# What's the difference between snoozing, acknowledging, and maintenance periods?

Oh Dear gives you three ways to silence notifications, each built for a different situation. In all three cases we keep running your checks in the background - you just decide what you want to hear about.

## Acknowledging: "I know, I'm on it"

Acknowledging is a *reaction* to an active incident. Your uptime check fails, the alert comes in, and you're already working on a fix. You don't need a reminder every few minutes.

- Works **per check** (e.g. only the uptime check, while certificate alerts stay active).
- Silences the repeated **problem** notifications for that incident.
- The **recovery notification still comes through**, and it ends the acknowledgement. If the same check fails again five minutes later, that's a new incident, and you'll be alerted right away.
- You pick a duration as a safety net: alerting resumes when the check recovers or when the timer runs out, whichever comes first.

You'll find the *Acknowledge* button on the check page whenever a check is failing, and you can acknowledge straight from the alert itself: the e-mail contains one-click links, and Slack and Telegram messages carry an acknowledge action.

## Snoozing: "keep quiet for a while"

Snoozing is *preventive*. The check is healthy, but you know noise is coming - a planned migration, a flaky third-party API, a DNS change you're rolling out.

- Works **per check**, just like acknowledging.
- It's the **full mute**: both the **problem** and the **recovery** notifications stay silent for the duration you pick. You muted it, so you own the quiet window, nothing at all comes through until the timer runs out.
- That's the difference from an acknowledgement: you acknowledge a problem you already know about (so you still want the "all clear"), but you snooze a check pre-emptively (so you don't want to hear from it at all).

The same button reads *Snooze* while a check is healthy and *Acknowledge* while it's failing, and the behaviour follows the label. Scheduled task (cron) checks have no recovery moment either way, so their silences always run until the timer expires.

## Maintenance periods: "mute the whole monitor"

A maintenance period is the heavyweight option for planned work on an entire site or server.

- Works **per monitor**: every check on that monitor goes quiet at once.
- Silences **both problem and recovery** notifications - nothing comes through until the window ends.
- Downtime during a maintenance period is **not recorded**, so your uptime numbers and reports stay clean.
- Can be scheduled ahead of time, one-off or recurring (think weekly deploy windows).

## Which one should you use?

| | Scope | Silences | Recovery notification | Affects uptime stats |
|---|---|---|---|---|
| **Acknowledge** | One check | Repeated problem alerts for the current incident | Still sent, ends the acknowledgement | No |
| **Snooze** | One check | Everything on that check | Muted until the timer runs out | No |
| **Maintenance period** | Whole monitor | Everything on every check | Muted | Downtime isn't recorded |

Rule of thumb: something broke and you're on it? Acknowledge. Expecting noise on one check? Snooze. Taking the whole site down on purpose? Plan a maintenance period.

For the step-by-step on silencing a single check, see [Can I snooze a single check on a monitor?](/docs/faq/can-i-snooze-a-single-check-on-a-monitor). Full docs: [How to acknowledge or snooze notifications](/docs/notifications/how-to-acknowledge-or-snooze-notifications).