Introduction
Understand maintenance windows, types, scope, activation logic, and best practices.
Maintenance windows allow you to suppress noise and communicate planned impact. They can target all monitors or a subset, and operate as one‑time or recurring schedules.
Types
Type | Description | Use Case |
---|---|---|
one-time | Single scheduled window with concrete start & end timestamps | Infrastructure migration |
recurring | Repeats weekly on selected days + start/end times (UTC) | Nightly patch window |
When a Maintenance Window Counts as Active
The system decides whether a maintenance window is active using these rules:
- One‑Time: Active if
start < now < end
. - Recurring: Active if today’s weekday in
daysOfWeek
AND current time betweenstartTime
andendTime
(UTC comparison pattern).
Expired one‑time maintenances older than a threshold are hidden from some listings to reduce clutter.
Scope
Field | Effect |
---|---|
allMonitors | If true, entire instance is covered; counts derive from total monitor count. |
monitorIds | Explicit subset if allMonitors is false. |
enabled | Toggle to activate/deactivate without deleting configuration. |
Counting Affected Monitors
If any maintenance has allMonitors: true
, affected count equals current monitor total. Otherwise, distinct union of all enabled maintenance monitorIds
is used.
Filtering & Search
List queries support:
- Text search (title / description) via full text index.
- Monitor filter (returns those affecting selected monitors or set to all).
- Enabled filter (active toggle state).
Best Practices
- Prefer recurring windows for consistent maintenance policies; reduces manual scheduling overhead.
- Keep titles user-friendly (“Planned Database Upgrade – EU Central”).
- Disable (rather than delete) windows frequently reused seasonally.
Incident vs Maintenance
Scenario | Use Maintenance? | Use Incident? |
---|---|---|
Planned upgrade | Yes | No |
Unexpected outage | No | Yes |
Emergency security patch | Often (if possible) | Possibly (if user impact occurs) |
Accurate classification improves reliability analytics: planned work shouldn’t dilute incident MTTR.