/
db_write.go
207 lines (187 loc) · 4.1 KB
/
db_write.go
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
// Copyright (c) 2012, Suryandaru Triandana <syndtr@gmail.com>
// All rights reserved.
//
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
package leveldb
import (
"time"
"github.com/borgenk/qdo/third_party/github.com/syndtr/goleveldb/leveldb/memdb"
"github.com/borgenk/qdo/third_party/github.com/syndtr/goleveldb/leveldb/opt"
)
func (d *DB) doWriteJournal(b *Batch) error {
w, err := d.journal.Next()
if err != nil {
return err
}
if _, err := w.Write(b.encode()); err != nil {
return err
}
if err := d.journal.Flush(); err != nil {
return err
}
if b.sync {
return d.journalWriter.Sync()
}
return nil
}
func (d *DB) writeJournal() {
defer d.closeWg.Done()
for {
select {
case _, _ = <-d.closeCh:
return
case b := <-d.journalCh:
if b != nil {
d.journalAckCh <- d.doWriteJournal(b)
}
}
}
}
func (d *DB) flush() (mem *memdb.DB, err error) {
s := d.s
delayed := false
flush := func() bool {
v := s.version()
defer v.release()
mem, _ = d.getMem()
switch {
case v.tLen(0) >= kL0_SlowdownWritesTrigger && !delayed:
delayed = true
time.Sleep(time.Millisecond)
case mem.Size() < s.o.GetWriteBuffer():
return false
case v.tLen(0) >= kL0_StopWritesTrigger:
delayed = true
err = d.wakeCompaction(2)
if err != nil {
return false
}
default:
// Wait for pending memdb compaction.
select {
case _, _ = <-d.closeCh:
err = ErrClosed
return false
case <-d.compMemAckCh:
case err = <-d.compErrCh:
return false
}
// Create new memdb and journal.
mem, err = d.newMem()
if err != nil {
return false
}
// Schedule memdb compaction.
d.compMemCh <- nil
return false
}
return true
}
start := time.Now()
for flush() {
}
if delayed {
s.logf("db@write delayed T·%v", time.Since(start))
}
return
}
// Write apply the given batch to the DB. The batch will be applied
// sequentially.
//
// It is safe to modify the contents of the arguments after Write returns.
func (d *DB) Write(b *Batch, wo *opt.WriteOptions) (err error) {
err = d.ok()
if err != nil || b == nil || b.len() == 0 {
return
}
b.init(wo.GetSync())
// The write happen synchronously.
select {
case _, _ = <-d.closeCh:
return ErrClosed
case d.writeCh <- b:
return <-d.writeAckCh
case d.writeLockCh <- struct{}{}:
}
merged := 0
defer func() {
<-d.writeLockCh
for i := 0; i < merged; i++ {
d.writeAckCh <- err
}
}()
mem, err := d.flush()
if err != nil {
return
}
// Calculate maximum size of the batch.
m := 1 << 20
if x := b.size(); x <= 128<<10 {
m = x + (128 << 10)
}
// Merge with other batch.
drain:
for b.size() <= m && !b.sync {
select {
case nb := <-d.writeCh:
b.append(nb)
merged++
default:
break drain
}
}
// Set batch first seq number relative from last seq.
b.seq = d.seq + 1
// Write journal concurrently if it is large enough.
if b.size() >= (128 << 10) {
// Push the write batch to the journal writer
select {
case _, _ = <-d.closeCh:
err = ErrClosed
return
case d.journalCh <- b:
// Write into memdb
b.memReplay(mem)
}
// Wait for journal writer
select {
case _, _ = <-d.closeCh:
err = ErrClosed
return
case err = <-d.journalAckCh:
if err != nil {
// Revert memdb if error detected
b.revertMemReplay(mem)
return
}
}
} else {
err = d.doWriteJournal(b)
if err != nil {
return
}
b.memReplay(mem)
}
// Set last seq number.
d.addSeq(uint64(b.len()))
return
}
// Put sets the value for the given key. It overwrites any previous value
// for that key; a DB is not a multi-map.
//
// It is safe to modify the contents of the arguments after Put returns.
func (d *DB) Put(key, value []byte, wo *opt.WriteOptions) error {
b := new(Batch)
b.Put(key, value)
return d.Write(b, wo)
}
// Delete deletes the value for the given key. It returns ErrNotFound if
// the DB does not contain the key.
//
// It is safe to modify the contents of the arguments after Delete returns.
func (d *DB) Delete(key []byte, wo *opt.WriteOptions) error {
b := new(Batch)
b.Delete(key)
return d.Write(b, wo)
}