> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zylon.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# System Monitor

> Enable live infrastructure and model-serving metrics in Backoffice.

Enable the System Monitor page in Backoffice to inspect live host-resource and inference-serving metrics.

### Configuration

| Key | Default | Values | Description |
| - | - | - | - |
| `features.systemMonitor.enabled` | `false` | `true`, `false` | Shows or hides **Backoffice > Platform > System Monitor**. |

### Example

```yaml theme={null}
features:
  systemMonitor:
    enabled: true
```

Run `sudo zylon-cli sync` after changing the configuration.

When the setting is enabled, Backoffice shows the navigation item and starts the live metrics stream when the page opens. When it is disabled, the navigation item is hidden and a direct visit to the page shows an unavailable state.

<Note>
  This setting controls Backoffice availability; it does not start or stop the backend metrics recorder. It is separate from `observability.platformMetrics` and `observability.usageMetrics`, which configure external observability and product telemetry.
</Note>

The chart supplies the backend with the Kubernetes node and Triton metrics endpoints used by the monitor. If Triton metrics are unavailable, GPU and per-model information can be empty or show zero values. CPU and RAM use Kubernetes node metrics when available and otherwise fall back to host readings.

See [System Monitor](/en/operator-manual/backoffice/system-monitor) for the metrics shown in Backoffice.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.