Monitoring: maintenance windows – planned work without false alerts

Monitoring has a new Maintenance section. Plan maintenance windows for your hosts there – and planned work no longer opens incidents or sends notifications.

How it works

• A window applies to the hosts you select – group-inherited checks included.
• You set start and end as points in time. For each of the two you choose independently whether it takes effect automatically at the planned moment or only when you click start or stop yourself.
• While a window is active the checks keep measuring – they just open no incidents, and nobody gets paged.

Honest numbers. The measured curve is never rewritten: you can still see afterwards what really happened during the maintenance. The charts mark the span as maintenance, and the metrics (downtime, MTTR, incidents) exclude it. When the window ends and the service is still not back, reality applies again immediately: the next check opens the incident and alerts as usual.

Also new for incidents

• Incidents can now be classified: confirmed outage or false alarm, with an optional note. False alarms drop out of the metrics; the charts mark the span.
• Past incidents the probes could not capture – during an outage of your connectivity, say – can be recorded manually. Only manually recorded entries can be deleted again.

You will find it all in the console under Monitoring → Maintenance, and per host in the incidents tab. If something is missing or looks off, do send me a note.