1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
|
/**
* 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.
*/
#define MONGO_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kSharding
#include "mongo/platform/basic.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/bson/util/bson_extract.h"
#include "mongo/client/connection_string.h"
#include "mongo/client/replica_set_monitor.h"
#include "mongo/db/auth/authorization_session.h"
#include "mongo/db/client.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbhelpers.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/ops/update.h"
#include "mongo/db/ops/update_lifecycle_impl.h"
#include "mongo/db/repl/optime.h"
#include "mongo/db/repl/replication_coordinator_global.h"
#include "mongo/db/s/collection_metadata.h"
#include "mongo/db/s/collection_sharding_state.h"
#include "mongo/db/s/metadata_loader.h"
#include "mongo/db/s/operation_sharding_state.h"
#include "mongo/db/s/sharded_connection_info.h"
#include "mongo/db/s/sharding_initialization_mongod.h"
#include "mongo/db/s/type_shard_identity.h"
#include "mongo/executor/network_interface_factory.h"
#include "mongo/executor/network_interface_thread_pool.h"
#include "mongo/executor/task_executor_pool.h"
#include "mongo/rpc/metadata/config_server_metadata.h"
#include "mongo/rpc/metadata/metadata_hook.h"
#include "mongo/s/catalog/sharding_catalog_client.h"
#include "mongo/s/catalog/type_chunk.h"
#include "mongo/s/chunk_version.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/client/sharding_network_connection_hook.h"
#include "mongo/s/config.h"
#include "mongo/s/grid.h"
#include "mongo/s/sharding_initialization.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
#include <chrono>
#include <ctime>
#include <iomanip>
#include <iostream>
namespace mongo {
using std::shared_ptr;
using std::string;
using std::vector;
using CallbackArgs = executor::TaskExecutor::CallbackArgs;
namespace {
const auto getShardingState = ServiceContext::declareDecoration<ShardingState>();
// Max number of concurrent config server refresh threads
const int kMaxConfigServerRefreshThreads = 3;
// Maximum number of times to try to refresh the collection metadata if conflicts are occurring
const int kMaxNumMetadataRefreshAttempts = 3;
/**
* Updates the config server field of the shardIdentity document with the given connection string
* if setName is equal to the config server replica set name.
*
* Note: This is intended to be used on a new thread that hasn't called Client::initThread.
* One example use case is for the ReplicaSetMonitor asynchronous callback when it detects changes
* to replica set membership.
*/
void updateShardIdentityConfigStringCB(const string& setName, const string& newConnectionString) {
auto configsvrConnStr = grid.shardRegistry()->getConfigServerConnectionString();
if (configsvrConnStr.getSetName() != setName) {
// Ignore all change notification for other sets that are not the config server.
return;
}
Client::initThread("updateShardIdentityConfigConnString");
auto uniqOpCtx = getGlobalServiceContext()->makeOperationContext(&cc());
auto status = ShardingState::get(uniqOpCtx.get())
->updateShardIdentityConfigString(uniqOpCtx.get(), newConnectionString);
if (!status.isOK() && !ErrorCodes::isNotMasterError(status.code())) {
warning() << "error encountered while trying to update config connection string to "
<< newConnectionString << causedBy(redact(status));
}
}
} // namespace
const std::set<std::string> ShardingState::_commandsThatInitializeShardingAwareness{
"_recvChunkStart", "mergeChunks", "moveChunk", "setShardVersion", "splitChunk"};
ShardingState::ShardingState()
: _initializationState(static_cast<uint32_t>(InitializationState::kNew)),
_initializationStatus(Status(ErrorCodes::InternalError, "Uninitialized value")),
_configServerTickets(kMaxConfigServerRefreshThreads),
_globalInit(&initializeGlobalShardingStateForMongod),
_scheduleWorkFn([](NamespaceString nss) {}) {}
ShardingState::~ShardingState() = default;
ShardingState* ShardingState::get(ServiceContext* serviceContext) {
return &getShardingState(serviceContext);
}
ShardingState* ShardingState::get(OperationContext* operationContext) {
return ShardingState::get(operationContext->getServiceContext());
}
bool ShardingState::enabled() const {
return _getInitializationState() == InitializationState::kInitialized;
}
ConnectionString ShardingState::getConfigServer(OperationContext* txn) {
invariant(enabled());
stdx::lock_guard<stdx::mutex> lk(_mutex);
return Grid::get(txn)->shardRegistry()->getConfigServerConnectionString();
}
string ShardingState::getShardName() {
invariant(enabled());
stdx::lock_guard<stdx::mutex> lk(_mutex);
return _shardName;
}
void ShardingState::shutDown(OperationContext* txn) {
bool mustEnterShutdownState = false;
{
stdx::unique_lock<stdx::mutex> lk(_mutex);
while (_getInitializationState() == InitializationState::kInitializing) {
_initializationFinishedCondition.wait(lk);
}
if (_getInitializationState() == InitializationState::kNew) {
_setInitializationState_inlock(InitializationState::kInitializing);
mustEnterShutdownState = true;
}
}
// Initialization completion must be signalled outside of the mutex
if (mustEnterShutdownState) {
_signalInitializationComplete(
Status(ErrorCodes::ShutdownInProgress,
"Sharding state unavailable because the system is shutting down"));
}
if (_getInitializationState() == InitializationState::kInitialized) {
grid.getExecutorPool()->shutdownAndJoin();
grid.catalogClient(txn)->shutDown(txn);
}
}
Status ShardingState::updateConfigServerOpTimeFromMetadata(OperationContext* txn) {
if (serverGlobalParams.clusterRole == ClusterRole::ConfigServer) {
// Nothing to do if we're a config server ourselves.
return Status::OK();
}
boost::optional<repl::OpTime> opTime = rpc::ConfigServerMetadata::get(txn).getOpTime();
if (opTime) {
if (!AuthorizationSession::get(txn->getClient())
->isAuthorizedForActionsOnResource(ResourcePattern::forClusterResource(),
ActionType::internal)) {
return Status(ErrorCodes::Unauthorized, "Unauthorized to update config opTime");
}
grid.advanceConfigOpTime(*opTime);
}
return Status::OK();
}
CollectionShardingState* ShardingState::getNS(const std::string& ns, OperationContext* txn) {
stdx::lock_guard<stdx::mutex> lk(_mutex);
CollectionShardingStateMap::iterator it = _collections.find(ns);
if (it == _collections.end()) {
auto inserted =
_collections.insert(make_pair(ns,
stdx::make_unique<CollectionShardingState>(
txn->getServiceContext(), NamespaceString(ns))));
invariant(inserted.second);
it = std::move(inserted.first);
}
return it->second.get();
}
void ShardingState::markCollectionsNotShardedAtStepdown() {
stdx::lock_guard<stdx::mutex> lk(_mutex);
for (auto& coll : _collections) {
auto& css = coll.second;
css->markNotShardedAtStepdown();
}
}
void ShardingState::setGlobalInitMethodForTest(GlobalInitFunc func) {
_globalInit = func;
}
void ShardingState::setScheduleCleanupFunctionForTest(RangeDeleterCleanupNotificationFunc fn) {
_scheduleWorkFn = fn;
}
void ShardingState::scheduleCleanup(const NamespaceString& nss) {
_scheduleWorkFn(nss);
}
Status ShardingState::onStaleShardVersion(OperationContext* txn,
const NamespaceString& nss,
const ChunkVersion& expectedVersion) {
invariant(!txn->lockState()->isLocked());
invariant(enabled());
LOG(2) << "metadata refresh requested for " << nss.ns() << " at shard version "
<< expectedVersion;
// Ensure any ongoing migrations have completed
auto& oss = OperationShardingState::get(txn);
oss.waitForMigrationCriticalSectionSignal(txn);
ChunkVersion collectionShardVersion;
// Fast path - check if the requested version is at a higher version than the current metadata
// version or a different epoch before verifying against config server.
ScopedCollectionMetadata currentMetadata;
{
AutoGetCollection autoColl(txn, nss, MODE_IS);
currentMetadata = CollectionShardingState::get(txn, nss)->getMetadata();
if (currentMetadata) {
collectionShardVersion = currentMetadata->getShardVersion();
}
if (collectionShardVersion.epoch() == expectedVersion.epoch() &&
collectionShardVersion >= expectedVersion) {
// Don't need to remotely reload if we're in the same epoch and the requested version is
// smaller than the one we know about. This means that the remote side is behind.
return Status::OK();
}
}
// At the first attempt try to use the currently loaded metadata and on subsequent attempts use
// the complete metadata
int numRefreshAttempts = 0;
while (true) {
numRefreshAttempts++;
auto refreshStatusAndVersion =
_refreshMetadata(txn, nss, (currentMetadata ? currentMetadata.getMetadata() : nullptr));
if (refreshStatusAndVersion.isOK()) {
LOG(1) << "Successfully refreshed metadata for " << nss.ns() << " to "
<< refreshStatusAndVersion.getValue();
return Status::OK();
}
if (refreshStatusAndVersion == ErrorCodes::RemoteChangeDetected &&
numRefreshAttempts < kMaxNumMetadataRefreshAttempts) {
currentMetadata = ScopedCollectionMetadata();
log() << "Refresh failed and will be retried as full reload "
<< refreshStatusAndVersion.getStatus();
continue;
}
return refreshStatusAndVersion.getStatus();
}
MONGO_UNREACHABLE;
}
Status ShardingState::refreshMetadataNow(OperationContext* txn,
const NamespaceString& nss,
ChunkVersion* latestShardVersion) {
auto refreshLatestShardVersionStatus = _refreshMetadata(txn, nss, nullptr);
if (!refreshLatestShardVersionStatus.isOK()) {
return refreshLatestShardVersionStatus.getStatus();
}
*latestShardVersion = refreshLatestShardVersionStatus.getValue();
return Status::OK();
}
void ShardingState::initializeFromConfigConnString(OperationContext* txn,
const string& configSvr,
const string shardName) {
{
stdx::lock_guard<stdx::mutex> lk(_mutex);
if (_getInitializationState() == InitializationState::kNew) {
uassert(18509,
"Unable to obtain host name during sharding initialization.",
!getHostName().empty());
ConnectionString configSvrConnStr = uassertStatusOK(ConnectionString::parse(configSvr));
_setInitializationState_inlock(InitializationState::kInitializing);
stdx::thread thread([this, configSvrConnStr, shardName] {
_initializeImpl(configSvrConnStr, shardName);
});
thread.detach();
}
}
uassertStatusOK(_waitForInitialization(txn->getDeadline()));
uassertStatusOK(reloadShardRegistryUntilSuccess(txn));
uassertStatusOK(updateConfigServerOpTimeFromMetadata(txn));
}
// NOTE: This method can be called inside a database lock so it should never take any database
// locks, perform I/O, or any long running operations.
Status ShardingState::initializeFromShardIdentity(OperationContext* txn,
const ShardIdentityType& shardIdentity) {
invariant(serverGlobalParams.clusterRole == ClusterRole::ShardServer);
Status validationStatus = shardIdentity.validate();
if (!validationStatus.isOK()) {
return Status(
validationStatus.code(),
str::stream()
<< "Invalid shard identity document found when initializing sharding state: "
<< validationStatus.reason());
}
log() << "initializing sharding state with: " << shardIdentity;
stdx::unique_lock<stdx::mutex> lk(_mutex);
// TODO: remove after v3.4.
// This is for backwards compatibility with old style initialization through metadata
// commands/setShardVersion, which can happen concurrently with an insert of a
// shardIdentity document to admin.system.version.
if (_getInitializationState() == InitializationState::kInitializing) {
auto waitStatus = _waitForInitialization_inlock(Date_t::max(), lk);
if (!waitStatus.isOK()) {
return waitStatus;
}
}
if (_getInitializationState() == InitializationState::kError) {
return {ErrorCodes::ManualInterventionRequired,
str::stream() << "Server's sharding metadata manager failed to initialize and will "
"remain in this state until the instance is manually reset"
<< causedBy(_initializationStatus)};
}
auto configSvrConnStr = shardIdentity.getConfigsvrConnString();
// TODO: remove after v3.4.
// This is for backwards compatibility with old style initialization through metadata
// commands/setShardVersion, which sets the shardName and configsvrConnectionString.
if (_getInitializationState() == InitializationState::kInitialized) {
if (_shardName != shardIdentity.getShardName()) {
return {ErrorCodes::InconsistentShardIdentity,
str::stream() << "shard name previously set as " << _shardName
<< " is different from stored: "
<< shardIdentity.getShardName()};
}
auto prevConfigsvrConnStr = grid.shardRegistry()->getConfigServerConnectionString();
if (prevConfigsvrConnStr.type() != ConnectionString::SET) {
return {ErrorCodes::UnsupportedFormat,
str::stream() << "config server connection string was previously initialized as"
" something that is not a replica set: "
<< prevConfigsvrConnStr.toString()};
}
if (prevConfigsvrConnStr.getSetName() != configSvrConnStr.getSetName()) {
return {ErrorCodes::InconsistentShardIdentity,
str::stream() << "config server connection string previously set as "
<< prevConfigsvrConnStr.toString()
<< " is different from stored: "
<< configSvrConnStr.toString()};
}
// The clusterId will only be unset if sharding state was initialized via the sharding
// metadata commands.
if (!_clusterId.isSet()) {
_clusterId = shardIdentity.getClusterId();
} else if (_clusterId != shardIdentity.getClusterId()) {
return {ErrorCodes::InconsistentShardIdentity,
str::stream() << "cluster id previously set as " << _clusterId
<< " is different from stored: "
<< shardIdentity.getClusterId()};
}
return Status::OK();
}
if (_getInitializationState() == InitializationState::kNew) {
ShardedConnectionInfo::addHook();
try {
Status status = _globalInit(txn, configSvrConnStr, generateDistLockProcessId(txn));
// TODO: remove after v3.4.
// This is for backwards compatibility with old style initialization through metadata
// commands/setShardVersion, which can happen concurrently with an insert of a
// shardIdentity document to admin.system.version.
if (status.isOK()) {
_setInitializationState_inlock(InitializationState::kInitialized);
ReplicaSetMonitor::setSynchronousConfigChangeHook(
&ConfigServer::replicaSetChangeShardRegistryUpdateHook);
ReplicaSetMonitor::setAsynchronousConfigChangeHook(
&updateShardIdentityConfigStringCB);
} else {
_initializationStatus = status;
_setInitializationState_inlock(InitializationState::kError);
}
_shardName = shardIdentity.getShardName();
_clusterId = shardIdentity.getClusterId();
_initializeRangeDeleterTaskExecutor();
return status;
} catch (const DBException& ex) {
auto errorStatus = ex.toStatus();
_setInitializationState_inlock(InitializationState::kError);
_initializationStatus = errorStatus;
return errorStatus;
}
}
MONGO_UNREACHABLE;
}
void ShardingState::_initializeImpl(ConnectionString configSvr, string shardName) {
Client::initThread("ShardingState initialization");
auto txn = cc().makeOperationContext();
// Do this initialization outside of the lock, since we are already protected by having entered
// the kInitializing state.
ShardedConnectionInfo::addHook();
try {
Status status = _globalInit(txn.get(), configSvr, generateDistLockProcessId(txn.get()));
if (status.isOK()) {
ReplicaSetMonitor::setSynchronousConfigChangeHook(
&ConfigServer::replicaSetChangeShardRegistryUpdateHook);
ReplicaSetMonitor::setAsynchronousConfigChangeHook(&updateShardIdentityConfigStringCB);
_initializeRangeDeleterTaskExecutor();
_shardName = shardName;
}
_signalInitializationComplete(status);
} catch (const DBException& ex) {
_signalInitializationComplete(ex.toStatus());
}
}
Status ShardingState::_waitForInitialization(Date_t deadline) {
if (enabled())
return Status::OK();
stdx::unique_lock<stdx::mutex> lk(_mutex);
return _waitForInitialization_inlock(deadline, lk);
}
Status ShardingState::_waitForInitialization_inlock(Date_t deadline,
stdx::unique_lock<stdx::mutex>& lk) {
{
while (_getInitializationState() == InitializationState::kInitializing ||
_getInitializationState() == InitializationState::kNew) {
if (deadline == Date_t::max()) {
_initializationFinishedCondition.wait(lk);
} else if (stdx::cv_status::timeout ==
_initializationFinishedCondition.wait_until(lk,
deadline.toSystemTimePoint())) {
return Status(ErrorCodes::ExceededTimeLimit,
"Initializing sharding state exceeded time limit");
}
}
}
auto initializationState = _getInitializationState();
if (initializationState == InitializationState::kInitialized) {
fassertStatusOK(34349, _initializationStatus);
return Status::OK();
}
if (initializationState == InitializationState::kError) {
return Status(ErrorCodes::ManualInterventionRequired,
str::stream()
<< "Server's sharding metadata manager failed to initialize and will "
"remain in this state until the instance is manually reset"
<< causedBy(_initializationStatus));
}
MONGO_UNREACHABLE;
}
ShardingState::InitializationState ShardingState::_getInitializationState() const {
return static_cast<InitializationState>(_initializationState.load());
}
void ShardingState::_setInitializationState_inlock(InitializationState newState) {
_initializationState.store(static_cast<uint32_t>(newState));
}
void ShardingState::_signalInitializationComplete(Status status) {
stdx::lock_guard<stdx::mutex> lk(_mutex);
invariant(_getInitializationState() == InitializationState::kInitializing);
if (!status.isOK()) {
_initializationStatus = status;
_setInitializationState_inlock(InitializationState::kError);
} else {
_initializationStatus = Status::OK();
_setInitializationState_inlock(InitializationState::kInitialized);
}
_initializationFinishedCondition.notify_all();
}
StatusWith<bool> ShardingState::initializeShardingAwarenessIfNeeded(OperationContext* txn) {
// In sharded readOnly mode, we ignore the shardIdentity document on disk and instead *require*
// a shardIdentity document to be passed through --overrideShardIdentity.
if (storageGlobalParams.readOnly) {
if (serverGlobalParams.clusterRole == ClusterRole::ShardServer) {
if (serverGlobalParams.overrideShardIdentity.isEmpty()) {
return {ErrorCodes::InvalidOptions,
"If started with --shardsvr in queryableBackupMode, a shardIdentity "
"document must be provided through --overrideShardIdentity"};
}
auto swOverrideShardIdentity =
ShardIdentityType::fromBSON(serverGlobalParams.overrideShardIdentity);
if (!swOverrideShardIdentity.isOK()) {
return swOverrideShardIdentity.getStatus();
}
auto status = initializeFromShardIdentity(txn, swOverrideShardIdentity.getValue());
if (!status.isOK()) {
return status;
}
return true;
} else {
// Error if --overrideShardIdentity is used but *not* started with --shardsvr.
if (!serverGlobalParams.overrideShardIdentity.isEmpty()) {
return {
ErrorCodes::InvalidOptions,
str::stream()
<< "Not started with --shardsvr, but a shardIdentity document was provided "
"through --overrideShardIdentity: "
<< serverGlobalParams.overrideShardIdentity};
}
return false;
}
}
// In sharded *non*-readOnly mode, error if --overrideShardIdentity is provided. Use the
// shardIdentity document on disk if one exists, but it is okay if no shardIdentity document is
// provided at all (sharding awareness will be initialized when a shardIdentity document is
// inserted).
else {
if (!serverGlobalParams.overrideShardIdentity.isEmpty()) {
return {
ErrorCodes::InvalidOptions,
str::stream() << "--overrideShardIdentity is only allowed in sharded "
"queryableBackupMode. If not in queryableBackupMode, you can edit "
"the shardIdentity document by starting the server *without* "
"--shardsvr, manually updating the shardIdentity document in the "
<< NamespaceString::kConfigCollectionNamespace.toString()
<< " collection, and restarting the server with --shardsvr."};
}
// Load the shardIdentity document from disk.
invariant(!txn->lockState()->isLocked());
BSONObj shardIdentityBSON;
try {
AutoGetCollection autoColl(txn, NamespaceString::kConfigCollectionNamespace, MODE_IS);
Helpers::findOne(txn,
autoColl.getCollection(),
BSON("_id" << ShardIdentityType::IdName),
shardIdentityBSON);
} catch (const DBException& ex) {
return ex.toStatus();
}
if (serverGlobalParams.clusterRole == ClusterRole::ShardServer) {
if (shardIdentityBSON.isEmpty()) {
warning() << "Started with --shardsvr, but no shardIdentity document was found on "
"disk in "
<< NamespaceString::kConfigCollectionNamespace
<< ". This most likely means this server has not yet been added to a "
"sharded cluster.";
return false;
}
auto swShardIdentity = ShardIdentityType::fromBSON(shardIdentityBSON);
if (!swShardIdentity.isOK()) {
return swShardIdentity.getStatus();
}
auto status = initializeFromShardIdentity(txn, swShardIdentity.getValue());
if (!status.isOK()) {
return status;
}
return true;
} else {
// Warn if a shardIdentity document is found on disk but *not* started with --shardsvr.
if (!shardIdentityBSON.isEmpty()) {
warning() << "Not started with --shardsvr, but a shardIdentity document was found "
"on disk in "
<< NamespaceString::kConfigCollectionNamespace << ": "
<< shardIdentityBSON;
}
return false;
}
}
}
StatusWith<ChunkVersion> ShardingState::_refreshMetadata(
OperationContext* txn, const NamespaceString& nss, const CollectionMetadata* metadataForDiff) {
invariant(!txn->lockState()->isLocked());
{
Status status = _waitForInitialization(txn->getDeadline());
if (!status.isOK())
return status;
}
// We can't reload if a shard name has not yet been set
{
stdx::lock_guard<stdx::mutex> lk(_mutex);
if (_shardName.empty()) {
string errMsg = str::stream() << "cannot refresh metadata for " << nss.ns()
<< " before shard name has been set";
warning() << errMsg;
return {ErrorCodes::NotYetInitialized, errMsg};
}
}
// The _configServerTickets serializes this process such that only a small number of threads can
// try to refresh at the same time in order to avoid overloading the config server
_configServerTickets.waitForTicket();
TicketHolderReleaser needTicketFrom(&_configServerTickets);
Timer t;
log() << "MetadataLoader loading chunks for " << nss.ns() << " based on: "
<< (metadataForDiff ? metadataForDiff->getCollVersion().toString() : "(empty)");
std::unique_ptr<CollectionMetadata> remoteMetadata(stdx::make_unique<CollectionMetadata>());
Status status = MetadataLoader::makeCollectionMetadata(txn,
grid.catalogClient(txn),
nss.ns(),
getShardName(),
metadataForDiff,
remoteMetadata.get());
if (!status.isOK() && status != ErrorCodes::NamespaceNotFound) {
warning() << "MetadataLoader failed after " << t.millis() << " ms"
<< causedBy(redact(status));
return status;
}
// Exclusive collection lock needed since we're now changing the metadata
ScopedTransaction transaction(txn, MODE_IX);
AutoGetCollection autoColl(txn, nss, MODE_IX, MODE_X);
auto css = CollectionShardingState::get(txn, nss);
if (status.isOK()) {
css->refreshMetadata(txn, std::move(remoteMetadata));
auto metadata = css->getMetadata();
log() << "MetadataLoader took " << t.millis() << " ms and found version "
<< metadata->getCollVersion();
return metadata->getShardVersion();
}
invariant(status == ErrorCodes::NamespaceNotFound);
css->refreshMetadata(txn, nullptr);
log() << "MetadataLoader took " << t.millis() << " ms and did not find the namespace";
return ChunkVersion::UNSHARDED();
}
StatusWith<ScopedRegisterDonateChunk> ShardingState::registerDonateChunk(
const MoveChunkRequest& args) {
return _activeMigrationsRegistry.registerDonateChunk(args);
}
StatusWith<ScopedRegisterReceiveChunk> ShardingState::registerReceiveChunk(
const NamespaceString& nss, const ChunkRange& chunkRange, const ShardId& fromShardId) {
return _activeMigrationsRegistry.registerReceiveChunk(nss, chunkRange, fromShardId);
}
boost::optional<NamespaceString> ShardingState::getActiveDonateChunkNss() {
return _activeMigrationsRegistry.getActiveDonateChunkNss();
}
BSONObj ShardingState::getActiveMigrationStatusReport(OperationContext* txn) {
return _activeMigrationsRegistry.getActiveMigrationStatusReport(txn);
}
void ShardingState::appendInfo(OperationContext* txn, BSONObjBuilder& builder) {
const bool isEnabled = enabled();
builder.appendBool("enabled", isEnabled);
if (!isEnabled)
return;
stdx::lock_guard<stdx::mutex> lk(_mutex);
builder.append("configServer",
grid.shardRegistry()->getConfigServerConnectionString().toString());
builder.append("shardName", _shardName);
builder.append("clusterId", _clusterId);
BSONObjBuilder versionB(builder.subobjStart("versions"));
for (CollectionShardingStateMap::const_iterator it = _collections.begin();
it != _collections.end();
++it) {
ScopedCollectionMetadata metadata = it->second->getMetadata();
if (metadata) {
versionB.appendTimestamp(it->first, metadata->getShardVersion().toLong());
} else {
versionB.appendTimestamp(it->first, ChunkVersion::UNSHARDED().toLong());
}
}
versionB.done();
}
bool ShardingState::needCollectionMetadata(OperationContext* txn, const string& ns) {
if (!enabled())
return false;
Client* client = txn->getClient();
// Shard version information received from mongos may either by attached to the Client or
// directly to the OperationContext.
return ShardedConnectionInfo::get(client, false) ||
OperationShardingState::get(txn).hasShardVersion();
}
Status ShardingState::updateShardIdentityConfigString(OperationContext* txn,
const std::string& newConnectionString) {
BSONObj updateObj(ShardIdentityType::createConfigServerUpdateObject(newConnectionString));
UpdateRequest updateReq(NamespaceString::kConfigCollectionNamespace);
updateReq.setQuery(BSON("_id" << ShardIdentityType::IdName));
updateReq.setUpdates(updateObj);
UpdateLifecycleImpl updateLifecycle(NamespaceString::kConfigCollectionNamespace);
updateReq.setLifecycle(&updateLifecycle);
try {
AutoGetOrCreateDb autoDb(txn, NamespaceString::kConfigCollectionNamespace.db(), MODE_X);
auto result = update(txn, autoDb.getDb(), updateReq);
if (result.numMatched == 0) {
warning() << "failed to update config string of shard identity document because "
<< "it does not exist. This shard could have been removed from the cluster";
} else {
LOG(2) << "Updated config server connection string in shardIdentity document to"
<< newConnectionString;
}
} catch (const DBException& exception) {
return exception.toStatus();
}
return Status::OK();
}
void ShardingState::_initializeRangeDeleterTaskExecutor() {
invariant(!_rangeDeleterTaskExecutor);
auto net =
executor::makeNetworkInterface("NetworkInterfaceCollectionRangeDeleter-TaskExecutor");
auto netPtr = net.get();
_rangeDeleterTaskExecutor = stdx::make_unique<executor::ThreadPoolTaskExecutor>(
stdx::make_unique<executor::NetworkInterfaceThreadPool>(netPtr), std::move(net));
}
executor::ThreadPoolTaskExecutor* ShardingState::getRangeDeleterTaskExecutor() {
return _rangeDeleterTaskExecutor.get();
}
/**
* Global free function.
*/
bool isMongos() {
return false;
}
} // namespace mongo
|