Do not sum per-process rolling event gauges across replicas

A per-process rolling-event gauge is not a fleet counter. Summing it after discarding instance and replica labels can retain overlapping source series and cross a threshold before the corresponding application logs show that many new events. Preserve resource identity during diagnosis, query raw series at the evaluation timestamp, and alert from a monotonic counter with restart-safe rate/increase semantics or from a deliberately deduplicated fleet aggregation. Do not infer unique calls from the aggregate: the gauge measures retained process-local event timestamps, not durable event identity.

Applies when: Grafana alerts aggregate OpenTelemetry gauges emitted independently by replicated workers. Keywords: Grafana Prometheus OpenTelemetry rolling gauge replica stale series alert aggregation