Alertmanager Notification Grouping: Balancing group_wait and group_interval for Latency

Prometheus Alertmanager utilizes group_by labels to aggregate alerts and prevent notification storms. To manage the delivery flow, the group_wait timer determines how long to wait for additional alerts to join a group before sending the first notification, while group_interval controls the wait time before sending a notification about new alerts added to an existing group.

In high-cardinality environments, there is a tension between reducing noise and maintaining low latency for critical alerts. A long group_wait reduces the number of initial messages but delays the first alert notification. Conversely, a short group_wait may lead to fragmented notifications if related alerts fire seconds apart.

Configuration Uncertainty

When managing multiple alert severity levels within the same grouping logic, it is unclear how to optimize these timers to ensure critical alerts bypass the wait period without breaking the aggregation of lower-priority noise.

• Does the current Alertmanager architecture support dynamic group_wait values based on alert labels?
• What is the recommended strategy for separating high-priority alerts from grouped low-priority alerts to avoid notification latency?