veza/veza-backend-api/internal/metrics/cache_hit_rate.go

package metrics

// Cache hit/miss counters per subsystem (v1.0.9 W3 Day 11).
//
// Three call-sites instrumented in v1.0.9:
//   - rate_limiter   — Redis INCR result classified as "hit" if the key
//                      already existed in the window (in-window request),
//                      "miss" if it was a new window (key just created).
//   - chat_pubsub    — "hit" on a successful Publish/Subscribe round-trip,
//                      "miss" on connection error (Redis unreachable).
//   - presence       — "hit" on a successful Get/Set/Del, "miss" on a key
//                      that didn't exist (presence stale or never set) or
//                      on an underlying Redis error.
//
// Subsystems are passed as labels rather than baked into separate metrics
// so dashboards can pivot. Cardinality is fixed at the three values above
// (plus future additions in W3+); never label by user_id / room_id /
// per-key — that would explode cardinality.

import (
	"github.com/prometheus/client_golang/prometheus"
	"github.com/prometheus/client_golang/prometheus/promauto"
)

var (
	cacheHits = promauto.NewCounterVec(
		prometheus.CounterOpts{
			Name: "veza_cache_hits_total",
			Help: "Total cache hits per subsystem",
		},
		[]string{"subsystem"},
	)

	cacheMisses = promauto.NewCounterVec(
		prometheus.CounterOpts{
			Name: "veza_cache_misses_total",
			Help: "Total cache misses per subsystem",
		},
		[]string{"subsystem"},
	)
)

// RecordCacheHit increments the hit counter for a subsystem. Subsystem
// must be one of the bounded set documented at file-level — adding a
// new value is a deliberate choice that should also update Grafana.
func RecordCacheHit(subsystem string) {
	cacheHits.WithLabelValues(subsystem).Inc()
}

// RecordCacheMiss increments the miss counter for a subsystem.
func RecordCacheMiss(subsystem string) {
	cacheMisses.WithLabelValues(subsystem).Inc()
}