lib/promscrape: reduce memory usage when scraping targets with big number of metrics alongside targets with small number of labels

Previously targets with big number of metrics and/or labels could generated too big buffers,
which then could be re-used when scraping targets with small number of metrics.
This resulted in memory waste.

Now big buffers are used only for targets with big number of metrics / labels,
while small buffers are used for targets with small number of metrics / labels.

lib/leveledbytebufferpool/pool.go

@@ -19,9 +19,6 @@ var pools [30]sync.Pool
 
 // Get returns byte buffer with the given capacity.
 func Get(capacity int) *bytesutil.ByteBuffer {
-	if capacity <= 0 {
-		capacity = 1
-	}
 	id, capacityNeeded := getPoolIdAndCapacity(capacity)
 	for i := 0; i < 2; i++ {
 		if id < 0 || id >= len(pools) {

lib/promscrape/scrapework.go

@@ -4,6 +4,7 @@ import (
 	"flag"
 	"fmt"
 	"math"
+	"math/bits"
 	"strings"
 	"sync"
 	"time"
@@ -136,6 +137,10 @@ type scrapeWork struct {
 	// prevBodyLen contains the previous response body length for the given scrape work.
 	// It is used as a hint in order to reduce memory usage for body buffers.
 	prevBodyLen int
+
+	// prevLabelsLen contains the number of all the labels generated during the previous scrape.
+	// It is used as a hint in order to reduce memory usage when parsing scrape responses.
+	prevLabelsLen int
 }
 
 func (sw *scrapeWork) run(stopCh <-chan struct{}) {
@@ -212,7 +217,7 @@ func (sw *scrapeWork) scrapeInternal(scrapeTimestamp, realTimestamp int64) error
 	scrapeDuration.Update(duration)
 	scrapeResponseSize.Update(float64(len(body.B)))
 	up := 1
-	wc := writeRequestCtxPool.Get().(*writeRequestCtx)
+	wc := writeRequestCtxPool.Get(sw.prevLabelsLen)
 	if err != nil {
 		up = 0
 		scrapesFailed.Inc()
@@ -241,6 +246,7 @@ func (sw *scrapeWork) scrapeInternal(scrapeTimestamp, realTimestamp int64) error
 	startTime := time.Now()
 	sw.PushData(&wc.writeRequest)
 	pushDataDuration.UpdateDuration(startTime)
+	sw.prevLabelsLen = len(wc.labels)
 	wc.reset()
 	writeRequestCtxPool.Put(wc)
 	// body must be released only after wc is released, since wc refers to body.
@@ -250,6 +256,50 @@ func (sw *scrapeWork) scrapeInternal(scrapeTimestamp, realTimestamp int64) error
 	return err
 }
 
+// leveldWriteRequestCtxPool allows reducing memory usage when writeRequesCtx
+// structs contain mixed number of labels.
+//
+// Its logic has been copied from leveledbytebufferpool.
+type leveledWriteRequestCtxPool struct {
+	pools [30]sync.Pool
+}
+
+func (lwp *leveledWriteRequestCtxPool) Get(labelsCapacity int) *writeRequestCtx {
+	id, capacityNeeded := lwp.getPoolIdAndCapacity(labelsCapacity)
+	for i := 0; i < 2; i++ {
+		if id < 0 || id >= len(lwp.pools) {
+			break
+		}
+		if v := lwp.pools[id].Get(); v != nil {
+			return v.(*writeRequestCtx)
+		}
+		id++
+	}
+	return &writeRequestCtx{
+		labels: make([]prompbmarshal.Label, 0, capacityNeeded),
+	}
+}
+
+func (lwp *leveledWriteRequestCtxPool) Put(wc *writeRequestCtx) {
+	capacity := cap(wc.labels)
+	id, _ := lwp.getPoolIdAndCapacity(capacity)
+	wc.reset()
+	lwp.pools[id].Put(wc)
+}
+
+func (lwp *leveledWriteRequestCtxPool) getPoolIdAndCapacity(size int) (int, int) {
+	size--
+	if size < 0 {
+		size = 0
+	}
+	size >>= 3
+	id := bits.Len(uint(size))
+	if id > len(lwp.pools) {
+		id = len(lwp.pools) - 1
+	}
+	return id, (1 << (id + 3))
+}
+
 type writeRequestCtx struct {
 	rows         parser.Rows
 	writeRequest prompbmarshal.WriteRequest
@@ -264,11 +314,7 @@ func (wc *writeRequestCtx) reset() {
 	wc.samples = wc.samples[:0]
 }
 
-var writeRequestCtxPool = &sync.Pool{
-	New: func() interface{} {
-		return &writeRequestCtx{}
-	},
-}
+var writeRequestCtxPool leveledWriteRequestCtxPool
 
 func (sw *scrapeWork) getSeriesAdded(wc *writeRequestCtx) int {
 	mPrev := sw.prevSeriesMap