receiver/prometheus: reject datapoints with duplicate label keys (#3408)

As of Prometheus 2.16.0 (released on 2020-02-13), datapoints with
duplicate label keys MUST be rejected and the Prometheus RemoteWrite
exporter tests were failing, but that was a red herring as the
real issue was really in the receiver.

Fixes open-telemetry/prometheus-interoperability-spec#44
Fixes #3407

receiver/prometheusreceiver/internal/metricsbuilder.go

@@ -18,6 +18,7 @@ import (
 	"errors"
 	"fmt"
 	"regexp"
+	"sort"
 	"strconv"
 	"strings"
 
@@ -87,6 +88,23 @@ func (b *metricBuilder) matchStartTimeMetric(metricName string) bool {
 
 // AddDataPoint is for feeding prometheus data complexValue in its processing order
 func (b *metricBuilder) AddDataPoint(ls labels.Labels, t int64, v float64) error {
+	// Any datapoint with duplicate labels MUST be rejected per:
+	// * https://github.com/open-telemetry/wg-prometheus/issues/44
+	// * https://github.com/open-telemetry/opentelemetry-collector/issues/3407
+	// as Prometheus rejects such too as of version 2.16.0, released on 2020-02-13.
+	seen := make(map[string]bool)
+	dupLabels := make([]string, 0, len(ls))
+	for _, label := range ls {
+		if _, ok := seen[label.Name]; ok {
+			dupLabels = append(dupLabels, label.Name)
+		}
+		seen[label.Name] = true
+	}
+	if len(dupLabels) != 0 {
+		sort.Strings(dupLabels)
+		return fmt.Errorf("invalid sample: non-unique label names: %q", dupLabels)
+	}
+
 	metricName := ls.Get(model.MetricNameLabel)
 	switch {
 	case metricName == "":

receiver/prometheusreceiver/internal/metricsbuilder_test.go

@@ -25,6 +25,7 @@ import (
 	"github.com/prometheus/prometheus/pkg/textparse"
 	"github.com/prometheus/prometheus/scrape"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 	"google.golang.org/protobuf/types/known/wrapperspb"
 )
 
@@ -1474,3 +1475,23 @@ func Test_heuristicalMetricAndKnownUnits(t *testing.T) {
 		})
 	}
 }
+
+// Ensure that we reject duplicate label keys. See https://github.com/open-telemetry/wg-prometheus/issues/44.
+func TestMetricBuilderDuplicateLabelKeysAreRejected(t *testing.T) {
+	mc := newMockMetadataCache(testMetadata)
+	mb := newMetricBuilder(mc, true, "", testLogger)
+
+	dupLabels := labels.Labels{
+		{Name: "__name__", Value: "test"},
+		{Name: "a", Value: "1"},
+		{Name: "a", Value: "1"},
+		{Name: "z", Value: "9"},
+		{Name: "z", Value: "1"},
+		{Name: "instance", Value: "0.0.0.0:8855"},
+		{Name: "job", Value: "test"},
+	}
+
+	err := mb.AddDataPoint(dupLabels, 1917, 1.0)
+	require.NotNil(t, err)
+	require.Contains(t, err.Error(), `invalid sample: non-unique label names: ["a" "z"]`)
+}