prometheus-go-code-review

Audit Prometheus metrics in Go by validating metric types (Counter, Gauge, Histogram) match measurement semantics, enforcing low-cardinality labels with proper unit suffixes, ensuring single registration, and confirming /metrics endpoint exposure. Use prometheus/client_golang to inspect definitions across services and adjust to follow best practices.

Common mistakes include using wrong metric types, high-cardinality labels causing memory issues, missing or incorrect unit suffixes, registering metrics multiple times causing panics, and misconfigured histogram buckets. This Skill identifies and fixes these patterns in prometheus/client_golang implementations.

Prevent race conditions by registering each metric exactly once during initialization, using a global registry or singleton pattern, and avoiding concurrent metric creation. This Skill ensures your Go service registers all prometheus/client_golang metrics safely before exposing /metrics.

Review histogram buckets to ensure they align with your latency or request-size distribution, avoid too many buckets causing cardinality explosion, and use sensible boundaries (e.g., milliseconds for latency). This Skill validates bucket choices across your Go instrumentation.

Yes, Prometheus instrumentation via prometheus/client_golang works across multiple Go services. This Skill audits metric consistency, label cardinality, naming conventions, and registration patterns to ensure reliability and observability best practices across your service fleet.