The uptime percentage

A Monitor’s Uptime card shows its uptime percentage over the last 24 hours, 7 days, 30 days and 90 days.

How it is worked out

The percentage is worked out from the time it was down, over the time we were watching it. A Monitor watched for a whole day and down for part of it scores that part of the day as down, and the rest as up.

The time it was down is the time its Incidents were open. Only the Uptime Check’s Incidents count. A Performance warning, an expired certificate or an expired domain does not change the figure, and neither does a Regional outage: one location that cannot reach your target was never downtime. The Uptime card says it in one line: “Incidents count against uptime. Regional outages don’t.”

An Incident counts from the moment it opened to the moment it ended, however it ended. An Incident that was flapping counts whole, short recoveries inside it included, because it stayed open through them.

Never 100 while there was downtime

The figure has two decimals, and while there was any downtime in the window, it is never shown as 100. 20 seconds down in 30 days prints as 99.99%, and not 100%.

Hours not measured

Some hours we cannot speak for. An hour is not measured when a run was due in it and none reached us, or when your target failed and too few locations could answer to confirm it, with no run that passed. Our own outage lands here, and so does the time you were down during it.

A not-measured hour leaves both sides of the figure: it is neither up nor down, and the time your target was down in it is not counted either. So our own outage never shows as a good hour for you. Because the figure leaves those hours out, their count is shown beside the figure, on the Uptime card and on each day of the bar under it.

The Incidents and the hours not measured are kept for good, so the figure never changes when old runs are deleted. Runs, charts and how long we keep them says how long the rest is kept.

Time we were not watching

Time when the Monitor was paused, or Suspended, or before it was created, is not watched at all. It is on neither side of the figure. On the bar of the last days, a day like that is grey, and says why: Paused, Suspended or Not monitored yet. A window we did not watch at all has no figure, and the card shows “—” and Not monitored yet.

An Incident reported in error

When one of our locations was faulty, we take back the Incident it opened: it was reported in error, and you receive a correction. A retracted Incident leaves the figure: it was never downtime. Its day returns to up on the bar.

It is not hidden. It stays on the Monitor’s Events tab, and on your Status page, marked as reported in error, so the alert you received still has a record.