fix(cache): Estimate size of posting lists

Author: xqqp

posting/mvcc.go

@@ -370,7 +370,7 @@ func (c *Cache) set(key []byte, i *CachePL) {
 		return
 	}
 	c.numCacheSave.Add(1)
-	c.data.Set(key, i, 1)
+	c.data.Set(key, i, 0)
 }
 
 func (c *Cache) del(key []byte) {
@@ -508,14 +508,18 @@ func initMemoryLayer(cacheSize int64, removeOnUpdate bool) *MemoryLayer {
 	ml.removeOnUpdate = removeOnUpdate
 	ml.statsHolder = NewStatsHolder()
 	if cacheSize > 0 {
-		cache, err := ristretto.NewCache[[]byte, *CachePL](&ristretto.Config[[]byte, *CachePL]{
+		cache, err := ristretto.NewCache(&ristretto.Config[[]byte, *CachePL]{
 			// Use 5% of cache memory for storing counters.
 			NumCounters: int64(float64(cacheSize) * 0.05 * 2),
 			MaxCost:     int64(float64(cacheSize) * 0.95),
 			BufferItems: 16,
 			Metrics:     true,
 			Cost: func(val *CachePL) int64 {
-				return 1
+				itemSize := int64(8)
+				if val.list != nil {
+					itemSize += int64(val.list.ApproximateSize())
+				}
+				return itemSize
 			},
 			ShouldUpdate: func(cur, prev *CachePL) bool {
 				return !(cur.list != nil && prev.list != nil && prev.list.maxTs > cur.list.maxTs)

posting/size.go

@@ -16,6 +16,97 @@ import (
 
 const sizeOfBucket = 144
 
+func (l *List) ApproximateSize() uint64 {
+	if l == nil {
+		return 0
+	}
+
+	l.RLock()
+	defer l.RUnlock()
+
+	var size uint64 = 4*8 + // safe mutex consists of 4 words.
+		1*8 + // plist pointer consists of 1 word.
+		1*8 + // mutation map pointer  consists of 1 word.
+		2*8 + // minTs and maxTs take 1 word each.
+		3*8 + // array take 3 words. so key array is 3 words.
+		3*8 + // array take 3 words. so cache array is 3 words.
+		1*8 // So far 11 words, in order to round the slab we're adding one more word.
+	// so far basic struct layout has been calculated.
+
+	// add byte array memory
+	size += uint64(cap(l.key)) + uint64(cap(l.cache))
+
+	size += approxPostingListSize(l.plist)
+
+	if l.mutationMap != nil {
+		size += l.mutationMap.ApproximateSize()
+	}
+
+	return size
+}
+
+func approxPostingListSize(list *pb.PostingList) uint64 {
+	if list == nil {
+		return 0
+	}
+
+	var size uint64 = 1*8 + // Pack consists of 1 word.
+		3*8 + // Postings array consists of 3 words.
+		1*8 + // CommitTs consists of 1 word.
+		3*8 // Splits array consists of 3 words.
+
+	// add pack size.
+	size += calculatePackSize(list.Pack)
+
+	// Each entry take one word.
+	// Adding each entry reference allocation.
+	size += uint64(cap(list.Postings)) * 8
+	for _, p := range list.Postings {
+		// add the size of each posting.
+		size += calculatePostingSize(p) * uint64(cap(list.Postings))
+		break
+	}
+
+	// Each entry take one word.
+	// Adding each entry size.
+	size += uint64(cap(list.Splits)) * 8
+
+	return size
+}
+
+func (m *MutableLayer) ApproximateSize() uint64 {
+	if m == nil {
+		return 0
+	}
+
+	var size uint64 = 2*8 + // committedEntries and currentEntries take 2 words each.
+		1*8 + // readTs takes 1 word.
+		1*8 + // deleteAllMarker takes 1 word.
+		1*8 + // committedUids takes 1 word.
+		1*8 + // committedUidsTime takes 1 word.
+		1*8 + // length takes 1 word.
+		1*8 + // lastEntry takes 1 word.
+		1*8 + // committedUidsTime takes 1 word.
+		1*8 + // isUidsCalculated takes 1 word.
+		1*8 // calculatedUids takes 1 word.
+	// so far basic struct layout has been calculated.
+
+	// Add each entry size of committedEntries.
+	size += uint64(len(m.committedEntries)) * 8
+	for _, v := range m.committedUids {
+		size += calculatePostingSize(v) * uint64(len(m.committedEntries))
+		break
+	}
+
+	size += approxPostingListSize(m.currentEntries)
+	size += approxPostingListSize(m.lastEntry)
+
+	size += uint64(len(m.currentUids)) * 8
+	size += uint64(cap(m.calculatedUids)) * 8
+
+	return size
+}
+
 // DeepSize computes the memory taken by a Posting List
 func (l *List) DeepSize() uint64 {
 	if l == nil {

worker/server_state.go

@@ -41,7 +41,7 @@ const (
 	ZeroLimitsDefaults = `uid-lease=0; refill-interval=30s; disable-admin-http=false;`
 	GraphQLDefaults    = `introspection=true; debug=false; extensions=true; poll-interval=1s; ` +
 		`lambda-url=;`
-	CacheDefaults        = `size-mb=1024; percentage=40,40,20; remove-on-update=false`
+	CacheDefaults        = `size-mb=4096; percentage=40,40,20; remove-on-update=false`
 	FeatureFlagsDefaults = `normalize-compatibility-mode=; enable-detailed-metrics=false`
 )