Settings every monitor has

These appear on every monitor type (except where noted). The fields specific to one type are on that type's own page.

Interval

How often the check runs, from 30 seconds up to once a day. The fastest you can pick depends on your plan:

Options faster than your plan allows are not shown, rather than shown and rejected.

Faster is not automatically better. It finds problems sooner and uses more of your monitor allowance of attention — a 30-second interval on something that blips weekly will tell you about every blip.

Timeout

How long one check waits before giving up. Choose from 5, 10, 15, 20, 30, 45 or 60 seconds.

A timeout is a failure, so this doubles as "how slow is too slow". Set it above your genuine worst case and below your patience. A page that normally answers in 300 ms does not need 60 seconds; giving it 60 just means an outage takes a minute longer to notice.

Failure threshold

How many consecutive failed checks open an incident. Default 2, counted across all regions in arrival order — see how monitoring works.

Regions

Where checks run from. New York is always on and cannot be removed. The others — New York, San Francisco, London, Frankfurt, Singapore, Bangalore, Sydney, Toronto, Amsterdam, Atlanta — depend on your plan; see the table in what UptimeCraft does.

Pick regions where your users are. Checking from Sydney a site that only serves Europe tells you about transpacific routing, not about your site.

Heartbeat monitors have no regions: nothing goes out to check.

Latency alerts

Tells you when responses get slow without calling the monitor down. Two fields: a threshold in milliseconds, and how many consecutive checks must exceed it.

It never changes the monitor's status and never opens an incident. A slow site is still up, and conflating the two means either paging for slowness or ignoring real outages. The consecutive-checks field matters here more than elsewhere — response times are noisy, and a single slow check is almost always nothing.

Not available on heartbeat monitors, which have no response time.

Pause and resume

Pausing stops checks. No results, no incidents, no alerts, and a gap in the chart for the period. A paused monitor still counts against your plan's monitor allowance.

For planned work, a maintenance window is usually the better tool: it keeps recording results so you can see what happened, and it starts and stops on its own.

Check now

Runs a check immediately without waiting for the interval. Useful after changing a setting, to confirm you fixed what you thought you fixed.

It is refused during a maintenance window.

Delete

Permanent, and takes the monitor's history with it. There is no undo and no archive. If you might want the history, pause instead.