Monitoring & Metrics
VCAL exposes native Prometheus metrics at /metrics.
Key Metrics
| Metric | Type | Description |
|---|---|---|
vcal_cache_hits_total | Counter | Cache Hit Ratio (%) |
vcal_tokens_saved_total | Counter | Tokens Saved (range) |
vcal_tokens_saved_total | Counter | Cost Savings (range) |
vcal_answers_cached | Gauge | Answers Cached |
vcal_cache_hits_total | Gauge | Requests (range) |
vcal_search_errors_total, vcal_batch_search_errors_total, vcal_insert_errors_total, vcal_upsert_errors_total, vcal_delete_errors_total | Counter | Errors (range) |
vcal_search_latency_seconds_bucket | Gauge | Server Search Latency (p50 / p95) |
vcal_evictions_total | Gauge | TTL Evictions (10m) |
vcal_evictions_total | Gauge | LRU Evictions (10m) |
Example Grafana Dashboard
-
Add Prometheus datasource:
URL: http://vcal-server:8080/metrics -
Import the provided dashboard JSON from
deploy/grafana/vcal-dashboard.json(request the file from VCAL Server Team). -
Visualize:
- Cache hit ratio
- Tokens saved (range)
- Costs saved (range)
- Answers cached
- Requests
- Errors
- Server Search Latency (p55 / p95)
- TTL Evictions
- LRU Evictions