---

name: rust-observability description: "Rust observability skill for tracing, metrics, structured logging, OpenTelemetry pipelines, and production incident diagnostics."---

Rust Observability Skill

Core Question

Which telemetry signals reduce mean time to detect and resolve production failures?

Signal Model

• Logs: contextual events and state snapshots.
• Traces: cross-service causal path and latency breakdown.
• Metrics: rate/error/duration + saturation and capacity.

Use all three intentionally, not interchangeably.

Instrumentation Priorities

1. Request entry/exit and key business operations.
2. Downstream calls (DB, cache, RPC, queue).
3. Retry/timeouts and circuit-breaker transitions.
4. Queue depth and worker saturation.

Logging Rules

• Structured logs only.
• No secrets/tokens/PII leakage.
• Stable field names for dashboards and alerts.

Metrics Rules

• Keep label cardinality bounded.
• Prefer RED + USE style baseline metrics.
• Track queue/worker/resource saturation explicitly.

Common Pitfalls

• High-cardinality labels (raw user/session IDs).
• Verbose debug logs on hot paths in production.
• Traces without propagated correlation context.
• Alert thresholds not aligned with SLOs.

Review Checklist

• Critical path spans and error fields exist.
• Metrics cover latency/errors/saturation.
• Correlation IDs flow across service boundaries.
• Logging policy prevents sensitive data exposure.
• Alerts map to actionable runbooks.

Verification Commands

cargo check
cargo test
cargo clippy
RUST_LOG=info cargo run

Related Skills

• rust-web

• rust-database

• rust-concurrency

Localized Reference

• Original Chinese version is preserved in

  SKILL_ZH.md

  .