Maintenance windows
Schedule a window before a deploy or planned outage and ViewPeek will stop emailing / Slacking / Discord-ing alerts for the duration. The agents keep checking and recording results, so your uptime numbers stay accurate and your public status page can show "Scheduled maintenance" instead of red squares.
What it is
A maintenance window is a time-bounded alert mute. It has:
- A name (shows in the list and audit log).
- An optional reason (free text, shows on the public status page).
- A start and end time (UTC — pick the absolute window, not your local clock).
- A scope — either a specific list of monitors or every monitor in your org.
During the window: checks run normally, results are written to the database, the dashboard shows live status. Email + Slack + Discord + webhook delivery is suppressed. When the window ends, alerting resumes automatically — no manual unmute.
Schedule one
- Go to Maintenance in the topbar.
- Click Schedule window.
- Name it (e.g.
API deploy — 2026-06-12) and optionally add a reason. - Pick start + end in UTC. Maximum window length is 30 days.
- Tick the specific monitors affected, or check Apply to every monitor in the org if you really want a blanket mute.
- Save. The window is live the moment the clock crosses the start time.
Scope: one monitor or the whole org
The default and recommended scope is per-monitor: tick the boxes for the specific monitors that will be affected by your maintenance. This way, if something unrelated breaks at the same time, you still get alerts on it.
The org-wide option exists for the rare cases when you really do want everything muted — e.g. a full-data-center maintenance affecting every endpoint. Use it sparingly; it's a "I know what I'm doing" switch.
Maintenance vs pause
| Maintenance window | Pause | |
|---|---|---|
| Checks run? | Yes — data preserved | No — agent skips the monitor |
| Alerts fire? | No | No |
| Uptime % during it? | Counts (down checks reduce uptime) | Doesn't count — no rows written |
| Best for | Planned work where you expect blips but want SLA accuracy | Long-term monitor retirement or "stop touching this for now" |
| Scope | One monitor, several monitors, or whole org | One monitor at a time |
Rule of thumb: maintenance for deploys, pause for "shouldn't be running right now".
On the status page
If you have a public status page set up and the maintenance window's monitors are exposed, the page will reflect the window — visitors see the maintenance reason instead of a red outage square. (This is the killer feature: planned work shouldn't make you look unreliable.)
Cancel early
If the deploy finishes faster than expected, hit Cancel on the row in the maintenance list — alerts resume immediately. Cancelled windows stay in the list (with a Cancelled badge) so the audit trail is preserved; click Delete to remove them entirely.
Next steps
- Schedule your first window.
- Pause individual monitors when you want to stop checking entirely.
- Set up a public status page so customers see scheduled maintenance instead of guessing.