client must approve results manually
deliveryMethod = ogpu.client.DeliveryMethod.MANUAL_CONFIRMATION

Use when: - Results need human review - Quality control is important - Tasks are high-value or critical

🚀 Publishing Your Source

Configure and publish your source to the OpenGPU network:

Chain Configuration

# Configure chain (testnet for development)
ogpu.client.ChainConfig.set_chain(ogpu.client.ChainId.OGPU_TESTNET)

# For mainnet deployment
ogpu.client.ChainConfig.set_chain(ogpu.client.ChainId.OGPU_MAINNET)

Publication Process

try:
    source_address = ogpu.client.publish_source(source_info)
    print(f"✅ Source published successfully!")
    print(f"Source address: {source_address}")
except Exception as e:
    print(f"❌ Failed to publish source: {e}")

Private Key Configuration

The SDK automatically uses your private key from environment variables:

export CLIENT_PRIVATE_KEY="your_private_key_here"

Or pass it explicitly:

source_address = ogpu.client.publish_source(
    source_info=source_info,
    private_key="your_private_key_here"
)

🔧 Validation Checklist

  • ✅ Docker Compose files are publicly accessible via HTTPS URLs
  • ✅ Private key is configured correctly
  • ✅ Wallet has sufficient $OGPU for gas
  • ✅ Chain configuration matches your target network
  • ✅ Docker Compose files are valid YAML and contain required services
  • ✅ Payment amounts are competitive with network rates
  • ✅ Timeout durations are reasonable for your use case

🎯 Configuration Examples

Quick Start Configuration

# Simple configuration for testing
simple_source = ogpu.client.SourceInfo(
    name="test-service",
    description="Simple test service",
    imageEnvs=ogpu.client.ImageEnvironments(
        cpu="https://cipfs.ogpuscan.io/ipfs/QmTestCompose"
    ),
    minPayment=Web3.to_wei(0.001, "ether"),
    maxExpiryDuration=300,  # 5 minutes
    deliveryMethod=ogpu.client.DeliveryMethod.FIRST_RESPONSE
)

Production Configuration

# Production-ready configuration
production_source = ogpu.client.SourceInfo(
    name="ai-image-processor",
    description="High-performance AI image processing service",
    logoUrl="https://mycompany.com/logo.png",
    imageEnvs=ogpu.client.ImageEnvironments(
        cpu="https://cipfs.ogpuscan.io/ipfs/QmProdCPUCompose",
        nvidia="https://cipfs.ogpuscan.io/ipfs/QmProdGPUCompose"
    ),
    minPayment=Web3.to_wei(0.02, "ether"),
    minAvailableLockup=Web3.to_wei(0.1, "ether"),
    maxExpiryDuration=1800,  # 30 minutes
    deliveryMethod=ogpu.client.DeliveryMethod.MANUAL_CONFIRMATION
)

🎯 Next Steps

After Configuration:

Advanced Topics:

🆘 Need Help?

Happy configuring! 🚀