Task · TSO-0126

Alert on sustained processor queue drops

Description

The health dashboard visualizes tailscale2otel.processor.dropped by signal and reason, but no rule watches confirmed record or span loss after queue saturation. Operators can respond by correcting capacity or deliberate load-shedding policy.

Acceptance Criteria

Definition of Done

Implementation Notes

Done in 7841543f. ts2o-processor-queue-drops: sum by (signal, reason) (rate(tailscale2otel_processor_dropped_total[10m])) > 0 for 10m, advisory, non-paging, policy optional. Runbook otlp-export-health, because the loss is upstream of the exporter and no export-failure rule can see it. Panel ‘Processor queue drops/s by signal & reason’. The fixture drives two series at once and asserts two separate alerts, pinning the per-signal-and-reason split: deliberate load-shedding and a saturated queue need different answers and collapsing them would hide which is happening. Negative-tested individually. Live-verified in the same push as TSO-0124.

View the source file on GitHub