Quickstart: Add metrics and tracing to your API

Q: How does tracing help with support?

Tagging spans with request_id lets you jump from a slow or failing trace straight to the exact request support can look up on their side.

2 min read

spanOne per call

request_idTag it on the span

REDRate, errors, duration

Instrument the client

async function tracedCall(url, init) {
  const span = tracer.startSpan('credicorp.call', { attributes: { 'http.url': url } });
  const start = performance.now();
  try {
    const res = await fetch(url, init);
    span.setAttribute('credicorp.request_id', res.headers.get('X-Request-Id'));
    span.setAttribute('http.status_code', res.status);
    metrics.histogram('credicorp.latency', performance.now() - start);
    if (!res.ok) metrics.increment('credicorp.errors');
    return res;
  } finally {
    span.end();
  }
}

Correlate with support

Because the span carries the request_id, a spike in your dashboard links directly to specific requests you can quote to support. This turns monitoring into actionable diagnostics.

Frequently asked questions

What metrics matter most?

Rate, errors and duration (the RED method). Add a saturation signal (your local rate-limit budget from RateLimit-Remaining) and you have full coverage.

How does tracing help with support?

Tagging spans with request_id lets you jump from a slow or failing trace straight to the exact request support can look up on their side.