Monitor kprobe
kprobe should be monitored like any production dependency.
Key metrics
| Metric | Meaning |
|---|---|
| agent events captured | Number of events read from kernel buffers. |
| agent events dropped | Events lost before delivery. |
| publish latency | Time to publish events into the pipeline. |
| pipeline lag | Delay between raw and enriched events. |
| storage write latency | Time to persist timelines and graph updates. |
| correlation rate | Percentage of events with service or trace context. |
| query latency | API 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