# Aggregation observability

Purpose
- Track Link-Not-Merge aggregation and overlay pipelines.

Metrics
- aggregation_ingest_latency_seconds{tenant,source,status}
- aggregation_conflict_total{tenant,advisory,product,reason}
- aggregation_overlay_cache_hits_total, aggregation_overlay_cache_misses_total
- aggregation_vex_gate_total{tenant,status}
- aggregation_queue_depth{tenant}

Traces
- Span: aggregation.process
- Attributes: tenant, advisory, product, vex_status, source_kind, overlay_version, cache_hit

Logs
- tenant, advisory, product, vex_status
- decision (merged, suppressed, dropped)
- reason, duration_ms, trace_id

SLOs
- Ingest latency p95 < 500ms per statement.
- Overlay cache hit rate > 80%.
- Error rate < 0.1% over 10 minutes.

Alerts
- HighConflictRate: aggregation_conflict_total delta > 100 per minute.
- QueueBacklog: aggregation_queue_depth > 10k for 5 minutes.
- LowCacheHit: cache hit rate < 60% for 10 minutes.

Offline posture
- Export metrics to local Prometheus scrape.
- Deterministic ordering preserved; cache warmers seeded from bundled fixtures.