Fastify Highlights Summary

Every highlight bullet from the 8 pages in this section, gathered on one page and grouped by the page it came from.

Fastify's Encapsulation Model

• Fastify's core abstraction is a plugin tree, not a flat chain - each register() creates a new, isolated context
• Encapsulation is the default; fastify-plugin is how you deliberately opt out of it
• Schemas compile once at startup into fast validators and serializers, not on every request
• avvio drives async plugin bootstrapping, guaranteeing dependency order before the server accepts traffic

Fastify Basics

• Fastify 5 is schema-first with compiled serializers for speed
• Plugins encapsulate routes, hooks, and decorators in isolated scopes
• Lifecycle hooks: onRequest, preHandler, onResponse
• Built-in JSON parsing and Pino logging out of the box
• Use fastify-plugin to break encapsulation when needed
• inject() enables fast HTTP tests without opening ports

JSON Schema Validation

• Use JSON Schema for every public API endpoint
• Response schemas compile to serializers for speed
• Always define response schemas to serialize faster
• Set additionalProperties false to reject unexpected fields
• Use TypeBox or Zod to keep schemas and TypeScript types in sync
• Reuse schemas with $ref to avoid slow compilation

Fastify Plugins

• Wrap shared plugins with fastify-plugin to lift decorators to parent scope
• register() isolates decorators to child scope; use fastify-plugin for globals
• @fastify/autoload discovers plugins by directory, loading plugins before routes
• Await app.register(dbPlugin) before routes to ensure database is ready
• Add onClose hook to cleanup connection pools when the app shuts down
• Default export async function is required for autoload to recognize plugins

Logging with Pino

• Pino provides structured JSON logging with request context for Fastify services
• Redact password and authorization headers from logs to prevent PII exposure
• Use pino-pretty only in development; JSON to stdout prevents production throughput loss
• Generate request IDs from headers to correlate logs across multiple services
• Use req.log.error with err object to preserve full error stack traces
• Pino writes JSON lines to stdout, parsed natively by log aggregators like Datadog

Testing Fastify Apps

• Use inject() for fast in-memory HTTP testing without ports
• Create testable Fastify apps with factory pattern (buildApp)
• inject() simulates HTTP requests through full Fastify lifecycle in-process
• Call app.close() and use fresh instances to isolate test state
• Test error paths with 400, 401, 404, 500 status responses explicitly
• Wrap mocked dependencies with fastify-plugin before route plugins

Fastify vs Express ADR

• Fastify handles 40-60k rps vs Express 15-20k rps in benchmarks
• Choose Express for team velocity unless throughput is proven critical
• Fastify compiles JSON serializers faster for response-heavy APIs
• Fastify includes JSON Schema validation; Express requires Zod or Ajv
• Express has the largest Node.js HTTP framework ecosystem
• NestJS 11 on Fastify optimizes enterprise monoliths with 50+ endpoints

Fastify Best Practices

• JSON Schema on every public route for validation and serialization
• One plugin per domain; use fastify-plugin for shared decorators
• App factory pattern for testable inject() tests
• Pino JSON logs to stdout; never pino-pretty in production
• await app.register() and app.ready() before accepting traffic
• Close connections in onClose hooks on shutdown

Y29kZWd1aWRlcy5pb3xjZ2lvODIzfDIwMjYwNw