DevOps & CloudSoftware Engineering

Monitoring and Observability for Web Applications

Set up proper monitoring with structured logging, error tracking, performance metrics, and alerting before your users find the bugs.

Mbeah Essilfie

May 22, 2026 at 06:04 AM

10 min read 791 views

Monitoring and Observability for Web Applications

Observability ≠ Monitoring

Monitoring tells you something is wrong (alert: 500 errors spiking). Observability helps you understand why (trace the request through your system).

The three pillars: Logs, Metrics, Traces.

Structured Logging

// ❌ Unstructured — hard to search and parse
console.log('User ' + userId + ' created post ' + postId)

// ✅ Structured — queryable and parseable
logger.info('Post created', {
  userId,
  postId,
  duration: elapsed,
  tags: post.tags,
})

A simple structured logger:

type LogLevel = 'debug' | 'info' | 'warn' | 'error'

function createLogger(service: string) {
  return {
    info: (message: string, meta?: Record<string, unknown>) =>
      emit('info', message, meta),
    error: (message: string, error?: Error, meta?: Record<string, unknown>) =>
      emit('error', message, { ...meta, error: error?.stack }),
  }

  function emit(level: LogLevel, message: string, meta?: Record<string, unknown>) {
    const entry = {
      timestamp: new Date().toISOString(),
      level,
      service,
      message,
      ...meta,
    }
    console.log(JSON.stringify(entry))
  }
}

const logger = createLogger('api')

Error Tracking

// server/plugins/error-tracking.ts
export default defineNitroPlugin((nitroApp) => {
  nitroApp.hooks.hook('error', (error, { event }) => {
    // Send to Sentry, Bugsnag, etc.
    reportError({
      error,
      url: event?.path,
      user: event?.context?.user?.id,
      timestamp: Date.now(),
    })
  })
})

Don't just log errors — alert on them. An error in your logs that nobody sees is the same as no error tracking at all.

Key Metrics to Track

Application metrics:

Request rate (requests/second)
Error rate (5xx responses / total)
Latency (p50, p95, p99)
Active users

Infrastructure metrics:

CPU / Memory usage
Database connection pool utilization
Queue depth

Health Check Endpoint

// server/api/health.get.ts
export default defineEventHandler(async () => {
  const checks = {
    database: await checkDb(),
    redis: await checkRedis(),
    storage: await checkS3(),
  }

  const healthy = Object.values(checks).every(c => c.status === 'ok')

  return {
    status: healthy ? 'healthy' : 'degraded',
    checks,
    version: process.env.APP_VERSION,
    uptime: process.uptime(),
  }
})