summaryrefslogtreecommitdiff
path: root/src/mongo/db/session.cpp
blob: 60ea0762b9918feb32fcd06a0ba33f71ebe56a8d (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
/**
 *    Copyright (C) 2017 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::kStorage

#include "mongo/platform/basic.h"

#include "mongo/db/session.h"

#include "mongo/db/catalog/index_catalog.h"
#include "mongo/db/concurrency/lock_state.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbdirectclient.h"
#include "mongo/db/index/index_access_method.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/ops/update.h"
#include "mongo/db/query/get_executor.h"
#include "mongo/db/repl/read_concern_args.h"
#include "mongo/db/retryable_writes_stats.h"
#include "mongo/db/transaction_history_iterator.h"
#include "mongo/stdx/memory.h"
#include "mongo/transport/transport_layer.h"
#include "mongo/util/fail_point_service.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"

namespace mongo {
namespace {

void fassertOnRepeatedExecution(OperationContext* opCtx,
                                const LogicalSessionId& lsid,
                                TxnNumber txnNumber,
                                StmtId stmtId,
                                const repl::OpTime& firstOpTime,
                                const repl::OpTime& secondOpTime) {
    severe() << "Statement id " << stmtId << " from transaction [ " << lsid.toBSON() << ":"
             << txnNumber << " ] was committed once with opTime " << firstOpTime
             << " and a second time with opTime " << secondOpTime
             << ". This indicates possible data corruption or server bug and the process will be "
                "terminated.";
    fassertFailed(40526);
}

struct ActiveTransactionHistory {
    boost::optional<SessionTxnRecord> lastTxnRecord;
    Session::CommittedStatementTimestampMap committedStatements;
    bool hasIncompleteHistory{false};
};

ActiveTransactionHistory fetchActiveTransactionHistory(OperationContext* opCtx,
                                                       const LogicalSessionId& lsid) {
    ActiveTransactionHistory result;

    result.lastTxnRecord = [&]() -> boost::optional<SessionTxnRecord> {
        DBDirectClient client(opCtx);
        auto result =
            client.findOne(NamespaceString::kSessionTransactionsTableNamespace.ns(),
                           {BSON(SessionTxnRecord::kSessionIdFieldName << lsid.toBSON())});
        if (result.isEmpty()) {
            return boost::none;
        }

        return SessionTxnRecord::parse(IDLParserErrorContext("parse latest txn record for session"),
                                       result);
    }();

    if (!result.lastTxnRecord) {
        return result;
    }

    auto it = TransactionHistoryIterator(result.lastTxnRecord->getLastWriteOpTime());
    while (it.hasNext()) {
        try {
            const auto entry = it.next(opCtx);
            invariant(entry.getStatementId());

            if (*entry.getStatementId() == kIncompleteHistoryStmtId) {
                // Only the dead end sentinel can have this id for oplog write history
                invariant(entry.getObject2());
                invariant(entry.getObject2()->woCompare(Session::kDeadEndSentinel) == 0);
                result.hasIncompleteHistory = true;
                continue;
            }

            const auto insertRes =
                result.committedStatements.emplace(*entry.getStatementId(), entry.getOpTime());
            if (!insertRes.second) {
                const auto& existingOpTime = insertRes.first->second;
                fassertOnRepeatedExecution(opCtx,
                                           lsid,
                                           result.lastTxnRecord->getTxnNum(),
                                           *entry.getStatementId(),
                                           existingOpTime,
                                           entry.getOpTime());
            }
        } catch (const DBException& ex) {
            if (ex.code() == ErrorCodes::IncompleteTransactionHistory) {
                result.hasIncompleteHistory = true;
                break;
            }

            throw;
        }
    }

    return result;
}

void updateSessionEntry(OperationContext* opCtx, const UpdateRequest& updateRequest) {
    // Current code only supports replacement update.
    dassert(UpdateDriver::isDocReplacement(updateRequest.getUpdates()));

    AutoGetCollection autoColl(opCtx, NamespaceString::kSessionTransactionsTableNamespace, MODE_IX);

    uassert(40527,
            str::stream() << "Unable to persist transaction state because the session transaction "
                             "collection is missing. This indicates that the "
                          << NamespaceString::kSessionTransactionsTableNamespace.ns()
                          << " collection has been manually deleted.",
            autoColl.getCollection());

    WriteUnitOfWork wuow(opCtx);

    auto collection = autoColl.getCollection();
    auto idIndex = collection->getIndexCatalog()->findIdIndex(opCtx);

    uassert(40672,
            str::stream() << "Failed to fetch _id index for "
                          << NamespaceString::kSessionTransactionsTableNamespace.ns(),
            idIndex);

    auto indexAccess = collection->getIndexCatalog()->getIndex(idIndex);
    // Since we are looking up a key inside the _id index, create a key object consisting of only
    // the _id field.
    auto idToFetch = updateRequest.getQuery().firstElement();
    auto toUpdateIdDoc = idToFetch.wrap();
    dassert(idToFetch.fieldNameStringData() == "_id"_sd);
    auto recordId = indexAccess->findSingle(opCtx, toUpdateIdDoc);
    auto startingSnapshotId = opCtx->recoveryUnit()->getSnapshotId();

    if (recordId.isNull()) {
        // Upsert case.
        auto status = collection->insertDocument(
            opCtx, InsertStatement(updateRequest.getUpdates()), nullptr, true, false);

        if (status == ErrorCodes::DuplicateKey) {
            throw WriteConflictException();
        }

        uassertStatusOK(status);
        wuow.commit();
        return;
    }

    auto originalRecordData = collection->getRecordStore()->dataFor(opCtx, recordId);
    auto originalDoc = originalRecordData.toBson();

    invariant(collection->getDefaultCollator() == nullptr);
    boost::intrusive_ptr<ExpressionContext> expCtx(new ExpressionContext(opCtx, nullptr));

    auto matcher = fassertStatusOK(
        40673, MatchExpressionParser::parse(updateRequest.getQuery(), std::move(expCtx)));
    if (!matcher->matchesBSON(originalDoc)) {
        // Document no longer match what we expect so throw WCE to make the caller re-examine.
        throw WriteConflictException();
    }

    OplogUpdateEntryArgs args;
    args.nss = NamespaceString::kSessionTransactionsTableNamespace;
    args.uuid = collection->uuid();
    args.update = updateRequest.getUpdates();
    args.criteria = toUpdateIdDoc;
    args.fromMigrate = false;

    collection->updateDocument(opCtx,
                               recordId,
                               Snapshotted<BSONObj>(startingSnapshotId, originalDoc),
                               updateRequest.getUpdates(),
                               true,   // enforceQuota
                               false,  // indexesAffected = false because _id is the only index
                               nullptr,
                               &args);

    wuow.commit();
}

// Failpoint which allows different failure actions to happen after each write. Supports the
// parameters below, which can be combined with each other (unless explicitly disallowed):
//
// closeConnection (bool, default = true): Closes the connection on which the write was executed.
// failBeforeCommitExceptionCode (int, default = not specified): If set, the specified exception
//      code will be thrown, which will cause the write to not commit; if not specified, the write
//      will be allowed to commit.
MONGO_FP_DECLARE(onPrimaryTransactionalWrite);

}  // namespace

const BSONObj Session::kDeadEndSentinel(BSON("$incompleteOplogHistory" << 1));

Session::Session(LogicalSessionId sessionId) : _sessionId(std::move(sessionId)) {}

void Session::refreshFromStorageIfNeeded(OperationContext* opCtx) {
    invariant(!opCtx->lockState()->isLocked());
    invariant(repl::ReadConcernArgs::get(opCtx).getLevel() ==
              repl::ReadConcernLevel::kLocalReadConcern);

    stdx::unique_lock<stdx::mutex> ul(_mutex);

    while (!_isValid) {
        const int numInvalidations = _numInvalidations;

        ul.unlock();

        auto activeTxnHistory = fetchActiveTransactionHistory(opCtx, _sessionId);

        ul.lock();

        // Protect against concurrent refreshes or invalidations
        if (!_isValid && _numInvalidations == numInvalidations) {
            _isValid = true;
            _lastWrittenSessionRecord = std::move(activeTxnHistory.lastTxnRecord);

            if (_lastWrittenSessionRecord) {
                _activeTxnNumber = _lastWrittenSessionRecord->getTxnNum();
                _activeTxnCommittedStatements = std::move(activeTxnHistory.committedStatements);
                _hasIncompleteHistory = activeTxnHistory.hasIncompleteHistory;
            }

            break;
        }
    }
}

void Session::beginOrContinueTxn(OperationContext* opCtx,
                                 TxnNumber txnNumber,
                                 boost::optional<bool> autocommit) {
    invariant(!opCtx->lockState()->isLocked());

    stdx::lock_guard<stdx::mutex> lg(_mutex);
    _beginOrContinueTxn(lg, txnNumber, autocommit);
}

void Session::beginOrContinueTxnOnMigration(OperationContext* opCtx, TxnNumber txnNumber) {
    invariant(!opCtx->lockState()->isLocked());

    stdx::lock_guard<stdx::mutex> lg(_mutex);
    _beginOrContinueTxnOnMigration(lg, txnNumber);
}


void Session::onWriteOpCompletedOnPrimary(OperationContext* opCtx,
                                          TxnNumber txnNumber,
                                          std::vector<StmtId> stmtIdsWritten,
                                          const repl::OpTime& lastStmtIdWriteOpTime,
                                          Date_t lastStmtIdWriteDate) {
    invariant(opCtx->lockState()->inAWriteUnitOfWork());

    stdx::unique_lock<stdx::mutex> ul(_mutex);

    // Sanity check that we don't double-execute statements
    for (const auto stmtId : stmtIdsWritten) {
        const auto stmtOpTime = _checkStatementExecuted(ul, txnNumber, stmtId);
        if (stmtOpTime) {
            fassertOnRepeatedExecution(
                opCtx, _sessionId, txnNumber, stmtId, *stmtOpTime, lastStmtIdWriteOpTime);
        }
    }

    const auto updateRequest =
        _makeUpdateRequest(ul, txnNumber, lastStmtIdWriteOpTime, lastStmtIdWriteDate);

    ul.unlock();

    repl::UnreplicatedWritesBlock doNotReplicateWrites(opCtx);

    updateSessionEntry(opCtx, updateRequest);
    _registerUpdateCacheOnCommit(
        opCtx, txnNumber, std::move(stmtIdsWritten), lastStmtIdWriteOpTime);
}

bool Session::onMigrateBeginOnPrimary(OperationContext* opCtx, TxnNumber txnNumber, StmtId stmtId) {
    beginOrContinueTxnOnMigration(opCtx, txnNumber);

    try {
        if (checkStatementExecuted(opCtx, txnNumber, stmtId)) {
            return false;
        }
    } catch (const DBException& ex) {
        // If the transaction chain was truncated on the recipient shard, then we
        // are most likely copying from a session that hasn't been touched on the
        // recipient shard for a very long time but could be recent on the donor.
        // We continue copying regardless to get the entire transaction from the donor.
        if (ex.code() != ErrorCodes::IncompleteTransactionHistory) {
            throw;
        }
        if (stmtId == kIncompleteHistoryStmtId) {
            return false;
        }
    }

    return true;
}

void Session::onMigrateCompletedOnPrimary(OperationContext* opCtx,
                                          TxnNumber txnNumber,
                                          std::vector<StmtId> stmtIdsWritten,
                                          const repl::OpTime& lastStmtIdWriteOpTime,
                                          Date_t lastStmtIdWriteDate) {
    invariant(opCtx->lockState()->inAWriteUnitOfWork());

    stdx::unique_lock<stdx::mutex> ul(_mutex);

    _checkValid(ul);
    _checkIsActiveTransaction(ul, txnNumber);

    const auto updateRequest =
        _makeUpdateRequest(ul, txnNumber, lastStmtIdWriteOpTime, lastStmtIdWriteDate);

    ul.unlock();

    repl::UnreplicatedWritesBlock doNotReplicateWrites(opCtx);

    updateSessionEntry(opCtx, updateRequest);
    _registerUpdateCacheOnCommit(
        opCtx, txnNumber, std::move(stmtIdsWritten), lastStmtIdWriteOpTime);
}

void Session::updateSessionRecordOnSecondary(OperationContext* opCtx,
                                             const SessionTxnRecord& sessionTxnRecord) {
    invariant(!opCtx->lockState()->isLocked());

    writeConflictRetry(
        opCtx, "Update session txn", NamespaceString::kSessionTransactionsTableNamespace.ns(), [&] {
            UpdateRequest updateRequest(NamespaceString::kSessionTransactionsTableNamespace);
            updateRequest.setQuery(BSON(SessionTxnRecord::kSessionIdFieldName
                                        << sessionTxnRecord.getSessionId().toBSON()));
            updateRequest.setUpdates(sessionTxnRecord.toBSON());
            updateRequest.setUpsert(true);

            repl::UnreplicatedWritesBlock doNotReplicateWrites(opCtx);

            Lock::DBLock configDBLock(opCtx, NamespaceString::kConfigDb, MODE_IX);
            WriteUnitOfWork wuow(opCtx);
            updateSessionEntry(opCtx, updateRequest);
            wuow.commit();
        });
}

void Session::invalidate() {
    stdx::lock_guard<stdx::mutex> lg(_mutex);
    _isValid = false;
    _numInvalidations++;

    _lastWrittenSessionRecord.reset();

    _activeTxnNumber = kUninitializedTxnNumber;
    _activeTxnCommittedStatements.clear();
    _hasIncompleteHistory = false;
}

repl::OpTime Session::getLastWriteOpTime(TxnNumber txnNumber) const {
    stdx::lock_guard<stdx::mutex> lg(_mutex);
    _checkValid(lg);
    _checkIsActiveTransaction(lg, txnNumber);

    if (!_lastWrittenSessionRecord || _lastWrittenSessionRecord->getTxnNum() != txnNumber)
        return {};

    return _lastWrittenSessionRecord->getLastWriteOpTime();
}

boost::optional<repl::OplogEntry> Session::checkStatementExecuted(OperationContext* opCtx,
                                                                  TxnNumber txnNumber,
                                                                  StmtId stmtId) const {
    const auto stmtTimestamp = [&] {
        stdx::lock_guard<stdx::mutex> lg(_mutex);
        return _checkStatementExecuted(lg, txnNumber, stmtId);
    }();

    if (!stmtTimestamp)
        return boost::none;

    TransactionHistoryIterator txnIter(*stmtTimestamp);
    while (txnIter.hasNext()) {
        const auto entry = txnIter.next(opCtx);
        invariant(entry.getStatementId());
        if (*entry.getStatementId() == stmtId)
            return entry;
    }

    MONGO_UNREACHABLE;
}

bool Session::checkStatementExecutedNoOplogEntryFetch(TxnNumber txnNumber, StmtId stmtId) const {
    stdx::lock_guard<stdx::mutex> lg(_mutex);
    return bool(_checkStatementExecuted(lg, txnNumber, stmtId));
}

void Session::_beginOrContinueTxn(WithLock wl,
                                  TxnNumber txnNumber,
                                  boost::optional<bool> autocommit) {
    _checkValid(wl);
    _checkTxnValid(wl, txnNumber);

    if (txnNumber == _activeTxnNumber) {
        // Continuing an existing transaction.
        uassert(ErrorCodes::IllegalOperation,
                "Specifying 'autocommit' is only allowed at the beginning of a transaction",
                autocommit == boost::none);

        return;
    }

    // Start a new transaction with an autocommit field
    _setActiveTxn(wl, txnNumber);
    _autocommit = (autocommit != boost::none) ? *autocommit : true;  // autocommit defaults to true
    _isSnapshotTxn = false;
}

void Session::_checkTxnValid(WithLock, TxnNumber txnNumber) const {
    uassert(ErrorCodes::TransactionTooOld,
            str::stream() << "Cannot start transaction " << txnNumber << " on session "
                          << getSessionId()
                          << " because a newer transaction "
                          << _activeTxnNumber
                          << " has already started.",
            txnNumber >= _activeTxnNumber);
}

void Session::stashTransactionResources(OperationContext* opCtx) {
    stdx::lock_guard<stdx::mutex> lg(_mutex);
    if (!_isSnapshotTxn) {
        return;
    }

    invariant(opCtx->hasStashedCursor());

    if (*opCtx->getTxnNumber() != _activeTxnNumber) {
        // The session is checked out, so _activeTxnNumber cannot advance due to a user operation.
        // However, when a chunk is migrated, session and transaction information is copied from the
        // donor shard to the recipient. This occurs outside of the check-out mechanism and can lead
        // to a higher _activeTxnNumber during the lifetime of a checkout. If that occurs, we abort
        // the current transaction. Note that it would indicate a user bug to have a newer
        // transaction on one shard while an older transaction is still active on another shard.
        uasserted(ErrorCodes::TransactionAborted,
                  str::stream() << "Transaction aborted. Active txnNumber is now "
                                << _activeTxnNumber);
    }

    opCtx->getWriteUnitOfWork()->release();
    opCtx->setWriteUnitOfWork(nullptr);

    _stashedLocker = opCtx->releaseLockState();
    _stashedRecoveryUnit.reset(opCtx->releaseRecoveryUnit());

    opCtx->setLockState(stdx::make_unique<DefaultLockerImpl>());
    opCtx->setRecoveryUnit(opCtx->getServiceContext()->getGlobalStorageEngine()->newRecoveryUnit(),
                           OperationContext::kNotInUnitOfWork);
}

void Session::unstashTransactionResources(OperationContext* opCtx) {
    stdx::lock_guard<stdx::mutex> lg(_mutex);
    if (opCtx->getTxnNumber() < _activeTxnNumber) {
        // The session is checked out, so _activeTxnNumber cannot advance due to a user operation.
        // However, when a chunk is migrated, session and transaction information is copied from the
        // donor shard to the recipient. This occurs outside of the check-out mechanism and can lead
        // to a higher _activeTxnNumber during the lifetime of a checkout. If that occurs, we abort
        // the current transaction. Note that it would indicate a user bug to have a newer
        // transaction on one shard while an older transaction is still active on another shard.
        _releaseStashedTransactionResources(lg, opCtx);
        uasserted(ErrorCodes::TransactionAborted,
                  str::stream() << "Transaction aborted. Active txnNumber is now "
                                << _activeTxnNumber);
        return;
    }

    if (_stashedLocker) {
        invariant(_stashedRecoveryUnit);
        opCtx->releaseLockState();
        opCtx->setLockState(std::move(_stashedLocker));
        opCtx->setRecoveryUnit(_stashedRecoveryUnit.release(),
                               OperationContext::RecoveryUnitState::kNotInUnitOfWork);
        opCtx->setWriteUnitOfWork(WriteUnitOfWork::createForSnapshotResume(opCtx));
    } else {
        auto readConcernArgs = repl::ReadConcernArgs::get(opCtx);
        if (readConcernArgs.getLevel() == repl::ReadConcernLevel::kSnapshotReadConcern) {
            opCtx->setWriteUnitOfWork(std::make_unique<WriteUnitOfWork>(opCtx));
            _isSnapshotTxn = true;
        }
    }
}

void Session::_releaseStashedTransactionResources(WithLock wl, OperationContext* opCtx) {
    if (opCtx->getWriteUnitOfWork()) {
        opCtx->setWriteUnitOfWork(nullptr);
    }

    _stashedRecoveryUnit.reset(nullptr);
    _stashedLocker.reset(nullptr);
    _isSnapshotTxn = false;
}

void Session::_beginOrContinueTxnOnMigration(WithLock wl, TxnNumber txnNumber) {
    _checkValid(wl);
    _checkTxnValid(wl, txnNumber);

    // Check for continuing an existing transaction
    if (txnNumber == _activeTxnNumber)
        return;

    _setActiveTxn(wl, txnNumber);
}

void Session::_setActiveTxn(WithLock, TxnNumber txnNumber) {
    _activeTxnNumber = txnNumber;
    _activeTxnCommittedStatements.clear();
    _hasIncompleteHistory = false;
}

void Session::_checkValid(WithLock) const {
    uassert(ErrorCodes::ConflictingOperationInProgress,
            str::stream() << "Session " << getSessionId()
                          << " was concurrently modified and the operation must be retried.",
            _isValid);
}

void Session::_checkIsActiveTransaction(WithLock, TxnNumber txnNumber) const {
    uassert(ErrorCodes::ConflictingOperationInProgress,
            str::stream() << "Cannot perform retryability check for transaction " << txnNumber
                          << " on session "
                          << getSessionId()
                          << " because a different transaction "
                          << _activeTxnNumber
                          << " is now active.",
            txnNumber == _activeTxnNumber);
}

boost::optional<repl::OpTime> Session::_checkStatementExecuted(WithLock wl,
                                                               TxnNumber txnNumber,
                                                               StmtId stmtId) const {
    _checkValid(wl);
    _checkIsActiveTransaction(wl, txnNumber);

    const auto it = _activeTxnCommittedStatements.find(stmtId);
    if (it == _activeTxnCommittedStatements.end()) {
        uassert(ErrorCodes::IncompleteTransactionHistory,
                str::stream() << "Incomplete history detected for transaction " << txnNumber
                              << " on session "
                              << _sessionId.toBSON(),
                !_hasIncompleteHistory);

        return boost::none;
    }

    invariant(_lastWrittenSessionRecord);
    invariant(_lastWrittenSessionRecord->getTxnNum() == txnNumber);

    return it->second;
}

UpdateRequest Session::_makeUpdateRequest(WithLock,
                                          TxnNumber newTxnNumber,
                                          const repl::OpTime& newLastWriteOpTime,
                                          Date_t newLastWriteDate) const {
    UpdateRequest updateRequest(NamespaceString::kSessionTransactionsTableNamespace);

    const auto updateBSON = [&] {
        SessionTxnRecord newTxnRecord;
        newTxnRecord.setSessionId(_sessionId);
        newTxnRecord.setTxnNum(newTxnNumber);
        newTxnRecord.setLastWriteOpTime(newLastWriteOpTime);
        newTxnRecord.setLastWriteDate(newLastWriteDate);
        return newTxnRecord.toBSON();
    }();
    updateRequest.setUpdates(updateBSON);

    if (_lastWrittenSessionRecord) {
        updateRequest.setQuery(BSON(SessionTxnRecord::kSessionIdFieldName
                                    << _sessionId.toBSON()
                                    << SessionTxnRecord::kTxnNumFieldName
                                    << _lastWrittenSessionRecord->getTxnNum()
                                    << SessionTxnRecord::kLastWriteOpTimeFieldName
                                    << _lastWrittenSessionRecord->getLastWriteOpTime()));
    } else {
        updateRequest.setQuery(updateBSON);
        updateRequest.setUpsert(true);
    }

    return updateRequest;
}

void Session::_registerUpdateCacheOnCommit(OperationContext* opCtx,
                                           TxnNumber newTxnNumber,
                                           std::vector<StmtId> stmtIdsWritten,
                                           const repl::OpTime& lastStmtIdWriteOpTime) {
    opCtx->recoveryUnit()->onCommit([
        this,
        opCtx,
        newTxnNumber,
        stmtIdsWritten = std::move(stmtIdsWritten),
        lastStmtIdWriteOpTime
    ] {
        RetryableWritesStats::get(opCtx)->incrementTransactionsCollectionWriteCount();

        stdx::lock_guard<stdx::mutex> lg(_mutex);

        if (!_isValid)
            return;

        // The cache of the last written record must always be advanced after a write so that
        // subsequent writes have the correct point to start from.
        if (!_lastWrittenSessionRecord) {
            _lastWrittenSessionRecord.emplace();

            _lastWrittenSessionRecord->setSessionId(_sessionId);
            _lastWrittenSessionRecord->setTxnNum(newTxnNumber);
            _lastWrittenSessionRecord->setLastWriteOpTime(lastStmtIdWriteOpTime);
        } else {
            if (newTxnNumber > _lastWrittenSessionRecord->getTxnNum())
                _lastWrittenSessionRecord->setTxnNum(newTxnNumber);

            if (lastStmtIdWriteOpTime > _lastWrittenSessionRecord->getLastWriteOpTime())
                _lastWrittenSessionRecord->setLastWriteOpTime(lastStmtIdWriteOpTime);
        }

        if (newTxnNumber > _activeTxnNumber) {
            // This call is necessary in order to advance the txn number and reset the cached state
            // in the case where just before the storage transaction commits, the cache entry gets
            // invalidated and immediately refreshed while there were no writes for newTxnNumber
            // yet. In this case _activeTxnNumber will be less than newTxnNumber and we will fail to
            // update the cache even though the write was successful.
            _beginOrContinueTxn(lg, newTxnNumber, boost::none);
        }

        if (newTxnNumber == _activeTxnNumber) {
            for (const auto stmtId : stmtIdsWritten) {
                if (stmtId == kIncompleteHistoryStmtId) {
                    _hasIncompleteHistory = true;
                    continue;
                }

                const auto insertRes =
                    _activeTxnCommittedStatements.emplace(stmtId, lastStmtIdWriteOpTime);
                if (!insertRes.second) {
                    const auto& existingOpTime = insertRes.first->second;
                    fassertOnRepeatedExecution(opCtx,
                                               _sessionId,
                                               newTxnNumber,
                                               stmtId,
                                               existingOpTime,
                                               lastStmtIdWriteOpTime);
                }
            }
        }
    });

    MONGO_FAIL_POINT_BLOCK(onPrimaryTransactionalWrite, customArgs) {
        const auto& data = customArgs.getData();

        const auto closeConnectionElem = data["closeConnection"];
        if (closeConnectionElem.eoo() || closeConnectionElem.Bool()) {
            opCtx->getClient()->session()->end();
        }

        const auto failBeforeCommitExceptionElem = data["failBeforeCommitExceptionCode"];
        if (!failBeforeCommitExceptionElem.eoo()) {
            const auto failureCode = ErrorCodes::Error(int(failBeforeCommitExceptionElem.Number()));
            uasserted(failureCode,
                      str::stream() << "Failing write for " << _sessionId << ":" << newTxnNumber
                                    << " due to failpoint. The write must not be reflected.");
        }
    }
}

}  // namespace mongo