Observability
Real-time monitoring, alerting, and diagnostics built into the platform.
Health Score
HyperSDK computes a system health score (0-100) based on:
| Metric | Weight | Source |
|---|---|---|
| CPU usage | 25% | /proc/stat |
| Memory usage | 25% | /proc/meminfo |
| Disk usage | 25% | /proc/diskstats + filesystem stats |
| Network health | 15% | /proc/net/dev |
| Service status | 10% | Process monitoring |
The score includes bottleneck detection — identifying the top contributing factor when health drops.
Smart Alerts
Configure threshold-based alerts for:
- CPU usage exceeding threshold
- Memory pressure and swap usage
- Disk space running low
- Network errors or packet loss
- Job failures
Alerts appear in the dashboard and are available via the /api/v1/system/alerts endpoint.
Explain Mode
When a metric is elevated, click Explain in the dashboard to see:
- Contributing factors ranked by impact
- Historical trend for the metric
- Recommended actions
Example: "CPU is at 87% — top contributors: migration job (62%), system indexing (15%), other processes (10%)"
Metrics
Dashboard Views
The dashboard provides 8 observability views:
| View | What It Shows |
|---|---|
| Health Overview | System health score with bottleneck indicator |
| Alerts | Active and historical alerts |
| Processes | Running processes with CPU/memory usage |
| Containers | Container status and resource consumption |
| Security | Failed logins, rate limit events, audit trail |
| Debug | Logs, request tracing, error analysis |
| Network | Interface stats, connection counts, bandwidth |
| Storage | Pool status, disk I/O, capacity planning |
Prometheus Export
curl -sk https://your-server:5080/api/v1/metrics
Returns metrics in Prometheus exposition format for integration with Grafana or other monitoring systems.
WebSocket
Real-time metrics stream via WebSocket:
const ws = new WebSocket('wss://your-server:5080/ws');
ws.onmessage = (event) => {
const data = JSON.parse(event.data);
// data.type: 'metrics', 'alert', 'job_update'
};
Carbon-Aware Scheduling
Schedule migration jobs during low-carbon grid periods:
curl -sk https://your-server:5080/api/v1/carbon/schedule
Returns optimal time windows based on real-time electricity grid carbon intensity data. Typical reduction: 30-50% CO2 per migration batch.
Schedule a Demo to see observability in action.