-
Notifications
You must be signed in to change notification settings - Fork 18
/
streams.go
322 lines (270 loc) · 8.28 KB
/
streams.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
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
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
package kstream
import (
"fmt"
"github.com/tryfix/errors"
"github.com/tryfix/kstream/consumer"
"github.com/tryfix/log"
"github.com/tryfix/metrics"
"sync"
)
type StreamInstance struct {
id string
streams map[string]*kStream // topic:topology
topics []string
processorPool *processorPool
numOfParallelConsumers int
stopping chan bool
stopped chan bool
logger log.Logger
consumer consumer.Builder
builder *StreamBuilder
reBalanceHandler consumer.ReBalanceHandler
}
type instancesOptions struct {
notifyOnSynced chan bool
reBalanceHandler consumer.ReBalanceHandler
}
func NotifyOnStart(c chan bool) InstancesOptions {
return func(config *instancesOptions) {
config.notifyOnSynced = c
}
}
func WithReBalanceHandler(h consumer.ReBalanceHandler) InstancesOptions {
return func(config *instancesOptions) {
config.reBalanceHandler = h
}
}
type InstancesOptions func(config *instancesOptions)
func (iOpts *instancesOptions) apply(options ...InstancesOptions) {
for _, o := range options {
o(iOpts)
}
}
type Instances struct {
streams map[int]*StreamInstance
globalTables map[string]*globalKTable
globalTableStream *globalTableStream
options *instancesOptions
logger log.Logger
builder *StreamBuilder
metricsReporter metrics.Reporter
}
func NewStreams(builder *StreamBuilder, options ...InstancesOptions) *Instances {
opts := new(instancesOptions)
opts.apply(options...)
instances := &Instances{
streams: make(map[int]*StreamInstance),
globalTables: builder.globalTables,
options: opts,
builder: builder,
metricsReporter: builder.metricsReporter,
logger: builder.logger.NewLog(log.Prefixed(`streams`)),
}
if len(builder.streams) < 1 {
return instances
}
for i := 1; i <= builder.config.ConsumerCount; i++ {
id := fmt.Sprintf(`streams-%d`, i)
logger := builder.logger.NewLog(log.Prefixed(id))
instance := &StreamInstance{
id: id,
streams: builder.streams,
processorPool: newProcessorPool(id, builder.streams, builder.defaultBuilders.changelog, logger, builder.metricsReporter),
numOfParallelConsumers: builder.config.ConsumerCount,
stopping: make(chan bool, 1),
stopped: make(chan bool, 1),
logger: logger,
consumer: builder.defaultBuilders.Consumer,
builder: builder,
}
if opts.reBalanceHandler != nil {
instance.reBalanceHandler = opts.reBalanceHandler
}
for t := range builder.streams {
instance.topics = append(instance.topics, t)
}
instances.streams[i] = instance
}
return instances
}
func (ins *Instances) Start() (err error) {
defer ins.logger.Info(`k-stream shutdown completed`)
/*var syncableReplicas []consumer.TopicPartition
// start changelog replica syncing
for _, stream := range ins.builder.streams {
if stream.changelog.enabled && stream.changelog.replicated {
for i := 0; i <= stream.changelog.topic.numOfPartitions-1; i++ {
// already allocated partition doesnt need a replica syncer
syncableReplicas = append(syncableReplicas, consumer.TopicPartition{
Topic: stream.changelog.topic.name, Partition: int32(i),
})
}
}
}
if len(syncableReplicas) > 0 {
if err := ins.builder.changelogReplicaManager.StartReplicas(syncableReplicas); err != nil {
ins.logger.Fatal( err)
}
}*/
wg := new(sync.WaitGroup)
// start global table streams
if len(ins.globalTables) > 0 {
//wg.Add(1)
ins.globalTableStream, err = newGlobalTableStream(ins.globalTables, &GlobalTableStreamConfig{
ConsumerBuilder: ins.builder.defaultBuilders.PartitionConsumer,
Logger: ins.logger,
KafkaAdmin: ins.builder.defaultBuilders.KafkaAdmin,
OffsetManager: ins.builder.defaultBuilders.OffsetManager,
Metrics: ins.metricsReporter,
BackendBuilder: ins.builder.defaultBuilders.Backend,
})
if err != nil {
return errors.WithPrevious(err, `global table stream start failed`)
}
ins.globalTableStream.StartStreams(wg)
if len(ins.streams) < 1 {
if ins.options.notifyOnSynced != nil {
ins.options.notifyOnSynced <- true
}
wg.Wait()
return nil
}
}
for _, instance := range ins.streams {
// start stream consumer
wg.Add(1)
go func(wg *sync.WaitGroup, i *StreamInstance) {
if err := i.Start(wg); err != nil {
ins.logger.Fatal(err)
}
}(wg, instance)
}
if ins.options.notifyOnSynced != nil {
ins.options.notifyOnSynced <- true
}
wg.Wait()
return nil
}
// Stop stops all the running Streams Instances and then GlobalTables
func (ins *Instances) Stop() {
if len(ins.streams) > 0 {
// stop all the streams first
wg := &sync.WaitGroup{}
for _, instance := range ins.streams {
wg.Add(1)
go func(w *sync.WaitGroup, i *StreamInstance) {
i.Stop()
w.Done()
}(wg, instance)
}
wg.Wait()
}
// stop running global tables
if len(ins.globalTables) > 0 {
ins.globalTableStream.stop()
}
}
// starts the high level consumer for all streams
func (s *StreamInstance) Start(wg *sync.WaitGroup) error {
defer wg.Done()
/*config.OnRebalanced = func(allocation consumer.Allocation) {
// consumer booting logic
if len(allocation.Assigned) > 0 {
logger.DefaultLogger.Warn( `partitions added`, allocation.Assigned)
for _, tp := range allocation.Assigned {
// stop started replicas
if s.streams[tp.Topic].changelog.replicated {
if err := s.builder.changelogReplicaManager.StopReplicas([]consumer.TopicPartition{
{Topic: s.streams[tp.Topic].changelog.topic.name, Partition: tp.Partition},
}); err != nil {
logger.DefaultLogger.Error( err)
}
}
if err := s.processorPool.addProcessor(tp); err != nil {
logger.DefaultLogger.Fatal( `allocation failed due to `, err)
}
pro := s.processorPool.Processor(tp)
if err := pro.boot(); err != nil {
logger.DefaultLogger.Fatal(`k-stream.consumer`,
fmt.Sprintf("cannot boot processor due to : %+v", err))
}
}
}
if len(allocation.Removed) > 0 {
logger.DefaultLogger.Warn( `partitions removed`, allocation.Removed)
// start changelog replica syncers
var syncableReplicas []consumer.TopicPartition
for _, tp := range allocation.Removed {
if s.streams[tp.Topic].changelog.replicated {
syncableReplicas = append(syncableReplicas, consumer.TopicPartition{
Topic: s.streams[tp.Topic].changelog.topic.name, Partition: tp.Partition,
})
}
}
if len(syncableReplicas) > 0 {
if err := s.builder.changelogReplicaManager.StartReplicas(syncableReplicas); err != nil {
logger.DefaultLogger.Error( fmt.Sprintf(`cannot start replicas due to %s`, err))
}
}
}
go func() {
if s.allocationNotify != nil {
s.allocationNotify <- allocation
}
}()
}*/
c, err := s.consumer.Build(consumer.BuilderWithId(fmt.Sprintf(`group_consumer_%s`, s.id)))
if err != nil {
return errors.WithPrevious(err, `cannot initiate consumer`)
}
reBalancer := &reBalanceHandler{
userHandler: s.reBalanceHandler,
logger: s.logger,
processors: s.processorPool,
builder: s.builder,
}
partitionConsumers, err := c.Consume(s.topics, reBalancer)
if err != nil {
return errors.WithPrevious(err, `cannot start consumer`)
}
consumerWg := new(sync.WaitGroup)
consumerWg.Add(1)
go func(consumerWg *sync.WaitGroup) {
<-s.stopping
if err := c.Close(); err != nil {
s.logger.Fatal(err)
}
consumerWg.Done()
s.stopped <- true
}(consumerWg)
go func() {
for err := range c.Errors() {
s.logger.Info(err)
}
}()
wgx := new(sync.WaitGroup)
for partition := range partitionConsumers {
wgx.Add(1)
go func(wg *sync.WaitGroup, p consumer.Partition) {
pro := s.processorPool.Processor(p.Partition())
go func(processor *processor) {
for record := range processor.changelogMarks {
if err := p.CommitOffset(record); err != nil {
s.logger.Error(fmt.Sprintf("cannot commit partition offset due to : %+v", err))
}
}
}(pro)
pro.records = p.Records()
pro.start()
wg.Done()
}(wgx, partition)
}
wgx.Wait()
consumerWg.Wait()
return nil
}
func (s *StreamInstance) Stop() {
s.stopping <- true
<-s.stopped
s.logger.Info(`instance stopped`)
}