Reported in error

Now and then, an Incident should never have opened. One of the locations that agreed it failed is later found to be faulty, and without that location the others do not agree. Then we take the Incident back: it was reported in error.

The same can happen to a Regional outage, when the one location that could not reach your address turns out to be at fault itself.

What the dashboard shows

A taken-back Incident stays in the dashboard, marked so you can tell it apart:

• Its card at the top of the Monitor’s page says Retracted and “Reported in error.”
• On the Events tab, its row is muted, its badge says retracted, and it carries one line: “Reported in error: one of our locations was faulty, and the others did not agree.”
• Under How uup decided, its last step says “Taken back later: one of our locations was faulty”.

A taken-back Regional outage carries its own line, naming the location: “Reported in error: the fault was in our own location in Frankfurt.”

Nothing hides it, because a switch that hid it would hide real outages too. You can still dismiss its card, as for any ended Incident.

The correction you receive

In place of a recovery message, we send a correction, to the same places the alert went. For a Monitor named Acme API, its title is “Correction: Acme API was not down”. It says that the alert was wrong, that the failure was seen through one of our locations that has since been found faulty, and that it will not count against your uptime. It ends: “You will get no recovery message for it.”

For a Regional outage, the correction is titled “Correction: Acme API could be reached from Frankfurt”, and says your uptime was untouched.

A Webhook receives them as the correction and regional_correction messages.

Your uptime

A taken-back Incident leaves your uptime: the time it covered counts as up, as if it had never opened, and that day’s bar is up again. It never turns that time into not measured, because the other locations’ runs still stand.

A Regional outage never counted against your uptime, so taking one back changes nothing there.

Your Status page

The Incident stays in your Status page’s Incident history, tagged Retracted and struck through, and its entry reads:

Acme API was reported as down in error.

Our monitoring was at fault. The service was working.

That day’s bar on the Status page is up again. A Regional outage never appears on a Status page, so there is nothing to change there.