Metrics Package

Overview

The Metrics package provides a flexible and efficient way to instrument Go applications for metrics collection. It supports various metric types and is designed around a singleton pattern to ensure global access and performance. The package defaults to a no-operation implementation for environments where metrics is not required or enabled.

Features

• Singleton Access: Ensures that the entire application uses a single instance of the metrics system.
• Extensible Interface: Supports multiple backend implementations.
• HTTP Metrics Handler: Easy integration with HTTP servers to expose metrics.
• Lazy Loading: Metrics are instantiated only when they are first used, which optimizes resource usage.

Metric Types

• CountMeter: Monotonically increasing counter that resets on application restart.
• CountVecMeter: CountMeter with support for labeling for dimensional data.
• GaugeMeter: Represents a metric that can go up or down.
• GaugeVecMeter: GaugeMeter with label support.
• HistogramMeter: Measures distributions of values into predefined buckets.
• HistogramVecMeter: HistogramMeter with label support.

Usage

Counters

To create a counter:

counter := metrics.Counter("request_count")
counter.Add(1)

For counters with labels:

counterVec := metrics.CounterVec("request_count_by_status", []string{"status"})
counterVec.AddWithLabel(1, map[string]string{"status": "200"})

Gauges

To create a gauge:

gauge := metrics.Gauge("current_users")
gauge.Gauge(5)

For gauges with labels:

gaugeVec := metrics.GaugeVec("current_users_by_tier", []string{"tier"})
gaugeVec.GaugeWithLabel(5, map[string]string{"tier": "premium"})

Histograms

To create a histogram:

histogram := metrics.Histogram("response_times_ms", metrics.Bucket10s)
histogram.Observe(350)

For histograms with labels:

histogramVec := metrics.HistogramVec("response_times_by_route_ms", []string{"route"}, metrics.BucketHTTPReqs)
histogramVec.ObserveWithLabels(350, map[string]string{"route": "/api/data"})

Lazy Loading

To defer the instantiation of any metric:

lazyHistogram := metrics.LazyLoadHistogram("response_times_ms", metrics.Bucket10s)
h := lazyHistogram() // Actual instantiation occurs here
h.Observe(500)

HTTP Handler

To expose metrics via HTTP:

http.Handle("/metrics", metrics.Handler())