Skip to content
numbrs docs

Relay Monitoring

Relay monitoring tracks the health and performance of any Nostr relay. numbrs connects to each relay on a 5-minute schedule and measures what actually matters — connection speed, event response time, and uptime over time.

Adding a relay

Section titled “Adding a relay”
  1. Click Relay Monitoring in the sidebar
  2. Click Add Relay
  3. Paste the WebSocket URL — e.g. wss://relay.example.com
  4. Optionally give it a friendly name
  5. Click Save

numbrs starts polling immediately. You’ll see the first data point within a minute.

What gets measured

Section titled “What gets measured”

Every check runs three measurements:

Connect latency — Time in milliseconds to complete the WebSocket handshake. This measures pure network and server overhead. High connect latency usually means server load or network routing issues.

Event latency — Time in milliseconds from sending a request to receiving the first EVENT response back. This is the number that matters most for client performance — it’s what users actually feel when they load their feed.

Uptime — Percentage of checks where the relay was reachable and responding correctly. Calculated over rolling 24-hour, 7-day, and 30-day windows.

When a check fails, numbrs records the error type — connection refused, timeout, TLS error, invalid response — so you can distinguish “server down” from “DNS broken” from “port blocked”.

Reading the relay detail page

Section titled “Reading the relay detail page”

On the relay detail page you’ll see:

  • Latency chart — connect and event latency over time. Spikes show problems; steady lines show health.
  • Uptime — percentages for each rolling window
  • Health score — a composite number based on recent uptime and latency. Higher is better.
  • Incident log — every outage with start time, end time, and duration
  • Check history — individual measurements with timestamps

Health score

Section titled “Health score”

The health score combines uptime and latency into a single number (0–100). A relay with 100% uptime but very high event latency scores lower than a fast relay. It’s a quick way to rank relays without reading the charts.

Setting up alerts

Section titled “Setting up alerts”

You can get notified when:

  • A relay goes down (first failed check)
  • Latency exceeds a threshold (e.g., event latency > 500ms)
  • Uptime drops below a percentage over a rolling window

Configure alerts from the relay detail page or from Alerts in the sidebar. See Alerts for details.

Notes on measurement accuracy

Section titled “Notes on measurement accuracy”

Checks run from numbrs’ servers, not from your location. The latency you see reflects the path from the check server to your relay — which may be different from what your clients experience.

If you’re in Europe and your relay is in Europe, the numbers will be representative. If your check server is in the US and your relay is in Asia, factor in the geographic distance.

For relays on Tor or other overlay networks, standard checks may not work as expected. Open an issue on GitHub if you need support for non-standard configurations.

Next steps

Section titled “Next steps”