Start a window

Open the machine and choose Start maintenance. Choose 15 minutes, one hour or a custom whole-minute duration from 1 to 10,080 minutes (seven days). Optionally add a short reason your team can read. The window starts immediately using server time.

Owners and operators can start, change or end maintenance. Viewers can see it. Future schedules, recurring windows and bulk maintenance are not included in the reviewed release.

What continues

Check-ins, actual connectivity, resource readings, service observations and incident history continue. A maintenance badge appears alongside the true state. An offline device stays visibly offline; maintenance does not make it appear healthy.

No agent update, service stop or re-enrollment is required. Leave the collector running so the team can confirm recovery during the work.

Notification behavior

New and queued submissions wait while the window is active. Existing queue records and attempt history are retained. Recovery cancels pending messages normally. Muted, canceled, failed, accepted or uncertain submissions are not resurrected by maintenance.

An email already in submission may finish. Starting maintenance cannot recall an accepted message, so start the window before disruptive work.

End and verify

Choose End now, adjust the duration or let the window expire. When it ends, only unresolved eligible problems resume notification. Resource and service mail waits for fresh readings. Offline follow-up timing waits thirty minutes after the window ends, including when you end it early.

A delayed-only offline policy still waits its delay; maintenance does not invent an immediate message. The server workers manage expiry even when no browser is open. An application outage can delay that evaluation until the workers recover.

After work, check the latest accepted check-in, current observations and incident recovery. Review the recorded start/change/end/expiry events. An incident closed by changing a rule is different from a measured recovery.