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
|
/**
* Copyright (C) 2012 10gen 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/bgsync.h"
#include <memory>
#include "mongo/base/counter.h"
#include "mongo/client/connection_pool.h"
#include "mongo/db/auth/authorization_session.h"
#include "mongo/db/client.h"
#include "mongo/db/commands/fsync.h"
#include "mongo/db/commands/server_status_metric.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/dbhelpers.h"
#include "mongo/db/operation_context_impl.h"
#include "mongo/db/repl/oplog.h"
#include "mongo/db/repl/oplog_interface_local.h"
#include "mongo/db/repl/oplogreader.h"
#include "mongo/db/repl/replication_coordinator_global.h"
#include "mongo/db/repl/replication_coordinator_impl.h"
#include "mongo/db/repl/rollback_source_impl.h"
#include "mongo/db/repl/rs_rollback.h"
#include "mongo/db/repl/rs_sync.h"
#include "mongo/db/stats/timer_stats.h"
#include "mongo/executor/network_interface_factory.h"
#include "mongo/rpc/metadata/repl_set_metadata.h"
#include "mongo/stdx/memory.h"
#include "mongo/util/concurrency/thread_pool.h"
#include "mongo/util/exit.h"
#include "mongo/util/fail_point_service.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
#include "mongo/util/scopeguard.h"
#include "mongo/util/time_support.h"
namespace mongo {
using std::string;
namespace repl {
namespace {
const char hashFieldName[] = "h";
int SleepToAllowBatchingMillis = 2;
const int BatchIsSmallish = 40000; // bytes
/**
* Returns new thread pool for thead pool task executor.
*/
std::unique_ptr<ThreadPool> makeThreadPool() {
ThreadPool::Options threadPoolOptions;
threadPoolOptions.poolName = "rsBackgroundSync";
return stdx::make_unique<ThreadPool>(threadPoolOptions);
}
/**
* Checks the criteria for rolling back.
* 'getNextOperation' returns the first result of the oplog tailing query.
* 'lastOpTimeFetched' should be consistent with the predicate in the query.
* Returns RemoteOplogStale if the oplog query has no results.
* Returns OplogStartMissing if we cannot find the timestamp of the last fetched operation in
* the remote oplog.
*/
Status checkRemoteOplogStart(stdx::function<StatusWith<BSONObj>()> getNextOperation,
OpTime lastOpTimeFetched,
long long lastHashFetched) {
auto result = getNextOperation();
if (!result.isOK()) {
// The GTE query from upstream returns nothing, so we're ahead of the upstream.
return Status(ErrorCodes::RemoteOplogStale,
"we are ahead of the sync source, will try to roll back");
}
BSONObj o = result.getValue();
OpTime opTime = fassertStatusOK(28778, OpTime::parseFromOplogEntry(o));
long long hash = o["h"].numberLong();
if (opTime != lastOpTimeFetched || hash != lastHashFetched) {
return Status(ErrorCodes::OplogStartMissing,
str::stream() << "our last op time fetched: " << lastOpTimeFetched.toString()
<< ". source's GTE: " << opTime.toString() << " hashes: ("
<< lastHashFetched << "/" << hash << ")");
}
return Status::OK();
}
} // namespace
MONGO_FP_DECLARE(rsBgSyncProduce);
MONGO_FP_DECLARE(stepDownWhileDrainingFailPoint);
BackgroundSync* BackgroundSync::s_instance = 0;
stdx::mutex BackgroundSync::s_mutex;
// The number and time spent reading batches off the network
static TimerStats getmoreReplStats;
static ServerStatusMetricField<TimerStats> displayBatchesRecieved("repl.network.getmores",
&getmoreReplStats);
// The oplog entries read via the oplog reader
static Counter64 opsReadStats;
static ServerStatusMetricField<Counter64> displayOpsRead("repl.network.ops", &opsReadStats);
// The bytes read via the oplog reader
static Counter64 networkByteStats;
static ServerStatusMetricField<Counter64> displayBytesRead("repl.network.bytes", &networkByteStats);
// The count of items in the buffer
static Counter64 bufferCountGauge;
static ServerStatusMetricField<Counter64> displayBufferCount("repl.buffer.count",
&bufferCountGauge);
// The size (bytes) of items in the buffer
static Counter64 bufferSizeGauge;
static ServerStatusMetricField<Counter64> displayBufferSize("repl.buffer.sizeBytes",
&bufferSizeGauge);
// The max size (bytes) of the buffer
static int bufferMaxSizeGauge = 256 * 1024 * 1024;
static ServerStatusMetricField<int> displayBufferMaxSize("repl.buffer.maxSizeBytes",
&bufferMaxSizeGauge);
BackgroundSyncInterface::~BackgroundSyncInterface() {}
namespace {
size_t getSize(const BSONObj& o) {
// SERVER-9808 Avoid Fortify complaint about implicit signed->unsigned conversion
return static_cast<size_t>(o.objsize());
}
} // namespace
BackgroundSync::BackgroundSync()
: _buffer(bufferMaxSizeGauge, &getSize),
_threadPoolTaskExecutor(makeThreadPool(), executor::makeNetworkInterface()),
_lastOpTimeFetched(Timestamp(std::numeric_limits<int>::max(), 0),
std::numeric_limits<long long>::max()),
_lastFetchedHash(0),
_pause(true),
_replCoord(getGlobalReplicationCoordinator()),
_initialSyncRequestedFlag(false),
_indexPrefetchConfig(PREFETCH_ALL) {}
BackgroundSync* BackgroundSync::get() {
stdx::unique_lock<stdx::mutex> lock(s_mutex);
if (s_instance == NULL && !inShutdown()) {
s_instance = new BackgroundSync();
}
return s_instance;
}
void BackgroundSync::shutdown() {
stdx::lock_guard<stdx::mutex> lock(_mutex);
// Clear the buffer in case the producerThread is waiting in push() due to a full queue.
invariant(inShutdown());
clearBuffer();
_pause = true;
}
void BackgroundSync::producerThread() {
Client::initThread("rsBackgroundSync");
AuthorizationSession::get(cc())->grantInternalAuthorization();
_threadPoolTaskExecutor.startup();
ON_BLOCK_EXIT([this]() {
_threadPoolTaskExecutor.shutdown();
_threadPoolTaskExecutor.join();
});
while (!inShutdown()) {
try {
_producerThread();
} catch (const DBException& e) {
std::string msg(str::stream() << "sync producer problem: " << e.toString());
error() << msg;
_replCoord->setMyHeartbeatMessage(msg);
} catch (const std::exception& e2) {
severe() << "sync producer exception: " << e2.what();
fassertFailed(28546);
}
}
stop();
}
void BackgroundSync::_producerThread() {
const MemberState state = _replCoord->getMemberState();
// we want to pause when the state changes to primary
if (_replCoord->isWaitingForApplierToDrain() || state.primary()) {
if (!isPaused()) {
stop();
}
if (_replCoord->isWaitingForApplierToDrain()) {
// Signal to consumers that we have entered the paused state if the signal isn't already
// in the queue.
const boost::optional<BSONObj> lastObjectPushed = _buffer.lastObjectPushed();
if (!lastObjectPushed || !lastObjectPushed->isEmpty()) {
const BSONObj sentinelDoc;
_buffer.pushEvenIfFull(sentinelDoc);
bufferCountGauge.increment();
bufferSizeGauge.increment(sentinelDoc.objsize());
}
}
sleepsecs(1);
return;
}
// TODO(spencer): Use a condition variable to await loading a config.
if (state.startup()) {
// Wait for a config to be loaded
sleepsecs(1);
return;
}
// We need to wait until initial sync has started.
if (_replCoord->getMyLastOptime().isNull()) {
sleepsecs(1);
return;
}
// we want to unpause when we're no longer primary
// start() also loads _lastOpTimeFetched, which we know is set from the "if"
OperationContextImpl txn;
if (isPaused()) {
start(&txn);
}
_produce(&txn);
}
void BackgroundSync::_produce(OperationContext* txn) {
// this oplog reader does not do a handshake because we don't want the server it's syncing
// from to track how far it has synced
{
stdx::unique_lock<stdx::mutex> lock(_mutex);
if (_lastOpTimeFetched.isNull()) {
// then we're initial syncing and we're still waiting for this to be set
lock.unlock();
sleepsecs(1);
// if there is no one to sync from
return;
}
if (_replCoord->isWaitingForApplierToDrain() || _replCoord->getMemberState().primary() ||
inShutdownStrict()) {
return;
}
}
while (MONGO_FAIL_POINT(rsBgSyncProduce)) {
sleepmillis(0);
}
// find a target to sync from the last optime fetched
OpTime lastOpTimeFetched;
{
stdx::unique_lock<stdx::mutex> lock(_mutex);
lastOpTimeFetched = _lastOpTimeFetched;
_syncSourceHost = HostAndPort();
}
OplogReader syncSourceReader;
syncSourceReader.connectToSyncSource(txn, lastOpTimeFetched, _replCoord);
// no server found
if (syncSourceReader.getHost().empty()) {
sleepsecs(1);
// if there is no one to sync from
return;
}
long long lastHashFetched;
{
stdx::lock_guard<stdx::mutex> lock(_mutex);
if (_pause) {
return;
}
lastOpTimeFetched = _lastOpTimeFetched;
lastHashFetched = _lastFetchedHash;
_syncSourceHost = syncSourceReader.getHost();
_replCoord->signalUpstreamUpdater();
}
const Milliseconds oplogSocketTimeout(OplogReader::kSocketTimeout);
const auto isV1ElectionProtocol = _replCoord->isV1ElectionProtocol();
// Under protocol version 1, make the awaitData timeout (maxTimeMS) dependent on the election
// timeout. This enables the sync source to communicate liveness of the primary to secondaries.
// Under protocol version 0, use a default timeout of 2 seconds for awaitData.
const Milliseconds fetcherMaxTimeMS(
isV1ElectionProtocol ? _replCoord->getConfig().getElectionTimeoutPeriod() / 2 : Seconds(2));
// Prefer host in oplog reader to _syncSourceHost because _syncSourceHost may be cleared
// if sync source feedback fails.
const HostAndPort source = syncSourceReader.getHost();
syncSourceReader.resetConnection();
// no more references to oplog reader from here on.
// If this status is not OK after the fetcher returns from wait(),
// proceed to execute rollback
Status remoteOplogStartStatus = Status::OK();
auto fetcherCallback = stdx::bind(&BackgroundSync::_fetcherCallback,
this,
stdx::placeholders::_1,
stdx::placeholders::_3,
stdx::cref(source),
lastOpTimeFetched,
lastHashFetched,
fetcherMaxTimeMS,
&remoteOplogStartStatus);
BSONObjBuilder cmdBob;
cmdBob.append("find", nsToCollectionSubstring(rsOplogName));
cmdBob.append("filter", BSON("ts" << BSON("$gte" << lastOpTimeFetched.getTimestamp())));
cmdBob.append("tailable", true);
cmdBob.append("oplogReplay", true);
cmdBob.append("awaitData", true);
cmdBob.append("maxTimeMS", durationCount<Milliseconds>(fetcherMaxTimeMS));
BSONObjBuilder metadataBob;
if (isV1ElectionProtocol) {
cmdBob.append("term", _replCoord->getTerm());
metadataBob.append(rpc::kReplSetMetadataFieldName, 1);
}
auto dbName = nsToDatabase(rsOplogName);
auto cmdObj = cmdBob.obj();
auto metadataObj = metadataBob.obj();
Fetcher fetcher(&_threadPoolTaskExecutor,
source,
dbName,
cmdObj,
fetcherCallback,
metadataObj,
_replCoord->getConfig().getElectionTimeoutPeriod());
auto scheduleStatus = fetcher.schedule();
if (!scheduleStatus.isOK()) {
warning() << "unable to schedule fetcher to read remote oplog on " << source << ": "
<< scheduleStatus;
return;
}
fetcher.wait();
// If the background sync is paused after the fetcher is started, we need to
// re-evaluate our sync source and oplog common point.
if (isPaused()) {
return;
}
// Execute rollback if necessary.
// Rollback is a synchronous operation that uses the task executor and may not be
// executed inside the fetcher callback.
if (!remoteOplogStartStatus.isOK()) {
const int messagingPortTags = 0;
ConnectionPool connectionPool(messagingPortTags);
std::unique_ptr<ConnectionPool::ConnectionPtr> connection;
auto getConnection =
[&connection, &connectionPool, oplogSocketTimeout, source]() -> DBClientBase* {
if (!connection.get()) {
connection.reset(new ConnectionPool::ConnectionPtr(
&connectionPool, source, Date_t::now(), oplogSocketTimeout));
};
return connection->get();
};
log() << "starting rollback: " << remoteOplogStartStatus;
_rollback(txn, source, getConnection);
stop();
}
}
void BackgroundSync::_fetcherCallback(const StatusWith<Fetcher::QueryResponse>& result,
BSONObjBuilder* bob,
const HostAndPort& source,
OpTime lastOpTimeFetched,
long long lastFetchedHash,
Milliseconds fetcherMaxTimeMS,
Status* remoteOplogStartStatus) {
// if target cut connections between connecting and querying (for
// example, because it stepped down) we might not have a cursor
if (!result.isOK()) {
return;
}
if (inShutdown()) {
return;
}
// Check if we have been paused.
if (isPaused()) {
return;
}
const auto& queryResponse = result.getValue();
bool syncSourceHasSyncSource = false;
OpTime sourcesLastOp;
// Forward metadata (containing liveness information) to replication coordinator.
bool receivedMetadata =
queryResponse.otherFields.metadata.hasElement(rpc::kReplSetMetadataFieldName);
if (receivedMetadata) {
auto metadataResult =
rpc::ReplSetMetadata::readFromMetadata(queryResponse.otherFields.metadata);
if (!metadataResult.isOK()) {
error() << "invalid replication metadata from sync source " << source << ": "
<< metadataResult.getStatus() << ": " << queryResponse.otherFields.metadata;
return;
}
const auto& metadata = metadataResult.getValue();
_replCoord->processReplSetMetadata(metadata);
if (metadata.getPrimaryIndex() != rpc::ReplSetMetadata::kNoPrimary) {
_replCoord->cancelAndRescheduleElectionTimeout();
}
syncSourceHasSyncSource = metadata.getSyncSourceIndex() != -1;
sourcesLastOp = metadata.getLastOpVisible();
}
const auto& documents = queryResponse.documents;
auto documentBegin = documents.cbegin();
auto documentEnd = documents.cend();
// Check start of remote oplog and, if necessary, stop fetcher to execute rollback.
if (queryResponse.first) {
auto getNextOperation = [&documentBegin, documentEnd]() -> StatusWith<BSONObj> {
if (documentBegin == documentEnd) {
return Status(ErrorCodes::OplogStartMissing, "remote oplog start missing");
}
return *(documentBegin++);
};
*remoteOplogStartStatus =
checkRemoteOplogStart(getNextOperation, lastOpTimeFetched, lastFetchedHash);
if (!remoteOplogStartStatus->isOK()) {
// Stop fetcher and execute rollback.
return;
}
// If this is the first batch and no rollback is needed, we should have advanced
// the document iterator.
invariant(documentBegin != documents.cbegin());
}
// process documents
int currentBatchMessageSize = 0;
for (auto documentIter = documentBegin; documentIter != documentEnd; ++documentIter) {
if (inShutdown()) {
return;
}
// If we are transitioning to primary state, we need to leave
// this loop in order to go into bgsync-pause mode.
if (_replCoord->isWaitingForApplierToDrain() || _replCoord->getMemberState().primary()) {
LOG(1) << "waiting for draining or we are primary, not adding more ops to buffer";
return;
}
// At this point, we are guaranteed to have at least one thing to read out
// of the fetcher.
const BSONObj& o = *documentIter;
currentBatchMessageSize += o.objsize();
opsReadStats.increment();
if (MONGO_FAIL_POINT(stepDownWhileDrainingFailPoint)) {
sleepsecs(20);
}
OCCASIONALLY {
LOG(2) << "bgsync buffer has " << _buffer.size() << " bytes";
}
bufferCountGauge.increment();
bufferSizeGauge.increment(getSize(o));
_buffer.push(o);
{
stdx::unique_lock<stdx::mutex> lock(_mutex);
_lastFetchedHash = o["h"].numberLong();
_lastOpTimeFetched = fassertStatusOK(28770, OpTime::parseFromOplogEntry(o));
LOG(3) << "lastOpTimeFetched: " << _lastOpTimeFetched;
}
}
// record time for each batch
getmoreReplStats.recordMillis(durationCount<Milliseconds>(queryResponse.elapsedMillis));
networkByteStats.increment(currentBatchMessageSize);
// Check some things periodically
// (whenever we run out of items in the
// current cursor batch)
if (currentBatchMessageSize > 0 && currentBatchMessageSize < BatchIsSmallish) {
// on a very low latency network, if we don't wait a little, we'll be
// getting ops to write almost one at a time. this will both be expensive
// for the upstream server as well as potentially defeating our parallel
// application of batches on the secondary.
//
// the inference here is basically if the batch is really small, we are
// "caught up".
//
sleepmillis(SleepToAllowBatchingMillis);
}
// If we are transitioning to primary state, we need to leave
// this loop in order to go into bgsync-pause mode.
if (_replCoord->isWaitingForApplierToDrain() || _replCoord->getMemberState().primary()) {
return;
}
// re-evaluate quality of sync target
if (_shouldChangeSyncSource(source, sourcesLastOp, syncSourceHasSyncSource)) {
return;
}
// Check if we have been paused.
if (isPaused()) {
return;
}
// We fill in 'bob' to signal the fetcher to process with another getMore.
invariant(bob);
bob->append("getMore", queryResponse.cursorId);
bob->append("collection", queryResponse.nss.coll());
bob->append("maxTimeMS", durationCount<Milliseconds>(fetcherMaxTimeMS));
if (receivedMetadata) {
bob->append("term", _replCoord->getTerm());
_replCoord->getLastCommittedOpTime().append(bob, "lastKnownCommittedOpTime");
}
}
bool BackgroundSync::_shouldChangeSyncSource(const HostAndPort& syncSource,
const OpTime& syncSourceLastOpTime,
bool syncSourceHasSyncSource) {
// is it even still around?
if (getSyncTarget().empty() || syncSource.empty()) {
return true;
}
// check other members: is any member's optime more than MaxSyncSourceLag seconds
// ahead of the current sync source?
return _replCoord->shouldChangeSyncSource(
syncSource, syncSourceLastOpTime, syncSourceHasSyncSource);
}
bool BackgroundSync::peek(BSONObj* op) {
return _buffer.peek(*op);
}
void BackgroundSync::waitForMore() {
BSONObj op;
// Block for one second before timing out.
// Ignore the value of the op we peeked at.
_buffer.blockingPeek(op, 1);
}
void BackgroundSync::consume() {
// this is just to get the op off the queue, it's been peeked at
// and queued for application already
BSONObj op = _buffer.blockingPop();
bufferCountGauge.decrement(1);
bufferSizeGauge.decrement(getSize(op));
}
void BackgroundSync::_rollback(OperationContext* txn,
const HostAndPort& source,
stdx::function<DBClientBase*()> getConnection) {
// Abort only when syncRollback detects we are in a unrecoverable state.
// In other cases, we log the message contained in the error status and retry later.
auto status = syncRollback(txn,
_replCoord->getMyLastOptime(),
OplogInterfaceLocal(txn, rsOplogName),
RollbackSourceImpl(getConnection, source, rsOplogName),
_replCoord);
if (status.isOK()) {
return;
}
if (ErrorCodes::UnrecoverableRollbackError == status.code()) {
fassertNoTrace(28723, status);
}
warning() << "rollback cannot proceed at this time (retrying later): " << status;
}
HostAndPort BackgroundSync::getSyncTarget() {
stdx::unique_lock<stdx::mutex> lock(_mutex);
return _syncSourceHost;
}
void BackgroundSync::clearSyncTarget() {
stdx::unique_lock<stdx::mutex> lock(_mutex);
_syncSourceHost = HostAndPort();
}
void BackgroundSync::cancelFetcher() {
_threadPoolTaskExecutor.cancelAllCommands();
}
void BackgroundSync::stop() {
stdx::lock_guard<stdx::mutex> lock(_mutex);
_pause = true;
_syncSourceHost = HostAndPort();
_lastOpTimeFetched = OpTime();
_lastFetchedHash = 0;
}
void BackgroundSync::start(OperationContext* txn) {
massert(16235, "going to start syncing, but buffer is not empty", _buffer.empty());
long long lastFetchedHash = _readLastAppliedHash(txn);
stdx::lock_guard<stdx::mutex> lk(_mutex);
_pause = false;
// reset _last fields with current oplog data
_lastOpTimeFetched = _replCoord->getMyLastOptime();
_lastFetchedHash = lastFetchedHash;
LOG(1) << "bgsync fetch queue set to: " << _lastOpTimeFetched << " " << _lastFetchedHash;
}
bool BackgroundSync::isPaused() const {
stdx::lock_guard<stdx::mutex> lock(_mutex);
return _pause;
}
void BackgroundSync::clearBuffer() {
_buffer.clear();
const auto count = bufferCountGauge.get();
bufferCountGauge.decrement(count);
const auto size = bufferSizeGauge.get();
bufferSizeGauge.decrement(size);
}
long long BackgroundSync::_readLastAppliedHash(OperationContext* txn) {
BSONObj oplogEntry;
try {
MONGO_WRITE_CONFLICT_RETRY_LOOP_BEGIN {
ScopedTransaction transaction(txn, MODE_IX);
Lock::DBLock lk(txn->lockState(), "local", MODE_X);
bool success = Helpers::getLast(txn, rsOplogName.c_str(), oplogEntry);
if (!success) {
// This can happen when we are to do an initial sync. lastHash will be set
// after the initial sync is complete.
return 0;
}
}
MONGO_WRITE_CONFLICT_RETRY_LOOP_END(txn, "readLastAppliedHash", rsOplogName);
} catch (const DBException& ex) {
severe() << "Problem reading " << rsOplogName << ": " << ex.toStatus();
fassertFailed(18904);
}
BSONElement hashElement = oplogEntry[hashFieldName];
if (hashElement.eoo()) {
severe() << "Most recent entry in " << rsOplogName << " missing \"" << hashFieldName
<< "\" field";
fassertFailed(18902);
}
if (hashElement.type() != NumberLong) {
severe() << "Expected type of \"" << hashFieldName << "\" in most recent " << rsOplogName
<< " entry to have type NumberLong, but found " << typeName(hashElement.type());
fassertFailed(18903);
}
return hashElement.safeNumberLong();
}
bool BackgroundSync::getInitialSyncRequestedFlag() {
stdx::lock_guard<stdx::mutex> lock(_initialSyncMutex);
return _initialSyncRequestedFlag;
}
void BackgroundSync::setInitialSyncRequestedFlag(bool value) {
stdx::lock_guard<stdx::mutex> lock(_initialSyncMutex);
_initialSyncRequestedFlag = value;
}
void BackgroundSync::pushTestOpToBuffer(const BSONObj& op) {
_buffer.push(op);
bufferCountGauge.increment();
bufferSizeGauge.increment(op.objsize());
}
} // namespace repl
} // namespace mongo
|