0.9.0: Initial Release

This is the initial release.

Collecting metrics

The core component of the metrics is the metrics.Collector interface. You can create a new instance of this interface by calling metrics.New() with a GeoIP lookup provider from the geoip library as a parameter. You can then dynamically create metrics:

m := metrics.New(geoip)
testCounter, err := m.CreateCounter(
    "test", // Name of the metric
    "MB", // Unit of the metric
    "This is a test", // Help text of the metric
)

You can then increment the counter:

testCounter.Increment()
testCounter.IncrementBy(5)

Alternatively, you can also create a CounterGeo to make a label automatically based on GeoIP lookup:

testCounter, err := m.CreateCounterGeo(
    "test", // Name of the metric
    "MB", // Unit of the metric
    "This is a test", // Help text of the metric
)
testCounter.Increment(net.ParseIP("127.0.0.1"))

If you need a metric that can be decremented or set directly you can use the Gauge type instead.

Using the metrics server

The metrics server exposes the collected metrics on an HTTP webserver in the Prometheus / OpenMetrics format. It requires the service library and a logger from the log library to work properly:

server := metrics.NewServer(
    metrics.Config{
        ServerConfiguration: http.ServerConfiguration{
            Listen:       "127.0.0.1:8080",
        },
        Enable:              true,
        Path:                "/metrics",
    },
    metricsCollector,
    logger,
)

lifecycle := service.NewLifecycle(server)
go func() {
    if err := lifecycle.Run(); err != nil {
        // Handle crash
    } 	
}()

//Later:
lifecycle.Stop(context.Background())

Alternatively, you can skip the full HTTP server and request a handler that you can embed in any Go HTTP server:

handler := metrics.NewHandler(
    "/metrics",
    metricsCollector
)
http.ListenAndServe("0.0.0.0:8080", handler)