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
|
package filesys
import (
"bytes"
"context"
"fmt"
"io"
"sync"
"time"
"github.com/chrislusf/seaweedfs/weed/glog"
"github.com/chrislusf/seaweedfs/weed/operation"
"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
"github.com/chrislusf/seaweedfs/weed/security"
)
type ContinuousDirtyPages struct {
intervals *ContinuousIntervals
f *File
lock sync.Mutex
}
func newDirtyPages(file *File) *ContinuousDirtyPages {
return &ContinuousDirtyPages{
intervals: &ContinuousIntervals{},
f: file,
}
}
func (pages *ContinuousDirtyPages) releaseResource() {
}
var counter = int32(0)
func (pages *ContinuousDirtyPages) AddPage(ctx context.Context, offset int64, data []byte) (chunks []*filer_pb.FileChunk, err error) {
pages.lock.Lock()
defer pages.lock.Unlock()
glog.V(3).Infof("%s AddPage [%d,%d)", pages.f.fullpath(), offset, offset+int64(len(data)))
if len(data) > int(pages.f.wfs.option.ChunkSizeLimit) {
// this is more than what buffer can hold.
return pages.flushAndSave(ctx, offset, data)
}
pages.intervals.AddInterval(data, offset)
var chunk *filer_pb.FileChunk
var hasSavedData bool
if pages.intervals.TotalSize() > pages.f.wfs.option.ChunkSizeLimit {
chunk, hasSavedData, err = pages.saveExistingLargestPageToStorage(ctx)
if hasSavedData {
chunks = append(chunks, chunk)
}
}
return
}
func (pages *ContinuousDirtyPages) flushAndSave(ctx context.Context, offset int64, data []byte) (chunks []*filer_pb.FileChunk, err error) {
var chunk *filer_pb.FileChunk
var newChunks []*filer_pb.FileChunk
// flush existing
if newChunks, err = pages.saveExistingPagesToStorage(ctx); err == nil {
if newChunks != nil {
chunks = append(chunks, newChunks...)
}
} else {
return
}
// flush the new page
if chunk, err = pages.saveToStorage(ctx, bytes.NewReader(data), offset, int64(len(data))); err == nil {
if chunk != nil {
glog.V(4).Infof("%s/%s flush big request [%d,%d) to %s", pages.f.dir.Path, pages.f.Name, chunk.Offset, chunk.Offset+int64(chunk.Size), chunk.FileId)
chunks = append(chunks, chunk)
}
} else {
glog.V(0).Infof("%s/%s failed to flush2 [%d,%d): %v", pages.f.dir.Path, pages.f.Name, chunk.Offset, chunk.Offset+int64(chunk.Size), err)
return
}
return
}
func (pages *ContinuousDirtyPages) FlushToStorage(ctx context.Context) (chunks []*filer_pb.FileChunk, err error) {
pages.lock.Lock()
defer pages.lock.Unlock()
return pages.saveExistingPagesToStorage(ctx)
}
func (pages *ContinuousDirtyPages) saveExistingPagesToStorage(ctx context.Context) (chunks []*filer_pb.FileChunk, err error) {
var hasSavedData bool
var chunk *filer_pb.FileChunk
for {
chunk, hasSavedData, err = pages.saveExistingLargestPageToStorage(ctx)
if !hasSavedData {
return chunks, err
}
if err == nil {
chunks = append(chunks, chunk)
} else {
return
}
}
}
func (pages *ContinuousDirtyPages) saveExistingLargestPageToStorage(ctx context.Context) (chunk *filer_pb.FileChunk, hasSavedData bool, err error) {
maxList := pages.intervals.RemoveLargestIntervalLinkedList()
if maxList == nil {
return nil, false, nil
}
chunk, err = pages.saveToStorage(ctx, maxList.ToReader(), maxList.Offset(), maxList.Size())
if err == nil {
hasSavedData = true
glog.V(3).Infof("%s saveToStorage [%d,%d) %s", pages.f.fullpath(), maxList.Offset(), maxList.Offset()+maxList.Size(), chunk.FileId)
} else {
glog.V(0).Infof("%s saveToStorage [%d,%d): %v", pages.f.fullpath(), maxList.Offset(), maxList.Offset()+maxList.Size(), err)
return
}
return
}
func (pages *ContinuousDirtyPages) saveToStorage(ctx context.Context, reader io.Reader, offset int64, size int64) (*filer_pb.FileChunk, error) {
var fileId, host string
var auth security.EncodedJwt
if err := pages.f.wfs.WithFilerClient(ctx, func(client filer_pb.SeaweedFilerClient) error {
request := &filer_pb.AssignVolumeRequest{
Count: 1,
Replication: pages.f.wfs.option.Replication,
Collection: pages.f.wfs.option.Collection,
TtlSec: pages.f.wfs.option.TtlSec,
DataCenter: pages.f.wfs.option.DataCenter,
}
resp, err := client.AssignVolume(ctx, request)
if err != nil {
glog.V(0).Infof("assign volume failure %v: %v", request, err)
return err
}
fileId, host, auth = resp.FileId, resp.Url, security.EncodedJwt(resp.Auth)
return nil
}); err != nil {
return nil, fmt.Errorf("filerGrpcAddress assign volume: %v", err)
}
fileUrl := fmt.Sprintf("http://%s/%s", host, fileId)
uploadResult, err := operation.Upload(fileUrl, pages.f.Name, reader, false, "", nil, auth)
if err != nil {
glog.V(0).Infof("upload data %v to %s: %v", pages.f.Name, fileUrl, err)
return nil, fmt.Errorf("upload data: %v", err)
}
if uploadResult.Error != "" {
glog.V(0).Infof("upload failure %v to %s: %v", pages.f.Name, fileUrl, err)
return nil, fmt.Errorf("upload result: %v", uploadResult.Error)
}
return &filer_pb.FileChunk{
FileId: fileId,
Offset: offset,
Size: uint64(size),
Mtime: time.Now().UnixNano(),
ETag: uploadResult.ETag,
}, nil
}
func max(x, y int64) int64 {
if x > y {
return x
}
return y
}
func min(x, y int64) int64 {
if x < y {
return x
}
return y
}
func (pages *ContinuousDirtyPages) ReadDirtyData(ctx context.Context, data []byte, startOffset int64) (offset int64, size int) {
pages.lock.Lock()
defer pages.lock.Unlock()
return pages.intervals.ReadData(data, startOffset)
}
|