ADR-022: Error Handling and Logging Strategy

Status

Proposed (aspirational)

Note: This ADR is aspirational guidance, not an enforced rule. Unlike Accepted ADRs — which the agent constitution treats as binding (halt-on-violation) — this one documents a recommended error-handling/logging strategy a cloner may adopt as their app grows. It is intentionally not wired into quality:ci or the halt gate. Promote to Accepted if/when the strategy is implemented in the template itself.

Context

The Astro starter template currently lacks a comprehensive error handling and logging strategy. As the application grows, we need consistent patterns for:

• Error handling: How to catch, log, and recover from errors
• Structured logging: What to log, at what level, and in what format
• Observability: How to monitor application health and debug issues
• Security: Ensuring sensitive data isn’t logged

This becomes especially critical for:

• Dynamic routes that may encounter missing content
• Form submissions that may fail
• Build-time operations (token generation, link validation)
• Production deployments where debugging is harder

Decision Drivers

• Debuggability: Developers need clear error messages to diagnose issues
• Security: Never log sensitive data (tokens, passwords, PII)
• Performance: Logging should have minimal overhead
• Observability: Errors should be trackable in production
• User Experience: Users should see helpful error messages, not stack traces

Considered Options

Option 1: Console.log Only (Current State)

Description: Ad-hoc console.log() and console.error() statements

Pros:

• Simple, no dependencies
• Works in browser and Node.js

Cons:

• No structure, hard to parse
• No log levels (debug vs error)
• No context (timestamps, request IDs)
• Can’t filter or search logs

Option 2: Structured Logging with Pino

Description: Use Pino for structured JSON logging

Pros:

• Fast (5x faster than Winston)
• Structured JSON output
• Log levels (trace, debug, info, warn, error, fatal)
• Child loggers with context
• Production-ready

Cons:

• Additional dependency
• Requires configuration
• JSON logs less readable in development

Option 3: Custom Logger Wrapper

Description: Build a thin wrapper around console with structure

Pros:

• No dependencies
• Full control over format
• Can add structure gradually

Cons:

• Reinventing the wheel
• Missing features (log rotation, transports)
• Maintenance burden

Decision

We will implement Option 2 (Structured Logging with Pino) for build scripts and server-side code, with the following patterns:

1. Error Handling Patterns

// ✅ Graceful error handling in dynamic routes
export async function getStaticPaths() {
  try {
    const posts = await getCollection('blog');
    return posts.map(post => ({
      params: { slug: post.slug },
      props: { post },
    }));
  } catch (error) {
    logger.error({ error, context: 'getStaticPaths' }, 'Failed to load blog posts');
    // Fail build - don't deploy broken site
    throw error;
  }
}

// ✅ Defensive error handling in pages
const { slug } = Astro.params;
const post = await getPost(slug);

if (!post) {
  logger.warn({ slug }, 'Post not found, redirecting to 404');
  return Astro.redirect('/404', 303);
}

2. Structured Logging

import pino from 'pino';

export const logger = pino({
  level: process.env.LOG_LEVEL || 'info',
  transport: {
    target: 'pino-pretty',
    options: {
      colorize: true,
      ignore: 'pid,hostname',
      translateTime: 'HH:MM:ss',
    },
  },
});

// Usage
logger.info({ userId: 123, action: 'login' }, 'User logged in');
logger.error({ error, context: 'database' }, 'Database connection failed');

3. Security: Redact Sensitive Data

// ✅ Redact sensitive fields
const logger = pino({
  redact: {
    paths: [
      'password',
      'token',
      'apiKey',
      'email',
      '*.password',
      '*.token',
      'req.headers.authorization',
    ],
    censor: '[REDACTED]',
  },
});

// Example: email is redacted
logger.info({ email: 'user@example.com', action: 'signup' }, 'User signed up');
// Output: { email: '[REDACTED]', action: 'signup', msg: 'User signed up' }

4. Build-Time Logging

import { logger } from './logger';

export async function buildTokens() {
  logger.info('Starting token build');

  try {
    const tokens = await loadTokens();
    logger.debug({ tokenCount: tokens.length }, 'Loaded tokens');

    const validated = validateContrast(tokens);
    logger.info({ validated: validated.length }, 'Validated token contrast');

    await writeTokens(validated);
    logger.info('Token build complete');
  } catch (error) {
    logger.error({ error }, 'Token build failed');
    process.exit(1);
  }
}

5. Client-Side Error Handling

// For client-side errors (Preact islands, forms)
window.addEventListener('error', (event) => {
  // Send to error tracking service (Sentry, LogRocket, etc.)
  console.error('Unhandled error:', event.error);

  // Don't log in production console (security)
  if (import.meta.env.PROD) {
    event.preventDefault();
  }
});

// Form error handling
async function handleSubmit(event: Event) {
  event.preventDefault();

  try {
    const response = await fetch('/api/contact', {
      method: 'POST',
      body: new FormData(event.target as HTMLFormElement),
    });

    if (!response.ok) {
      throw new Error(`HTTP ${response.status}: ${response.statusText}`);
    }

    // Success
    window.location.href = '/thank-you';
  } catch (error) {
    // Show user-friendly error
    showError('Failed to send message. Please try again.');

    // Log for debugging (only in dev)
    if (import.meta.env.DEV) {
      console.error('Form submission failed:', error);
    }
  }
}

Consequences

Positive

• Better debugging: Structured logs are searchable and filterable
• Security: Sensitive data is automatically redacted
• Observability: Can integrate with log aggregation tools (Datadog, Splunk)
• Performance: Pino is fast (minimal overhead)
• Consistency: All errors logged in same format

Negative

• Dependency: Adds Pino and pino-pretty to devDependencies
• Learning curve: Team needs to learn structured logging
• JSON logs: Less readable in development (mitigated by pino-pretty)

Neutral

• Build size: No impact (logging is build-time only)
• Runtime: No client-side logging library (uses native console)

Validation

• Build logs: Check that token build, link validation use structured logging
• Error scenarios: Test 404 handling, form errors, missing content
• Security: Verify sensitive data is redacted in logs
• Performance: Ensure logging doesn’t slow down builds

Implementation Checklist

• Install Pino and pino-pretty
• Create scripts/logger.ts with redaction config
• Update scripts/build-tokens.ts to use logger
• Update scripts/check-review-dates.mjs to use logger
• Add error handling to dynamic routes ([slug].astro)
• Document logging patterns in CONTRIBUTING.md
• Add log level configuration to .env.example

References

• Pino Documentation
• Structured Logging Best Practices
• OWASP Logging Cheat Sheet

Related ADRs

• ADR-011: Dynamic Route Error Handling (implements graceful degradation)
• ADR-021: Contact Form Progressive Enhancement (form error handling)

Date: 2025-11-15
Participants: Development Team
Outcome: Proposed