Telescope

Statuses and alert timing

What up, down, and degraded mean, and why an alert fires only after several failed checks in a row instead of the first one.

Two things surprise people most: what degraded means, and why a brief outage didn't page anyone. Both come down to how Telescope classifies a check and when it decides to alert.

What the statuses mean

Every monitor is always in one of three states:

StatusMeaning
UpThe check passed — your service is healthy.
DegradedThe service is reachable but the check couldn't complete cleanly — most often a firewall (WAF) blocking the checker, or an SSL/domain that's valid but nearing expiry. Worth a look, not an outage.
DownThe check failed — an error status, a timeout, an unreachable host, or a failed assertion.

If a monitor is degraded because of a firewall, see Monitor down but site up. For SSL and domain degraded states, see SSL and domain monitors.

Why my site was down but I got no alert

Why it happens. Telescope doesn't alert on the first failed check. Each monitor has a failure threshold — the number of consecutive failed checks required before it's considered down and an alert is created. This is deliberate: it filters out single transient blips so one flaky check doesn't wake you at 3am.

The default threshold is 3 consecutive failures. So a service that fails once or twice and recovers never crosses the line — and never alerts.

CheckResultConsecutive failuresAlert?
1Down1No — below threshold
2Down2No — below threshold
3Down3Yes — alert created
4Down4No — already alerted
5Up0 (reset)Recovery alert

How to fix. Tune the threshold to the service:

  • For a critical, must-know-instantly service, drop it to 1 so the first failure alerts.
  • For a flaky endpoint you don't want to hear about on every hiccup, raise it.

Detection time follows your interval

The threshold is counted in checks, not minutes. With a 3-failure threshold and a 60-second interval, a real outage takes about three checks — a few minutes — to confirm before it alerts. A shorter interval confirms faster.

One alert per outage

Once an alert fires for a monitor, Telescope won't create another for the same ongoing issue — a site down for an hour is one alert, not one per check. When the monitor recovers, its active alerts are resolved automatically, a recovery notification is sent, and the failure counter resets to zero. A new alert is only created if it goes down again afterward.

See Alerts for the full lifecycle and Notification channels for where alerts are delivered.

On this page