seaweedfs/weed/storage/needle/needle_read_write.go

459 lines
14 KiB
Go
Raw Normal View History

2019-04-19 04:43:36 +00:00
package needle
2012-12-21 05:57:03 +00:00
import (
"bytes"
2012-12-21 06:32:21 +00:00
"errors"
2013-01-17 08:22:18 +00:00
"fmt"
"github.com/chrislusf/seaweedfs/weed/glog"
2022-02-04 09:07:14 +00:00
"github.com/chrislusf/seaweedfs/weed/stats"
"github.com/chrislusf/seaweedfs/weed/storage/backend"
. "github.com/chrislusf/seaweedfs/weed/storage/types"
"github.com/chrislusf/seaweedfs/weed/util"
"io"
"math"
"sync"
2012-12-21 05:57:03 +00:00
)
const (
2020-06-20 05:45:27 +00:00
FlagIsCompressed = 0x01
FlagHasName = 0x02
FlagHasMime = 0x04
FlagHasLastModifiedDate = 0x08
FlagHasTtl = 0x10
FlagHasPairs = 0x20
2015-12-15 06:38:58 +00:00
FlagIsChunkManifest = 0x80
LastModifiedBytesLength = 5
TtlBytesLength = 2
)
var ErrorSizeMismatch = errors.New("size mismatch")
2018-07-24 08:36:04 +00:00
func (n *Needle) DiskSize(version Version) int64 {
return GetActualSize(n.Size, version)
}
2017-01-06 18:22:20 +00:00
var bufPool = sync.Pool{
New: func() interface{} {
return new(bytes.Buffer)
},
}
func (n *Needle) prepareWriteBuffer(version Version, writeBytes *bytes.Buffer) (Size, int64, error) {
writeBytes.Reset()
2013-01-17 08:22:18 +00:00
switch version {
case Version1:
2019-04-19 07:39:34 +00:00
header := make([]byte, NeedleHeaderSize)
CookieToBytes(header[0:CookieSize], n.Cookie)
NeedleIdToBytes(header[CookieSize:CookieSize+NeedleIdSize], n.Id)
n.Size = Size(len(n.Data))
SizeToBytes(header[CookieSize+NeedleIdSize:CookieSize+NeedleIdSize+SizeSize], n.Size)
size := n.Size
actualSize := NeedleHeaderSize + int64(n.Size)
writeBytes.Write(header)
writeBytes.Write(n.Data)
2018-07-24 08:36:04 +00:00
padding := PaddingLength(n.Size, version)
util.Uint32toBytes(header[0:NeedleChecksumSize], n.Checksum.Value())
2021-07-01 08:21:14 +00:00
writeBytes.Write(header[0 : NeedleChecksumSize+padding])
return size, actualSize, nil
2018-07-24 08:36:04 +00:00
case Version2, Version3:
2019-04-19 07:39:34 +00:00
header := make([]byte, NeedleHeaderSize+TimestampSize) // adding timestamp to reuse it and avoid extra allocation
CookieToBytes(header[0:CookieSize], n.Cookie)
NeedleIdToBytes(header[CookieSize:CookieSize+NeedleIdSize], n.Id)
if len(n.Name) >= math.MaxUint8 {
2018-12-11 04:57:51 +00:00
n.NameSize = math.MaxUint8
} else {
n.NameSize = uint8(len(n.Name))
}
n.DataSize, n.MimeSize = uint32(len(n.Data)), uint8(len(n.Mime))
if n.DataSize > 0 {
n.Size = 4 + Size(n.DataSize) + 1
if n.HasName() {
n.Size = n.Size + 1 + Size(n.NameSize)
}
if n.HasMime() {
n.Size = n.Size + 1 + Size(n.MimeSize)
}
if n.HasLastModifiedDate() {
n.Size = n.Size + LastModifiedBytesLength
}
if n.HasTtl() {
n.Size = n.Size + TtlBytesLength
}
if n.HasPairs() {
n.Size += 2 + Size(n.PairsSize)
}
} else {
n.Size = 0
}
SizeToBytes(header[CookieSize+NeedleIdSize:CookieSize+NeedleIdSize+SizeSize], n.Size)
writeBytes.Write(header[0:NeedleHeaderSize])
if n.DataSize > 0 {
util.Uint32toBytes(header[0:4], n.DataSize)
writeBytes.Write(header[0:4])
writeBytes.Write(n.Data)
util.Uint8toBytes(header[0:1], n.Flags)
writeBytes.Write(header[0:1])
if n.HasName() {
util.Uint8toBytes(header[0:1], n.NameSize)
writeBytes.Write(header[0:1])
writeBytes.Write(n.Name[:n.NameSize])
}
if n.HasMime() {
util.Uint8toBytes(header[0:1], n.MimeSize)
writeBytes.Write(header[0:1])
writeBytes.Write(n.Mime)
}
if n.HasLastModifiedDate() {
util.Uint64toBytes(header[0:8], n.LastModified)
2021-07-01 08:21:14 +00:00
writeBytes.Write(header[8-LastModifiedBytesLength : 8])
}
2015-07-09 06:21:53 +00:00
if n.HasTtl() && n.Ttl != nil {
n.Ttl.ToBytes(header[0:TtlBytesLength])
writeBytes.Write(header[0:TtlBytesLength])
}
if n.HasPairs() {
util.Uint16toBytes(header[0:2], n.PairsSize)
writeBytes.Write(header[0:2])
writeBytes.Write(n.Pairs)
}
}
2018-07-24 08:36:04 +00:00
padding := PaddingLength(n.Size, version)
util.Uint32toBytes(header[0:NeedleChecksumSize], n.Checksum.Value())
2018-07-24 08:36:04 +00:00
if version == Version2 {
2021-07-01 08:21:14 +00:00
writeBytes.Write(header[0 : NeedleChecksumSize+padding])
2018-07-24 08:36:04 +00:00
} else {
// version3
util.Uint64toBytes(header[NeedleChecksumSize:NeedleChecksumSize+TimestampSize], n.AppendAtNs)
2021-07-01 08:21:14 +00:00
writeBytes.Write(header[0 : NeedleChecksumSize+TimestampSize+padding])
2018-07-24 08:36:04 +00:00
}
2017-01-06 18:22:20 +00:00
return Size(n.DataSize), GetActualSize(n.Size, version), nil
2012-12-21 08:36:55 +00:00
}
return 0, 0, fmt.Errorf("Unsupported Version! (%d)", version)
}
func (n *Needle) Append(w backend.BackendStorageFile, version Version) (offset uint64, size Size, actualSize int64, err error) {
if end, _, e := w.GetStat(); e == nil {
defer func(w backend.BackendStorageFile, off int64) {
if err != nil {
if te := w.Truncate(end); te != nil {
2019-12-09 03:44:16 +00:00
glog.V(0).Infof("Failed to truncate %s back to %d with error: %v", w.Name(), end, te)
}
}
}(w, end)
offset = uint64(end)
} else {
err = fmt.Errorf("Cannot Read Current Volume Position: %v", e)
return
}
if offset >= MaxPossibleVolumeSize && n.Size.IsValid() {
2020-06-16 07:27:48 +00:00
err = fmt.Errorf("Volume Size %d Exeededs %d", offset, MaxPossibleVolumeSize)
return
}
bytesBuffer := bufPool.Get().(*bytes.Buffer)
defer bufPool.Put(bytesBuffer)
size, actualSize, err = n.prepareWriteBuffer(version, bytesBuffer)
if err == nil {
_, err = w.WriteAt(bytesBuffer.Bytes(), int64(offset))
}
return offset, size, actualSize, err
2012-12-21 05:57:03 +00:00
}
func WriteNeedleBlob(w backend.BackendStorageFile, dataSlice []byte, size Size, appendAtNs uint64, version Version) (offset uint64, err error) {
if end, _, e := w.GetStat(); e == nil {
defer func(w backend.BackendStorageFile, off int64) {
if err != nil {
if te := w.Truncate(end); te != nil {
glog.V(0).Infof("Failed to truncate %s back to %d with error: %v", w.Name(), end, te)
}
}
}(w, end)
offset = uint64(end)
} else {
err = fmt.Errorf("Cannot Read Current Volume Position: %v", e)
return
}
if version == Version3 {
tsOffset := NeedleHeaderSize + size + NeedleChecksumSize
2021-04-02 09:22:26 +00:00
util.Uint64toBytes(dataSlice[tsOffset:tsOffset+TimestampSize], appendAtNs)
}
if err == nil {
_, err = w.WriteAt(dataSlice, int64(offset))
}
return
}
func ReadNeedleBlob(r backend.BackendStorageFile, offset int64, size Size, version Version) (dataSlice []byte, err error) {
dataSize := GetActualSize(size, version)
dataSlice = make([]byte, int(dataSize))
2021-02-20 20:39:25 +00:00
var n int
n, err = r.ReadAt(dataSlice, offset)
if err != nil && int64(n) == dataSize {
err = nil
}
if err != nil {
fileSize, _, _ := r.GetStat()
2021-09-02 19:32:46 +00:00
glog.Errorf("%s read %d dataSize %d offset %d fileSize %d: %v", r.Name(), n, dataSize, offset, fileSize, err)
2021-02-20 20:39:25 +00:00
}
return dataSlice, err
}
2019-05-27 18:59:03 +00:00
// ReadBytes hydrates the needle from the bytes buffer, with only n.Id is set.
func (n *Needle) ReadBytes(bytes []byte, offset int64, size Size, version Version) (err error) {
n.ParseNeedleHeader(bytes)
if n.Size != size {
// cookie is not always passed in for this API. Use size to do preliminary checking.
if OffsetSize == 4 && offset < int64(MaxPossibleVolumeSize) {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorSizeMismatchOffsetSize).Inc()
glog.Errorf("entry not found1: offset %d found id %x size %d, expected size %d", offset, n.Id, n.Size, size)
return ErrorSizeMismatch
}
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorSizeMismatch).Inc()
2020-08-05 05:09:07 +00:00
return fmt.Errorf("entry not found: offset %d found id %x size %d, expected size %d", offset, n.Id, n.Size, size)
}
2013-01-17 08:22:18 +00:00
switch version {
case Version1:
2019-04-19 07:39:34 +00:00
n.Data = bytes[NeedleHeaderSize : NeedleHeaderSize+size]
2018-07-24 08:36:04 +00:00
case Version2, Version3:
2019-04-19 07:39:34 +00:00
err = n.readNeedleDataVersion2(bytes[NeedleHeaderSize : NeedleHeaderSize+int(n.Size)])
2012-12-21 05:57:03 +00:00
}
2019-04-30 03:22:19 +00:00
if err != nil && err != io.EOF {
2019-03-19 12:34:43 +00:00
return err
}
if size > 0 {
checksum := util.BytesToUint32(bytes[NeedleHeaderSize+size : NeedleHeaderSize+size+NeedleChecksumSize])
newChecksum := NewCRC(n.Data)
if checksum != newChecksum.Value() {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorCRC).Inc()
return errors.New("CRC error! Data On Disk Corrupted")
}
n.Checksum = newChecksum
}
2018-07-24 08:36:04 +00:00
if version == Version3 {
2019-04-19 07:39:34 +00:00
tsOffset := NeedleHeaderSize + size + NeedleChecksumSize
2018-07-24 08:38:08 +00:00
n.AppendAtNs = util.BytesToUint64(bytes[tsOffset : tsOffset+TimestampSize])
2018-07-24 08:36:04 +00:00
}
return nil
2012-12-21 05:57:03 +00:00
}
2019-05-27 18:59:03 +00:00
// ReadData hydrates the needle from the file, with only n.Id is set.
func (n *Needle) ReadData(r backend.BackendStorageFile, offset int64, size Size, version Version) (err error) {
2019-05-27 18:59:03 +00:00
bytes, err := ReadNeedleBlob(r, offset, size, version)
if err != nil {
return err
}
return n.ReadBytes(bytes, offset, size, version)
}
func (n *Needle) ParseNeedleHeader(bytes []byte) {
n.Cookie = BytesToCookie(bytes[0:CookieSize])
2018-07-24 08:38:08 +00:00
n.Id = BytesToNeedleId(bytes[CookieSize : CookieSize+NeedleIdSize])
n.Size = BytesToSize(bytes[CookieSize+NeedleIdSize : NeedleHeaderSize])
}
2019-03-19 12:34:43 +00:00
func (n *Needle) readNeedleDataVersion2(bytes []byte) (err error) {
2012-12-21 10:20:38 +00:00
index, lenBytes := 0, len(bytes)
if index < lenBytes {
2018-07-24 08:38:08 +00:00
n.DataSize = util.BytesToUint32(bytes[index : index+4])
2012-12-21 10:20:38 +00:00
index = index + 4
if int(n.DataSize)+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 1)
}
2018-07-24 08:38:08 +00:00
n.Data = bytes[index : index+int(n.DataSize)]
2012-12-21 10:20:38 +00:00
index = index + int(n.DataSize)
n.Flags = bytes[index]
index = index + 1
}
if index < lenBytes && n.HasName() {
n.NameSize = uint8(bytes[index])
index = index + 1
if int(n.NameSize)+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 2)
2019-03-19 12:34:43 +00:00
}
2018-07-24 08:38:08 +00:00
n.Name = bytes[index : index+int(n.NameSize)]
index = index + int(n.NameSize)
}
if index < lenBytes && n.HasMime() {
n.MimeSize = uint8(bytes[index])
index = index + 1
if int(n.MimeSize)+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 3)
2019-03-19 12:34:43 +00:00
}
2018-07-24 08:38:08 +00:00
n.Mime = bytes[index : index+int(n.MimeSize)]
index = index + int(n.MimeSize)
}
if index < lenBytes && n.HasLastModifiedDate() {
if LastModifiedBytesLength+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 4)
2019-03-19 12:34:43 +00:00
}
2018-07-24 08:38:08 +00:00
n.LastModified = util.BytesToUint64(bytes[index : index+LastModifiedBytesLength])
index = index + LastModifiedBytesLength
}
if index < lenBytes && n.HasTtl() {
if TtlBytesLength+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 5)
2019-03-19 12:34:43 +00:00
}
2018-07-24 08:38:08 +00:00
n.Ttl = LoadTTLFromBytes(bytes[index : index+TtlBytesLength])
index = index + TtlBytesLength
}
if index < lenBytes && n.HasPairs() {
if 2+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 6)
2019-03-19 12:34:43 +00:00
}
2018-07-24 08:38:08 +00:00
n.PairsSize = util.BytesToUint16(bytes[index : index+2])
index += 2
if int(n.PairsSize)+index > lenBytes {
stats.VolumeServerRequestCounter.WithLabelValues(stats.ErrorIndexOutOfRange).Inc()
return fmt.Errorf("index out of range %d", 7)
2019-03-19 12:34:43 +00:00
}
end := index + int(n.PairsSize)
n.Pairs = bytes[index:end]
index = end
}
2019-03-19 12:34:43 +00:00
return nil
}
func ReadNeedleHeader(r backend.BackendStorageFile, version Version, offset int64) (n *Needle, bytes []byte, bodyLength int64, err error) {
2012-12-21 06:32:21 +00:00
n = new(Needle)
2018-07-24 08:36:04 +00:00
if version == Version1 || version == Version2 || version == Version3 {
2019-04-19 07:39:34 +00:00
bytes = make([]byte, NeedleHeaderSize)
var count int
count, err = r.ReadAt(bytes, offset)
if count <= 0 || err != nil {
return nil, bytes, 0, err
2012-12-21 06:32:21 +00:00
}
n.ParseNeedleHeader(bytes)
2018-07-24 08:36:04 +00:00
bodyLength = NeedleBodyLength(n.Size, version)
2012-12-21 08:36:55 +00:00
}
2012-12-21 08:36:55 +00:00
return
}
func PaddingLength(needleSize Size, version Version) Size {
2018-07-24 08:36:04 +00:00
if version == Version3 {
// this is same value as version2, but just listed here for clarity
2019-04-19 07:39:34 +00:00
return NeedlePaddingSize - ((NeedleHeaderSize + needleSize + NeedleChecksumSize + TimestampSize) % NeedlePaddingSize)
2018-07-24 08:36:04 +00:00
}
2019-04-19 07:39:34 +00:00
return NeedlePaddingSize - ((NeedleHeaderSize + needleSize + NeedleChecksumSize) % NeedlePaddingSize)
2018-07-24 08:36:04 +00:00
}
func NeedleBodyLength(needleSize Size, version Version) int64 {
2018-07-24 08:36:04 +00:00
if version == Version3 {
return int64(needleSize) + NeedleChecksumSize + TimestampSize + int64(PaddingLength(needleSize, version))
}
return int64(needleSize) + NeedleChecksumSize + int64(PaddingLength(needleSize, version))
}
2012-12-21 08:36:55 +00:00
//n should be a needle already read the header
//the input stream will read until next file entry
func (n *Needle) ReadNeedleBody(r backend.BackendStorageFile, version Version, offset int64, bodyLength int64) (bytes []byte, err error) {
2019-03-18 06:28:43 +00:00
if bodyLength <= 0 {
return nil, nil
}
2019-04-19 02:22:13 +00:00
bytes = make([]byte, bodyLength)
if _, err = r.ReadAt(bytes, offset); err != nil {
return
}
err = n.ReadNeedleBodyBytes(bytes, version)
return
}
func (n *Needle) ReadNeedleBodyBytes(needleBody []byte, version Version) (err error) {
if len(needleBody) <= 0 {
return nil
}
2013-01-17 08:22:18 +00:00
switch version {
case Version1:
2019-04-19 02:22:13 +00:00
n.Data = needleBody[:n.Size]
2012-12-21 08:36:55 +00:00
n.Checksum = NewCRC(n.Data)
2018-07-24 08:36:04 +00:00
case Version2, Version3:
2019-04-19 02:22:13 +00:00
err = n.readNeedleDataVersion2(needleBody[0:n.Size])
n.Checksum = NewCRC(n.Data)
2019-03-18 06:28:43 +00:00
if version == Version3 {
2019-03-19 12:34:43 +00:00
tsOffset := n.Size + NeedleChecksumSize
2019-04-19 02:22:13 +00:00
n.AppendAtNs = util.BytesToUint64(needleBody[tsOffset : tsOffset+TimestampSize])
2019-03-18 06:28:43 +00:00
}
default:
2019-04-19 02:22:13 +00:00
err = fmt.Errorf("unsupported version %d!", version)
2012-12-21 05:57:03 +00:00
}
2012-12-21 06:32:21 +00:00
return
2012-12-21 05:57:03 +00:00
}
2020-06-20 05:45:27 +00:00
func (n *Needle) IsCompressed() bool {
return n.Flags&FlagIsCompressed > 0
}
2020-06-20 05:45:27 +00:00
func (n *Needle) SetIsCompressed() {
n.Flags = n.Flags | FlagIsCompressed
}
func (n *Needle) HasName() bool {
return n.Flags&FlagHasName > 0
}
func (n *Needle) SetHasName() {
n.Flags = n.Flags | FlagHasName
}
func (n *Needle) HasMime() bool {
return n.Flags&FlagHasMime > 0
}
func (n *Needle) SetHasMime() {
n.Flags = n.Flags | FlagHasMime
}
func (n *Needle) HasLastModifiedDate() bool {
return n.Flags&FlagHasLastModifiedDate > 0
}
func (n *Needle) SetHasLastModifiedDate() {
n.Flags = n.Flags | FlagHasLastModifiedDate
}
func (n *Needle) HasTtl() bool {
return n.Flags&FlagHasTtl > 0
}
func (n *Needle) SetHasTtl() {
n.Flags = n.Flags | FlagHasTtl
}
func (n *Needle) IsChunkedManifest() bool {
2015-12-15 06:38:58 +00:00
return n.Flags&FlagIsChunkManifest > 0
}
2015-12-15 06:38:58 +00:00
func (n *Needle) SetIsChunkManifest() {
n.Flags = n.Flags | FlagIsChunkManifest
}
func (n *Needle) HasPairs() bool {
return n.Flags&FlagHasPairs != 0
}
func (n *Needle) SetHasPairs() {
n.Flags = n.Flags | FlagHasPairs
}
2019-04-19 04:43:36 +00:00
func GetActualSize(size Size, version Version) int64 {
2019-04-19 07:39:34 +00:00
return NeedleHeaderSize + NeedleBodyLength(size, version)
2019-04-19 04:43:36 +00:00
}