System Requirements and Environment Setup

Hardware Recommendations

• Minimum: 4-core CPU with 8GB RAM
• Recommended for high concurrency: 16GB RAM
• Disk space: ≥20GB for dependencies and model files

System Initialization

# Update system packages
sudo apt update && sudo apt upgrade -y
# Install development tools
sudo apt install -y build-essential curl git
# Configure firewall (open essential ports only)
sudo ufw allow 22/tcp   # SSH
sudo ufw allow 3000/tcp # Default web port
sudo ufw enable

User Management

# Create dedicated service user
sudo adduser ai-service
sudo usermod -aG sudo ai-service
su - ai-service

Node.js Environment Configuration

Version Management with NVM

# Install NVM
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash
source ~/.bashrc
# Verify installation
command -v nvm
# Install LTS version
nvm install 20
nvm alias default 20
node -v  # Should show v20.x.x

Version Management Best Practices

• Production: Use even-numbered LTS versions (e.g., v20.x)
• Development: Track latest versions
• Project-specific versions: Use .nvmrc files

AI Inference Platform Installation

Global Installation

# Install latest version
npm install -g openclaw@latest --unsafe-perm
# Verify installation
openclaw --version

Troubleshooting Common Issues

Directory Conflicts:

# Clean residual files
rm -rf ~/.npm-global/lib/node_modules/openclaw
npm cache clean --force
# Reinstall
npm install -g openclaw@latest

Permission Errors:

1. Fix ownership:

   sudo chown -R $USER ~/.npm

2. Or configure non-root npm:

   mkdir ~/.npm-global
   npm config set prefix '~/.npm-global'
   echo 'export PATH=~/.npm-global/bin:$PATH' >> ~/.bashrc
   source ~/.bashrc

Large Model Integration

Model Service Configuration

1. Obtain API key from model provider
2. Verify service access permissions
3. Record service endpoint URL

Interactive Configuration

openclaw config

Follow prompts to configure:

1. Provider type: Select “Custom Provider”
2. Connection parameters:
   • Name: Custom identifier (e.g., zhipu-glm)
   • Endpoint: https://api.example.com/v4 (example)
   • API Key: Provider-issued authentication key

Configuration Validation

Check ~/.openclaw/config.json for:

{
  "providers": {
    "zhipu-glm": {
      "type": "custom",
      "endpoint": "https://api.example.com/v4",
      "apiKey": "sk-xxxxxxxxxxxxxxxx"
    }
  }
}

Service Deployment and Security

Development Server

openclaw serve --port 3000

Verify with:

curl http://localhost:3000/health
# Should return {"status":"ok"}

Production Deployment

1. Process management with PM2:

   npm install -g pm2
   pm2 start `which openclaw` --name ai-service -- serve --port 3000
   pm2 save
   pm2 startup

2. Nginx reverse proxy configuration:

   server {
    listen 80;
    server_name ai.example.com;
    location / {
        proxy_pass http://localhost:3000;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
   }

Secure Access Options

SSH Tunneling:

# Local tunnel setup
ssh -L 3000:localhost:3000 user@server-ip
# Browser access
http://localhost:3000

VPN Access:

1. Deploy WireGuard VPN
2. Configure firewall to allow 3000 only from VPN subnet
3. Access services through VPN connection

Performance Optimization

Monitoring Setup

# Install monitoring tools
sudo apt install -y htop sysstat
# Configure Node.js monitoring
npm install -g pm2-logrotate
pm2 set pm2-logrotate:max_size 10M

Performance Tuning

1. Adjust Node.js memory:

   export NODE_OPTIONS="--max-old-space-size=4096"

2. Implement rate limiting:

   {
   "rateLimit": {
    "max": 100,
    "windowMs": 60000
   }
   }

Troubleshooting Guide

Common Error Codes

Code	Cause	Solution
502	Service down	Check service status/logs
403	Auth failed	Verify API key validity
504	Timeout	Adjust model timeout settings

Log Analysis

1. Service logs:

   pm2 logs ai-service

2. Enable debug logging:

   export DEBUG=openclaw:*
   openclaw serve

3. System logs:

   journalctl -u nginx --no-pager -n 100

Advanced Features

Multi-Model Support

Update configuration:

{
  "providers": {
    "model-a": {...},
    "model-b": {...}
  }
}

Custom Routing

Create routes.js:

module.exports = (app) => {
  app.get('/custom', (req, res) => {
    res.send('Custom endpoint');
  });
};

Start with:

openclaw serve --routes ./routes.js

Authentication Integration

1. Install Passport.js:

   npm install passport passport-jwt

2. Configure JWT middleware

Maintenance and Upgrades

Version Upgrades

# Check for updates
npm outdated -g openclaw
# Perform upgrade
npm update -g openclaw
# Verify version
openclaw --version

Backup Strategy

1. Configuration backup:

   tar czvf openclaw-config-$(date +%Y%m%d).tar.gz ~/.openclaw/

2. Database backup (if applicable):
   ```bash

   Locate database file

   find / -name “openclaw.db” 2>/dev/null

Perform backup

cp /path/to/openclaw.db /backups/
```

Best Practices Summary

1. Security First: Always access admin interfaces through VPN/SSH tunnels
2. Resource Isolation: Create separate users for different model services
3. Monitoring: Configure alerts for critical metrics
4. Maintenance: Establish weekly maintenance windows
5. Documentation: Maintain detailed deployment records

This standardized approach enables developers to quickly deploy secure, production-ready AI inference platforms on Ubuntu 24.04 while maintaining flexibility for customization. The solution balances development convenience with production stability requirements.