[receiver/prometheusreceiver] Preserve bucketless histograms during NHCB conversion

.chloggen/fix-prometheusreceiver-bucketless-nhcb.yaml

@@ -0,0 +1,9 @@
+change_type: bug_fix
+
+component: receiver/prometheus
+
+note: Prevent the Prometheus receiver from dropping classic histograms without explicit bucket boundaries when `convert_classic_histograms_to_nhcb` is enabled and classic histograms are not retained.
+
+issues: [49893]
+
+change_logs: [user]

receiver/prometheusreceiver/internal/metricfamily.go

@@ -104,25 +104,25 @@ func (mg *metricGroup) toDistributionPoint(dest pmetric.HistogramDataPointSlice)
 		return
 	}
 
+	pointIsStale := value.IsStaleNaN(mg.sum) || value.IsStaleNaN(mg.count)
+	// NHCB conversion loses whether _count was absent. Reject the inconsistent
+	// form that identifies a sum-only classic histogram.
+	if mg.isNHCB && !pointIsStale && mg.count == 0 && mg.sum != 0 {
+		return
+	}
+
 	mg.sortPoints()
 
 	var bounds []float64
 	var bucketCounts []uint64
-	pointIsStale := value.IsStaleNaN(mg.sum) || value.IsStaleNaN(mg.count)
 
 	if mg.isNHCB {
 		switch {
 		case mg.hValue != nil:
-			if len(mg.hValue.CustomValues) == 0 {
-				return
-			}
 			bounds = make([]float64, len(mg.hValue.CustomValues))
 			copy(bounds, mg.hValue.CustomValues)
 			bucketCounts = convertNHCBBDeltBuckets(mg.hValue)
 		case mg.fhValue != nil:
-			if len(mg.fhValue.CustomValues) == 0 {
-				return
-			}
 			bounds = make([]float64, len(mg.fhValue.CustomValues))
 			copy(bounds, mg.fhValue.CustomValues)
 			bucketCounts = convertNHCBAbsoluteBuckets(mg.fhValue)

receiver/prometheusreceiver/internal/metricfamily_test.go

@@ -328,6 +328,49 @@ func TestMetricGroupData_toNHCBDistributionUnitTest(t *testing.T) {
 				return point
 			},
 		},
+		{
+			name:                "integer NHCB without explicit bounds",
+			metricName:          "histogram",
+			intervalStartTimeMs: 11,
+			labels:              labels.FromMap(map[string]string{"a": "A", "b": "B"}),
+			integerHistogram: &histogram.Histogram{
+				Schema:          histogram.CustomBucketsSchema,
+				Count:           42,
+				Sum:             123.5,
+				PositiveSpans:   []histogram.Span{{Offset: 0, Length: 1}},
+				PositiveBuckets: []int64{42},
+			},
+			want: func() pmetric.HistogramDataPoint {
+				point := pmetric.NewHistogramDataPoint()
+				point.SetCount(42)
+				point.SetSum(123.5)
+				point.SetTimestamp(pcommon.Timestamp(11 * time.Millisecond))
+				point.BucketCounts().FromRaw([]uint64{42})
+				attributes := point.Attributes()
+				attributes.PutStr("a", "A")
+				attributes.PutStr("b", "B")
+				return point
+			},
+		},
+		{
+			name:                "integer NHCB without explicit bounds with zero count",
+			metricName:          "histogram",
+			intervalStartTimeMs: 11,
+			labels:              labels.FromMap(map[string]string{"a": "A", "b": "B"}),
+			integerHistogram: &histogram.Histogram{
+				Schema: histogram.CustomBucketsSchema,
+			},
+			want: func() pmetric.HistogramDataPoint {
+				point := pmetric.NewHistogramDataPoint()
+				point.SetSum(0)
+				point.SetTimestamp(pcommon.Timestamp(11 * time.Millisecond))
+				point.BucketCounts().FromRaw([]uint64{0})
+				attributes := point.Attributes()
+				attributes.PutStr("a", "A")
+				attributes.PutStr("b", "B")
+				return point
+			},
+		},
 		{
 			name:                "integer NHCB that is stale",
 			metricName:          "histogram",
@@ -351,6 +394,26 @@ func TestMetricGroupData_toNHCBDistributionUnitTest(t *testing.T) {
 				return point
 			},
 		},
+		{
+			name:                "integer NHCB without explicit bounds that is stale",
+			metricName:          "histogram",
+			intervalStartTimeMs: 11,
+			labels:              labels.FromMap(map[string]string{"a": "A", "b": "B"}),
+			integerHistogram: &histogram.Histogram{
+				Schema: histogram.CustomBucketsSchema,
+				Sum:    math.Float64frombits(value.StaleNaN),
+			},
+			want: func() pmetric.HistogramDataPoint {
+				point := pmetric.NewHistogramDataPoint()
+				point.SetTimestamp(pcommon.Timestamp(11 * time.Millisecond))
+				point.SetFlags(pmetric.DefaultDataPointFlags.WithNoRecordedValue(true))
+				point.BucketCounts().FromRaw([]uint64{0})
+				attributes := point.Attributes()
+				attributes.PutStr("a", "A")
+				attributes.PutStr("b", "B")
+				return point
+			},
+		},
 		{
 			name:                "float NHCB",
 			metricName:          "histogram",
@@ -376,6 +439,63 @@ func TestMetricGroupData_toNHCBDistributionUnitTest(t *testing.T) {
 				return point
 			},
 		},
+		{
+			name:                "float NHCB without explicit bounds",
+			metricName:          "histogram",
+			intervalStartTimeMs: 12,
+			labels:              labels.FromMap(map[string]string{"a": "A"}),
+			floatHistogram: &histogram.FloatHistogram{
+				Schema:          histogram.CustomBucketsSchema,
+				Count:           42,
+				Sum:             123.5,
+				PositiveSpans:   []histogram.Span{{Offset: 0, Length: 1}},
+				PositiveBuckets: []float64{42},
+			},
+			want: func() pmetric.HistogramDataPoint {
+				point := pmetric.NewHistogramDataPoint()
+				point.SetCount(42)
+				point.SetSum(123.5)
+				point.SetTimestamp(pcommon.Timestamp(12 * time.Millisecond))
+				point.BucketCounts().FromRaw([]uint64{42})
+				point.Attributes().PutStr("a", "A")
+				return point
+			},
+		},
+		{
+			name:                "float NHCB without explicit bounds with zero count",
+			metricName:          "histogram",
+			intervalStartTimeMs: 12,
+			labels:              labels.FromMap(map[string]string{"a": "A"}),
+			floatHistogram: &histogram.FloatHistogram{
+				Schema: histogram.CustomBucketsSchema,
+			},
+			want: func() pmetric.HistogramDataPoint {
+				point := pmetric.NewHistogramDataPoint()
+				point.SetSum(0)
+				point.SetTimestamp(pcommon.Timestamp(12 * time.Millisecond))
+				point.BucketCounts().FromRaw([]uint64{0})
+				point.Attributes().PutStr("a", "A")
+				return point
+			},
+		},
+		{
+			name:                "float NHCB without explicit bounds that is stale",
+			metricName:          "histogram",
+			intervalStartTimeMs: 12,
+			labels:              labels.FromMap(map[string]string{"a": "A"}),
+			floatHistogram: &histogram.FloatHistogram{
+				Schema: histogram.CustomBucketsSchema,
+				Sum:    math.Float64frombits(value.StaleNaN),
+			},
+			want: func() pmetric.HistogramDataPoint {
+				point := pmetric.NewHistogramDataPoint()
+				point.SetTimestamp(pcommon.Timestamp(12 * time.Millisecond))
+				point.SetFlags(pmetric.DefaultDataPointFlags.WithNoRecordedValue(true))
+				point.BucketCounts().FromRaw([]uint64{0})
+				point.Attributes().PutStr("a", "A")
+				return point
+			},
+		},
 		{
 			name:                "integer NHCB with negative boundaries",
 			metricName:          "histogram",
@@ -427,6 +547,44 @@ func TestMetricGroupData_toNHCBDistributionUnitTest(t *testing.T) {
 	}
 }
 
+func TestMetricGroupData_toNHCBDistributionRejectsNonzeroSumWithZeroCount(t *testing.T) {
+	tests := []struct {
+		name             string
+		integerHistogram *histogram.Histogram
+		floatHistogram   *histogram.FloatHistogram
+	}{
+		{
+			name: "integer NHCB",
+			integerHistogram: &histogram.Histogram{
+				Schema: histogram.CustomBucketsSchema,
+				Sum:    123.5,
+			},
+		},
+		{
+			name: "float NHCB",
+			floatHistogram: &histogram.FloatHistogram{
+				Schema: histogram.CustomBucketsSchema,
+				Sum:    123.5,
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			mp := newMetricFamily("histogram", mc, zap.NewNop(), false, false)
+			lbls := labels.FromMap(map[string]string{"a": "A"})
+			sRef, _ := getSeriesRefWithoutScopeLabels(nil, lbls, mp.mtype)
+
+			err := mp.addNHCBSeries(sRef, "histogram", lbls, 11, tt.integerHistogram, tt.floatHistogram)
+			require.NoError(t, err)
+
+			sl := pmetric.NewMetricSlice()
+			mp.appendMetric(sl, false)
+			require.Zero(t, sl.Len())
+		})
+	}
+}
+
 func TestMetricGroupData_toExponentialDistributionUnitTest(t *testing.T) {
 	type scrape struct {
 		at         int64

receiver/prometheusreceiver/metrics_receiver_test.go

@@ -1471,6 +1471,107 @@ func TestCoreMetricsEndToEnd(t *testing.T) {
 	testComponent(t, targets, nil)
 }
 
+func TestBucketlessHistogramNHCBConversion(t *testing.T) {
+	tests := []struct {
+		name                   string
+		alwaysScrapeClassic    bool
+		expectedHistogramCount int
+	}{
+		{
+			name:                   "classic histogram not retained",
+			alwaysScrapeClassic:    false,
+			expectedHistogramCount: 1,
+		},
+		{
+			name:                   "classic histogram retained",
+			alwaysScrapeClassic:    true,
+			expectedHistogramCount: 2,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			targets := []*testData{
+				{
+					name: "bucketless-histogram",
+					pages: []mockPrometheusResponse{
+						{
+							code: 200,
+							data: `
+# HELP demo_seconds A histogram without explicit buckets.
+# TYPE demo_seconds histogram
+demo_seconds_sum 123.5
+demo_seconds_count 42
+`,
+						},
+					},
+					validateFunc: func(t *testing.T, td *testData, resourceMetrics []pmetric.ResourceMetrics) {
+						verifyNumValidScrapeResults(t, td, resourceMetrics)
+						require.NotEmpty(t, resourceMetrics)
+						require.Equal(t, expectedScrapeMetricCount+tt.expectedHistogramCount, metricsCount(resourceMetrics[0]))
+
+						histogramCount := 0
+						for _, metric := range getMetrics(resourceMetrics[0]) {
+							if metric.Name() != "demo_seconds" {
+								continue
+							}
+							require.Equal(t, pmetric.MetricTypeHistogram, metric.Type())
+							require.Equal(t, 1, metric.Histogram().DataPoints().Len())
+							compareHistogram(42, 123.5, nil, []uint64{42})(t, metric.Histogram().DataPoints().At(0))
+							histogramCount++
+						}
+						require.Equal(t, tt.expectedHistogramCount, histogramCount)
+					},
+				},
+			}
+
+			testComponent(t, targets, nil, func(cfg *PromConfig) {
+				enabled := true
+				alwaysScrapeClassic := tt.alwaysScrapeClassic
+				for _, scrapeConfig := range cfg.ScrapeConfigs {
+					scrapeConfig.ConvertClassicHistogramsToNHCB = &enabled
+					scrapeConfig.AlwaysScrapeClassicHistograms = &alwaysScrapeClassic
+				}
+			})
+		})
+	}
+}
+
+func TestSumOnlyHistogramNHCBConversion(t *testing.T) {
+	targets := []*testData{
+		{
+			name: "sum-only-histogram",
+			pages: []mockPrometheusResponse{
+				{
+					code: 200,
+					data: `
+# HELP demo_seconds A histogram without a count.
+# TYPE demo_seconds histogram
+demo_seconds_sum 123.5
+`,
+				},
+			},
+			validateFunc: func(t *testing.T, td *testData, resourceMetrics []pmetric.ResourceMetrics) {
+				verifyNumValidScrapeResults(t, td, resourceMetrics)
+				require.NotEmpty(t, resourceMetrics)
+				require.Equal(t, expectedScrapeMetricCount, metricsCount(resourceMetrics[0]))
+				for _, metric := range getMetrics(resourceMetrics[0]) {
+					require.NotEqual(t, "demo_seconds", metric.Name())
+				}
+			},
+		},
+	}
+
+	testComponent(t, targets, nil, func(cfg *PromConfig) {
+		enabled := true
+		alwaysScrapeClassic := false
+		for _, scrapeConfig := range cfg.ScrapeConfigs {
+			scrapeConfig.ConvertClassicHistogramsToNHCB = &enabled
+			scrapeConfig.AlwaysScrapeClassicHistograms = &alwaysScrapeClassic
+		}
+	})
+}
+
 // metric type is defined as 'untyped' in the first metric
 // and, type hint is missing in the 2nd metric
 var untypedMetrics = `