Installing Sentinel
Hardware, software prerequisites and installation methods.
Hardware
| Component | Minimum | Recommended |
|---|---|---|
| CPU | 8 cores | 16+ cores (Xeon / EPYC) |
| RAM | 16 GB | 32 GB |
| Storage | 100 GB SSD | 500 GB+ NVMe for recordings |
| GPU | NVIDIA, 8 GB VRAM | NVIDIA T4 / A10G / RTX 3080+ |
| Network | 100 Mbps | 1 Gbps for multi-camera streams |
Supported environments
- Tested: Ubuntu 22.04 and 24.04 LTS on bare metal; AWS EC2 g4dn.xlarge (T4)
- Compatible: AWS EC2 g5.xlarge (A10G), Azure NC6s v3, on-premises NVIDIA servers
- Docker Desktop (macOS / Windows): development and demonstration only, without GPU
Software prerequisites
docker >= 24.0
docker-compose-plugin >= 2.20
python3 >= 3.10
ffmpeg >= 4.4
openssl
# GPU
nvidia-driver >= 525
nvidia-container-toolkit >= 1.14Installation
Sentinel ships with a one-command installer that starts all Docker services, plus a manual step-by-step method. MediaMTX runs on the host for direct access to the network and GPU. After installation, sign in to the web console, change the initial password immediately, then configure system settings, recording storage and cameras. Installation packages and licence keys are provided by VedaCloud.ai — email connect@vedacloud.ai.
Adding cameras
Add cameras from the web console, register RTSP sources with MediaMTX for re-streaming, or use ONVIF auto-discovery through the VMS connector.