System Health
Monitor live service status, RAM, CPU, and network metrics for your ragistry environment with automatic streaming updates.
Reading the System Health page
Reading the System Health page
The System Health page (sidebar → Health) gives you a real-time view of all backend services that power ragistry.
Live updates
The page connects via Server-Sent Events (SSE) to the backend and updates automatically every few seconds — no manual refresh needed. The current time of last update is shown as a clock badge in the page header.
If the connection drops, the page reconnects automatically after 5 seconds. A "Verbindung zum Server verloren" message appears while reconnecting.
Service status grid
The grid shows the status of each core service. A green indicator means the service is healthy; red means there is a problem.
Core services monitored:
- Chat — the conversation handling service
- Database — the data storage layer
- Embedding — the vector embedding service
- LLM — the language model inference service
System resource metrics
Three cards at the bottom show raw server metrics:
| Card | Metrics shown |
|---|---|
| RAM usage | Current utilisation % and MB used |
| CPU usage | Current utilisation % and number of threads |
| Network | Active connections count and status (Active / Inactive) |
Colour coding:
- 🟢 Green: healthy range
- 🟡 Yellow: caution
- 🔴 Red: high load
Time period selector
Use the dropdown in the page header to change the period shown in the service status history charts:
- Last hour (
1h) - Last 24 hours (
24h) ← default - Last 7 days (
7d) - Last 30 days (
30d)
What to do when a service shows unhealthy
What to do when a service shows unhealthy
If the System Health page shows a red indicator for any service, here is what it means and what to do.
| Unhealthy service | Impact | Recommended action |
|---|---|---|
| Chat | Users cannot start or continue conversations | Contact support immediately |
| Database | Data cannot be read or written; most features break | Contact support immediately |
| Embedding | New documents and pages cannot be indexed; existing conversations still work | Contact support if the issue persists more than a few minutes |
| LLM | The AI cannot generate responses; conversations fail | Contact support immediately |
Transient vs. persistent issues
A brief (< 1 minute) red indicator may be a temporary blip — the page reconnects and self-corrects automatically. If the indicator stays red for several minutes:
- Note the Last Updated timestamp in the page header.
- Take a screenshot of the Health page.
- Contact support at [email protected] with the screenshot and a description of any user-visible symptoms.
The dashboard "System" card
The home dashboard also shows a single System status card (Healthy / Problems). If it shows "Problems", navigate to Health for a per-service breakdown.