-
Notifications
You must be signed in to change notification settings - Fork 262
/
peertask_conductor.go
1558 lines (1402 loc) · 46.2 KB
/
peertask_conductor.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
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* Copyright 2022 The Dragonfly Authors
*
* 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 peer
import (
"bytes"
"context"
"errors"
"fmt"
"io"
"runtime/debug"
"sync"
"time"
semconv "go.opentelemetry.io/otel/semconv/v1.7.0"
"go.opentelemetry.io/otel/trace"
"go.uber.org/atomic"
"golang.org/x/time/rate"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/credentials"
"google.golang.org/grpc/status"
commonv1 "d7y.io/api/pkg/apis/common/v1"
errordetailsv1 "d7y.io/api/pkg/apis/errordetails/v1"
schedulerv1 "d7y.io/api/pkg/apis/scheduler/v1"
"d7y.io/dragonfly/v2/client/config"
"d7y.io/dragonfly/v2/client/daemon/metrics"
"d7y.io/dragonfly/v2/client/daemon/storage"
"d7y.io/dragonfly/v2/internal/dferrors"
logger "d7y.io/dragonfly/v2/internal/dflog"
"d7y.io/dragonfly/v2/pkg/digest"
"d7y.io/dragonfly/v2/pkg/idgen"
nethttp "d7y.io/dragonfly/v2/pkg/net/http"
"d7y.io/dragonfly/v2/pkg/rpc/common"
schedulerclient "d7y.io/dragonfly/v2/pkg/rpc/scheduler/client"
"d7y.io/dragonfly/v2/pkg/source"
)
const (
// TODO implement peer task health check
// reasonContextCanceled = "context canceled"
// reasonReScheduleTimeout = "wait more available peers from scheduler timeout"
reasonScheduleTimeout = "wait first peer packet from scheduler timeout"
reasonPeerGoneFromScheduler = "scheduler says client should disconnect"
reasonBackSourceDisabled = "download from source disabled"
failedReasonNotSet = "unknown"
)
var _ Task = (*peerTaskConductor)(nil)
// peerTaskConductor will fetch all pieces from other peers and send pieces info to broker
type peerTaskConductor struct {
TaskOption
*logger.SugaredLoggerOnWith
// ctx is with span info for tracing
// we use successCh and failCh mark task success or fail
ctx context.Context
ctxCancel context.CancelFunc
// piece download uses this context
pieceDownloadCtx context.Context
// when back source, cancel all piece download action
pieceDownloadCancel context.CancelFunc
// request is the original PeerTaskRequest
request *schedulerv1.PeerTaskRequest
// needBackSource indicates downloading resource from instead of other peers
needBackSource *atomic.Bool
seed bool
peerTaskManager *peerTaskManager
storage storage.TaskStorageDriver
schedulerClient schedulerclient.V1
// peer task meta info
peerID string
taskID string
totalPiece *atomic.Int32
digest *atomic.String
contentLength *atomic.Int64
completedLength *atomic.Int64
usedTraffic *atomic.Uint64
header atomic.Value
broker *pieceBroker
sizeScope commonv1.SizeScope
singlePiece *schedulerv1.SinglePiece
tinyData *TinyData
// peerPacketStream stands schedulerclient.PeerPacketStream from scheduler
peerPacketStream schedulerv1.Scheduler_ReportPieceResultClient
legacyPeerCount *atomic.Int64
// pieceTaskSyncManager syncs piece task from other peers
pieceTaskSyncManager *pieceTaskSyncManager
// same actions must be done only once, like close done channel and so on
statusOnce sync.Once
// done channel will be closed when peer task success
successCh chan struct{}
// fail channel will be closed after peer task fail
failCh chan struct{}
// span stands open telemetry trace span
span trace.Span
// failedReason will be set when peer task failed
failedReason string
// failedReason will be set when peer task failed
failedCode commonv1.Code
// readyPieces stands all downloaded pieces
readyPieces *Bitmap
// lock used by piece result manage, when update readyPieces, lock first
readyPiecesLock sync.RWMutex
// runningPieces stands all downloading pieces
runningPieces *Bitmap
// lock used by piece download worker
runningPiecesLock sync.Mutex
// requestedPieces stands all pieces requested from peers
requestedPieces *Bitmap
// lock used by piece download worker
requestedPiecesLock sync.RWMutex
// lock used by send piece result
sendPieceResultLock sync.Mutex
// trafficShaper used to automatically allocate bandwidth for every peer task
trafficShaper TrafficShaper
// limiter will be used when enable per peer task rate limit
limiter *rate.Limiter
startTime time.Time
// subtask only
parent *peerTaskConductor
rg *nethttp.Range
sourceErrorStatus *status.Status
}
type TaskOption struct {
// PeerHost info about current PeerHost
PeerHost *schedulerv1.PeerHost
// PieceManager will be used for downloading piece
PieceManager PieceManager
StorageManager storage.Manager
// schedule options
SchedulerOption config.SchedulerOption
CalculateDigest bool
GRPCCredentials credentials.TransportCredentials
GRPCDialTimeout time.Duration
// WatchdogTimeout > 0 indicates to start watch dog for every single peer task
WatchdogTimeout time.Duration
}
func (ptm *peerTaskManager) newPeerTaskConductor(
ctx context.Context,
request *schedulerv1.PeerTaskRequest,
limit rate.Limit,
parent *peerTaskConductor,
rg *nethttp.Range,
seed bool) *peerTaskConductor {
// use a new context with span info
ctx = trace.ContextWithSpan(context.Background(), trace.SpanFromContext(ctx))
ctx, span := tracer.Start(ctx, config.SpanPeerTask, trace.WithSpanKind(trace.SpanKindClient))
span.SetAttributes(config.AttributePeerHost.String(ptm.PeerHost.Id))
span.SetAttributes(semconv.NetHostIPKey.String(ptm.PeerHost.Ip))
span.SetAttributes(config.AttributePeerID.String(request.PeerId))
span.SetAttributes(semconv.HTTPURLKey.String(request.Url))
taskID := idgen.TaskIDV1(request.Url, request.UrlMeta)
request.TaskId = taskID
// init log with values
var (
log *logger.SugaredLoggerOnWith
traceID = span.SpanContext().TraceID()
)
logKV := []any{
"peer", request.PeerId,
"task", taskID,
"component", "PeerTask",
}
if traceID.IsValid() {
logKV = append(logKV, "trace", traceID.String())
}
log = logger.With(logKV...)
span.SetAttributes(config.AttributeTaskID.String(taskID))
ctx, cancel := context.WithCancel(ctx)
ptc := &peerTaskConductor{
TaskOption: ptm.TaskOption,
peerTaskManager: ptm,
request: request,
startTime: time.Now(),
ctx: ctx,
ctxCancel: cancel,
broker: newPieceBroker(),
peerID: request.PeerId,
taskID: taskID,
successCh: make(chan struct{}),
failCh: make(chan struct{}),
legacyPeerCount: atomic.NewInt64(0),
span: span,
readyPieces: NewBitmap(),
runningPieces: NewBitmap(),
requestedPieces: NewBitmap(),
failedReason: failedReasonNotSet,
failedCode: commonv1.Code_UnknownError,
contentLength: atomic.NewInt64(-1),
totalPiece: atomic.NewInt32(-1),
digest: atomic.NewString(""),
trafficShaper: ptm.trafficShaper,
limiter: rate.NewLimiter(limit, int(limit)),
completedLength: atomic.NewInt64(0),
usedTraffic: atomic.NewUint64(0),
SugaredLoggerOnWith: log,
seed: seed,
parent: parent,
rg: rg,
}
ptc.pieceDownloadCtx, ptc.pieceDownloadCancel = context.WithCancel(ptc.ctx)
return ptc
}
// register to scheduler, if error and disable auto back source, return error, otherwise return nil
func (pt *peerTaskConductor) register() error {
pt.Debugf("request overview, pid: %s, url: %s, filter: %s, tag: %s, range: %s, digest: %s, header: %#v",
pt.request.PeerId, pt.request.Url, pt.request.UrlMeta.Filter, pt.request.UrlMeta.Tag, pt.request.UrlMeta.Range, pt.request.UrlMeta.Digest, pt.request.UrlMeta.Header)
// trace register
regCtx, cancel := context.WithTimeout(pt.ctx, pt.SchedulerOption.ScheduleTimeout.Duration)
defer cancel()
regCtx, regSpan := tracer.Start(regCtx, config.SpanRegisterTask)
var (
needBackSource bool
sizeScope commonv1.SizeScope
singlePiece *schedulerv1.SinglePiece
tinyData *TinyData
)
pt.Infof("step 1: peer %s start to register", pt.request.PeerId)
pt.schedulerClient = pt.peerTaskManager.SchedulerClient
result, err := pt.schedulerClient.RegisterPeerTask(regCtx, pt.request)
regSpan.RecordError(err)
regSpan.End()
if err != nil {
if err == context.DeadlineExceeded {
pt.Errorf("scheduler did not response in %s", pt.SchedulerOption.ScheduleTimeout.Duration)
}
pt.Errorf("step 1: peer %s register failed: %s", pt.request.PeerId, err)
if pt.SchedulerOption.DisableAutoBackSource {
// when peer register failed, some actions need to do with peerPacketStream
pt.peerPacketStream = &dummyPeerPacketStream{}
pt.Errorf("register peer task failed: %s, peer id: %s, auto back source disabled", err, pt.request.PeerId)
pt.span.RecordError(err)
pt.cancel(commonv1.Code_SchedError, err.Error())
return err
}
needBackSource = true
// can not detect source or scheduler error, create a new dummy scheduler client
pt.schedulerClient = &dummySchedulerClient{}
result = &schedulerv1.RegisterResult{TaskId: pt.taskID}
pt.Warnf("register peer task failed: %s, peer id: %s, try to back source", err, pt.request.PeerId)
} else {
pt.Infof("register task success, SizeScope: %s", commonv1.SizeScope_name[int32(result.SizeScope)])
}
var header map[string]string
if !needBackSource {
sizeScope = result.SizeScope
switch result.SizeScope {
case commonv1.SizeScope_NORMAL:
pt.span.SetAttributes(config.AttributePeerTaskSizeScope.String("normal"))
case commonv1.SizeScope_SMALL:
pt.span.SetAttributes(config.AttributePeerTaskSizeScope.String("small"))
if piece, ok := result.DirectPiece.(*schedulerv1.RegisterResult_SinglePiece); ok {
singlePiece = piece.SinglePiece
}
if result.ExtendAttribute != nil {
header = result.ExtendAttribute.Header
}
case commonv1.SizeScope_TINY:
pt.span.SetAttributes(config.AttributePeerTaskSizeScope.String("tiny"))
if piece, ok := result.DirectPiece.(*schedulerv1.RegisterResult_PieceContent); ok {
tinyData = &TinyData{
TaskID: result.TaskId,
PeerID: pt.request.PeerId,
Content: piece.PieceContent,
}
} else {
err = errors.New("scheduler return tiny piece but can not parse piece content")
// when peer register failed, some actions need to do with peerPacketStream
pt.peerPacketStream = &dummyPeerPacketStream{}
pt.span.RecordError(err)
pt.Errorf("%s", err)
pt.cancel(commonv1.Code_SchedError, err.Error())
return err
}
if result.ExtendAttribute != nil {
header = result.ExtendAttribute.Header
}
case commonv1.SizeScope_EMPTY:
tinyData = &TinyData{
TaskID: result.TaskId,
PeerID: pt.request.PeerId,
Content: []byte{},
}
pt.span.SetAttributes(config.AttributePeerTaskSizeScope.String("empty"))
if result.ExtendAttribute != nil {
header = result.ExtendAttribute.Header
}
}
}
peerPacketStream, err := pt.schedulerClient.ReportPieceResult(pt.ctx, pt.request)
pt.Infof("step 2: start report piece result")
if err != nil {
// when peer register failed, some actions need to do with peerPacketStream
pt.peerPacketStream = &dummyPeerPacketStream{}
pt.span.RecordError(err)
pt.cancel(commonv1.Code_SchedError, err.Error())
return err
}
pt.peerPacketStream = peerPacketStream
pt.sizeScope = sizeScope
pt.singlePiece = singlePiece
pt.tinyData = tinyData
pt.needBackSource = atomic.NewBool(needBackSource)
if len(header) > 0 {
pt.SetHeader(header)
}
return nil
}
func (pt *peerTaskConductor) start() error {
// when is seed task, setup back source
if pt.seed {
pt.peerPacketStream = &dummyPeerPacketStream{}
pt.schedulerClient = &dummySchedulerClient{}
pt.sizeScope = commonv1.SizeScope_NORMAL
pt.needBackSource = atomic.NewBool(true)
} else {
// register to scheduler
if err := pt.register(); err != nil {
return err
}
}
pt.trafficShaper.AddTask(pt.peerTaskManager.getRunningTaskKey(pt.taskID, pt.peerID), pt)
go pt.broker.Start()
go pt.pullPieces()
return nil
}
func (pt *peerTaskConductor) GetPeerID() string {
return pt.peerID
}
func (pt *peerTaskConductor) GetTaskID() string {
return pt.taskID
}
func (pt *peerTaskConductor) GetStorage() storage.TaskStorageDriver {
return pt.storage
}
func (pt *peerTaskConductor) GetContentLength() int64 {
return pt.contentLength.Load()
}
func (pt *peerTaskConductor) SetContentLength(i int64) {
pt.contentLength.Store(i)
}
func (pt *peerTaskConductor) AddTraffic(n uint64) {
pt.usedTraffic.Add(n)
}
func (pt *peerTaskConductor) GetTraffic() uint64 {
return pt.usedTraffic.Load()
}
func (pt *peerTaskConductor) GetTotalPieces() int32 {
return pt.totalPiece.Load()
}
func (pt *peerTaskConductor) SetTotalPieces(i int32) {
pt.totalPiece.Store(i)
}
func (pt *peerTaskConductor) SetPieceMd5Sign(md5 string) {
pt.digest.Store(md5)
}
func (pt *peerTaskConductor) GetPieceMd5Sign() string {
return pt.digest.Load()
}
func (pt *peerTaskConductor) SetHeader(header map[string]string) {
var hdr = &source.Header{}
for k, v := range header {
hdr.Set(k, v)
}
pt.header.Store(hdr)
}
func (pt *peerTaskConductor) GetHeader() *source.Header {
hdr := pt.header.Load()
if hdr != nil {
return hdr.(*source.Header)
}
return nil
}
func (pt *peerTaskConductor) Context() context.Context {
return pt.ctx
}
func (pt *peerTaskConductor) Log() *logger.SugaredLoggerOnWith {
return pt.SugaredLoggerOnWith
}
func (pt *peerTaskConductor) UpdateSourceErrorStatus(st *status.Status) {
pt.sourceErrorStatus = st
}
func (pt *peerTaskConductor) cancel(code commonv1.Code, reason string) {
pt.statusOnce.Do(func() {
pt.failedCode = code
pt.failedReason = reason
pt.fail()
})
}
func (pt *peerTaskConductor) cancelNotRegisterred(code commonv1.Code, reason string) {
pt.statusOnce.Do(func() {
pt.failedCode = code
pt.failedReason = reason
metrics.PeerTaskFailedCount.WithLabelValues(metrics.FailTypeInit).Add(1)
pt.peerTaskManager.PeerTaskDone(pt.taskID, pt.peerID)
pt.span.SetAttributes(config.AttributePeerTaskSuccess.Bool(false))
pt.span.SetAttributes(config.AttributePeerTaskCode.Int(int(pt.failedCode)))
pt.span.SetAttributes(config.AttributePeerTaskMessage.String(pt.failedReason))
close(pt.failCh)
pt.broker.Stop()
pt.span.End()
pt.pieceDownloadCancel()
if pt.pieceTaskSyncManager != nil {
pt.pieceTaskSyncManager.cancel()
}
})
}
// only use when receive back source code from scheduler
func (pt *peerTaskConductor) markBackSource() {
pt.needBackSource.Store(true)
}
// only use when legacy get piece from peers schedule timeout
func (pt *peerTaskConductor) forceBackSource() {
pt.needBackSource.Store(true)
pt.backSource()
}
func (pt *peerTaskConductor) backSource() {
// cancel all piece download
pt.pieceDownloadCancel()
// cancel all sync pieces
if pt.pieceTaskSyncManager != nil {
pt.pieceTaskSyncManager.cancel()
}
ctx, span := tracer.Start(pt.ctx, config.SpanBackSource)
pt.SetContentLength(-1)
err := pt.PieceManager.DownloadSource(ctx, pt, pt.request, pt.rg)
if err != nil {
pt.Errorf("download from source error: %s", err)
span.SetAttributes(config.AttributePeerTaskSuccess.Bool(false))
span.RecordError(err)
if isBackSourceError(err) {
pt.cancel(commonv1.Code_ClientBackSourceError, err.Error())
} else {
pt.cancel(commonv1.Code_ClientError, err.Error())
}
span.End()
return
}
pt.Done()
pt.Infof("download from source ok")
span.SetAttributes(config.AttributePeerTaskSuccess.Bool(true))
span.End()
return
}
func (pt *peerTaskConductor) pullPieces() {
if pt.needBackSource.Load() {
pt.backSource()
return
}
switch pt.sizeScope {
case commonv1.SizeScope_EMPTY:
pt.storeEmptyPeerTask()
case commonv1.SizeScope_TINY:
pt.storeTinyPeerTask()
case commonv1.SizeScope_SMALL:
pt.pullSinglePiece()
case commonv1.SizeScope_NORMAL:
pt.pullPiecesWithP2P()
default:
pt.cancel(commonv1.Code_ClientError, fmt.Sprintf("unknown size scope: %d", pt.sizeScope))
}
}
func (pt *peerTaskConductor) pullPiecesWithP2P() {
var (
// keep same size with pt.failedPieceCh for avoiding deadlock
pieceRequestQueue = NewPieceDispatcher(config.DefaultPieceDispatcherRandomRatio, pt.Log())
)
ctx, cancel := context.WithCancel(pt.ctx)
pt.pieceTaskSyncManager = &pieceTaskSyncManager{
ctx: ctx,
ctxCancel: cancel,
peerTaskConductor: pt,
pieceRequestQueue: pieceRequestQueue,
workers: map[string]*pieceTaskSynchronizer{},
}
pt.receivePeerPacket(pieceRequestQueue)
}
func (pt *peerTaskConductor) storeEmptyPeerTask() {
pt.SetContentLength(0)
pt.SetTotalPieces(0)
ctx := pt.ctx
var err error
storageDriver, err := pt.StorageManager.RegisterTask(ctx,
&storage.RegisterTaskRequest{
PeerTaskMetadata: storage.PeerTaskMetadata{
PeerID: pt.peerID,
TaskID: pt.taskID,
},
DesiredLocation: "",
ContentLength: 0,
TotalPieces: 0,
})
pt.storage = storageDriver
if err != nil {
pt.Errorf("register tiny data storage failed: %s", err)
pt.cancel(commonv1.Code_ClientError, err.Error())
return
}
if err = pt.UpdateStorage(); err != nil {
pt.Errorf("update tiny data storage failed: %s", err)
pt.cancel(commonv1.Code_ClientError, err.Error())
return
}
pt.Debug("store empty metadata")
pt.Done()
}
func (pt *peerTaskConductor) storeTinyPeerTask() {
contentLength := int64(len(pt.tinyData.Content))
pt.SetContentLength(contentLength)
pt.SetTotalPieces(1)
ctx := pt.ctx
var err error
storageDriver, err := pt.StorageManager.RegisterTask(ctx,
&storage.RegisterTaskRequest{
PeerTaskMetadata: storage.PeerTaskMetadata{
PeerID: pt.tinyData.PeerID,
TaskID: pt.tinyData.TaskID,
},
DesiredLocation: "",
ContentLength: contentLength,
TotalPieces: 1,
// TODO check digest
})
pt.storage = storageDriver
if err != nil {
pt.Errorf("register tiny data storage failed: %s", err)
pt.cancel(commonv1.Code_ClientError, err.Error())
return
}
n, err := pt.GetStorage().WritePiece(ctx,
&storage.WritePieceRequest{
PeerTaskMetadata: storage.PeerTaskMetadata{
PeerID: pt.tinyData.PeerID,
TaskID: pt.tinyData.TaskID,
},
PieceMetadata: storage.PieceMetadata{
Num: 0,
Md5: "",
Offset: 0,
Range: nethttp.Range{
Start: 0,
Length: contentLength,
},
Style: 0,
},
UnknownLength: false,
Reader: bytes.NewBuffer(pt.tinyData.Content),
GenMetadata: func(n int64) (int32, int64, bool) {
return 1, contentLength, true
},
})
if err != nil {
pt.Errorf("write tiny data storage failed: %s", err)
pt.cancel(commonv1.Code_ClientError, err.Error())
return
}
if n != contentLength {
pt.Errorf("write tiny data storage failed, want: %d, wrote: %d", contentLength, n)
pt.cancel(commonv1.Code_ClientError, err.Error())
return
}
err = pt.UpdateStorage()
if err != nil {
pt.Errorf("update tiny data storage failed: %s", err)
pt.cancel(commonv1.Code_ClientError, err.Error())
return
}
pt.Debugf("store tiny data, len: %d", contentLength)
pt.PublishPieceInfo(0, uint32(contentLength))
}
func (pt *peerTaskConductor) receivePeerPacket(pieceRequestQueue PieceDispatcher) {
var (
lastNotReadyPiece int32 = 0
peerPacket *schedulerv1.PeerPacket
err error
firstPacketReceived bool
firstPacketDone = make(chan bool)
)
// only record first schedule result
// other schedule result will record as an event in peer task span
_, firstPeerSpan := tracer.Start(pt.ctx, config.SpanFirstSchedule)
defer func() {
if !firstPacketReceived {
firstPeerSpan.End()
}
if pt.needBackSource.Load() {
return
}
select {
case <-pt.successCh:
case <-pt.failCh:
default:
pt.Errorf("receivePeerPacket exit, but peer task not success or fail")
pt.Fail()
}
}()
go pt.waitFirstPeerPacket(firstPacketDone)
loop:
for {
select {
case <-pt.successCh:
pt.Infof("peer task success, stop wait peer packet from scheduler")
break loop
case <-pt.failCh:
pt.Infof("peer task fail, stop wait peer packet from scheduler")
break loop
default:
}
peerPacket, err = pt.peerPacketStream.Recv()
if err == io.EOF {
pt.Debugf("peerPacketStream closed")
break loop
}
if err != nil {
// some errors, like commonv1.Code_SchedReregister, after reregister success,
// we can continue to receive peer packet from the new scheduler
cont := pt.confirmReceivePeerPacketError(err)
if cont {
continue
}
if !firstPacketReceived {
firstPeerSpan.RecordError(err)
}
break loop
}
pt.Debugf("receive peerPacket %v", peerPacket)
if peerPacket.Code != commonv1.Code_Success {
if peerPacket.Code == commonv1.Code_SchedNeedBackSource {
pt.forceBackSource()
pt.Infof("receive back source code")
return
}
pt.Errorf("receive peer packet with error: %d", peerPacket.Code)
if pt.isExitPeerPacketCode(peerPacket) {
pt.Errorf(pt.failedReason)
pt.cancel(pt.failedCode, pt.failedReason)
if !firstPacketReceived {
firstPeerSpan.RecordError(fmt.Errorf(pt.failedReason))
}
pt.span.AddEvent("receive exit peer packet",
trace.WithAttributes(config.AttributePeerPacketCode.Int(int(peerPacket.Code))))
pt.span.RecordError(fmt.Errorf(pt.failedReason))
break
} else {
pt.span.AddEvent("receive not success peer packet",
trace.WithAttributes(config.AttributePeerPacketCode.Int(int(peerPacket.Code))))
}
continue
}
if peerPacket.MainPeer == nil && peerPacket.CandidatePeers == nil {
pt.Warnf("scheduler client send a peerPacket with empty peers")
continue
}
pt.Infof("receive new peer packet, main peer: %s, parallel count: %d",
peerPacket.MainPeer.PeerId, peerPacket.ParallelCount)
pt.span.AddEvent("receive new peer packet",
trace.WithAttributes(config.AttributeMainPeer.String(peerPacket.MainPeer.PeerId)))
if !firstPacketReceived {
pt.initDownloadPieceWorkers(peerPacket.ParallelCount, pieceRequestQueue)
firstPeerSpan.SetAttributes(config.AttributeMainPeer.String(peerPacket.MainPeer.PeerId))
firstPeerSpan.End()
}
lastNotReadyPiece = pt.updateSynchronizers(lastNotReadyPiece, peerPacket)
if !firstPacketReceived {
// trigger legacy get piece once to avoid first schedule timeout
firstPacketReceived = true
close(firstPacketDone)
}
}
}
// updateSynchronizers will convert peers to synchronizer, if failed, will update failed peers to schedulerv1.PeerPacket
func (pt *peerTaskConductor) updateSynchronizers(lastNum int32, p *schedulerv1.PeerPacket) int32 {
desiredPiece, ok := pt.getNextNotReadyPieceNum(lastNum)
if !ok {
pt.Infof("all pieces is ready, peer task completed, skip to synchronize")
p.MainPeer = nil
p.CandidatePeers = nil
return desiredPiece
}
var peers = []*schedulerv1.PeerPacket_DestPeer{p.MainPeer}
peers = append(peers, p.CandidatePeers...)
pt.pieceTaskSyncManager.syncPeers(peers, desiredPiece)
return desiredPiece
}
func (pt *peerTaskConductor) confirmReceivePeerPacketError(err error) (cont bool) {
select {
case <-pt.successCh:
return false
case <-pt.failCh:
return false
default:
}
var (
failedCode = commonv1.Code_UnknownError
failedReason string
)
// extract DfError for grpc status
err = dferrors.ConvertGRPCErrorToDfError(err)
de, ok := err.(*dferrors.DfError)
if ok {
switch de.Code {
case commonv1.Code_SchedNeedBackSource:
pt.forceBackSource()
pt.Infof("receive back source code")
return false
case commonv1.Code_SchedReregister:
pt.Infof("receive reregister code")
regErr := pt.register()
if regErr == nil {
pt.Infof("reregister ok")
return true
}
pt.Errorf("reregister to scheduler error: %s", regErr)
fallthrough
default:
failedCode = de.Code
failedReason = de.Message
pt.Errorf("receive peer packet failed: %s", pt.failedReason)
}
} else {
pt.Errorf("receive peer packet failed: %s", err)
}
pt.cancel(failedCode, failedReason)
return false
}
func (pt *peerTaskConductor) isExitPeerPacketCode(pp *schedulerv1.PeerPacket) bool {
switch pp.Code {
case commonv1.Code_ResourceLacked, commonv1.Code_BadRequest,
commonv1.Code_PeerTaskNotFound, commonv1.Code_UnknownError, commonv1.Code_RequestTimeOut:
// 1xxx
pt.failedCode = pp.Code
pt.failedReason = fmt.Sprintf("receive exit peer packet with code %d", pp.Code)
return true
case commonv1.Code_SchedError, commonv1.Code_SchedTaskStatusError, commonv1.Code_SchedPeerNotFound, commonv1.Code_SchedForbidden:
// 5xxx
pt.failedCode = pp.Code
pt.failedReason = fmt.Sprintf("receive exit peer packet with code %d", pp.Code)
return true
case commonv1.Code_SchedPeerGone:
pt.failedReason = reasonPeerGoneFromScheduler
pt.failedCode = commonv1.Code_SchedPeerGone
return true
case commonv1.Code_CDNTaskRegistryFail:
// 6xxx
pt.failedCode = pp.Code
pt.failedReason = fmt.Sprintf("receive exit peer packet with code %d", pp.Code)
return true
case commonv1.Code_BackToSourceAborted:
st := status.Newf(codes.Aborted, "source response is not valid")
st, err := st.WithDetails(pp.GetSourceError())
if err != nil {
pt.Errorf("convert source error details error: %s", err.Error())
return false
}
pt.sourceErrorStatus = st
return true
}
return false
}
func (pt *peerTaskConductor) pullSinglePiece() {
pt.Infof("single piece, dest peer id: %s, piece num: %d, size: %d",
pt.singlePiece.DstPid, pt.singlePiece.PieceInfo.PieceNum, pt.singlePiece.PieceInfo.RangeSize)
ctx, span := tracer.Start(pt.ctx, fmt.Sprintf(config.SpanDownloadPiece, pt.singlePiece.PieceInfo.PieceNum))
span.SetAttributes(config.AttributePiece.Int(int(pt.singlePiece.PieceInfo.PieceNum)))
pt.SetContentLength(int64(pt.singlePiece.PieceInfo.RangeSize))
pt.SetTotalPieces(1)
pt.SetPieceMd5Sign(digest.SHA256FromStrings(pt.singlePiece.PieceInfo.PieceMd5))
request := &DownloadPieceRequest{
storage: pt.GetStorage(),
piece: pt.singlePiece.PieceInfo,
log: pt.Log(),
TaskID: pt.GetTaskID(),
PeerID: pt.GetPeerID(),
DstPid: pt.singlePiece.DstPid,
DstAddr: pt.singlePiece.DstAddr,
}
if result, err := pt.PieceManager.DownloadPiece(ctx, request); err == nil {
pt.reportSuccessResult(request, result)
pt.PublishPieceInfo(request.piece.PieceNum, request.piece.RangeSize)
span.SetAttributes(config.AttributePieceSuccess.Bool(true))
span.End()
pt.Infof("single piece download success")
} else {
// fallback to download from other peers
span.RecordError(err)
span.SetAttributes(config.AttributePieceSuccess.Bool(false))
span.End()
pt.Warnf("single piece download failed, switch to download from other peers")
pt.ReportPieceResult(request, result, err)
pt.pullPiecesWithP2P()
}
}
func (pt *peerTaskConductor) updateMetadata(piecePacket *commonv1.PiecePacket) {
// update total piece
var metadataChanged bool
if piecePacket.TotalPiece > pt.GetTotalPieces() {
metadataChanged = true
pt.SetTotalPieces(piecePacket.TotalPiece)
pt.Debugf("update total piece count: %d, dst peer %s", piecePacket.TotalPiece, piecePacket.DstPid)
}
// update digest
if len(piecePacket.PieceMd5Sign) > 0 && len(pt.GetPieceMd5Sign()) == 0 {
metadataChanged = true
pt.SetPieceMd5Sign(piecePacket.PieceMd5Sign)
pt.Debugf("update digest: %s, dst peer %s", piecePacket.PieceMd5Sign, piecePacket.DstPid)
}
// update content length
if piecePacket.ContentLength > -1 && pt.GetContentLength() == -1 {
metadataChanged = true
pt.SetContentLength(piecePacket.ContentLength)
pt.span.SetAttributes(config.AttributeTaskContentLength.Int64(piecePacket.ContentLength))
pt.Debugf("update content length: %d, dst peer %s", piecePacket.ContentLength, piecePacket.DstPid)
} else if piecePacket.ContentLength > -1 && piecePacket.ContentLength != pt.GetContentLength() {
// corrupt data check
reason := fmt.Sprintf("corrupt data - content length did not match, current: %d, from piece packet: %d",
pt.GetContentLength(), piecePacket.ContentLength)
pt.Errorf(reason)
pt.cancel(commonv1.Code_ClientError, reason)
return
}
if piecePacket.ExtendAttribute != nil && len(piecePacket.ExtendAttribute.Header) > 0 && pt.GetHeader() == nil {
metadataChanged = true
pt.SetHeader(piecePacket.ExtendAttribute.Header)
pt.Debugf("update response header: %#v, dst peer %s", piecePacket.ExtendAttribute.Header, piecePacket.DstPid)
}
if metadataChanged {
err := pt.UpdateStorage()
if err != nil {
pt.Errorf("update storage error: %s", err)
}
}
}
func (pt *peerTaskConductor) initDownloadPieceWorkers(count int32, pieceRequestQueue PieceDispatcher) {
if count < 1 {
count = 4
}
for i := int32(0); i < count; i++ {
go pt.downloadPieceWorker(i, pieceRequestQueue)
}
}
func (pt *peerTaskConductor) waitFirstPeerPacket(done chan bool) {
// wait first available peer
select {
case <-pt.successCh:
pt.Infof("peer task succeed, no need to wait first peer")
return
case <-pt.failCh:
pt.Warnf("peer task failed, no need to wait first peer")
return
case <-done:
pt.Debugf("first peer packet received")
return
case <-time.After(pt.SchedulerOption.ScheduleTimeout.Duration):
if pt.SchedulerOption.DisableAutoBackSource {
pt.cancel(commonv1.Code_ClientScheduleTimeout, reasonBackSourceDisabled)
err := fmt.Errorf("%s, auto back source disabled", pt.failedReason)
pt.span.RecordError(err)
pt.Errorf(err.Error())
return
}
pt.Warnf("start download from source due to %s", reasonScheduleTimeout)
pt.span.AddEvent("back source due to schedule timeout")
pt.forceBackSource()
return
}
}
func (pt *peerTaskConductor) downloadPieceWorker(id int32, requests PieceDispatcher) {
for {
request, err := requests.Get()
if errors.Is(err, ErrNoValidPieceTemporarily) {
continue
}
if err != nil {
pt.Infof("piece download queue cancelled, peer download worker #%d exit, err: %v", id, err)
return
}
pt.readyPiecesLock.RLock()
if pt.readyPieces.IsSet(request.piece.PieceNum) {
pt.readyPiecesLock.RUnlock()
pt.Log().Debugf("piece %d is already downloaded, skip", request.piece.PieceNum)
continue
}
pt.readyPiecesLock.RUnlock()
result := pt.downloadPiece(id, request)
if result != nil {