AI Hub with SearXNG, Open WebUI, ComfyUI, and Ollama

This repository contains Docker Compose configurations for setting up an AI hub that includes:

• SearXNG (Privacy-focused metasearch engine)
• Open WebUI (Web interface for Ollama)
• ComfyUI (Advanced Generate Image Interface)
• Ollama (Local LLM runner)

Components

Name	Description	Port
SearXNG	Privacy-focused metasearch engine	8080
Open WebUI	Web interface for Ollama	3000
ComfyUI	Stable Diffusion interface	8188
Ollama	Local LLM runner	11434 (internal)
Caddy	Reverse proxy	80, 443
Valkey	In-memory database	internal

Prerequisites

1. Install Docker
2. Install Docker Compose
3. For ComfyUI: NVIDIA GPU with appropriate drivers installed
4. Git
5. Mandatory | Go to the open webui docs before you run any docker compose, you need to follow the steps to create the necessary files to run searxng with open webui. Open WebUI Docs - SearXNG

Docker Compose Files

There are three docker compose, those are:

1. docker-compose-w-ollama-comfyui.yaml - This docker compose file includes all the services, SearXNG, Open WebUI, ComfyUI, and Ollama.
2. docker-compose-w-ollama.yaml - This docker compose file includes SearXNG, Open WebUI, and Ollama.
3. docker-compose.yaml - This docker compose file includes only SearXNG with SearXNG. (This option is better if you run ollama in a server like runpod, and it exposes the port 11434 through a URL, so, you can use this docker in your local machine keeping the searxng and open webui configurations in your local machine)

Quick Start (Local Development)

1. Clone the repository:

git clone https://github.com/spacecodee/ai-hub-searxng-docker.git
cd ai-hub-searxng-docker

2. Create a .env file in the root directory:

SEARXNG_HOSTNAME=localhost

# Adjust these values based on your system's capabilities
SEARXNG_UWSGI_WORKERS=6
SEARXNG_UWSGI_THREADS=6

3. Start the services:

docker compose -f docker-compose-w-ollama-comfyui.yaml up -d

Accessing the Services

After starting the services, you can access them at:

• SearXNG: http://localhost:8080
• Open WebUI: http://localhost:3000
• ComfyUI: http://localhost:8188

If you want to run this docker-compose in your local machine that is connected to any server, you need to forward the ports using ssh:

ssh -L 8080:localhost:8080 -L 3000:localhost:3000 -L 8188:localhost:8188 user@server-ip

Configuring Open WebUI with SearXNG

Open WebUI can be integrated with SearXNG to provide web search capabilities. To set this up:

1. Access Open WebUI at http://localhost:3000
2. Go to Settings > Extensions
3. Enable the Web Search extension
4. Configure the SearXNG settings:
   • SearXNG URL: http://your-ip-or-container-name:8080/search?q=<query> (Must be like that)
   • Search Path: /search

For more detailed information about the Open WebUI and SearXNG integration, visit the official documentation.

ComfyUI Information

The ComfyUI installation includes:

• Base ComfyUI
• ComfyUI-Manager (automatically installed on startup)
• Support for custom nodes

ComfyUI Directory Structure:

comfyui/
├── models/      # Place your models here
├── output/      # Generated images appear here
├── input/       # Place input images here
└── custom_nodes/ # Custom nodes directory

Updating

To update the services:

# Pull the latest changes
git pull

# Rebuild and restart the containers
docker compose -f docker-compose-w-ollama-comfyui.yaml down
docker compose -f docker-compose-w-ollama-comfyui.yaml up -d --build

Troubleshooting

Viewing Logs

To view logs for all services:

docker compose -f docker-compose-w-ollama-comfyui.yaml logs -f

For specific services:

# SearXNG logs
docker compose -f docker-compose-w-ollama-comfyui.yaml logs -f searxng

# ComfyUI logs
docker compose -f docker-compose-w-ollama-comfyui.yaml logs -f comfyui

# Open WebUI logs
docker compose -f docker-compose-w-ollama-comfyui.yaml logs -f openwebui

# Ollama logs
docker compose -f docker-compose-w-ollama-comfyui.yaml logs -f ollama

Common Issues

1. ComfyUI Manager not installing: Check the ComfyUI container logs for any error messages during startup.
2. GPU not detected: Ensure NVIDIA drivers are properly installed and the NVIDIA Container Toolkit is configured.
3. Services not starting: Check if the required ports are available and not used by other applications.

Important Notes

• This configuration is intended for local development and testing purposes.
• For production deployments, additional security measures should be implemented:
  • Change default passwords
  • Configure proper SSL/TLS certificates
  • Implement authentication
  • Adjust worker and thread counts based on server capacity
  • Configure proper firewalls and access controls

License

This project is licensed under the MIT License - see the LICENSE file for details.