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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
|
/**
* 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.
*/
#include "mongo/platform/basic.h"
#include "mongo/base/checked_cast.h"
#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/bson/mutable/document.h"
#include "mongo/bson/mutable/element.h"
#include "mongo/db/catalog/database_holder.h"
#include "mongo/db/catalog/document_validation.h"
#include "mongo/db/client.h"
#include "mongo/db/commands.h"
#include "mongo/db/commands/update_metrics.h"
#include "mongo/db/commands/write_commands_common.h"
#include "mongo/db/curop.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/json.h"
#include "mongo/db/lasterror.h"
#include "mongo/db/matcher/doc_validation_error.h"
#include "mongo/db/matcher/extensions_callback_real.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/ops/delete_request_gen.h"
#include "mongo/db/ops/parsed_delete.h"
#include "mongo/db/ops/parsed_update.h"
#include "mongo/db/ops/write_ops.h"
#include "mongo/db/ops/write_ops_exec.h"
#include "mongo/db/pipeline/aggregate_command_gen.h"
#include "mongo/db/pipeline/lite_parsed_pipeline.h"
#include "mongo/db/query/explain.h"
#include "mongo/db/query/get_executor.h"
#include "mongo/db/repl/repl_client_info.h"
#include "mongo/db/repl/replication_coordinator.h"
#include "mongo/db/repl/tenant_migration_access_blocker_registry.h"
#include "mongo/db/repl/tenant_migration_conflict_info.h"
#include "mongo/db/retryable_writes_stats.h"
#include "mongo/db/stats/counters.h"
#include "mongo/db/storage/duplicate_key_error_info.h"
#include "mongo/db/storage/durable_catalog.h"
#include "mongo/db/timeseries/bucket_catalog.h"
#include "mongo/db/transaction_participant.h"
#include "mongo/db/views/view_catalog.h"
#include "mongo/db/write_concern.h"
#include "mongo/logv2/redaction.h"
#include "mongo/s/stale_exception.h"
#include "mongo/util/fail_point.h"
#include "mongo/util/string_map.h"
namespace mongo {
namespace {
MONGO_FAIL_POINT_DEFINE(hangWriteBeforeWaitingForMigrationDecision);
MONGO_FAIL_POINT_DEFINE(hangTimeseriesInsertBeforeCommit);
MONGO_FAIL_POINT_DEFINE(failTimeseriesInsert);
void redactTooLongLog(mutablebson::Document* cmdObj, StringData fieldName) {
namespace mmb = mutablebson;
mmb::Element root = cmdObj->root();
mmb::Element field = root.findFirstChildNamed(fieldName);
// If the cmdObj is too large, it will be a "too big" message given by CachedBSONObj.get()
if (!field.ok()) {
return;
}
// Redact the log if there are more than one documents or operations.
if (field.countChildren() > 1) {
field.setValueInt(field.countChildren()).transitional_ignore();
}
}
bool shouldSkipOutput(OperationContext* opCtx) {
const WriteConcernOptions& writeConcern = opCtx->getWriteConcern();
return writeConcern.wMode.empty() && writeConcern.wNumNodes == 0 &&
(writeConcern.syncMode == WriteConcernOptions::SyncMode::NONE ||
writeConcern.syncMode == WriteConcernOptions::SyncMode::UNSET);
}
/**
* Returns true if 'ns' is a time-series collection. That is, this namespace is backed by a
* time-series buckets collection.
*/
bool isTimeseries(OperationContext* opCtx, const NamespaceString& ns) {
// If the buckets collection exists now, the time-series insert path will check for the
// existence of the buckets collection later on with a lock.
// If this check is concurrent with the creation of a time-series collection and the buckets
// collection does not yet exist, this check may return false unnecessarily. As a result, an
// insert attempt into the time-series namespace will either succeed or fail, depending on who
// wins the race.
auto bucketsNs = ns.makeTimeseriesBucketsNamespace();
return CollectionCatalog::get(opCtx)
->lookupCollectionByNamespaceForRead(opCtx, bucketsNs)
.get();
}
// Default for control.version in time-series bucket collection.
const int kTimeseriesControlVersion = 1;
/**
* Transforms a single time-series insert to an update request on an existing bucket.
*/
write_ops::UpdateOpEntry makeTimeseriesUpdateOpEntry(
std::shared_ptr<BucketCatalog::WriteBatch> batch, const BSONObj& metadata) {
BSONObjBuilder updateBuilder;
{
if (!batch->min().isEmpty() || !batch->max().isEmpty()) {
BSONObjBuilder controlBuilder(updateBuilder.subobjStart(
str::stream() << doc_diff::kSubDiffSectionFieldPrefix << "control"));
if (!batch->min().isEmpty()) {
controlBuilder.append(
str::stream() << doc_diff::kSubDiffSectionFieldPrefix << "min", batch->min());
}
if (!batch->max().isEmpty()) {
controlBuilder.append(
str::stream() << doc_diff::kSubDiffSectionFieldPrefix << "max", batch->max());
}
}
}
{
// doc_diff::kSubDiffSectionFieldPrefix + <field name> => {<index_0>: ..., <index_1>: ...}
StringDataMap<BSONObjBuilder> dataFieldBuilders;
auto metadataElem = metadata.firstElement();
DecimalCounter<uint32_t> count(batch->numPreviouslyCommittedMeasurements());
for (const auto& doc : batch->measurements()) {
for (const auto& elem : doc) {
auto key = elem.fieldNameStringData();
if (metadataElem && key == metadataElem.fieldNameStringData()) {
continue;
}
auto& builder = dataFieldBuilders[key];
builder.appendAs(elem, count);
}
++count;
}
// doc_diff::kSubDiffSectionFieldPrefix + <field name>
BSONObjBuilder dataBuilder(updateBuilder.subobjStart("sdata"));
BSONObjBuilder newDataFieldsBuilder;
for (auto& pair : dataFieldBuilders) {
// Existing 'data' fields with measurements require different treatment from fields
// not observed before (missing from control.min and control.max).
if (batch->newFieldNamesToBeInserted().count(pair.first)) {
newDataFieldsBuilder.append(pair.first, pair.second.obj());
}
}
auto newDataFields = newDataFieldsBuilder.obj();
if (!newDataFields.isEmpty()) {
dataBuilder.append(doc_diff::kInsertSectionFieldName, newDataFields);
}
for (auto& pair : dataFieldBuilders) {
// Existing 'data' fields with measurements require different treatment from fields
// not observed before (missing from control.min and control.max).
if (!batch->newFieldNamesToBeInserted().count(pair.first)) {
dataBuilder.append(doc_diff::kSubDiffSectionFieldPrefix + pair.first.toString(),
BSON(doc_diff::kInsertSectionFieldName << pair.second.obj()));
}
}
}
write_ops::UpdateModification u(updateBuilder.obj(), write_ops::UpdateModification::DiffTag{});
write_ops::UpdateOpEntry update(BSON("_id" << batch->bucket()->id()), std::move(u));
invariant(!update.getMulti(), batch->bucket()->id().toString());
invariant(!update.getUpsert(), batch->bucket()->id().toString());
return update;
}
/**
* Returns the single-element array to use as the vector of documents for inserting a new bucket.
*/
BSONArray makeTimeseriesInsertDocument(std::shared_ptr<BucketCatalog::WriteBatch> batch,
const BSONObj& metadata) {
auto metadataElem = metadata.firstElement();
StringDataMap<BSONObjBuilder> dataBuilders;
DecimalCounter<uint32_t> count;
for (const auto& doc : batch->measurements()) {
for (const auto& elem : doc) {
auto key = elem.fieldNameStringData();
if (metadataElem && key == metadataElem.fieldNameStringData()) {
continue;
}
dataBuilders[key].appendAs(elem, count);
}
++count;
}
BSONArrayBuilder builder;
{
BSONObjBuilder bucketBuilder(builder.subobjStart());
bucketBuilder.append("_id", batch->bucket()->id());
{
BSONObjBuilder bucketControlBuilder(bucketBuilder.subobjStart("control"));
bucketControlBuilder.append("version", kTimeseriesControlVersion);
bucketControlBuilder.append("min", batch->min());
bucketControlBuilder.append("max", batch->max());
}
if (metadataElem) {
bucketBuilder.appendAs(metadataElem, "meta");
}
{
BSONObjBuilder bucketDataBuilder(bucketBuilder.subobjStart("data"));
for (auto& dataBuilder : dataBuilders) {
bucketDataBuilder.append(dataBuilder.first, dataBuilder.second.obj());
}
}
}
return builder.arr();
}
/**
* Returns true if the time-series write is retryable.
*/
bool isTimeseriesWriteRetryable(OperationContext* opCtx) {
if (!opCtx->getTxnNumber()) {
return false;
}
if (opCtx->inMultiDocumentTransaction()) {
return false;
}
return true;
}
BucketCatalog::CombineWithInsertsFromOtherClients canCombineWithInsertsFromOtherClients(
OperationContext* opCtx) {
return isTimeseriesWriteRetryable(opCtx)
? BucketCatalog::CombineWithInsertsFromOtherClients::kDisallow
: BucketCatalog::CombineWithInsertsFromOtherClients::kAllow;
}
bool checkFailTimeseriesInsertFailPoint(const BSONObj& metadata) {
bool shouldFailInsert = false;
failTimeseriesInsert.executeIf(
[&](const BSONObj&) { shouldFailInsert = true; },
[&](const BSONObj& data) {
BSONElementComparator comp(BSONElementComparator::FieldNamesMode::kIgnore, nullptr);
return comp.compare(data["metadata"], metadata.firstElement()) == 0;
});
return shouldFailInsert;
}
template <typename T>
boost::optional<BSONObj> generateError(OperationContext* opCtx,
const StatusWith<T>& result,
int index,
size_t numErrors) {
auto status = result.getStatus();
if (status.isOK()) {
return boost::none;
}
auto errorMessage = [numErrors, errorSize = size_t(0)](StringData rawMessage) mutable {
// Start truncating error messages once both of these limits are exceeded.
constexpr size_t kErrorSizeTruncationMin = 1024 * 1024;
constexpr size_t kErrorCountTruncationMin = 2;
if (errorSize >= kErrorSizeTruncationMin && numErrors >= kErrorCountTruncationMin) {
return ""_sd;
}
errorSize += rawMessage.size();
return rawMessage;
};
BSONSizeTracker errorsSizeTracker;
BSONObjBuilder error(errorsSizeTracker);
error.append("index", index);
if (auto staleInfo = status.template extraInfo<StaleConfigInfo>()) {
error.append("code", int(ErrorCodes::StaleShardVersion)); // Different from exception!
{
BSONObjBuilder errInfo(error.subobjStart("errInfo"));
staleInfo->serialize(&errInfo);
}
} else if (ErrorCodes::DocumentValidationFailure == status.code() && status.extraInfo()) {
auto docValidationError =
status.template extraInfo<doc_validation_error::DocumentValidationFailureInfo>();
error.append("code", static_cast<int>(ErrorCodes::DocumentValidationFailure));
error.append("errInfo", docValidationError->getDetails());
} else if (ErrorCodes::isTenantMigrationError(status.code())) {
if (ErrorCodes::TenantMigrationConflict == status.code()) {
auto migrationConflictInfo = status.template extraInfo<TenantMigrationConflictInfo>();
hangWriteBeforeWaitingForMigrationDecision.pauseWhileSet(opCtx);
auto mtab = migrationConflictInfo->getTenantMigrationAccessBlocker();
auto migrationStatus =
mtab->waitUntilCommittedOrAborted(opCtx, migrationConflictInfo->getOperationType());
mtab->recordTenantMigrationError(migrationStatus);
error.append("code", static_cast<int>(migrationStatus.code()));
// We want to append an empty errmsg for the errors after the first one, so let the
// code below that appends errmsg do that.
if (status.reason() != "") {
error.append("errmsg", errorMessage(migrationStatus.reason()));
}
} else {
error.append("code", int(status.code()));
}
} else {
error.append("code", int(status.code()));
if (auto const extraInfo = status.extraInfo()) {
extraInfo->serialize(&error);
}
}
// Skip appending errmsg if it has already been appended like in the case of
// TenantMigrationConflict.
if (!error.hasField("errmsg")) {
error.append("errmsg", errorMessage(status.reason()));
}
return error.obj();
}
/**
* Contains hooks that are used by 'populateReply' method.
*/
struct PopulateReplyHooks {
// Called for each 'SingleWriteResult' processed by 'populateReply' method.
std::function<void(const SingleWriteResult&, int)> singleWriteResultHandler;
// Called after all 'SingleWriteResult' processing is completed by 'populateReply' method.
// This is called as the last method.
std::function<void()> postProcessHandler;
};
/**
* Method to populate a write command reply message. It takes 'result' parameter as an input
* source and populate the fields of 'cmdReply'.
*/
template <typename CommandReplyType>
void populateReply(OperationContext* opCtx,
bool continueOnError,
size_t opsInBatch,
write_ops_exec::WriteResult result,
CommandReplyType* cmdReply,
boost::optional<PopulateReplyHooks> hooks = boost::none) {
invariant(cmdReply);
if (shouldSkipOutput(opCtx))
return;
if (continueOnError) {
invariant(!result.results.empty());
const auto& lastResult = result.results.back();
if (lastResult == ErrorCodes::StaleDbVersion ||
ErrorCodes::isStaleShardVersionError(lastResult.getStatus()) ||
ErrorCodes::isTenantMigrationError(lastResult.getStatus())) {
// For ordered:false commands we need to duplicate these error results for all ops
// after we stopped. See handleError() in write_ops_exec.cpp for more info.
//
// Omit the reason from the duplicate unordered responses so it doesn't consume BSON
// object space
result.results.resize(opsInBatch, lastResult.getStatus().withReason(""));
}
}
long long nVal = 0;
std::vector<BSONObj> errors;
for (size_t i = 0; i < result.results.size(); i++) {
if (auto error = generateError(opCtx, result.results[i], i, errors.size())) {
errors.push_back(*error);
continue;
}
const auto& opResult = result.results[i].getValue();
nVal += opResult.getN(); // Always there.
// Handle custom processing of each result.
if (hooks && hooks->singleWriteResultHandler)
hooks->singleWriteResultHandler(opResult, i);
}
auto& replyBase = cmdReply->getWriteReplyBase();
replyBase.setN(nVal);
if (!errors.empty()) {
replyBase.setWriteErrors(errors);
}
// writeConcernError field is handled by command processor.
{
// Undocumented repl fields that mongos depends on.
auto* replCoord = repl::ReplicationCoordinator::get(opCtx->getServiceContext());
const auto replMode = replCoord->getReplicationMode();
if (replMode != repl::ReplicationCoordinator::modeNone) {
replyBase.setOpTime(repl::ReplClientInfo::forClient(opCtx->getClient()).getLastOp());
if (replMode == repl::ReplicationCoordinator::modeReplSet) {
replyBase.setElectionId(replCoord->getElectionId());
}
}
}
// Call the called-defined post processing handler.
if (hooks && hooks->postProcessHandler)
hooks->postProcessHandler();
}
void transactionChecks(OperationContext* opCtx, const NamespaceString& ns) {
if (!opCtx->inMultiDocumentTransaction())
return;
uassert(50791,
str::stream() << "Cannot write to system collection " << ns.toString()
<< " within a transaction.",
!ns.isSystem() || ns.isPrivilegeCollection());
auto replCoord = repl::ReplicationCoordinator::get(opCtx);
uassert(50790,
str::stream() << "Cannot write to unreplicated collection " << ns.toString()
<< " within a transaction.",
!replCoord->isOplogDisabledFor(opCtx, ns));
}
class CmdInsert final : public write_ops::InsertCmdVersion1Gen<CmdInsert> {
public:
AllowedOnSecondary secondaryAllowed(ServiceContext*) const final {
return AllowedOnSecondary::kNever;
}
void snipForLogging(mutablebson::Document* cmdObj) const final {
redactTooLongLog(cmdObj, "documents");
}
std::string help() const final {
return "insert documents";
}
ReadWriteType getReadWriteType() const final {
return Command::ReadWriteType::kWrite;
}
bool collectsResourceConsumptionMetrics() const final {
return true;
}
bool shouldAffectCommandCounter() const final {
return false;
}
class Invocation final : public InvocationBaseGen {
public:
Invocation(OperationContext* opCtx,
const Command* command,
const OpMsgRequest& opMsgRequest)
: InvocationBaseGen(opCtx, command, opMsgRequest) {
InsertOp::validate(request());
}
bool supportsWriteConcern() const final {
return true;
}
NamespaceString ns() const final {
return request().getNamespace();
}
write_ops::InsertReply typedRun(OperationContext* opCtx) final try {
transactionChecks(opCtx, ns());
write_ops::InsertReply insertReply;
if (isTimeseries(opCtx, ns())) {
// Re-throw parsing exceptions to be consistent with CmdInsert::Invocation's
// constructor.
try {
_performTimeseriesWrites(opCtx, &insertReply);
} catch (DBException& ex) {
ex.addContext(str::stream() << "time-series insert failed: " << ns().ns());
throw;
}
return insertReply;
}
auto reply = write_ops_exec::performInserts(opCtx, request());
populateReply(opCtx,
!request().getWriteCommandBase().getOrdered(),
request().getDocuments().size(),
std::move(reply),
&insertReply);
return insertReply;
} catch (const DBException& ex) {
LastError::get(opCtx->getClient()).setLastError(ex.code(), ex.reason());
throw;
}
private:
void doCheckAuthorization(OperationContext* opCtx) const final try {
auth::checkAuthForInsertCommand(AuthorizationSession::get(opCtx->getClient()),
request().getBypassDocumentValidation(),
request());
} catch (const DBException& ex) {
LastError::get(opCtx->getClient()).setLastError(ex.code(), ex.reason());
throw;
}
StatusWith<SingleWriteResult> _getTimeseriesSingleWriteResult(
const write_ops_exec::WriteResult& reply) const {
invariant(reply.results.size() == 1,
str::stream() << "Unexpected number of results (" << reply.results.size()
<< ") for insert on time-series collection " << ns());
return reply.results[0];
}
StatusWith<SingleWriteResult> _performTimeseriesInsert(
OperationContext* opCtx,
std::shared_ptr<BucketCatalog::WriteBatch> batch,
const BSONObj& metadata,
const boost::optional<std::vector<StmtId>> stmtIds) const {
if (checkFailTimeseriesInsertFailPoint(metadata)) {
return {ErrorCodes::FailPointEnabled,
"Failed time-series insert due to failTimeseriesInsert fail point"};
}
auto bucketsNs = ns().makeTimeseriesBucketsNamespace();
BSONObjBuilder builder;
builder.append(write_ops::Insert::kCommandName, bucketsNs.coll());
// The schema validation configured in the bucket collection is intended for direct
// operations by end users and is not applicable here.
builder.append(write_ops::Insert::kBypassDocumentValidationFieldName, true);
if (stmtIds) {
builder.append(write_ops::Insert::kStmtIdsFieldName, *stmtIds);
}
builder.append(write_ops::Insert::kDocumentsFieldName,
makeTimeseriesInsertDocument(batch, metadata));
auto request = OpMsgRequest::fromDBAndBody(bucketsNs.db(), builder.obj());
auto timeseriesInsertBatch =
write_ops::Insert::parse({"CmdInsert::_performTimeseriesInsert"}, request);
return _getTimeseriesSingleWriteResult(write_ops_exec::performInserts(
opCtx, timeseriesInsertBatch, write_ops_exec::InsertType::kTimeseries));
}
StatusWith<SingleWriteResult> _performTimeseriesUpdate(
OperationContext* opCtx,
std::shared_ptr<BucketCatalog::WriteBatch> batch,
const BSONObj& metadata,
const boost::optional<std::vector<StmtId>> stmtIds) const {
if (checkFailTimeseriesInsertFailPoint(metadata)) {
return {ErrorCodes::FailPointEnabled,
"Failed time-series insert due to failTimeseriesInsert fail point"};
}
auto update = makeTimeseriesUpdateOpEntry(batch, metadata);
write_ops::Update timeseriesUpdateBatch(ns().makeTimeseriesBucketsNamespace(),
{update});
write_ops::WriteCommandBase writeCommandBase;
// The schema validation configured in the bucket collection is intended for direct
// operations by end users and is not applicable here.
writeCommandBase.setBypassDocumentValidation(true);
if (stmtIds) {
writeCommandBase.setStmtIds(*stmtIds);
}
timeseriesUpdateBatch.setWriteCommandBase(std::move(writeCommandBase));
return _getTimeseriesSingleWriteResult(write_ops_exec::performUpdates(
opCtx, timeseriesUpdateBatch, write_ops_exec::UpdateType::kTimeseries));
}
void _commitTimeseriesBucket(OperationContext* opCtx,
std::shared_ptr<BucketCatalog::WriteBatch> batch,
size_t start,
size_t index,
const boost::optional<std::vector<StmtId>>& stmtIds,
std::vector<BSONObj>* errors,
boost::optional<repl::OpTime>* opTime,
boost::optional<OID>* electionId,
std::vector<size_t>* updatesToRetry) const {
auto& bucketCatalog = BucketCatalog::get(opCtx);
auto metadata = bucketCatalog.getMetadata(batch->bucket());
bucketCatalog.prepareCommit(batch);
auto result = batch->numPreviouslyCommittedMeasurements() == 0
? _performTimeseriesInsert(opCtx, batch, metadata, stmtIds)
: _performTimeseriesUpdate(opCtx, batch, metadata, stmtIds);
if (auto error = generateError(opCtx, result, start + index, errors->size())) {
errors->push_back(*error);
bucketCatalog.abort(batch);
return;
}
if (batch->numPreviouslyCommittedMeasurements() != 0 &&
result.getValue().getNModified() == 0) {
// No document in the buckets collection was found to update, meaning that it was
// removed.
bucketCatalog.abort(batch);
updatesToRetry->push_back(index);
return;
}
auto* replCoord = repl::ReplicationCoordinator::get(opCtx->getServiceContext());
const auto replMode = replCoord->getReplicationMode();
*opTime = replMode != repl::ReplicationCoordinator::modeNone
? boost::make_optional(
repl::ReplClientInfo::forClient(opCtx->getClient()).getLastOp())
: boost::none;
*electionId = replMode == repl::ReplicationCoordinator::modeReplSet
? boost::make_optional(replCoord->getElectionId())
: boost::none;
bucketCatalog.finish(
batch, BucketCatalog::CommitInfo{std::move(result), *opTime, *electionId});
}
/**
* Writes to the underlying system.buckets collection. Returns the indices, relative to
* 'start', of the batch which were attempted in an update operation, but found no bucket to
* update. These indices can be passed as the 'indices' parameter in a subsequent
* call to this function, in order to to be retried.
*/
std::vector<size_t> _performUnorderedTimeseriesWrites(
OperationContext* opCtx,
size_t start,
size_t numDocs,
std::vector<BSONObj>* errors,
boost::optional<repl::OpTime>* opTime,
boost::optional<OID>* electionId,
bool* containsRetry,
const std::vector<size_t>& indices) const {
auto& bucketCatalog = BucketCatalog::get(opCtx);
auto bucketsNs = ns().makeTimeseriesBucketsNamespace();
// Holding this shared pointer to the collection guarantees that the collator is not
// invalidated.
auto bucketsColl =
CollectionCatalog::get(opCtx)->lookupCollectionByNamespaceForRead(opCtx, bucketsNs);
uassert(ErrorCodes::NamespaceNotFound,
"Could not find time-series buckets collection for write",
bucketsColl);
uassert(ErrorCodes::InvalidOptions,
"Time-series buckets collection is missing time-series options",
bucketsColl->getTimeseriesOptions());
std::vector<std::pair<std::shared_ptr<BucketCatalog::WriteBatch>, size_t>> batches;
stdx::unordered_map<BucketCatalog::Bucket*, std::vector<StmtId>> bucketStmtIds;
auto insert = [&](size_t index) {
invariant(start + index < request().getDocuments().size());
auto stmtId = request().getStmtId().value_or(0) + start + index;
if (isTimeseriesWriteRetryable(opCtx) &&
TransactionParticipant::get(opCtx).checkStatementExecutedNoOplogEntryFetch(
stmtId)) {
RetryableWritesStats::get(opCtx)->incrementRetriedStatementsCount();
*containsRetry = true;
return;
}
auto result = bucketCatalog.insert(opCtx,
ns(),
bucketsColl->getDefaultCollator(),
*bucketsColl->getTimeseriesOptions(),
request().getDocuments()[start + index],
canCombineWithInsertsFromOtherClients(opCtx));
if (auto error = generateError(opCtx, result, index, errors->size())) {
errors->push_back(*error);
} else {
const auto& batch = result.getValue();
batches.emplace_back(batch, index);
if (isTimeseriesWriteRetryable(opCtx) && !request().getStmtIds()) {
bucketStmtIds[batch->bucket()].push_back(stmtId);
}
}
};
if (!indices.empty()) {
std::for_each(indices.begin(), indices.end(), insert);
} else {
for (size_t i = 0; i < numDocs; i++) {
insert(i);
}
}
hangTimeseriesInsertBeforeCommit.pauseWhileSet();
std::vector<size_t> updatesToRetry;
for (auto& [batch, index] : batches) {
bool shouldCommit = batch->claimCommitRights();
if (shouldCommit) {
auto stmtIds =
[&, bucket = batch->bucket()]() -> boost::optional<std::vector<StmtId>> {
if (!isTimeseriesWriteRetryable(opCtx)) {
return boost::none;
} else if (auto& stmtIds = request().getStmtIds()) {
return stmtIds;
} else {
return bucketStmtIds[bucket];
}
}();
_commitTimeseriesBucket(opCtx,
batch,
start,
index,
stmtIds,
errors,
opTime,
electionId,
&updatesToRetry);
batch.reset();
}
}
for (const auto& [batch, index] : batches) {
if (!batch) {
continue;
}
auto swCommitInfo = batch->getResult();
if (!swCommitInfo.isOK()) {
invariant(swCommitInfo.getStatus() == ErrorCodes::TimeseriesBucketCleared,
str::stream()
<< "Got unexpected error (" << swCommitInfo.getStatus()
<< ") waiting for time-series bucket to be committed for " << ns()
<< ": " << redact(request().toBSON({})));
updatesToRetry.push_back(index);
continue;
}
const auto& commitInfo = swCommitInfo.getValue();
if (auto error =
generateError(opCtx, commitInfo.result, start + index, errors->size())) {
errors->push_back(*error);
}
if (commitInfo.opTime) {
*opTime = std::max(opTime->value_or(repl::OpTime()), *commitInfo.opTime);
}
if (commitInfo.electionId) {
*electionId = std::max(electionId->value_or(OID()), *commitInfo.electionId);
}
}
return updatesToRetry;
}
void _performTimeseriesWritesSubset(OperationContext* opCtx,
size_t start,
size_t numDocs,
std::vector<BSONObj>* errors,
boost::optional<repl::OpTime>* opTime,
boost::optional<OID>* electionId,
bool* containsRetry) const {
std::vector<size_t> updatesToRetry;
do {
updatesToRetry = _performUnorderedTimeseriesWrites(opCtx,
start,
numDocs,
errors,
opTime,
electionId,
containsRetry,
updatesToRetry);
} while (!updatesToRetry.empty());
}
void _performTimeseriesWrites(OperationContext* opCtx,
write_ops::InsertReply* insertReply) const {
auto& curOp = *CurOp::get(opCtx);
ON_BLOCK_EXIT([&] {
// This is the only part of finishCurOp we need to do for inserts because they reuse
// the top-level curOp. The rest is handled by the top-level entrypoint.
curOp.done();
Top::get(opCtx->getServiceContext())
.record(opCtx,
request().getNamespace().ns(),
LogicalOp::opInsert,
Top::LockType::WriteLocked,
durationCount<Microseconds>(curOp.elapsedTimeExcludingPauses()),
curOp.isCommand(),
curOp.getReadWriteType());
});
std::vector<BSONObj> errors;
boost::optional<repl::OpTime> opTime;
boost::optional<OID> electionId;
bool containsRetry = false;
auto& baseReply = insertReply->getWriteReplyBase();
if (request().getOrdered()) {
baseReply.setN(request().getDocuments().size());
for (size_t i = 0; i < request().getDocuments().size(); ++i) {
_performTimeseriesWritesSubset(
opCtx, i, 1, &errors, &opTime, &electionId, &containsRetry);
if (!errors.empty()) {
baseReply.setN(i);
break;
}
}
} else {
_performTimeseriesWritesSubset(opCtx,
0,
request().getDocuments().size(),
&errors,
&opTime,
&electionId,
&containsRetry);
baseReply.setN(request().getDocuments().size() - errors.size());
}
if (!errors.empty()) {
baseReply.setWriteErrors(errors);
}
if (opTime) {
baseReply.setOpTime(*opTime);
}
if (electionId) {
baseReply.setElectionId(*electionId);
}
if (containsRetry) {
RetryableWritesStats::get(opCtx)->incrementRetriedCommandsCount();
}
}
};
} cmdInsert;
class CmdUpdate final : public write_ops::UpdateCmdVersion1Gen<CmdUpdate> {
public:
AllowedOnSecondary secondaryAllowed(ServiceContext*) const final {
return AllowedOnSecondary::kNever;
}
void snipForLogging(mutablebson::Document* cmdObj) const final {
redactTooLongLog(cmdObj, "updates");
}
std::string help() const final {
return "update documents";
}
ReadWriteType getReadWriteType() const final {
return Command::ReadWriteType::kWrite;
}
bool collectsResourceConsumptionMetrics() const final {
return true;
}
bool shouldAffectCommandCounter() const final {
return false;
}
class Invocation final : public InvocationBaseGen {
public:
Invocation(OperationContext* opCtx,
const Command* command,
const OpMsgRequest& opMsgRequest)
: InvocationBaseGen(opCtx, command, opMsgRequest), _commandObj(opMsgRequest.body) {
UpdateOp::validate(request());
invariant(_commandObj.isOwned());
// Extend the lifetime of `updates` to allow asynchronous mirroring.
if (auto seq = opMsgRequest.getSequence("updates"_sd); seq && !seq->objs.empty()) {
// Current design ignores contents of `updates` array except for the first entry.
// Assuming identical collation for all elements in `updates`, future design could
// use the disjunction primitive (i.e, `$or`) to compile all queries into a single
// filter. Such a design also requires a sound way of combining hints.
invariant(seq->objs.front().isOwned());
_updateOpObj = seq->objs.front();
}
}
bool supportsWriteConcern() const final {
return true;
}
NamespaceString ns() const final {
return request().getNamespace();
}
bool getBypass() const {
return request().getBypassDocumentValidation();
}
bool supportsReadMirroring() const override {
return true;
}
void appendMirrorableRequest(BSONObjBuilder* bob) const override {
auto extractQueryDetails = [](const BSONObj& update, BSONObjBuilder* bob) -> void {
// "filter", "hint", and "collation" fields are optional.
if (update.isEmpty())
return;
// The constructor verifies the following.
invariant(update.isOwned());
if (update.hasField("q"))
bob->append("filter", update["q"].Obj());
if (update.hasField("hint") && !update["hint"].Obj().isEmpty())
bob->append("hint", update["hint"].Obj());
if (update.hasField("collation") && !update["collation"].Obj().isEmpty())
bob->append("collation", update["collation"].Obj());
};
invariant(!_commandObj.isEmpty());
bob->append("find", _commandObj["update"].String());
extractQueryDetails(_updateOpObj, bob);
bob->append("batchSize", 1);
bob->append("singleBatch", true);
}
write_ops::UpdateReply typedRun(OperationContext* opCtx) final try {
transactionChecks(opCtx, ns());
write_ops::UpdateReply updateReply;
long long nModified = 0;
// Tracks the upserted information. The memory of this variable gets moved in the
// 'postProcessHandler' and should not be accessed afterwards.
std::vector<write_ops::Upserted> upsertedInfoVec;
auto reply = write_ops_exec::performUpdates(opCtx, request());
// Handler to process each 'SingleWriteResult'.
auto singleWriteHandler = [&](const SingleWriteResult& opResult, int index) {
nModified += opResult.getNModified();
BSONSizeTracker upsertInfoSizeTracker;
if (auto idElement = opResult.getUpsertedId().firstElement())
upsertedInfoVec.emplace_back(write_ops::Upserted(index, idElement));
};
// Handler to do the post-processing.
auto postProcessHandler = [&]() {
updateReply.setNModified(nModified);
if (!upsertedInfoVec.empty())
updateReply.setUpserted(std::move(upsertedInfoVec));
};
populateReply(opCtx,
!request().getWriteCommandBase().getOrdered(),
request().getUpdates().size(),
std::move(reply),
&updateReply,
PopulateReplyHooks{singleWriteHandler, postProcessHandler});
// Collect metrics.
for (auto&& update : request().getUpdates()) {
// If this was a pipeline style update, record that pipeline-style was used and
// which stages were being used.
auto& updateMod = update.getU();
if (updateMod.type() == write_ops::UpdateModification::Type::kPipeline) {
AggregateCommand aggCmd(request().getNamespace(),
updateMod.getUpdatePipeline());
LiteParsedPipeline pipeline(aggCmd);
pipeline.tickGlobalStageCounters();
CmdUpdate::updateMetrics.incrementExecutedWithAggregationPipeline();
}
// If this command had arrayFilters option, record that it was used.
if (update.getArrayFilters()) {
CmdUpdate::updateMetrics.incrementExecutedWithArrayFilters();
}
}
return updateReply;
} catch (const DBException& ex) {
LastError::get(opCtx->getClient()).setLastError(ex.code(), ex.reason());
throw;
}
private:
void doCheckAuthorization(OperationContext* opCtx) const final try {
auth::checkAuthForUpdateCommand(AuthorizationSession::get(opCtx->getClient()),
request().getBypassDocumentValidation(),
request());
} catch (const DBException& ex) {
LastError::get(opCtx->getClient()).setLastError(ex.code(), ex.reason());
throw;
}
void explain(OperationContext* opCtx,
ExplainOptions::Verbosity verbosity,
rpc::ReplyBuilderInterface* result) override {
uassert(ErrorCodes::InvalidLength,
"explained write batches must be of size 1",
request().getUpdates().size() == 1);
UpdateRequest updateRequest(request().getUpdates()[0]);
updateRequest.setNamespaceString(request().getNamespace());
updateRequest.setLegacyRuntimeConstants(request().getLegacyRuntimeConstants().value_or(
Variables::generateRuntimeConstants(opCtx)));
updateRequest.setLetParameters(request().getLet());
updateRequest.setYieldPolicy(PlanYieldPolicy::YieldPolicy::YIELD_AUTO);
updateRequest.setExplain(verbosity);
const ExtensionsCallbackReal extensionsCallback(opCtx,
&updateRequest.getNamespaceString());
ParsedUpdate parsedUpdate(opCtx, &updateRequest, extensionsCallback);
uassertStatusOK(parsedUpdate.parseRequest());
// Explains of write commands are read-only, but we take write locks so that timing
// info is more accurate.
AutoGetCollection collection(opCtx, request().getNamespace(), MODE_IX);
auto exec = uassertStatusOK(getExecutorUpdate(&CurOp::get(opCtx)->debug(),
&collection.getCollection(),
&parsedUpdate,
verbosity));
auto bodyBuilder = result->getBodyBuilder();
Explain::explainStages(exec.get(),
collection.getCollection(),
verbosity,
BSONObj(),
_commandObj,
&bodyBuilder);
}
BSONObj _commandObj;
// Holds a shared pointer to the first entry in `updates` array.
BSONObj _updateOpObj;
};
// Update related command execution metrics.
static UpdateMetrics updateMetrics;
} cmdUpdate;
UpdateMetrics CmdUpdate::updateMetrics{"update"};
class CmdDelete final : public write_ops::DeleteCmdVersion1Gen<CmdDelete> {
public:
AllowedOnSecondary secondaryAllowed(ServiceContext*) const final {
return AllowedOnSecondary::kNever;
}
void snipForLogging(mutablebson::Document* cmdObj) const final {
redactTooLongLog(cmdObj, "deletes");
}
std::string help() const final {
return "delete documents";
}
ReadWriteType getReadWriteType() const final {
return Command::ReadWriteType::kWrite;
}
bool collectsResourceConsumptionMetrics() const final {
return true;
}
bool shouldAffectCommandCounter() const final {
return false;
}
class Invocation final : public InvocationBaseGen {
public:
Invocation(OperationContext* opCtx,
const Command* command,
const OpMsgRequest& opMsgRequest)
: InvocationBaseGen(opCtx, command, opMsgRequest), _commandObj(opMsgRequest.body) {
DeleteOp::validate(request());
}
bool supportsWriteConcern() const final {
return true;
}
NamespaceString ns() const final {
return request().getNamespace();
}
write_ops::DeleteReply typedRun(OperationContext* opCtx) final try {
transactionChecks(opCtx, ns());
write_ops::DeleteReply deleteReply;
auto reply = write_ops_exec::performDeletes(opCtx, request());
populateReply(opCtx,
!request().getWriteCommandBase().getOrdered(),
request().getDeletes().size(),
std::move(reply),
&deleteReply);
return deleteReply;
} catch (const DBException& ex) {
LastError::get(opCtx->getClient()).setLastError(ex.code(), ex.reason());
throw;
}
private:
void doCheckAuthorization(OperationContext* opCtx) const final try {
auth::checkAuthForDeleteCommand(AuthorizationSession::get(opCtx->getClient()),
request().getBypassDocumentValidation(),
request());
} catch (const DBException& ex) {
LastError::get(opCtx->getClient()).setLastError(ex.code(), ex.reason());
throw;
}
void explain(OperationContext* opCtx,
ExplainOptions::Verbosity verbosity,
rpc::ReplyBuilderInterface* result) override {
uassert(ErrorCodes::InvalidLength,
"explained write batches must be of size 1",
request().getDeletes().size() == 1);
auto deleteRequest = DeleteRequest{};
deleteRequest.setNsString(request().getNamespace());
deleteRequest.setLegacyRuntimeConstants(request().getLegacyRuntimeConstants().value_or(
Variables::generateRuntimeConstants(opCtx)));
deleteRequest.setLet(request().getLet());
deleteRequest.setQuery(request().getDeletes()[0].getQ());
deleteRequest.setCollation(write_ops::collationOf(request().getDeletes()[0]));
deleteRequest.setMulti(request().getDeletes()[0].getMulti());
deleteRequest.setYieldPolicy(PlanYieldPolicy::YieldPolicy::YIELD_AUTO);
deleteRequest.setHint(request().getDeletes()[0].getHint());
deleteRequest.setIsExplain(true);
ParsedDelete parsedDelete(opCtx, &deleteRequest);
uassertStatusOK(parsedDelete.parseRequest());
// Explains of write commands are read-only, but we take write locks so that timing
// info is more accurate.
AutoGetCollection collection(opCtx, request().getNamespace(), MODE_IX);
// Explain the plan tree.
auto exec = uassertStatusOK(getExecutorDelete(&CurOp::get(opCtx)->debug(),
&collection.getCollection(),
&parsedDelete,
verbosity));
auto bodyBuilder = result->getBodyBuilder();
Explain::explainStages(exec.get(),
collection.getCollection(),
verbosity,
BSONObj(),
_commandObj,
&bodyBuilder);
}
const BSONObj& _commandObj;
};
} cmdDelete;
} // namespace
} // namespace mongo
|