Correlating Logs, Metrics, and Traces

In this lesson, you will learn how to bridge the gap between disparate telemetry signals by embedding trace identifiers directly into your application logs.

The "Three Pillars" Disconnect

In complex Node.js microservices, metrics tell you that something is wrong (e.g., a spike in latency), traces show you where the bottleneck exists in the call chain, and logs provide the why (the granular execution details). However, if you have to manually copy a traceId from your distributed tracing tool and search for it in a separate logging aggregator, your mean time to resolution (MTTR) will suffer.

To achieve seamless navigation, we must "link" these signals. The standard approach is to inject the current trace_id and span_id into every log record generated during a request.

Implementing Correlation

When using OpenTelemetry, you can extract the current span context and attach it to your logger. If you are using a library like pino or winston, you should use a format that includes these IDs.

Here is a conceptual example using the OpenTelemetry API to augment your log output:

const api = require('@opentelemetry/api');

function logWithTrace(message, level = 'info') {
  const span = api.trace.getSpan(api.context.active());
  const traceId = span?.spanContext().traceId;
  const spanId = span?.spanContext().spanId;

  // Now every log entry contains the trace identifiers
  logger[level]({
    message,
    trace_id: traceId,
    span_id: spanId
  });
}

Navigating the UI

Once your logs include trace_id as a searchable field, modern observability platforms (like Grafana, Datadog, or Honeycomb) detect this pattern. They will automatically render the trace_id as a clickable link. When an engineer clicks that ID in the log viewer, the UI performs a deep-link navigation directly to the waterfall visualization of that specific trace.

By ensuring your structured logs include the trace_id and span_id at the moment of execution, you transform your logs from isolated text blobs into actionable entry points that lead directly to the full context of a distributed request.

Key Takeaway: Correlating logs and traces requires embedding active trace_ids into your log metadata, allowing for instant "one-click" navigation from a specific error event to its distributed execution flow.