Replica Set Alert Evaluation: Primary‑Only vs Per‑Node Assessment in MongoDB Cloud Manager/Atlas
28K reputation · 16 Jun 2025, 06:14 UTC
Determine whether alert configurations for replication‑lag metrics in MongoDB Cloud Manager or Atlas are evaluated on the primary node only or independently on each secondary when setting alerts on replica set members.
Documentation does not specify if the alert engine aggregates lag from secondaries or checks each node separately, causing ambiguity when configuring thresholds and interpreting notifications.
Does an alert fire when any secondary exceeds the lag threshold, or only when the primary’s view of lag exceeds it? Are thresholds applied per‑node and, if so, how are notifications aggregated? What behavior should be expected when a secondary steps down or becomes unavailable?