Monitoring

​

Forward DNS monitoring

• latency: last, average, p95, a sparkline,
• packet loss,
• share of queries it received,
• errors by kind: request timeout (no answer in time), unreachable (connection refused or no route), and TLS (a failed handshake or certificate, often something intercepting the connection to a DoT or DoH upstream).

• a warning banner appears on every page,
• a toast pops up when it goes down and when it recovers,
• an entry is added to the incident log (the bell in the top bar), kept for 90 days.

​

Traffic alerts

​

Daily summary

​

Getting alerts

dnz notify alerts --nxdomain-percent 40 --rate-limited 500 --summary-hour 7
dnz notify incidents        # upstream outages and recoveries
dnz stats --range 30d       # the dashboard's traffic summary in the terminal