metrics - lightweight package for exporting metrics in Prometheus format

Features

• Lightweight. Has minimal number of third-party dependencies and all these deps are small. See this article for details.
• Easy to use. See the API docs.
• Fast.
• Allows exporting distinct metric sets via distinct endpoints. See Set.
• Supports easy-to-use histograms, which just work without any tuning. Read more about VictoriaMetrics histograms at this article.
• Can push metrics to VictoriaMetrics or to any other remote storage, which accepts metrics in Prometheus text exposition format. See these docs.

Limitations

• It doesn't implement advanced functionality from github.com/prometheus/client_golang.

Usage

import "github.com/VictoriaMetrics/metrics"

// Register various metrics.
// Metric name may contain labels in Prometheus format - see below.
var (
	// Register counter without labels.
	requestsTotal = metrics.NewCounter("requests_total")

	// Register summary with a single label.
	requestDuration = metrics.NewSummary(`requests_duration_seconds{path="/foobar/baz"}`)

	// Register gauge with two labels.
	queueSize = metrics.NewGauge(`queue_size{queue="foobar",topic="baz"}`, func() float64 {
		return float64(foobarQueue.Len())
	})

	// Register histogram with a single label.
	responseSize = metrics.NewHistogram(`response_size{path="/foo/bar"}`)
)

// ...
func requestHandler() {
	// Increment requestTotal counter.
	requestsTotal.Inc()

	startTime := time.Now()
	processRequest()
	// Update requestDuration summary.
	requestDuration.UpdateDuration(startTime)

	// Update responseSize histogram.
	responseSize.Update(responseSize)
}

// Expose the registered metrics at `/metrics` path.
http.HandleFunc("/metrics", func(w http.ResponseWriter, req *http.Request) {
	metrics.WritePrometheus(w, true)
})

// ... or push registered metrics every 10 seconds to http://victoria-metrics:8428/api/v1/import/prometheus
// with the added `instance="foobar"` label to all the pushed metrics.
metrics.InitPush("http://victoria-metrics:8428/api/v1/import/prometheus", 10*time.Second, `instance="foobar"`, true)

By default, exposed metrics do not have TYPE or HELP meta information. Call ExposeMetadata(true) in order to generate TYPE and HELP meta information per each metric.

See docs for more info.

Users

• Metrics has been extracted from VictoriaMetrics sources. See this article for more info about VictoriaMetrics.

FAQ

Why the metrics API isn't compatible with github.com/prometheus/client_golang?

Because the github.com/prometheus/client_golang is too complex and is hard to use.

Why the metrics.WritePrometheus doesn't expose documentation for each metric?

Because this documentation is ignored by Prometheus. The documentation is for users. Just give meaningful names to the exported metrics or add comments in the source code or in other suitable place explaining each metric exposed from your application.

How to implement CounterVec in metrics?

Just use GetOrCreateCounter instead of CounterVec.With. See this example for details.

Why Histogram buckets contain vmrange labels instead of le labels like in Prometheus histograms?

Buckets with vmrange labels occupy less disk space compared to Promethes-style buckets with le labels, because vmrange buckets don't include counters for the previous ranges. VictoriaMetrics provides prometheus_buckets function, which converts vmrange buckets to Prometheus-style buckets with le labels. This is useful for building heatmaps in Grafana. Additionally, its' histogram_quantile function transparently handles histogram buckets with vmrange labels.