PRW2: add label validations

Signed-off-by: SungJin1212 <tjdwls1201@gmail.com>

Author: SungJin1212

pkg/util/push/push.go

@@ -2,12 +2,14 @@ package push
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"net/http"
 	"strconv"
 
 	"github.com/go-kit/log/level"
 	"github.com/prometheus/client_golang/exp/api/remote"
+	"github.com/prometheus/common/model"
 	"github.com/prometheus/prometheus/model/labels"
 	writev2 "github.com/prometheus/prometheus/prompb/io/prometheus/write/v2"
 	"github.com/prometheus/prometheus/util/compression"
@@ -165,15 +167,26 @@ func setPRW2RespHeader(w http.ResponseWriter, samples, histograms, exemplars int
 	w.Header().Set(rw20WrittenExemplarsHeader, strconv.FormatInt(exemplars, 10))
 }
 
-func convertV2RequestToV1(req *writev2.Request) (cortexpb.PreallocWriteRequest, error) {
-	var v1Req cortexpb.PreallocWriteRequest
+func convertV2RequestToV1(req *writev2.Request) (v1Req cortexpb.PreallocWriteRequest, err error) {
+	var (
+		badLabelErrs []error
+	)
+
 	v1Timeseries := make([]cortexpb.PreallocTimeseries, 0, len(req.Timeseries))
 	var v1Metadata []*cortexpb.MetricMetadata
 
 	b := labels.NewScratchBuilder(0)
 	symbols := req.Symbols
 	for _, v2Ts := range req.Timeseries {
 		lbs := v2Ts.ToLabels(&b, symbols)
+		// PRW2 spec allows UTF-8.
+		// c.f. https://prometheus.io/docs/specs/prw/remote_write_spec_2_0/#series-labels
+		if !lbs.Has(labels.MetricName) || !lbs.IsValid(model.UTF8Validation) {
+			badLabelErrs = append(badLabelErrs, fmt.Errorf("invalid metric name or labels, got %v", lbs.String()))
+			continue
+		} else if duplicateLabel, hasDuplicate := lbs.HasDuplicateLabelNames(); hasDuplicate {
+			badLabelErrs = append(badLabelErrs, fmt.Errorf("invalid labels for series, labels %v, duplicated label %s", lbs.String(), duplicateLabel))
+		}
 		v1Timeseries = append(v1Timeseries, cortexpb.PreallocTimeseries{
 			TimeSeries: &cortexpb.TimeSeries{
 				Labels:     cortexpb.FromLabelsToLabelAdapters(lbs),
@@ -192,6 +205,10 @@ func convertV2RequestToV1(req *writev2.Request) (cortexpb.PreallocWriteRequest,
 		}
 	}
 
+	if len(badLabelErrs) > 0 {
+		return v1Req, errors.Join(badLabelErrs...)
+	}
+
 	v1Req.Timeseries = v1Timeseries
 	v1Req.Metadata = v1Metadata
 

pkg/util/push/push_test.go

@@ -160,6 +160,57 @@ func Benchmark_convertV2RequestToV1(b *testing.B) {
 	}
 }
 
+func Test_convertV2RequestToV1_InvalidLabels(t *testing.T) {
+	symbols := []string{"", "__name__", "test_metric", "b", "c", "baz", "qux", "d", "e", "foo", "bar", "f", "g", "h", "i", "Test gauge for test purposes", "Maybe op/sec who knows (:", "Test counter for test purposes"}
+
+	tests := []struct {
+		desc           string
+		ts             []writev2.TimeSeries
+		expectedErrMsg string
+	}{
+		{
+			desc: "no metric name",
+			ts: []writev2.TimeSeries{
+				{
+					LabelsRefs: []uint32{2, 3},
+					Samples:    []writev2.Sample{{Value: 2, Timestamp: 2}},
+				},
+			},
+			expectedErrMsg: "invalid metric name or labels, got {test_metric=\"b\"}",
+		},
+		{
+			desc: "no label",
+			ts: []writev2.TimeSeries{
+				{
+					LabelsRefs: []uint32{},
+					Samples:    []writev2.Sample{{Value: 2, Timestamp: 2}},
+				},
+			},
+			expectedErrMsg: "invalid metric name or labels, got {}",
+		},
+		{
+			desc: "duplicate label",
+			ts: []writev2.TimeSeries{
+				{
+					LabelsRefs: []uint32{1, 2, 2, 2, 2, 2},
+					Samples:    []writev2.Sample{{Value: 2, Timestamp: 2}},
+				},
+			},
+			expectedErrMsg: "invalid labels for series, labels {__name__=\"test_metric\", test_metric=\"test_metric\", test_metric=\"test_metric\"}, duplicated label test_metric",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.desc, func(t *testing.T) {
+			var v2Req writev2.Request
+			v2Req.Symbols = symbols
+			v2Req.Timeseries = test.ts
+			_, err := convertV2RequestToV1(&v2Req)
+			require.EqualError(t, err, test.expectedErrMsg)
+		})
+	}
+}
+
 func Test_convertV2RequestToV1(t *testing.T) {
 	var v2Req writev2.Request