Add a metric for when a tenant has structured metadata sanitized at

ingestion time.

Signed-off-by: Callum Styan <callumstyan@gmail.com>

Author: cstyan

pkg/distributor/distributor.go

@@ -167,10 +167,11 @@ type Distributor struct {
 	RequestParserWrapper push.RequestParserWrapper
 
 	// metrics
-	ingesterAppends        *prometheus.CounterVec
-	ingesterAppendTimeouts *prometheus.CounterVec
-	replicationFactor      prometheus.Gauge
-	streamShardCount       prometheus.Counter
+	ingesterAppends                       *prometheus.CounterVec
+	ingesterAppendTimeouts                *prometheus.CounterVec
+	replicationFactor                     prometheus.Gauge
+	streamShardCount                      prometheus.Counter
+	tenantPushSanitizedStructuredMetadata *prometheus.CounterVec
 
 	usageTracker   push.UsageTracker
 	ingesterTasks  chan pushIngesterTask
@@ -284,6 +285,11 @@ func New(
 			Name:      "stream_sharding_count",
 			Help:      "Total number of times the distributor has sharded streams",
 		}),
+		tenantPushSanitizedStructuredMetadata: promauto.With(registerer).NewCounterVec(prometheus.CounterOpts{
+			Namespace: constants.Loki,
+			Name:      "distributor_push_structured_metadata_sanitized_total",
+			Help:      "The total number of times we've had to sanitize structured metadata (names or values) at ingestion time per tenant.",
+		}, []string{"tenant"}),
 		kafkaAppends: promauto.With(registerer).NewCounterVec(prometheus.CounterOpts{
 			Namespace: constants.Loki,
 			Name:      "distributor_kafka_appends_total",
@@ -527,9 +533,14 @@ func (d *Distributor) Push(ctx context.Context, req *logproto.PushRequest) (*log
 					continue
 				}
 
+				var normalized string
 				structuredMetadata := logproto.FromLabelAdaptersToLabels(entry.StructuredMetadata)
 				for i := range entry.StructuredMetadata {
-					structuredMetadata[i].Name = otlptranslate.NormalizeLabel(structuredMetadata[i].Name)
+					normalized = otlptranslate.NormalizeLabel(structuredMetadata[i].Name)
+					if normalized != structuredMetadata[i].Name {
+						structuredMetadata[i].Name = normalized
+						d.tenantPushSanitizedStructuredMetadata.WithLabelValues(tenantID).Inc()
+					}
 					if strings.ContainsRune(structuredMetadata[i].Value, utf8.RuneError) {
 						structuredMetadata[i].Value = strings.Map(removeInvalidUtf, structuredMetadata[i].Value)
 					}

pkg/distributor/distributor_test.go

@@ -3,6 +3,7 @@ package distributor
 import (
 	"context"
 	"fmt"
+	"github.com/prometheus/client_golang/prometheus/testutil"
 	"math"
 	"math/rand"
 	"net/http"
@@ -1961,22 +1962,27 @@ func TestDistributor_StructuredMetadataSanitization(t *testing.T) {
 	for _, tc := range []struct {
 		req              *logproto.PushRequest
 		expectedResponse *logproto.PushResponse
+		numSanitizations float64
 	}{
 		{
 			makeWriteRequestWithLabels(10, 10, []string{`{foo="bar"}`}, true, false, false),
 			success,
+			0,
 		},
 		{
 			makeWriteRequestWithLabels(10, 10, []string{`{foo="bar"}`}, true, true, false),
 			success,
+			10,
 		},
 		{
 			makeWriteRequestWithLabels(10, 10, []string{`{foo="bar"}`}, true, false, true),
 			success,
+			10,
 		},
 		{
 			makeWriteRequestWithLabels(10, 10, []string{`{foo="bar"}`}, true, true, true),
 			success,
+			20,
 		},
 	} {
 		distributors, _ := prepare(t, 1, 5, limits, nil)
@@ -1988,5 +1994,7 @@ func TestDistributor_StructuredMetadataSanitization(t *testing.T) {
 		response, err := distributors[0].Push(ctx, &request)
 		require.NoError(t, err)
 		assert.Equal(t, tc.expectedResponse, response)
+		//fmt.Println("counter: ", distributors[0].tenantPushSanitizedStructuredMetadata.WithLabelValues())
+		assert.Equal(t, tc.numSanitizations, testutil.ToFloat64(distributors[0].tenantPushSanitizedStructuredMetadata.WithLabelValues("test")))
 	}
 }