testing ADR - architectural decisions

docs/adr/0004-standardized-application-wide-structured-logging.md

@@ -0,0 +1,85 @@
+# ADR-004: Standardized Application-Wide Structured Logging
+
+**Date**: 2025-12-12
+
+**Status**: Proposed
+
+## Context
+
+Our application currently uses a `logger` service, but the implementation of logging varies significantly across different modules. The `errorHandler` middleware produces high-quality, structured JSON logs for errors, but logging within route handlers and service layers is often ad-hoc, using plain strings or inconsistent object structures.
+
+This inconsistency leads to several problems:
+1.  **Difficult Debugging**: It is hard to trace a single user request through the system or correlate events related to a specific operation.
+2.  **Ineffective Log Analysis**: Inconsistent log formats make it difficult to effectively query, filter, and create dashboards in a log management system (like Datadog, Splunk, or the ELK stack).
+3.  **Security Risks**: There is no enforced standard for redacting sensitive information (like passwords or tokens) in logs outside of the `errorHandler`, increasing the risk of accidental data exposure.
+4.  **Missing Context**: Logs often lack crucial context, such as a unique request ID, the authenticated user's ID, or the source IP address, making them less useful for diagnosing issues.
+
+## Decision
+
+We will adopt a standardized, application-wide structured logging policy. All log entries MUST be in JSON format and adhere to a consistent schema.
+
+1.  **Request-Scoped Logger with Context**: We will create a middleware that runs at the beginning of the request lifecycle. This middleware will:
+    *   Generate a unique `request_id` for each incoming request.
+    *   Create a request-scoped logger instance (a "child logger") that automatically includes the `request_id`, `user_id` (if authenticated), and `ip_address` in every log message it generates.
+    *   Attach this child logger to the `req` object (e.g., `req.log`).
+
+2.  **Mandatory Use of Request-Scoped Logger**: All route handlers and any service functions called by them **MUST** use the request-scoped logger (`req.log`) instead of the global logger instance. This ensures all logs for a given request are automatically correlated.
+
+3.  **Standardized Log Schema**: All log messages should follow a base schema. The logger configuration will be updated to enforce this.
+    *   **Base Fields**: `level`, `timestamp`, `message`, `request_id`, `user_id`, `ip_address`.
+    *   **Error Fields**: When logging an error, the log entry MUST include an `error` object with `name`, `message`, and `stack`.
+
+4.  **Standardized Logging Practices**:
+    *   **INFO**: Log key business events, such as `User logged in` or `Flyer processed`.
+    *   **WARN**: Log recoverable errors or unusual situations that do not break the request, such as `Client Error: 404 on GET /api/non-existent-route` or `Retrying failed database connection`.
+    *   **ERROR**: Log only unhandled or server-side errors that cause a request to fail (typically handled by the `errorHandler`). Avoid logging expected client errors (like 4xx) at this level.
+    *   **DEBUG**: Log detailed diagnostic information useful during development, such as function entry/exit points or variable states.
+
+### Example Usage:
+
+```typescript
+// In a new middleware file: logger.middleware.ts
+import { logger } from '../services/logger.server';
+import { randomUUID } from 'crypto';
+
+export const requestLogger = (req, res, next) => {
+  const requestId = randomUUID();
+  // Create a child logger with context for this request
+  req.log = logger.child({
+    request_id: requestId,
+    user_id: req.user?.user_id, // Assumes user is attached by auth middleware
+    ip_address: req.ip,
+  });
+  next();
+};
+
+// In server/app setup:
+// app.use(requestLogger); // Add this early in the middleware chain
+
+// In a route handler:
+router.get('/:id', async (req, res, next) => {
+    // Use the request-scoped logger
+    req.log.info({ flyerId: req.params.id }, 'Fetching flyer by ID');
+    try {
+        // ... business logic ...
+        res.json(flyer);
+    } catch (error) {
+        // The error itself will be logged with full context by the errorHandler
+        next(error);
+    }
+});
+```
+
+## Consequences
+
+### Positive
+
+*   **Enhanced Observability**: Every log line from a single request can be instantly grouped and analyzed, dramatically speeding up debugging.
+*   **Improved Security**: Centralizing the addition of context (like `user_id`) reduces the chance of developers manually logging sensitive data.
+*   **Scalable Log Management**: Consistent JSON logs are easily ingested and indexed by any modern log aggregation tool.
+*   **Clearer Code**: Removes the need to manually pass contextual information (like user ID) down to service functions just for logging purposes.
+
+### Negative
+
+*   **Refactoring Effort**: Requires adding the `requestLogger` middleware and refactoring all routes and services to use `req.log` instead of the global `logger`.
+*   **Slight Performance Overhead**: Creating a child logger for every request adds a minor performance cost, though this is negligible for most modern logging libraries.