generate ec01~ec14, generate ecx file with sorted needle values

weed/storage/erasure_coding/ec_encoder.go

@@ -1,9 +1,11 @@
 package erasure_coding
 
 import (
+	"fmt"
 	"io"
 	"os"
 
+	"github.com/chrislusf/seaweedfs/weed/glog"
 	"github.com/klauspost/reedsolomon"
 )
 
@@ -14,7 +16,46 @@ const (
 	ErasureCodingSmallBlockSize = 1024 * 1024        // 1MB
 )
 
-func encodeData(file *os.File, enc reedsolomon.Encoder, startOffset, blockSize int64, buffers [][]byte) error {
+func encodeData(file *os.File, enc reedsolomon.Encoder, startOffset, blockSize int64, buffers [][]byte, outputs []*os.File) error {
+
+	bufferSize := int64(len(buffers[0]))
+	batchCount := blockSize/bufferSize
+	if blockSize%bufferSize!=0 {
+		glog.Fatalf("unexpected block size %d buffer size %d", blockSize, bufferSize)
+	}
+
+	for b := int64(0); b < batchCount; b++ {
+		err := encodeDataOneBatch(file, enc, startOffset+b*bufferSize, blockSize, buffers, outputs)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+func openEcFiles(baseFileName string) (files []*os.File, err error){
+	for i := 0; i< DataShardsCount+ParityShardsCount; i++{
+		fname := fmt.Sprintf("%s.ec%02d", baseFileName, i+1)
+		f, err := os.OpenFile(fname, os.O_TRUNC|os.O_CREATE|os.O_WRONLY, 0644)
+		if err != nil {
+			return files, fmt.Errorf("failed to open file %s: %v", fname, err)
+		}
+		files = append(files, f)
+	}
+	return
+}
+
+func closeEcFiles(files []*os.File){
+	for _, f := range files{
+		if f != nil {
+			f.Close()
+		}
+	}
+}
+
+
+func encodeDataOneBatch(file *os.File, enc reedsolomon.Encoder, startOffset, blockSize int64, buffers [][]byte, outputs []*os.File) error {
 
 	// read data into buffers
 	for i := 0; i < DataShardsCount; i++ {
@@ -36,5 +77,12 @@ func encodeData(file *os.File, enc reedsolomon.Encoder, startOffset, blockSize i
 		return err
 	}
 
+	for i := 0; i < DataShardsCount+ParityShardsCount; i++ {
+		_, err := outputs[i].Write(buffers[i])
+		if err != nil {
+			return err
+		}
+	}
+
 	return nil
 }

weed/storage/erasure_coding/ec_test.go

@@ -1,9 +1,13 @@
 package erasure_coding
 
 import (
+	"fmt"
 	"os"
 	"testing"
 
+	"github.com/chrislusf/seaweedfs/weed/storage"
+	"github.com/chrislusf/seaweedfs/weed/storage/needle_map"
+	"github.com/chrislusf/seaweedfs/weed/storage/types"
 	"github.com/klauspost/reedsolomon"
 )
 
@@ -11,8 +15,9 @@ func TestEncodingDecoding(t *testing.T) {
 	largeBlockSize := int64(10000)
 	smallBlockSize := int64(100)
 	bufferSize := 50
+	baseFileName := "1"
 
-	file, err := os.OpenFile("1.dat", os.O_RDONLY, 0)
+	file, err := os.OpenFile(baseFileName+".dat", os.O_RDONLY, 0)
 	if err != nil {
 		t.Logf("failed to open dat file: %v", err)
 	}
@@ -22,31 +27,95 @@ func TestEncodingDecoding(t *testing.T) {
 		t.Logf("failed to stat dat file: %v", err)
 	}
 
-	remainingSize := fi.Size()
-	var processedSize int64
+	err = encodeDatFile(fi.Size(), err, baseFileName, bufferSize, largeBlockSize, file, smallBlockSize)
+	if err != nil {
+		t.Logf("failed to stat dat file: %v", err)
+	}
+	file.Close()
 
-	enc, err := reedsolomon.New(DataShardsCount, ParityShardsCount)
+	err = writeSortedEcxFiles(baseFileName)
 	if err != nil {
-		t.Logf("failed to create encoder: %v", err)
+		t.Logf("writeSortedEcxFiles: %v", err)
 	}
 
-	buffers := make([][]byte, DataShardsCount+ParityShardsCount)
+	err = validateFiles(baseFileName)
+	if err != nil {
+		t.Logf("writeSortedEcxFiles: %v", err)
+	}
 
+}
+
+func encodeDatFile(remainingSize int64, err error, baseFileName string, bufferSize int, largeBlockSize int64, file *os.File, smallBlockSize int64) error {
+	var processedSize int64
+	enc, err := reedsolomon.New(DataShardsCount, ParityShardsCount)
+	if err != nil {
+		return fmt.Errorf("failed to create encoder: %v", err)
+	}
+	buffers := make([][]byte, DataShardsCount+ParityShardsCount)
+	outputs, err := openEcFiles(baseFileName)
+	defer closeEcFiles(outputs)
+	if err != nil {
+		return fmt.Errorf("failed to open dat file: %v", err)
+	}
 	for i, _ := range buffers {
 		buffers[i] = make([]byte, bufferSize)
 	}
-
 	for remainingSize > largeBlockSize*DataShardsCount {
-		encodeData(file, enc, processedSize, largeBlockSize, buffers)
+		err = encodeData(file, enc, processedSize, largeBlockSize, buffers, outputs)
+		if err != nil {
+			return fmt.Errorf("failed to encode large chunk data: %v", err)
+		}
 		remainingSize -= largeBlockSize * DataShardsCount
 		processedSize += largeBlockSize * DataShardsCount
 	}
-
 	for remainingSize > 0 {
-		encodeData(file, enc, processedSize, smallBlockSize, buffers)
+		encodeData(file, enc, processedSize, smallBlockSize, buffers, outputs)
+		if err != nil {
+			return fmt.Errorf("failed to encode small chunk data: %v", err)
+		}
 		remainingSize -= smallBlockSize * DataShardsCount
 		processedSize += smallBlockSize * DataShardsCount
 	}
+	return nil
+}
+
+func writeSortedEcxFiles(baseFileName string) (e error) {
+
+	var indexFile *os.File
+	if indexFile, e = os.OpenFile(baseFileName+".idx", os.O_RDONLY, 0644); e != nil {
+		return fmt.Errorf("cannot read Volume Index %s.idx: %v", baseFileName, e)
+	}
+
+	cm := needle_map.NewCompactMap()
+	storage.WalkIndexFile(indexFile, func(key types.NeedleId, offset types.Offset, size uint32) error {
+		if !offset.IsZero() && size != types.TombstoneFileSize {
+			cm.Set(key, offset, size)
+		} else {
+			cm.Delete(key)
+		}
+		return nil
+	})
+
+	ecxFile, err := os.OpenFile(baseFileName+".ecx", os.O_TRUNC|os.O_CREATE|os.O_WRONLY, 0644)
+	if err != nil {
+		return fmt.Errorf("failed to open dat file: %v", err)
+	}
+	defer ecxFile.Close()
+
+	err = cm.AscendingVisit(func(value needle_map.NeedleValue) error {
+		bytes := value.ToBytes()
+		_, writeErr := ecxFile.Write(bytes)
+		return writeErr
+	})
+
+	if err != nil {
+		return fmt.Errorf("failed to open dat file: %v", err)
+	}
 
+	return nil
 }
 
+func validateFiles(baseFileName string) error {
+	return nil
+
+}

weed/storage/needle_map.go

@@ -62,10 +62,7 @@ func IdxFileEntry(bytes []byte) (key NeedleId, offset Offset, size uint32) {
 	return
 }
 func (nm *baseNeedleMapper) appendToIndexFile(key NeedleId, offset Offset, size uint32) error {
-	bytes := make([]byte, NeedleIdSize+OffsetSize+SizeSize)
-	NeedleIdToBytes(bytes[0:NeedleIdSize], key)
-	OffsetToBytes(bytes[NeedleIdSize:NeedleIdSize+OffsetSize], offset)
-	util.Uint32toBytes(bytes[NeedleIdSize+OffsetSize:NeedleIdSize+OffsetSize+SizeSize], size)
+	bytes := needle_map.ToBytes(key, offset, size)
 
 	nm.indexFileAccessLock.Lock()
 	defer nm.indexFileAccessLock.Unlock()
@@ -76,6 +73,7 @@ func (nm *baseNeedleMapper) appendToIndexFile(key NeedleId, offset Offset, size
 	_, err := nm.indexFile.Write(bytes)
 	return err
 }
+
 func (nm *baseNeedleMapper) IndexFileContent() ([]byte, error) {
 	nm.indexFileAccessLock.Lock()
 	defer nm.indexFileAccessLock.Unlock()

weed/storage/needle_map/btree_map.go

@@ -43,7 +43,7 @@ func (cm *BtreeMap) Get(key NeedleId) (*NeedleValue, bool) {
 }
 
 // Visit visits all entries or stop if any error when visiting
-func (cm *BtreeMap) Visit(visit func(NeedleValue) error) (ret error) {
+func (cm *BtreeMap) AscendingVisit(visit func(NeedleValue) error) (ret error) {
 	cm.tree.Ascend(func(item btree.Item) bool {
 		needle := item.(NeedleValue)
 		ret = visit(needle)

weed/storage/needle_map/compact_map.go

@@ -244,24 +244,37 @@ func (cm *CompactMap) binarySearchCompactSection(key NeedleId) int {
 }
 
 // Visit visits all entries or stop if any error when visiting
-func (cm *CompactMap) Visit(visit func(NeedleValue) error) error {
+func (cm *CompactMap) AscendingVisit(visit func(NeedleValue) error) error {
 	for _, cs := range cm.list {
 		cs.RLock()
-		for i, v := range cs.overflow {
-			if err := visit(toNeedleValue(cs.overflowExtra[i], v, cs)); err != nil {
+		var i, j int
+		for i, j = 0, 0; i < len(cs.overflow) && j < len(cs.values) && j<cs.counter; {
+			if cs.overflow[i].Key < cs.values[j].Key {
+				if err := visit(toNeedleValue(cs.overflowExtra[i], cs.overflow[i], cs)); err != nil {
+					cs.RUnlock()
+					return err
+				}
+				i++
+			}else if cs.overflow[i].Key == cs.values[j].Key {
+				j++
+			}else{
+				if err := visit(toNeedleValue(cs.valuesExtra[j], cs.values[j], cs)); err != nil {
+					cs.RUnlock()
+					return err
+				}
+				j++
+			}
+		}
+		for ;i < len(cs.overflow);i++{
+			if err := visit(toNeedleValue(cs.overflowExtra[i], cs.overflow[i], cs)); err != nil {
 				cs.RUnlock()
 				return err
 			}
 		}
-		for i, v := range cs.values {
-			if i >= cs.counter {
-				break
-			}
-			if _, _, found := cs.findOverflowEntry(v.Key); !found {
-				if err := visit(toNeedleValue(cs.valuesExtra[i], v, cs)); err != nil {
-					cs.RUnlock()
-					return err
-				}
+		for ; j < len(cs.values)&& j<cs.counter;j++{
+			if err := visit(toNeedleValue(cs.valuesExtra[j], cs.values[j], cs)); err != nil {
+				cs.RUnlock()
+				return err
 			}
 		}
 		cs.RUnlock()
@@ -279,10 +292,10 @@ func toNeedleValue(snve SectionalNeedleValueExtra, snv SectionalNeedleValue, cs
 
 func (nv NeedleValue) toSectionalNeedleValue(cs *CompactSection) (SectionalNeedleValue, SectionalNeedleValueExtra) {
 	return SectionalNeedleValue{
-			SectionalNeedleId(nv.Key - cs.start),
-			nv.Offset.OffsetLower,
-			nv.Size,
-		}, SectionalNeedleValueExtra{
-			nv.Offset.OffsetHigher,
-		}
+		SectionalNeedleId(nv.Key - cs.start),
+		nv.Offset.OffsetLower,
+		nv.Size,
+	}, SectionalNeedleValueExtra{
+		nv.Offset.OffsetHigher,
+	}
 }

weed/storage/needle_map/compact_map_test.go

@@ -19,7 +19,7 @@ func TestOverflow2(t *testing.T) {
 	m.Set(NeedleId(150158), ToOffset(8), 3000073)
 	m.Set(NeedleId(150162), ToOffset(8), 3000073)
 
-	m.Visit(func(value NeedleValue) error {
+	m.AscendingVisit(func(value NeedleValue) error {
 		println("needle key:", value.Key)
 		return nil
 	})

weed/storage/needle_map/needle_value.go

@@ -2,6 +2,7 @@ package needle_map
 
 import (
 	. "github.com/chrislusf/seaweedfs/weed/storage/types"
+	"github.com/chrislusf/seaweedfs/weed/util"
 	"github.com/google/btree"
 )
 
@@ -15,3 +16,15 @@ func (this NeedleValue) Less(than btree.Item) bool {
 	that := than.(NeedleValue)
 	return this.Key < that.Key
 }
+
+func (nv NeedleValue) ToBytes() []byte {
+	return ToBytes(nv.Key, nv.Offset, nv.Size)
+}
+
+func ToBytes(key NeedleId, offset Offset, size uint32) []byte {
+	bytes := make([]byte, NeedleIdSize+OffsetSize+SizeSize)
+	NeedleIdToBytes(bytes[0:NeedleIdSize], key)
+	OffsetToBytes(bytes[NeedleIdSize:NeedleIdSize+OffsetSize], offset)
+	util.Uint32toBytes(bytes[NeedleIdSize+OffsetSize:NeedleIdSize+OffsetSize+SizeSize], size)
+	return bytes
+}

weed/storage/needle_map/needle_value_map.go

@@ -8,5 +8,5 @@ type NeedleValueMap interface {
 	Set(key NeedleId, offset Offset, size uint32) (oldOffset Offset, oldSize uint32)
 	Delete(key NeedleId) uint32
 	Get(key NeedleId) (*NeedleValue, bool)
-	Visit(visit func(NeedleValue) error) error
+	AscendingVisit(visit func(NeedleValue) error) error
 }