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
|
/**
* Copyright (C) 2015 MongoDB Inc.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License, version 3,
* as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
* As a special exception, the copyright holders give permission to link the
* code of portions of this program with the OpenSSL library under certain
* conditions as described in each individual source file and distribute
* linked combinations including the program with the OpenSSL library. You
* must comply with the GNU Affero General Public License in all respects for
* all of the code used other than as permitted herein. If you modify file(s)
* with this exception, you may extend this exception to your version of the
* file(s), but you are not obligated to do so. If you do not wish to do so,
* delete this exception statement from your version. If you delete this
* exception statement from all source files in the program, then also delete
* it in the license file.
*/
#pragma once
#include <cstdint>
#include <iosfwd>
#include <memory>
#include "mongo/base/status.h"
#include "mongo/bson/bsonobj.h"
#include "mongo/bson/timestamp.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/repl/callback_completion_guard.h"
#include "mongo/db/repl/collection_cloner.h"
#include "mongo/db/repl/data_replicator_external_state.h"
#include "mongo/db/repl/multiapplier.h"
#include "mongo/db/repl/oplog_buffer.h"
#include "mongo/db/repl/oplog_fetcher.h"
#include "mongo/db/repl/optime.h"
#include "mongo/db/repl/rollback_checker.h"
#include "mongo/db/repl/sync_source_selector.h"
#include "mongo/stdx/condition_variable.h"
#include "mongo/stdx/functional.h"
#include "mongo/stdx/mutex.h"
#include "mongo/util/fail_point_service.h"
#include "mongo/util/net/hostandport.h"
namespace mongo {
namespace repl {
// TODO: Remove forward declares once we remove rs_initialsync.cpp and other dependents.
// Failpoint which fails initial sync and leaves an oplog entry in the buffer.
MONGO_FP_FORWARD_DECLARE(failInitSyncWithBufferedEntriesLeft);
// Failpoint which causes the initial sync function to hang before copying databases.
MONGO_FP_FORWARD_DECLARE(initialSyncHangBeforeCopyingDatabases);
// Failpoint which causes the initial sync function to hang before calling shouldRetry on a failed
// operation.
MONGO_FP_FORWARD_DECLARE(initialSyncHangBeforeGettingMissingDocument);
// Failpoint which stops the applier.
MONGO_FP_FORWARD_DECLARE(rsSyncApplyStop);
struct InitialSyncState;
struct MemberState;
class StorageInterface;
struct DataReplicatorOptions {
/** Function to return optime of last operation applied on this node */
using GetMyLastOptimeFn = stdx::function<OpTime()>;
/** Function to update optime of last operation applied on this node */
using SetMyLastOptimeFn = stdx::function<void(const OpTime&)>;
/** Function to sets this node into a specific follower mode. */
using SetFollowerModeFn = stdx::function<bool(const MemberState&)>;
/** Function to get this node's slaveDelay. */
using GetSlaveDelayFn = stdx::function<Seconds()>;
// Error and retry values
Milliseconds syncSourceRetryWait{1000};
Milliseconds initialSyncRetryWait{1000};
Seconds blacklistSyncSourcePenaltyForNetworkConnectionError{10};
Minutes blacklistSyncSourcePenaltyForOplogStartMissing{10};
// DataReplicator waits this long before retrying getApplierBatchCallback() if there are
// currently no operations available to apply or if the 'rsSyncApplyStop' failpoint is active.
// This default value is based on the duration in BackgroundSync::waitForMore() and
// SyncTail::tryPopAndWaitForMore().
Milliseconds getApplierBatchCallbackRetryWait{1000};
// Batching settings.
std::uint32_t replBatchLimitBytes = 512 * 1024 * 1024;
std::uint32_t replBatchLimitOperations = 5000;
// Replication settings
NamespaceString localOplogNS = NamespaceString("local.oplog.rs");
NamespaceString remoteOplogNS = NamespaceString("local.oplog.rs");
GetMyLastOptimeFn getMyLastOptime;
SetMyLastOptimeFn setMyLastOptime;
GetSlaveDelayFn getSlaveDelay;
SyncSourceSelector* syncSourceSelector = nullptr;
// The oplog fetcher will restart the oplog tailing query this many times on non-cancellation
// failures.
std::uint32_t oplogFetcherMaxFetcherRestarts = 0;
std::string toString() const {
return str::stream() << "DataReplicatorOptions -- "
<< " localOplogNs: " << localOplogNS.toString()
<< " remoteOplogNS: " << remoteOplogNS.toString();
}
};
/**
* The data replicator provides services to keep collection in sync by replicating
* changes via an oplog source to the local system storage.
*
* This class will use existing machinery like the Executor to schedule work and
* network tasks, as well as provide serial access and synchronization of state.
*
*
* Entry Points:
* -- startup: Start initial sync.
*/
class DataReplicator {
MONGO_DISALLOW_COPYING(DataReplicator);
public:
/**
* Callback function to report last applied optime (with hash) of initial sync.
*/
typedef stdx::function<void(const StatusWith<OpTimeWithHash>& lastApplied)> OnCompletionFn;
/**
* Callback completion guard for data replicator.
*/
using OnCompletionGuard = CallbackCompletionGuard<StatusWith<OpTimeWithHash>>;
struct InitialSyncAttemptInfo {
int durationMillis;
Status status;
HostAndPort syncSource;
std::string toString() const;
BSONObj toBSON() const;
void append(BSONObjBuilder* builder) const;
};
struct Stats {
std::uint32_t failedInitialSyncAttempts{0};
std::uint32_t maxFailedInitialSyncAttempts{0};
Date_t initialSyncStart;
Date_t initialSyncEnd;
std::vector<DataReplicator::InitialSyncAttemptInfo> initialSyncAttemptInfos;
std::string toString() const;
BSONObj toBSON() const;
void append(BSONObjBuilder* builder) const;
};
DataReplicator(DataReplicatorOptions opts,
std::unique_ptr<DataReplicatorExternalState> dataReplicatorExternalState,
StorageInterface* storage,
const OnCompletionFn& onCompletion);
virtual ~DataReplicator();
/**
* Returns true if an initial sync is currently running or in the process of shutting down.
*/
bool isActive() const;
/**
* Starts initial sync process, with the provided number of attempts
*/
Status startup(OperationContext* opCtx, std::uint32_t maxAttempts) noexcept;
/**
* Shuts down replication if "start" has been called, and blocks until shutdown has completed.
*/
Status shutdown();
/**
* Block until inactive.
*/
void join();
/**
* Returns internal state in a loggable format.
*/
std::string getDiagnosticString() const;
/**
* Returns stats about the progress of initial sync. If initial sync is not in progress it
* returns summary statistics for what occurred during initial sync.
*/
BSONObj getInitialSyncProgress() const;
/**
* Overrides how executor schedules database work.
*
* For testing only.
*/
void setScheduleDbWorkFn_forTest(const CollectionCloner::ScheduleDbWorkFn& scheduleDbWorkFn);
// State transitions:
// PreStart --> Running --> ShuttingDown --> Complete
// It is possible to skip intermediate states. For example, calling shutdown() when the data
// replicator has not started will transition from PreStart directly to Complete.
enum class State { kPreStart, kRunning, kShuttingDown, kComplete };
/**
* Returns current data replicator state.
* For testing only.
*/
State getState_forTest() const;
private:
/**
* Returns true if we are still processing initial sync tasks (_state is either Running or
* Shutdown).
*/
bool _isActive_inlock() const;
/**
* Cancels all outstanding work.
* Used by shutdown() and CompletionGuard::setResultAndCancelRemainingWork().
*/
void _cancelRemainingWork_inlock();
/**
* Returns true if the data replicator has received a shutdown request (_state is ShuttingDown).
*/
bool _isShuttingDown() const;
bool _isShuttingDown_inlock() const;
/**
* Initial sync flowchart:
*
* start()
* |
* |
* V
* _setUp_inlock()
* |
* |
* V
* _startInitialSyncAttemptCallback()
* |
* |
* |<-------+
* | |
* | | (bad sync source)
* | |
* V |
* _chooseSyncSourceCallback()
* |
* |
* | (good sync source found)
* |
* |
* V
* _recreateOplogAndDropReplicatedDatabases()
* |
* |
* V
* _rollbackCheckerResetCallback()
* |
* |
* V
* _lastOplogEntryFetcherCallbackForBeginTimestamp()
* |
* |
* +------------------------------+
* | |
* | |
* V V
* _oplogFetcherCallback() _databasesClonerCallback
* | |
* | |
* | V
* | _lastOplogEntryFetcherCallbackForStopTimestamp()
* | | |
* | | |
* | (no ops to apply) | | (have ops to apply)
* | | |
* | | V
* | | _getNextApplierBatchCallback()<-----+
* | | | ^ |
* | | | | |
* | | | (no docs fetched | |
* | | | and end ts not | |
* | | | reached) | |
* | | | | |
* | | V | |
* | | _multiApplierCallback()-----+ |
* | | | | |
* | | | | |
* | | | | (docs fetched) | (end ts not
* | | | | | reached)
* | | | V |
* | | | _lastOplogEntryFetcherCallbackAfter-
* | | | FetchingMissingDocuments()
* | | | |
* | | | |
* | (reached end timestamp)
* | | | |
* | V V V
* | _rollbackCheckerCheckForRollbackCallback()
* | |
* | |
* +------------------------------+
* |
* |
* V
* _finishInitialSyncAttempt()
* |
* |
* V
* _finishCallback()
*/
/**
* Sets up internal state to begin initial sync.
*/
void _setUp_inlock(OperationContext* opCtx, std::uint32_t initialSyncMaxAttempts);
/**
* Tears down internal state before reporting final status to caller.
*/
void _tearDown_inlock(OperationContext* opCtx, const StatusWith<OpTimeWithHash>& lastApplied);
/**
* Callback to start a single initial sync attempt.
*/
void _startInitialSyncAttemptCallback(const executor::TaskExecutor::CallbackArgs& callbackArgs,
std::uint32_t initialSyncAttempt,
std::uint32_t initialSyncMaxAttempts);
/**
* Callback to obtain sync source from sync source selector.
* For every initial sync attempt, we will try up to 'numInitialSyncConnectAttempts' times (at
* an interval of '_opts.syncSourceRetryWait' ms) to obtain a valid sync source before giving up
* and returning ErrorCodes::InitialSyncOplogSourceMissing.
*/
void _chooseSyncSourceCallback(const executor::TaskExecutor::CallbackArgs& callbackArgs,
std::uint32_t chooseSyncSourceAttempt,
std::uint32_t chooseSyncSourceMaxAttempts,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* This function does the following:
* 1.) Drop oplog.
* 2.) Drop user databases (replicated dbs).
* 3.) Create oplog.
*/
Status _recreateOplogAndDropReplicatedDatabases();
/**
* Callback for rollback checker's first replSetGetRBID command before starting data cloning.
*/
void _rollbackCheckerResetCallback(const RollbackChecker::Result& result,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for first '_lastOplogEntryFetcher' callback. A successful response lets us
* determine the starting point for tailing the oplog using the OplogFetcher as well as
* setting a reference point for the state of the sync source's oplog when data cloning
* completes.
*/
void _lastOplogEntryFetcherCallbackForBeginTimestamp(
const StatusWith<Fetcher::QueryResponse>& result,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for oplog fetcher.
*/
void _oplogFetcherCallback(const Status& status,
const OpTimeWithHash& lastFetched,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for DatabasesCloner.
*/
void _databasesClonerCallback(const Status& status,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for second '_lastOplogEntryFetcher' callback. This is scheduled to obtain the stop
* timestamp after DatabasesCloner has completed and enables us to determine if the oplog on
* the sync source has advanced since we started cloning the databases.
*/
void _lastOplogEntryFetcherCallbackForStopTimestamp(
const StatusWith<Fetcher::QueryResponse>& result,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback to obtain next batch of operations to apply.
*/
void _getNextApplierBatchCallback(const executor::TaskExecutor::CallbackArgs& callbackArgs,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for MultiApplier completion.
*/
void _multiApplierCallback(const Status& status,
OpTimeWithHash lastApplied,
std::uint32_t numApplied,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for third '_lastOplogEntryFetcher' callback. This is scheduled after MultiApplier
* completed successfully and missing documents were fetched from the sync source while
* DataReplicatorExternalState::_multiInitialSyncApply() was processing operations.
* This callback will update InitialSyncState::stopTimestamp on success.
*/
void _lastOplogEntryFetcherCallbackAfterFetchingMissingDocuments(
const StatusWith<Fetcher::QueryResponse>& result,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Callback for rollback checker's last replSetGetRBID command after cloning data and applying
* operations.
*/
void _rollbackCheckerCheckForRollbackCallback(
const RollbackChecker::Result& result,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Reports result of current initial sync attempt. May schedule another initial sync attempt
* depending on shutdown state and whether we've exhausted all initial sync retries.
*/
void _finishInitialSyncAttempt(const StatusWith<OpTimeWithHash>& lastApplied);
/**
* Invokes completion callback and transitions state to State::kComplete.
*/
void _finishCallback(StatusWith<OpTimeWithHash> lastApplied);
// Obtains a valid sync source from the sync source selector.
// Returns error if a sync source cannot be found.
StatusWith<HostAndPort> _chooseSyncSource_inlock();
/**
* Pushes documents from oplog fetcher to blocking queue for
* applier to consume.
*
* Returns a status even though it always returns OK, to conform the interface OplogFetcher
* expects for the EnqueueDocumentsFn.
*/
Status _enqueueDocuments(Fetcher::Documents::const_iterator begin,
Fetcher::Documents::const_iterator end,
const OplogFetcher::DocumentsInfo& info);
BSONObj _getInitialSyncProgress_inlock() const;
StatusWith<MultiApplier::Operations> _getNextApplierBatch_inlock();
/**
* Schedules a fetcher to get the last oplog entry from the sync source.
*/
Status _scheduleLastOplogEntryFetcher_inlock(Fetcher::CallbackFn callback);
/**
* Checks the current oplog application progress (begin and end timestamps).
* If necessary, schedules a _getNextApplierBatchCallback() task.
* If the stop and end timestamps are inconsistent or if there is an issue scheduling the task,
* we set the error status in 'onCompletionGuard' and shut down the OplogFetcher.
* Passes 'lock' through to completion guard.
*/
void _checkApplierProgressAndScheduleGetNextApplierBatch_inlock(
const stdx::lock_guard<stdx::mutex>& lock,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Schedules a rollback checker to get the rollback ID after data cloning or applying. This
* helps us check if a rollback occurred on the sync source.
* If we fail to schedule the rollback checker, we set the error status in 'onCompletionGuard'
* and shut down the OplogFetcher.
* Passes 'lock' through to completion guard.
*/
void _scheduleRollbackCheckerCheckForRollback_inlock(
const stdx::lock_guard<stdx::mutex>& lock,
std::shared_ptr<OnCompletionGuard> onCompletionGuard);
/**
* Checks the given status (or embedded status inside the callback args) and current data
* replicator shutdown state. If the given status is not OK or if we are shutting down, returns
* a new error status that should be passed to _finishCallback. The reason in the new error
* status will include 'message'.
* Otherwise, returns Status::OK().
*/
Status _checkForShutdownAndConvertStatus_inlock(
const executor::TaskExecutor::CallbackArgs& callbackArgs, const std::string& message);
Status _checkForShutdownAndConvertStatus_inlock(const Status& status,
const std::string& message);
/**
* Schedules work to be run by the task executor.
* Saves handle if work was successfully scheduled.
* Returns scheduleWork status (without the handle).
*/
Status _scheduleWorkAndSaveHandle_inlock(const executor::TaskExecutor::CallbackFn& work,
executor::TaskExecutor::CallbackHandle* handle,
const std::string& name);
Status _scheduleWorkAtAndSaveHandle_inlock(Date_t when,
const executor::TaskExecutor::CallbackFn& work,
executor::TaskExecutor::CallbackHandle* handle,
const std::string& name);
/**
* Cancels task executor callback handle if not null.
*/
void _cancelHandle_inlock(executor::TaskExecutor::CallbackHandle handle);
/**
* Starts up component and checks data replicator's shutdown state at the same time.
* If component's startup() fails, resets 'component' (which is assumed to be a unique_ptr
* to the component type).
*/
template <typename Component>
Status _startupComponent_inlock(Component& component);
/**
* Shuts down component if not null.
*/
template <typename Component>
void _shutdownComponent_inlock(Component& component);
// Counts how many documents have been refetched from the source in the current batch.
AtomicUInt32 _fetchCount;
//
// All member variables are labeled with one of the following codes indicating the
// synchronization rules for accessing them.
//
// (R) Read-only in concurrent operation; no synchronization required.
// (S) Self-synchronizing; access in any way from any context.
// (M) Reads and writes guarded by _mutex
// (X) Reads and writes must be performed in a callback in _exec
// (MX) Must hold _mutex and be in a callback in _exec to write; must either hold
// _mutex or be in a callback in _exec to read.
mutable stdx::mutex _mutex; // (S)
const DataReplicatorOptions _opts; // (R)
std::unique_ptr<DataReplicatorExternalState> _dataReplicatorExternalState; // (R)
executor::TaskExecutor* _exec; // (R)
StorageInterface* _storage; // (R)
// This is invoked with the final status of the initial sync. If startup() fails, this callback
// is never invoked. The caller gets the last applied optime with hash when the initial sync
// completes successfully or an error status.
// '_onCompletion' is cleared on completion (in _finishCallback()) in order to release any
// resources that might be held by the callback function object.
OnCompletionFn _onCompletion; // (M)
// Handle to currently scheduled _startInitialSyncAttemptCallback() task.
executor::TaskExecutor::CallbackHandle _startInitialSyncAttemptHandle; // (M)
// Handle to currently scheduled _chooseSyncSourceCallback() task.
executor::TaskExecutor::CallbackHandle _chooseSyncSourceHandle; // (M)
// RollbackChecker to get rollback ID before and after each initial sync attempt.
std::unique_ptr<RollbackChecker> _rollbackChecker; // (M)
// Handle returned from RollbackChecker::reset().
RollbackChecker::CallbackHandle _getBaseRollbackIdHandle; // (M)
// Handle returned from RollbackChecker::checkForRollback().
RollbackChecker::CallbackHandle _getLastRollbackIdHandle; // (M)
// Handle to currently scheduled _getNextApplierBatchCallback() task.
executor::TaskExecutor::CallbackHandle _getNextApplierBatchHandle; // (M)
std::unique_ptr<InitialSyncState> _initialSyncState; // (M)
std::unique_ptr<OplogFetcher> _oplogFetcher; // (S)
std::unique_ptr<Fetcher> _lastOplogEntryFetcher; // (S)
std::unique_ptr<MultiApplier> _applier; // (M)
HostAndPort _syncSource; // (M)
OpTimeWithHash _lastFetched; // (MX)
OpTimeWithHash _lastApplied; // (MX)
std::unique_ptr<OplogBuffer> _oplogBuffer; // (M)
// Used to signal changes in _state.
mutable stdx::condition_variable _stateCondition;
// Current data replicator state. See comments for State enum class for details.
State _state = State::kPreStart; // (M)
// Passed to CollectionCloner via DatabasesCloner.
CollectionCloner::ScheduleDbWorkFn _scheduleDbWorkFn; // (M)
// Contains stats on the current initial sync request (includes all attempts).
// To access these stats in a user-readable format, use getInitialSyncProgress().
Stats _stats; // (M)
};
} // namespace repl
} // namespace mongo
|