summaryrefslogtreecommitdiff
path: root/src/mongo/db/repl/replication_coordinator_external_state_impl.cpp
blob: f140580fbcc9549dbff236824cc555a5fd86e30a (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
/**
 *    Copyright (C) 2014 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/replication_coordinator_external_state_impl.h"

#include <string>

#include "mongo/base/init.h"
#include "mongo/base/status_with.h"
#include "mongo/bson/oid.h"
#include "mongo/db/catalog/database.h"
#include "mongo/db/catalog/database_holder.h"
#include "mongo/db/client.h"
#include "mongo/db/concurrency/d_concurrency.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/dbhelpers.h"
#include "mongo/db/jsobj.h"
#include "mongo/db/op_observer.h"
#include "mongo/db/repl/bgsync.h"
#include "mongo/db/repl/isself.h"
#include "mongo/db/repl/last_vote.h"
#include "mongo/db/repl/master_slave.h"
#include "mongo/db/repl/member_state.h"
#include "mongo/db/repl/oplog.h"
#include "mongo/db/repl/oplog_buffer_blocking_queue.h"
#include "mongo/db/repl/oplog_buffer_collection.h"
#include "mongo/db/repl/repl_settings.h"
#include "mongo/db/repl/replication_coordinator_global.h"
#include "mongo/db/repl/rs_initialsync.h"
#include "mongo/db/repl/rs_sync.h"
#include "mongo/db/repl/snapshot_thread.h"
#include "mongo/db/repl/storage_interface.h"
#include "mongo/db/repl/sync_tail.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/db/s/sharding_state_recovery.h"
#include "mongo/db/server_parameters.h"
#include "mongo/db/service_context.h"
#include "mongo/db/storage/storage_engine.h"
#include "mongo/executor/network_interface.h"
#include "mongo/executor/network_interface_factory.h"
#include "mongo/executor/thread_pool_task_executor.h"
#include "mongo/s/balancer/balancer.h"
#include "mongo/s/catalog/sharding_catalog_manager.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/cluster_identity_loader.h"
#include "mongo/s/grid.h"
#include "mongo/stdx/functional.h"
#include "mongo/stdx/memory.h"
#include "mongo/stdx/thread.h"
#include "mongo/transport/session.h"
#include "mongo/transport/transport_layer.h"
#include "mongo/util/assert_util.h"
#include "mongo/util/concurrency/thread_pool.h"
#include "mongo/util/exit.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
#include "mongo/util/net/hostandport.h"
#include "mongo/util/net/listen.h"

namespace mongo {
namespace repl {

namespace {
const char configCollectionName[] = "local.system.replset";
const char configDatabaseName[] = "local";
const char lastVoteCollectionName[] = "local.replset.election";
const char lastVoteDatabaseName[] = "local";
const char meCollectionName[] = "local.me";
const char meDatabaseName[] = "local";
const char tsFieldName[] = "ts";

const char kCollectionOplogBufferName[] = "collection";
const char kBlockingQueueOplogBufferName[] = "inMemoryBlockingQueue";

// Set this to true to force background creation of snapshots even if --enableMajorityReadConcern
// isn't specified. This can be used for A-B benchmarking to find how much overhead
// repl::SnapshotThread introduces.
MONGO_EXPORT_STARTUP_SERVER_PARAMETER(enableReplSnapshotThread, bool, false);

MONGO_EXPORT_STARTUP_SERVER_PARAMETER(use3dot2InitialSync, bool, true);

// Set this to specify whether to use a collection to buffer the oplog on the destination server
// during initial sync to prevent rolling over the oplog.
MONGO_EXPORT_STARTUP_SERVER_PARAMETER(initialSyncOplogBuffer,
                                      std::string,
                                      kBlockingQueueOplogBufferName);

MONGO_INITIALIZER(initialSyncOplogBuffer)(InitializerContext*) {
    if ((initialSyncOplogBuffer != kCollectionOplogBufferName) &&
        (initialSyncOplogBuffer != kBlockingQueueOplogBufferName)) {
        return Status(ErrorCodes::BadValue,
                      "unsupported initial sync oplog buffer option: " + initialSyncOplogBuffer);
    }
    if (use3dot2InitialSync && (initialSyncOplogBuffer == kCollectionOplogBufferName)) {
        return Status(ErrorCodes::BadValue,
                      "cannot use collection oplog buffer without --setParameter "
                      "use3dot2InitialSync=false");
    }

    return Status::OK();
}

/**
 * Returns new thread pool for thread pool task executor.
 */
std::unique_ptr<ThreadPool> makeThreadPool() {
    ThreadPool::Options threadPoolOptions;
    threadPoolOptions.poolName = "replication";
    threadPoolOptions.onCreateThread = [](const std::string& threadName) {
        Client::initThread(threadName.c_str());
    };
    return stdx::make_unique<ThreadPool>(threadPoolOptions);
}

}  // namespace

ReplicationCoordinatorExternalStateImpl::ReplicationCoordinatorExternalStateImpl(
    StorageInterface* storageInterface)
    : _storageInterface(storageInterface) {
    uassert(ErrorCodes::BadValue, "A StorageInterface is required.", _storageInterface);
}
ReplicationCoordinatorExternalStateImpl::~ReplicationCoordinatorExternalStateImpl() {}

bool ReplicationCoordinatorExternalStateImpl::isInitialSyncFlagSet(OperationContext* txn) {
    return _storageInterface->getInitialSyncFlag(txn);
}

void ReplicationCoordinatorExternalStateImpl::startInitialSync(OnInitialSyncFinishedFn finished) {
    _initialSyncThread.reset(new stdx::thread{[finished, this]() {
        Client::initThreadIfNotAlready("initial sync");
        // Do initial sync.
        syncDoInitialSync(this);
        finished();
    }});
}

void ReplicationCoordinatorExternalStateImpl::runOnInitialSyncThread(
    stdx::function<void(OperationContext* txn)> run) {
    _initialSyncThread.reset(new stdx::thread{[run, this]() {
        Client::initThreadIfNotAlready("initial sync");
        auto txn = cc().makeOperationContext();
        invariant(txn);
        invariant(txn->getClient());
        run(txn.get());
    }});
}

void ReplicationCoordinatorExternalStateImpl::startSteadyStateReplication(OperationContext* txn) {
    invariant(!_bgSync);
    log() << "Starting replication fetcher thread";
    _bgSync = stdx::make_unique<BackgroundSync>(this, makeSteadyStateOplogBuffer(txn));
    _bgSync->startup(txn);

    log() << "Starting replication applier threads";
    invariant(!_applierThread);
    _applierThread.reset(new stdx::thread(stdx::bind(&runSyncThread, _bgSync.get())));
    log() << "Starting replication reporter thread";
    invariant(!_syncSourceFeedbackThread);
    _syncSourceFeedbackThread.reset(new stdx::thread(stdx::bind(
        &SyncSourceFeedback::run, &_syncSourceFeedback, _taskExecutor.get(), _bgSync.get())));
}

void ReplicationCoordinatorExternalStateImpl::startThreads(const ReplSettings& settings) {
    stdx::lock_guard<stdx::mutex> lk(_threadMutex);
    if (_startedThreads) {
        return;
    }
    log() << "Starting replication storage threads";
    if (settings.isMajorityReadConcernEnabled() || enableReplSnapshotThread) {
        _snapshotThread = SnapshotThread::start(getGlobalServiceContext());
    }
    getGlobalServiceContext()->getGlobalStorageEngine()->setJournalListener(this);

    _taskExecutor = stdx::make_unique<executor::ThreadPoolTaskExecutor>(
        makeThreadPool(), executor::makeNetworkInterface("NetworkInterfaceASIO-RS"));
    _taskExecutor->startup();

    _writerPool = SyncTail::makeWriterPool();

    _storageInterface->startup();

    _startedThreads = true;
}

void ReplicationCoordinatorExternalStateImpl::startMasterSlave(OperationContext* txn) {
    repl::startMasterSlave(txn);
}

void ReplicationCoordinatorExternalStateImpl::shutdown(OperationContext* txn) {
    stdx::lock_guard<stdx::mutex> lk(_threadMutex);
    if (_startedThreads) {
        log() << "Stopping replication applier threads";
        if (_syncSourceFeedbackThread) {
            _syncSourceFeedback.shutdown();
            _syncSourceFeedbackThread->join();
        }
        if (_applierThread) {
            _applierThread->join();
        }

        if (_bgSync) {
            _bgSync->shutdown(txn);
            _bgSync->join(txn);
        }
        if (_snapshotThread)
            _snapshotThread->shutdown();

        _taskExecutor->shutdown();
        _taskExecutor->join();
        _storageInterface->shutdown();
    }
}

executor::TaskExecutor* ReplicationCoordinatorExternalStateImpl::getTaskExecutor() const {
    return _taskExecutor.get();
}

Status ReplicationCoordinatorExternalStateImpl::initializeReplSetStorage(OperationContext* txn,
                                                                         const BSONObj& config) {
    try {
        createOplog(txn);

        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            ScopedTransaction scopedXact(txn, MODE_X);
            Lock::GlobalWrite globalWrite(txn->lockState());

            WriteUnitOfWork wuow(txn);
            Helpers::putSingleton(txn, configCollectionName, config);
            const auto msgObj = BSON("msg"
                                     << "initiating set");
            getGlobalServiceContext()->getOpObserver()->onOpMessage(txn, msgObj);
            wuow.commit();
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(txn, "initiate oplog entry", "local.oplog.rs");
    } catch (const DBException& ex) {
        return ex.toStatus();
    }
    return Status::OK();
}

void ReplicationCoordinatorExternalStateImpl::logTransitionToPrimaryToOplog(OperationContext* txn) {
    MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
        ScopedTransaction scopedXact(txn, MODE_X);

        WriteUnitOfWork wuow(txn);
        txn->getClient()->getServiceContext()->getOpObserver()->onOpMessage(txn,
                                                                            BSON("msg"
                                                                                 << "new primary"));
        wuow.commit();
    }
    MONGO_WRITE_CONFLICT_RETRY_LOOP_END(
        txn, "logging transition to primary to oplog", "local.oplog.rs");
}

void ReplicationCoordinatorExternalStateImpl::forwardSlaveProgress() {
    _syncSourceFeedback.forwardSlaveProgress();
}

OID ReplicationCoordinatorExternalStateImpl::ensureMe(OperationContext* txn) {
    std::string myname = getHostName();
    OID myRID;
    {
        ScopedTransaction transaction(txn, MODE_IX);
        Lock::DBLock lock(txn->lockState(), meDatabaseName, MODE_X);

        BSONObj me;
        // local.me is an identifier for a server for getLastError w:2+
        // TODO: handle WriteConflictExceptions below
        if (!Helpers::getSingleton(txn, meCollectionName, me) || !me.hasField("host") ||
            me["host"].String() != myname) {
            myRID = OID::gen();

            // clean out local.me
            Helpers::emptyCollection(txn, meCollectionName);

            // repopulate
            BSONObjBuilder b;
            b.append("_id", myRID);
            b.append("host", myname);
            Helpers::putSingleton(txn, meCollectionName, b.done());
        } else {
            myRID = me["_id"].OID();
        }
    }
    return myRID;
}

StatusWith<BSONObj> ReplicationCoordinatorExternalStateImpl::loadLocalConfigDocument(
    OperationContext* txn) {
    try {
        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            BSONObj config;
            if (!Helpers::getSingleton(txn, configCollectionName, config)) {
                return StatusWith<BSONObj>(
                    ErrorCodes::NoMatchingDocument,
                    str::stream() << "Did not find replica set configuration document in "
                                  << configCollectionName);
            }
            return StatusWith<BSONObj>(config);
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(txn, "load replica set config", configCollectionName);
    } catch (const DBException& ex) {
        return StatusWith<BSONObj>(ex.toStatus());
    }
}

Status ReplicationCoordinatorExternalStateImpl::storeLocalConfigDocument(OperationContext* txn,
                                                                         const BSONObj& config) {
    try {
        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            ScopedTransaction transaction(txn, MODE_IX);
            Lock::DBLock dbWriteLock(txn->lockState(), configDatabaseName, MODE_X);
            Helpers::putSingleton(txn, configCollectionName, config);
            return Status::OK();
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(txn, "save replica set config", configCollectionName);
    } catch (const DBException& ex) {
        return ex.toStatus();
    }
}

StatusWith<LastVote> ReplicationCoordinatorExternalStateImpl::loadLocalLastVoteDocument(
    OperationContext* txn) {
    try {
        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            BSONObj lastVoteObj;
            if (!Helpers::getSingleton(txn, lastVoteCollectionName, lastVoteObj)) {
                return StatusWith<LastVote>(ErrorCodes::NoMatchingDocument,
                                            str::stream()
                                                << "Did not find replica set lastVote document in "
                                                << lastVoteCollectionName);
            }
            LastVote lastVote;
            lastVote.initialize(lastVoteObj);
            return StatusWith<LastVote>(lastVote);
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(
            txn, "load replica set lastVote", lastVoteCollectionName);
    } catch (const DBException& ex) {
        return StatusWith<LastVote>(ex.toStatus());
    }
}

Status ReplicationCoordinatorExternalStateImpl::storeLocalLastVoteDocument(
    OperationContext* txn, const LastVote& lastVote) {
    BSONObj lastVoteObj = lastVote.toBSON();
    try {
        MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
            ScopedTransaction transaction(txn, MODE_IX);
            Lock::DBLock dbWriteLock(txn->lockState(), lastVoteDatabaseName, MODE_X);
            Helpers::putSingleton(txn, lastVoteCollectionName, lastVoteObj);
            return Status::OK();
        }
        MONGO_WRITE_CONFLICT_RETRY_LOOP_END(
            txn, "save replica set lastVote", lastVoteCollectionName);
        MONGO_UNREACHABLE;
    } catch (const DBException& ex) {
        return ex.toStatus();
    }
}

void ReplicationCoordinatorExternalStateImpl::setGlobalTimestamp(const Timestamp& newTime) {
    setNewTimestamp(newTime);
}

void ReplicationCoordinatorExternalStateImpl::cleanUpLastApplyBatch(OperationContext* txn) {
    if (_storageInterface->getInitialSyncFlag(txn)) {
        return;  // Initial Sync will take over so no cleanup is needed.
    }
    auto mv = _storageInterface->getMinValid(txn);

    if (!mv.start.isNull()) {
        // If we are in the middle of a batch, and recoveringm then we need to truncate the oplog.
        LOG(2) << "Recovering from a failed apply batch, start:" << mv.start.toBSON();
        truncateOplogTo(txn, mv.start.getTimestamp());
    }
}

StatusWith<OpTime> ReplicationCoordinatorExternalStateImpl::loadLastOpTime(OperationContext* txn) {
    // TODO: handle WriteConflictExceptions below
    try {
        // If we are doing an initial sync do not read from the oplog.
        if (_storageInterface->getInitialSyncFlag(txn)) {
            return {ErrorCodes::InitialSyncFailure, "In the middle of an initial sync."};
        }

        BSONObj oplogEntry;
        if (!Helpers::getLast(txn, rsOplogName.c_str(), oplogEntry)) {
            return StatusWith<OpTime>(ErrorCodes::NoMatchingDocument,
                                      str::stream() << "Did not find any entries in "
                                                    << rsOplogName);
        }
        BSONElement tsElement = oplogEntry[tsFieldName];
        if (tsElement.eoo()) {
            return StatusWith<OpTime>(ErrorCodes::NoSuchKey,
                                      str::stream() << "Most recent entry in " << rsOplogName
                                                    << " missing \""
                                                    << tsFieldName
                                                    << "\" field");
        }
        if (tsElement.type() != bsonTimestamp) {
            return StatusWith<OpTime>(ErrorCodes::TypeMismatch,
                                      str::stream() << "Expected type of \"" << tsFieldName
                                                    << "\" in most recent "
                                                    << rsOplogName
                                                    << " entry to have type Timestamp, but found "
                                                    << typeName(tsElement.type()));
        }
        return OpTime::parseFromOplogEntry(oplogEntry);
    } catch (const DBException& ex) {
        return StatusWith<OpTime>(ex.toStatus());
    }
}

bool ReplicationCoordinatorExternalStateImpl::isSelf(const HostAndPort& host, ServiceContext* ctx) {
    return repl::isSelf(host, ctx);
}

HostAndPort ReplicationCoordinatorExternalStateImpl::getClientHostAndPort(
    const OperationContext* txn) {
    return HostAndPort(txn->getClient()->clientAddress(true));
}

void ReplicationCoordinatorExternalStateImpl::closeConnections() {
    getGlobalServiceContext()->getTransportLayer()->endAllSessions(transport::Session::kKeepOpen);
}

void ReplicationCoordinatorExternalStateImpl::killAllUserOperations(OperationContext* txn) {
    ServiceContext* environment = txn->getServiceContext();
    environment->killAllUserOperations(txn, ErrorCodes::InterruptedDueToReplStateChange);
}

void ReplicationCoordinatorExternalStateImpl::shardingOnStepDownHook() {
    if (serverGlobalParams.clusterRole == ClusterRole::ConfigServer) {
        Balancer::get(getGlobalServiceContext())->stopThread();
    }

    ShardingState::get(getGlobalServiceContext())->clearCollectionMetadata();
}

void ReplicationCoordinatorExternalStateImpl::shardingOnDrainingStateHook(OperationContext* txn) {
    auto status = ShardingStateRecovery::recover(txn);

    if (status == ErrorCodes::ShutdownInProgress) {
        // Note: callers of this method don't expect exceptions, so throw only unexpected fatal
        // errors.
        return;
    }

    if (!status.isOK()) {
        fassertFailedWithStatus(40107, status);
    }

    if (serverGlobalParams.clusterRole == ClusterRole::ConfigServer) {
        status = Grid::get(txn)->catalogManager()->initializeConfigDatabaseIfNeeded(txn);
        if (!status.isOK() && status != ErrorCodes::AlreadyInitialized) {
            if (status == ErrorCodes::ShutdownInProgress ||
                status == ErrorCodes::InterruptedAtShutdown) {
                // Don't fassert if we're mid-shutdown, let the shutdown happen gracefully.
                return;
            }

            fassertFailedWithStatus(40184,
                                    Status(status.code(),
                                           str::stream()
                                               << "Failed to initialize config database on config "
                                                  "server's first transition to primary"
                                               << causedBy(status)));
        }

        if (status != ErrorCodes::AlreadyInitialized) {
            // Load the clusterId into memory. Use local readConcern, since we can't use majority
            // readConcern in drain mode because the global lock prevents replication. This is
            // safe, since if the clusterId write is rolled back, any writes that depend on it will
            // also be rolled back.
            ClusterIdentityLoader::get(txn)->getClusterId(
                txn, repl::ReadConcernLevel::kLocalReadConcern);
        }

        // Free any leftover locks from previous instantiations
        auto distLockManager = Grid::get(txn)->catalogClient(txn)->getDistLockManager();
        distLockManager->unlockAll(txn, distLockManager->getProcessID());

        // If this is a config server node becoming a primary, start the balancer
        auto balancer = Balancer::get(txn);

        // We need to join the balancer here, because it might have been running at a previous time
        // when this node was a primary.
        balancer->joinThread();
        balancer->startThread(txn);
    } else if (ShardingState::get(txn)->enabled()) {
        const auto configsvrConnStr =
            Grid::get(txn)->shardRegistry()->getConfigShard()->getConnString();
        auto status = ShardingState::get(txn)->updateShardIdentityConfigString(
            txn, configsvrConnStr.toString());
        if (!status.isOK()) {
            warning() << "error encountered while trying to update config connection string to "
                      << configsvrConnStr << causedBy(status);
        }
    }

    // There is a slight chance that some stale metadata might have been loaded before the latest
    // optime has been recovered, so throw out everything that we have up to now
    ShardingState::get(txn)->clearCollectionMetadata();
}

void ReplicationCoordinatorExternalStateImpl::signalApplierToChooseNewSyncSource() {
    if (_bgSync) {
        _bgSync->clearSyncTarget();
    }
}

void ReplicationCoordinatorExternalStateImpl::signalApplierToCancelFetcher() {
    invariant(_bgSync);
    _bgSync->cancelFetcher();
}

void ReplicationCoordinatorExternalStateImpl::dropAllTempCollections(OperationContext* txn) {
    std::vector<std::string> dbNames;
    StorageEngine* storageEngine = getGlobalServiceContext()->getGlobalStorageEngine();
    storageEngine->listDatabases(&dbNames);

    for (std::vector<std::string>::iterator it = dbNames.begin(); it != dbNames.end(); ++it) {
        // The local db is special because it isn't replicated. It is cleared at startup even on
        // replica set members.
        if (*it == "local")
            continue;
        LOG(2) << "Removing temporary collections from " << *it;
        Database* db = dbHolder().get(txn, *it);
        // Since we must be holding the global lock during this function, if listDatabases
        // returned this dbname, we should be able to get a reference to it - it can't have
        // been dropped.
        invariant(db);
        db->clearTmpCollections(txn);
    }
}

void ReplicationCoordinatorExternalStateImpl::dropAllSnapshots() {
    if (auto manager = getGlobalServiceContext()->getGlobalStorageEngine()->getSnapshotManager())
        manager->dropAllSnapshots();
}

void ReplicationCoordinatorExternalStateImpl::updateCommittedSnapshot(SnapshotName newCommitPoint) {
    auto manager = getGlobalServiceContext()->getGlobalStorageEngine()->getSnapshotManager();
    invariant(manager);  // This should never be called if there is no SnapshotManager.
    manager->setCommittedSnapshot(newCommitPoint);
}

void ReplicationCoordinatorExternalStateImpl::forceSnapshotCreation() {
    if (_snapshotThread)
        _snapshotThread->forceSnapshot();
}

bool ReplicationCoordinatorExternalStateImpl::snapshotsEnabled() const {
    return _snapshotThread != nullptr;
}

void ReplicationCoordinatorExternalStateImpl::notifyOplogMetadataWaiters() {
    signalOplogWaiters();
}

double ReplicationCoordinatorExternalStateImpl::getElectionTimeoutOffsetLimitFraction() const {
    return replElectionTimeoutOffsetLimitFraction;
}

bool ReplicationCoordinatorExternalStateImpl::isReadCommittedSupportedByStorageEngine(
    OperationContext* txn) const {
    auto storageEngine = txn->getServiceContext()->getGlobalStorageEngine();
    // This should never be called if the storage engine has not been initialized.
    invariant(storageEngine);
    return storageEngine->getSnapshotManager();
}

StatusWith<OpTime> ReplicationCoordinatorExternalStateImpl::multiApply(
    OperationContext* txn,
    MultiApplier::Operations ops,
    MultiApplier::ApplyOperationFn applyOperation) {
    return repl::multiApply(txn, _writerPool.get(), std::move(ops), applyOperation);
}

void ReplicationCoordinatorExternalStateImpl::multiSyncApply(MultiApplier::OperationPtrs* ops) {
    // SyncTail* argument is not used by repl::multiSyncApply().
    repl::multiSyncApply(ops, nullptr);
}

void ReplicationCoordinatorExternalStateImpl::multiInitialSyncApply(
    MultiApplier::OperationPtrs* ops, const HostAndPort& source) {

    // repl::multiInitialSyncApply uses SyncTail::shouldRetry() (and implicitly getMissingDoc())
    // to fetch missing documents during initial sync. Therefore, it is fine to construct SyncTail
    // with invalid BackgroundSync, MultiSyncApplyFunc and writerPool arguments because we will not
    // be accessing any SyncTail functionality that require these constructor parameters.
    SyncTail syncTail(nullptr, SyncTail::MultiSyncApplyFunc(), nullptr);
    syncTail.setHostname(source.toString());
    repl::multiInitialSyncApply(ops, &syncTail);
}

std::unique_ptr<OplogBuffer> ReplicationCoordinatorExternalStateImpl::makeInitialSyncOplogBuffer(
    OperationContext* txn) const {
    if (initialSyncOplogBuffer == kCollectionOplogBufferName) {
        return stdx::make_unique<OplogBufferCollection>(StorageInterface::get(txn));
    } else {
        return stdx::make_unique<OplogBufferBlockingQueue>();
    }
}

std::unique_ptr<OplogBuffer> ReplicationCoordinatorExternalStateImpl::makeSteadyStateOplogBuffer(
    OperationContext* txn) const {
    return stdx::make_unique<OplogBufferBlockingQueue>();
}

bool ReplicationCoordinatorExternalStateImpl::shouldUseDataReplicatorInitialSync() const {
    return !use3dot2InitialSync;
}

JournalListener::Token ReplicationCoordinatorExternalStateImpl::getToken() {
    return repl::getGlobalReplicationCoordinator()->getMyLastAppliedOpTime();
}

void ReplicationCoordinatorExternalStateImpl::onDurable(const JournalListener::Token& token) {
    repl::getGlobalReplicationCoordinator()->setMyLastDurableOpTimeForward(token);
}

}  // namespace repl
}  // namespace mongo