1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
|
package weed_server
import (
"bytes"
"crypto/md5"
"fmt"
"hash"
"io"
"net/http"
"strconv"
"sync"
"time"
"slices"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/operation"
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/security"
"github.com/seaweedfs/seaweedfs/weed/stats"
"github.com/seaweedfs/seaweedfs/weed/util"
)
var bufPool = sync.Pool{
New: func() interface{} {
return new(bytes.Buffer)
},
}
func (fs *FilerServer) uploadRequestToChunks(w http.ResponseWriter, r *http.Request, reader io.Reader, chunkSize int32, fileName, contentType string, contentLength int64, so *operation.StorageOption) (fileChunks []*filer_pb.FileChunk, md5Hash hash.Hash, chunkOffset int64, uploadErr error, smallContent []byte) {
query := r.URL.Query()
isAppend := isAppend(r)
if query.Has("offset") {
offset := query.Get("offset")
offsetInt, err := strconv.ParseInt(offset, 10, 64)
if err != nil || offsetInt < 0 {
err = fmt.Errorf("invalid 'offset': '%s'", offset)
return nil, nil, 0, err, nil
}
if isAppend && offsetInt > 0 {
err = fmt.Errorf("cannot set offset when op=append")
return nil, nil, 0, err, nil
}
chunkOffset = offsetInt
}
return fs.uploadReaderToChunks(reader, chunkOffset, chunkSize, fileName, contentType, isAppend, so)
}
func (fs *FilerServer) uploadReaderToChunks(reader io.Reader, startOffset int64, chunkSize int32, fileName, contentType string, isAppend bool, so *operation.StorageOption) (fileChunks []*filer_pb.FileChunk, md5Hash hash.Hash, chunkOffset int64, uploadErr error, smallContent []byte) {
md5Hash = md5.New()
chunkOffset = startOffset
var partReader = io.NopCloser(io.TeeReader(reader, md5Hash))
var wg sync.WaitGroup
var bytesBufferCounter int64 = 4
bytesBufferLimitChan := make(chan struct{}, bytesBufferCounter)
var fileChunksLock sync.Mutex
var uploadErrLock sync.Mutex
for {
// need to throttle used byte buffer
bytesBufferLimitChan <- struct{}{}
// As long as there is an error in the upload of one chunk, it can be terminated early
// uploadErr may be modified in other go routines, lock is needed to avoid race condition
uploadErrLock.Lock()
if uploadErr != nil {
<-bytesBufferLimitChan
uploadErrLock.Unlock()
break
}
uploadErrLock.Unlock()
bytesBuffer := bufPool.Get().(*bytes.Buffer)
limitedReader := io.LimitReader(partReader, int64(chunkSize))
bytesBuffer.Reset()
dataSize, err := bytesBuffer.ReadFrom(limitedReader)
// data, err := io.ReadAll(limitedReader)
if err != nil || dataSize == 0 {
bufPool.Put(bytesBuffer)
<-bytesBufferLimitChan
if err != nil {
uploadErrLock.Lock()
if uploadErr == nil {
uploadErr = err
}
uploadErrLock.Unlock()
}
break
}
if chunkOffset == 0 && !isAppend {
if dataSize < fs.option.SaveToFilerLimit {
chunkOffset += dataSize
smallContent = make([]byte, dataSize)
bytesBuffer.Read(smallContent)
bufPool.Put(bytesBuffer)
<-bytesBufferLimitChan
stats.FilerHandlerCounter.WithLabelValues(stats.ContentSaveToFiler).Inc()
break
}
} else {
stats.FilerHandlerCounter.WithLabelValues(stats.AutoChunk).Inc()
}
wg.Add(1)
go func(offset int64, buf *bytes.Buffer) {
defer func() {
bufPool.Put(buf)
<-bytesBufferLimitChan
wg.Done()
}()
chunks, toChunkErr := fs.dataToChunk(fileName, contentType, buf.Bytes(), offset, so)
if toChunkErr != nil {
uploadErrLock.Lock()
if uploadErr == nil {
uploadErr = toChunkErr
}
uploadErrLock.Unlock()
}
if chunks != nil {
fileChunksLock.Lock()
fileChunksSize := len(fileChunks) + len(chunks)
for _, chunk := range chunks {
fileChunks = append(fileChunks, chunk)
glog.V(4).Infof("uploaded %s chunk %d to %s [%d,%d)", fileName, fileChunksSize, chunk.FileId, offset, offset+int64(chunk.Size))
}
fileChunksLock.Unlock()
}
}(chunkOffset, bytesBuffer)
// reset variables for the next chunk
chunkOffset = chunkOffset + dataSize
// if last chunk was not at full chunk size, but already exhausted the reader
if dataSize < int64(chunkSize) {
break
}
}
wg.Wait()
if uploadErr != nil {
glog.V(0).Infof("upload file %s error: %v", fileName, uploadErr)
for _, chunk := range fileChunks {
glog.V(4).Infof("purging failed uploaded %s chunk %s [%d,%d)", fileName, chunk.FileId, chunk.Offset, chunk.Offset+int64(chunk.Size))
}
fs.filer.DeleteUncommittedChunks(fileChunks)
return nil, md5Hash, 0, uploadErr, nil
}
slices.SortFunc(fileChunks, func(a, b *filer_pb.FileChunk) int {
return int(a.Offset - b.Offset)
})
return fileChunks, md5Hash, chunkOffset, nil, smallContent
}
func (fs *FilerServer) doUpload(urlLocation string, limitedReader io.Reader, fileName string, contentType string, pairMap map[string]string, auth security.EncodedJwt) (*operation.UploadResult, error, []byte) {
stats.FilerHandlerCounter.WithLabelValues(stats.ChunkUpload).Inc()
start := time.Now()
defer func() {
stats.FilerRequestHistogram.WithLabelValues(stats.ChunkUpload).Observe(time.Since(start).Seconds())
}()
uploadOption := &operation.UploadOption{
UploadUrl: urlLocation,
Filename: fileName,
Cipher: fs.option.Cipher,
IsInputCompressed: false,
MimeType: contentType,
PairMap: pairMap,
Jwt: auth,
}
uploader, err := operation.NewUploader()
if err != nil {
return nil, err, []byte{}
}
uploadResult, err, data := uploader.Upload(limitedReader, uploadOption)
if uploadResult != nil && uploadResult.RetryCount > 0 {
stats.FilerHandlerCounter.WithLabelValues(stats.ChunkUploadRetry).Add(float64(uploadResult.RetryCount))
}
return uploadResult, err, data
}
func (fs *FilerServer) dataToChunk(fileName, contentType string, data []byte, chunkOffset int64, so *operation.StorageOption) ([]*filer_pb.FileChunk, error) {
dataReader := util.NewBytesReader(data)
// retry to assign a different file id
var fileId, urlLocation string
var auth security.EncodedJwt
var uploadErr error
var uploadResult *operation.UploadResult
var failedFileChunks []*filer_pb.FileChunk
err := util.Retry("filerDataToChunk", func() error {
// assign one file id for one chunk
fileId, urlLocation, auth, uploadErr = fs.assignNewFileInfo(so)
if uploadErr != nil {
glog.V(4).Infof("retry later due to assign error: %v", uploadErr)
stats.FilerHandlerCounter.WithLabelValues(stats.ChunkAssignRetry).Inc()
return uploadErr
}
// upload the chunk to the volume server
uploadResult, uploadErr, _ = fs.doUpload(urlLocation, dataReader, fileName, contentType, nil, auth)
if uploadErr != nil {
glog.V(4).Infof("retry later due to upload error: %v", uploadErr)
stats.FilerHandlerCounter.WithLabelValues(stats.ChunkDoUploadRetry).Inc()
fid, _ := filer_pb.ToFileIdObject(fileId)
fileChunk := filer_pb.FileChunk{
FileId: fileId,
Offset: chunkOffset,
Fid: fid,
}
failedFileChunks = append(failedFileChunks, &fileChunk)
return uploadErr
}
return nil
})
if err != nil {
glog.Errorf("upload error: %v", err)
return failedFileChunks, err
}
// if last chunk exhausted the reader exactly at the border
if uploadResult.Size == 0 {
return nil, nil
}
return []*filer_pb.FileChunk{uploadResult.ToPbFileChunk(fileId, chunkOffset, time.Now().UnixNano())}, nil
}
|
