n8n-Ollama-WebUI Stack

This repository provides a complete self-hosted workflow automation and AI environment featuring:

• n8n – visual workflow automation
• Ollama – local LLM model server
• Open-WebUI – browser UI for chatting with Ollama models
• PostgreSQL – persistent database backend for n8n
• Nginx – TLS-terminating reverse proxy for all services

All services run on Ubuntu using Docker Compose, with optional GPU support for Ollama.

---

Directory layout

.
├── docker-compose.yml        # All services: n8n, postgres, ollama, open-webui, nginx
├── nginx/
│   └── default.conf          # Nginx reverse proxy config
├── certs/
│   ├── fullchain.pem         # SSL certificate
│   └── privkey.pem           # SSL private key
├── .env                      # Environment variables
├── db_data/                  # Postgres volume
├── n8n_data/                 # n8n persistent data
├── webui_data/               # Open-WebUI data
└── ollama/                   # Ollama model and config cache

---

Prerequisites

• Ubuntu 22.04+ host
• Docker Engine + Docker Compose plugin
• IPv4 connectivity on your public interface (IPv6 may be used only for management)
• Valid TLS certificate and key under certs/ (or self-signed for testing)

---

Quick start

1. Clone this repository

git clone https://github.com/<yourname>/n8n-ollama-stack.git
cd n8n-mcp-agents

2. Prepare environment file

Copy .env.template → .env and edit:

POSTGRES_USER=n8n
POSTGRES_PASSWORD=n8n
POSTGRES_DB=n8n

# Random secret for n8n encryption
N8N_ENCRYPTION_KEY=$(openssl rand -hex 32)

# Timezone and hostname
TZ=America/New_York
N8N_HOST=localhost
WEBHOOK_URL=https://<your-public-hostname>:8443/
MODEL_NAME=llama3

3. Launch the stack

docker compose up -d

The first start will pull images and initialize Postgres and Ollama.

---

Service endpoints

Service	Port	Description / URL
Nginx reverse proxy	443	Entry point for all HTTPS traffic
n8n	5678	Exposed via Nginx at / → https:/// or directly http://:5678/
Open-WebUI	8080	Exposed via Nginx at /webui/ → https:///webui/ or directly http://:8080/
Ollama API	11434	Exposed via Nginx at /ollama/ → https:///ollama/
PostgreSQL	5432	Local only, used by n8n

Note: Nginx proxies internally to 127.0.0.1 (IPv4). Ensure your host networking prefers IPv4 (see below).

---

GPU support (optional)

If your host has an NVIDIA GPU and drivers installed:

docker compose up -d ollama

The ollama service already includes:

deploy:
  resources:
    reservations:
      devices:
        - capabilities: [gpu]

so Docker will pass GPU devices automatically if available.

---

Logs & troubleshooting

docker compose logs -f nginx
docker compose logs -f n8n
docker compose logs -f open-webui
docker compose logs -f ollama

To test upstream reachability:

curl -vk https://<your-host>/ollama/api/tags
curl -vk https://<your-host>/
curl -vk https://<your-host>/webui/

If 502 Bad Gateway appears, verify that:

• Nginx proxy_pass uses 127.0.0.1 (not localhost)
• Services are listening on 0.0.0.0 or 127.0.0.1 ports
• IPv4 preference (/etc/gai.conf) is applied

---

Shut down and cleanup

docker compose down

To remove volumes (wipe Postgres/n8n data):

docker compose down -v

---

License

MIT License © 2025 Komal Thareja