wip

lib/logstorage/arena.go

@@ -4,6 +4,7 @@ import (
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/slicesutil"
 )
 
 func getArena() *arena {
@@ -29,8 +30,12 @@ func (a *arena) reset() {
 	a.b = a.b[:0]
 }
 
+func (a *arena) preallocate(n int) {
+	a.b = slicesutil.ExtendCapacity(a.b, n)
+}
+
 func (a *arena) sizeBytes() int {
-	return len(a.b)
+	return cap(a.b)
 }
 
 func (a *arena) copyBytes(b []byte) []byte {
@@ -41,9 +46,8 @@ func (a *arena) copyBytes(b []byte) []byte {
 	ab := a.b
 	abLen := len(ab)
 	ab = append(ab, b...)
-	result := ab[abLen:]
 	a.b = ab
-	return result
+	return ab[abLen:]
 }
 
 func (a *arena) copyBytesToString(b []byte) string {

lib/logstorage/arena_test.go

@@ -35,7 +35,7 @@ func TestArena(t *testing.T) {
 			}
 		}
 
-		if n := a.sizeBytes(); n != valuesLen {
+		if n := a.sizeBytes(); n < valuesLen {
 			t.Fatalf("unexpected arena size; got %d; want %d", n, valuesLen)
 		}
 
@@ -47,7 +47,7 @@ func TestArena(t *testing.T) {
 				t.Fatalf("unexpected len(b); got %d; want %d", len(b), j)
 			}
 			valuesLen += j
-			if n := a.sizeBytes(); n != valuesLen {
+			if n := a.sizeBytes(); n < valuesLen {
 				t.Fatalf("unexpected arena size; got %d; want %d", n, valuesLen)
 			}
 			for k := range b {

lib/logstorage/bitmap.go

@@ -3,6 +3,7 @@ package logstorage
 import (
 	"math/bits"
 	"sync"
+	"unsafe"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/slicesutil"
@@ -37,6 +38,10 @@ func (bm *bitmap) reset() {
 	bm.bitsLen = 0
 }
 
+func (bm *bitmap) sizeBytes() int {
+	return int(unsafe.Sizeof(*bm)) + cap(bm.a)*int(unsafe.Sizeof(bm.a[0]))
+}
+
 func (bm *bitmap) copyFrom(src *bitmap) {
 	bm.reset()
 
@@ -149,7 +154,8 @@ func (bm *bitmap) forEachSetBit(f func(idx int) bool) {
 // forEachSetBitReadonly calls f for each set bit
 func (bm *bitmap) forEachSetBitReadonly(f func(idx int)) {
 	if bm.areAllBitsSet() {
-		for i := range bm.bitsLen {
+		n := bm.bitsLen
+		for i := 0; i < n; i++ {
 			f(i)
 		}
 		return

lib/logstorage/bitmap_timing_test.go

@@ -5,7 +5,7 @@ import (
 )
 
 func BenchmarkBitmapForEachSetBitReadonly(b *testing.B) {
-	const bitsLen = 64*1024
+	const bitsLen = 64 * 1024
 
 	b.Run("no-zero-bits", func(b *testing.B) {
 		bm := getBitmap(bitsLen)
@@ -34,7 +34,7 @@ func BenchmarkBitmapForEachSetBitReadonly(b *testing.B) {
 }
 
 func BenchmarkBitmapForEachSetBit(b *testing.B) {
-	const bitsLen = 64*1024
+	const bitsLen = 64 * 1024
 
 	b.Run("no-zero-bits-noclear", func(b *testing.B) {
 		bm := getBitmap(bitsLen)

lib/logstorage/block_search.go

@@ -168,9 +168,9 @@ func (bs *blockSearch) search(bsw *blockSearchWork) {
 
 	// fetch the requested columns to bs.br.
 	if bs.bsw.so.needAllColumns {
-		bs.br.fetchAllColumns(bs, bm)
+		bs.br.initAllColumns()
 	} else {
-		bs.br.fetchRequestedColumns(bs, bm)
+		bs.br.initRequestedColumns()
 	}
 }
 

lib/logstorage/pipe_sort.go

@@ -232,13 +232,13 @@ func (shard *pipeSortProcessorShard) writeBlock(br *blockResult) {
 			{
 				c: &blockResultColumn{
 					valueType:     valueTypeString,
-					encodedValues: rc.values,
+					valuesEncoded: rc.values,
 				},
 				i64Values: i64Values,
 				f64Values: f64Values,
 			},
 		}
-		shard.stateSizeBudget -= len(rc.buf) + int(unsafe.Sizeof(byColumns[0])+unsafe.Sizeof(*byColumns[0].c))
+		shard.stateSizeBudget -= rc.sizeBytes() + int(unsafe.Sizeof(byColumns[0])+unsafe.Sizeof(*byColumns[0].c))
 
 		// Append br to shard.blocks.
 		shard.blocks = append(shard.blocks, sortBlock{
@@ -260,8 +260,8 @@ func (shard *pipeSortProcessorShard) writeBlock(br *blockResult) {
 				continue
 			}
 			if c.isConst {
-				bc.i64Values = shard.createInt64Values(c.encodedValues)
+				bc.i64Values = shard.createInt64Values(c.valuesEncoded)
-				bc.f64Values = shard.createFloat64Values(c.encodedValues)
+				bc.f64Values = shard.createFloat64Values(c.valuesEncoded)
 				continue
 			}
 
@@ -610,8 +610,8 @@ func sortBlockLess(shardA *pipeSortProcessorShard, rowIdxA int, shardB *pipeSort
 
 		if cA.c.isConst && cB.c.isConst {
 			// Fast path - compare const values
-			ccA := cA.c.encodedValues[0]
+			ccA := cA.c.valuesEncoded[0]
-			ccB := cB.c.encodedValues[0]
+			ccB := cB.c.valuesEncoded[0]
 			if ccA == ccB {
 				continue
 			}

lib/logstorage/pipe_stats.go

@@ -182,14 +182,14 @@ func (shard *pipeStatsProcessorShard) writeBlock(br *blockResult) {
 		c := br.getColumnByName(bf.name)
 		if c.isConst {
 			// Fast path for column with constant value.
-			v := br.getBucketedValue(c.encodedValues[0], bf)
+			v := br.getBucketedValue(c.valuesEncoded[0], bf)
 			shard.keyBuf = encoding.MarshalBytes(shard.keyBuf[:0], bytesutil.ToUnsafeBytes(v))
 			psg := shard.getPipeStatsGroup(shard.keyBuf)
 			shard.stateSizeBudget -= psg.updateStatsForAllRows(br)
 			return
 		}
 
-		values := c.getBucketedValues(br, bf)
+		values := c.getValuesBucketed(br, bf)
 		if areConstValues(values) {
 			// Fast path for column with constant values.
 			shard.keyBuf = encoding.MarshalBytes(shard.keyBuf[:0], bytesutil.ToUnsafeBytes(values[0]))
@@ -216,7 +216,7 @@ func (shard *pipeStatsProcessorShard) writeBlock(br *blockResult) {
 	columnValues := shard.columnValues[:0]
 	for _, bf := range byFields {
 		c := br.getColumnByName(bf.name)
-		values := c.getBucketedValues(br, bf)
+		values := c.getValuesBucketed(br, bf)
 		columnValues = append(columnValues, values)
 	}
 	shard.columnValues = columnValues

lib/logstorage/stats_avg.go

@@ -57,7 +57,7 @@ func (sap *statsAvgProcessor) updateStatsForRow(br *blockResult, rowIdx int) int
 	if sap.sa.containsStar {
 		// Scan all the fields for the given row
 		for _, c := range br.getColumns() {
-			f, ok := c.getFloatValueAtRow(rowIdx)
+			f, ok := c.getFloatValueAtRow(br, rowIdx)
 			if ok {
 				sap.sum += f
 				sap.count++
@@ -67,7 +67,7 @@ func (sap *statsAvgProcessor) updateStatsForRow(br *blockResult, rowIdx int) int
 		// Scan only the given fields for the given row
 		for _, field := range sap.sa.fields {
 			c := br.getColumnByName(field)
-			f, ok := c.getFloatValueAtRow(rowIdx)
+			f, ok := c.getFloatValueAtRow(br, rowIdx)
 			if ok {
 				sap.sum += f
 				sap.count++

lib/logstorage/stats_count.go

@@ -49,7 +49,7 @@ func (scp *statsCountProcessor) updateStatsForAllRows(br *blockResult) int {
 		// Fast path for count(single_column)
 		c := br.getColumnByName(fields[0])
 		if c.isConst {
-			if c.encodedValues[0] != "" {
+			if c.valuesEncoded[0] != "" {
 				scp.rowsCount += uint64(len(br.timestamps))
 			}
 			return 0
@@ -60,7 +60,7 @@ func (scp *statsCountProcessor) updateStatsForAllRows(br *blockResult) int {
 		}
 		switch c.valueType {
 		case valueTypeString:
-			for _, v := range c.encodedValues {
+			for _, v := range c.getValuesEncoded(br) {
 				if v != "" {
 					scp.rowsCount++
 				}
@@ -72,7 +72,7 @@ func (scp *statsCountProcessor) updateStatsForAllRows(br *blockResult) int {
 				scp.rowsCount += uint64(len(br.timestamps))
 				return 0
 			}
-			for _, v := range c.encodedValues {
+			for _, v := range c.getValuesEncoded(br) {
 				if int(v[0]) != zeroDictIdx {
 					scp.rowsCount++
 				}
@@ -95,7 +95,7 @@ func (scp *statsCountProcessor) updateStatsForAllRows(br *blockResult) int {
 	for _, f := range fields {
 		c := br.getColumnByName(f)
 		if c.isConst {
-			if c.encodedValues[0] != "" {
+			if c.valuesEncoded[0] != "" {
 				scp.rowsCount += uint64(len(br.timestamps))
 				return 0
 			}
@@ -105,18 +105,21 @@ func (scp *statsCountProcessor) updateStatsForAllRows(br *blockResult) int {
 			scp.rowsCount += uint64(len(br.timestamps))
 			return 0
 		}
+
 		switch c.valueType {
 		case valueTypeString:
+			valuesEncoded := c.getValuesEncoded(br)
 			bm.forEachSetBit(func(i int) bool {
-				return c.encodedValues[i] == ""
+				return valuesEncoded[i] == ""
 			})
 		case valueTypeDict:
 			if !slices.Contains(c.dictValues, "") {
 				scp.rowsCount += uint64(len(br.timestamps))
 				return 0
 			}
+			valuesEncoded := c.getValuesEncoded(br)
 			bm.forEachSetBit(func(i int) bool {
-				dictIdx := c.encodedValues[i][0]
+				dictIdx := valuesEncoded[i][0]
 				return c.dictValues[dictIdx] == ""
 			})
 		case valueTypeUint8, valueTypeUint16, valueTypeUint32, valueTypeUint64, valueTypeFloat64, valueTypeIPv4, valueTypeTimestampISO8601:
@@ -144,7 +147,7 @@ func (scp *statsCountProcessor) updateStatsForRow(br *blockResult, rowIdx int) i
 		// Fast path for count(single_column)
 		c := br.getColumnByName(fields[0])
 		if c.isConst {
-			if c.encodedValues[0] != "" {
+			if c.valuesEncoded[0] != "" {
 				scp.rowsCount++
 			}
 			return 0
@@ -155,12 +158,14 @@ func (scp *statsCountProcessor) updateStatsForRow(br *blockResult, rowIdx int) i
 		}
 		switch c.valueType {
 		case valueTypeString:
-			if v := c.encodedValues[rowIdx]; v != "" {
+			valuesEncoded := c.getValuesEncoded(br)
+			if v := valuesEncoded[rowIdx]; v != "" {
 				scp.rowsCount++
 			}
 			return 0
 		case valueTypeDict:
-			dictIdx := c.encodedValues[rowIdx][0]
+			valuesEncoded := c.getValuesEncoded(br)
+			dictIdx := valuesEncoded[rowIdx][0]
 			if v := c.dictValues[dictIdx]; v != "" {
 				scp.rowsCount++
 			}

lib/logstorage/stats_count_empty.go

@@ -53,7 +53,7 @@ func (scp *statsCountEmptyProcessor) updateStatsForAllRows(br *blockResult) int
 		// Fast path for count_empty(single_column)
 		c := br.getColumnByName(fields[0])
 		if c.isConst {
-			if c.encodedValues[0] == "" {
+			if c.valuesEncoded[0] == "" {
 				scp.rowsCount += uint64(len(br.timestamps))
 			}
 			return 0
@@ -63,7 +63,7 @@ func (scp *statsCountEmptyProcessor) updateStatsForAllRows(br *blockResult) int
 		}
 		switch c.valueType {
 		case valueTypeString:
-			for _, v := range c.encodedValues {
+			for _, v := range c.getValuesEncoded(br) {
 				if v == "" {
 					scp.rowsCount++
 				}
@@ -74,7 +74,7 @@ func (scp *statsCountEmptyProcessor) updateStatsForAllRows(br *blockResult) int
 			if zeroDictIdx < 0 {
 				return 0
 			}
-			for _, v := range c.encodedValues {
+			for _, v := range c.getValuesEncoded(br) {
 				if int(v[0]) == zeroDictIdx {
 					scp.rowsCount++
 				}
@@ -96,7 +96,7 @@ func (scp *statsCountEmptyProcessor) updateStatsForAllRows(br *blockResult) int
 	for _, f := range fields {
 		c := br.getColumnByName(f)
 		if c.isConst {
-			if c.encodedValues[0] == "" {
+			if c.valuesEncoded[0] == "" {
 				scp.rowsCount += uint64(len(br.timestamps))
 				return 0
 			}
@@ -107,15 +107,17 @@ func (scp *statsCountEmptyProcessor) updateStatsForAllRows(br *blockResult) int
 		}
 		switch c.valueType {
 		case valueTypeString:
+			valuesEncoded := c.getValuesEncoded(br)
 			bm.forEachSetBit(func(i int) bool {
-				return c.encodedValues[i] == ""
+				return valuesEncoded[i] == ""
 			})
 		case valueTypeDict:
 			if !slices.Contains(c.dictValues, "") {
 				return 0
 			}
+			valuesEncoded := c.getValuesEncoded(br)
 			bm.forEachSetBit(func(i int) bool {
-				dictIdx := c.encodedValues[i][0]
+				dictIdx := valuesEncoded[i][0]
 				return c.dictValues[dictIdx] == ""
 			})
 		case valueTypeUint8, valueTypeUint16, valueTypeUint32, valueTypeUint64, valueTypeFloat64, valueTypeIPv4, valueTypeTimestampISO8601:
@@ -145,7 +147,7 @@ func (scp *statsCountEmptyProcessor) updateStatsForRow(br *blockResult, rowIdx i
 		// Fast path for count_empty(single_column)
 		c := br.getColumnByName(fields[0])
 		if c.isConst {
-			if c.encodedValues[0] == "" {
+			if c.valuesEncoded[0] == "" {
 				scp.rowsCount++
 			}
 			return 0
@@ -155,12 +157,14 @@ func (scp *statsCountEmptyProcessor) updateStatsForRow(br *blockResult, rowIdx i
 		}
 		switch c.valueType {
 		case valueTypeString:
-			if v := c.encodedValues[rowIdx]; v == "" {
+			valuesEncoded := c.getValuesEncoded(br)
+			if v := valuesEncoded[rowIdx]; v == "" {
 				scp.rowsCount++
 			}
 			return 0
 		case valueTypeDict:
-			dictIdx := c.encodedValues[rowIdx][0]
+			valuesEncoded := c.getValuesEncoded(br)
+			dictIdx := valuesEncoded[rowIdx][0]
 			if v := c.dictValues[dictIdx]; v == "" {
 				scp.rowsCount++
 			}

lib/logstorage/stats_count_uniq.go

@@ -122,7 +122,7 @@ func (sup *statsCountUniqProcessor) updateStatsForAllRows(br *blockResult) int {
 		}
 		if c.isConst {
 			// count unique const values
-			v := c.encodedValues[0]
+			v := c.valuesEncoded[0]
 			if v == "" {
 				// Do not count empty values
 				return stateSizeIncrease
@@ -156,7 +156,7 @@ func (sup *statsCountUniqProcessor) updateStatsForAllRows(br *blockResult) int {
 			return stateSizeIncrease
 		}
 
-		// Count unique values across encodedValues
+		// Count unique values across values
 		values := c.getValues(br)
 		keyBuf := sup.keyBuf[:0]
 		for i, v := range values {
@@ -278,7 +278,7 @@ func (sup *statsCountUniqProcessor) updateStatsForRow(br *blockResult, rowIdx in
 		}
 		if c.isConst {
 			// count unique const values
-			v := c.encodedValues[0]
+			v := c.valuesEncoded[0]
 			if v == "" {
 				// Do not count empty values
 				return stateSizeIncrease
@@ -295,7 +295,8 @@ func (sup *statsCountUniqProcessor) updateStatsForRow(br *blockResult, rowIdx in
 		}
 		if c.valueType == valueTypeDict {
 			// count unique non-zero c.dictValues
-			dictIdx := c.encodedValues[rowIdx][0]
+			valuesEncoded := c.getValuesEncoded(br)
+			dictIdx := valuesEncoded[rowIdx][0]
 			v := c.dictValues[dictIdx]
 			if v == "" {
 				// Do not count empty values

lib/logstorage/stats_max.go

@@ -108,14 +108,14 @@ func (smp *statsMaxProcessor) updateStateForColumn(br *blockResult, c *blockResu
 	}
 	if c.isConst {
 		// Special case for const column
-		v := c.encodedValues[0]
+		v := c.valuesEncoded[0]
 		smp.updateStateString(v)
 		return
 	}
 
 	switch c.valueType {
 	case valueTypeString:
-		for _, v := range c.encodedValues {
+		for _, v := range c.getValuesEncoded(br) {
 			smp.updateStateString(v)
 		}
 	case valueTypeDict:
@@ -124,23 +124,23 @@ func (smp *statsMaxProcessor) updateStateForColumn(br *blockResult, c *blockResu
 		}
 	case valueTypeUint8, valueTypeUint16, valueTypeUint32, valueTypeUint64:
 		bb := bbPool.Get()
-		bb.B = marshalUint64String(bb.B[:0], c.maxValue)
+		bb.B = marshalUint64String(bb.B[:0], c.ch.maxValue)
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	case valueTypeFloat64:
-		f := math.Float64frombits(c.maxValue)
+		f := math.Float64frombits(c.ch.maxValue)
 		bb := bbPool.Get()
 		bb.B = marshalFloat64String(bb.B[:0], f)
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	case valueTypeIPv4:
 		bb := bbPool.Get()
-		bb.B = marshalIPv4String(bb.B[:0], uint32(c.maxValue))
+		bb.B = marshalIPv4String(bb.B[:0], uint32(c.ch.maxValue))
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	case valueTypeTimestampISO8601:
 		bb := bbPool.Get()
-		bb.B = marshalTimestampISO8601String(bb.B[:0], int64(c.maxValue))
+		bb.B = marshalTimestampISO8601String(bb.B[:0], int64(c.ch.maxValue))
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	default:

lib/logstorage/stats_min.go

@@ -108,14 +108,14 @@ func (smp *statsMinProcessor) updateStateForColumn(br *blockResult, c *blockResu
 	}
 	if c.isConst {
 		// Special case for const column
-		v := c.encodedValues[0]
+		v := c.valuesEncoded[0]
 		smp.updateStateString(v)
 		return
 	}
 
 	switch c.valueType {
 	case valueTypeString:
-		for _, v := range c.encodedValues {
+		for _, v := range c.getValuesEncoded(br) {
 			smp.updateStateString(v)
 		}
 	case valueTypeDict:
@@ -124,23 +124,23 @@ func (smp *statsMinProcessor) updateStateForColumn(br *blockResult, c *blockResu
 		}
 	case valueTypeUint8, valueTypeUint16, valueTypeUint32, valueTypeUint64:
 		bb := bbPool.Get()
-		bb.B = marshalUint64String(bb.B[:0], c.minValue)
+		bb.B = marshalUint64String(bb.B[:0], c.ch.minValue)
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	case valueTypeFloat64:
-		f := math.Float64frombits(c.minValue)
+		f := math.Float64frombits(c.ch.minValue)
 		bb := bbPool.Get()
 		bb.B = marshalFloat64String(bb.B[:0], f)
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	case valueTypeIPv4:
 		bb := bbPool.Get()
-		bb.B = marshalIPv4String(bb.B[:0], uint32(c.minValue))
+		bb.B = marshalIPv4String(bb.B[:0], uint32(c.ch.minValue))
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	case valueTypeTimestampISO8601:
 		bb := bbPool.Get()
-		bb.B = marshalTimestampISO8601String(bb.B[:0], int64(c.minValue))
+		bb.B = marshalTimestampISO8601String(bb.B[:0], int64(c.ch.minValue))
 		smp.updateStateBytes(bb.B)
 		bbPool.Put(bb)
 	default:

lib/logstorage/stats_quantile.go

@@ -2,11 +2,15 @@ package logstorage
 
 import (
 	"fmt"
+	"math"
 	"slices"
 	"strconv"
 	"unsafe"
 
 	"github.com/valyala/fastrand"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 )
 
 type statsQuantile struct {
@@ -38,27 +42,16 @@ type statsQuantileProcessor struct {
 }
 
 func (sqp *statsQuantileProcessor) updateStatsForAllRows(br *blockResult) int {
-	h := &sqp.h
 	stateSizeIncrease := 0
 
 	if sqp.sq.containsStar {
 		for _, c := range br.getColumns() {
-			for _, v := range c.getValues(br) {
+			stateSizeIncrease += sqp.updateStateForColumn(br, c)
-				f, ok := tryParseFloat64(v)
-				if ok {
-					stateSizeIncrease += h.update(f)
-				}
-			}
 		}
 	} else {
 		for _, field := range sqp.sq.fields {
 			c := br.getColumnByName(field)
-			for _, v := range c.getValues(br) {
+			stateSizeIncrease += sqp.updateStateForColumn(br, c)
-				f, ok := tryParseFloat64(v)
-				if ok {
-					stateSizeIncrease += h.update(f)
-				}
-			}
 		}
 	}
 
@@ -71,7 +64,7 @@ func (sqp *statsQuantileProcessor) updateStatsForRow(br *blockResult, rowIdx int
 
 	if sqp.sq.containsStar {
 		for _, c := range br.getColumns() {
-			f, ok := c.getFloatValueAtRow(rowIdx)
+			f, ok := c.getFloatValueAtRow(br, rowIdx)
 			if ok {
 				stateSizeIncrease += h.update(f)
 			}
@@ -79,7 +72,7 @@ func (sqp *statsQuantileProcessor) updateStatsForRow(br *blockResult, rowIdx int
 	} else {
 		for _, field := range sqp.sq.fields {
 			c := br.getColumnByName(field)
-			f, ok := c.getFloatValueAtRow(rowIdx)
+			f, ok := c.getFloatValueAtRow(br, rowIdx)
 			if ok {
 				stateSizeIncrease += h.update(f)
 			}
@@ -89,6 +82,85 @@ func (sqp *statsQuantileProcessor) updateStatsForRow(br *blockResult, rowIdx int
 	return stateSizeIncrease
 }
 
+func (sqp *statsQuantileProcessor) updateStateForColumn(br *blockResult, c *blockResultColumn) int {
+	h := &sqp.h
+	stateSizeIncrease := 0
+
+	if c.isConst {
+		f, ok := tryParseFloat64(c.valuesEncoded[0])
+		if ok {
+			for range br.timestamps {
+				stateSizeIncrease += h.update(f)
+			}
+		}
+		return stateSizeIncrease
+	}
+	if c.isTime {
+		return 0
+	}
+
+	switch c.valueType {
+	case valueTypeString:
+		for _, v := range c.getValues(br) {
+			f, ok := tryParseFloat64(v)
+			if ok {
+				stateSizeIncrease += h.update(f)
+			}
+		}
+	case valueTypeDict:
+		dictValues := c.dictValues
+		a := encoding.GetFloat64s(len(dictValues))
+		for i, v := range dictValues {
+			f, ok := tryParseFloat64(v)
+			if !ok {
+				f = nan
+			}
+			a.A[i] = f
+		}
+		for _, v := range c.getValuesEncoded(br) {
+			idx := v[0]
+			f := a.A[idx]
+			if !math.IsNaN(f) {
+				h.update(f)
+			}
+		}
+		encoding.PutFloat64s(a)
+	case valueTypeUint8:
+		for _, v := range c.getValuesEncoded(br) {
+			n := unmarshalUint8(v)
+			h.update(float64(n))
+		}
+	case valueTypeUint16:
+		for _, v := range c.getValuesEncoded(br) {
+			n := unmarshalUint16(v)
+			h.update(float64(n))
+		}
+	case valueTypeUint32:
+		for _, v := range c.getValuesEncoded(br) {
+			n := unmarshalUint32(v)
+			h.update(float64(n))
+		}
+	case valueTypeUint64:
+		for _, v := range c.getValuesEncoded(br) {
+			n := unmarshalUint64(v)
+			h.update(float64(n))
+		}
+	case valueTypeFloat64:
+		for _, v := range c.getValuesEncoded(br) {
+			f := unmarshalFloat64(v)
+			if !math.IsNaN(f) {
+				h.update(f)
+			}
+		}
+	case valueTypeIPv4:
+	case valueTypeTimestampISO8601:
+	default:
+		logger.Panicf("BUG: unexpected valueType=%d", c.valueType)
+	}
+
+	return stateSizeIncrease
+}
+
 func (sqp *statsQuantileProcessor) mergeState(sfp statsProcessor) {
 	src := sfp.(*statsQuantileProcessor)
 	sqp.h.mergeState(&src.h)

lib/logstorage/stats_sum.go

@@ -38,27 +38,13 @@ func (ssp *statsSumProcessor) updateStatsForAllRows(br *blockResult) int {
 	if ssp.ss.containsStar {
 		// Sum all the columns
 		for _, c := range br.getColumns() {
-			f, count := c.sumValues(br)
+			ssp.updateStateForColumn(br, c)
-			if count > 0 {
-				if math.IsNaN(ssp.sum) {
-					ssp.sum = f
-				} else {
-					ssp.sum += f
-				}
-			}
 		}
 	} else {
 		// Sum the requested columns
 		for _, field := range ssp.ss.fields {
 			c := br.getColumnByName(field)
-			f, count := c.sumValues(br)
+			ssp.updateStateForColumn(br, c)
-			if count > 0 {
-				if math.IsNaN(ssp.sum) {
-					ssp.sum = f
-				} else {
-					ssp.sum += f
-				}
-			}
 		}
 	}
 	return 0
@@ -68,32 +54,39 @@ func (ssp *statsSumProcessor) updateStatsForRow(br *blockResult, rowIdx int) int
 	if ssp.ss.containsStar {
 		// Sum all the fields for the given row
 		for _, c := range br.getColumns() {
-			f, ok := c.getFloatValueAtRow(rowIdx)
+			f, ok := c.getFloatValueAtRow(br, rowIdx)
 			if ok {
-				if math.IsNaN(ssp.sum) {
+				ssp.updateState(f)
-					ssp.sum = f
-				} else {
-					ssp.sum += f
-				}
 			}
 		}
 	} else {
 		// Sum only the given fields for the given row
 		for _, field := range ssp.ss.fields {
 			c := br.getColumnByName(field)
-			f, ok := c.getFloatValueAtRow(rowIdx)
+			f, ok := c.getFloatValueAtRow(br, rowIdx)
 			if ok {
-				if math.IsNaN(ssp.sum) {
+				ssp.updateState(f)
-					ssp.sum = f
-				} else {
-					ssp.sum += f
-				}
 			}
 		}
 	}
 	return 0
 }
 
+func (ssp *statsSumProcessor) updateStateForColumn(br *blockResult, c *blockResultColumn) {
+	f, count := c.sumValues(br)
+	if count > 0 {
+		ssp.updateState(f)
+	}
+}
+
+func (ssp *statsSumProcessor) updateState(f float64) {
+	if math.IsNaN(ssp.sum) {
+		ssp.sum = f
+	} else {
+		ssp.sum += f
+	}
+}
+
 func (ssp *statsSumProcessor) mergeState(sfp statsProcessor) {
 	src := sfp.(*statsSumProcessor)
 	ssp.sum += src.sum

lib/logstorage/stats_uniq_values.go

@@ -68,7 +68,7 @@ func (sup *statsUniqValuesProcessor) updateStatsForAllRowsColumn(c *blockResultC
 	stateSizeIncrease := 0
 	if c.isConst {
 		// collect unique const values
-		v := c.encodedValues[0]
+		v := c.valuesEncoded[0]
 		if v == "" {
 			// skip empty values
 			return stateSizeIncrease
@@ -141,7 +141,7 @@ func (sup *statsUniqValuesProcessor) updateStatsForRowColumn(c *blockResultColum
 	stateSizeIncrease := 0
 	if c.isConst {
 		// collect unique const values
-		v := c.encodedValues[0]
+		v := c.valuesEncoded[0]
 		if v == "" {
 			// skip empty values
 			return stateSizeIncrease
@@ -155,7 +155,8 @@ func (sup *statsUniqValuesProcessor) updateStatsForRowColumn(c *blockResultColum
 	}
 	if c.valueType == valueTypeDict {
 		// collect unique non-zero c.dictValues
-		dictIdx := c.encodedValues[rowIdx][0]
+		valuesEncoded := c.getValuesEncoded(br)
+		dictIdx := valuesEncoded[rowIdx][0]
 		v := c.dictValues[dictIdx]
 		if v == "" {
 			// skip empty values

lib/logstorage/stats_values.go

@@ -61,7 +61,7 @@ func (svp *statsValuesProcessor) updateStatsForAllRows(br *blockResult) int {
 func (svp *statsValuesProcessor) updateStatsForAllRowsColumn(c *blockResultColumn, br *blockResult) int {
 	stateSizeIncrease := 0
 	if c.isConst {
-		v := strings.Clone(c.encodedValues[0])
+		v := strings.Clone(c.valuesEncoded[0])
 		stateSizeIncrease += len(v)
 
 		values := svp.values
@@ -81,7 +81,7 @@ func (svp *statsValuesProcessor) updateStatsForAllRowsColumn(c *blockResultColum
 		}
 
 		values := svp.values
-		for _, encodedValue := range c.encodedValues {
+		for _, encodedValue := range c.getValuesEncoded(br) {
 			idx := encodedValue[0]
 			values = append(values, dictValues[idx])
 		}
@@ -128,7 +128,7 @@ func (svp *statsValuesProcessor) updateStatsForRow(br *blockResult, rowIdx int)
 func (svp *statsValuesProcessor) updateStatsForRowColumn(c *blockResultColumn, br *blockResult, rowIdx int) int {
 	stateSizeIncrease := 0
 	if c.isConst {
-		v := strings.Clone(c.encodedValues[0])
+		v := strings.Clone(c.valuesEncoded[0])
 		stateSizeIncrease += len(v)
 
 		svp.values = append(svp.values, v)
@@ -138,7 +138,8 @@ func (svp *statsValuesProcessor) updateStatsForRowColumn(c *blockResultColumn, b
 	}
 	if c.valueType == valueTypeDict {
 		// collect unique non-zero c.dictValues
-		dictIdx := c.encodedValues[rowIdx][0]
+		valuesEncoded := c.getValuesEncoded(br)
+		dictIdx := valuesEncoded[rowIdx][0]
 		v := strings.Clone(c.dictValues[dictIdx])
 		stateSizeIncrease += len(v)