Where to place logs to trace a bug through your system

Log Placement Is a Design Decision

Random log placement creates noise. Strategic placement answers a specific question: given any bug report, can you trace the execution path through your system from the log output alone? If not, your logs are not useful.

The Entry-Exit Pattern

Log at the boundary of every significant operation: when it starts (with all relevant input context) and when it ends (with outcome -- success, error, or result summary). This gives you a complete trace of what happened and how long it took.

async function processOrder(orderId) {
  logger.info({ orderId }, 'processOrder: start');
  try {
    const order = await db.getOrder(orderId);
    logger.info({ orderId, itemCount: order.items.length }, 'processOrder: order fetched');
    const result = await chargeCard(order);
    logger.info({ orderId, chargeId: result.id }, 'processOrder: payment complete');
    return result;
  } catch (err) {
    logger.error({ orderId, err }, 'processOrder: failed');
    throw err;
  }
}

Correlation IDs

In systems with multiple services or parallel requests, logs from different operations interleave. A correlation ID -- a unique identifier assigned at the start of a request and passed through every log entry for that request -- lets you filter all logs for a single operation. Generate one UUID at the API entry point and thread it through every subsequent call. Without it, logs from concurrent users are indistinguishable.