Files
ts-immich-machine-learning/README.md
2025-11-01 15:24:17 -04:00

108 lines
3.9 KiB
Markdown

# Immich Machine Learning with Tailscale Integration
![Immich Machine Learning with Tailscale](https://immich.app/img/immich-logo-inline-light.png "Immich Machine Learning")
This project sets up a Immich Machine Learning instance with Tailscale VPN integration using Docker Compose. It creates a secure, private network connection for your Immich Machine Learning instance using Tailscale.
## Prerequisites
- Docker and Docker Compose installed on your system
- A Tailscale account and auth key (get one from https://login.tailscale.com/admin/authkeys)
- Basic understanding of Docker and networking concepts
## Project Structure
```
ts-immich-machine-learning/
├── docker-compose.yml
├── tailscale/
├── tailscale-data/ # Persistent Tailscale state
└── config/ # Tailscale configuration files
```
## Setup Instructions
1. **Clone the Repository**
```bash
git clone https://gitea.damconsulting.llc/DAM/ts-immich-machine-learning
cd ts-immich-machine-learning
```
2. Create Required Directories
```bash
mkdir -p tailscale/tailscale-data
```
3. Configure Tailscale
- Replace `{{YOUR_TAILSCALE_AUTHKEY}}` in the docker-compose.yml with your actual Tailscale auth key
- Optionally, update the file in `tailscale/config/serve.json` if you need specific Tailscale serve configurations
- CAUTION: Changing `"${TS_CERT_DOMAIN}:443": false` to `true` will expose the service to the internet
4. Configure Immich Machine Learning
- See the [documentation](https://docs.immich.app/features/ml-hardware-acceleration) for configuration options
5. Start the Services
```bash
docker compose up -d
```
6. Wait for Certificate to propagate [~2m]
7. Configure Immich
- After starting the services your service should be available via tailnet at https://immich-machine-learning.{{YOUR_TAILNET_DOMAIN}}.ts.net ie https://immich-machine-learning.tail12345.ts.net/
- add the new machine learning service to the immich configuration json under the machine learning section
## Services
### immich-machine-learning-ts (Tailscale)
- Runs Tailscale VPN client
- Image: tailscale/tailscale:latest
- Container name: immich-machine-learning-ts
- Hostname: immich-machine-learning
- Requires NET_ADMIN and SYS_MODULE capabilities
- Persists state in ./tailscale/tailscale-data
- Uses configuration from ./tailscale/config
### immich-machine-learning
- Depends on immich-machine-learning-ts service
## Usage
- After starting the services your service should be available via tailnet at `https://immich-machine-learning.{{YOUR_TAILNET_DOMAIN}}.ts.net` ie `https://immich-machine-learning.tail12345.ts.net/`
- To manually get the Tailscale IP/hostname of your container:
```bash
docker logs immich-machine-learning-ts
```
Look for the Tailscale IP address in the logs.
## Optional Features
- Uncomment and adjust the ports mapping if you need direct access (without Tailscale):
```yaml
ports:
- 3003:3003
```
- Stopping the Services
```bash
docker compose down
```
## Troubleshooting
- Check container logs:
```bash
docker logs immich-machine-learning-ts
docker logs immich-machine-learning
```
- Ensure your Tailscale auth key is valid and not expired
- Verify the configuration files have proper permissions
- Make sure required directories exist before starting
## Notes
- The Immich Machine Learning service uses the Tailscale service's network stack via `network_mode: service:immich-machine-learning-ts`
- Direct port mapping is disabled by default as Tailscale handles the networking
- Services restart automatically unless explicitly stopped
- For more information:
- Tailscale documentation: https://tailscale.com/kb/
- Immich Machine Learning [documentation](https://docs.immich.app/features/ml-hardware-acceleration)
- Immich Machine Learning [repository](https://github.com/immich-app/immich/tree/main/machine-learning)