Monitor kprobe

kprobe should be monitored like any production dependency.

Key metrics

MetricMeaning
agent events capturedNumber of events read from kernel buffers.
agent events droppedEvents lost before delivery.
publish latencyTime to publish events into the pipeline.
pipeline lagDelay between raw and enriched events.
storage write latencyTime to persist timelines and graph updates.
correlation ratePercentage of events with service or trace context.
query latencyAPI response time for timelines and graphs.

Alerts

Alert on:

  • agents missing from nodes
  • sustained event drops
  • high pipeline lag
  • storage write failures
  • correlation rate drop
  • API unavailable