Checkmate
DocumentationUptime monitoring

Status window

Control how many failed checks it takes before a monitor is marked down.

A monitor's status does not flip on a single failed check. Checkmate evaluates a sliding window of recent checks and only changes the status when enough of them agree. This is what stops a brief network blip from raising an incident and it is the first thing to adjust if a monitor flaps.

Settings

Both fields are in the Incidents section of the monitor's configure screen.

FieldRangeDefault
Number of checks in sliding window1-205
Percentage threshold1-10060

The threshold answers: what percentage of checks in the window must fail (or succeed) before the status changes?

How it works

With the defaults, a window of 5 and a threshold of 60%, a monitor goes down once 3 of the last 5 checks have failed and comes back up once 60% of the window is passing again.

statusWindow: [up, up, down, down, down]   ← 3 of 5 failed = 60%
                                              monitor goes down

A single failure in an otherwise healthy window changes nothing.

Tuning

  • Flapping monitors: raise the window size, the threshold or both. A window of 10 at 60% needs six consecutive-ish failures and tolerates transient errors much better.
  • Faster alerting: lower both. A window of 1 at 100% marks the monitor down on the first failed check, with no smoothing at all.
  • Remember the window is measured in checks, not time. A window of 5 on a 1-minute interval spans about 5 minutes; the same window on a 1-hour interval spans 5 hours.
Note

Heartbeat monitors ignore these settings and always use a window of 1 at 100%, because a missed heartbeat is already unambiguous.

See also

On this page

Was this page helpful?