Observability
MuninnDB ships with full observability: Prometheus metrics, structured JSON logging, health/readiness endpoints, and Go pprof profiling.
Health Endpoints
| Endpoint | Description |
|---|---|
| GET /health | Overall health: all workers running, storage reachable |
| GET /ready | Readiness check: ready to accept traffic |
| GET /metrics | Prometheus metrics scrape endpoint |
| GET /debug/pprof/ | Go pprof profiling (CPU, heap, goroutines) |
Prometheus Metrics
Key metrics exported at /metrics:
muninn_activate_duration_ms— Activation query latency histogrammuninn_store_duration_ms— Write latency histogrammuninn_engrams_total— Total engrams by statemuninn_decay_worker_last_run— Last decay run timestampmuninn_hebbian_updates_total— Hebbian weight updates countmuninn_cache_hits_total— L1 cache hit/miss ratiomuninn_triggers_fired_total— Semantic trigger events
Structured Logging
All log output is structured JSON via Go's stdlib slog. Pipe to any log aggregator (Loki, CloudWatch, Datadog).
{"time":"2026-02-21T10:00:00Z","level":"INFO","msg":"activate","vault":"default","duration_ms":12,"results":5,"context_len":24}