Files
n8n-mcp/docs/HTTP_DEPLOYMENT.md
czlonkowski 149b59a541 docs: comprehensive documentation update for production deployment
- Updated README.md with clear setup instructions and recent updates
- Simplified Claude Desktop setup guide with troubleshooting
- Enhanced HTTP deployment guide for production use
- Streamlined troubleshooting guide with quick fixes
- Added mcp-http-client.js for Node.js 16 compatibility
- Fixed stdio mode console output corruption

Key improvements:
- Clear distinction between local and remote deployment
- Node.js 18+ requirement for mcp-remote clearly documented
- USE_FIXED_HTTP=true prominently featured for v2.3.2
- Production deployment best practices
- Multi-user service considerations

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-06-15 01:11:02 +02:00

431 lines
8.7 KiB
Markdown

# HTTP Deployment Guide for n8n-MCP
Deploy n8n-MCP as a remote HTTP server to provide n8n knowledge to Claude from anywhere.
## 🎯 Overview
n8n-MCP HTTP mode enables:
- ☁️ Cloud deployment (VPS, Docker, Kubernetes)
- 🌐 Remote access from any Claude Desktop client
- 🔒 Token-based authentication
- ⚡ Production-ready performance (~12ms response time)
- 🔧 Fixed implementation (v2.3.2) for stability
## 📋 Prerequisites
**Server Requirements:**
- Node.js 16+ or Docker
- 512MB RAM minimum
- Public IP or domain name
- (Recommended) SSL certificate for HTTPS
**Client Requirements:**
- Claude Desktop
- Node.js 18+ (for mcp-remote)
- Or Claude Pro/Team (for native remote MCP)
## 🚀 Quick Start
### Option 1: Docker Deployment (Recommended)
```bash
# 1. Create environment file
cat > .env << EOF
AUTH_TOKEN=$(openssl rand -base64 32)
USE_FIXED_HTTP=true
MCP_MODE=http
PORT=3000
EOF
# 2. Deploy with Docker
docker run -d \
--name n8n-mcp \
--restart unless-stopped \
--env-file .env \
-p 3000:3000 \
ghcr.io/czlonkowski/n8n-mcp:latest
# 3. Verify deployment
curl http://localhost:3000/health
```
### Option 2: Manual Installation
```bash
# 1. Clone and setup
git clone https://github.com/czlonkowski/n8n-mcp.git
cd n8n-mcp
npm install
npm run build
npm run rebuild
# 2. Configure environment
export MCP_MODE=http
export USE_FIXED_HTTP=true # Important: Use fixed implementation
export AUTH_TOKEN=$(openssl rand -base64 32)
export PORT=3000
# 3. Start server
npm run start:http
```
💡 **Save your AUTH_TOKEN** - clients will need it to connect!
## ⚙️ Configuration
### Required Environment Variables
| Variable | Description | Example |
|----------|-------------|------|
| `MCP_MODE` | Must be set to `http` | `http` |
| `USE_FIXED_HTTP` | **Important**: Set to `true` for v2.3.2 fixes | `true` |
| `AUTH_TOKEN` | Secure token (32+ characters) | `generated-token` |
### Optional Settings
| Variable | Description | Default |
|----------|-------------|---------|
| `PORT` | Server port | `3000` |
| `HOST` | Bind address | `0.0.0.0` |
| `LOG_LEVEL` | Log verbosity | `info` |
| `NODE_ENV` | Environment | `production` |
## 🔐 Security Setup
### Authentication
All requests require Bearer token authentication:
```bash
# Test authentication
curl -H "Authorization: Bearer $AUTH_TOKEN" \
https://your-server.com/health
```
### SSL/HTTPS (Strongly Recommended)
Use a reverse proxy for SSL termination:
**Nginx example:**
```nginx
server {
listen 443 ssl;
server_name your-domain.com;
ssl_certificate /path/to/cert.pem;
ssl_certificate_key /path/to/key.pem;
location /mcp {
proxy_pass http://localhost:3000;
proxy_set_header Authorization $http_authorization;
}
}
```
**Caddy example (automatic HTTPS):**
```caddy
your-domain.com {
reverse_proxy /mcp localhost:3000
}
```
## 💻 Client Configuration
### For All Claude Desktop Users
**Requirements**: Node.js 18+ installed locally
```json
{
"mcpServers": {
"n8n-remote": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://your-server.com/mcp",
"--header",
"Authorization: Bearer ${AUTH_TOKEN}"
],
"env": {
"AUTH_TOKEN": "your-auth-token-here"
}
}
}
}
```
### For Claude Pro/Team Users
Use native remote MCP support:
1. Go to Settings > Integrations
2. Add your MCP server URL
3. Complete OAuth flow (if implemented)
⚠️ **Note**: Direct config file entries won't work for remote servers in Pro/Team.
## 🌐 Production Deployment
### Docker Compose Setup
```yaml
version: '3.8'
services:
n8n-mcp:
image: ghcr.io/czlonkowski/n8n-mcp:latest
container_name: n8n-mcp
restart: unless-stopped
environment:
MCP_MODE: http
USE_FIXED_HTTP: true
AUTH_TOKEN: ${AUTH_TOKEN:?AUTH_TOKEN required}
NODE_ENV: production
LOG_LEVEL: info
ports:
- "127.0.0.1:3000:3000" # Bind to localhost only
volumes:
- n8n-mcp-data:/app/data
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:3000/health"]
interval: 30s
timeout: 10s
retries: 3
deploy:
resources:
limits:
memory: 512M
reservations:
memory: 256M
volumes:
n8n-mcp-data:
```
### Systemd Service (Linux)
Create `/etc/systemd/system/n8n-mcp.service`:
```ini
[Unit]
Description=n8n-MCP HTTP Server
After=network.target
[Service]
Type=simple
User=n8n-mcp
WorkingDirectory=/opt/n8n-mcp
ExecStart=/usr/bin/node dist/mcp/index.js
Restart=always
RestartSec=10
# Environment
Environment="MCP_MODE=http"
Environment="USE_FIXED_HTTP=true"
Environment="NODE_ENV=production"
EnvironmentFile=/opt/n8n-mcp/.env
# Security
NoNewPrivileges=true
PrivateTmp=true
[Install]
WantedBy=multi-user.target
```
Enable:
```bash
sudo systemctl enable n8n-mcp
sudo systemctl start n8n-mcp
```
## 📡 Monitoring & Maintenance
### Health Checks
```bash
# Basic health check
curl https://your-server.com/health
# Response:
{
"status": "ok",
"mode": "http-fixed",
"version": "2.3.2",
"uptime": 3600,
"memory": {
"used": 45,
"total": 512,
"unit": "MB"
}
}
```
### Monitoring with Prometheus
```yaml
# prometheus.yml
scrape_configs:
- job_name: 'n8n-mcp'
static_configs:
- targets: ['localhost:3000']
metrics_path: '/health'
bearer_token: 'your-auth-token'
```
### Log Management
```bash
# Docker logs
docker logs -f n8n-mcp --tail 100
# Systemd logs
journalctl -u n8n-mcp -f
# Log rotation (Docker)
docker run -d \
--log-driver json-file \
--log-opt max-size=10m \
--log-opt max-file=3 \
n8n-mcp
```
## 🔒 Security Best Practices
### 1. Token Management
```bash
# Generate strong tokens
openssl rand -base64 32
# Rotate tokens regularly
AUTH_TOKEN_NEW=$(openssl rand -base64 32)
docker exec n8n-mcp env AUTH_TOKEN=$AUTH_TOKEN_NEW
```
### 2. Network Security
-**Always use HTTPS** in production
-**Firewall rules** to limit access
-**VPN** for internal deployments
-**Rate limiting** at proxy level
### 3. Container Security
```bash
# Run as non-root user (already configured)
# Read-only filesystem
docker run --read-only \
--tmpfs /tmp \
-v n8n-mcp-data:/app/data \
n8n-mcp
# Security scanning
docker scan ghcr.io/czlonkowski/n8n-mcp:latest
```
## 🔍 Troubleshooting
### Common Issues
**"Stream is not readable" error:**
- ✅ Solution: Ensure `USE_FIXED_HTTP=true` is set
- This is fixed in v2.3.2
**"TransformStream is not defined" (client-side):**
- 🔄 Update Node.js to v18+ on client machine
- Or use Docker stdio mode instead
**Connection refused:**
```bash
# Check server is running
curl http://localhost:3000/health
# Check Docker status
docker ps
docker logs n8n-mcp
# Check firewall
sudo ufw status
```
**Authentication failed:**
- Verify AUTH_TOKEN matches exactly
- Check for extra spaces or quotes
- Test with curl first
### Debug Mode
```bash
# Enable debug logging
LOG_LEVEL=debug docker run ...
# Test MCP endpoint directly
curl -X POST https://your-server.com/mcp \
-H "Authorization: Bearer $AUTH_TOKEN" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","method":"list_nodes","params":{"limit":5},"id":1}'
```
## 🚀 Scaling & Performance
### Performance Metrics
- Average response time: **~12ms**
- Memory usage: **~50-100MB**
- Concurrent connections: **100+**
- Database queries: **<5ms** with FTS5
### Horizontal Scaling
The server is stateless - scale easily:
```yaml
# Docker Swarm example
deploy:
replicas: 3
update_config:
parallelism: 1
delay: 10s
restart_policy:
condition: on-failure
```
### Optimization Tips
1. **Use Docker** for consistent performance
2. **Enable HTTP/2** in your reverse proxy
3. **Set up CDN** for static assets
4. **Monitor memory** usage over time
## 👥 Multi-User Service Considerations
While n8n-MCP is designed for single-user deployments, you can build a multi-user service:
1. **Use this as a core engine** with your own auth layer
2. **Deploy multiple instances** with different tokens
3. **Add user management** in your proxy layer
4. **Implement rate limiting** per user
See [Architecture Guide](./ARCHITECTURE.md) for building multi-user services.
## 📦 Updates & Maintenance
```bash
# Update to latest version
docker pull ghcr.io/czlonkowski/n8n-mcp:latest
docker compose up -d
# Backup database
docker cp n8n-mcp:/app/data/nodes.db ./backup-$(date +%Y%m%d).db
# Restore database
docker cp ./backup.db n8n-mcp:/app/data/nodes.db
docker restart n8n-mcp
```
## 🆘 Getting Help
- 📚 [Full Documentation](https://github.com/czlonkowski/n8n-mcp)
- 🐛 [Report Issues](https://github.com/czlonkowski/n8n-mcp/issues)
- 💬 [Discussions](https://github.com/czlonkowski/n8n-mcp/discussions)