Analytics — System Health
Navigate to: Dashboard → System Health
What it shows: Real-time health of every configured LLM provider, the system process, and gateway routes. Flags degraded or unreachable providers before they affect users.
Provider health states:
| State | Meaning |
|---|---|
| Healthy | Error rate < 10 %, at least one enabled key |
| Warning | Error rate 10–50 % |
| Degraded | Error rate > 50 % |
| No Keys | Provider configured but no API keys added |
| All Disabled | Keys exist but all are toggled off |
| Unhealthy | Cannot reach provider endpoint |
What the page shows:
- System uptime, process memory (RSS), hostname
- Per-provider cards with: key count, error rate, last request timestamp, circuit breaker state
- Gateway route health table (route name, status, avg latency, requests/min, uptime %)
Step-by-step:
- Click System Health in the sidebar.
- Review the provider cards — any card in Warning or Degraded state needs attention.
- For a degraded provider, click Check Now on the provider card to trigger an on-demand health ping.
- If a provider shows No Keys, navigate to API Key Management and add at least one key.
- For circuit breaker state Open, the provider is being bypassed. It will auto-recover after the cooldown window (default 60 s).
How to test:
- Disable all API keys for one provider in API Key Management.
- Reload System Health — that provider’s card should show All Disabled.
- Re-enable a key and click Check Now — status should return to Healthy.
API endpoints:
GET /health— full system healthGET /admin/gateway/health— gateway route healthPOST /admin/gateway/routes/:id/health/check— trigger on-demand check