Maintenance windows

Planned work — a deploy, a database migration, a provider upgrade — will make your checks fail. Without a heads-up, UptimeStatus would open incidents, page on-call, and dock your uptime percentage for work you scheduled on purpose. A maintenance window tells UptimeStatus "this is expected."

What a window does

For the monitors attached to it, during the scheduled window:

No paging. A failing check inside the window does not open an incident or fire notifications.
No uptime dent. Time inside the window is excluded from the uptime percentage and the status-page uptime-history bars, so a planned outage never shows as downtime.
Public "under maintenance" state. Any status page that includes an affected component displays an under maintenance banner for the duration, so your users know it's intentional.

When the window ends, normal monitoring resumes automatically. If a monitor is still failing after the window closes, it opens an incident as usual.

Creating a window

In the dashboard, go to Maintenance and click Schedule window.
Set the start and end times (and a timezone). One-off windows cover a single planned change. For regular work like weekly reboots, add a recurrence as a cron expression (for example 0 2 * * 0 for every Sunday at 02:00 UTC, or @daily); each occurrence keeps the same duration as the start-to-end you set here.
Add a short description — this is the message shown on the status page (e.g. "Scheduled database upgrade, expect ~10 min of downtime").
Attach the monitors the work affects. Only attached monitors are silenced; everything else keeps alerting normally, so an unrelated outage during your maintenance window still pages.
Save. The window is now active on schedule.

Tips

Attach the smallest set of monitors that the work actually touches. Over-attaching hides real, unrelated outages.
Pad the window slightly on both ends — start it a few minutes before the change and end it a few minutes after, so a slow rollout or a lingering cache doesn't page you on the boundary.
Publish the description on customer-facing pages ahead of time. Subscribers to a status page that shows an affected monitor are emailed automatically when a window is within about a day away, so they hear about planned work before it starts. Recurring windows notify subscribers ahead of each occurrence.