Grouping and throttling consolidates related alerts and limits notification frequency so one operational event produces one useful notification instead of a flood. In Alertmanager, it follows inhibition, silencing, and routing, then precedes delivery through receivers such as email, Slack, PagerDuty, or SNS. section 7.2.1
Use it to make alert delivery operationally manageable:
- Group alerts by a shared label such as env, keeping related firing alerts together. section 7.2.1
- Throttle repeated notifications so an ongoing event does not repeatedly interrupt responders. section 7.2.1
- Keep grouping distinct from inhibition and silencing: those suppress notifications, while grouping and throttling shape delivery for active events. section 7.2.1
Prometheus self-monitoring
A Prometheus rule fires when total HTTP requests exceed 40. Alertmanager groups alerts by env and sends the resulting notification to a webhook, providing one event-oriented delivery rather than repeated individual alerts. section 7.2.1
Listing 7.4 Alertmanager configuration
route: group_by: ["env"] #1 receiver: "mywebhook" #2 receivers: - name: "mywebhook" webhook_configs: #3 - url: "http://webhook:80/anything"
Inspect the Alertmanager configuration showing group_by and receiver selection. section 7.2.1
