OpenHomeLab

A consolidated, well-organized Docker Compose homelab. 65+ services across ten categories — custom AI apps, AI image generation/editing, LLM inference, media server, photo/NVR, home automation, monitoring, utilities, network, dev tools, and infrastructure — managed from a single repository.

Philosophy

• One repo, every service. No more hunting through 15+ separate repos to find a compose file.
• Directory-per-service. Each service is isolated in services/<category>/<name>/ with its own docker-compose.yml and .env.example.
• No secrets in Git..env files are gitignored. .env.example files document every variable.
• Makefile orchestration. Start, stop, and inspect any service with a single command.
• Shared network. All services join the homelab Docker network for easy inter-service communication.

Quick Start

# 1. Clone
git clone https://github.com/davidamacey/OpenHomeLab.git
cd OpenHomeLab
# 2. Create the shared Docker network (one-time)
make network
# 3. Configure a service
cp services/media/immich/.env.example services/media/immich/.env
# Edit the .env file with your passwords and paths# 4. Start it
make up SERVICE=media/immich
# 5. Check status
make status

Service Catalog

Custom AI Projects (services/ai/)

These are custom-built applications with published Docker images. They require building from source or pulling from Docker Hub.

Service	Port(s)	GPU	Description
OpenTranscribe	5173–5181	0	AI transcription: speech-to-text with diarization, search
OpenAudio	5473–5479	0/2	Audio analysis: separation, classification, transcription
OpenSpeakers	5283–5286	0	TTS and voice cloning (multi-model, hot-swap)
VibeVoice	8300	0	Real-time voice conversion (single container)
Video Upscaler	-	0	Batch upscale old TV/film footage to 1080p/4K
Ring Detector	-	1	Doorbell event capture, storage, ML classification
OpenProcessor	4600–4608	2	Triton visual search: YOLO, face, OCR, CLIP + OpenSearch

Note: Custom projects use images from davidamacey/* on Docker Hub. See each service's .env.example for build and image tag instructions.

AI Image Generation & Editing (services/ai/)

Service	Port	GPU	Description
ComfyUI	8188	0	AI image generation with Flux models
Stable Diffusion	8600	2	SDXL TensorRT backend + web UI
IOPaint	7800	2	AI object removal and inpainting (LaMa, MAT, PowerPaint, BrushNet)

LLM / Inference Services (services/llm/)

Service	Port(s)	GPU	Description
Open WebUI	8010	-	Chat interface (connects to vLLM)
vLLM 20B	8012	2	GPT-OSS-20B served via vLLM (OpenAI-compatible)
vLLM 120B	8011	0+2	GPT-OSS-120B, tensor-parallel (disabled by default)
Apache Tika	8009	-	Document/OCR text extraction
Triton API	8000–8002	2	NVIDIA Triton model serving
YOLO API	8200	-	Object detection API (custom build)

Media Server (services/media/)

Service	Port(s)	GPU	Description
Jellyfin	8096	1	Free open-source media server (movies, TV, music)
Radarr	7878	-	Automatic movie collection manager
Sonarr	8989	-	Automatic TV show collection manager
Prowlarr	9696	-	Centralised indexer manager for arr apps
Bazarr	6767	-	Automatic subtitle downloader
Jellyseerr	5055	-	Media request and discovery UI for Jellyfin
qBittorrent	7460/6881	-	Feature-rich BitTorrent download client
SABnzbd	7470	-	Usenet binary newsreader and download client
Kavita	7480	-	Self-hosted manga, comics, and book reader
Audiobookshelf	13378	-	Audiobook and podcast server with mobile apps
Tdarr	8265/8266	2	Distributed video transcoding automation (NVENC)
MediaCMS	7950	-	Private video platform — upload/stream your own videos (YouTube-style)
PeerTube	7960	-	Full-featured private video platform: channels, playlists, live, API

Photo / NVR (services/media/, services/home/)

Service	Port(s)	GPU	Description
Immich	2283	1	Self-hosted photo & video management
Frigate	5000/8554/8555	1	GPU-accelerated NVR with object detect

Home Automation (services/home/)

Service	Port(s)	GPU	Description
Home Assistant	8123	-	Open-source home automation hub (3000+ integrations)
Node-RED	1880	-	Visual flow-based IoT and automation programming
Mosquitto	1883/9001	-	Lightweight MQTT message broker for IoT devices
Zigbee2MQTT	8885	-	Bridge Zigbee devices to MQTT (no hub required)

Home Management (services/home/)

Service	Port	GPU	Description
Homebox	3100	-	Home inventory management
Tandoor	9092	-	Recipe manager with PostgreSQL
Nextcloud	8080	-	Self-hosted cloud storage

Monitoring (services/monitoring/)

Service	Port(s)	GPU	Description
Uptime Kuma	3001	-	Service uptime monitoring and status pages
Grafana	3002	-	Metrics dashboards (Prometheus + node-exporter + cAdvisor + DCGM)
Prometheus	9090	-	Metrics collection (part of grafana stack)
DCGM Exporter	9400	-	NVIDIA GPU metrics exporter

Utilities (services/utilities/)

Service	Port(s)	GPU	Description
Stirling PDF	8080	-	40+ PDF manipulation tools
draw.io	8081/8443	-	Self-hosted diagramming
ntfy	80	-	Push notification server
Paperless-ngx	8070	-	Document management with OCR
Vaultwarden	8222	-	Bitwarden-compatible password vault
BookStack	6875	-	Self-hosted wiki and documentation
Duplicati	8210	-	Backup with cloud storage support
n8n	5678	-	Workflow automation with 400+ integrations
IT Tools	5620	-	100+ online developer tools (offline)
Memos	5230	-	Privacy-first lightweight note-taking

Network (services/network/)

Service	Port(s)	GPU	Description
AdGuard Home	3080/5353	-	Network-wide DNS ad and tracker blocking
wg-easy	51821/51820	-	WireGuard VPN with a simple web UI
Tailscale	-	-	Zero-config mesh VPN and subnet router (no port forwarding needed)
Cloudflare Zero Trust	-	-	WARP Connector — private LAN access for WARP clients via Cloudflare
Speedtest Tracker	7900	-	Scheduled internet speed monitoring with history

Dev Tools (services/dev/)

Service	Port(s)	GPU	Description
Gitea	3500/2222	-	Lightweight self-hosted Git service
code-server	3600	-	VS Code in the browser
Woodpecker CI	3610	-	Lightweight CI/CD with GitHub Actions syntax

Infrastructure (services/infra/)

Service	Port(s)	GPU	Description
Nginx Proxy Manager	80/81/443	-	Reverse proxy with GUI and SSL
Traefik	80/443/8090	-	Cloud-native reverse proxy with labels
Authentik	9300/9443	-	Identity provider with SSO/OIDC
Portainer	9444/9445	-	Docker management GUI
Heimdall	9000/9001	-	Application dashboard
Homepage	3000	-	Modern customizable dashboard
Cloudflare Tunnel	-	-	Secure remote access via Cloudflare
Watchtower	-	-	Automatic container image updates

Makefile Commands

# Single service
make up SERVICE=ai/comfyui # start
make down SERVICE=ai/comfyui # stop
make logs SERVICE=llm/open-webui # tail logs
make restart SERVICE=media/immich # restart
make pull SERVICE=ai/comfyui # pull latest image# Category
make up-category CATEGORY=ai # start all AI services
make down-category CATEGORY=llm # stop all LLM services# Global
make up-all # start everything
make down-all # stop everything
make pull-all # update all images
make status # show running containers
make gpu # GPU allocation + nvidia-smi
make network # create homelab network
make help# show all targets

GPU Configuration

This repo is configured for a three-GPU setup:

GPU	Model	VRAM	Assigned Services
0	RTX A6000	48 GB	ComfyUI, vLLM 120B (with GPU 2), OpenTranscribe, OpenAudio, OpenSpeakers, VibeVoice, Video Upscaler
1	RTX 3080 Ti	12 GB	Immich ML, Frigate NVR, Ring Detector
2	RTX A6000	48 GB	vLLM 20B, Triton, Stable Diffusion, OpenProcessor, OpenAudio (alt)

To adapt for your hardware: update the device_ids field in each GPU-enabled service's docker-compose.yml. See docs/gpu-allocation.md for the full allocation table.

Directory Structure

OpenHomeLab/
├── Makefile # Orchestration: up/down/logs/status/gpu
├── .env.example # Global variables: TZ, PUID, NAS_PATH
├── services/
│ ├── ai/
│ │ ├── comfyui/ # Flux image generation (GPU 0)
│ │ ├── stable-diffusion/ # SDXL TensorRT (GPU 2)
│ │ ├── iopaint/ # AI object removal & inpainting (GPU 2)
│ │ ├── opentranscribe/ # AI transcription platform (GPU 0)
│ │ ├── openaudio/ # Audio analysis platform (GPU 0/2)
│ │ ├── openspeakers/ # TTS & voice cloning (GPU 0)
│ │ ├── vibevoice/ # Real-time voice conversion (GPU 0)
│ │ ├── video-upscaler/ # Batch video upscaling (GPU 0)
│ │ ├── ring-detector/ # Doorbell ML classifier (GPU 1)
│ │ └── openprocessor/ # Triton visual search pipeline (GPU 2)
│ ├── llm/
│ │ ├── open-webui/ # vLLM + Open WebUI + Tika stack
│ │ └── triton/ # Triton server + YOLO API
│ ├── media/
│ │ ├── immich/ # Photo management (GPU 1)
│ │ ├── jellyfin/ # Media server: movies, TV, music (GPU 1)
│ │ ├── mediacms/ # Private video platform (YouTube-style)
│ │ ├── peertube/ # Full-featured private video platform
│ │ ├── radarr/ # Automatic movie downloader
│ │ ├── sonarr/ # Automatic TV show downloader
│ │ ├── prowlarr/ # Centralised indexer manager
│ │ ├── bazarr/ # Automatic subtitle downloader
│ │ ├── jellyseerr/ # Media request management
│ │ ├── qbittorrent/ # BitTorrent client
│ │ ├── sabnzbd/ # Usenet downloader
│ │ ├── kavita/ # Manga, comics, and book reader
│ │ ├── audiobookshelf/ # Audiobook and podcast server
│ │ └── tdarr/ # Video transcoding automation (GPU 2)
│ ├── home/
│ │ ├── homeassistant/ # Home automation hub (host networking)
│ │ ├── node-red/ # Visual IoT automation flows
│ │ ├── mosquitto/ # MQTT broker for IoT devices
│ │ ├── zigbee2mqtt/ # Zigbee-to-MQTT bridge
│ │ ├── homebox/ # Inventory management
│ │ ├── tandoor/ # Recipe manager
│ │ ├── nextcloud/ # Cloud storage
│ │ └── frigate/ # NVR camera system (GPU 1)
│ ├── monitoring/
│ │ ├── uptime-kuma/ # Uptime monitoring
│ │ └── grafana/ # Metrics: Prometheus + Grafana + Loki + DCGM
│ ├── network/
│ │ ├── adguard-home/ # DNS ad and tracker blocking
│ │ ├── wg-easy/ # WireGuard VPN with web UI
│ │ ├── tailscale/ # Mesh VPN subnet router (no port forwarding)
│ │ ├── cloudflare-zero-trust/ # WARP connector for private LAN access
│ │ └── speedtest-tracker/ # Scheduled internet speed monitoring
│ ├── dev/
│ │ ├── gitea/ # Self-hosted Git service
│ │ ├── code-server/ # VS Code in the browser
│ │ └── woodpecker-ci/ # CI/CD pipeline server + agent
│ ├── utilities/
│ │ ├── stirling-pdf/ # PDF tools
│ │ ├── drawio/ # Diagrams
│ │ ├── ntfy/ # Push notifications
│ │ ├── paperless-ngx/ # Document management
│ │ ├── vaultwarden/ # Password vault
│ │ ├── bookstack/ # Wiki / documentation
│ │ ├── duplicati/ # Backup to cloud or local
│ │ ├── n8n/ # Workflow automation (400+ integrations)
│ │ ├── it-tools/ # 100+ developer tools, fully offline
│ │ └── memos/ # Lightweight note-taking
│ └── infra/
│ ├── nginx-proxy/ # Reverse proxy (GUI)
│ ├── traefik/ # Reverse proxy (label-based)
│ ├── authentik/ # SSO / identity provider
│ ├── portainer/ # Docker management GUI
│ ├── heimdall/ # App dashboard
│ ├── homepage/ # App dashboard (alternative)
│ ├── cloudflared/ # Cloudflare Tunnel (remote access)
│ └── watchtower/ # Auto-updates
├── scripts/
│ ├── deploy.sh # Deploy all or specific services
│ ├── stop.sh # Stop services
│ ├── status.sh # Show running status
│ ├── backup.sh # Backup service volumes
│ └── gpu-check.sh # GPU allocation overview
└── docs/
├── gpu-allocation.md # GPU assignment strategy
├── port-map.md # All ports used by all services
├── backup-strategy.md # Backup procedures
└── adding-a-service.md # How to add a new service

Environment Variables

Each service has a .env.example file listing its required variables. Copy it to .env and fill in your values before starting the service.

Global variables (shared across services) are documented in the root .env.example:

Variable	Default	Description
TZ	America/New_York	Timezone
PUID	1000	User ID for file ownership
PGID	1000	Group ID for file ownership
NAS_PATH	/mnt/nas	Base path to storage volume
DATA_PATH	/mnt/nas/appdata	Container app data

Port Conflicts

A few services share default ports (80, 8080). See docs/port-map.md for the full port inventory and conflict resolution guidance. The recommended solution is to run nginx-proxy-manager or traefik on 80/443 and proxy all other services through it.

Documentation

• GPU Allocation — which GPU each service uses and why
• Port Map — every port, every service
• Backup Strategy — how to back up and restore
• Networking Guide — local DNS, reverse proxy, Cloudflare Tunnel, and Immich phone sync
• Adding a Service — template and conventions for new services

Related Projects

The application source code for custom services lives in separate repositories:

• OpenTranscribe — AI transcription with diarization and search
• OpenAudio — Audio analysis and classification platform
• OpenSpeakers — TTS and voice cloning
• OpenProcessor — Triton visual search pipeline (YOLO, face, OCR, CLIP)
• OpenRingDetector — Self-hosted Ring doorbell AI detection

This repo contains only deployment configs — docker-compose.yml files that pull published images. Application development happens in the source repos.

License

MIT — use, adapt, and share freely.