Logging & Observability

Logger

The Logger interface provides structured logging capabilities throughout your Zuplo API gateway. The logger is accessible via the context.log property and supports multiple log levels.

Interface

Code(ts)
 
interface Logger {
  debug(...messages: unknown[]): void;
  info(...messages: unknown[]): void;
  warn(...messages: unknown[]): void;
  error(...messages: unknown[]): void;
  log(...messages: unknown[]): void;
}

Log Levels

The logger supports the following levels (from least to most severe):

debug - Detailed information for debugging
info - General informational messages
warn - Warning messages for potentially problematic situations
error - Error messages for failures and exceptions
log - Alias for info level

Usage

The logger is available on the ZuploContext object:

Code(ts)
 
import { ZuploContext, ZuploRequest } from "@zuplo/runtime";

export default async function handler(
  request: ZuploRequest,
  context: ZuploContext,
) {
  // Log at different levels
  context.log.debug("Detailed debug information");
  context.log.info("Request received", { path: request.url });
  context.log.warn("Deprecated endpoint accessed");
  context.log.error("Failed to process request", { error: "Invalid input" });

  return new Response("OK");
}

Structured Logging

The logger accepts multiple arguments and automatically serializes objects:

Code(ts)
 
// Log simple strings
context.log.info("User authenticated");

// Log objects
context.log.info({
  userId: "user-123",
  action: "login",
  timestamp: Date.now(),
});

// Log multiple arguments
context.log.info(
  "Processing request",
  {
    method: request.method,
    path: request.url,
  },
  "additional info",
);

// Log arrays
context.log.debug(["step1", "step2", "step3"]);

// Log errors
try {
  await riskyOperation();
} catch (error) {
  context.log.error("Operation failed", {
    error: error.message,
    stack: error.stack,
  });
}

Automatic Request ID

The logger automatically includes the request ID with every log entry:

Code(ts)
 
context.log.info("Processing payment");
// Output includes: { "message": "Processing payment", "requestId": "req_abc123..." }

Environment Configuration

Log levels vary by environment:

Development/Preview: Typically set to debug or info level
Production: Typically set to error level only

Best Practices

1. Use Appropriate Log Levels

Code(ts)
 
// Debug - detailed information for troubleshooting
context.log.debug("Cache lookup", { key: cacheKey, ttl: 300 });

// Info - general application flow
context.log.info("Order created", { orderId: "order-123", total: 99.99 });

// Warn - concerning but not critical
context.log.warn("Rate limit approaching", { current: 95, limit: 100 });

// Error - failures and exceptions
context.log.error("Payment failed", {
  orderId: "order-123",
  errorCode: "INSUFFICIENT_FUNDS",
});

2. Include Contextual Information

Code(ts)
 
export default async function handler(
  request: ZuploRequest,
  context: ZuploContext,
) {
  const startTime = Date.now();
  const userId = request.user?.sub;

  context.log.info("Request started", {
    userId,
    method: request.method,
    path: new URL(request.url).pathname,
    userAgent: request.headers.get("user-agent"),
  });

  try {
    const result = await processRequest(request);

    context.log.info("Request completed", {
      userId,
      duration: Date.now() - startTime,
      status: "success",
    });

    return new Response(JSON.stringify(result));
  } catch (error) {
    context.log.error("Request failed", {
      userId,
      duration: Date.now() - startTime,
      error: error.message,
      stack: error.stack,
    });

    throw error;
  }
}

3. Avoid Logging Sensitive Data

Code(ts)
 
// DON'T log sensitive information
context.log.info("User login", {
  email: user.email,
  password: user.password, // Never log passwords!
});

// DO log safe identifiers
context.log.info("User login", {
  userId: user.id,
  email: user.email.replace(/(.{2}).*(@.*)/, "$1***$2"), // Partially masked
});

4. Use Structured Data for Better Querying

Code(ts)
 
// Good - structured data that can be queried
context.log.info("API call completed", {
  service: "payment-processor",
  operation: "charge",
  duration: 1234,
  status: "success",
  amount: 99.99,
  currency: "USD",
});

// Less useful - unstructured string
context.log.info(`Payment of $99.99 USD processed in 1234ms`);

Integration with Log Management

Logs are automatically forwarded to your configured log management solution (DataDog, Loki, etc.). The structured format makes it easy to:

Search and filter logs
Create alerts based on log patterns
Generate metrics from log data
Trace requests across services

Performance Considerations

Logging has minimal performance impact
Logs are processed asynchronously
In production, use error level to reduce log volume
Avoid logging large objects or sensitive data