Push reduce one hash operation of Labels.

Signed-off-by: tanghengjian <[email protected]>

Author: t00350320

CHANGELOG.md

@@ -54,6 +54,7 @@
 * [ENHANCEMENT] Enhance traces with hostname information. #4898
 * [ENHANCEMENT] Improve the documentation around limits. #4905
 * [ENHANCEMENT] Distributor: cache user overrides to reduce lock contention. #4904
+* [ENHANCEMENT] Push reduce one hash operation of Labels. #4945
 * [FEATURE] Compactor: Added `-compactor.block-files-concurrency` allowing to configure number of go routines for download/upload block files during compaction. #4784
 * [FEATURE] Compactor: Added `-compactor.blocks-fetch-concurrency` allowing to configure number of go routines for blocks during compaction. #4787
 * [FEATURE] Compactor: Added configurations for Azure MSI in blocks-storage, ruler-storage and alertmanager-storage. #4818

pkg/ingester/active_series.go

@@ -1,17 +1,12 @@
 package ingester
 
 import (
-	"hash"
 	"math"
 	"sync"
 	"time"
 
-	"github.com/cespare/xxhash"
-	"github.com/prometheus/common/model"
 	"github.com/prometheus/prometheus/model/labels"
 	"go.uber.org/atomic"
-
-	"github.com/cortexproject/cortex/pkg/util"
 )
 
 const (
@@ -53,30 +48,10 @@ func NewActiveSeries() *ActiveSeries {
 }
 
 // Updates series timestamp to 'now'. Function is called to make a copy of labels if entry doesn't exist yet.
-func (c *ActiveSeries) UpdateSeries(series labels.Labels, now time.Time, labelsCopy func(labels.Labels) labels.Labels) {
-	fp := fingerprint(series)
-	stripeID := fp % numActiveSeriesStripes
-
-	c.stripes[stripeID].updateSeriesTimestamp(now, series, fp, labelsCopy)
-}
-
-var sep = []byte{model.SeparatorByte}
-
-var hashPool = sync.Pool{New: func() interface{} { return xxhash.New() }}
-
-func fingerprint(series labels.Labels) uint64 {
-	sum := hashPool.Get().(hash.Hash64)
-	defer hashPool.Put(sum)
-
-	sum.Reset()
-	for _, label := range series {
-		_, _ = sum.Write(util.YoloBuf(label.Name))
-		_, _ = sum.Write(sep)
-		_, _ = sum.Write(util.YoloBuf(label.Value))
-		_, _ = sum.Write(sep)
-	}
+func (c *ActiveSeries) UpdateSeries(series labels.Labels, hash uint64, now time.Time, labelsCopy func(labels.Labels) labels.Labels) {
+	stripeID := hash % numActiveSeriesStripes
 
-	return sum.Sum64()
+	c.stripes[stripeID].updateSeriesTimestamp(now, series, hash, labelsCopy)
 }
 
 // Purge removes expired entries from the cache. This function should be called

pkg/ingester/active_series_test.go

@@ -8,27 +8,33 @@ import (
 	"sync"
 	"testing"
 	"time"
+	"unsafe"
 
 	"github.com/prometheus/prometheus/model/labels"
 	"github.com/stretchr/testify/assert"
 )
 
 func copyFn(l labels.Labels) labels.Labels { return l }
 
+func FromLabelAdaptersToLabels(ls []labels.Label) labels.Labels {
+	return *(*labels.Labels)(unsafe.Pointer(&ls))
+}
+
 func TestActiveSeries_UpdateSeries(t *testing.T) {
 	ls1 := []labels.Label{{Name: "a", Value: "1"}}
 	ls2 := []labels.Label{{Name: "a", Value: "2"}}
 
 	c := NewActiveSeries()
 	assert.Equal(t, 0, c.Active())
-
-	c.UpdateSeries(ls1, time.Now(), copyFn)
+	labels1Hash := FromLabelAdaptersToLabels(ls1).Hash()
+	labels2Hash := FromLabelAdaptersToLabels(ls2).Hash()
+	c.UpdateSeries(ls1, labels1Hash, time.Now(), copyFn)
 	assert.Equal(t, 1, c.Active())
 
-	c.UpdateSeries(ls1, time.Now(), copyFn)
+	c.UpdateSeries(ls1, labels1Hash, time.Now(), copyFn)
 	assert.Equal(t, 1, c.Active())
 
-	c.UpdateSeries(ls2, time.Now(), copyFn)
+	c.UpdateSeries(ls2, labels2Hash, time.Now(), copyFn)
 	assert.Equal(t, 2, c.Active())
 }
 
@@ -46,7 +52,7 @@ func TestActiveSeries_Purge(t *testing.T) {
 		c := NewActiveSeries()
 
 		for i := 0; i < len(series); i++ {
-			c.UpdateSeries(series[i], time.Unix(int64(i), 0), copyFn)
+			c.UpdateSeries(series[i], FromLabelAdaptersToLabels(series[i]).Hash(), time.Unix(int64(i), 0), copyFn)
 		}
 
 		c.Purge(time.Unix(int64(ttl+1), 0))
@@ -62,24 +68,25 @@ func TestActiveSeries_PurgeOpt(t *testing.T) {
 	metric := labels.NewBuilder(labels.FromStrings("__name__", "logs"))
 	ls1 := metric.Set("_", "ypfajYg2lsv").Labels(nil)
 	ls2 := metric.Set("_", "KiqbryhzUpn").Labels(nil)
-
+	labels1Hash := FromLabelAdaptersToLabels(ls1).Hash()
+	labels2Hash := FromLabelAdaptersToLabels(ls2).Hash()
 	c := NewActiveSeries()
 
 	now := time.Now()
-	c.UpdateSeries(ls1, now.Add(-2*time.Minute), copyFn)
-	c.UpdateSeries(ls2, now, copyFn)
+	c.UpdateSeries(ls1, labels1Hash, now.Add(-2*time.Minute), copyFn)
+	c.UpdateSeries(ls2, labels2Hash, now, copyFn)
 	c.Purge(now)
 
 	assert.Equal(t, 1, c.Active())
 
-	c.UpdateSeries(ls1, now.Add(-1*time.Minute), copyFn)
-	c.UpdateSeries(ls2, now, copyFn)
+	c.UpdateSeries(ls1, labels1Hash, now.Add(-1*time.Minute), copyFn)
+	c.UpdateSeries(ls2, labels2Hash, now, copyFn)
 	c.Purge(now)
 
 	assert.Equal(t, 1, c.Active())
 
 	// This will *not* update the series, since there is already newer timestamp.
-	c.UpdateSeries(ls2, now.Add(-1*time.Minute), copyFn)
+	c.UpdateSeries(ls2, labels2Hash, now.Add(-1*time.Minute), copyFn)
 	c.Purge(now)
 
 	assert.Equal(t, 1, c.Active())
@@ -116,7 +123,7 @@ func benchmarkActiveSeriesConcurrencySingleSeries(b *testing.B, goroutines int)
 
 			for ix := 0; ix < max; ix++ {
 				now = now.Add(time.Duration(ix) * time.Millisecond)
-				c.UpdateSeries(series, now, copyFn)
+				c.UpdateSeries(series, FromLabelAdaptersToLabels(series).Hash(), now, copyFn)
 			}
 		}()
 	}
@@ -145,7 +152,7 @@ func BenchmarkActiveSeries_UpdateSeries(b *testing.B) {
 
 	b.ResetTimer()
 	for ix := 0; ix < b.N; ix++ {
-		c.UpdateSeries(series[ix], time.Unix(0, now+int64(ix)), copyFn)
+		c.UpdateSeries(series[ix], FromLabelAdaptersToLabels(series[ix]).Hash(), time.Unix(0, now+int64(ix)), copyFn)
 	}
 }
 
@@ -175,9 +182,9 @@ func benchmarkPurge(b *testing.B, twice bool) {
 		// Prepare series
 		for ix, s := range series {
 			if ix < numExpiresSeries {
-				c.UpdateSeries(s, now.Add(-time.Minute), copyFn)
+				c.UpdateSeries(s, FromLabelAdaptersToLabels(s).Hash(), now.Add(-time.Minute), copyFn)
 			} else {
-				c.UpdateSeries(s, now, copyFn)
+				c.UpdateSeries(s, FromLabelAdaptersToLabels(s).Hash(), now, copyFn)
 			}
 		}
 

pkg/ingester/ingester.go

@@ -963,7 +963,9 @@ func (i *Ingester) Push(ctx context.Context, req *cortexpb.WriteRequest) (*corte
 		// has sorted labels once hit the ingester).
 
 		// Look up a reference for this series.
-		ref, copiedLabels := app.GetRef(cortexpb.FromLabelAdaptersToLabels(ts.Labels))
+		tsLabels := cortexpb.FromLabelAdaptersToLabels(ts.Labels)
+		tsLabelsHash := tsLabels.Hash()
+		ref, copiedLabels := app.GetRef(tsLabels, tsLabelsHash)
 
 		// To find out if any sample was added to this series, we keep old value.
 		oldSucceededSamplesCount := succeededSamplesCount
@@ -1034,7 +1036,7 @@ func (i *Ingester) Push(ctx context.Context, req *cortexpb.WriteRequest) (*corte
 		}
 
 		if i.cfg.ActiveSeriesMetricsEnabled && succeededSamplesCount > oldSucceededSamplesCount {
-			db.activeSeries.UpdateSeries(cortexpb.FromLabelAdaptersToLabels(ts.Labels), startAppend, func(l labels.Labels) labels.Labels {
+			db.activeSeries.UpdateSeries(tsLabels, tsLabelsHash, startAppend, func(l labels.Labels) labels.Labels {
 				// we must already have copied the labels if succeededSamplesCount has been incremented.
 				return copiedLabels
 			})