/
worker.go
540 lines (475 loc) · 13.9 KB
/
worker.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
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
// Copyright 2022-2023 Tigris Data, Inc.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package workers
import (
"context"
"fmt"
"math/rand"
"runtime/debug"
"sync"
"time"
jsoniter "github.com/json-iterator/go"
"github.com/rs/zerolog/log"
api "github.com/tigrisdata/tigris/api/server/v1"
"github.com/tigrisdata/tigris/schema"
"github.com/tigrisdata/tigris/server/config"
"github.com/tigrisdata/tigris/server/metadata"
"github.com/tigrisdata/tigris/server/metrics"
"github.com/tigrisdata/tigris/server/services/v1/database"
"github.com/tigrisdata/tigris/server/transaction"
"github.com/tigrisdata/tigris/store/search"
ulog "github.com/tigrisdata/tigris/util/log"
)
const (
MAX_ERROR_COUNT = 10
LEASE_TIME = 3 * time.Minute
PEAK_JOB_ITEMS = 5 // number of items to fetch from the queue to see which one to select
QUEUE_UPDATE_PERIOD = 5 * time.Minute
)
type WorkerTestTask struct {
Sleep time.Duration `json:"sleep"`
NumErrors int `json:"error_count"`
ShouldStopWorker bool `json:"should_stop_worker"`
}
type Worker struct {
id uint
queue *metadata.QueueSubspace
txMgr *transaction.Manager
tenantMgr *metadata.TenantManager
searchStore search.Store
errCount uint
// Indicated to the worker to shutdown
done chan struct{}
// How long the worker sleeps between checking queue
sleepTime time.Duration
itemEvent chan<- Event
hearbeatChan chan<- uint
}
type Event struct {
Success bool
Item metadata.QueueItem
WorkerId uint
}
func newEvent(success bool, item metadata.QueueItem, workerId uint) Event {
return Event{
success,
item,
workerId,
}
}
func newWorker(id uint, queue *metadata.QueueSubspace, txMgr *transaction.Manager, tenantMgr *metadata.TenantManager, searchStore search.Store, sleepTime time.Duration, itemEvent chan<- Event, heartbeatChan chan<- uint) *Worker {
return &Worker{
id: id,
queue: queue,
txMgr: txMgr,
tenantMgr: tenantMgr,
searchStore: searchStore,
errCount: 0,
done: make(chan struct{}, 1),
sleepTime: sleepTime,
itemEvent: itemEvent,
hearbeatChan: heartbeatChan,
}
}
// Add a slightly random jitter for each worker
// This helps avoid a situation where every worker
// starts or peaks at the exact time and all tries to read from the job queue.
func (w *Worker) jitterSleep() {
//nolint:gosec // it is ok to use a weak random generator here
jitterSleep := rand.Intn(500)
time.Sleep(time.Duration(jitterSleep)*time.Millisecond + w.sleepTime)
}
func (w *Worker) Start() {
defer catchWorkerPanics()
w.jitterSleep()
w.Loop()
}
func catchWorkerPanics() {
if r := recover(); r != nil {
stackTrace := string(debug.Stack())
if err, ok := r.(error); ok {
log.Err(err).Str("stacktrace", stackTrace).Msg("worker panic")
} else {
log.Error().Str("stacktrack", stackTrace).Msgf("worker panic %v", err)
}
}
}
func (w *Worker) Loop() {
hbTicker := time.NewTicker(2 * w.sleepTime)
for {
select {
case <-w.done:
log.Info().Msgf("Worker %d shutting down", w.id)
hbTicker.Stop()
return
case <-hbTicker.C:
w.hearbeatChan <- w.id
default:
err := w.peekAndProcess()
if err != nil {
w.errCount++
log.Err(err).Msgf("Worker %d error while processing", w.id)
metrics.IncFailedJobError()
// if the worker is getting to many errors in a row
// shut the worker down and restart a new one
if w.errCount >= MAX_ERROR_COUNT {
log.Err(err).Msgf("Worker %d exceeded error count and shutting down", w.id)
return
}
} else {
w.errCount = 0
}
}
w.jitterSleep()
}
}
func (w *Worker) Stop() {
w.done <- struct{}{}
}
func (w *Worker) peekAndProcess() error {
ctx := context.Background()
tx, err := w.txMgr.StartTx(ctx)
if err != nil {
return err
}
items, err := w.queue.Peek(ctx, tx, PEAK_JOB_ITEMS)
if err != nil {
return err
}
if len(items) == 0 {
return tx.Rollback(ctx)
}
selectedItem, err := w.queue.ObtainLease(ctx, tx, &items[0], LEASE_TIME)
log.Debug().Msgf("Worker %d: processing task %s", w.id, selectedItem.Id)
if err != nil {
return err
}
if err = tx.Commit(ctx); err != nil {
return err
}
if err = w.processItem(selectedItem); err != nil {
log.Err(err).Msgf("Worker %d: failed to process %s", w.id, selectedItem.Id)
return w.handleFailedProcessing(ctx, selectedItem)
}
log.Info().Msgf("Worker %d: Completed %s", w.id, selectedItem.Id)
w.itemEvent <- newEvent(true, *selectedItem, w.id)
return nil
}
func (w *Worker) handleFailedProcessing(ctx context.Context, selectedItem *metadata.QueueItem) error {
selectedItem.ErrorCount++
tx, err := w.txMgr.StartTx(ctx)
if err != nil {
return err
}
// The queue item has had to many errors the job will remove it from the queue
if selectedItem.ErrorCount >= MAX_ERROR_COUNT {
log.Err(err).Msgf("Worker %d: Max fail count, dropping item %s from the queue", w.id, selectedItem.Id)
w.itemEvent <- newEvent(false, *selectedItem, w.id)
if err = w.queue.Dequeue(ctx, tx, selectedItem); err != nil {
return err
}
} else {
if err = w.queue.Requeue(ctx, tx, selectedItem, w.sleepTime*time.Duration(selectedItem.ErrorCount)); err != nil {
return err
}
}
return tx.Commit(ctx)
}
func (w *Worker) processItem(queueItem *metadata.QueueItem) error {
switch queueItem.TaskType {
case metadata.BUILD_INDEX_QUEUE_TASK:
return w.buildIndexTask(queueItem)
case metadata.TEST_QUEUE_TASK:
return w.testQueueTask(queueItem)
case metadata.BUILD_SEARCH_INDEX_TASK:
return w.buildSearchTask(queueItem)
}
return fmt.Errorf("unknown job type")
}
func (w *Worker) testQueueTask(item *metadata.QueueItem) error {
ctx := context.Background()
var testTask WorkerTestTask
if err := jsoniter.Unmarshal(item.Data, &testTask); err != nil {
return err
}
if testTask.ShouldStopWorker {
testTask.ShouldStopWorker = false
item.Data, _ = jsoniter.Marshal(testTask)
w.Stop()
return fmt.Errorf("forced worker stop %d", w.id)
}
if testTask.NumErrors > 0 {
testTask.NumErrors--
item.Data, _ = jsoniter.Marshal(testTask)
return fmt.Errorf("test error generated %d", testTask.NumErrors)
}
time.Sleep(testTask.Sleep)
tx, err := w.txMgr.StartTx(ctx)
if err != nil {
return err
}
if err = w.queue.Complete(ctx, tx, item); err != nil {
return err
}
return tx.Commit(ctx)
}
func (w *Worker) buildIndexTask(queueItem *metadata.QueueItem) error {
var task metadata.IndexBuildTask
if err := jsoniter.Unmarshal(queueItem.Data, &task); err != nil {
return err
}
ctx := context.Background()
dbBranch := metadata.NewDatabaseNameWithBranch(task.ProjName, task.Branch)
tenant, err := w.tenantMgr.GetTenant(ctx, task.NamespaceId)
if err != nil {
return err
}
project, err := tenant.GetProject(task.ProjName)
if err != nil {
return err
}
db, err := project.GetDatabase(dbBranch)
if err != nil {
return err
}
coll := db.GetCollection(task.CollName)
if coll == nil {
return fmt.Errorf("could not find collection \"%s\"", task.CollName)
}
// Indexer will only index indexes that are not `INDEX_ACTIVE` state
indexer := database.NewSecondaryIndexer(coll, true)
// Extend the lease of the queue item so that another worker does not
// try and process it
progressUpdate := func(ctx context.Context, tx transaction.Tx) error {
return w.queue.RenewLease(ctx, tx, queueItem, LEASE_TIME)
}
if err = indexer.BuildCollection(ctx, w.txMgr, progressUpdate); err != nil {
return err
}
for _, index := range coll.SecondaryIndexes.All {
index.State = schema.INDEX_ACTIVE
}
tx, err := w.txMgr.StartTx(ctx)
if err != nil {
return err
}
if err = tenant.UpdateCollectionIndexes(ctx, tx, db, coll.Name, coll.SecondaryIndexes.All); ulog.E(err) {
return err
}
if err = w.queue.Complete(ctx, tx, queueItem); ulog.E(err) {
return err
}
return tx.Commit(ctx)
}
func (w *Worker) buildSearchTask(queueItem *metadata.QueueItem) error {
var task metadata.IndexBuildTask
if err := jsoniter.Unmarshal(queueItem.Data, &task); err != nil {
return err
}
ctx := context.Background()
dbBranch := metadata.NewDatabaseNameWithBranch(task.ProjName, task.Branch)
tenant, err := w.tenantMgr.GetTenant(ctx, task.NamespaceId)
if err != nil {
return err
}
project, err := tenant.GetProject(task.ProjName)
if err != nil {
return err
}
db, err := project.GetDatabase(dbBranch)
if err != nil {
return err
}
coll := db.GetCollection(task.CollName)
if coll == nil {
return fmt.Errorf("could not find collection \"%s\"", task.CollName)
}
req := &api.BuildCollectionSearchIndexRequest{
Branch: task.Branch,
Collection: task.CollName,
Project: task.ProjName,
}
qr := database.NewQueryRunnerFactory(w.txMgr, nil, w.searchStore)
searchIndexer := qr.GetSearchIndexRunner(req, &metrics.WriteQueryMetrics{}, nil)
searchIndexer.ProgressUpdate = func(ctx context.Context) error {
tx, err := w.txMgr.StartTx(ctx)
if err != nil {
return err
}
if err = w.queue.RenewLease(ctx, tx, queueItem, LEASE_TIME); err != nil {
return err
}
return tx.Commit(ctx)
}
_, _, err = searchIndexer.ReadOnly(ctx, tenant)
if err != nil {
return err
}
tx, err := w.txMgr.StartTx(ctx)
if err != nil {
return err
}
if err = tenant.UpdateSearchStatus(ctx, tx, db, coll, schema.SearchIndexActive); err != nil {
return err
}
if err = w.queue.Complete(ctx, tx, queueItem); ulog.E(err) {
return err
}
return tx.Commit(ctx)
}
type WorkerInfo struct {
worker *Worker
lastHearbeat time.Time
}
type WorkerPool struct {
sync.Mutex
maxWorkers uint
queue *metadata.QueueSubspace
txMgr *transaction.Manager
tenantMgr *metadata.TenantManager
searchStore search.Store
workers []*WorkerInfo
nextWorkerId uint
workerSleepTime time.Duration
poolSleepTime time.Duration
heartbeatChan chan uint
stopChan chan struct{}
eventChan chan Event
eventListeners []chan<- Event
}
type Complete func(item *metadata.QueueItem)
func NewWorkerPool(maxWorkers uint, queue *metadata.QueueSubspace, txMgr *transaction.Manager, tenantMgr *metadata.TenantManager, searchStore search.Store, workerSleepTime time.Duration, poolSleepTime time.Duration) *WorkerPool {
return &WorkerPool{
maxWorkers: maxWorkers,
queue: queue,
txMgr: txMgr,
tenantMgr: tenantMgr,
searchStore: searchStore,
workers: make([]*WorkerInfo, 0),
nextWorkerId: 0,
workerSleepTime: workerSleepTime,
poolSleepTime: poolSleepTime,
heartbeatChan: make(chan uint, maxWorkers*3),
stopChan: make(chan struct{}, 1),
eventChan: make(chan Event, maxWorkers*3),
// This is for testing
eventListeners: make([]chan<- Event, 0),
}
}
func (pool *WorkerPool) Start() error {
for i := uint(0); i < pool.maxWorkers; i++ {
pool.nextWorkerId = i
pool.workers = append(pool.workers, pool.newWorker(i))
}
go pool.Loop()
return nil
}
func (pool *WorkerPool) newWorker(id uint) *WorkerInfo {
worker := newWorker(id, pool.queue, pool.txMgr, pool.tenantMgr, pool.searchStore, pool.workerSleepTime, pool.eventChan, pool.heartbeatChan)
go worker.Start()
return &WorkerInfo{
worker: worker,
lastHearbeat: time.Now(),
}
}
// Internal testing for worker events.
func (pool *WorkerPool) subscribe(listener chan<- Event) {
pool.Lock()
defer pool.Unlock()
pool.eventListeners = append(pool.eventListeners, listener)
}
func (pool *WorkerPool) notify(event Event) {
pool.Lock()
defer pool.Unlock()
for _, listener := range pool.eventListeners {
select {
case listener <- event:
continue
case <-time.After(5 * time.Millisecond):
continue
}
}
}
func (pool *WorkerPool) Loop() {
ticker := time.NewTicker(pool.poolSleepTime)
queueSizeCheck := time.NewTicker(QUEUE_UPDATE_PERIOD)
for {
select {
case <-pool.stopChan:
log.Info().Msg("shutting down worker pool")
return
case id := <-pool.heartbeatChan:
pool.rxHeartbeats(id)
case event := <-pool.eventChan:
pool.notify(event)
case <-queueSizeCheck.C:
pool.updateQueueSizeMetric()
case <-ticker.C:
pool.checkHeartbeats()
}
}
}
func (pool *WorkerPool) updateQueueSizeMetric() {
if !config.DefaultConfig.Metrics.Queue.Enabled {
return
}
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
tx, err := pool.txMgr.StartTx(ctx)
if err != nil {
log.Err(err).Msg("failed to start tx for queue metric size")
}
items, err := pool.queue.GetAll(ctx, tx)
if err != nil {
log.Err(err).Msg("failed to get queue items for queue metric size")
}
metrics.SetQueueSize(len(items))
err = tx.Rollback(ctx)
if err != nil {
log.Err(err).Msg("failed to rollback for queue metric size")
}
}
func (pool *WorkerPool) rxHeartbeats(workerId uint) {
pool.Lock()
defer pool.Unlock()
for _, info := range pool.workers {
if info.worker.id == workerId {
info.lastHearbeat = time.Now()
return
}
}
log.Error().Msgf("received heartbeat from missing worker id %d", workerId)
}
func (pool *WorkerPool) checkHeartbeats() {
pool.Lock()
defer pool.Unlock()
now := time.Now()
for i, info := range pool.workers {
if now.Sub(info.lastHearbeat) > 5*pool.workerSleepTime {
info.worker.Stop()
pool.nextWorkerId++
metrics.IncFailedJobError()
log.Error().Msgf("No response from worker %d adding new worker", info.worker.id)
pool.workers[i] = pool.newWorker(pool.nextWorkerId)
}
}
}
func (pool *WorkerPool) Stop() {
pool.Lock()
defer pool.Unlock()
for _, info := range pool.workers {
info.worker.Stop()
}
pool.stopChan <- struct{}{}
}