Logging Basics

8 examples to get you started with Logging for Node.js backends - 6 basic and 2 intermediate.

Prerequisites

npm install pino
npm install -D pino-pretty typescript@5.6 tsx

Production services on Node 24.18.0 should emit structured JSON to stdout - aggregators (Datadog, Loki, CloudWatch) parse one object per line.

Basic Examples

1. Structured JSON Instead of String Concatenation

Machines parse fields; humans grep level and requestId.

import pino from "pino";
 
const logger = pino({ level: "info" });
 
logger.info({ userId: "u_123", action: "login_success" }, "user logged in");
// {"level":30,"userId":"u_123","action":"login_success","msg":"user logged in",...}

First argument is a context object; second is a short human message.
String templates like `User ${id} logged in` lose searchable fields.
level 30 is info in Pino's numeric mapping.

Related: pino - child loggers and redaction

2. Log Levels Mean Something

Pick the level that matches operator response, not developer mood.

logger.fatal({ err }, "database unreachable at boot");
logger.error({ err, orderId }, "payment capture failed");
logger.warn({ retryCount: 3 }, "downstream timeout, retrying");
logger.info({ requestId }, "request completed");
logger.debug({ query }, "sql executed");

error - needs investigation or user impact.
warn - degraded but recovering (retries, fallbacks).
info - normal business and request lifecycle events.
debug - development only or sampled in production.

Related: Logging Best Practices - prod log level policy

3. Log Errors with the `err` Key

Pino's standard serializer captures stack traces.

try {
  await chargeCard(orderId);
} catch (err) {
  logger.error({ err, orderId }, "charge failed");
}

{ err } triggers pino.stdSerializers.err - stack and type preserved.
logger.error(err) without object wrapper loses field structure.
Map to HTTP 5xx after logging - log first, then respond generically.

Related: Error Response Standards - client vs log detail

4. Ban `console.log` in Application Code

Unstructured lines break JSON-only log pipelines.

// eslint no-console: error
import pino from "pino";
const log = pino();
log.info({ event: "worker_started" });

console.log bypasses level filters and redaction.
ESLint no-console in src/ with exception for CLI scripts.
Tests can use pino.destination({ sync: true }) for capture.

Related: Logging with Pino on Fastify - framework integration

5. Include Request Metadata on Every HTTP Log

Method, path, status, and duration power SLO dashboards.

logger.info({
  event: "request_complete",
  method: "GET",
  path: "/users",
  statusCode: 200,
  durationMs: 42,
  requestId: "req_abc",
});

Consistent field names across services (requestId, not reqId in one app and rid in another).
Duration from reply.elapsedTime (Fastify) or middleware timing (Express).
Health checks can be sampled or excluded to reduce noise.

Related: Request Correlation IDs - x-request-id

6. One Logger Instance Per Process

Create at module scope; avoid pino() per request.

// src/logger.ts
import pino from "pino";
 
export const logger = pino({
  level: process.env.LOG_LEVEL ?? "info",
  base: { service: "orders-api", version: process.env.APP_VERSION },
});

base fields appear on every line - service name for multi-tenant log indexes.
Child loggers add per-request fields without reconfiguring the root.
Pass logger to workers via import, not global mutation.

Related: pino - configuration options

Intermediate Examples

7. Child Logger for Request Context

Bind requestId once; all downstream logs inherit it.

const child = logger.child({ requestId: "req_abc", tenantId: "t_1" });
child.info({ action: "fetch_order" });
child.info({ action: "send_receipt" });
// both lines include requestId and tenantId

Create child in HTTP middleware or Fastify onRequest hook.
After auth, req.log = req.log.child({ userId: user.id }).
Child loggers are cheap - prefer over manual spread on every call.

Related: AsyncLocalStorage for Context - non-HTTP workers

8. Redact Sensitive Paths

Keep passwords and tokens out of log storage.

const logger = pino({
  redact: {
    paths: ["req.headers.authorization", "req.body.password", "email"],
    censor: "[REDACTED]",
  },
});

Redaction runs at serialize time - safer than hoping developers remember.
List paths in a security review checklist.
GDPR and PCI often require email/phone redaction or hashing in logs.

Related: PII Redaction & Compliance - retention policies

Node.js API Documentation - Complete reference for core modules and APIs.
Node.js Best Practices (GitHub) - Curated production best practices.
DigitalOcean Node.js Tutorials - Production-oriented how-tos.
Node.js the Right Way - Server-side JavaScript fundamentals.
Node.js Cookbook, 5th Edition - Practical server-side recipes.

Stack versions: This page was written for Node.js 24.18.0 (Active LTS), npm 10+, TypeScript 5.6+, Express 5, Fastify 5, and NestJS 11.

Prerequisites

npm install pino
npm install -D pino-pretty typescript@5.6 tsx

Production services on Node 24.18.0 should emit structured JSON to stdout - aggregators (Datadog, Loki, CloudWatch) parse one object per line.

Basic Examples

1. Structured JSON Instead of String Concatenation

Machines parse fields; humans grep level and requestId.

import pino from "pino";
 
const logger = pino({ level: "info" });
 
logger.info({ userId: "u_123", action: "login_success" }, "user logged in");
// {"level":30,"userId":"u_123","action":"login_success","msg":"user logged in",...}

First argument is a context object; second is a short human message.
String templates like `User ${id} logged in` lose searchable fields.
level 30 is info in Pino's numeric mapping.

Related: pino - child loggers and redaction

2. Log Levels Mean Something

Pick the level that matches operator response, not developer mood.

logger.fatal({ err }, "database unreachable at boot");
logger.error({ err, orderId }, "payment capture failed");
logger.warn({ retryCount: 3 }, "downstream timeout, retrying");
logger.info({ requestId }, "request completed");
logger.debug({ query }, "sql executed");

error - needs investigation or user impact.
warn - degraded but recovering (retries, fallbacks).
info - normal business and request lifecycle events.
debug - development only or sampled in production.

Related: Logging Best Practices - prod log level policy

3. Log Errors with the `err` Key

Pino's standard serializer captures stack traces.

try {
  await chargeCard(orderId);
} catch (err) {
  logger.error({ err, orderId }, "charge failed");
}

{ err } triggers pino.stdSerializers.err - stack and type preserved.
logger.error(err) without object wrapper loses field structure.
Map to HTTP 5xx after logging - log first, then respond generically.

Related: Error Response Standards - client vs log detail

4. Ban `console.log` in Application Code

Unstructured lines break JSON-only log pipelines.

// eslint no-console: error
import pino from "pino";
const log = pino();
log.info({ event: "worker_started" });

console.log bypasses level filters and redaction.
ESLint no-console in src/ with exception for CLI scripts.
Tests can use pino.destination({ sync: true }) for capture.

Related: Logging with Pino on Fastify - framework integration

5. Include Request Metadata on Every HTTP Log

Method, path, status, and duration power SLO dashboards.

logger.info({
  event: "request_complete",
  method: "GET",
  path: "/users",
  statusCode: 200,
  durationMs: 42,
  requestId: "req_abc",
});

Consistent field names across services (requestId, not reqId in one app and rid in another).
Duration from reply.elapsedTime (Fastify) or middleware timing (Express).
Health checks can be sampled or excluded to reduce noise.

Related: Request Correlation IDs - x-request-id

6. One Logger Instance Per Process

Create at module scope; avoid pino() per request.

// src/logger.ts
import pino from "pino";
 
export const logger = pino({
  level: process.env.LOG_LEVEL ?? "info",
  base: { service: "orders-api", version: process.env.APP_VERSION },
});

base fields appear on every line - service name for multi-tenant log indexes.
Child loggers add per-request fields without reconfiguring the root.
Pass logger to workers via import, not global mutation.

Related: pino - configuration options

Intermediate Examples

7. Child Logger for Request Context

Bind requestId once; all downstream logs inherit it.

const child = logger.child({ requestId: "req_abc", tenantId: "t_1" });
child.info({ action: "fetch_order" });
child.info({ action: "send_receipt" });
// both lines include requestId and tenantId

Create child in HTTP middleware or Fastify onRequest hook.
After auth, req.log = req.log.child({ userId: user.id }).
Child loggers are cheap - prefer over manual spread on every call.

Related: AsyncLocalStorage for Context - non-HTTP workers

8. Redact Sensitive Paths

Keep passwords and tokens out of log storage.

const logger = pino({
  redact: {
    paths: ["req.headers.authorization", "req.body.password", "email"],
    censor: "[REDACTED]",
  },
});

Redaction runs at serialize time - safer than hoping developers remember.
List paths in a security review checklist.
GDPR and PCI often require email/phone redaction or hashing in logs.

Related: PII Redaction & Compliance - retention policies

Node.js API Documentation - Complete reference for core modules and APIs.
Node.js Best Practices (GitHub) - Curated production best practices.
DigitalOcean Node.js Tutorials - Production-oriented how-tos.
Node.js the Right Way - Server-side JavaScript fundamentals.
Node.js Cookbook, 5th Edition - Practical server-side recipes.

Stack versions: This page was written for Node.js 24.18.0 (Active LTS), npm 10+, TypeScript 5.6+, Express 5, Fastify 5, and NestJS 11.

Logging Basics

Search Documentation

Logging Basics