Include offending series labels in missing-metric-name push errors

CHANGELOG.md

@@ -47,6 +47,7 @@
 * [ENHANCEMENT] Ring: Cache `ShuffleShardWithLookback` subrings. The cached entry is invalidated on topology change or once `now` reaches the earliest `RegisteredTimestamp + lookbackPeriod` of any included instance. #7628
 * [ENHANCEMENT] Query Frontend: Rename `time_taken` field to `time_taken_ms` and make it return millisecond count. #7649
 * [ENHANCEMENT] Update prometheus alertmanager version to v0.33.0. #7647
+* [ENHANCEMENT] Distributor: Include the offending series labels in "sample missing metric name" push errors so operators can identify which series is missing a metric name. #7657
 * [ENHANCEMENT] Querier/Ingester: Detach ingester series from gRPC buffers to reduce heap. #7670
 * [BUGFIX] Querier: Fix queryWithRetry and labelsWithRetry returning (nil, nil) on cancelled context by propagating ctx.Err(). #7370
 * [BUGFIX] Metrics Helper: Fix non-deterministic bucket order in merged histograms by sorting buckets after map iteration, matching Prometheus client library behavior. #7380

pkg/distributor/distributor_test.go

@@ -2186,7 +2186,7 @@ func TestDistributor_Push_LabelRemoval_RemovingNameLabelWillError(t *testing.T)
 	req := mockWriteRequest([]labels.Labels{tc.inputSeries}, 1, 1, false)
 	_, err = ds[0].Push(ctx, req)
 	require.Error(t, err)
-	assert.Equal(t, "rpc error: code = Code(400) desc = sample missing metric name", err.Error())
+	assert.Equal(t, `rpc error: code = Code(400) desc = sample missing metric name metric: "{cluster=\"one\"}"`, err.Error())
 }
 
 func TestDistributor_Push_ShouldGuaranteeShardingTokenConsistencyOverTheTime(t *testing.T) {

pkg/util/validation/errors.go

@@ -134,14 +134,18 @@ func (e *tooManyLabelsError) Error() string {
 		len(e.series), e.limit, cortexpb.FromLabelAdaptersToMetric(e.series).String())
 }
 
-type noMetricNameError struct{}
+type noMetricNameError struct {
+	series string
+}
 
-func newNoMetricNameError() ValidationError {
-	return &noMetricNameError{}
+func newNoMetricNameError(series []cortexpb.LabelAdapter) ValidationError {
+	return &noMetricNameError{
+		series: formatLabelSet(series),
+	}
 }
 
 func (e *noMetricNameError) Error() string {
-	return "sample missing metric name"
+	return fmt.Sprintf("sample missing metric name metric: %.200q", e.series)
 }
 
 type invalidMetricNameError struct {

pkg/util/validation/validate.go

@@ -285,7 +285,7 @@ func ValidateMetricName(limits *Limits, ls []cortexpb.LabelAdapter, nameValidati
 	}
 	unsafeMetricName, err := extract.UnsafeMetricNameFromLabelAdapters(ls)
 	if err != nil {
-		return newNoMetricNameError(), missingMetricName
+		return newNoMetricNameError(ls), missingMetricName
 	}
 	if !nameValidationScheme.IsValidMetricName(unsafeMetricName) {
 		return newInvalidMetricNameError(unsafeMetricName), invalidMetricName

pkg/util/validation/validate_test.go

@@ -81,7 +81,8 @@ func TestValidateMetricName(t *testing.T) {
 		expectErr error
 		reason    string
 	}{
-		{map[model.LabelName]model.LabelValue{}, newNoMetricNameError(), missingMetricName},
+		{map[model.LabelName]model.LabelValue{}, newNoMetricNameError(cortexpb.FromMetricsToLabelAdapters(model.Metric{})), missingMetricName},
+		{map[model.LabelName]model.LabelValue{"foo": "bar"}, newNoMetricNameError(cortexpb.FromMetricsToLabelAdapters(model.Metric{"foo": "bar"})), missingMetricName},
 		{map[model.LabelName]model.LabelValue{model.MetricNameLabel: ""}, newInvalidMetricNameError(""), invalidMetricName},
 		{map[model.LabelName]model.LabelValue{model.MetricNameLabel: " "}, newInvalidMetricNameError(" "), invalidMetricName},
 		{map[model.LabelName]model.LabelValue{model.MetricNameLabel: "test.\xc5.metric"}, newInvalidMetricNameError("test.\xc5.metric"), invalidMetricName},
@@ -102,6 +103,17 @@ func TestValidateMetricName(t *testing.T) {
 	assert.Empty(t, reason)
 }
 
+func TestNoMetricNameError_ReportsOffendingSeries(t *testing.T) {
+	cfg := new(Limits)
+	cfg.EnforceMetricName = true
+
+	err, reason := ValidateMetricName(cfg, cortexpb.FromMetricsToLabelAdapters(model.Metric{"foo": "bar"}), model.LegacyValidation)
+	require.Error(t, err)
+	assert.Equal(t, missingMetricName, reason)
+	// The error must surface the offending series so operators can identify it (see issue #5802).
+	assert.Equal(t, `sample missing metric name metric: "{foo=\"bar\"}"`, err.Error())
+}
+
 func TestValidateLabels(t *testing.T) {
 	cfg := new(Limits)
 	userID := "testUser"