logging-design-patternslisted

# Logging Design Patterns Rules for structured, secure, and cost-effective logging in production Node.js/TypeScript services. **Research:** `docs/deepresearch/reports/Logging Design Best Practices.md` ## Non-Negotiable Rules | # | Rule | Violation Example | Correct Pattern | |---|------|-------------------|-----------------| | 1 | **Structured JSON, never printf** | `` console.log(`User ${id} failed`) `` | `logger.info({ user_id: id, event: 'login_failed' })` | | 2 | **snake_case log fields** | `{ userId, requestId }` | `{ user_id, request_id }` (OTel/Datadog convention) | | 3 | **Strict log level semantics** | `logger.info('DB pool exhausted')` | `logger.warn({ pool_usage: 95 }, 'High DB pool usage')` | | 4 | **Correlation ID on every log** | Log lines with no trace context | Include `trace_id` and `span_id` from OTel span | | 5 | **Never log PII or secrets** | `logger.info({ body: req.body })` | Use `pino-redact` with explicit allowlist paths | | 6 | **No console.log in production** | `console.log(data)` | `logger.info({ ... }, 'message')` | | 7 | **Logs for events, metrics for aggregates** | `` logger.info(`took ${ms}ms`) `` | Emit histogram metric + log the discrete event | | 8 | **AsyncLocalStorage for context** | Manually passing request_id through call chain | `als.run({ request_id }, () => next())` | | 9 | **Never log 100% at high QPS** | Logging every request at INFO in hot paths | Tail-based sampling (errors + slow requests always) | | 10 | **Canonical log line