A high-performance NVR digesting dozens of multidimensional video streams generates incredible friction on local hardware. OpenNVR implements pure bare-metal monitoring logic to ensure your NVR engine doesn’t buckle under sudden throughput spikes or rapid GPU saturation.

Bare-Metal Network Telemetry

Network saturation is the leading cause of packet-loss and dropped camera chunks.

OpenNVR explicitly escapes its containerized boundaries to query routing statistics directly from your Host Linux environment. By opening the Infrastructure Analytics page, you can dynamically view real-time payload velocities passing through exact Network Interface Cards (NICs), allowing you to cleanly distinguish your local Camera LAN (eth1) traffic from your outbound Corporate LAN (eth0) traffic.

Transparent Stack-Traces

In legacy NVRs, diagnosing a configuration failure requires pulling massive, obfuscated text files via USB drives. OpenNVR pipes its native developer logs directly into the frontend UI.

• FastAPI Backend Traces: Instantly access HTTP 500 anomalies, failed adapter JSON unmarshaling, or failed WebRTC protocol bindings directly from the GUI. Look for the real-time log visualizer within the system health dashboard.
• MediaMTX Diagnostic Feeds: Monitor exact UDP datagram losses or camera RTSP authorization failures occurring natively inside the proxy layer.

VRAM & Hardware Saturation

If your edge node utilizes local AI Adapters (such as lightweight kai-c inferencing), ensuring you haven’t overdrawn the GPU’s memory allocation is paramount. The internal metric daemon periodically polls the kernel (e.g., nvidia-smi interfaces where applicable) and charts physical VRAM allocation explicitly inside the dashboard. This powerful telemetry proactively alerts you before you bind a massive model that exceeds your physical node’s constraints.