summaryrefslogtreecommitdiff
path: root/src/mongo/db/s/migration_util.cpp
blob: c7eb2183d85930dd3d6f35533eb12f19c8899ef1 (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
/**
 *    Copyright (C) 2019-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_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kSharding

#include "mongo/platform/basic.h"

#include "mongo/db/s/migration_util.h"

#include "mongo/base/error_codes.h"
#include "mongo/bson/bsonobj.h"
#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/client/query.h"
#include "mongo/db/catalog/collection_catalog_helper.h"
#include "mongo/db/catalog_raii.h"
#include "mongo/db/dbdirectclient.h"
#include "mongo/db/logical_session_cache.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/ops/write_ops.h"
#include "mongo/db/repl/repl_client_info.h"
#include "mongo/db/repl/replication_coordinator.h"
#include "mongo/db/s/active_migrations_registry.h"
#include "mongo/db/s/collection_sharding_runtime.h"
#include "mongo/db/s/migration_coordinator.h"
#include "mongo/db/s/shard_filtering_metadata_refresh.h"
#include "mongo/db/s/wait_for_majority_service.h"
#include "mongo/db/write_concern.h"
#include "mongo/executor/task_executor_pool.h"
#include "mongo/executor/thread_pool_task_executor.h"
#include "mongo/logv2/log.h"
#include "mongo/rpc/get_status_from_command_result.h"
#include "mongo/s/catalog/type_chunk.h"
#include "mongo/s/client/shard.h"
#include "mongo/s/grid.h"
#include "mongo/s/request_types/ensure_chunk_version_is_greater_than_gen.h"
#include "mongo/util/exit.h"
#include "mongo/util/log.h"

namespace mongo {
namespace migrationutil {
namespace {

MONGO_FAIL_POINT_DEFINE(hangBeforeFilteringMetadataRefresh);
MONGO_FAIL_POINT_DEFINE(hangInEnsureChunkVersionIsGreaterThanInterruptible);
MONGO_FAIL_POINT_DEFINE(hangInEnsureChunkVersionIsGreaterThanThenSimulateErrorUninterruptible);
MONGO_FAIL_POINT_DEFINE(hangInRefreshFilteringMetadataUntilSuccessInterruptible);
MONGO_FAIL_POINT_DEFINE(hangInRefreshFilteringMetadataUntilSuccessThenSimulateErrorUninterruptible);

const char kSourceShard[] = "source";
const char kDestinationShard[] = "destination";
const char kIsDonorShard[] = "isDonorShard";
const char kChunk[] = "chunk";
const char kCollection[] = "collection";

const WriteConcernOptions kMajorityWriteConcern(WriteConcernOptions::kMajority,
                                                WriteConcernOptions::SyncMode::UNSET,
                                                WriteConcernOptions::kNoTimeout);

template <typename Cmd>
void sendToRecipient(OperationContext* opCtx,
                     const ShardId& recipientId,
                     const Cmd& cmd,
                     const BSONObj& passthroughFields = {}) {
    auto recipientShard =
        uassertStatusOK(Grid::get(opCtx)->shardRegistry()->getShard(opCtx, recipientId));

    auto cmdBSON = cmd.toBSON(passthroughFields);
    LOGV2_DEBUG(22023, 1, "Sending request {cmd} to recipient.", "cmd"_attr = cmdBSON);

    auto response = recipientShard->runCommandWithFixedRetryAttempts(
        opCtx,
        ReadPreferenceSetting{ReadPreference::PrimaryOnly},
        cmd.getDbName().toString(),
        cmdBSON,
        Shard::RetryPolicy::kIdempotent);

    uassertStatusOK(Shard::CommandResponse::getEffectiveStatus(response));
}

}  // namespace

BSONObj makeMigrationStatusDocument(const NamespaceString& nss,
                                    const ShardId& fromShard,
                                    const ShardId& toShard,
                                    const bool& isDonorShard,
                                    const BSONObj& min,
                                    const BSONObj& max) {
    BSONObjBuilder builder;
    builder.append(kSourceShard, fromShard.toString());
    builder.append(kDestinationShard, toShard.toString());
    builder.append(kIsDonorShard, isDonorShard);
    builder.append(kChunk, BSON(ChunkType::min(min) << ChunkType::max(max)));
    builder.append(kCollection, nss.ns());
    return builder.obj();
}

Query overlappingRangeQuery(const ChunkRange& range, const UUID& uuid) {
    return QUERY(RangeDeletionTask::kCollectionUuidFieldName
                 << uuid << RangeDeletionTask::kRangeFieldName + "." + ChunkRange::kMinKey << LT
                 << range.getMax() << RangeDeletionTask::kRangeFieldName + "." + ChunkRange::kMaxKey
                 << GT << range.getMin());
}

bool checkForConflictingDeletions(OperationContext* opCtx,
                                  const ChunkRange& range,
                                  const UUID& uuid) {
    PersistentTaskStore<RangeDeletionTask> store(opCtx, NamespaceString::kRangeDeletionNamespace);

    return store.count(opCtx, overlappingRangeQuery(range, uuid)) > 0;
}

ExecutorFuture<void> submitRangeDeletionTask(OperationContext* opCtx,
                                             const RangeDeletionTask& deletionTask) {
    const auto serviceContext = opCtx->getServiceContext();
    return ExecutorFuture<void>(Grid::get(serviceContext)->getExecutorPool()->getFixedExecutor())
        .then([=] {
            ThreadClient tc(kRangeDeletionThreadName, serviceContext);
            {
                stdx::lock_guard<Client> lk(*tc.get());
                tc->setSystemOperationKillable(lk);
            }
            auto uniqueOpCtx = tc->makeOperationContext();
            auto opCtx = uniqueOpCtx.get();

            boost::optional<AutoGetCollection> autoColl;
            autoColl.emplace(opCtx, deletionTask.getNss(), MODE_IS);

            auto css = CollectionShardingRuntime::get(opCtx, deletionTask.getNss());
            if (!css->getCurrentMetadataIfKnown() || !css->getCurrentMetadata()->isSharded() ||
                !css->getCurrentMetadata()->uuidMatches(deletionTask.getCollectionUuid())) {
                // If the collection's filtering metadata is not known, is unsharded, or its UUID
                // does not match the UUID of the deletion task, force a filtering metadata refresh
                // once, because this node may have just stepped up and therefore may have a stale
                // cache.
                LOGV2(
                    22024,
                    "Filtering metadata for namespace in deletion task "
                    "{deletionTask}{css_getCurrentMetadataIfKnown_css_getCurrentMetadata_isSharded_"
                    "has_UUID_that_does_not_match_UUID_of_the_deletion_task_is_unsharded_is_not_"
                    "known}, forcing a refresh of {deletionTask_getNss}",
                    "deletionTask"_attr = deletionTask.toBSON(),
                    "css_getCurrentMetadataIfKnown_css_getCurrentMetadata_isSharded_has_UUID_that_does_not_match_UUID_of_the_deletion_task_is_unsharded_is_not_known"_attr =
                        (css->getCurrentMetadataIfKnown()
                             ? (css->getCurrentMetadata()->isSharded()
                                    ? " has UUID that does not match UUID of the deletion task"
                                    : " is unsharded")
                             : " is not known"),
                    "deletionTask_getNss"_attr = deletionTask.getNss());

                // TODO (SERVER-46075): Add an asynchronous version of
                // forceShardFilteringMetadataRefresh to avoid blocking on the network in the
                // thread pool.
                autoColl.reset();
                refreshFilteringMetadataUntilSuccess(opCtx, deletionTask.getNss());
            }

            autoColl.emplace(opCtx, deletionTask.getNss(), MODE_IS);
            uassert(
                ErrorCodes::RangeDeletionAbandonedBecauseCollectionWithUUIDDoesNotExist,
                str::stream() << "Even after forced refresh, filtering metadata for namespace in "
                                 "deletion task "
                              << (css->getCurrentMetadata()->isSharded()
                                      ? " has UUID that does not match UUID of the deletion task"
                                      : " is unsharded"),
                css->getCurrentMetadata()->isSharded() &&
                    css->getCurrentMetadata()->uuidMatches(deletionTask.getCollectionUuid()));

            LOGV2(22026,
                  "Submitting range deletion task {deletionTask}",
                  "deletionTask"_attr = deletionTask.toBSON());

            const auto whenToClean = deletionTask.getWhenToClean() == CleanWhenEnum::kNow
                ? CollectionShardingRuntime::kNow
                : CollectionShardingRuntime::kDelayed;

            return css->cleanUpRange(deletionTask.getRange(), whenToClean);
        })
        .onError([=](const Status status) {
            ThreadClient tc(kRangeDeletionThreadName, serviceContext);
            {
                stdx::lock_guard<Client> lk(*tc.get());
                tc->setSystemOperationKillable(lk);
            }
            auto uniqueOpCtx = tc->makeOperationContext();
            auto opCtx = uniqueOpCtx.get();

            LOGV2(22027,
                  "Failed to submit range deletion task "
                  "{deletionTask}{causedBy_status}",
                  "deletionTask"_attr = deletionTask.toBSON(),
                  "causedBy_status"_attr = causedBy(status));

            if (status == ErrorCodes::RangeDeletionAbandonedBecauseCollectionWithUUIDDoesNotExist) {
                deleteRangeDeletionTaskLocally(
                    opCtx, deletionTask.getId(), ShardingCatalogClient::kLocalWriteConcern);
            }

            // Note, we use onError and make it return its input status, because ExecutorFuture does
            // not support tapError.
            return status;
        });
}

void submitPendingDeletions(OperationContext* opCtx) {
    PersistentTaskStore<RangeDeletionTask> store(opCtx, NamespaceString::kRangeDeletionNamespace);

    auto query = QUERY("pending" << BSON("$exists" << false));

    std::vector<RangeDeletionTask> invalidRanges;
    store.forEach(opCtx, query, [&opCtx, &invalidRanges](const RangeDeletionTask& deletionTask) {
        migrationutil::submitRangeDeletionTask(opCtx, deletionTask).getAsync([](auto) {});
        return true;
    });
}

void resubmitRangeDeletionsOnStepUp(ServiceContext* serviceContext) {
    LOGV2(22028, "Starting pending deletion submission thread.");

    auto executor = Grid::get(serviceContext)->getExecutorPool()->getFixedExecutor();

    ExecutorFuture<void>(executor).getAsync([serviceContext](const Status& status) {
        ThreadClient tc("ResubmitRangeDeletions", serviceContext);
        {
            stdx::lock_guard<Client> lk(*tc.get());
            tc->setSystemOperationKillable(lk);
        }

        auto opCtx = tc->makeOperationContext();

        submitPendingDeletions(opCtx.get());
    });
}

void dropRangeDeletionsCollection(OperationContext* opCtx) {
    DBDirectClient client(opCtx);
    client.dropCollection(NamespaceString::kRangeDeletionNamespace.toString(),
                          WriteConcerns::kMajorityWriteConcern);
}

template <typename Callable>
void forEachOrphanRange(OperationContext* opCtx, const NamespaceString& nss, Callable&& handler) {
    AutoGetCollection autoColl(opCtx, nss, MODE_IX);

    const auto css = CollectionShardingRuntime::get(opCtx, nss);
    const auto metadata = css->getCurrentMetadata();
    const auto emptyChunkMap =
        RangeMap{SimpleBSONObjComparator::kInstance.makeBSONObjIndexedMap<BSONObj>()};

    if (!metadata->isSharded()) {
        LOGV2(22029,
              "Upgrade: skipping orphaned range enumeration for {nss}, collection is not sharded",
              "nss"_attr = nss);
        return;
    }

    auto startingKey = metadata->getMinKey();

    while (true) {
        auto range = metadata->getNextOrphanRange(emptyChunkMap, startingKey);
        if (!range) {
            LOGV2_DEBUG(22030,
                        2,
                        "Upgrade: Completed orphaned range enumeration for {nss} starting from "
                        "{startingKey}, no orphan ranges remain",
                        "nss"_attr = nss.toString(),
                        "startingKey"_attr = redact(startingKey));

            return;
        }

        handler(*range);

        startingKey = range->getMax();
    }
}

void submitOrphanRanges(OperationContext* opCtx, const NamespaceString& nss, const UUID& uuid) {
    try {
        auto version = forceShardFilteringMetadataRefresh(opCtx, nss, true);

        if (version == ChunkVersion::UNSHARDED())
            return;

        LOGV2_DEBUG(22031,
                    2,
                    "Upgrade: Cleaning up existing orphans for {nss} : {uuid}",
                    "nss"_attr = nss,
                    "uuid"_attr = uuid);

        std::vector<RangeDeletionTask> deletions;
        forEachOrphanRange(opCtx, nss, [&deletions, &opCtx, &nss, &uuid](const auto& range) {
            // Since this is not part of an active migration, the migration UUID and the donor shard
            // are set to unused values so that they don't conflict.
            RangeDeletionTask task(
                UUID::gen(), nss, uuid, ShardId("fromFCVUpgrade"), range, CleanWhenEnum::kDelayed);
            deletions.emplace_back(task);
        });

        if (deletions.empty())
            return;

        PersistentTaskStore<RangeDeletionTask> store(opCtx,
                                                     NamespaceString::kRangeDeletionNamespace);

        for (const auto& task : deletions) {
            LOGV2_DEBUG(22032,
                        2,
                        "Upgrade: Submitting range for cleanup: {task_getRange} from {nss}",
                        "task_getRange"_attr = task.getRange(),
                        "nss"_attr = nss);
            store.add(opCtx, task);
        }
    } catch (ExceptionFor<ErrorCodes::NamespaceNotFound>& e) {
        LOGV2(22033,
              "Upgrade: Failed to cleanup orphans for {nss} because the namespace was not found: "
              "{e_what}, the collection must have been dropped",
              "nss"_attr = nss,
              "e_what"_attr = e.what());
    }
}

void submitOrphanRangesForCleanup(OperationContext* opCtx) {
    auto& catalog = CollectionCatalog::get(opCtx);
    const auto& dbs = catalog.getAllDbNames();

    for (const auto& dbName : dbs) {
        if (dbName == NamespaceString::kLocalDb)
            continue;

        for (auto collIt = catalog.begin(dbName); collIt != catalog.end(); ++collIt) {
            auto uuid = collIt.uuid().get();
            auto nss = catalog.lookupNSSByUUID(opCtx, uuid).get();
            LOGV2_DEBUG(22034, 2, "Upgrade: processing collection: {nss}", "nss"_attr = nss);

            submitOrphanRanges(opCtx, nss, uuid);
        }
    }
}

void persistMigrationCoordinatorLocally(OperationContext* opCtx,
                                        const MigrationCoordinatorDocument& migrationDoc) {
    PersistentTaskStore<MigrationCoordinatorDocument> store(
        opCtx, NamespaceString::kMigrationCoordinatorsNamespace);
    try {
        store.add(opCtx, migrationDoc);
    } catch (const ExceptionFor<ErrorCodes::DuplicateKey>&) {
        // Convert a DuplicateKey error to an anonymous error.
        uasserted(
            31374,
            str::stream() << "While attempting to write migration information for migration "
                          << ", found document with the same migration id. Attempted migration: "
                          << migrationDoc.toBSON());
    }
}

void persistRangeDeletionTaskLocally(OperationContext* opCtx,
                                     const RangeDeletionTask& deletionTask) {
    PersistentTaskStore<RangeDeletionTask> store(opCtx, NamespaceString::kRangeDeletionNamespace);
    try {
        store.add(opCtx, deletionTask);
    } catch (const ExceptionFor<ErrorCodes::DuplicateKey>&) {
        // Convert a DuplicateKey error to an anonymous error.
        uasserted(31375,
                  str::stream() << "While attempting to write range deletion task for migration "
                                << ", found document with the same migration id. Attempted range "
                                   "deletion task: "
                                << deletionTask.toBSON());
    }
}

void persistCommitDecision(OperationContext* opCtx, const UUID& migrationId) {
    PersistentTaskStore<MigrationCoordinatorDocument> store(
        opCtx, NamespaceString::kMigrationCoordinatorsNamespace);
    store.update(
        opCtx,
        QUERY(MigrationCoordinatorDocument::kIdFieldName << migrationId),
        BSON("$set" << BSON(MigrationCoordinatorDocument::kDecisionFieldName << "committed")));
}

void persistAbortDecision(OperationContext* opCtx, const UUID& migrationId) {
    PersistentTaskStore<MigrationCoordinatorDocument> store(
        opCtx, NamespaceString::kMigrationCoordinatorsNamespace);
    store.update(
        opCtx,
        QUERY(MigrationCoordinatorDocument::kIdFieldName << migrationId),
        BSON("$set" << BSON(MigrationCoordinatorDocument::kDecisionFieldName << "aborted")));
}

void deleteRangeDeletionTaskOnRecipient(OperationContext* opCtx,
                                        const ShardId& recipientId,
                                        const UUID& migrationId) {
    write_ops::Delete deleteOp(NamespaceString::kRangeDeletionNamespace);
    write_ops::DeleteOpEntry query(BSON(RangeDeletionTask::kIdFieldName << migrationId),
                                   false /*multi*/);
    deleteOp.setDeletes({query});

    sendToRecipient(opCtx,
                    recipientId,
                    deleteOp,
                    BSON(WriteConcernOptions::kWriteConcernField << WriteConcernOptions::Majority));
}

void deleteRangeDeletionTaskLocally(OperationContext* opCtx,
                                    const UUID& deletionTaskId,
                                    const WriteConcernOptions& writeConcern) {
    PersistentTaskStore<RangeDeletionTask> store(opCtx, NamespaceString::kRangeDeletionNamespace);
    store.remove(opCtx, QUERY(RangeDeletionTask::kIdFieldName << deletionTaskId), writeConcern);
}

void deleteRangeDeletionTasksForCollectionLocally(OperationContext* opCtx,
                                                  const UUID& collectionUuid) {
    PersistentTaskStore<RangeDeletionTask> store(opCtx, NamespaceString::kRangeDeletionNamespace);
    store.remove(opCtx, QUERY(RangeDeletionTask::kCollectionUuidFieldName << collectionUuid));
}

void markAsReadyRangeDeletionTaskOnRecipient(OperationContext* opCtx,
                                             const ShardId& recipientId,
                                             const UUID& migrationId) {
    write_ops::Update updateOp(NamespaceString::kRangeDeletionNamespace);
    auto queryFilter = BSON(RangeDeletionTask::kIdFieldName << migrationId);
    auto updateModification = write_ops::UpdateModification(
        BSON("$unset" << BSON(RangeDeletionTask::kPendingFieldName << "")));
    write_ops::UpdateOpEntry updateEntry(queryFilter, updateModification);
    updateEntry.setMulti(false);
    updateEntry.setUpsert(false);
    updateOp.setUpdates({updateEntry});

    sendToRecipient(opCtx,
                    recipientId,
                    updateOp,
                    BSON(WriteConcernOptions::kWriteConcernField << WriteConcernOptions::Majority));
}

void advanceTransactionOnRecipient(OperationContext* opCtx,
                                   const ShardId& recipientId,
                                   const LogicalSessionId& lsid,
                                   TxnNumber txnNumber) {
    write_ops::Update updateOp(NamespaceString::kServerConfigurationNamespace);
    auto queryFilter = BSON("_id"
                            << "migrationCoordinatorStats");
    auto updateModification = write_ops::UpdateModification(BSON("$inc" << BSON("count" << 1)));

    write_ops::UpdateOpEntry updateEntry(queryFilter, updateModification);
    updateEntry.setMulti(false);
    updateEntry.setUpsert(true);
    updateOp.setUpdates({updateEntry});

    auto passthroughFields =
        BSON(WriteConcernOptions::kWriteConcernField << WriteConcernOptions::Majority << "lsid"
                                                     << lsid.toBSON() << "txnNumber" << txnNumber);
    sendToRecipient(opCtx, recipientId, updateOp, passthroughFields);
}

void markAsReadyRangeDeletionTaskLocally(OperationContext* opCtx, const UUID& migrationId) {
    PersistentTaskStore<RangeDeletionTask> store(opCtx, NamespaceString::kRangeDeletionNamespace);
    auto query = QUERY(RangeDeletionTask::kIdFieldName << migrationId);
    auto update = BSON("$unset" << BSON(RangeDeletionTask::kPendingFieldName << ""));

    store.update(opCtx, query, update);
}

void deleteMigrationCoordinatorDocumentLocally(OperationContext* opCtx, const UUID& migrationId) {
    PersistentTaskStore<MigrationCoordinatorDocument> store(
        opCtx, NamespaceString::kMigrationCoordinatorsNamespace);
    store.remove(opCtx,
                 QUERY(MigrationCoordinatorDocument::kIdFieldName << migrationId),
                 {1, WriteConcernOptions::SyncMode::UNSET, Seconds(0)});
}

void ensureChunkVersionIsGreaterThan(OperationContext* opCtx,
                                     const ChunkRange& range,
                                     const ChunkVersion& preMigrationChunkVersion) {
    ConfigsvrEnsureChunkVersionIsGreaterThan ensureChunkVersionIsGreaterThanRequest;
    ensureChunkVersionIsGreaterThanRequest.setDbName(NamespaceString::kAdminDb);
    ensureChunkVersionIsGreaterThanRequest.setMinKey(range.getMin());
    ensureChunkVersionIsGreaterThanRequest.setMaxKey(range.getMax());
    ensureChunkVersionIsGreaterThanRequest.setVersion(preMigrationChunkVersion);
    const auto ensureChunkVersionIsGreaterThanRequestBSON =
        ensureChunkVersionIsGreaterThanRequest.toBSON({});

    const auto term = repl::ReplicationCoordinator::get(opCtx)->getTerm();

    for (int attempts = 1;; attempts++) {
        try {
            auto newClient =
                opCtx->getServiceContext()->makeClient("ensureChunkVersionIsGreaterThan");
            {
                stdx::lock_guard<Client> lk(*newClient.get());
                newClient->setSystemOperationKillable(lk);
            }
            AlternativeClientRegion acr(newClient);
            auto newOpCtxPtr = cc().makeOperationContext();
            auto newOpCtx = newOpCtxPtr.get();

            hangInEnsureChunkVersionIsGreaterThanInterruptible.pauseWhileSet(newOpCtx);

            const auto ensureChunkVersionIsGreaterThanResponse =
                Grid::get(newOpCtx)
                    ->shardRegistry()
                    ->getConfigShard()
                    ->runCommandWithFixedRetryAttempts(
                        newOpCtx,
                        ReadPreferenceSetting{ReadPreference::PrimaryOnly},
                        "admin",
                        ensureChunkVersionIsGreaterThanRequestBSON,
                        Shard::RetryPolicy::kIdempotent);
            const auto ensureChunkVersionIsGreaterThanStatus =
                Shard::CommandResponse::getEffectiveStatus(ensureChunkVersionIsGreaterThanResponse);

            uassertStatusOK(ensureChunkVersionIsGreaterThanStatus);

            if (hangInEnsureChunkVersionIsGreaterThanThenSimulateErrorUninterruptible
                    .shouldFail()) {
                hangInEnsureChunkVersionIsGreaterThanThenSimulateErrorUninterruptible
                    .pauseWhileSet();
                uasserted(
                    ErrorCodes::InternalError,
                    "simulate an error response for _configsvrEnsureChunkVersionIsGreaterThan");
            }
            break;
        } catch (const DBException& ex) {
            // If the server is already doing a clean shutdown, join the shutdown.
            if (globalInShutdownDeprecated()) {
                shutdown(waitForShutdown());
            }

            // If this node has stepped down, stop retrying.
            uassert(
                ErrorCodes::InterruptedDueToReplStateChange,
                "Stepped down while trying to send ensureChunkVersionIsGreaterThan to recover a "
                "migration commit decision",
                repl::ReplicationCoordinator::get(opCtx)->getMemberState() ==
                        repl::MemberState::RS_PRIMARY &&
                    term == repl::ReplicationCoordinator::get(opCtx)->getTerm());

            LOGV2(22035,
                  "_configsvrEnsureChunkVersionIsGreaterThan failed after {attempts} attempts "
                  "{causedBy_ex_toStatus} . Will try again.",
                  "attempts"_attr = attempts,
                  "causedBy_ex_toStatus"_attr = causedBy(redact(ex.toStatus())));
        }
    }
}

void refreshFilteringMetadataUntilSuccess(OperationContext* opCtx, const NamespaceString& nss) {
    const auto term = repl::ReplicationCoordinator::get(opCtx)->getTerm();

    for (int attempts = 1;; attempts++) {
        try {
            auto newClient =
                opCtx->getServiceContext()->makeClient("refreshFilteringMetadataUntilSuccess");
            {
                stdx::lock_guard<Client> lk(*newClient.get());
                newClient->setSystemOperationKillable(lk);
            }
            AlternativeClientRegion acr(newClient);
            auto newOpCtxPtr = cc().makeOperationContext();
            auto newOpCtx = newOpCtxPtr.get();

            hangInRefreshFilteringMetadataUntilSuccessInterruptible.pauseWhileSet(newOpCtx);

            forceShardFilteringMetadataRefresh(newOpCtx, nss, true);

            if (hangInRefreshFilteringMetadataUntilSuccessThenSimulateErrorUninterruptible
                    .shouldFail()) {
                hangInRefreshFilteringMetadataUntilSuccessThenSimulateErrorUninterruptible
                    .pauseWhileSet();
                uasserted(ErrorCodes::InternalError,
                          "simulate an error response for forceShardFilteringMetadataRefresh");
            }
            break;
        } catch (const ExceptionFor<ErrorCodes::NamespaceNotFound>&) {
            // A filtering metadata refresh can throw NamespaceNotFound if the database was dropped
            // from the cluster.
            break;
        } catch (const DBException& ex) {
            // If the server is already doing a clean shutdown, join the shutdown.
            if (globalInShutdownDeprecated()) {
                shutdown(waitForShutdown());
            }

            // If this node has stepped down, stop retrying.
            uassert(ErrorCodes::InterruptedDueToReplStateChange,
                    "Stepped down while trying to force a filtering metadata refresh to recover a "
                    "migration commit decision",
                    repl::ReplicationCoordinator::get(opCtx)->getMemberState() ==
                            repl::MemberState::RS_PRIMARY &&
                        term == repl::ReplicationCoordinator::get(opCtx)->getTerm());

            LOGV2(22036,
                  "Failed to refresh metadata for {nss_ns} after {attempts} attempts "
                  "{causedBy_ex_toStatus}. Will try to refresh again.",
                  "nss_ns"_attr = nss.ns(),
                  "attempts"_attr = attempts,
                  "causedBy_ex_toStatus"_attr = causedBy(redact(ex.toStatus())));
        }
    }
}

void resumeMigrationCoordinationsOnStepUp(ServiceContext* serviceContext) {
    LOGV2(22037, "Starting migration coordinator stepup recovery thread.");

    auto executor = Grid::get(serviceContext)->getExecutorPool()->getFixedExecutor();
    ExecutorFuture<void>(executor)
        .then([serviceContext] {
            ThreadClient tc("MigrationCoordinatorStepupRecovery", serviceContext);
            {
                stdx::lock_guard<Client> lk(*tc.get());
                tc->setSystemOperationKillable(lk);
            }
            auto uniqueOpCtx = tc->makeOperationContext();
            auto opCtx = uniqueOpCtx.get();

            // Wait for the latest OpTime to be majority committed to ensure any decision that is
            // read is on the true branch of history.
            // Note (Esha): I don't think this is strictly required for correctness, but it is
            // difficult to reason about, and being pessimistic by waiting for the decision to be
            // majority committed does not cost much, since stepup should be rare. It *is* required
            // that this node ensure a decision that it itself recovers is majority committed. For
            // example, it is possible that this node is a stale primary, and the true primary has
            // already sent a *commit* decision and re-received a chunk containing the minKey of
            // this migration. In this case, this node would see that the minKey is still owned and
            // assume the migration *aborted*. If this node communicated the abort decision to the
            // recipient, the recipient (if it had not heard the decision yet) would delete data
            // that the recipient actually owns. (The recipient does not currently wait to hear the
            // range deletion decision for the first migration before being able to donate (any
            // part of) the chunk again.)
            auto& replClientInfo = repl::ReplClientInfo::forClient(opCtx->getClient());
            replClientInfo.setLastOpToSystemLastOpTime(opCtx);
            const auto lastOpTime = replClientInfo.getLastOp();
            LOGV2(22038,
                  "Waiting for OpTime {lastOpTime} to become majority committed",
                  "lastOpTime"_attr = lastOpTime);
            return WaitForMajorityService::get(serviceContext).waitUntilMajority(lastOpTime);
        })
        .thenRunOn(executor)
        .then([serviceContext]() {
            ThreadClient tc("MigrationCoordinatorStepupRecovery", serviceContext);
            {
                stdx::lock_guard<Client> lk(*tc.get());
                tc->setSystemOperationKillable(lk);
            }
            auto uniqueOpCtx = tc->makeOperationContext();
            auto opCtx = uniqueOpCtx.get();

            PersistentTaskStore<MigrationCoordinatorDocument> store(
                opCtx, NamespaceString::kMigrationCoordinatorsNamespace);
            Query query;
            store.forEach(opCtx, query, [&opCtx](const MigrationCoordinatorDocument& doc) {
                LOGV2(22039, "Recovering migration {doc}", "doc"_attr = doc.toBSON());

                // Create a MigrationCoordinator to complete the coordination.
                MigrationCoordinator coordinator(doc.getId(),
                                                 doc.getMigrationSessionId(),
                                                 doc.getLsid(),
                                                 doc.getDonorShardId(),
                                                 doc.getRecipientShardId(),
                                                 doc.getNss(),
                                                 doc.getCollectionUuid(),
                                                 doc.getRange(),
                                                 doc.getPreMigrationChunkVersion(),
                                                 false /* waitForDelete */);

                if (doc.getDecision()) {
                    // The decision is already known.
                    coordinator.setMigrationDecision(
                        (*doc.getDecision()) == DecisionEnum::kCommitted
                            ? MigrationCoordinator::Decision::kCommitted
                            : MigrationCoordinator::Decision::kAborted);
                    coordinator.completeMigration(opCtx);
                    return true;
                }

                // The decision is not known. Recover the decision from the config server.

                ensureChunkVersionIsGreaterThan(
                    opCtx, doc.getRange(), doc.getPreMigrationChunkVersion());

                hangBeforeFilteringMetadataRefresh.pauseWhileSet();

                refreshFilteringMetadataUntilSuccess(opCtx, doc.getNss());

                auto refreshedMetadata = [&] {
                    AutoGetCollection autoColl(opCtx, doc.getNss(), MODE_IS);
                    auto* const css = CollectionShardingRuntime::get(opCtx, doc.getNss());
                    return css->getCurrentMetadataIfKnown();
                }();

                if (!refreshedMetadata || !(*refreshedMetadata)->isSharded() ||
                    !(*refreshedMetadata)->uuidMatches(doc.getCollectionUuid())) {
                    LOGV2(
                        22040,
                        "Even after forced refresh, filtering metadata for namespace in "
                        "migration coordinator doc "
                        "{doc}{refreshedMetadata_refreshedMetadata_isSharded_is_not_known_has_UUID_"
                        "that_does_not_match_the_collection_UUID_in_the_coordinator_doc}. Deleting "
                        "the range deletion tasks on the donor and recipient as "
                        "well as the migration coordinator document on this node.",
                        "doc"_attr = doc.toBSON(),
                        "refreshedMetadata_refreshedMetadata_isSharded_is_not_known_has_UUID_that_does_not_match_the_collection_UUID_in_the_coordinator_doc"_attr =
                            (!refreshedMetadata || !(*refreshedMetadata)->isSharded()
                                 ? "is not known"
                                 : "has UUID that does not match the collection UUID in the "
                                   "coordinator doc"));

                    // TODO (SERVER-45707): Test that range deletion tasks are eventually
                    // deleted even if the collection is dropped before migration coordination
                    // is resumed.
                    deleteRangeDeletionTaskOnRecipient(
                        opCtx, doc.getRecipientShardId(), doc.getId());
                    deleteRangeDeletionTaskLocally(opCtx, doc.getId());
                    coordinator.forgetMigration(opCtx);
                    return true;
                }

                if ((*refreshedMetadata)->keyBelongsToMe(doc.getRange().getMin())) {
                    coordinator.setMigrationDecision(MigrationCoordinator::Decision::kAborted);
                } else {
                    coordinator.setMigrationDecision(MigrationCoordinator::Decision::kCommitted);
                }
                coordinator.completeMigration(opCtx);
                return true;
            });
        })
        .getAsync([](const Status& status) {
            if (!status.isOK()) {
                LOGV2(22041,
                      "Failed to resume coordinating migrations on stepup {causedBy_status}",
                      "causedBy_status"_attr = causedBy(status));
            }
        });
}

}  // namespace migrationutil
}  // namespace mongo