/
replication_coordinator_impl_heartbeat.cpp
1256 lines (1116 loc) · 56.4 KB
/
replication_coordinator_impl_heartbeat.cpp
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 (C) 2018-present MongoDB, Inc.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the Server Side Public License, version 1,
* as published by MongoDB, Inc.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* Server Side Public License for more details.
*
* You should have received a copy of the Server Side Public License
* along with this program. If not, see
* <http://www.mongodb.com/licensing/server-side-public-license>.
*
* As a special exception, the copyright holders give permission to link the
* code of portions of this program with the OpenSSL library under certain
* conditions as described in each individual source file and distribute
* linked combinations including the program with the OpenSSL library. You
* must comply with the Server Side Public License in all respects for
* all of the code used other than as permitted herein. If you modify file(s)
* with this exception, you may extend this exception to your version of the
* file(s), but you are not obligated to do so. If you do not wish to do so,
* delete this exception statement from your version. If you delete this
* exception statement from all source files in the program, then also delete
* it in the license file.
*/
#define MONGO_LOGV2_DEFAULT_COMPONENT ::mongo::logv2::LogComponent::kReplication
#define LOGV2_FOR_ELECTION(ID, DLEVEL, MESSAGE, ...) \
LOGV2_DEBUG_OPTIONS( \
ID, DLEVEL, {logv2::LogComponent::kReplicationElection}, MESSAGE, ##__VA_ARGS__)
#define LOGV2_FOR_HEARTBEATS(ID, DLEVEL, MESSAGE, ...) \
LOGV2_DEBUG_OPTIONS( \
ID, DLEVEL, {logv2::LogComponent::kReplicationHeartbeats}, MESSAGE, ##__VA_ARGS__)
#include "mongo/platform/basic.h"
#include <algorithm>
#include <functional>
#include "mongo/base/status.h"
#include "mongo/db/commands/test_commands_enabled.h"
#include "mongo/db/index_builds_coordinator.h"
#include "mongo/db/kill_sessions_local.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/repl/heartbeat_response_action.h"
#include "mongo/db/repl/repl_server_parameters_gen.h"
#include "mongo/db/repl/repl_set_config_checks.h"
#include "mongo/db/repl/repl_set_heartbeat_args_v1.h"
#include "mongo/db/repl/repl_set_heartbeat_response.h"
#include "mongo/db/repl/replica_set_aware_service.h"
#include "mongo/db/repl/replication_coordinator_impl.h"
#include "mongo/db/repl/replication_metrics.h"
#include "mongo/db/repl/replication_process.h"
#include "mongo/db/repl/topology_coordinator.h"
#include "mongo/db/repl/vote_requester.h"
#include "mongo/db/service_context.h"
#include "mongo/db/storage/control/journal_flusher.h"
#include "mongo/logv2/log.h"
#include "mongo/platform/mutex.h"
#include "mongo/rpc/get_status_from_command_result.h"
#include "mongo/rpc/metadata/repl_set_metadata.h"
#include "mongo/util/assert_util.h"
#include "mongo/util/fail_point.h"
#include "mongo/util/str.h"
#include "mongo/util/time_support.h"
namespace mongo {
namespace repl {
namespace {
MONGO_FAIL_POINT_DEFINE(blockHeartbeatStepdown);
MONGO_FAIL_POINT_DEFINE(blockHeartbeatReconfigFinish);
MONGO_FAIL_POINT_DEFINE(hangAfterTrackingNewHandleInHandleHeartbeatResponseForTest);
MONGO_FAIL_POINT_DEFINE(waitForPostActionCompleteInHbReconfig);
} // namespace
using executor::RemoteCommandRequest;
Milliseconds ReplicationCoordinatorImpl::_getRandomizedElectionOffset_inlock() {
long long electionTimeout = durationCount<Milliseconds>(_rsConfig.getElectionTimeoutPeriod());
long long randomOffsetUpperBound =
electionTimeout * _externalState->getElectionTimeoutOffsetLimitFraction();
// Avoid divide by zero error in random number generator.
if (randomOffsetUpperBound == 0) {
return Milliseconds(0);
}
return Milliseconds{_nextRandomInt64_inlock(randomOffsetUpperBound)};
}
void ReplicationCoordinatorImpl::_doMemberHeartbeat(executor::TaskExecutor::CallbackArgs cbData,
const HostAndPort& target) {
stdx::lock_guard<Latch> lk(_mutex);
_untrackHeartbeatHandle_inlock(cbData.myHandle);
if (cbData.status == ErrorCodes::CallbackCanceled) {
return;
}
const Date_t now = _replExecutor->now();
BSONObj heartbeatObj;
Milliseconds timeout(0);
const std::pair<ReplSetHeartbeatArgsV1, Milliseconds> hbRequest =
_topCoord->prepareHeartbeatRequestV1(now, _settings.ourSetName(), target);
heartbeatObj = hbRequest.first.toBSON();
timeout = hbRequest.second;
const RemoteCommandRequest request(
target, "admin", heartbeatObj, BSON(rpc::kReplSetMetadataFieldName << 1), nullptr, timeout);
const executor::TaskExecutor::RemoteCommandCallbackFn callback =
[=](const executor::TaskExecutor::RemoteCommandCallbackArgs& cbData) {
return _handleHeartbeatResponse(cbData);
};
LOGV2_FOR_HEARTBEATS(4615670,
2,
"Sending heartbeat (requestId: {requestId}) to {target} {heartbeatObj}",
"Sending heartbeat",
"requestId"_attr = request.id,
"target"_attr = target,
"heartbeatObj"_attr = heartbeatObj);
_trackHeartbeatHandle_inlock(
_replExecutor->scheduleRemoteCommand(request, callback), HeartbeatState::kSent, target);
}
void ReplicationCoordinatorImpl::_scheduleHeartbeatToTarget_inlock(const HostAndPort& target,
Date_t when) {
LOGV2_FOR_HEARTBEATS(4615618,
2,
"Scheduling heartbeat to {target} at {when}",
"Scheduling heartbeat",
"target"_attr = target,
"when"_attr = when);
_trackHeartbeatHandle_inlock(
_replExecutor->scheduleWorkAt(when,
[=](const executor::TaskExecutor::CallbackArgs& cbData) {
_doMemberHeartbeat(cbData, target);
}),
HeartbeatState::kScheduled,
target);
}
void ReplicationCoordinatorImpl::handleHeartbeatResponse_forTest(BSONObj response,
int targetIndex,
Milliseconds ping) {
// Make a handle to a valid no-op.
CallbackHandle handle = uassertStatusOK(
_replExecutor->scheduleWork([=](const executor::TaskExecutor::CallbackArgs& args) {}));
RemoteCommandRequest request;
{
stdx::unique_lock<Latch> lk(_mutex);
request.target = _rsConfig.getMemberAt(targetIndex).getHostAndPort();
// Simulate preparing a heartbeat request so that the target's ping stats are initialized.
_topCoord->prepareHeartbeatRequestV1(
_replExecutor->now(), _rsConfig.getReplSetName(), request.target);
// Pretend we sent a request so that _untrackHeartbeatHandle_inlock succeeds.
_trackHeartbeatHandle_inlock(handle, HeartbeatState::kSent, request.target);
}
executor::TaskExecutor::ResponseStatus status(response, ping);
executor::TaskExecutor::RemoteCommandCallbackArgs cbData(
_replExecutor.get(), handle, request, status);
hangAfterTrackingNewHandleInHandleHeartbeatResponseForTest.pauseWhileSet();
_handleHeartbeatResponse(cbData);
}
void ReplicationCoordinatorImpl::_handleHeartbeatResponse(
const executor::TaskExecutor::RemoteCommandCallbackArgs& cbData) {
stdx::unique_lock<Latch> lk(_mutex);
// remove handle from queued heartbeats
_untrackHeartbeatHandle_inlock(cbData.myHandle);
// Parse and validate the response. At the end of this step, if responseStatus is OK then
// hbResponse is valid.
Status responseStatus = cbData.response.status;
const HostAndPort& target = cbData.request.target;
if (responseStatus == ErrorCodes::CallbackCanceled) {
LOGV2_FOR_HEARTBEATS(4615619,
2,
"Received response to heartbeat (requestId: {requestId}) from "
"{target} but the heartbeat was cancelled.",
"Received response to heartbeat, but the heartbeat was cancelled",
"requestId"_attr = cbData.request.id,
"target"_attr = target);
return;
}
ReplSetHeartbeatResponse hbResponse;
BSONObj resp;
if (responseStatus.isOK()) {
resp = cbData.response.data;
responseStatus = hbResponse.initialize(resp, _topCoord->getTerm());
StatusWith<rpc::ReplSetMetadata> replMetadata =
rpc::ReplSetMetadata::readFromMetadata(cbData.response.data);
LOGV2_FOR_HEARTBEATS(
4615620,
2,
"Received response to heartbeat (requestId: {requestId}) from {target}, {response}",
"Received response to heartbeat",
"requestId"_attr = cbData.request.id,
"target"_attr = target,
"response"_attr = resp);
// Reject heartbeat responses (and metadata) from nodes with mismatched replica set IDs.
// It is problematic to perform this check in the heartbeat reconfiguring logic because it
// is possible for two mismatched replica sets to have the same replica set name and
// configuration version. A heartbeat reconfiguration would not take place in that case.
// Additionally, this is where we would stop further processing of the metadata from an
// unknown replica set.
if (replMetadata.isOK() && _rsConfig.isInitialized() && _rsConfig.hasReplicaSetId() &&
replMetadata.getValue().getReplicaSetId().isSet() &&
_rsConfig.getReplicaSetId() != replMetadata.getValue().getReplicaSetId()) {
responseStatus =
Status(ErrorCodes::InvalidReplicaSetConfig,
str::stream()
<< "replica set IDs do not match, ours: " << _rsConfig.getReplicaSetId()
<< "; remote node's: " << replMetadata.getValue().getReplicaSetId());
// Ignore metadata.
replMetadata = responseStatus;
}
if (replMetadata.isOK()) {
// It is safe to update our commit point via heartbeat propagation as long as the
// the new commit point we learned of is on the same branch of history as our own
// oplog.
if (_getMemberState_inlock().arbiter() ||
(!_getMemberState_inlock().startup() && !_getMemberState_inlock().startup2())) {
// The node that sent the heartbeat is not guaranteed to be our sync source.
const bool fromSyncSource = false;
_advanceCommitPoint(
lk, replMetadata.getValue().getLastOpCommitted(), fromSyncSource);
}
// Asynchronous stepdown could happen, but it will wait for _mutex and execute
// after this function, so we cannot and don't need to wait for it to finish.
_processReplSetMetadata_inlock(replMetadata.getValue());
}
// Arbiters are always expected to report null durable optimes (and wall times).
// If that is not the case here, make sure to correct these times before ingesting them.
auto memberInConfig = _rsConfig.findMemberByHostAndPort(target);
if ((hbResponse.hasState() && hbResponse.getState().arbiter()) ||
(_rsConfig.isInitialized() && memberInConfig && memberInConfig->isArbiter())) {
if (hbResponse.hasDurableOpTime() &&
(!hbResponse.getDurableOpTime().isNull() ||
hbResponse.getDurableOpTimeAndWallTime().wallTime != Date_t())) {
LOGV2_FOR_HEARTBEATS(
5662000,
1,
"Received non-null durable optime/walltime for arbiter from heartbeat. "
"Ignoring value(s).",
"target"_attr = target,
"durableOpTimeAndWallTime"_attr = hbResponse.getDurableOpTimeAndWallTime());
hbResponse.unsetDurableOpTimeAndWallTime();
}
}
}
const Date_t now = _replExecutor->now();
Microseconds networkTime(0);
StatusWith<ReplSetHeartbeatResponse> hbStatusResponse(hbResponse);
if (responseStatus.isOK()) {
networkTime = cbData.response.elapsed.value_or(Microseconds{0});
// TODO(sz) Because the term is duplicated in ReplSetMetaData, we can get rid of this
// and update tests.
const auto& hbResponse = hbStatusResponse.getValue();
_updateTerm_inlock(hbResponse.getTerm());
// Postpone election timeout if we have a successful heartbeat response from the primary.
if (hbResponse.hasState() && hbResponse.getState().primary() &&
hbResponse.getTerm() == _topCoord->getTerm()) {
LOGV2_FOR_ELECTION(
4615659, 4, "Postponing election timeout due to heartbeat from primary");
_cancelAndRescheduleElectionTimeout_inlock();
}
} else {
LOGV2_FOR_HEARTBEATS(4615621,
2,
"Error in heartbeat (requestId: {requestId}) to {target}, "
"response status: {error}",
"Error in heartbeat",
"requestId"_attr = cbData.request.id,
"target"_attr = target,
"error"_attr = responseStatus);
hbStatusResponse = StatusWith<ReplSetHeartbeatResponse>(responseStatus);
}
// Leaving networkTime units as ms since the average ping calulation may be affected.
HeartbeatResponseAction action = _topCoord->processHeartbeatResponse(
now, duration_cast<Milliseconds>(networkTime), target, hbStatusResponse);
if (action.getAction() == HeartbeatResponseAction::NoAction && hbStatusResponse.isOK() &&
hbStatusResponse.getValue().hasState() &&
hbStatusResponse.getValue().getState() != MemberState::RS_PRIMARY) {
if (action.getAdvancedOpTimeOrUpdatedConfig()) {
// If a member's opTime has moved forward or config is newer, try to update the
// lastCommitted. Even if we've only updated the config, this is still safe.
_updateLastCommittedOpTimeAndWallTime(lk);
// Wake up replication waiters on optime changes or updated configs.
_wakeReadyWaiters(lk);
} else if (action.getBecameElectable() && _topCoord->isSteppingDown()) {
// Try to wake up the stepDown waiter when a new node becomes electable.
_wakeReadyWaiters(lk);
}
}
// When receiving a heartbeat response indicating that the remote is in a state past
// STARTUP_2, the primary will initiate a reconfig to remove the 'newlyAdded' field for that
// node (if present). This field is normally set when we add new members with votes:1 to the
// set.
if (_getMemberState_inlock().primary() && hbStatusResponse.isOK() &&
hbStatusResponse.getValue().hasState()) {
auto remoteState = hbStatusResponse.getValue().getState();
if (remoteState == MemberState::RS_SECONDARY || remoteState == MemberState::RS_RECOVERING ||
remoteState == MemberState::RS_ROLLBACK) {
const auto mem = _rsConfig.findMemberByHostAndPort(target);
if (mem && mem->isNewlyAdded()) {
// 'NewlyAdded' field can only exist if automatic reconfig is supported, with the
// exception of upgrading/downgrading fcv document. And, it's safe to have that
// exception because a node can't downgrade the binary version until its FCV
// document is fully downgraded. So, its impossible for a node with downgraded
// binaries to have on-disk repl config with 'newlyAdded' fields.
invariant(
_supportsAutomaticReconfig() ||
serverGlobalParams.featureCompatibility.isGreaterThan(
ServerGlobalParams::FeatureCompatibility::Version::kFullyDowngradedTo44));
const auto memId = mem->getId();
auto status = _replExecutor->scheduleWork(
[=](const executor::TaskExecutor::CallbackArgs& cbData) {
_reconfigToRemoveNewlyAddedField(
cbData, memId, _rsConfig.getConfigVersionAndTerm());
});
if (!status.isOK()) {
LOGV2_DEBUG(4634500,
1,
"Failed to schedule work for removing 'newlyAdded' field.",
"memberId"_attr = memId.getData(),
"error"_attr = status.getStatus());
} else {
LOGV2_DEBUG(4634501,
1,
"Scheduled automatic reconfig to remove 'newlyAdded' field.",
"memberId"_attr = memId.getData());
}
}
}
}
// Abort catchup if we have caught up to the latest known optime after heartbeat refreshing.
if (_catchupState) {
_catchupState->signalHeartbeatUpdate_inlock();
}
// Cancel catchup takeover if the last applied write by any node in the replica set was made
// in the current term, which implies that the primary has caught up.
bool catchupTakeoverScheduled = _catchupTakeoverCbh.isValid();
if (responseStatus.isOK() && catchupTakeoverScheduled && hbResponse.hasAppliedOpTime()) {
const auto& hbLastAppliedOpTime = hbResponse.getAppliedOpTime();
if (hbLastAppliedOpTime.getTerm() == _topCoord->getTerm()) {
_cancelCatchupTakeover_inlock();
}
}
_scheduleHeartbeatToTarget_inlock(target, std::max(now, action.getNextHeartbeatStartDate()));
_handleHeartbeatResponseAction_inlock(action, hbStatusResponse, std::move(lk));
}
stdx::unique_lock<Latch> ReplicationCoordinatorImpl::_handleHeartbeatResponseAction_inlock(
const HeartbeatResponseAction& action,
const StatusWith<ReplSetHeartbeatResponse>& responseStatus,
stdx::unique_lock<Latch> lock) {
invariant(lock.owns_lock());
switch (action.getAction()) {
case HeartbeatResponseAction::NoAction:
// Update the cached member state if different than the current topology member state
if (_memberState != _topCoord->getMemberState()) {
const PostMemberStateUpdateAction postUpdateAction =
_updateMemberStateFromTopologyCoordinator(lock);
lock.unlock();
_performPostMemberStateUpdateAction(postUpdateAction);
lock.lock();
}
break;
case HeartbeatResponseAction::Reconfig:
invariant(responseStatus.isOK());
_scheduleHeartbeatReconfig(lock, responseStatus.getValue().getConfig());
break;
case HeartbeatResponseAction::RetryReconfig:
_scheduleHeartbeatReconfig(lock, _rsConfig);
break;
case HeartbeatResponseAction::StepDownSelf:
invariant(action.getPrimaryConfigIndex() == _selfIndex);
if (_topCoord->prepareForUnconditionalStepDown()) {
LOGV2(21475, "Stepping down from primary in response to heartbeat");
_stepDownStart();
} else {
LOGV2_DEBUG(21476,
2,
"Heartbeat would have triggered a stepdown, but we're already in the "
"process of stepping down");
}
break;
case HeartbeatResponseAction::PriorityTakeover: {
// Don't schedule a priority takeover if any takeover is already scheduled.
if (!_priorityTakeoverCbh.isValid() && !_catchupTakeoverCbh.isValid()) {
// Add randomized offset to calculated priority takeover delay.
Milliseconds priorityTakeoverDelay = _rsConfig.getPriorityTakeoverDelay(_selfIndex);
Milliseconds randomOffset = _getRandomizedElectionOffset_inlock();
_priorityTakeoverWhen = _replExecutor->now() + priorityTakeoverDelay + randomOffset;
LOGV2_FOR_ELECTION(4615601,
0,
"Scheduling priority takeover at {when}",
"Scheduling priority takeover",
"when"_attr = _priorityTakeoverWhen);
_priorityTakeoverCbh = _scheduleWorkAt(
_priorityTakeoverWhen, [=](const mongo::executor::TaskExecutor::CallbackArgs&) {
_startElectSelfIfEligibleV1(StartElectionReasonEnum::kPriorityTakeover);
});
}
break;
}
case HeartbeatResponseAction::CatchupTakeover: {
// Don't schedule a catchup takeover if any takeover is already scheduled.
if (!_catchupTakeoverCbh.isValid() && !_priorityTakeoverCbh.isValid()) {
Milliseconds catchupTakeoverDelay = _rsConfig.getCatchUpTakeoverDelay();
_catchupTakeoverWhen = _replExecutor->now() + catchupTakeoverDelay;
LOGV2_FOR_ELECTION(4615648,
0,
"Scheduling catchup takeover at {when}",
"Scheduling catchup takeover",
"when"_attr = _catchupTakeoverWhen);
_catchupTakeoverCbh = _scheduleWorkAt(
_catchupTakeoverWhen, [=](const mongo::executor::TaskExecutor::CallbackArgs&) {
_startElectSelfIfEligibleV1(StartElectionReasonEnum::kCatchupTakeover);
});
}
break;
}
}
return lock;
}
namespace {
/**
* This callback is purely for logging and has no effect on any other operations
*/
void remoteStepdownCallback(const executor::TaskExecutor::RemoteCommandCallbackArgs& cbData) {
const Status status = cbData.response.status;
if (status == ErrorCodes::CallbackCanceled) {
return;
}
if (status.isOK()) {
LOGV2_DEBUG(21477,
1,
"stepdown of primary({primary}) succeeded with response -- "
"{response}",
"Stepdown of primary succeeded",
"primary"_attr = cbData.request.target,
"response"_attr = cbData.response.data);
} else {
LOGV2_WARNING(21486,
"stepdown of primary({primary}) failed due to {error}",
"Stepdown of primary failed",
"primary"_attr = cbData.request.target,
"error"_attr = cbData.response.status);
}
}
} // namespace
executor::TaskExecutor::EventHandle ReplicationCoordinatorImpl::_stepDownStart() {
auto finishEvent = _makeEvent();
if (!finishEvent) {
return finishEvent;
}
_replExecutor
->scheduleWork([=](const executor::TaskExecutor::CallbackArgs& cbData) {
_stepDownFinish(cbData, finishEvent);
})
.status_with_transitional_ignore();
return finishEvent;
}
void ReplicationCoordinatorImpl::_stepDownFinish(
const executor::TaskExecutor::CallbackArgs& cbData,
const executor::TaskExecutor::EventHandle& finishedEvent) {
if (cbData.status == ErrorCodes::CallbackCanceled) {
return;
}
if (MONGO_unlikely(blockHeartbeatStepdown.shouldFail())) {
// This log output is used in js tests so please leave it.
LOGV2(21479,
"stepDown - blockHeartbeatStepdown fail point enabled. "
"Blocking until fail point is disabled.");
auto inShutdown = [&] {
stdx::lock_guard<Latch> lk(_mutex);
return _inShutdown;
};
while (MONGO_unlikely(blockHeartbeatStepdown.shouldFail()) && !inShutdown()) {
mongo::sleepsecs(1);
}
}
auto opCtx = cc().makeOperationContext();
// kill all write operations which are no longer safe to run on step down. Also, operations that
// have taken global lock in S mode and operations blocked on prepare conflict will be killed to
// avoid 3-way deadlock between read, prepared transaction and step down thread.
AutoGetRstlForStepUpStepDown arsd(
this, opCtx.get(), ReplicationCoordinator::OpsKillingStateTransitionEnum::kStepDown);
stdx::unique_lock<Latch> lk(_mutex);
// This node has already stepped down due to reconfig. So, signal anyone who is waiting on the
// step down event.
if (!_topCoord->isSteppingDownUnconditionally()) {
_replExecutor->signalEvent(finishedEvent);
return;
}
// We need to release the mutex before yielding locks for prepared transactions, which might
// check out sessions, to avoid deadlocks with checked-out sessions accessing this mutex.
lk.unlock();
yieldLocksForPreparedTransactions(opCtx.get());
invalidateSessionsForStepdown(opCtx.get());
lk.lock();
// Clear the node's election candidate metrics since it is no longer primary.
ReplicationMetrics::get(opCtx.get()).clearElectionCandidateMetrics();
_topCoord->finishUnconditionalStepDown();
// Update _canAcceptNonLocalWrites.
_updateWriteAbilityFromTopologyCoordinator(lk, opCtx.get());
const auto action = _updateMemberStateFromTopologyCoordinator(lk);
if (_pendingTermUpdateDuringStepDown) {
TopologyCoordinator::UpdateTermResult result;
_updateTerm_inlock(*_pendingTermUpdateDuringStepDown, &result);
// We've just stepped down due to the "term", so it's impossible to step down again
// for the same term.
invariant(result != TopologyCoordinator::UpdateTermResult::kTriggerStepDown);
_pendingTermUpdateDuringStepDown = boost::none;
}
lk.unlock();
_performPostMemberStateUpdateAction(action);
_replExecutor->signalEvent(finishedEvent);
}
bool ReplicationCoordinatorImpl::_shouldStepDownOnReconfig(WithLock,
const ReplSetConfig& newConfig,
StatusWith<int> myIndex) {
return _memberState.primary() &&
!(myIndex.isOK() && newConfig.getMemberAt(myIndex.getValue()).isElectable());
}
void ReplicationCoordinatorImpl::_scheduleHeartbeatReconfig(WithLock lk,
const ReplSetConfig& newConfig) {
if (_inShutdown) {
return;
}
switch (_rsConfigState) {
case kConfigUninitialized:
case kConfigSteady:
LOGV2_FOR_HEARTBEATS(4615622,
1,
"Received new config via heartbeat with {newConfigVersionAndTerm}",
"Received new config via heartbeat",
"newConfigVersionAndTerm"_attr =
newConfig.getConfigVersionAndTerm());
break;
case kConfigInitiating:
case kConfigReconfiguring:
case kConfigHBReconfiguring:
LOGV2_FOR_HEARTBEATS(
4615623,
1,
"Ignoring new configuration with {newConfigVersionAndTerm} because "
"already in the midst of a configuration process.",
"Ignoring new configuration because we are already in the midst of a configuration "
"process",
"newConfigVersionAndTerm"_attr = newConfig.getConfigVersionAndTerm());
return;
case kConfigPreStart:
case kConfigStartingUp:
case kConfigReplicationDisabled:
LOGV2_FATAL(18807,
"Reconfiguration request occurred while _rsConfigState == "
"{_rsConfigState}; aborting.",
"Aborting reconfiguration request",
"_rsConfigState"_attr = int(_rsConfigState));
}
// Allow force reconfigs to proceed even if we are not a writable primary yet.
if (_memberState.primary() && !_readWriteAbility->canAcceptNonLocalWrites(lk) &&
newConfig.getConfigTerm() != OpTime::kUninitializedTerm) {
LOGV2_FOR_HEARTBEATS(
4794900,
1,
"Not scheduling a heartbeat reconfig since we are in PRIMARY state but "
"cannot accept writes yet.");
return;
}
// Prevent heartbeat reconfigs from running concurrently with an election.
if (_topCoord->getRole() == TopologyCoordinator::Role::kCandidate) {
LOGV2_FOR_HEARTBEATS(
482570, 1, "Not scheduling a heartbeat reconfig when running for election");
return;
}
_setConfigState_inlock(kConfigHBReconfiguring);
invariant(!_rsConfig.isInitialized() ||
_rsConfig.getConfigVersionAndTerm() < newConfig.getConfigVersionAndTerm() ||
_selfIndex < 0);
_replExecutor
->scheduleWork([=](const executor::TaskExecutor::CallbackArgs& cbData) {
_heartbeatReconfigStore(cbData, newConfig);
})
.status_with_transitional_ignore();
}
void ReplicationCoordinatorImpl::_heartbeatReconfigStore(
const executor::TaskExecutor::CallbackArgs& cbd, const ReplSetConfig& newConfig) {
if (cbd.status.code() == ErrorCodes::CallbackCanceled) {
LOGV2(21480,
"The callback to persist the replica set configuration was canceled - the "
"configuration was not persisted but was used: {newConfig}",
"The callback to persist the replica set configuration was canceled - the "
"configuration was not persisted but was used",
"newConfig"_attr = newConfig.toBSON());
return;
}
const StatusWith<int> myIndex = validateConfigForHeartbeatReconfig(
_externalState.get(), newConfig, getGlobalServiceContext());
if (myIndex.getStatus() == ErrorCodes::NodeNotFound) {
stdx::lock_guard<Latch> lk(_mutex);
// If this node absent in newConfig, and this node was not previously initialized,
// return to kConfigUninitialized immediately, rather than storing the config and
// transitioning into the RS_REMOVED state. See SERVER-15740.
if (!_rsConfig.isInitialized()) {
invariant(_rsConfigState == kConfigHBReconfiguring);
LOGV2_FOR_HEARTBEATS(4615625,
1,
"Ignoring new configuration in heartbeat response because we "
"are uninitialized and not a member of the new configuration");
_setConfigState_inlock(kConfigUninitialized);
return;
}
}
bool shouldStartDataReplication = false;
if (!myIndex.getStatus().isOK() && myIndex.getStatus() != ErrorCodes::NodeNotFound) {
LOGV2_WARNING(21487,
"Not persisting new configuration in heartbeat response to disk because "
"it is invalid: {error}",
"Not persisting new configuration in heartbeat response to disk because "
"it is invalid",
"error"_attr = myIndex.getStatus());
} else {
LOGV2_FOR_HEARTBEATS(4615626,
2,
"Config with {newConfigVersionAndTerm} validated for "
"reconfig; persisting to disk.",
"Config validated for reconfig; persisting to disk",
"newConfigVersionAndTerm"_attr = newConfig.getConfigVersionAndTerm());
auto opCtx = cc().makeOperationContext();
// Don't write the no-op for config learned via heartbeats.
auto status = _externalState->storeLocalConfigDocument(
opCtx.get(), newConfig.toBSON(), false /* writeOplog */);
// Wait for durability of the new config document.
JournalFlusher::get(opCtx.get())->waitForJournalFlush();
bool isFirstConfig;
{
stdx::lock_guard<Latch> lk(_mutex);
isFirstConfig = !_rsConfig.isInitialized();
if (!status.isOK()) {
LOGV2_ERROR(21488,
"Ignoring new configuration in heartbeat response because we failed to"
" write it to stable storage; {error}",
"Ignoring new configuration in heartbeat response because we failed to"
" write it to stable storage",
"error"_attr = status);
invariant(_rsConfigState == kConfigHBReconfiguring);
if (isFirstConfig) {
_setConfigState_inlock(kConfigUninitialized);
} else {
_setConfigState_inlock(kConfigSteady);
}
return;
}
}
bool isArbiter = myIndex.isOK() && myIndex.getValue() != -1 &&
newConfig.getMemberAt(myIndex.getValue()).isArbiter();
if (isArbiter) {
ReplicaSetAwareServiceRegistry::get(_service).onBecomeArbiter();
}
if (!isArbiter && myIndex.isOK() && myIndex.getValue() != -1) {
shouldStartDataReplication = true;
}
LOGV2_FOR_HEARTBEATS(
4615627,
2,
"New configuration with {newConfigVersionAndTerm} persisted "
"to local storage; installing new config in memory",
"New configuration persisted to local storage; installing new config in memory",
"newConfigVersionAndTerm"_attr = newConfig.getConfigVersionAndTerm());
}
_heartbeatReconfigFinish(cbd, newConfig, myIndex);
// Start data replication after the config has been installed.
if (shouldStartDataReplication) {
auto opCtx = cc().makeOperationContext();
_replicationProcess->getConsistencyMarkers()->initializeMinValidDocument(opCtx.get());
_externalState->startThreads();
_startDataReplication(opCtx.get());
}
}
void ReplicationCoordinatorImpl::_heartbeatReconfigFinish(
const executor::TaskExecutor::CallbackArgs& cbData,
const ReplSetConfig& newConfig,
StatusWith<int> myIndex) {
if (cbData.status == ErrorCodes::CallbackCanceled) {
return;
}
if (MONGO_unlikely(blockHeartbeatReconfigFinish.shouldFail())) {
LOGV2_FOR_HEARTBEATS(4615628,
0,
"blockHeartbeatReconfigFinish fail point enabled. Rescheduling "
"_heartbeatReconfigFinish until fail point is disabled");
_replExecutor
->scheduleWorkAt(_replExecutor->now() + Milliseconds{10},
[=](const executor::TaskExecutor::CallbackArgs& cbData) {
_heartbeatReconfigFinish(cbData, newConfig, myIndex);
})
.status_with_transitional_ignore();
return;
}
// Do not conduct an election during a reconfig, as the node may not be electable post-reconfig.
// If there is an election in-progress, there can be at most one. No new election can happen as
// we have already set our ReplicationCoordinatorImpl::_rsConfigState state to
// "kConfigReconfiguring" which prevents new elections from happening.
{
stdx::lock_guard<Latch> lk(_mutex);
if (auto electionFinishedEvent = _cancelElectionIfNeeded(lk)) {
LOGV2_FOR_HEARTBEATS(4615629,
0,
"Waiting for election to complete before finishing reconfig to "
"config with {newConfigVersionAndTerm}",
"Waiting for election to complete before finishing reconfig",
"newConfigVersionAndTerm"_attr =
newConfig.getConfigVersionAndTerm());
// Wait for the election to complete and the node's Role to be set to follower.
_replExecutor
->onEvent(electionFinishedEvent,
[=](const executor::TaskExecutor::CallbackArgs& cbData) {
_heartbeatReconfigFinish(cbData, newConfig, myIndex);
})
.status_with_transitional_ignore();
return;
}
}
auto opCtx = cc().makeOperationContext();
boost::optional<AutoGetRstlForStepUpStepDown> arsd;
stdx::unique_lock<Latch> lk(_mutex);
if (_shouldStepDownOnReconfig(lk, newConfig, myIndex)) {
_topCoord->prepareForUnconditionalStepDown();
lk.unlock();
// Primary node will be either unelectable or removed after the configuration change.
// So, finish the reconfig under RSTL, so that the step down occurs safely.
arsd.emplace(
this, opCtx.get(), ReplicationCoordinator::OpsKillingStateTransitionEnum::kStepDown);
lk.lock();
if (_topCoord->isSteppingDownUnconditionally()) {
invariant(opCtx->lockState()->isRSTLExclusive());
LOGV2(21481,
"Stepping down from primary, because we received a new config via heartbeat");
// We need to release the mutex before yielding locks for prepared transactions, which
// might check out sessions, to avoid deadlocks with checked-out sessions accessing
// this mutex.
lk.unlock();
yieldLocksForPreparedTransactions(opCtx.get());
invalidateSessionsForStepdown(opCtx.get());
lk.lock();
// Clear the node's election candidate metrics since it is no longer primary.
ReplicationMetrics::get(opCtx.get()).clearElectionCandidateMetrics();
// Update _canAcceptNonLocalWrites.
_updateWriteAbilityFromTopologyCoordinator(lk, opCtx.get());
} else {
// Release the rstl lock as the node might have stepped down due to
// other unconditional step down code paths like learning new term via heartbeat &
// liveness timeout. And, no new election can happen as we have already set our
// ReplicationCoordinatorImpl::_rsConfigState state to "kConfigReconfiguring" which
// prevents new elections from happening. So, its safe to release the RSTL lock.
arsd.reset();
}
}
invariant(_rsConfigState == kConfigHBReconfiguring);
invariant(!_rsConfig.isInitialized() ||
_rsConfig.getConfigVersionAndTerm() < newConfig.getConfigVersionAndTerm() ||
_selfIndex < 0);
if (!myIndex.isOK()) {
switch (myIndex.getStatus().code()) {
case ErrorCodes::NodeNotFound:
LOGV2(21482,
"Cannot find self in new replica set configuration; I must be removed; "
"{error}",
"Cannot find self in new replica set configuration; I must be removed",
"error"_attr = myIndex.getStatus());
break;
case ErrorCodes::InvalidReplicaSetConfig:
LOGV2_ERROR(21489,
"Several entries in new config represent this node; "
"Removing self until an acceptable configuration arrives; {error}",
"Several entries in new config represent this node; "
"Removing self until an acceptable configuration arrives",
"error"_attr = myIndex.getStatus());
break;
default:
LOGV2_ERROR(21490,
"Could not validate configuration received from remote node; "
"Removing self until an acceptable configuration arrives; {error}",
"Could not validate configuration received from remote node; "
"Removing self until an acceptable configuration arrives",
"error"_attr = myIndex.getStatus());
break;
}
myIndex = StatusWith<int>(-1);
}
const ReplSetConfig oldConfig = _rsConfig;
// If we do not have an index, we should pass -1 as our index to avoid falsely adding ourself to
// the data structures inside of the TopologyCoordinator.
const int myIndexValue = myIndex.getStatus().isOK() ? myIndex.getValue() : -1;
const PostMemberStateUpdateAction action =
_setCurrentRSConfig(lk, opCtx.get(), newConfig, myIndexValue);
lk.unlock();
_performPostMemberStateUpdateAction(action);
if (MONGO_unlikely(waitForPostActionCompleteInHbReconfig.shouldFail())) {
// Used in tests that wait for the post member state update action to complete.
// eg. Closing connections upon being removed.
LOGV2(5286701, "waitForPostActionCompleteInHbReconfig failpoint enabled");
}
}
void ReplicationCoordinatorImpl::_trackHeartbeatHandle_inlock(
const StatusWith<executor::TaskExecutor::CallbackHandle>& handle,
HeartbeatState hbState,
const HostAndPort& target) {
if (handle.getStatus() == ErrorCodes::ShutdownInProgress) {
return;
}
fassert(18912, handle.getStatus());
// The target's HostAndPort should be safe to store, because it cannot change without a
// reconfig. On reconfig, all current heartbeats get cancelled and new requests are sent out, so
// there should not be a situation where the target node's HostAndPort changes but this
// heartbeat handle remains active.
_heartbeatHandles.push_back({handle.getValue(), hbState, target});
}
void ReplicationCoordinatorImpl::_untrackHeartbeatHandle_inlock(
const executor::TaskExecutor::CallbackHandle& handle) {
const auto newEnd = std::remove_if(_heartbeatHandles.begin(),
_heartbeatHandles.end(),
[&](auto& hbHandle) { return hbHandle.handle == handle; });
invariant(newEnd != _heartbeatHandles.end());
_heartbeatHandles.erase(newEnd, _heartbeatHandles.end());
}
void ReplicationCoordinatorImpl::_cancelHeartbeats_inlock() {
LOGV2_FOR_HEARTBEATS(4615630, 2, "Cancelling all heartbeats");
for (const auto& hbHandle : _heartbeatHandles) {
_replExecutor->cancel(hbHandle.handle);
}
// Heartbeat callbacks will remove themselves from _heartbeatHandles when they execute with
// CallbackCanceled status, so it's better to leave the handles in the list, for now.
if (_handleLivenessTimeoutCbh.isValid()) {
_replExecutor->cancel(_handleLivenessTimeoutCbh);
}
}
void ReplicationCoordinatorImpl::restartScheduledHeartbeats_forTest() {
stdx::unique_lock<Latch> lk(_mutex);
invariant(getTestCommandsEnabled());
_restartScheduledHeartbeats_inlock();
};
void ReplicationCoordinatorImpl::_restartScheduledHeartbeats_inlock() {
LOGV2_FOR_HEARTBEATS(5031800, 2, "Restarting all scheduled heartbeats");
const Date_t now = _replExecutor->now();
stdx::unordered_set<HostAndPort> restartedTargets;
for (auto& hbHandle : _heartbeatHandles) {
// Only cancel heartbeats that are scheduled. If a heartbeat request has already been
// sent, we should wait for the response instead.
if (hbHandle.hbState != HeartbeatState::kScheduled) {
continue;
}
LOGV2_FOR_HEARTBEATS(5031802, 2, "Restarting heartbeat", "target"_attr = hbHandle.target);
_replExecutor->cancel(hbHandle.handle);
// Track the members that we have cancelled heartbeats.
restartedTargets.insert(hbHandle.target);
}
for (auto target : restartedTargets) {
_scheduleHeartbeatToTarget_inlock(target, now);
_topCoord->restartHeartbeat(now, target);
}
}
void ReplicationCoordinatorImpl::_startHeartbeats_inlock() {
const Date_t now = _replExecutor->now();
_seedList.clear();
for (int i = 0; i < _rsConfig.getNumMembers(); ++i) {
if (i == _selfIndex) {
continue;
}
auto target = _rsConfig.getMemberAt(i).getHostAndPort();
_scheduleHeartbeatToTarget_inlock(target, now);
_topCoord->restartHeartbeat(now, target);
}
_scheduleNextLivenessUpdate_inlock();
}
void ReplicationCoordinatorImpl::_handleLivenessTimeout(
const executor::TaskExecutor::CallbackArgs& cbData) {
stdx::unique_lock<Latch> lk(_mutex);
// Only reset the callback handle if it matches, otherwise more will be coming through
if (cbData.myHandle == _handleLivenessTimeoutCbh) {
_handleLivenessTimeoutCbh = CallbackHandle();
}
if (!cbData.status.isOK()) {
return;
}
// Scan liveness table for problems and mark nodes as down by calling into topocoord.
HeartbeatResponseAction action = _topCoord->checkMemberTimeouts(_replExecutor->now());
// Don't mind potential asynchronous stepdown as this is the last step of
// liveness check.
lk = _handleHeartbeatResponseAction_inlock(
action, StatusWith(ReplSetHeartbeatResponse()), std::move(lk));
_scheduleNextLivenessUpdate_inlock();