Response

Incident Conditions

When a monitor counts as down or degraded, and when that opens an incident. You set it per monitor on its Alerting tab.

Overview

Every healthcheck is in one of three states. Healthy when its checks pass. Down when enough regions confirm it failing: that always opens a critical incident. Degraded when some regions fail or it's slower than you expect: the monitor shows it, and it opens an incident only if you ask for that.

Each monitor sets this on its Alerting tab: how many regions must confirm an outage, what counts as slow, and whether degraded alerts. What happens once an incident opens, such as who is notified, where and how, is up to your workflows.

Conditions Create. Workflows Notify.

These settings decide when an incident opens. Workflows decide what happens next. You can change how you get notified without touching detection, and the other way around.

How a Failure Is Confirmed

A single failed check never opens an incident. When a check fails, the region that ran it rechecks it, 10 seconds apart: twice by default, or once to three times with Recheck a failure. Only a failure every run agrees on counts as a confirmed failure. The moment one region confirms a failure, Uptimy asks the monitor's other regions to check right away instead of waiting for their next scheduled run, so with the default rechecks an outage is confirmed within about a minute whatever the check interval. Fewer rechecks alert sooner; more rechecks mean fewer false alarms.

Browser (Playwright) checks run in one region and rerun once after a failure before it counts. Each run can take up to 2 minutes, so Recheck a failure is fixed at once for them.

Down

A monitor is down when enough of its regions confirm failing. By default that's a majority of the regions it runs in, so one region's network trouble can't look like an outage:

RegionsMajority
11
22
32
43
53

Change it with Confirm from: down as soon as one region confirms, or only when all of them do. Going down opens a critical incident, dated from when the failures began rather than when they were confirmed.

It's back up when enough regions succeed twice in a row again, so one lucky check doesn't end an outage. With a lower Confirm from, more regions have to recover: with "1 of 4", all four must be passing again. The incident resolves at the first of those successes.

Degraded

A monitor that isn't down is degraded when:

  • Some regions fail: fewer than needed for down confirm failing. A region whose failure hasn't been confirmed yet also shows the monitor as degraded, but never alerts.
  • It's slow: you set Slow above a number of milliseconds over a window of 1 minute to 1 hour, and the p95 response time of its successful checks across its regions in that window is above it. It needs at least three results in the window, so one slow check never counts, and failed checks (timeouts, errors) aren't response times. Slow above is off until you set it.
  • Its certificate is expiring: an SSL/TLS check passes with a warning while its certificate is inside the check's expiry warning window. One region's warning is enough, and the monitor is healthy again once no region's latest result carries one.

Alert When Degraded

Off by default. Turned on, becoming degraded opens a medium incident and runs your workflows, and the incident resolves when the monitor is healthy again.

If a degraded monitor goes down, it stays the same incident. It becomes critical, takes the outage's title, cause and regions, and runs the workflows that react to new incidents again. An outage pages the same way whether or not a degraded alert came first, and a workflow's Deduplicate step won't swallow it.

Route by Severity

Down is always critical and degraded is medium. Add a Filter node on severity to page your on-call for outages and send degraded alerts to a Slack channel.

Settings

SettingDefaultAPI field
Confirm fromA majority of its regionsconfirmRegions (null for a majority)
Recheck a failure2 timesrecheckAttempts (1 to 3)
Slow aboveOffslowThresholdMs (50 to 120,000; null for off)
Slow window5 minutesslowWindowMinutes (1 to 60)
Alert when degradedOffalertOnDegraded

The API fields belong to the healthcheck, so you can also set them with uptimyctl healthchecks upsert (see the CLI docs).

In Workflows

A trigger on Incident created runs for outages and, with degraded alerts on, for degraded monitors. A degraded incident's workflow can read why it was opened:

  • {{event.metadata.degraded.reason}}: regions, slow or warning
  • {{event.metadata.degraded.p95Ms}} and {{event.metadata.degraded.slowThresholdMs}}: the measured p95 and your threshold, for a slow one
  • {{event.metadata.degraded.warning}}: the check's warning, such as certificate expires in 12 days (2026-11-26), for a warning

Maintenance and Agent Monitors

While a maintenance window that lists the monitor is in progress, no incident opens.

Monitors run by an Uptimy Agent are decided by the agent itself: Cloud takes the status it reports, and these settings don't apply to them.