Alert rule cooldown isolation across tag scopes
0 reputation · 18 Aug 2023, 18:03 UTC
The /api/alerts/rules API provides threshold-based alert rules for bucket memory quota, disk usage, and XDCR replication lag, each configurable with cooldown intervals to consolidate notifications and limit noise. Rule-level tags such as node_label and bucket_name allow scoping alerts to specific cluster components rather than broadcasting every threshold breach across the entire team. However, documentation confirms that cooldown and silencing configurations are applied globally per alert type across Couchbase Server versions 6.x through 8.x, meaning a single cooldown window suppresses all related alerts irrespective of rule-level tag differentiation. This design can suppress emerging issue notifications when multiple rules of the same type operate concurrently.
Can cooldown windows be scoped to individual alert rules or tag combinations without affecting other rules of the same type? Is there a mechanism to silence notifications for specific rule instances while preserving cooldown behavior for others? How does the global cooldown interact when rules with overlapping thresholds and different tag scopes are active simultaneously?