summaryrefslogtreecommitdiff
path: root/src/mongo/db/repl/storage_interface_impl.cpp
blob: 17d1077c559ee656178d6f8abbdc840cf0390759 (plain)
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
/**
 *    Copyright (C) 2015 MongoDB Inc.
 *
 *    This program is free software: you can redistribute it and/or  modify
 *    it under the terms of the GNU Affero General Public License, version 3,
 *    as published by the Free Software Foundation.
 *
 *    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
 *    GNU Affero General Public License for more details.
 *
 *    You should have received a copy of the GNU Affero General Public License
 *    along with this program.  If not, see <http://www.gnu.org/licenses/>.
 *
 *    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 GNU Affero General 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_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kReplication

#include "mongo/platform/basic.h"

#include "mongo/db/repl/storage_interface_impl.h"

#include <algorithm>
#include <boost/optional.hpp>
#include <utility>

#include "mongo/base/status.h"
#include "mongo/base/status_with.h"
#include "mongo/base/string_data.h"
#include "mongo/bson/bsonobj.h"
#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/bson/util/bson_extract.h"
#include "mongo/db/auth/authorization_manager.h"
#include "mongo/db/catalog/collection.h"
#include "mongo/db/catalog/collection_catalog_entry.h"
#include "mongo/db/catalog/database.h"
#include "mongo/db/catalog/document_validation.h"
#include "mongo/db/catalog/index_catalog.h"
#include "mongo/db/catalog/index_create.h"
#include "mongo/db/client.h"
#include "mongo/db/concurrency/d_concurrency.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/curop.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbhelpers.h"
#include "mongo/db/exec/delete.h"
#include "mongo/db/exec/update.h"
#include "mongo/db/jsobj.h"
#include "mongo/db/keypattern.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/ops/delete_request.h"
#include "mongo/db/ops/parsed_update.h"
#include "mongo/db/ops/update_request.h"
#include "mongo/db/query/get_executor.h"
#include "mongo/db/query/internal_plans.h"
#include "mongo/db/repl/collection_bulk_loader_impl.h"
#include "mongo/db/repl/oplog.h"
#include "mongo/db/repl/replication_coordinator.h"
#include "mongo/db/repl/replication_coordinator_global.h"
#include "mongo/db/repl/rollback_gen.h"
#include "mongo/db/repl/task_runner.h"
#include "mongo/db/service_context.h"
#include "mongo/util/assert_util.h"
#include "mongo/util/concurrency/old_thread_pool.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"

namespace mongo {
namespace repl {

const char StorageInterfaceImpl::kDefaultRollbackIdNamespace[] = "local.system.rollback.id";
const char StorageInterfaceImpl::kRollbackIdFieldName[] = "rollbackId";
const char StorageInterfaceImpl::kRollbackIdDocumentId[] = "rollbackId";

namespace {
using UniqueLock = stdx::unique_lock<stdx::mutex>;

const auto kIdIndexName = "_id_"_sd;

}  // namespace

StorageInterfaceImpl::StorageInterfaceImpl()
    : _rollbackIdNss(StorageInterfaceImpl::kDefaultRollbackIdNamespace) {}

StatusWith<int> StorageInterfaceImpl::getRollbackID(OperationContext* opCtx) {
    BSONObjBuilder bob;
    bob.append("_id", kRollbackIdDocumentId);
    auto id = bob.obj();

    try {
        auto rbidDoc = findById(opCtx, _rollbackIdNss, id["_id"]);
        if (!rbidDoc.isOK()) {
            return rbidDoc.getStatus();
        }

        auto rbid = RollbackID::parse(IDLParserErrorContext("RollbackID"), rbidDoc.getValue());
        invariant(rbid.get_id() == kRollbackIdDocumentId);
        return rbid.getRollbackId();
    } catch (...) {
        return exceptionToStatus();
    }

    MONGO_UNREACHABLE;
}

Status StorageInterfaceImpl::initializeRollbackID(OperationContext* opCtx) {
    auto status = createCollection(opCtx, _rollbackIdNss, CollectionOptions());
    if (!status.isOK()) {
        return status;
    }

    RollbackID rbid;
    rbid.set_id(kRollbackIdDocumentId);
    rbid.setRollbackId(0);

    BSONObjBuilder bob;
    rbid.serialize(&bob);
    return insertDocument(opCtx, _rollbackIdNss, bob.done());
}

Status StorageInterfaceImpl::incrementRollbackID(OperationContext* opCtx) {
    // This is safe because this is only called during rollback, and you can not have two
    // rollbacks at once.
    auto rbid = getRollbackID(opCtx);
    if (!rbid.isOK()) {
        return rbid.getStatus();
    }

    // If we would go over the integer limit, reset the Rollback ID to 0.
    BSONObjBuilder updateBob;
    if (rbid.getValue() == std::numeric_limits<int>::max()) {
        BSONObjBuilder setBob(updateBob.subobjStart("$set"));
        setBob.append(kRollbackIdFieldName, 0);
    } else {
        BSONObjBuilder incBob(updateBob.subobjStart("$inc"));
        incBob.append(kRollbackIdFieldName, 1);
    }

    // Since the Rollback ID is in a singleton collection, we can fix the _id field.
    BSONObjBuilder bob;
    bob.append("_id", kRollbackIdDocumentId);
    auto id = bob.obj();
    Status status = upsertById(opCtx, _rollbackIdNss, id["_id"], updateBob.obj());

    // We wait until durable so that we are sure the Rollback ID is updated before rollback ends.
    if (status.isOK()) {
        opCtx->recoveryUnit()->waitUntilDurable();
    }
    return status;
}

StatusWith<std::unique_ptr<CollectionBulkLoader>>
StorageInterfaceImpl::createCollectionForBulkLoading(
    const NamespaceString& nss,
    const CollectionOptions& options,
    const BSONObj idIndexSpec,
    const std::vector<BSONObj>& secondaryIndexSpecs) {

    LOG(2) << "StorageInterfaceImpl::createCollectionForBulkLoading called for ns: " << nss.ns();
    auto threadPool =
        stdx::make_unique<OldThreadPool>(1, str::stream() << "InitialSyncInserters-" << nss.ns());
    std::unique_ptr<TaskRunner> runner = stdx::make_unique<TaskRunner>(threadPool.get());

    // Setup cond_var for signalling when done.
    std::unique_ptr<CollectionBulkLoader> loaderToReturn;
    Collection* collection;

    auto status = runner->runSynchronousTask([&](OperationContext* opCtx) -> Status {
        // We are not replicating nor validating writes under this OperationContext*.
        // The OperationContext* is used for all writes to the (newly) cloned collection.
        UnreplicatedWritesBlock uwb(opCtx);
        documentValidationDisabled(opCtx) = true;

        // Retry if WCE.
        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            // Get locks and create the collection.
            auto db = stdx::make_unique<AutoGetOrCreateDb>(opCtx, nss.db(), MODE_IX);
            auto coll = stdx::make_unique<AutoGetCollection>(opCtx, nss, MODE_X);
            collection = coll->getCollection();

            if (collection) {
                return {ErrorCodes::NamespaceExists, "Collection already exists."};
            }

            // Create the collection.
            WriteUnitOfWork wunit(opCtx);
            collection = db->getDb()->createCollection(opCtx, nss.ns(), options, false);
            invariant(collection);
            wunit.commit();

            // Build empty capped indexes.  Capped indexes cannot be build by the MultiIndexBlock
            // because the cap might delete documents off the back while we are inserting them into
            // the front.
            if (options.capped) {
                WriteUnitOfWork wunit(opCtx);
                if (!idIndexSpec.isEmpty()) {
                    auto status = collection->getIndexCatalog()->createIndexOnEmptyCollection(
                        opCtx, idIndexSpec);
                    if (!status.getStatus().isOK()) {
                        return status.getStatus();
                    }
                }
                for (auto&& spec : secondaryIndexSpecs) {
                    auto status =
                        collection->getIndexCatalog()->createIndexOnEmptyCollection(opCtx, spec);
                    if (!status.getStatus().isOK()) {
                        return status.getStatus();
                    }
                }
                wunit.commit();
            }


            coll = stdx::make_unique<AutoGetCollection>(opCtx, nss, MODE_IX);

            // Move locks into loader, so it now controls their lifetime.
            auto loader = stdx::make_unique<CollectionBulkLoaderImpl>(opCtx,
                                                                      collection,
                                                                      options.capped ? BSONObj()
                                                                                     : idIndexSpec,
                                                                      std::move(threadPool),
                                                                      std::move(runner),
                                                                      std::move(db),
                                                                      std::move(coll));

            // Move the loader into the StatusWith.
            loaderToReturn = std::move(loader);
            return Status::OK();
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, "beginCollectionClone", nss.ns());
        MONGO_UNREACHABLE;
    });

    if (!status.isOK()) {
        return status;
    }

    invariant(collection);

    status = loaderToReturn->init(collection,
                                  options.capped ? std::vector<BSONObj>() : secondaryIndexSpecs);
    if (!status.isOK()) {
        return status;
    }
    return std::move(loaderToReturn);
}


Status StorageInterfaceImpl::insertDocument(OperationContext* opCtx,
                                            const NamespaceString& nss,
                                            const BSONObj& doc) {
    return insertDocuments(opCtx, nss, {doc});
}

namespace {

/**
 * Returns Collection* from database RAII object.
 * Returns NamespaceNotFound if the database or collection does not exist.
 */
template <typename AutoGetCollectionType>
StatusWith<Collection*> getCollection(const AutoGetCollectionType& autoGetCollection,
                                      const NamespaceString& nss,
                                      const std::string& message) {
    if (!autoGetCollection.getDb()) {
        return {ErrorCodes::NamespaceNotFound,
                str::stream() << "Database [" << nss.db() << "] not found. " << message};
    }

    auto collection = autoGetCollection.getCollection();
    if (!collection) {
        return {ErrorCodes::NamespaceNotFound,
                str::stream() << "Collection [" << nss.ns() << "] not found. " << message};
    }

    return collection;
}

Status insertDocumentsSingleBatch(OperationContext* opCtx,
                                  const NamespaceString& nss,
                                  std::vector<BSONObj>::const_iterator begin,
                                  std::vector<BSONObj>::const_iterator end) {
    AutoGetCollection autoColl(opCtx, nss, MODE_IX);

    auto collectionResult =
        getCollection(autoColl, nss, "The collection must exist before inserting documents.");
    if (!collectionResult.isOK()) {
        return collectionResult.getStatus();
    }
    auto collection = collectionResult.getValue();

    WriteUnitOfWork wunit(opCtx);
    OpDebug* const nullOpDebug = nullptr;
    auto status = collection->insertDocuments(opCtx, begin, end, nullOpDebug, false);
    if (!status.isOK()) {
        return status;
    }
    wunit.commit();

    return Status::OK();
}

}  // namespace

Status StorageInterfaceImpl::insertDocuments(OperationContext* opCtx,
                                             const NamespaceString& nss,
                                             const std::vector<BSONObj>& docs) {
    if (docs.size() > 1U) {
        try {
            if (insertDocumentsSingleBatch(opCtx, nss, docs.cbegin(), docs.cend()).isOK()) {
                return Status::OK();
            }
        } catch (...) {
            // Ignore this failure and behave as-if we never tried to do the combined batch insert.
            // The loop below will handle reporting any non-transient errors.
        }
    }

    // Try to insert the batch one-at-a-time because the batch failed all-at-once inserting.
    for (auto it = docs.cbegin(); it != docs.cend(); ++it) {
        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            auto status = insertDocumentsSingleBatch(opCtx, nss, it, it + 1);
            if (!status.isOK()) {
                return status;
            }
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(
            opCtx, "StorageInterfaceImpl::insertDocuments", nss.ns());
    }

    return Status::OK();
}

Status StorageInterfaceImpl::dropReplicatedDatabases(OperationContext* opCtx) {
    dropAllDatabasesExceptLocal(opCtx);
    return Status::OK();
}

Status StorageInterfaceImpl::createOplog(OperationContext* opCtx, const NamespaceString& nss) {
    mongo::repl::createOplog(opCtx, nss.ns(), true);
    return Status::OK();
}

StatusWith<size_t> StorageInterfaceImpl::getOplogMaxSize(OperationContext* opCtx,
                                                         const NamespaceString& nss) {
    AutoGetCollectionForReadCommand autoColl(opCtx, nss);
    auto collectionResult = getCollection(autoColl, nss, "Your oplog doesn't exist.");
    if (!collectionResult.isOK()) {
        return collectionResult.getStatus();
    }
    auto collection = collectionResult.getValue();

    const auto options = collection->getCatalogEntry()->getCollectionOptions(opCtx);
    if (!options.capped)
        return {ErrorCodes::BadValue, str::stream() << nss.ns() << " isn't capped"};

    return options.cappedSize;
}

Status StorageInterfaceImpl::createCollection(OperationContext* opCtx,
                                              const NamespaceString& nss,
                                              const CollectionOptions& options) {
    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        AutoGetOrCreateDb databaseWriteGuard(opCtx, nss.db(), MODE_X);
        auto db = databaseWriteGuard.getDb();
        invariant(db);
        if (db->getCollection(opCtx, nss)) {
            return {ErrorCodes::NamespaceExists,
                    str::stream() << "Collection " << nss.ns() << " already exists."};
        }
        WriteUnitOfWork wuow(opCtx);
        try {
            auto coll = db->createCollection(opCtx, nss.ns(), options);
            invariant(coll);
        } catch (const UserException& ex) {
            return ex.toStatus();
        }
        wuow.commit();
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, "StorageInterfaceImpl::createCollection", nss.ns());
    return Status::OK();
}

Status StorageInterfaceImpl::dropCollection(OperationContext* opCtx, const NamespaceString& nss) {
    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        AutoGetDb autoDB(opCtx, nss.db(), MODE_X);
        if (!autoDB.getDb()) {
            // Database does not exist - nothing to do.
            return Status::OK();
        }
        WriteUnitOfWork wunit(opCtx);
        const auto status = autoDB.getDb()->dropCollectionEvenIfSystem(opCtx, nss);
        if (status.isOK()) {
            wunit.commit();
        }
        return status;
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, "StorageInterfaceImpl::dropCollection", nss.ns());
}

namespace {

/**
 * Returns DeleteStageParams for deleteOne with fetch.
 */
DeleteStageParams makeDeleteStageParamsForDeleteDocuments() {
    DeleteStageParams deleteStageParams;
    deleteStageParams.isMulti = true;
    deleteStageParams.returnDeleted = true;
    return deleteStageParams;
}

/**
 * Shared implementation between findDocuments, deleteDocuments, and _findOrDeleteById.
 * _findOrDeleteById is used by findById, and deleteById.
 */
enum class FindDeleteMode { kFind, kDelete };
StatusWith<std::vector<BSONObj>> _findOrDeleteDocuments(
    OperationContext* opCtx,
    const NamespaceString& nss,
    boost::optional<StringData> indexName,
    StorageInterface::ScanDirection scanDirection,
    const BSONObj& startKey,
    const BSONObj& endKey,
    BoundInclusion boundInclusion,
    std::size_t limit,
    FindDeleteMode mode) {
    auto isFind = mode == FindDeleteMode::kFind;
    auto opStr = isFind ? "StorageInterfaceImpl::find" : "StorageInterfaceImpl::delete";

    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        auto collectionAccessMode = isFind ? MODE_IS : MODE_IX;
        AutoGetCollection autoColl(opCtx, nss, collectionAccessMode);
        auto collectionResult = getCollection(
            autoColl, nss, str::stream() << "Unable to proceed with " << opStr << ".");
        if (!collectionResult.isOK()) {
            return collectionResult.getStatus();
        }
        auto collection = collectionResult.getValue();

        auto isForward = scanDirection == StorageInterface::ScanDirection::kForward;
        auto direction = isForward ? InternalPlanner::FORWARD : InternalPlanner::BACKWARD;

        std::unique_ptr<PlanExecutor, PlanExecutor::Deleter> planExecutor;
        if (!indexName) {
            if (!startKey.isEmpty()) {
                return {ErrorCodes::NoSuchKey,
                        "non-empty startKey not allowed for collection scan"};
            }
            if (boundInclusion != BoundInclusion::kIncludeStartKeyOnly) {
                return {ErrorCodes::InvalidOptions,
                        "bound inclusion must be BoundInclusion::kIncludeStartKeyOnly for "
                        "collection scan"};
            }
            // Use collection scan.
            planExecutor = isFind
                ? InternalPlanner::collectionScan(
                      opCtx, nss.ns(), collection, PlanExecutor::NO_YIELD, direction)
                : InternalPlanner::deleteWithCollectionScan(
                      opCtx,
                      collection,
                      makeDeleteStageParamsForDeleteDocuments(),
                      PlanExecutor::NO_YIELD,
                      direction);
        } else {
            // Use index scan.
            auto indexCatalog = collection->getIndexCatalog();
            invariant(indexCatalog);
            bool includeUnfinishedIndexes = false;
            IndexDescriptor* indexDescriptor =
                indexCatalog->findIndexByName(opCtx, *indexName, includeUnfinishedIndexes);
            if (!indexDescriptor) {
                return {ErrorCodes::IndexNotFound,
                        str::stream() << "Index not found, ns:" << nss.ns() << ", index: "
                                      << *indexName};
            }
            if (indexDescriptor->isPartial()) {
                return {ErrorCodes::IndexOptionsConflict,
                        str::stream() << "Partial index is not allowed for this operation, ns:"
                                      << nss.ns()
                                      << ", index: "
                                      << *indexName};
            }

            KeyPattern keyPattern(indexDescriptor->keyPattern());
            auto minKey = Helpers::toKeyFormat(keyPattern.extendRangeBound({}, false));
            auto maxKey = Helpers::toKeyFormat(keyPattern.extendRangeBound({}, true));
            auto bounds =
                isForward ? std::make_pair(minKey, maxKey) : std::make_pair(maxKey, minKey);
            if (!startKey.isEmpty()) {
                bounds.first = startKey;
            }
            if (!endKey.isEmpty()) {
                bounds.second = endKey;
            }
            planExecutor = isFind
                ? InternalPlanner::indexScan(opCtx,
                                             collection,
                                             indexDescriptor,
                                             bounds.first,
                                             bounds.second,
                                             boundInclusion,
                                             PlanExecutor::NO_YIELD,
                                             direction,
                                             InternalPlanner::IXSCAN_FETCH)
                : InternalPlanner::deleteWithIndexScan(opCtx,
                                                       collection,
                                                       makeDeleteStageParamsForDeleteDocuments(),
                                                       indexDescriptor,
                                                       bounds.first,
                                                       bounds.second,
                                                       boundInclusion,
                                                       PlanExecutor::NO_YIELD,
                                                       direction);
        }

        std::vector<BSONObj> docs;
        while (docs.size() < limit) {
            BSONObj doc;
            auto state = planExecutor->getNext(&doc, nullptr);
            if (PlanExecutor::ADVANCED == state) {
                docs.push_back(doc.getOwned());
            } else {
                invariant(PlanExecutor::IS_EOF == state);
                break;
            }
        }
        return docs;
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, opStr, nss.ns());
    MONGO_UNREACHABLE;
}

StatusWith<BSONObj> _findOrDeleteById(OperationContext* opCtx,
                                      const NamespaceString& nss,
                                      const BSONElement& idKey,
                                      FindDeleteMode mode) {
    auto wrappedIdKey = idKey.wrap("");
    auto result = _findOrDeleteDocuments(opCtx,
                                         nss,
                                         kIdIndexName,
                                         StorageInterface::ScanDirection::kForward,
                                         wrappedIdKey,
                                         wrappedIdKey,
                                         BoundInclusion::kIncludeBothStartAndEndKeys,
                                         1U,
                                         mode);
    if (!result.isOK()) {
        return result.getStatus();
    }
    const auto& docs = result.getValue();
    if (docs.empty()) {
        return {ErrorCodes::NoSuchKey, str::stream() << "No document found with _id: " << idKey};
    }

    return docs.front();
}

}  // namespace

StatusWith<std::vector<BSONObj>> StorageInterfaceImpl::findDocuments(
    OperationContext* opCtx,
    const NamespaceString& nss,
    boost::optional<StringData> indexName,
    ScanDirection scanDirection,
    const BSONObj& startKey,
    BoundInclusion boundInclusion,
    std::size_t limit) {
    return _findOrDeleteDocuments(opCtx,
                                  nss,
                                  indexName,
                                  scanDirection,
                                  startKey,
                                  {},
                                  boundInclusion,
                                  limit,
                                  FindDeleteMode::kFind);
}

StatusWith<std::vector<BSONObj>> StorageInterfaceImpl::deleteDocuments(
    OperationContext* opCtx,
    const NamespaceString& nss,
    boost::optional<StringData> indexName,
    ScanDirection scanDirection,
    const BSONObj& startKey,
    BoundInclusion boundInclusion,
    std::size_t limit) {
    return _findOrDeleteDocuments(opCtx,
                                  nss,
                                  indexName,
                                  scanDirection,
                                  startKey,
                                  {},
                                  boundInclusion,
                                  limit,
                                  FindDeleteMode::kDelete);
}

StatusWith<BSONObj> StorageInterfaceImpl::findSingleton(OperationContext* opCtx,
                                                        const NamespaceString& nss) {
    auto result = findDocuments(opCtx,
                                nss,
                                boost::none,  // Collection scan.
                                StorageInterface::ScanDirection::kForward,
                                {},  // Start at the beginning of the collection.
                                BoundInclusion::kIncludeStartKeyOnly,
                                2U);  // Ask for 2 documents to ensure it's a singleton.
    if (!result.isOK()) {
        return result.getStatus();
    }

    const auto& docs = result.getValue();
    if (docs.empty()) {
        return {ErrorCodes::CollectionIsEmpty,
                str::stream() << "No document found in namespace: " << nss.ns()};
    } else if (docs.size() != 1U) {
        return {ErrorCodes::TooManyMatchingDocuments,
                str::stream() << "More than singleton document found in namespace: " << nss.ns()};
    }

    return docs.front();
}

StatusWith<BSONObj> StorageInterfaceImpl::findById(OperationContext* opCtx,
                                                   const NamespaceString& nss,
                                                   const BSONElement& idKey) {
    return _findOrDeleteById(opCtx, nss, idKey, FindDeleteMode::kFind);
}

StatusWith<BSONObj> StorageInterfaceImpl::deleteById(OperationContext* opCtx,
                                                     const NamespaceString& nss,
                                                     const BSONElement& idKey) {
    return _findOrDeleteById(opCtx, nss, idKey, FindDeleteMode::kDelete);
}

namespace {

/**
 * Checks _id key passed to upsertById and returns a query document for UpdateRequest.
 */
StatusWith<BSONObj> makeUpsertQuery(const BSONElement& idKey) {
    auto query = BSON("_id" << idKey);

    // With the ID hack, only simple _id queries are allowed. Otherwise, UpdateStage will fail with
    // a fatal assertion.
    if (!CanonicalQuery::isSimpleIdQuery(query)) {
        return {ErrorCodes::InvalidIdField,
                str::stream() << "Unable to update document with a non-simple _id query: "
                              << query};
    }

    return query;
}

Status _upsertWithQuery(OperationContext* opCtx,
                        const NamespaceString& nss,
                        const BSONObj& query,
                        const BSONObj& update) {
    UpdateRequest request(nss);
    request.setQuery(query);
    request.setUpdates(update);
    request.setUpsert(true);
    invariant(!request.isMulti());  // We only want to update one document for performance.
    invariant(!request.shouldReturnAnyDocs());
    invariant(PlanExecutor::NO_YIELD == request.getYieldPolicy());

    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        // ParsedUpdate needs to be inside the write conflict retry loop because it may create a
        // CanonicalQuery whose ownership will be transferred to the plan executor in
        // getExecutorUpdate().
        ParsedUpdate parsedUpdate(opCtx, &request);
        auto parsedUpdateStatus = parsedUpdate.parseRequest();
        if (!parsedUpdateStatus.isOK()) {
            return parsedUpdateStatus;
        }

        AutoGetCollection autoColl(opCtx, nss, MODE_IX);
        auto collectionResult = getCollection(
            autoColl,
            nss,
            str::stream() << "Unable to update documents in " << nss.ns() << " using query "
                          << query);
        if (!collectionResult.isOK()) {
            return collectionResult.getStatus();
        }
        auto collection = collectionResult.getValue();

        auto planExecutorResult =
            mongo::getExecutorUpdate(opCtx, nullptr, collection, &parsedUpdate);
        if (!planExecutorResult.isOK()) {
            return planExecutorResult.getStatus();
        }
        auto planExecutor = std::move(planExecutorResult.getValue());

        return planExecutor->executePlan();
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, "_upsertWithQuery", nss.ns());

    MONGO_UNREACHABLE;
}

}  // namespace

Status StorageInterfaceImpl::upsertById(OperationContext* opCtx,
                                        const NamespaceString& nss,
                                        const BSONElement& idKey,
                                        const BSONObj& update) {
    // Validate and construct an _id query for UpdateResult.
    // The _id key will be passed directly to IDHackStage.
    auto queryResult = makeUpsertQuery(idKey);
    if (!queryResult.isOK()) {
        return queryResult.getStatus();
    }
    auto query = queryResult.getValue();

    UpdateRequest request(nss);
    request.setQuery(query);
    request.setUpdates(update);
    request.setUpsert(true);
    invariant(!request.isMulti());  // This follows from using an exact _id query.
    invariant(!request.shouldReturnAnyDocs());
    invariant(PlanExecutor::NO_YIELD == request.getYieldPolicy());

    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        // ParsedUpdate needs to be inside the write conflict retry loop because it contains
        // the UpdateDriver whose state may be modified while we are applying the update.
        ParsedUpdate parsedUpdate(opCtx, &request);
        auto parsedUpdateStatus = parsedUpdate.parseRequest();
        if (!parsedUpdateStatus.isOK()) {
            return parsedUpdateStatus;
        }

        AutoGetCollection autoColl(opCtx, nss, MODE_IX);
        auto collectionResult = getCollection(autoColl, nss, "Unable to update document.");
        if (!collectionResult.isOK()) {
            return collectionResult.getStatus();
        }
        auto collection = collectionResult.getValue();

        // We're using the ID hack to perform the update so we have to disallow collections
        // without an _id index.
        auto descriptor = collection->getIndexCatalog()->findIdIndex(opCtx);
        if (!descriptor) {
            return {ErrorCodes::IndexNotFound,
                    "Unable to update document in a collection without an _id index."};
        }

        UpdateStageParams updateStageParams(
            parsedUpdate.getRequest(), parsedUpdate.getDriver(), nullptr);
        auto planExecutor = InternalPlanner::updateWithIdHack(opCtx,
                                                              collection,
                                                              updateStageParams,
                                                              descriptor,
                                                              idKey.wrap(""),
                                                              parsedUpdate.yieldPolicy());

        return planExecutor->executePlan();
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, "StorageInterfaceImpl::upsertById", nss.ns());

    MONGO_UNREACHABLE;
}

Status StorageInterfaceImpl::putSingleton(OperationContext* opCtx,
                                          const NamespaceString& nss,
                                          const BSONObj& update) {
    return _upsertWithQuery(opCtx, nss, {}, update);
}

Status StorageInterfaceImpl::deleteByFilter(OperationContext* opCtx,
                                            const NamespaceString& nss,
                                            const BSONObj& filter) {
    DeleteRequest request(nss);
    request.setQuery(filter);
    request.setMulti(true);
    request.setYieldPolicy(PlanExecutor::NO_YIELD);

    // This disables the legalClientSystemNS() check in getExecutorDelete() which is used to
    // disallow client deletes from unrecognized system collections.
    request.setGod();

    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        // ParsedDelete needs to be inside the write conflict retry loop because it may create a
        // CanonicalQuery whose ownership will be transferred to the plan executor in
        // getExecutorDelete().
        ParsedDelete parsedDelete(opCtx, &request);
        auto parsedDeleteStatus = parsedDelete.parseRequest();
        if (!parsedDeleteStatus.isOK()) {
            return parsedDeleteStatus;
        }

        AutoGetCollection autoColl(opCtx, nss, MODE_IX);
        auto collectionResult = getCollection(
            autoColl,
            nss,
            str::stream() << "Unable to delete documents in " << nss.ns() << " using filter "
                          << filter);
        if (!collectionResult.isOK()) {
            return collectionResult.getStatus();
        }
        auto collection = collectionResult.getValue();

        auto planExecutorResult =
            mongo::getExecutorDelete(opCtx, nullptr, collection, &parsedDelete);
        if (!planExecutorResult.isOK()) {
            return planExecutorResult.getStatus();
        }
        auto planExecutor = std::move(planExecutorResult.getValue());

        return planExecutor->executePlan();
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(opCtx, "StorageInterfaceImpl::deleteByFilter", nss.ns());

    MONGO_UNREACHABLE;
}

StatusWith<StorageInterface::CollectionSize> StorageInterfaceImpl::getCollectionSize(
    OperationContext* opCtx, const NamespaceString& nss) {
    AutoGetCollectionForRead autoColl(opCtx, nss);

    auto collectionResult =
        getCollection(autoColl, nss, "Unable to get total size of documents in collection.");
    if (!collectionResult.isOK()) {
        return collectionResult.getStatus();
    }
    auto collection = collectionResult.getValue();

    return collection->dataSize(opCtx);
}

StatusWith<StorageInterface::CollectionCount> StorageInterfaceImpl::getCollectionCount(
    OperationContext* opCtx, const NamespaceString& nss) {
    AutoGetCollectionForRead autoColl(opCtx, nss);

    auto collectionResult =
        getCollection(autoColl, nss, "Unable to get number of documents in collection.");
    if (!collectionResult.isOK()) {
        return collectionResult.getStatus();
    }
    auto collection = collectionResult.getValue();

    return collection->numRecords(opCtx);
}

Status StorageInterfaceImpl::isAdminDbValid(OperationContext* opCtx) {
    AutoGetDb autoDB(opCtx, "admin", MODE_X);
    auto adminDb = autoDB.getDb();
    if (!adminDb) {
        return Status::OK();
    }

    Collection* const usersCollection =
        adminDb->getCollection(opCtx, AuthorizationManager::usersCollectionNamespace);
    const bool hasUsers =
        usersCollection && !Helpers::findOne(opCtx, usersCollection, BSONObj(), false).isNull();
    Collection* const adminVersionCollection =
        adminDb->getCollection(opCtx, AuthorizationManager::versionCollectionNamespace);
    BSONObj authSchemaVersionDocument;
    if (!adminVersionCollection ||
        !Helpers::findOne(opCtx,
                          adminVersionCollection,
                          AuthorizationManager::versionDocumentQuery,
                          authSchemaVersionDocument)) {
        if (!hasUsers) {
            // It's OK to have no auth version document if there are no user documents.
            return Status::OK();
        }
        std::string msg = str::stream()
            << "During initial sync, found documents in "
            << AuthorizationManager::usersCollectionNamespace.ns()
            << " but could not find an auth schema version document in "
            << AuthorizationManager::versionCollectionNamespace.ns() << ".  "
            << "This indicates that the primary of this replica set was not successfully "
               "upgraded to schema version "
            << AuthorizationManager::schemaVersion26Final
            << ", which is the minimum supported schema version in this version of MongoDB";
        return {ErrorCodes::AuthSchemaIncompatible, msg};
    }
    long long foundSchemaVersion;
    Status status = bsonExtractIntegerField(authSchemaVersionDocument,
                                            AuthorizationManager::schemaVersionFieldName,
                                            &foundSchemaVersion);
    if (!status.isOK()) {
        std::string msg = str::stream()
            << "During initial sync, found malformed auth schema version document: "
            << status.toString() << "; document: " << authSchemaVersionDocument;
        return {ErrorCodes::AuthSchemaIncompatible, msg};
    }
    if ((foundSchemaVersion != AuthorizationManager::schemaVersion26Final) &&
        (foundSchemaVersion != AuthorizationManager::schemaVersion28SCRAM)) {
        std::string msg = str::stream()
            << "During initial sync, found auth schema version " << foundSchemaVersion
            << ", but this version of MongoDB only supports schema versions "
            << AuthorizationManager::schemaVersion26Final << " and "
            << AuthorizationManager::schemaVersion28SCRAM;
        return {ErrorCodes::AuthSchemaIncompatible, msg};
    }

    return Status::OK();
}

}  // namespace repl
}  // namespace mongo