n8n with PostgreSQL (pgvector) and Cloudflare Tunnel

A complete Docker Compose setup for running n8n locally with:

• PostgreSQL with pgvector extension for vector database capabilities
• Automatic database initialization with separate databases for n8n and documents
• Cloudflare Tunnel for webhook support and external access

Features

• 🚀 n8n workflow automation with PostgreSQL backend
• 🔍 Vector database support using pgvector extension
• 🌐 External webhook access via Cloudflare Tunnel
• 📊 Dual database setup: Main n8n database + separate documents database
• 🔧 Automated initialization with custom schema and tables

Prerequisites

• Docker and Docker Compose installed
• Cloudflare account with tunnel token (for webhook functionality)

Quick Start

1. Configure environment variables

   IMPORTANT: Copy the example.env file and name it .env and adjust the variables:

   # Database Configuration
   POSTGRES_USER=your_admin_user
   POSTGRES_PASSWORD=your_secure_password
   POSTGRES_DB=n8n
   POSTGRES_DOCUMENTS_DB=documents
   
   # n8n Configuration
   N8N_HOST=your-domain.com
   N8N_PROTOCOL=https
   N8N_WEBHOOK_URL=https://your-domain.com
   
    # Get the token from the cloudflare tunnel for the N8N_HOST
   TUNNEL_TOKEN=your_cloudflare_tunnel_token

2. Start the services

   docker-compose up -d

3. Access n8n

   • Local: http://localhost:5678
   • External: https://your-domain.com (via Cloudflare Tunnel)

4. Stop the services

   docker-compose stop

Database Structure

The setup automatically creates two databases:

Main Database (n8n)

• Used by n8n for storing workflows, executions, and settings
• Configured via environment variables in n8n service

Documents Database (documents)

• Dedicated vector database for AI/ML workflows
• Includes pgvector extension for similarity search
• Schema defined in postgres/sql/schema.sql
• Perfect for RAG (Retrieval-Augmented Generation) workflows

Services

PostgreSQL (pgvector/pgvector:pg16)

• Port: 5432
• Extensions: pgvector for vector operations
• Initialization: Runs postgres/init-data.sh on first startup
• Health Check: Ensures database is ready before starting n8n

n8n (docker.n8n.io/n8nio/n8n)

• Port: 5678
• Database: PostgreSQL backend
• Storage: Persistent volume for user data
• Dependencies: Waits for PostgreSQL health check

Cloudflare Tunnel (cloudflare/cloudflared:latest)

• Purpose: Exposes n8n to the internet for webhook functionality
• Configuration: Uses TUNNEL_TOKEN from environment
• Benefits: No need to open firewall ports or configure reverse proxy

Vector Database Usage

The documents database is configured with pgvector extension, making it perfect for:

• Embeddings storage for AI workflows
• Similarity search operations
• RAG implementations in n8n
• Document indexing and retrieval

Example n8n workflow usage:

1. Store document embeddings in the documents database
2. Perform similarity searches using pgvector functions
3. Integrate with AI nodes for intelligent document retrieval

Customization

Database Schema

Modify postgres/sql/schema.sql to customize the documents database structure.

Initialization Script

Update postgres/init-data.sh to add custom database setup logic.

Environment Variables

All configuration is centralized in .env file for easy customization.

Troubleshooting

Database Connection Issues

# Check PostgreSQL logs
docker-compose logs postgres

# Verify database health
docker-compose exec postgres pg_isready -U admin -d n8n

Cloudflare Tunnel Issues

# Check tunnel logs
docker-compose logs cloudflared

# Verify tunnel token is valid

n8n Access Issues

# Check n8n logs
docker-compose logs n8n

# Verify n8n is healthy
curl http://localhost:5678/healthz

Data Persistence

• Database data: Stored in Docker volume db_storage
• n8n user data: Stored in Docker volume n8n_storage
• Volumes persist between container restarts