Get early access
Features · Alerting & Incidents · Maintenance Windows
Alerting & Incidents · Maintenance Windows

Suppress alerts on purpose,
not by muting and forgetting.

Planned maintenance, one-off windows, recurring schedules. Scoped to hosts, tags or groups. Active maintenance shows on the status page automatically. No "remember to unmute" risk.

scheduled
db-prod-01 tonight 02:00 · 2h
tag:staging recurring Sun 03:00
web-prod-04 active now · 38m left
alerts during suppressed, not lost
paused toggle, no edit dance
— Why this matters

Planned downtime should not page anyone.

01

Recurring means recurring.

Daily, weekly with a multi-day picker, or monthly — including overnight ranges like 22:00 to 02:00, which is when maintenance actually happens.

02

Pause without deleting.

A master on/off per window suppresses instantly and keeps the schedule. Deleting a window to skip one night is how schedules get lost.

— What's included

How a window behaves.

One-off or recurring
A start and end, or a pattern with an optional end date.
Overnight ranges
22:00 to 02:00 is a single window, not two.
Scoped
Tenant-wide or per server.
Suppression by type
Alerts are held during the window, not lost.
Active toggle
Independent of the schedule, switchable from the row without opening anything.
Honest badges
Active, Scheduled, Completed, Idle between occurrences, Paused.
— Works with

Connected capabilities

Smart Alerts
What gets suppressed.
→
Public Status Pages
Where active maintenance is shown to customers.
→
Auto-Recovery
Optional suppression target during maintenance.
→
Audit Log
Window history for compliance evidence.
→
— FAQ

Questions, answered honestly.

Can I open a window from the CLI? +
hostatlas maintenance open --scope tag:role:database --duration 1h --reason "vacuum".
What happens to an alert that fires after the window ends but during warm-up? +
Held back until warm-up ends. If still firing, page goes through.
Do auto-recovery actions still run? +
Configurable per window: usually yes, because a maintenance window is exactly when self-heal helps. Override for risky ops.

Plan downtime without losing track of it.

Maintenance windows on every plan; recurring schedules from Starter.

Get early access Back to Alerting & Incidents