How to Self-Host Shinobi with Docker Compose

What Is Shinobi?

Shinobi is an open-source video management system (VMS) written in Node.js. It supports IP cameras via RTSP, MJPEG, and HLS, with features including motion detection, recording, live streaming via WebSocket, multi-user/multi-tenant support, and a plugin system for AI object detection. Shinobi was created as a modern alternative to ZoneMinder, with a cleaner web interface and easier Docker deployment.

Prerequisites

A Linux server (Ubuntu 22.04+ recommended)
Docker and Docker Compose installed (guide)
2 GB+ of free RAM
IP cameras with RTSP support
Storage space for recordings
A domain name (optional, for remote access)

Docker Compose Configuration

Create a docker-compose.yml file:

services:
  shinobi:
    image: shinobisystems/shinobi:dev
    container_name: shinobi
    restart: unless-stopped
    ports:
      - "8080:8080"  # Web UI
    environment:
      ADMIN_USER: [email protected]        # CHANGE: Admin email
      ADMIN_PASSWORD: your-admin-password     # CHANGE: Use a strong password
      CRON_KEY: your-cron-key-here            # CHANGE: Random string for cron API
      PLUGINKEY_MOTION: your-motion-key       # CHANGE: Random string for motion plugin
      PLUGINKEY_OPENCV: your-opencv-key       # CHANGE: Random string for OpenCV plugin
      PLUGINKEY_OPENALPR: your-openalpr-key   # CHANGE: Random string for ALPR plugin
    volumes:
      - shinobi-config:/config
      - shinobi-customAutoLoad:/home/Shinobi/libs/customAutoLoad
      - shinobi-database:/var/lib/mysql
      - /path/to/recordings:/home/Shinobi/videos  # CHANGE: Your recordings path
      - /dev/shm/shinobiStreams:/dev/shm/streams
    depends_on:
      - shinobi-db
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080"]
      interval: 30s
      timeout: 10s
      retries: 3

  shinobi-db:
    image: mariadb:11
    container_name: shinobi-db
    restart: unless-stopped
    environment:
      MYSQL_ROOT_PASSWORD: your-root-password       # CHANGE: Use a strong password
      MYSQL_DATABASE: ccio
      MYSQL_USER: majesticflame
      MYSQL_PASSWORD: your-db-password              # CHANGE: Use a strong password
    volumes:
      - shinobi-db-data:/var/lib/mysql
    healthcheck:
      test: ["CMD", "healthcheck.sh", "--connect", "--innodb_initialized"]
      interval: 10s
      timeout: 5s
      retries: 5

volumes:
  shinobi-config:
  shinobi-customAutoLoad:
  shinobi-database:
  shinobi-db-data:

Configuration notes:

Shinobi uses dev tag for the latest build. The project doesn’t use semantic version tags on Docker Hub — dev is the rolling release. Pin by digest if you need reproducibility.
MariaDB is required for camera config, user management, and event metadata.
/dev/shm/streams is used for live streaming data. The shared memory mount ensures low-latency video delivery.
Plugin keys are random strings used to authenticate built-in plugins (motion detection, OpenCV, ALPR). Generate them with openssl rand -hex 16.

Start the stack:

docker compose up -d

Initial Setup

Open http://your-server:8080 in your browser
Log in with the super admin credentials:
- URL: http://your-server:8080/super
- Email: The ADMIN_USER from docker-compose
- Password: The ADMIN_PASSWORD from docker-compose
In the super admin panel, create a regular user account:
- Click Accounts → Add
- Enter email, password, and set storage limits
Log out and log in at http://your-server:8080 with the regular user account
Add your first camera:
- Click the + button in the top menu
- Set Mode to “Record” or “Watch Only”
- Under Input, set:
  - Type: H.264 / H.265
  - Full URL Path: rtsp://user:pass@camera-ip:554/stream1
- Under Stream, set:
  - Type: HLS or WebSocket (WebSocket is lower latency)
- Click Save

Camera Modes

Mode	Description
Disabled	Camera is off
Watch Only	Live view, no recording, no detection
Record	Continuous recording to disk
Idle	Camera connected but not streaming/recording

Configuration

Motion Detection

Shinobi handles motion detection through its plugin system:

In camera settings, go to the Detector tab
Enable the detector and set:
- Send Frames: Yes
- How to Record: Traditional (saves to disk)
- Trigger Record: Yes (record only when motion is detected)
- Detection Region: Draw zones on the camera view
Set sensitivity:
- Indifference: Lower = more sensitive (default: 5)
- Max Sensitivity: Upper limit for motion detection

Storage Limits

Control disk usage per user:

In the super admin panel (/super), edit the user account
Set Storage Limit in MB (e.g., 50000 for ~50 GB)
Shinobi automatically purges oldest recordings when the limit is reached

Alternatively, set per-camera retention:

In camera settings → Recording tab
Set Days to keep videos (e.g., 14)
Shinobi deletes recordings older than this automatically

API Access

Shinobi provides a REST API for integration:

# Get list of monitors (cameras)
curl http://your-server:8080/API_KEY/monitor/GROUP_KEY

# Get a snapshot from a camera
curl http://your-server:8080/API_KEY/jpeg/GROUP_KEY/MONITOR_ID/snapshot.jpg

API keys are generated per user in account settings.

Advanced Configuration (Optional)

AI Object Detection (OpenCV Plugin)

Shinobi supports object detection via the built-in OpenCV plugin:

The plugin runs inside the Shinobi container
Enable it in camera settings → Detector tab
Set Detector Plugin to “OpenCV”
Configure detection classes (person, car, etc.)

For better detection accuracy, use TensorFlow or YOLO models via the plugin system. Note that Shinobi’s AI plugin documentation is sparse — you may need to experiment or consult community guides.

Multi-Tenant Setup

Shinobi’s multi-tenant capability is its standout feature:

Create separate user accounts in the super admin panel
Each user has their own cameras, recordings, and storage quota
Users cannot see each other’s cameras or recordings
Set per-user API keys for integrations

This makes Shinobi suitable for small businesses or shared housing where different tenants manage their own cameras.

ONVIF Camera Discovery

Shinobi supports ONVIF for automatic camera discovery:

In the main interface, click the ONVIF probe button
Shinobi scans the local network for ONVIF-compatible cameras
Select a discovered camera to auto-fill its RTSP URL and configuration

Reverse Proxy

For Nginx Proxy Manager:

Scheme: http
Forward Hostname: shinobi
Forward Port: 8080
WebSocket Support: ON (required for live streaming)

WebSocket support is critical — Shinobi’s live view uses WebSocket connections for low-latency streaming.

See Reverse Proxy Setup for detailed instructions.

Backup

Critical data to back up:

MariaDB database: Camera configurations, user accounts, event metadata
/config/: Shinobi configuration files
/home/Shinobi/videos/: Recorded video files (large)

# Backup database
docker exec shinobi-db mysqldump -u majesticflame -p'your-db-password' ccio > shinobi-backup.sql

# Backup config
docker cp shinobi:/config ./shinobi-config-backup/

See Backup Strategy for a comprehensive backup approach.

Troubleshooting

Camera shows spinning/loading but no video

Symptom: Camera is added but live view shows a loading spinner. Fix: Verify the RTSP URL works outside Shinobi:

ffplay -rtsp_transport tcp rtsp://user:pass@camera-ip:554/stream

Try changing the stream type from WebSocket to HLS in camera settings. Some browsers block WebSocket connections on non-HTTPS pages.

High CPU usage

Symptom: Node.js process consuming 100% CPU. Fix: Reduce the number of cameras streaming simultaneously. Lower the stream resolution in camera settings. Disable sub-streams if you enabled multiple stream outputs per camera. Check if a detector plugin is running CPU-heavy inference.

Recordings not saving

Symptom: Camera shows live view but no recordings appear. Fix: Verify the camera mode is set to “Record” (not “Watch Only” or “Idle”). Check that the recordings volume has write permissions and free space. Check Shinobi logs:

docker logs shinobi 2>&1 | grep -i error

Cannot access super admin panel

Symptom: /super returns a login page but credentials don’t work. Fix: The super admin credentials are set via environment variables on first run. If you changed them after initial setup, the database values take precedence. Reset via MariaDB:

docker exec -it shinobi-db mysql -u root -p ccio

Then update the Users table with new hashed credentials.

Resource Requirements

RAM: 1-2 GB for 3-5 cameras. 4 GB+ for 10+ cameras.
CPU: Node.js is single-threaded for the main process, but FFmpeg spawns per-camera processes. 2-4 cores recommended for 5+ cameras.
Disk: 10-50 GB per camera per week depending on resolution and recording mode.
Network: Each RTSP stream uses 2-8 Mbps.

Verdict

Shinobi is a functional NVR with a modern web interface and useful multi-tenant capabilities. Its Node.js architecture and WebSocket-based streaming make it more web-friendly than older alternatives like ZoneMinder.

However, the development pace has slowed significantly, documentation is sparse, and the project is essentially a one-person operation. The Docker image uses rolling dev tags with no semantic versioning, making reproducible deployments harder.

For new NVR deployments in 2026, Frigate is the recommended choice — better AI detection, better documentation, larger community, and more active development. Consider Shinobi if you specifically need multi-tenant user management or prefer web-based camera configuration over YAML files.

Frequently Asked Questions

Is Shinobi still maintained?

Shinobi receives commits on GitLab, but the pace has slowed. There are no formal version releases — the project uses rolling updates on the dev Docker tag. It’s functional but the long-term outlook is uncertain compared to Frigate’s active development.

Can Shinobi integrate with Home Assistant?

Not natively. You can use Shinobi’s API to create camera entities in Home Assistant, but it requires manual configuration. Frigate’s Home Assistant integration is significantly more polished and feature-rich.

What’s the difference between Shinobi CE and Pro?

Shinobi CE (Community Edition) is the open-source version. Shinobi Pro was a paid version with additional features, but the distinction has been blurred in recent years. The Docker image available on Docker Hub includes all features.