What a maintenance window is

View as Markdown

A maintenance window is a scheduled time range in which you tell HostTracker “this downtime is expected” for one or more monitors. Use it for planned deploys, server migrations, certificate swaps, provider maintenance - any change you know will make a monitor look down.

Why you would use one

Without a window, a planned deploy that takes your site offline for a few minutes looks exactly like a real outage: your team is alerted, and the downtime lowers the monitor’s uptime percentage and its SLA figures. A window tells HostTracker in advance, so for each covered monitor it can:

Hold back alerts - no down notification goes out while the window runs.
Keep the downtime out of statistics - the time is counted as maintenance instead of downtime, so uptime percentages and reports stay honest about unplanned outages.
Tell your visitors - the window shows on your status pages as planned maintenance instead of an unexplained outage.

You choose alerts, statistics or both per monitor, so one window can silence a web server completely while only holding back alerts for a dependent API whose uptime you still want measured.

What a window is made of

Part	What it means
Name	A short description you recognise later (“Database migration”).
Schedule	One time (a start and an end) or Weekly (chosen days of the week, the same time of day each time).
Time zone	The wall clock the times are read in. Weekly windows keep their local time across daylight-saving changes.
Monitors	The monitors the window covers, each with its own Alerts and Stats choice.
Show on status page	Whether visitors to your status pages see it (on by default).
Active / paused	A paused window stays saved but does nothing.

Every setting, its default and its API field: Create a maintenance window.

What a window does not do

It does not stop checking. Checks run on schedule and every result is recorded in the check log; a window only changes what those results trigger and how they count.
It does not hide incidents. A down episode during a window is still recorded, marked as under maintenance.
It does not extend itself. When the end time passes, suppression stops immediately; if the monitor is still down, one down alert goes out then. Extend the window before it ends if the work runs long.
It does not follow tags. A window covers the monitors you list when you save it - there is no “every monitor tagged X” option. To cover a tag, list the tagged monitors and add their ids; monitors tagged later are not included until you update the window. Recipe: Cover every monitor with a tag.

Full detail: What a maintenance window suppresses.

Plans and limits

Maintenance windows work on every plan. There is no cap on how many windows you have, how long they last (from 60 seconds up to years) or how many monitors one covers.

Ways to manage windows

In the app: Maintenance in the sidebar (/maintenance).
Through the REST API: GET/POST /maintenance, GET/PATCH/DELETE /maintenance/{id}, with the monitor:read / monitor:write scopes.
From an AI assistant through MCP: list_maintenance, create_maintenance, update_maintenance, delete_maintenance.
As code with the Terraform provider (hosttracker_maintenance).

Maintenance windows vs status page maintenance

Two different things share the word “maintenance”:

A maintenance window (this section) covers monitors and changes alerting and statistics. With Show on status page on, it also appears on your status pages automatically.
A scheduled maintenance declared on a status page is an announcement you write on one page for its visitors and subscribers. It does not change alerting or statistics. See Post announcements and incident updates.

Use a window when monitors will really go down; add a status page announcement when you also want to send your own words to the page’s subscribers.