获得普罗米修斯度量属性

获取 Prometheus 度量属性通常需要使用低级别的 protobuf 定义，因为 Desc(ription) 类型的字段未导出。为了绕过此限制，可以使用注册表方法。首先，将度量注册到注册表中。然后，通过收集注册表中的度量并访问 MetricType、Name、Help、Label、Bucket 等属性，可以访问度量的属性。通过验证属性与预期值相匹配，单元测试可以用来确保度量配置正确。

假设我有一个直方图指标：

requestDuration = prometheus.NewHistogramVec(prometheus.HistogramOpts{
    Name:    "example_request_duration_seconds",
    Help:    "Histogram for the runtime of a simple example function.",
    Buckets: prometheus.LinearBuckets(0.01, 0.01, 10),
}, []string{"label1"})

我想访问帮助、标签、存储桶、constlabels、目标、maxage 等属性。我该怎么做？

我基本上是在尝试编写单元测试来验证这些属性与预期的属性。

正确答案

由于 Desc(ription) type 的所有字段均未导出，因此您必须使用注册表绕过较低级别的 protobuf 定义：

package main

import (
    "testing"

    "github.com/prometheus/client_golang/prometheus"
    dto "github.com/prometheus/client_model/go"
    "github.com/stretchr/testify/assert"
    "github.com/stretchr/testify/require"
)

var requestDuration = prometheus.NewHistogramVec(prometheus.HistogramOpts{
    Name:    "example_request_duration_seconds",
    Help:    "Histogram for the runtime of a simple example function.",
    Buckets: prometheus.LinearBuckets(0.01, 0.01, 10),
}, []string{"label1"})

func TestMetrics(t *testing.T) {
    reg := prometheus.NewPedanticRegistry()
    err := reg.Register(requestDuration)
    require.NoError(t, err)

    // The built-in collectors must observe at least one values before they
    // are exported.
    //
    // This With call implicitly verifies that your labels are as expected.
    // If your program setup initializes metrics you can omit this and verify
    // labels directly as demonstrated below. Initializing metrics is best
    // practice; see
    // https://prometheus.io/docs/practices/instrumentation/#avoid-missing-metrics).
    requestDuration.With(prometheus.Labels{"label1": "foo"}).Observe(0)

    metrics, err := reg.Gather()
    require.NoError(t, err)
    require.Len(t, metrics, 1)

    assert.Equal(t, dto.MetricType_HISTOGRAM, metrics[0].GetType())
    assert.Equal(t, "example_request_duration_seconds", metrics[0].GetName())
    assert.Equal(t, "Histogram for the runtime of a simple example function.", metrics[0].GetHelp())

    // Label verification only required if Observe call above can be omitted.
    labels := metrics[0].GetMetric()[0].GetLabel()
    require.Len(t, labels, 1)
    assert.Equal(t, "label1", labels[0].GetName())

    buckets := metrics[0].GetMetric()[0].GetHistogram().GetBucket()
    require.Len(t, buckets, 10)
    assert.InDelta(t, 0.01, buckets[0].GetUpperBound(), 1e-6)
    assert.InDelta(t, 0.02, buckets[1].GetUpperBound(), 1e-6)
    // ...
    assert.InDelta(t, 0.09, buckets[8].GetUpperBound(), 1e-6)
    assert.InDelta(t, 0.10, buckets[9].GetUpperBound(), 1e-6)
}

今天关于《获得普罗米修斯度量属性》的内容就介绍到这里了，是不是学起来一目了然！想要了解更多关于的内容请关注golang学习网公众号！