-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathbuffer.go
More file actions
229 lines (213 loc) · 4.51 KB
/
buffer.go
File metadata and controls
229 lines (213 loc) · 4.51 KB
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
package pgbuffer
import (
"database/sql"
"errors"
"github.com/sirupsen/logrus"
"io/ioutil"
"sync"
"time"
"github.com/lib/pq"
)
func NewBuffer(db *sql.DB, cfg *Config) (*Buffer, error) {
if cfg.Logger == nil {
cfg.Logger = logrus.New()
cfg.Logger.SetOutput(ioutil.Discard)
}
b := &Buffer{
db: db,
writeChan: make(chan *writePayload, 1000),
logger: cfg.Logger,
cfg: cfg,
stopSignal: make(chan struct{}, 2),
flushSignal: make(chan struct{}, 2),
}
if err := b.validate(); err != nil {
return nil, err
}
//Setup some defaults
if cfg.Limit == 0 {
cfg.Limit = 1000
}
if cfg.Workers == 0 {
cfg.Workers = 1
}
b.data = make(map[string]*BufferedData)
for _, t := range cfg.Tables {
if t.Limit == 0 { // if its not set use the global limit
t.Limit = cfg.Limit
}
t.data = make([][]interface{}, t.Limit)
t.position = 0
t.LastWrite = time.Now()
b.data[t.Table] = t
}
return b, nil
}
func (b *Buffer) validate() error {
if len(b.cfg.Tables) == 0 {
return errors.New("no table config provided")
}
for _, t := range b.cfg.Tables {
if err := b.validateTable(t); err != nil {
return err
}
}
return nil
}
func (b *Buffer) validateTable(t *BufferedData) error {
tx, err := b.db.Begin()
if err != nil {
return err
}
defer tx.Rollback()
cpin := pq.CopyIn(t.Table, t.Columns...)
stmt, err := tx.Prepare(cpin)
if err != nil {
return err
}
stmt.Close()
return nil
}
func (b *Buffer) Stop() {
b.stopSignal <- struct{}{}
}
func (b *Buffer) FlushAll() {
b.flushSignal <- struct{}{}
}
func (b *Buffer) Run() {
//Control loop
//TODO: refactor
CONTROL:
for {
select {
case data := <-b.writeChan:
if buff, ok := b.data[data.table]; ok {
if buff.position == buff.Limit {
data := buff.data
buff.data = make([][]interface{}, buff.Limit)
buff.position = 0
if len(data) > 0 {
go b.Flush(buff, data)
}
}
buff.data[buff.position] = data.rows
buff.position++
}
case <-b.stopSignal:
for _, buff := range b.data {
buff.LastWrite = time.Now()
b.Flush(buff, buff.data)
}
break CONTROL
case <-b.flushSignal:
for _, buff := range b.data {
buff.LastWrite = time.Now()
data := buff.data
buff.data = make([][]interface{}, buff.Limit)
buff.position = 0
if len(data) > 0 {
go b.Flush(buff, data)
}
}
}
}
}
func (b *Buffer) Flush(buff *BufferedData, data [][]interface{}) {
buff.Lock()
defer buff.Unlock()
st := time.Now()
var wg sync.WaitGroup
if len(data) <= b.cfg.Workers {
wg.Add(1)
go b.flushBatch(&wg, buff.Table, buff.Columns, data)
} else {
var chunks = make([][][]interface{}, 0)
chunkSize := len(data) / b.cfg.Workers
for i := 0; i < b.cfg.Workers; i++ {
temp := make([][]interface{}, 0)
idx := i * chunkSize
end := i*chunkSize + chunkSize
for x := range data[idx:end] {
temp = append(temp, data[x])
}
if i == b.cfg.Workers {
for x := range data[idx:] {
temp = append(temp, data[x])
}
}
chunks = append(chunks, temp)
}
if b.logger != nil {
b.logger.Infof("workers %v, chunks %v\n", b.cfg.Workers, len(chunks))
}
for i := 0; i < b.cfg.Workers; i++ {
wg.Add(1)
if b.logger != nil {
b.logger.Infof("flushing %v %v records\n", i, len(chunks[i]))
}
go b.flushBatch(&wg, buff.Table, buff.Columns, chunks[i])
}
}
wg.Wait()
if b.logger != nil {
b.logger.Infof("flushed %d records in %v", len(data), time.Since(st))
}
}
func (b *Buffer) flushBatch(wg *sync.WaitGroup, table string, columns []string, data [][]interface{}) {
defer wg.Done()
tx, err := b.db.Begin()
if err != nil {
if b.logger != nil {
b.logger.Error(err)
}
return
}
defer tx.Rollback()
cpin := pq.CopyIn(table, columns...)
stmt, err := tx.Prepare(cpin)
if err != nil {
if b.logger != nil {
b.logger.Errorf("prepare failed %v", err)
}
return
}
for _, row := range data {
if len(row) > 0 {
_, err := stmt.Exec(row...)
if err != nil {
if b.logger != nil {
b.logger.Errorf("exec failed: %v", err)
}
return
}
}
}
_, err = stmt.Exec()
if err != nil {
if b.logger != nil {
b.logger.Errorf("end exec failed %v", err)
}
return
}
//err = stmt.Close()
//if err != nil {
// if b.logger != nil {
// b.logger.Error(err)
// }
// return
//}
err = tx.Commit()
if err != nil {
if b.logger != nil {
b.logger.Errorf("commit failure %v", err)
}
return
}
}
func (b *Buffer) Write(table string, rows ...interface{}) {
p := &writePayload{
table: table,
rows: rows,
}
b.writeChan <- p
}