Support for Examplar Metrics

[kirankumar-grootan]

It should be a good to have support for examplar metrics

func (ps *FiberPrometheus) RegisterAt(app fiber.Router, url string, handlers ...fiber.Handler) {
	ps.defaultURL = url

	h := append(handlers, adaptor.HTTPHandler(promhttp.HandlerFor(ps.gatherer, promhttp.HandlerOpts{
		EnableOpenMetrics: true,
	})))
	app.Get(ps.defaultURL, h...)
}

func (ps *FiberPrometheus) Middleware(ctx *fiber.Ctx) error {
        ...

	// Update the request duration histogram
	elapsed := float64(time.Since(start).Nanoseconds()) / 1e9

        // Extract the trace ID from the user context to use it as an exemplar if available.
        traceID := trace.SpanContextFromContext(ctx.UserContext()).TraceID()

	// Retrieve the histogram metric for request duration, labeled with status code, method, and path.
	histogram := ps.requestDuration.WithLabelValues(statusCode, method, path)

	// Check if the trace ID is valid to potentially use it as an exemplar.
	if traceID.IsValid() {
	// If the histogram supports exemplars, record the observed duration with the trace ID as an exemplar.
		if histogramExemplar, ok := histogram.(prometheus.ExemplarObserver); ok {
			histogramExemplar.ObserveWithExemplar(elapsed, prometheus.Labels{"traceID": traceID.String()})
		}
	}

	// Record the observed duration in the histogram.
	histogram.Observe(elapsed)

	return err
}

IDK if there will be a performance hit on this. But it's working. Happy to hear better way of handling this

[github-actions]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[kirankumar-grootan]

Not stale

[gaby]

@kirankumar-grootan Doesn't this require also having opentelemetry as a dependency?

[kirankumar-grootan]

yes @gaby "go.opentelemetry.io/otel/trace"

sry for not being active for long time

PR: #223

[gaby]

yes @gaby "go.opentelemetry.io/otel/trace"

sry for not being active for long time

PR: #223

Left a comment in the PR