Snooze, Acknowledge, and Resolve Flows
Who this is for
Anyone managing active alerts and wanting to understand how to properly handle a firing alert without just waiting for it to auto-resolve.
What you will complete
Use snooze to suppress an alert temporarily, acknowledge an alert to signal it is being handled, and understand when alerts auto-resolve vs. require manual action.
Before you begin
- At least one alert rule must be active and ideally one must be currently firing.
- Admin or Owner role to snooze and acknowledge alerts.
- Navigate to Alerts → Events tab.
The three alert states
| State | Indicator | Meaning |
|---|---|---|
| Firing | Red / destructive badge | The threshold condition is actively met right now |
| Snoozed | Amber / yellow badge | An admin suppressed notifications for a set period |
| Resolved | Green badge | The metric dropped back past the hysteresis threshold |
Snooze: suppress notifications temporarily
Use snooze when you know about the alert and need time to fix it without being repeatedly notified.
When to use snooze:
- You are in the middle of a maintenance task that is intentionally pushing CPU high
- You are waiting for a deployment to finish that is causing elevated metrics
- You want to suppress repeat notifications for a known issue overnight
Snooze options:
- 15 minutes
- 1 hour
- 4 hours
- 24 hours
Step-by-step: snooze an alert
- Go to Alerts → Events tab.
- Find the firing alert event you want to snooze.
- Click the Snooze button on the alert row.
- Select a snooze duration from the dropdown: 15 min, 1 hour, 4 hours, or 24 hours.
- The alert status changes to Snoozed and the badge turns amber.
- During the snooze period, no new notifications are sent for this alert.
- If the condition is still active when the snooze expires, the alert returns to Firing and notifications resume.
What snooze does not do: Snooze does not fix the underlying problem. The alert returns to Firing after the snooze period if the metric is still elevated.
Acknowledge: signal that someone is handling it
Acknowledge marks an alert as "seen and being worked on." This is useful for team coordination when multiple admins receive the same alert notification.
Step-by-step: acknowledge an alert
- Go to Alerts → Events tab.
- Find the firing alert.
- Click the Acknowledge (check/hand icon) button.
- The alert event is logged as acknowledged by your user at this timestamp.
- Other team members can see the alert has been acknowledged and by whom.
Acknowledging does not suppress further notifications. Combine acknowledge with snooze if you also want to silence repeat notifications.
Resolve: auto vs manual
Auto-resolve
Most alerts resolve automatically when the metric recovers past the hysteresis threshold. For example, if you have a CPU > 85% alert with the default 10% hysteresis, the alert auto-resolves when CPU drops below 75%.
You do not need to do anything for auto-resolve — the system handles it.
When auto-resolve does not happen
Some conditions do not self-resolve:
- Disk usage does not decrease unless files are deleted
- Load average may not recover without intervention (service restart or scaling)
For these cases, fix the underlying condition first, then the alert will auto-resolve once the metric recovers.
What success looks like
- After snoozed: the alert badge turns amber and the snooze duration is shown. No new notifications arrive during the snooze window.
- After the underlying condition is fixed: the metric drops, and within the next sampling interval (60 seconds), the alert transitions to Resolved and turns green.
- The Alert Events tab shows the full history: fired, snoozed (if applicable), resolved, with timestamps.
Common errors and fixes
"I snoozed the alert but it keeps sending notifications" Cause: You may have snoozed the alert in the dashboard but the notification channel has a separate configuration that sends on new events. Fix: Check that the snooze was applied correctly — the alert status should show "Snoozed" in the Events tab.
"Alert shows as Resolved but the metric is still high" Cause: The metric briefly dipped below the hysteresis threshold and the alert auto-resolved, but then climbed again — triggering a new alert event. Fix: Check the alert history. Multiple fire/resolve cycles with a short gap between them indicate flapping. Increase the resolve threshold or the alert duration. See KB-06-04.
"Alert shows as Firing but I fixed the issue 10 minutes ago" Cause: The metric has not yet dropped past the hysteresis threshold, or the next monitoring sample has not been processed yet. Fix: Wait 1–2 minutes for the next sampling cycle. If still Firing after 5 minutes, verify the underlying condition is truly resolved.