Use slice pooling to populate the query stream response

pkg/ingester/ingester.go

@@ -33,6 +33,7 @@ import (
 	"github.com/prometheus/prometheus/tsdb/chunkenc"
 	"github.com/prometheus/prometheus/tsdb/chunks"
 	"github.com/prometheus/prometheus/tsdb/wlog"
+	"github.com/prometheus/prometheus/util/zeropool"
 	"github.com/thanos-io/objstore"
 	"github.com/thanos-io/thanos/pkg/block/metadata"
 	"github.com/thanos-io/thanos/pkg/shipper"
@@ -95,6 +96,8 @@ const (
 var (
 	errExemplarRef      = errors.New("exemplars not ingested because series not already present")
 	errIngesterStopping = errors.New("ingester stopping")
+
+	tsChunksPool zeropool.Pool[[]client.TimeSeriesChunk]
 )
 
 // Config for an Ingester.
@@ -2045,7 +2048,8 @@ func (i *Ingester) queryStreamChunks(ctx context.Context, db *userTSDB, from, th
 		return 0, 0, 0, 0, ss.Err()
 	}
 
-	chunkSeries := make([]client.TimeSeriesChunk, 0, queryStreamBatchSize)
+	chunkSeries := getTimeSeriesChunksSlice()
+	defer putTimeSeriesChunksSlice(chunkSeries)
 	batchSizeBytes := 0
 	var it chunks.Iterator
 	for ss.Next() {
@@ -3062,6 +3066,31 @@ func (i *Ingester) ModeHandler(w http.ResponseWriter, r *http.Request) {
 	_, _ = w.Write([]byte(respMsg))
 }
 
+func (i *Ingester) getInstanceLimits() *InstanceLimits {
+	// Don't apply any limits while starting. We especially don't want to apply series in memory limit while replaying WAL.
+	if i.State() == services.Starting {
+		return nil
+	}
+
+	if i.cfg.InstanceLimitsFn == nil {
+		return defaultInstanceLimits
+	}
+
+	l := i.cfg.InstanceLimitsFn()
+	if l == nil {
+		return defaultInstanceLimits
+	}
+
+	return l
+}
+
+// stopIncomingRequests is called during the shutdown process.
+func (i *Ingester) stopIncomingRequests() {
+	i.stoppedMtx.Lock()
+	defer i.stoppedMtx.Unlock()
+	i.stopped = true
+}
+
 // metadataQueryRange returns the best range to query for metadata queries based on the timerange in the ingester.
 func metadataQueryRange(queryStart, queryEnd int64, db *userTSDB, queryIngestersWithin time.Duration) (mint, maxt int64, err error) {
 	if queryIngestersWithin > 0 {
@@ -3119,27 +3148,16 @@ func wrappedTSDBIngestExemplarErr(ingestErr error, timestamp model.Time, seriesL
 	)
 }
 
-func (i *Ingester) getInstanceLimits() *InstanceLimits {
-	// Don't apply any limits while starting. We especially don't want to apply series in memory limit while replaying WAL.
-	if i.State() == services.Starting {
-		return nil
-	}
-
-	if i.cfg.InstanceLimitsFn == nil {
-		return defaultInstanceLimits
+func getTimeSeriesChunksSlice() []client.TimeSeriesChunk {
+	if p := tsChunksPool.Get(); p != nil {
+		return p
 	}
 
-	l := i.cfg.InstanceLimitsFn()
-	if l == nil {
-		return defaultInstanceLimits
-	}
-
-	return l
+	return make([]client.TimeSeriesChunk, 0, queryStreamBatchSize)
 }
 
-// stopIncomingRequests is called during the shutdown process.
-func (i *Ingester) stopIncomingRequests() {
-	i.stoppedMtx.Lock()
-	defer i.stoppedMtx.Unlock()
-	i.stopped = true
+func putTimeSeriesChunksSlice(p []client.TimeSeriesChunk) {
+	if p != nil {
+		tsChunksPool.Put(p[:0])
+	}
 }

pkg/ingester/ingester_test.go

@@ -3424,10 +3424,25 @@ func (m *mockQueryStreamServer) Context() context.Context {
 }
 
 func BenchmarkIngester_QueryStream_Chunks(b *testing.B) {
-	benchmarkQueryStream(b)
+	tc := []struct {
+		samplesCount, seriesCount int
+	}{
+		{samplesCount: 10, seriesCount: 10},
+		{samplesCount: 10, seriesCount: 50},
+		{samplesCount: 10, seriesCount: 100},
+		{samplesCount: 50, seriesCount: 10},
+		{samplesCount: 50, seriesCount: 50},
+		{samplesCount: 50, seriesCount: 100},
+	}
+
+	for _, c := range tc {
+		b.Run(fmt.Sprintf("samplesCount=%v; seriesCount=%v", c.samplesCount, c.seriesCount), func(b *testing.B) {
+			benchmarkQueryStream(b, c.samplesCount, c.seriesCount)
+		})
+	}
 }
 
-func benchmarkQueryStream(b *testing.B) {
+func benchmarkQueryStream(b *testing.B, samplesCount, seriesCount int) {
 	cfg := defaultIngesterTestConfig(b)
 
 	// Create ingester.
@@ -3444,7 +3459,6 @@ func benchmarkQueryStream(b *testing.B) {
 	// Push series.
 	ctx := user.InjectOrgID(context.Background(), userID)
 
-	const samplesCount = 1000
 	samples := make([]cortexpb.Sample, 0, samplesCount)
 
 	for i := 0; i < samplesCount; i++ {
@@ -3454,15 +3468,14 @@ func benchmarkQueryStream(b *testing.B) {
 		})
 	}
 
-	const seriesCount = 100
 	for s := 0; s < seriesCount; s++ {
 		_, err = i.Push(ctx, writeRequestSingleSeries(labels.Labels{{Name: labels.MetricName, Value: "foo"}, {Name: "l", Value: strconv.Itoa(s)}}, samples))
 		require.NoError(b, err)
 	}
 
 	req := &client.QueryRequest{
 		StartTimestampMs: 0,
-		EndTimestampMs:   samplesCount + 1,
+		EndTimestampMs:   int64(samplesCount + 1),
 
 		Matchers: []*client.LabelMatcher{{
 			Type:  client.EQUAL,
@@ -3474,6 +3487,7 @@ func benchmarkQueryStream(b *testing.B) {
 	mockStream := &mockQueryStreamServer{ctx: ctx}
 
 	b.ResetTimer()
+	b.ReportAllocs()
 
 	for ix := 0; ix < b.N; ix++ {
 		err := i.QueryStream(req, mockStream)