seaweedfs/weed/storage/erasure_coding/ec_test.go

211 lines
5.3 KiB
Go
Raw Normal View History

2019-05-15 08:02:00 +00:00
package erasure_coding
import (
2019-05-19 10:01:58 +00:00
"bytes"
"fmt"
"math/rand"
2019-05-15 08:02:00 +00:00
"os"
"testing"
"github.com/klauspost/reedsolomon"
"github.com/seaweedfs/seaweedfs/weed/storage/needle_map"
"github.com/seaweedfs/seaweedfs/weed/storage/types"
2019-05-15 08:02:00 +00:00
)
2019-05-19 10:01:58 +00:00
const (
largeBlockSize = 10000
smallBlockSize = 100
)
2019-05-15 08:02:00 +00:00
func TestEncodingDecoding(t *testing.T) {
bufferSize := 50
baseFileName := "1"
2019-05-15 08:02:00 +00:00
2019-05-19 10:01:58 +00:00
err := generateEcFiles(baseFileName, bufferSize, largeBlockSize, smallBlockSize)
2019-05-15 08:02:00 +00:00
if err != nil {
2019-05-19 10:01:58 +00:00
t.Logf("generateEcFiles: %v", err)
2019-05-15 08:02:00 +00:00
}
err = WriteSortedFileFromIdx(baseFileName, ".ecx")
2019-05-15 08:02:00 +00:00
if err != nil {
t.Logf("WriteSortedFileFromIdx: %v", err)
2019-05-15 08:02:00 +00:00
}
2019-05-19 10:01:58 +00:00
err = validateFiles(baseFileName)
if err != nil {
t.Logf("WriteSortedFileFromIdx: %v", err)
}
2019-05-15 08:02:00 +00:00
2019-05-20 04:16:01 +00:00
removeGeneratedFiles(baseFileName)
2019-05-19 10:01:58 +00:00
}
func validateFiles(baseFileName string) error {
2020-02-05 05:16:34 +00:00
nm, err := readNeedleMap(baseFileName)
2019-05-19 10:01:58 +00:00
if err != nil {
2020-02-05 05:16:34 +00:00
return fmt.Errorf("readNeedleMap: %v", err)
2019-05-19 10:01:58 +00:00
}
defer nm.Close()
2019-05-19 10:01:58 +00:00
datFile, err := os.OpenFile(baseFileName+".dat", os.O_RDONLY, 0)
if err != nil {
return fmt.Errorf("failed to open dat file: %v", err)
}
defer datFile.Close()
fi, err := datFile.Stat()
if err != nil {
return fmt.Errorf("failed to stat dat file: %v", err)
}
ecFiles, err := openEcFiles(baseFileName, true)
if err != nil {
return fmt.Errorf("error opening ec files: %w", err)
}
2019-05-19 10:01:58 +00:00
defer closeEcFiles(ecFiles)
2020-02-05 05:16:34 +00:00
err = nm.AscendingVisit(func(value needle_map.NeedleValue) error {
2019-05-19 10:01:58 +00:00
return assertSame(datFile, fi.Size(), ecFiles, value.Offset, value.Size)
})
if err != nil {
return fmt.Errorf("failed to check ec files: %v", err)
}
return nil
}
func assertSame(datFile *os.File, datSize int64, ecFiles []*os.File, offset types.Offset, size types.Size) error {
2019-05-19 10:01:58 +00:00
data, err := readDatFile(datFile, offset, size)
if err != nil {
return fmt.Errorf("failed to read dat file: %v", err)
}
ecData, err := readEcFile(datSize, ecFiles, offset, size)
if err != nil {
return fmt.Errorf("failed to read ec file: %v", err)
}
if bytes.Compare(data, ecData) != 0 {
return fmt.Errorf("unexpected data read")
}
return nil
2019-05-19 10:01:58 +00:00
}
func readDatFile(datFile *os.File, offset types.Offset, size types.Size) ([]byte, error) {
2019-05-19 10:01:58 +00:00
data := make([]byte, size)
n, err := datFile.ReadAt(data, offset.ToActualOffset())
2019-05-19 10:01:58 +00:00
if err != nil {
return nil, fmt.Errorf("failed to ReadAt dat file: %v", err)
}
if n != int(size) {
return nil, fmt.Errorf("unexpected read size %d, expected %d", n, size)
}
return data, nil
}
func readEcFile(datSize int64, ecFiles []*os.File, offset types.Offset, size types.Size) (data []byte, err error) {
2019-05-19 10:01:58 +00:00
intervals := LocateData(largeBlockSize, smallBlockSize, datSize, offset.ToActualOffset(), size)
for i, interval := range intervals {
2019-05-27 18:59:03 +00:00
if d, e := readOneInterval(interval, ecFiles); e != nil {
return nil, e
} else {
if i == 0 {
data = d
} else {
data = append(data, d...)
}
}
}
return data, nil
}
2019-05-27 18:59:03 +00:00
func readOneInterval(interval Interval, ecFiles []*os.File) (data []byte, err error) {
2019-05-29 06:48:39 +00:00
ecFileIndex, ecFileOffset := interval.ToShardIdAndOffset(largeBlockSize, smallBlockSize)
data = make([]byte, interval.Size)
err = readFromFile(ecFiles[ecFileIndex], data, ecFileOffset)
{ // do some ec testing
2019-05-27 18:59:03 +00:00
ecData, err := readFromOtherEcFiles(ecFiles, int(ecFileIndex), ecFileOffset, interval.Size)
if err != nil {
return nil, fmt.Errorf("ec reconstruct error: %v", err)
}
if bytes.Compare(data, ecData) != 0 {
return nil, fmt.Errorf("ec compare error")
}
}
return
}
func readFromOtherEcFiles(ecFiles []*os.File, ecFileIndex int, ecFileOffset int64, size types.Size) (data []byte, err error) {
enc, err := reedsolomon.New(DataShardsCount, ParityShardsCount)
if err != nil {
return nil, fmt.Errorf("failed to create encoder: %v", err)
}
2019-05-25 09:02:44 +00:00
bufs := make([][]byte, TotalShardsCount)
for i := 0; i < DataShardsCount; {
2019-05-25 09:02:44 +00:00
n := int(rand.Int31n(TotalShardsCount))
if n == ecFileIndex || bufs[n] != nil {
continue
}
bufs[n] = make([]byte, size)
i++
}
for i, buf := range bufs {
if buf == nil {
continue
}
err = readFromFile(ecFiles[i], buf, ecFileOffset)
if err != nil {
return
}
}
if err = enc.ReconstructData(bufs); err != nil {
return nil, err
}
return bufs[ecFileIndex], nil
}
func readFromFile(file *os.File, data []byte, ecFileOffset int64) (err error) {
_, err = file.ReadAt(data, ecFileOffset)
return
2019-05-19 10:01:58 +00:00
}
2019-05-20 04:16:01 +00:00
func removeGeneratedFiles(baseFileName string) {
for i := 0; i < DataShardsCount+ParityShardsCount; i++ {
2019-05-29 07:34:30 +00:00
fname := fmt.Sprintf("%s.ec%02d", baseFileName, i)
2019-05-20 04:16:01 +00:00
os.Remove(fname)
}
os.Remove(baseFileName + ".ecx")
2019-05-20 04:16:01 +00:00
}
2019-05-19 10:01:58 +00:00
func TestLocateData(t *testing.T) {
intervals := LocateData(largeBlockSize, smallBlockSize, DataShardsCount*largeBlockSize+1, DataShardsCount*largeBlockSize, 1)
2019-05-19 10:01:58 +00:00
if len(intervals) != 1 {
t.Errorf("unexpected interval size %d", len(intervals))
}
2019-05-27 18:59:03 +00:00
if !intervals[0].sameAs(Interval{0, 0, 1, false, 1}) {
2019-05-19 10:01:58 +00:00
t.Errorf("unexpected interval %+v", intervals[0])
}
intervals = LocateData(largeBlockSize, smallBlockSize, DataShardsCount*largeBlockSize+1, DataShardsCount*largeBlockSize/2+100, DataShardsCount*largeBlockSize+1-DataShardsCount*largeBlockSize/2-100)
2019-05-19 10:01:58 +00:00
fmt.Printf("%+v\n", intervals)
}
2019-05-19 10:01:58 +00:00
func (this Interval) sameAs(that Interval) bool {
return this.IsLargeBlock == that.IsLargeBlock &&
this.InnerBlockOffset == that.InnerBlockOffset &&
this.BlockIndex == that.BlockIndex &&
this.Size == that.Size
}