Skip to content

Docker.md

Latest commit

 

History

History
446 lines (353 loc) · 11 KB

Docker.md

File metadata and controls

446 lines (353 loc) · 11 KB

🚀 Succinct Prover Node - Docker Installation & Usage Guide

  • This script is built on a version previously written by Yinger it works the same way 0xCRASHOUT - Guide
  • In this doc, I’ll focus more on detailed issues, execution commands, and related tools so you can better manage your Provers.
  • Big thanks to Yinger 🐐

📋 Table of Contents

🔧 Prerequisites

Financial Requirements

  • 1,000 testPROVE tokens - Required for staking
  • Sepolia ETH - For transaction fees (~0.1 ETH recommended)
  • Fresh wallet address - Dedicated wallet for security

Account Setup

  1. Create Prover Account: Visit Succinct Staking Portal
  2. Generate Prover Address: Complete prover creation to receive your EVM address
  3. Stake Tokens: Stake your 1,000 testPROVE at Staking Interface

💻 System Requirements

Hardware Requirements

Component Minimum Recommended
GPU NVIDIA RTX 3090 RTX 4090 or better
RAM 16GB 16GB+
Storage 100GB SSD 100GB NVMe SSD
CPU 8 cores 16+ cores
Network 100 Mbps 1 Gbps+ or 10 Gbps+ is better

Software Requirements

Software Version Notes
Ubuntu 20.04+ LTS recommended
NVIDIA Driver 555+ For CUDA 12.5+ support
Docker 20.10+ Latest stable version
NVIDIA Container Toolkit 1.17.8+ GPU container support

⚠️ Important Notes

  • Virtual Machine: Must be Ubuntu-based
  • Cloud Platforms: Avoid Pytorch/NVIDIA CUDA templates (Docker incompatible)
  • GPU Access: Direct GPU passthrough required

🚀 Quick Start

One-Command Setup (Recommended)

wget https://raw.githubusercontent.com/cerberus-node/succinct/main/auto-docker.sh && chmod +x auto-docker.sh && sudo ./auto-docker.sh

📖 Detailed Setup Process

Step 1: Download & Prepare Script

# Download the setup script from GitHub
wget https://raw.githubusercontent.com/cerberus-node/succinct/main/auto-docker.sh

# Make it executable
chmod +x auto-docker.sh

# Verify script integrity (optional)
sha256sum auto-docker.sh

Step 2: Run Installation Script

# Execute with root privileges
sudo ./auto-docker.sh

Step 3: Installation Flow

The script will perform these operations automatically:

  1. 🔍 System Validation

    • Check root privileges
    • Validate OS compatibility
    • Test internet connectivity
    • Verify disk space (minimum 10GB)
    • Check virtualization support

  2. 📦 Package Updates

    • Update system packages
    • Install essential dependencies
    • Configure non-interactive mode

  3. 🐳 Docker Installation

    • Add Docker official repository
    • Install Docker CE components
    • Configure Docker service
    • Add user to docker group

  4. 🎮 NVIDIA Driver Setup

    • Detect current driver version
    • Install/update to version 555+
    • Configure CUDA support
    • Handle reboot if required

  5. 🔧 NVIDIA Container Toolkit

    • Install specific version (1.17.8-1)
    • Configure Docker runtime
    • Test GPU container access

  6. 📥 Docker Image Pull

    • Download latest Succinct prover image
    • Verify image integrity

Step 4: Configuration Input

When prompted, provide:

Prover Address

Enter your Prover Address (0x...): 0x000000000000000000000000000000000000000
  • Format: 0x followed by 40 hexadecimal characters
  • Source: "My Prover" page on staking portal
  • Verification: Must match your created prover

Private Key

Enter your Private Key (without 0x prefix): abc123...
  • Format: 64 hexadecimal characters
  • Security: Use dedicated wallet only
  • Verification: Wallet must have staked 1,000 testPROVE

Step 5: Prover Launch

The script will automatically launch your prover with optimal settings.

🔧 Manual Docker Configuration

If you prefer manual setup or need custom configuration:

Environment Variables

# Core configuration
export PROVE_PER_BPGU=1.01
export PGUS_PER_SECOND=10485606
export PROVER_ADDRESS="0x000000000000000000000000000000000000000"
export PRIVATE_KEY="abc123def456..."

# Network configuration
export SUCCINCT_RPC_URL="https://rpc.sepolia.succinct.xyz"
export NETWORK_PRIVATE_KEY="$PRIVATE_KEY"

Docker Command

docker run \
    --gpus all \
    --network host \
    --restart unless-stopped \
    --name succinct-prover-$(date +%s) \
    -e NETWORK_PRIVATE_KEY="$PRIVATE_KEY" \
    -v /var/run/docker.sock:/var/run/docker.sock \
    public.ecr.aws/succinct-labs/spn-node:latest-gpu \
    prove \
    --rpc-url "$SUCCINCT_RPC_URL" \
    --throughput "$PGUS_PER_SECOND" \
    --bid "$PROVE_PER_BPGU" \
    --private-key "$PRIVATE_KEY" \
    --prover "$PROVER_ADDRESS"

Parameter Explanation

Parameter Description Default Value
--gpus all Enable all GPU access Required
--network host Use host networking Required
--restart unless-stopped Auto-restart policy Recommended
--throughput Processing speed (PGUs/sec) 10485606
--bid Bid price per BPGU 1.01
--rpc-url Succinct RPC endpoint Fixed

🔧 Advanced Configuration

Performance Optimization

# High-performance settings
export PGUS_PER_SECOND=15000000  # Increase for powerful GPUs
export PROVE_PER_BPGU=1.05       # Competitive bidding

# Resource limits
docker run \
    --gpus all \
    --memory="32g" \
    --cpus="8" \
    --network host \
    --restart unless-stopped \
    # ... rest of configuration

Custom Calibration

# Run calibration to optimize settings
docker run \
    --gpus all \
    --network host \
    -e NETWORK_PRIVATE_KEY="$PRIVATE_KEY" \
    -v /var/run/docker.sock:/var/run/docker.sock \
    public.ecr.aws/succinct-labs/spn-node:latest-gpu \
    calibrate \
    --usd-cost-per-hour 0.80 \
    --utilization-rate 0.5 \
    --profit-margin 0.1 \
    --prove-price 1.00

Multi-GPU Setup

# Specify specific GPUs
docker run --gpus '"device=0,1"' \
    # ... rest of configuration

# GPU resource allocation
docker run --gpus all \
    --device-cgroup-rule='c 195:* rmw' \
    # ... rest of configuration

🐛 Troubleshooting

Common Issues & Solutions

1. Script Permission Denied

# Problem: Permission denied when running script
# Solution:
chmod +x Auto-docker.sh
sudo ./Auto-docker.sh

2. NVIDIA Driver Issues

# Problem: Driver version too old
# Check current version:
nvidia-smi

# Manual driver update:
sudo apt update
sudo apt install -y nvidia-driver-555
sudo reboot

3. Docker Permission Issues

# Problem: Docker daemon not accessible
# Solution:
sudo usermod -aG docker $USER
newgrp docker
# or logout/login

4. GPU Not Detected in Container

# Test GPU access:
docker run --gpus all nvidia/cuda:12.0-base-ubuntu20.04 nvidia-smi

# If fails, reinstall NVIDIA Container Toolkit:
sudo apt-get purge nvidia-container-toolkit
sudo apt-get install nvidia-container-toolkit
sudo systemctl restart docker

5. Network Connectivity Issues

# Test RPC connectivity:
curl -X POST https://rpc.sepolia.succinct.xyz \
    -H "Content-Type: application/json" \
    -d '{"jsonrpc":"2.0","method":"eth_blockNumber","params":[],"id":1}'

# Check firewall:
sudo ufw status
sudo ufw allow out 443

Log Analysis

# View setup logs:
tail -f /var/log/succinct-prover-setup.log

# View container logs:
docker logs succinct-prover-$(date +%s)

# System resource monitoring:
htop
nvidia-smi -l 1

📊 Monitoring & Maintenance

Health Checks

# Check container status
docker ps -a | grep succinct

# Monitor GPU usage
watch -n 1 nvidia-smi

# Check system resources
htop
df -h

Performance Monitoring

# Container stats
docker stats

# Network monitoring
netstat -tulpn | grep docker

# Disk usage
du -sh /var/lib/docker/

Backup & Recovery

# Backup configuration
cp Auto-docker.sh /backup/
env | grep -E "(PROVER_ADDRESS|PROVE_PER_BPGU)" > /backup/prover-config.env

# Container recovery
docker restart succinct-prover-container-name

Updates

# Update Docker image
docker pull public.ecr.aws/succinct-labs/spn-node:latest-gpu

# Restart with new image
docker stop succinct-prover-container-name
docker rm succinct-prover-container-name
# Re-run docker command with new image

🔐 Security Best Practices

Wallet Security

  • ✅ Use dedicated wallet for prover operations only
  • ✅ Store private keys securely (hardware wallet recommended)
  • ✅ Regular key rotation for long-term operations
  • ❌ Never share private keys or commit to version control

System Security

# Enable firewall
sudo ufw enable
sudo ufw allow ssh
sudo ufw allow out 443
sudo ufw allow out 80

# Secure Docker daemon
sudo systemctl enable docker
sudo chmod 660 /var/run/docker.sock

# Regular updates
sudo apt update && sudo apt upgrade -y

Network Security

  • Use VPN when possible
  • Monitor network traffic for anomalies
  • Regular security audits of container configurations

Access Control

# Limit sudo access
sudo visudo
# Add: username ALL=(ALL) NOPASSWD: /path/to/Auto-docker.sh

# Secure log files
sudo chmod 640 /var/log/succinct-prover-setup.log

📚 Additional Resources

Official Documentation

Community Support

Useful Commands

# Quick status check
docker ps && nvidia-smi && df -h

# Emergency stop
docker stop $(docker ps -q --filter ancestor=public.ecr.aws/succinct-labs/spn-node:latest-gpu)

# Clean restart
./Auto-docker.sh

⚡ Quick Reference Card

Task Command
Start Setup sudo ./Auto-docker.sh
Check Status docker ps
View Logs docker logs container-name
GPU Status nvidia-smi
Stop Prover docker stop container-name
Restart Prover docker restart container-name
Update Image docker pull public.ecr.aws/succinct-labs/spn-node:latest-gpu

💡 Pro Tip: Always test your setup on a development environment before deploying to production. Keep your staking wallet secure and monitor your prover's performance regularly.