YepAPI/skills

Logging & Monitoring

Rules

• Structured logging: JSON format with timestamp, level, message, requestId, userId fields
• Log levels: error (broken), warn (degraded), info (business events), debug (development only)
• Use Pino for Node.js logging — fast, structured, JSON by default
• Sentry for error tracking: initialize early, capture unhandled exceptions, add user context and breadcrumbs
• Correlate logs with request IDs — generate a UUID per request, pass through all service calls
• Never log secrets, passwords, tokens, or full credit card numbers — redact sensitive fields
• Alert on error rate spikes, not individual errors — use thresholds (e.g., >5% error rate in 5 min)
• Axiom/Datadog: ship logs via OTLP or native SDK, add custom metrics for business KPIs
• Health check endpoint: GET /health returning { status: "ok" } with DB connectivity check
• Monitor response times at p50, p95, p99 — alert on p95 degradation

Patterns

// Pino structured logging
import pino from "pino";
const logger = pino({ level: process.env.LOG_LEVEL || "info" });

logger.info({ userId, action: "checkout", orderId }, "Order placed");
logger.error({ err, requestId }, "Payment processing failed");

// Sentry init (Next.js)
import * as Sentry from "@sentry/nextjs";
Sentry.init({ dsn: process.env.SENTRY_DSN, tracesSampleRate: 0.1 });

Avoid

• console.log in production — use a structured logger
• Logging every request body — log summaries, not payloads
• Ignoring unhandled promise rejections — always capture them
• Alert fatigue from noisy alerts — tune thresholds, group similar errors