
- Planned maintenance windows (patching, scaling, configuration changes).
- Controlled testing or deployments where known alerts will fire.
- Scheduled jobs or transient conditions that you do not want to notify on.

- Open the Silences section in Alertmanager (or your alerting UI).
- Click “New Silence”.
- Provide:
- A start time and duration, or an explicit end time.
- One or more matchers (labels) that select which alerts will be silenced.
- The creator/author name and an optional comment describing the reason.
- Save the silence.
Example: silence all alerts from the
ec2 job
=~ syntax if supported, e.g. job=~"ec2|aws".)
Silences suppress notifications—they do not stop alert rules from being evaluated or recorded. Alerts will still fire and be visible in the Alertmanager UI and Prometheus; only notifications (emails, Slack messages, etc.) are suppressed while the silence is active.
- The Silences tab lists all configured silences with their matchers, creator, comments, start/end times, and current status.
- Use this list to verify active silences, extend durations, edit matchers, or delete silences when maintenance completes.
- Consider tagging silences with clear comments and creator information so on-call engineers can quickly understand why a silence exists.
Be careful with broad matchers (for example, matching on a high-level label like
environment="production"). Broad silences can suppress many unrelated alerts and may hide critical issues. Prefer the narrowest set of matchers that describe the intended scope.- Alertmanager Silences documentation: https://prometheus.io/docs/alerting/latest/alertmanager/#silences
- Alertmanager matchers and syntax: https://prometheus.io/docs/alerting/latest/alertmanager/#matching-using-labels