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
|
/**
* Copyright (C) 2020-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::kSharding
#include "mongo/platform/basic.h"
#include "mongo/db/audit.h"
#include "mongo/db/auth/authorization_session.h"
#include "mongo/db/catalog_raii.h"
#include "mongo/db/commands/feature_compatibility_version.h"
#include "mongo/db/persistent_task_store.h"
#include "mongo/db/query/collation/collator_factory_interface.h"
#include "mongo/db/s/collection_sharding_runtime.h"
#include "mongo/db/s/create_collection_coordinator.h"
#include "mongo/db/s/recoverable_critical_section_service.h"
#include "mongo/db/s/shard_key_util.h"
#include "mongo/db/s/sharding_ddl_util.h"
#include "mongo/db/s/sharding_logging.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/db/timeseries/timeseries_lookup.h"
#include "mongo/logv2/log.h"
#include "mongo/rpc/get_status_from_command_result.h"
#include "mongo/s/cluster_commands_helpers.h"
#include "mongo/s/cluster_write.h"
#include "mongo/s/grid.h"
#include "mongo/s/request_types/shard_collection_gen.h"
namespace mongo {
namespace {
struct OptionsAndIndexes {
BSONObj options;
std::vector<BSONObj> indexSpecs;
BSONObj idIndexSpec;
};
OptionsAndIndexes getCollectionOptionsAndIndexes(OperationContext* opCtx,
const NamespaceStringOrUUID& nssOrUUID) {
DBDirectClient localClient(opCtx);
BSONObj idIndex;
BSONObjBuilder optionsBob;
auto all =
localClient.getCollectionInfos(nssOrUUID.dbname(), BSON("info.uuid" << *nssOrUUID.uuid()));
// There must be a collection at this time.
invariant(!all.empty());
auto& entry = all.front();
if (entry["options"].isABSONObj()) {
optionsBob.appendElements(entry["options"].Obj());
}
optionsBob.append(entry["info"]["uuid"]);
idIndex = entry["idIndex"].Obj().getOwned();
auto indexSpecsList = localClient.getIndexSpecs(nssOrUUID, false, 0);
return {optionsBob.obj(),
std::vector<BSONObj>(std::begin(indexSpecsList), std::end(indexSpecsList)),
idIndex};
}
/**
* Compares the proposed shard key with the shard key of the collection's existing zones
* to ensure they are a legal combination.
*/
void validateShardKeyAgainstExistingZones(OperationContext* opCtx,
const BSONObj& proposedKey,
const ShardKeyPattern& shardKeyPattern,
const std::vector<TagsType>& tags) {
for (const auto& tag : tags) {
BSONObjIterator tagMinFields(tag.getMinKey());
BSONObjIterator tagMaxFields(tag.getMaxKey());
BSONObjIterator proposedFields(proposedKey);
while (tagMinFields.more() && proposedFields.more()) {
BSONElement tagMinKeyElement = tagMinFields.next();
BSONElement tagMaxKeyElement = tagMaxFields.next();
uassert(ErrorCodes::InvalidOptions,
str::stream() << "the min and max of the existing zone " << tag.getMinKey()
<< " -->> " << tag.getMaxKey() << " have non-matching keys",
tagMinKeyElement.fieldNameStringData() ==
tagMaxKeyElement.fieldNameStringData());
BSONElement proposedKeyElement = proposedFields.next();
bool match = ((tagMinKeyElement.fieldNameStringData() ==
proposedKeyElement.fieldNameStringData()) &&
((tagMinFields.more() && proposedFields.more()) ||
(!tagMinFields.more() && !proposedFields.more())));
uassert(ErrorCodes::InvalidOptions,
str::stream() << "the proposed shard key " << proposedKey.toString()
<< " does not match with the shard key of the existing zone "
<< tag.getMinKey() << " -->> " << tag.getMaxKey(),
match);
// If the field is hashed, make sure that the min and max values are of supported type.
uassert(
ErrorCodes::InvalidOptions,
str::stream() << "cannot do hash sharding with the proposed key "
<< proposedKey.toString() << " because there exists a zone "
<< tag.getMinKey() << " -->> " << tag.getMaxKey()
<< " whose boundaries are not of type NumberLong, MinKey or MaxKey",
!ShardKeyPattern::isHashedPatternEl(proposedKeyElement) ||
(ShardKeyPattern::isValidHashedValue(tagMinKeyElement) &&
ShardKeyPattern::isValidHashedValue(tagMaxKeyElement)));
}
}
}
std::vector<TagsType> getTagsAndValidate(OperationContext* opCtx,
const NamespaceString& nss,
const BSONObj& proposedKey,
const ShardKeyPattern& shardKeyPattern) {
// Read zone info
const auto catalogClient = Grid::get(opCtx)->catalogClient();
auto tags = uassertStatusOK(catalogClient->getTagsForCollection(opCtx, nss));
if (!tags.empty()) {
validateShardKeyAgainstExistingZones(opCtx, proposedKey, shardKeyPattern, tags);
}
return tags;
}
bool checkIfCollectionIsEmpty(OperationContext* opCtx, const NamespaceString& nss) {
// Use find with predicate instead of count in order to ensure that the count
// command doesn't just consult the cached metadata, which may not always be
// correct
DBDirectClient localClient(opCtx);
return localClient.findOne(nss.ns(), Query()).isEmpty();
}
int getNumShards(OperationContext* opCtx) {
const auto shardRegistry = Grid::get(opCtx)->shardRegistry();
shardRegistry->reload(opCtx);
return shardRegistry->getNumShards(opCtx);
}
BSONObj getCollation(OperationContext* opCtx,
const NamespaceString& nss,
const boost::optional<BSONObj>& collation) {
// Ensure the collation is valid. Currently we only allow the simple collation.
std::unique_ptr<CollatorInterface> requestedCollator = nullptr;
if (collation) {
requestedCollator =
uassertStatusOK(CollatorFactoryInterface::get(opCtx->getServiceContext())
->makeFromBSON(collation.value()));
uassert(ErrorCodes::BadValue,
str::stream() << "The collation for shardCollection must be {locale: 'simple'}, "
<< "but found: " << collation.value(),
!requestedCollator);
}
AutoGetCollection autoColl(opCtx, nss, MODE_IS, AutoGetCollectionViewMode::kViewsForbidden);
const auto actualCollator = [&]() -> const CollatorInterface* {
const auto& coll = autoColl.getCollection();
if (coll) {
uassert(
ErrorCodes::InvalidOptions, "can't shard a capped collection", !coll->isCapped());
return coll->getDefaultCollator();
}
return nullptr;
}();
if (!requestedCollator && !actualCollator)
return BSONObj();
auto actualCollatorBSON = actualCollator->getSpec().toBSON();
if (!collation) {
auto actualCollatorFilter =
uassertStatusOK(CollatorFactoryInterface::get(opCtx->getServiceContext())
->makeFromBSON(actualCollatorBSON));
uassert(ErrorCodes::BadValue,
str::stream() << "If no collation was specified, the collection collation must be "
"{locale: 'simple'}, "
<< "but found: " << actualCollatorBSON,
!actualCollatorFilter);
}
return actualCollatorBSON;
}
void removeChunks(OperationContext* opCtx, const UUID& uuid) {
BatchWriteExecStats stats;
BatchedCommandResponse response;
BatchedCommandRequest deleteRequest([&]() {
write_ops::DeleteCommandRequest deleteOp(ChunkType::ConfigNS);
deleteOp.setWriteCommandRequestBase([] {
write_ops::WriteCommandRequestBase writeCommandBase;
writeCommandBase.setOrdered(false);
return writeCommandBase;
}());
deleteOp.setDeletes(
std::vector{write_ops::DeleteOpEntry(BSON(ChunkType::collectionUUID << uuid), false)});
return deleteOp;
}());
deleteRequest.setWriteConcern(ShardingCatalogClient::kMajorityWriteConcern.toBSON());
cluster::write(opCtx, deleteRequest, &stats, &response, boost::none);
uassertStatusOK(response.toStatus());
}
void upsertChunks(OperationContext* opCtx, std::vector<ChunkType>& chunks) {
BatchWriteExecStats stats;
BatchedCommandResponse response;
BatchedCommandRequest updateRequest([&]() {
write_ops::UpdateCommandRequest updateOp(ChunkType::ConfigNS);
std::vector<write_ops::UpdateOpEntry> entries;
entries.reserve(chunks.size());
for (const auto& chunk : chunks) {
write_ops::UpdateOpEntry entry(
BSON(ChunkType::collectionUUID << chunk.getCollectionUUID() << ChunkType::shard
<< chunk.getShard() << ChunkType::min
<< chunk.getMin()),
write_ops::UpdateModification::parseFromClassicUpdate(chunk.toConfigBSON()));
entry.setUpsert(true);
entry.setMulti(false);
entries.push_back(entry);
}
updateOp.setUpdates(entries);
return updateOp;
}());
updateRequest.setWriteConcern(ShardingCatalogClient::kMajorityWriteConcern.toBSON());
cluster::write(opCtx, updateRequest, &stats, &response, boost::none);
uassertStatusOK(response.toStatus());
}
void updateCatalogEntry(OperationContext* opCtx, const NamespaceString& nss, CollectionType& coll) {
BatchWriteExecStats stats;
BatchedCommandResponse response;
BatchedCommandRequest updateRequest([&]() {
write_ops::UpdateCommandRequest updateOp(CollectionType::ConfigNS);
updateOp.setUpdates({[&] {
write_ops::UpdateOpEntry entry;
entry.setQ(BSON(CollectionType::kNssFieldName << nss.ns()));
entry.setU(write_ops::UpdateModification::parseFromClassicUpdate(coll.toBSON()));
entry.setUpsert(true);
entry.setMulti(false);
return entry;
}()});
return updateOp;
}());
updateRequest.setWriteConcern(ShardingCatalogClient::kMajorityWriteConcern.toBSON());
try {
cluster::write(opCtx, updateRequest, &stats, &response, boost::none);
uassertStatusOK(response.toStatus());
} catch (const DBException&) {
// If an error happens when contacting the config server, we don't know if the update
// succeded or not, which might cause the local shard version to differ from the config
// server, so we clear the metadata to allow another operation to refresh it.
UninterruptibleLockGuard noInterrupt(opCtx->lockState());
AutoGetCollection autoColl(opCtx, nss, MODE_IX);
CollectionShardingRuntime::get(opCtx, nss)->clearFilteringMetadata(opCtx);
throw;
}
}
void broadcastDropCollection(OperationContext* opCtx,
const NamespaceString& nss,
const std::shared_ptr<executor::TaskExecutor>& executor) {
const auto primaryShardId = ShardingState::get(opCtx)->shardId();
const ShardsvrDropCollectionParticipant dropCollectionParticipant(nss);
auto participants = Grid::get(opCtx)->shardRegistry()->getAllShardIds(opCtx);
// Remove prymary shard from participants
participants.erase(std::remove(participants.begin(), participants.end(), primaryShardId),
participants.end());
sharding_ddl_util::sendAuthenticatedCommandToShards(
opCtx,
nss.db(),
CommandHelpers::appendMajorityWriteConcern(dropCollectionParticipant.toBSON({})),
participants,
executor);
}
} // namespace
CreateCollectionCoordinator::CreateCollectionCoordinator(ShardingDDLCoordinatorService* service,
const BSONObj& initialState)
: ShardingDDLCoordinator(service, initialState),
_doc(CreateCollectionCoordinatorDocument::parse(
IDLParserErrorContext("CreateCollectionCoordinatorDocument"), initialState)),
_critSecReason(BSON("command"
<< "createCollection"
<< "ns" << nss().toString() << "request"
<< _doc.getCreateCollectionRequest().toBSON())) {}
boost::optional<BSONObj> CreateCollectionCoordinator::reportForCurrentOp(
MongoProcessInterface::CurrentOpConnectionsMode connMode,
MongoProcessInterface::CurrentOpSessionsMode sessionMode) noexcept {
BSONObjBuilder cmdBob;
if (const auto& optComment = getForwardableOpMetadata().getComment()) {
cmdBob.append(optComment.get().firstElement());
}
cmdBob.append("request", _doc.getCreateCollectionRequest().toBSON());
BSONObjBuilder bob;
bob.append("type", "op");
bob.append("desc", "CreateCollectionCoordinator");
bob.append("op", "command");
bob.append("ns", nss().toString());
bob.append("command", cmdBob.obj());
bob.append("currentPhase", _doc.getPhase());
bob.append("active", true);
return bob.obj();
}
void CreateCollectionCoordinator::checkIfOptionsConflict(const BSONObj& doc) const {
// If we have two shard collections on the same namespace, then the arguments must be the same.
const auto otherDoc = CreateCollectionCoordinatorDocument::parse(
IDLParserErrorContext("RenameCollectionCoordinatorDocument"), doc);
uassert(ErrorCodes::ConflictingOperationInProgress,
"Another create collection with different arguments is already running for the same "
"namespace",
SimpleBSONObjComparator::kInstance.evaluate(
_doc.getCreateCollectionRequest().toBSON() ==
otherDoc.getCreateCollectionRequest().toBSON()));
}
ExecutorFuture<void> CreateCollectionCoordinator::_runImpl(
std::shared_ptr<executor::ScopedTaskExecutor> executor,
const CancellationToken& token) noexcept {
return ExecutorFuture<void>(**executor)
.then([this, anchor = shared_from_this()] {
_shardKeyPattern = ShardKeyPattern(*_doc.getShardKey());
if (_doc.getPhase() < Phase::kCommit) {
auto opCtxHolder = cc().makeOperationContext();
auto* opCtx = opCtxHolder.get();
getForwardableOpMetadata().setOn(opCtx);
_checkCommandArguments(opCtx);
}
})
.then(_executePhase(
Phase::kCommit,
[this, executor = executor, token, anchor = shared_from_this()] {
auto opCtxHolder = cc().makeOperationContext();
auto* opCtx = opCtxHolder.get();
getForwardableOpMetadata().setOn(opCtx);
if (_recoveredFromDisk) {
// If a stedown happened it could've ocurred while waiting for majority when
// writing config.collections. If the refresh happens before this write is
// majority committed, we will only see the data on config.chunks but not on
// config.collections, so we need to serialize the refresh with the collection
// creation.
sharding_ddl_util::linearizeCSRSReads(opCtx);
}
// Log the start of the event only if we're not recovering.
_logStartCreateCollection(opCtx);
// Quick check (without critical section) to see if another create collection
// already succeeded.
if (auto createCollectionResponseOpt =
sharding_ddl_util::checkIfCollectionAlreadySharded(
opCtx,
nss(),
_shardKeyPattern->getKeyPattern().toBSON(),
getCollation(opCtx, nss(), _doc.getCollation()),
_doc.getUnique().value_or(false))) {
_result = createCollectionResponseOpt;
// The collection was already created and commited but there was a
// stepdown after the commit.
RecoverableCriticalSectionService::get(opCtx)
->releaseRecoverableCriticalSection(
opCtx,
nss(),
_critSecReason,
ShardingCatalogClient::kMajorityWriteConcern);
return;
}
// Entering the critical section. From this point on, the writes are blocked. Before
// calling this method, we need the coordinator document to be persisted (and hence
// the kCheck state), otherwise nothing will release the critical section in the
// presence of a stepdown.
RecoverableCriticalSectionService::get(opCtx)
->acquireRecoverableCriticalSectionBlockWrites(
opCtx, nss(), _critSecReason, ShardingCatalogClient::kMajorityWriteConcern);
if (_recoveredFromDisk) {
auto uuid = sharding_ddl_util::getCollectionUUID(opCtx, nss());
// If the collection can be found locally, then we clean up the config.chunks
// collection.
if (uuid) {
LOGV2_DEBUG(5458704,
1,
"Removing partial changes from previous run",
"namespace"_attr = nss());
removeChunks(opCtx, *uuid);
broadcastDropCollection(opCtx, nss(), **executor);
}
}
_collectionEmpty = checkIfCollectionIsEmpty(opCtx, nss());
_createCollectionAndIndexes(opCtx);
_createPolicyAndChunks(opCtx);
audit::logShardCollection(opCtx->getClient(),
nss().ns(),
*_doc.getCreateCollectionRequest().getShardKey(),
*_doc.getCreateCollectionRequest().getUnique());
if (_splitPolicy->isOptimized()) {
// Block reads/writes from here on if we need to create
// the collection on other shards, this way we prevent
// reads/writes that should be redirected to another
// shard.
RecoverableCriticalSectionService::get(opCtx)
->promoteRecoverableCriticalSectionToBlockAlsoReads(
opCtx,
nss(),
_critSecReason,
ShardingCatalogClient::kMajorityWriteConcern);
_createCollectionOnNonPrimaryShards(opCtx);
_commit(opCtx);
}
RecoverableCriticalSectionService::get(opCtx)->releaseRecoverableCriticalSection(
opCtx, nss(), _critSecReason, ShardingCatalogClient::kMajorityWriteConcern);
if (!_splitPolicy->isOptimized()) {
_commit(opCtx);
}
_finalize(opCtx);
}))
.then([this] {
auto opCtxHolder = cc().makeOperationContext();
auto* opCtx = opCtxHolder.get();
getForwardableOpMetadata().setOn(opCtx);
_logEndCreateCollection(opCtx);
})
.onError([this, anchor = shared_from_this()](const Status& status) {
if (!status.isA<ErrorCategory::NotPrimaryError>() &&
!status.isA<ErrorCategory::ShutdownError>()) {
LOGV2_ERROR(5458702,
"Error running create collection",
"namespace"_attr = nss(),
"error"_attr = redact(status));
auto opCtxHolder = cc().makeOperationContext();
auto* opCtx = opCtxHolder.get();
RecoverableCriticalSectionService::get(opCtx)->releaseRecoverableCriticalSection(
opCtx, nss(), _critSecReason, ShardingCatalogClient::kMajorityWriteConcern);
}
return status;
});
}
void CreateCollectionCoordinator::_checkCommandArguments(OperationContext* opCtx) {
LOGV2_DEBUG(5277902, 2, "Create collection _checkCommandArguments", "namespace"_attr = nss());
const auto dbEnabledForSharding = [&, this] {
// The modification of the 'sharded' flag for the db does not imply a database version
// change so we can't use the DatabaseShardingState to look it up. Instead we will do a
// first attempt through the catalog cache and if it is unset we will attempt another time
// after a forced catalog cache refresh.
auto catalogCache = Grid::get(opCtx)->catalogCache();
auto dbInfo = uassertStatusOK(catalogCache->getDatabase(opCtx, nss().db()));
if (dbInfo.shardingEnabled()) {
return true;
}
dbInfo = uassertStatusOK(catalogCache->getDatabaseWithRefresh(opCtx, nss().db()));
return dbInfo.shardingEnabled();
}();
uassert(ErrorCodes::IllegalOperation,
str::stream() << "sharding not enabled for db " << nss().db(),
dbEnabledForSharding);
if (nss().db() == NamespaceString::kConfigDb) {
// Only allowlisted collections in config may be sharded (unless we are in test mode)
uassert(ErrorCodes::IllegalOperation,
"only special collections in the config db may be sharded",
nss() == NamespaceString::kLogicalSessionsNamespace);
}
// Ensure that hashed and unique are not both set.
uassert(ErrorCodes::InvalidOptions,
"Hashed shard keys cannot be declared unique. It's possible to ensure uniqueness on "
"the hashed field by declaring an additional (non-hashed) unique index on the field.",
!_shardKeyPattern->isHashedPattern() || !_doc.getUnique().value_or(false));
// Ensure that a time-series collection cannot be sharded
uassert(ErrorCodes::IllegalOperation,
str::stream() << "can't shard time-series collection " << nss(),
!timeseries::getTimeseriesOptions(opCtx, nss()));
// Ensure the namespace is valid.
uassert(ErrorCodes::IllegalOperation,
"can't shard system namespaces",
!nss().isSystem() || nss() == NamespaceString::kLogicalSessionsNamespace ||
nss().isTemporaryReshardingCollection());
if (_doc.getNumInitialChunks()) {
// Ensure numInitialChunks is within valid bounds.
// Cannot have more than 8192 initial chunks per shard. Setting a maximum of 1,000,000
// chunks in total to limit the amount of memory this command consumes so there is less
// danger of an OOM error.
const int maxNumInitialChunksForShards =
Grid::get(opCtx)->shardRegistry()->getNumShardsNoReload() * 8192;
const int maxNumInitialChunksTotal = 1000 * 1000; // Arbitrary limit to memory consumption
int numChunks = _doc.getNumInitialChunks().value();
uassert(ErrorCodes::InvalidOptions,
str::stream() << "numInitialChunks cannot be more than either: "
<< maxNumInitialChunksForShards << ", 8192 * number of shards; or "
<< maxNumInitialChunksTotal,
numChunks >= 0 && numChunks <= maxNumInitialChunksForShards &&
numChunks <= maxNumInitialChunksTotal);
}
if (nss().db() == NamespaceString::kConfigDb) {
auto configShard = Grid::get(opCtx)->shardRegistry()->getConfigShard();
auto findReponse = uassertStatusOK(
configShard->exhaustiveFindOnConfig(opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
repl::ReadConcernLevel::kMajorityReadConcern,
nss(),
BSONObj(),
BSONObj(),
1));
auto numDocs = findReponse.docs.size();
// If this is a collection on the config db, it must be empty to be sharded.
uassert(ErrorCodes::IllegalOperation,
"collections in the config db must be empty to be sharded",
numDocs == 0);
}
}
void CreateCollectionCoordinator::_createCollectionAndIndexes(OperationContext* opCtx) {
LOGV2_DEBUG(
5277903, 2, "Create collection _createCollectionAndIndexes", "namespace"_attr = nss());
_collation = getCollation(opCtx, nss(), _doc.getCollation());
const auto indexCreated = shardkeyutil::validateShardKeyIndexExistsOrCreateIfPossible(
opCtx,
nss(),
*_shardKeyPattern,
_collation,
_doc.getUnique().value_or(false),
shardkeyutil::ValidationBehaviorsShardCollection(opCtx));
auto replClientInfo = repl::ReplClientInfo::forClient(opCtx->getClient());
if (!indexCreated) {
replClientInfo.setLastOpToSystemLastOpTime(opCtx);
}
// Wait until the index is majority written, to prevent having the collection commited to the
// config server, but the index creation rolled backed on stepdowns.
WriteConcernResult ignoreResult;
uassertStatusOK(waitForWriteConcern(opCtx,
replClientInfo.getLastOp(),
ShardingCatalogClient::kMajorityWriteConcern,
&ignoreResult));
_collectionUUID = *sharding_ddl_util::getCollectionUUID(opCtx, nss());
}
void CreateCollectionCoordinator::_createPolicyAndChunks(OperationContext* opCtx) {
LOGV2_DEBUG(5277904, 2, "Create collection _createChunks", "namespace"_attr = nss());
_splitPolicy = InitialSplitPolicy::calculateOptimizationStrategy(
opCtx,
*_shardKeyPattern,
_doc.getNumInitialChunks() ? *_doc.getNumInitialChunks() : 0,
_doc.getPresplitHashedZones() ? *_doc.getPresplitHashedZones() : false,
_doc.getInitialSplitPoints(),
getTagsAndValidate(opCtx, nss(), _shardKeyPattern->toBSON(), *_shardKeyPattern),
getNumShards(opCtx),
*_collectionEmpty);
_initialChunks = _splitPolicy->createFirstChunks(
opCtx,
*_shardKeyPattern,
{nss(),
*_collectionUUID,
ShardingState::get(opCtx)->shardId(),
ChunkEntryFormat::getForVersionCallerGuaranteesFCVStability(
ServerGlobalParams::FeatureCompatibility::Version::kVersion50)});
// There must be at least one chunk.
invariant(!_initialChunks.chunks.empty());
_numChunks = _initialChunks.chunks.size();
}
void CreateCollectionCoordinator::_createCollectionOnNonPrimaryShards(OperationContext* opCtx) {
LOGV2_DEBUG(5277905,
2,
"Create collection _createCollectionOnNonPrimaryShards",
"namespace"_attr = nss());
std::vector<AsyncRequestsSender::Request> requests;
std::set<ShardId> initializedShards;
auto dbPrimaryShardId = ShardingState::get(opCtx)->shardId();
NamespaceStringOrUUID nssOrUUID{nss().db().toString(), *_collectionUUID};
auto [collOptions, indexes, idIndex] = getCollectionOptionsAndIndexes(opCtx, nssOrUUID);
for (const auto& chunk : _initialChunks.chunks) {
const auto& chunkShardId = chunk.getShard();
if (chunkShardId == dbPrimaryShardId ||
initializedShards.find(chunkShardId) != initializedShards.end()) {
continue;
}
ShardsvrCreateCollectionParticipant createCollectionParticipantRequest(nss());
createCollectionParticipantRequest.setCollectionUUID(*_collectionUUID);
createCollectionParticipantRequest.setOptions(collOptions);
createCollectionParticipantRequest.setIdIndex(idIndex);
createCollectionParticipantRequest.setIndexes(indexes);
requests.emplace_back(
chunkShardId,
createCollectionParticipantRequest.toBSON(
BSON("writeConcern" << ShardingCatalogClient::kMajorityWriteConcern.toBSON())));
initializedShards.emplace(chunkShardId);
}
if (!requests.empty()) {
auto responses = gatherResponses(opCtx,
nss().db(),
ReadPreferenceSetting(ReadPreference::PrimaryOnly),
Shard::RetryPolicy::kIdempotent,
requests);
// If any shards fail to create the collection, fail the entire shardCollection command
// (potentially leaving incomplely created sharded collection)
for (const auto& response : responses) {
auto shardResponse = uassertStatusOKWithContext(
std::move(response.swResponse),
str::stream() << "Unable to create collection " << nss().ns() << " on "
<< response.shardId);
auto status = getStatusFromCommandResult(shardResponse.data);
uassertStatusOK(status.withContext(str::stream()
<< "Unable to create collection " << nss().ns()
<< " on " << response.shardId));
auto wcStatus = getWriteConcernStatusFromCommandResult(shardResponse.data);
uassertStatusOK(wcStatus.withContext(str::stream()
<< "Unable to create collection " << nss().ns()
<< " on " << response.shardId));
}
}
}
void CreateCollectionCoordinator::_commit(OperationContext* opCtx) {
LOGV2_DEBUG(5277906, 2, "Create collection _commit", "namespace"_attr = nss());
// Upsert Chunks.
upsertChunks(opCtx, _initialChunks.chunks);
CollectionType coll(nss(),
_initialChunks.collVersion().epoch(),
_initialChunks.collVersion().getTimestamp(),
Date_t::now(),
*_collectionUUID);
coll.setKeyPattern(_shardKeyPattern->getKeyPattern());
if (_collation) {
coll.setDefaultCollation(_collation.value());
}
if (_doc.getUnique()) {
coll.setUnique(*_doc.getUnique());
}
updateCatalogEntry(opCtx, nss(), coll);
}
void CreateCollectionCoordinator::_finalize(OperationContext* opCtx) {
LOGV2_DEBUG(5277907, 2, "Create collection _finalize", "namespace"_attr = nss());
try {
forceShardFilteringMetadataRefresh(opCtx, nss());
} catch (const DBException&) {
// If the refresh fails, then set the shard version to UNKNOWN and let a future operation to
// refresh the metadata.
UninterruptibleLockGuard noInterrupt(opCtx->lockState());
AutoGetCollection autoColl(opCtx, nss(), MODE_IX);
CollectionShardingRuntime::get(opCtx, nss())->clearFilteringMetadata(opCtx);
}
// Is it really necessary to refresh all shards? or can I assume that the shard version will be
// unknown and refreshed eventually?
auto shardRegistry = Grid::get(opCtx)->shardRegistry();
auto dbPrimaryShardId = ShardingState::get(opCtx)->shardId();
std::set<ShardId> shardsRefreshed;
for (const auto& chunk : _initialChunks.chunks) {
const auto& chunkShardId = chunk.getShard();
if (chunkShardId == dbPrimaryShardId ||
shardsRefreshed.find(chunkShardId) != shardsRefreshed.end()) {
continue;
}
auto shard = uassertStatusOK(shardRegistry->getShard(opCtx, chunkShardId));
try {
auto refreshCmdResponse = uassertStatusOK(shard->runCommandWithFixedRetryAttempts(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
"admin",
BSON("_flushRoutingTableCacheUpdates" << nss().ns()),
Seconds{30},
Shard::RetryPolicy::kIdempotent));
uassertStatusOK(refreshCmdResponse.commandStatus);
} catch (const DBException& ex) {
LOGV2_WARNING(5277909,
"Could not refresh shard",
"shardId"_attr = shard->getId(),
"error"_attr = redact(ex.reason()));
}
shardsRefreshed.emplace(chunkShardId);
}
LOGV2(5277901,
"Created initial chunk(s)",
"namespace"_attr = nss(),
"numInitialChunks"_attr = _initialChunks.chunks.size(),
"initialCollectionVersion"_attr = _initialChunks.collVersion());
auto result = CreateCollectionResponse(
_initialChunks.chunks[_initialChunks.chunks.size() - 1].getVersion());
result.setCollectionUUID(_collectionUUID);
_result = std::move(result);
LOGV2(5458701,
"Collection created",
"namespace"_attr = nss(),
"UUID"_attr = _result->getCollectionUUID(),
"version"_attr = _result->getCollectionVersion());
}
void CreateCollectionCoordinator::_logStartCreateCollection(OperationContext* opCtx) {
BSONObjBuilder collectionDetail;
collectionDetail.append("shardKey", *_doc.getCreateCollectionRequest().getShardKey());
collectionDetail.append("collection", nss().ns());
collectionDetail.append("primary", ShardingState::get(opCtx)->shardId().toString());
ShardingLogging::get(opCtx)->logChange(
opCtx, "shardCollection.start", nss().ns(), collectionDetail.obj());
}
void CreateCollectionCoordinator::_logEndCreateCollection(OperationContext* opCtx) {
BSONObjBuilder collectionDetail;
_result->getCollectionUUID()->appendToBuilder(&collectionDetail, "uuid");
collectionDetail.append("version", _result->getCollectionVersion().toString());
if (_collectionEmpty)
collectionDetail.append("empty", *_collectionEmpty);
if (_numChunks)
collectionDetail.appendNumber("numChunks", static_cast<long long>(*_numChunks));
ShardingLogging::get(opCtx)->logChange(
opCtx, "shardCollection.end", nss().ns(), collectionDetail.obj());
}
// Phase change and document handling API.
void CreateCollectionCoordinator::_insertCoordinatorDocument(CoordDoc&& doc) {
auto docBSON = _doc.toBSON();
auto coorMetadata = doc.getShardingDDLCoordinatorMetadata();
coorMetadata.setRecoveredFromDisk(true);
doc.setShardingDDLCoordinatorMetadata(coorMetadata);
auto opCtx = cc().makeOperationContext();
PersistentTaskStore<CoordDoc> store(NamespaceString::kShardingDDLCoordinatorsNamespace);
store.add(opCtx.get(), doc, WriteConcerns::kMajorityWriteConcern);
_doc = std::move(doc);
}
void CreateCollectionCoordinator::_updateCoordinatorDocument(CoordDoc&& newDoc) {
auto opCtx = cc().makeOperationContext();
PersistentTaskStore<CoordDoc> store(NamespaceString::kShardingDDLCoordinatorsNamespace);
store.update(opCtx.get(),
BSON(CoordDoc::kIdFieldName << _doc.getId().toBSON()),
newDoc.toBSON(),
WriteConcerns::kMajorityWriteConcern);
_doc = std::move(newDoc);
}
void CreateCollectionCoordinator::_enterPhase(Phase newPhase) {
CoordDoc newDoc(_doc);
newDoc.setPhase(newPhase);
LOGV2_DEBUG(5565600,
2,
"Create collection coordinator phase transition",
"namespace"_attr = nss(),
"newPhase"_attr = CreateCollectionCoordinatorPhase_serializer(newDoc.getPhase()),
"oldPhase"_attr = CreateCollectionCoordinatorPhase_serializer(_doc.getPhase()));
if (_doc.getPhase() == Phase::kUnset) {
_insertCoordinatorDocument(std::move(newDoc));
return;
}
_updateCoordinatorDocument(std::move(newDoc));
}
} // namespace mongo
|