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
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
|
/**
* Copyright (C) 2013 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::kWrite
#include "mongo/platform/basic.h"
#include "mongo/db/commands/write_commands/batch_executor.h"
#include <boost/scoped_ptr.hpp>
#include <memory>
#include "mongo/base/error_codes.h"
#include "mongo/db/catalog/database.h"
#include "mongo/db/clientcursor.h"
#include "mongo/db/commands.h"
#include "mongo/db/instance.h"
#include "mongo/db/introspect.h"
#include "mongo/db/lasterror.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/catalog/database_holder.h"
#include "mongo/db/catalog/index_create.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/exec/delete.h"
#include "mongo/db/exec/update.h"
#include "mongo/db/ops/delete_request.h"
#include "mongo/db/ops/parsed_delete.h"
#include "mongo/db/ops/parsed_update.h"
#include "mongo/db/ops/insert.h"
#include "mongo/db/ops/update_lifecycle_impl.h"
#include "mongo/db/query/get_executor.h"
#include "mongo/db/query/plan_executor.h"
#include "mongo/db/query/query_knobs.h"
#include "mongo/db/repl/oplog.h"
#include "mongo/db/repl/repl_settings.h"
#include "mongo/db/repl/replication_coordinator_global.h"
#include "mongo/db/server_parameters.h"
#include "mongo/db/stats/counters.h"
#include "mongo/db/operation_context_impl.h"
#include "mongo/db/write_concern.h"
#include "mongo/s/collection_metadata.h"
#include "mongo/s/d_state.h"
#include "mongo/s/shard_key_pattern.h"
#include "mongo/s/write_ops/batched_upsert_detail.h"
#include "mongo/s/write_ops/write_error_detail.h"
#include "mongo/util/elapsed_tracker.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
namespace mongo {
using boost::scoped_ptr;
using std::auto_ptr;
using std::endl;
using std::string;
using std::vector;
namespace {
/**
* Data structure to safely hold and clean up results of single write operations.
*/
class WriteOpResult {
MONGO_DISALLOW_COPYING(WriteOpResult);
public:
WriteOpResult() {}
WriteOpStats& getStats() { return _stats; }
WriteErrorDetail* getError() { return _error.get(); }
WriteErrorDetail* releaseError() { return _error.release(); }
void setError(WriteErrorDetail* error) { _error.reset(error); }
private:
WriteOpStats _stats;
std::auto_ptr<WriteErrorDetail> _error;
};
} // namespace
// TODO: Determine queueing behavior we want here
MONGO_EXPORT_SERVER_PARAMETER( queueForMigrationCommit, bool, true );
using mongoutils::str::stream;
WriteBatchExecutor::WriteBatchExecutor( OperationContext* txn,
const WriteConcernOptions& wc,
OpCounters* opCounters,
LastError* le ) :
_txn(txn),
_defaultWriteConcern( wc ),
_opCounters( opCounters ),
_le( le ),
_stats( new WriteBatchStats ) {
}
static WCErrorDetail* toWriteConcernError( const Status& wcStatus,
const WriteConcernResult& wcResult ) {
WCErrorDetail* wcError = new WCErrorDetail;
wcError->setErrCode( wcStatus.code() );
wcError->setErrMessage( wcStatus.reason() );
if ( wcResult.wTimedOut )
wcError->setErrInfo( BSON( "wtimeout" << true ) );
return wcError;
}
static WriteErrorDetail* toWriteError( const Status& status ) {
WriteErrorDetail* error = new WriteErrorDetail;
// TODO: Complex transform here?
error->setErrCode( status.code() );
error->setErrMessage( status.reason() );
return error;
}
static void toBatchError( const Status& status, BatchedCommandResponse* response ) {
response->clear();
response->setErrCode( status.code() );
response->setErrMessage( status.reason() );
response->setOk( false );
dassert( response->isValid(NULL) );
}
static void noteInCriticalSection( WriteErrorDetail* staleError ) {
BSONObjBuilder builder;
if ( staleError->isErrInfoSet() )
builder.appendElements( staleError->getErrInfo() );
builder.append( "inCriticalSection", true );
staleError->setErrInfo( builder.obj() );
}
// static
Status WriteBatchExecutor::validateBatch( const BatchedCommandRequest& request ) {
// Validate namespace
const NamespaceString& nss = request.getNSS();
if ( !nss.isValid() ) {
return Status( ErrorCodes::InvalidNamespace,
nss.ns() + " is not a valid namespace" );
}
// Make sure we can write to the namespace
Status allowedStatus = userAllowedWriteNS( nss );
if ( !allowedStatus.isOK() ) {
return allowedStatus;
}
// Validate insert index requests
// TODO: Push insert index requests through createIndex once all upgrade paths support it
string errMsg;
if ( request.isInsertIndexRequest() && !request.isValidIndexRequest( &errMsg ) ) {
return Status( ErrorCodes::InvalidOptions, errMsg );
}
return Status::OK();
}
void WriteBatchExecutor::executeBatch( const BatchedCommandRequest& request,
BatchedCommandResponse* response ) {
// Validate namespace
Status isValid = validateBatch(request);
if (!isValid.isOK()) {
toBatchError( isValid, response );
return;
}
// Validate write concern
// TODO: Lift write concern parsing out of this entirely
WriteConcernOptions writeConcern;
BSONObj wcDoc;
if ( request.isWriteConcernSet() ) {
wcDoc = request.getWriteConcern();
}
Status wcStatus = Status::OK();
if ( wcDoc.isEmpty() ) {
// The default write concern if empty is w : 1
// Specifying w : 0 is/was allowed, but is interpreted identically to w : 1
writeConcern = _defaultWriteConcern;
if ( writeConcern.wNumNodes == 0 && writeConcern.wMode.empty() ) {
writeConcern.wNumNodes = 1;
}
}
else {
wcStatus = writeConcern.parse( wcDoc );
}
if ( wcStatus.isOK() ) {
wcStatus = validateWriteConcern( writeConcern );
}
if ( !wcStatus.isOK() ) {
toBatchError( wcStatus, response );
return;
}
if ( writeConcern.syncMode == WriteConcernOptions::JOURNAL ||
writeConcern.syncMode == WriteConcernOptions::FSYNC ) {
_txn->recoveryUnit()->goingToAwaitCommit();
}
if ( request.sizeWriteOps() == 0u ) {
toBatchError( Status( ErrorCodes::InvalidLength,
"no write ops were included in the batch" ),
response );
return;
}
// Validate batch size
if ( request.sizeWriteOps() > BatchedCommandRequest::kMaxWriteBatchSize ) {
toBatchError( Status( ErrorCodes::InvalidLength,
stream() << "exceeded maximum write batch size of "
<< BatchedCommandRequest::kMaxWriteBatchSize ),
response );
return;
}
//
// End validation
//
bool silentWC = writeConcern.wMode.empty() && writeConcern.wNumNodes == 0
&& writeConcern.syncMode == WriteConcernOptions::NONE;
Timer commandTimer;
OwnedPointerVector<WriteErrorDetail> writeErrorsOwned;
vector<WriteErrorDetail*>& writeErrors = writeErrorsOwned.mutableVector();
OwnedPointerVector<BatchedUpsertDetail> upsertedOwned;
vector<BatchedUpsertDetail*>& upserted = upsertedOwned.mutableVector();
//
// Apply each batch item, possibly bulking some items together in the write lock.
// Stops on error if batch is ordered.
//
bulkExecute( request, &upserted, &writeErrors );
//
// Try to enforce the write concern if everything succeeded (unordered or ordered)
// OR if something succeeded and we're unordered.
//
auto_ptr<WCErrorDetail> wcError;
bool needToEnforceWC = writeErrors.empty()
|| ( !request.getOrdered()
&& writeErrors.size() < request.sizeWriteOps() );
if ( needToEnforceWC ) {
_txn->getCurOp()->setMessage( "waiting for write concern" );
WriteConcernResult res;
Status status = waitForWriteConcern( _txn, writeConcern, _txn->getClient()->getLastOp(), &res );
if ( !status.isOK() ) {
wcError.reset( toWriteConcernError( status, res ) );
}
}
//
// Refresh metadata if needed
//
bool staleBatch = !writeErrors.empty()
&& writeErrors.back()->getErrCode() == ErrorCodes::StaleShardVersion;
if ( staleBatch ) {
const BatchedRequestMetadata* requestMetadata = request.getMetadata();
dassert( requestMetadata );
// Make sure our shard name is set or is the same as what was set previously
if ( shardingState.setShardName( requestMetadata->getShardName() ) ) {
//
// First, we refresh metadata if we need to based on the requested version.
//
ChunkVersion latestShardVersion;
shardingState.refreshMetadataIfNeeded( _txn,
request.getTargetingNS(),
requestMetadata->getShardVersion(),
&latestShardVersion );
// Report if we're still changing our metadata
// TODO: Better reporting per-collection
if ( shardingState.inCriticalMigrateSection() ) {
noteInCriticalSection( writeErrors.back() );
}
if ( queueForMigrationCommit ) {
//
// Queue up for migration to end - this allows us to be sure that clients will
// not repeatedly try to refresh metadata that is not yet written to the config
// server. Not necessary for correctness.
// Exposed as optional parameter to allow testing of queuing behavior with
// different network timings.
//
const ChunkVersion& requestShardVersion = requestMetadata->getShardVersion();
//
// Only wait if we're an older version (in the current collection epoch) and
// we're not write compatible, implying that the current migration is affecting
// writes.
//
if ( requestShardVersion.isOlderThan( latestShardVersion ) &&
!requestShardVersion.isWriteCompatibleWith( latestShardVersion ) ) {
while ( shardingState.inCriticalMigrateSection() ) {
log() << "write request to old shard version "
<< requestMetadata->getShardVersion().toString()
<< " waiting for migration commit" << endl;
shardingState.waitTillNotInCriticalSection( 10 /* secs */);
}
}
}
}
else {
// If our shard name is stale, our version must have been stale as well
dassert( writeErrors.size() == request.sizeWriteOps() );
}
}
//
// Construct response
//
response->setOk( true );
if ( !silentWC ) {
if ( upserted.size() ) {
response->setUpsertDetails( upserted );
}
if ( writeErrors.size() ) {
response->setErrDetails( writeErrors );
}
if ( wcError.get() ) {
response->setWriteConcernError( wcError.release() );
}
repl::ReplicationCoordinator* replCoord = repl::getGlobalReplicationCoordinator();
const repl::ReplicationCoordinator::Mode replMode = replCoord->getReplicationMode();
if (replMode != repl::ReplicationCoordinator::modeNone) {
response->setLastOp( _txn->getClient()->getLastOp() );
if (replMode == repl::ReplicationCoordinator::modeReplSet) {
response->setElectionId(replCoord->getElectionId());
}
}
// Set the stats for the response
response->setN( _stats->numInserted + _stats->numUpserted + _stats->numMatched
+ _stats->numDeleted );
if ( request.getBatchType() == BatchedCommandRequest::BatchType_Update )
response->setNModified( _stats->numModified );
}
dassert( response->isValid( NULL ) );
}
// Translates write item type to wire protocol op code.
// Helper for WriteBatchExecutor::applyWriteItem().
static int getOpCode( BatchedCommandRequest::BatchType writeType ) {
switch ( writeType ) {
case BatchedCommandRequest::BatchType_Insert:
return dbInsert;
case BatchedCommandRequest::BatchType_Update:
return dbUpdate;
default:
dassert( writeType == BatchedCommandRequest::BatchType_Delete );
return dbDelete;
}
return 0;
}
static void buildStaleError( const ChunkVersion& shardVersionRecvd,
const ChunkVersion& shardVersionWanted,
WriteErrorDetail* error ) {
// Write stale error to results
error->setErrCode( ErrorCodes::StaleShardVersion );
BSONObjBuilder infoB;
shardVersionWanted.addToBSON( infoB, "vWanted" );
error->setErrInfo( infoB.obj() );
string errMsg = stream() << "stale shard version detected before write, received "
<< shardVersionRecvd.toString() << " but local version is "
<< shardVersionWanted.toString();
error->setErrMessage( errMsg );
}
static bool checkShardVersion(OperationContext* txn,
ShardingState* shardingState,
const BatchedCommandRequest& request,
WriteOpResult* result) {
const NamespaceString& nss = request.getTargetingNSS();
dassert(txn->lockState()->isCollectionLockedForMode(nss.ns(), MODE_IX));
ChunkVersion requestShardVersion =
request.isMetadataSet() && request.getMetadata()->isShardVersionSet() ?
request.getMetadata()->getShardVersion() : ChunkVersion::IGNORED();
if ( shardingState->enabled() ) {
CollectionMetadataPtr metadata = shardingState->getCollectionMetadata( nss.ns() );
if ( !ChunkVersion::isIgnoredVersion( requestShardVersion ) ) {
ChunkVersion shardVersion =
metadata ? metadata->getShardVersion() : ChunkVersion::UNSHARDED();
if ( !requestShardVersion.isWriteCompatibleWith( shardVersion ) ) {
result->setError(new WriteErrorDetail);
buildStaleError(requestShardVersion, shardVersion, result->getError());
return false;
}
}
}
return true;
}
static bool checkIsMasterForDatabase(const NamespaceString& ns, WriteOpResult* result) {
if (!repl::getGlobalReplicationCoordinator()->canAcceptWritesForDatabase(
ns.db())) {
WriteErrorDetail* errorDetail = new WriteErrorDetail;
result->setError(errorDetail);
errorDetail->setErrCode(ErrorCodes::NotMaster);
errorDetail->setErrMessage("Not primary while writing to " + ns.toString());
return false;
}
return true;
}
static void buildUniqueIndexError( const BSONObj& keyPattern,
const BSONObj& indexPattern,
WriteErrorDetail* error ) {
error->setErrCode( ErrorCodes::CannotCreateIndex );
string errMsg = stream() << "cannot create unique index over " << indexPattern
<< " with shard key pattern " << keyPattern;
error->setErrMessage( errMsg );
}
static bool checkIndexConstraints(OperationContext* txn,
ShardingState* shardingState,
const BatchedCommandRequest& request,
WriteOpResult* result) {
const NamespaceString& nss = request.getTargetingNSS();
dassert(txn->lockState()->isCollectionLockedForMode(nss.ns(), MODE_IX));
if ( !request.isUniqueIndexRequest() )
return true;
if ( shardingState->enabled() ) {
CollectionMetadataPtr metadata = shardingState->getCollectionMetadata( nss.ns() );
if ( metadata ) {
ShardKeyPattern shardKeyPattern(metadata->getKeyPattern());
if (!shardKeyPattern.isUniqueIndexCompatible(request.getIndexKeyPattern())) {
result->setError(new WriteErrorDetail);
buildUniqueIndexError(metadata->getKeyPattern(),
request.getIndexKeyPattern(),
result->getError());
return false;
}
}
}
return true;
}
//
// HELPERS FOR CUROP MANAGEMENT AND GLOBAL STATS
//
static void beginCurrentOp( CurOp* currentOp, Client* client, const BatchItemRef& currWrite ) {
// Execute the write item as a child operation of the current operation.
// This is not done by out callers
// Set up the child op with more info
HostAndPort remote =
client->hasRemote() ? client->getRemote() : HostAndPort( "0.0.0.0", 0 );
// TODO Modify CurOp "wrapped" constructor to take an opcode, so calling .reset()
// is unneeded
currentOp->reset( remote, getOpCode( currWrite.getRequest()->getBatchType() ) );
currentOp->ensureStarted();
currentOp->setNS( currWrite.getRequest()->getNS() );
currentOp->debug().ns = currentOp->getNS();
currentOp->debug().op = currentOp->getOp();
if ( currWrite.getOpType() == BatchedCommandRequest::BatchType_Insert ) {
currentOp->setQuery( currWrite.getDocument() );
currentOp->debug().query = currWrite.getDocument();
currentOp->debug().ninserted = 0;
}
else if ( currWrite.getOpType() == BatchedCommandRequest::BatchType_Update ) {
currentOp->setQuery( currWrite.getUpdate()->getQuery() );
currentOp->debug().query = currWrite.getUpdate()->getQuery();
currentOp->debug().updateobj = currWrite.getUpdate()->getUpdateExpr();
// Note: debug().nMatched, nModified and nmoved are set internally in update
}
else {
dassert( currWrite.getOpType() == BatchedCommandRequest::BatchType_Delete );
currentOp->setQuery( currWrite.getDelete()->getQuery() );
currentOp->debug().query = currWrite.getDelete()->getQuery();
currentOp->debug().ndeleted = 0;
}
}
void WriteBatchExecutor::incOpStats( const BatchItemRef& currWrite ) {
if ( currWrite.getOpType() == BatchedCommandRequest::BatchType_Insert ) {
_opCounters->gotInsert();
}
else if ( currWrite.getOpType() == BatchedCommandRequest::BatchType_Update ) {
_opCounters->gotUpdate();
}
else {
dassert( currWrite.getOpType() == BatchedCommandRequest::BatchType_Delete );
_opCounters->gotDelete();
}
}
void WriteBatchExecutor::incWriteStats( const BatchItemRef& currWrite,
const WriteOpStats& stats,
const WriteErrorDetail* error,
CurOp* currentOp ) {
if ( currWrite.getOpType() == BatchedCommandRequest::BatchType_Insert ) {
_stats->numInserted += stats.n;
_le->nObjects = stats.n;
currentOp->debug().ninserted += stats.n;
}
else if ( currWrite.getOpType() == BatchedCommandRequest::BatchType_Update ) {
if ( stats.upsertedID.isEmpty() ) {
_stats->numMatched += stats.n;
_stats->numModified += stats.nModified;
}
else {
++_stats->numUpserted;
}
if ( !error ) {
_le->recordUpdate( stats.upsertedID.isEmpty() && stats.n > 0,
stats.n,
stats.upsertedID );
}
}
else {
dassert( currWrite.getOpType() == BatchedCommandRequest::BatchType_Delete );
_stats->numDeleted += stats.n;
if ( !error ) {
_le->recordDelete( stats.n );
}
currentOp->debug().ndeleted += stats.n;
}
if (error && !_le->disabled) {
_le->raiseError(error->getErrCode(), error->getErrMessage().c_str());
}
}
static void finishCurrentOp( OperationContext* txn,
CurOp* currentOp,
WriteErrorDetail* opError ) {
currentOp->done();
int executionTime = currentOp->debug().executionTime = currentOp->totalTimeMillis();
currentOp->debug().recordStats();
if (currentOp->getOp() == dbInsert) {
// This is a wrapped operation, so make sure to count this part of the op
// SERVER-13339: Properly fix the handling of context in the insert path.
// Right now it caches client contexts in ExecInsertsState, unlike the
// update and remove operations.
currentOp->recordGlobalTime(txn->lockState()->isWriteLocked(),
currentOp->totalTimeMicros());
}
if ( opError ) {
currentOp->debug().exceptionInfo = ExceptionInfo( opError->getErrMessage(),
opError->getErrCode() );
LOG(3) << " Caught Assertion in " << opToString( currentOp->getOp() )
<< ", continuing " << causedBy( opError->getErrMessage() ) << endl;
}
bool logAll = logger::globalLogDomain()->shouldLog(logger::LogComponent::kWrite,
logger::LogSeverity::Debug(1));
bool logSlow = executionTime
> ( serverGlobalParams.slowMS + currentOp->getExpectedLatencyMs() );
if ( logAll || logSlow ) {
LOG(0) << currentOp->debug().report( *currentOp ) << endl;
}
if ( currentOp->shouldDBProfile( executionTime ) ) {
profile( txn, *txn->getClient(), currentOp->getOp(), *currentOp );
}
}
// END HELPERS
//
// CORE WRITE OPERATIONS (declaration)
// These functions write to the database and return stats and zero or one of:
// - page fault
// - error
//
static void singleInsert( OperationContext* txn,
const BSONObj& docToInsert,
Collection* collection,
WriteOpResult* result );
static void singleCreateIndex( OperationContext* txn,
const BSONObj& indexDesc,
WriteOpResult* result );
static void multiUpdate( OperationContext* txn,
const BatchItemRef& updateItem,
WriteOpResult* result );
static void multiRemove( OperationContext* txn,
const BatchItemRef& removeItem,
WriteOpResult* result );
//
// WRITE EXECUTION
// In general, the exec* operations manage db lock state and stats before dispatching to the
// core write operations, which are *only* responsible for performing a write and reporting
// success or failure.
//
/**
* Representation of the execution state of execInserts. Used by a single
* execution of execInserts in a single thread.
*/
class WriteBatchExecutor::ExecInsertsState {
MONGO_DISALLOW_COPYING(ExecInsertsState);
public:
/**
* Constructs a new instance, for performing inserts described in "aRequest".
*/
explicit ExecInsertsState(OperationContext* txn,
const BatchedCommandRequest* aRequest);
/**
* Acquires the write lock and client context needed to perform the current write operation.
* Returns true on success, after which it is safe to use the "context" and "collection"
* members. It is safe to call this function if this instance already holds the write lock.
*
* On failure, writeLock, context and collection will be NULL/clear.
*/
bool lockAndCheck(WriteOpResult* result);
/**
* Releases the client context and write lock acquired by lockAndCheck. Safe to call
* regardless of whether or not this state object currently owns the lock.
*/
void unlock();
/**
* Returns true if this executor has the lock on the target database.
*/
bool hasLock() { return _writeLock.get(); }
/**
* Gets the lock-holding object. Only valid if hasLock().
*/
Lock::DBLock& getLock() { return *_writeLock; }
/**
* Gets the target collection for the batch operation. Value is undefined
* unless hasLock() is true.
*/
Collection* getCollection() { return _collection; }
OperationContext* txn;
// Request object describing the inserts.
const BatchedCommandRequest* request;
// Index of the current insert operation to perform.
size_t currIndex;
// Translation of insert documents in "request" into insert-ready forms. This vector has a
// correspondence with elements of the "request", and "currIndex" is used to
// index both.
std::vector<StatusWith<BSONObj> > normalizedInserts;
private:
bool _lockAndCheckImpl(WriteOpResult* result, bool intentLock=true);
ScopedTransaction _transaction;
// Guard object for the write lock on the target database.
scoped_ptr<Lock::DBLock> _writeLock;
scoped_ptr<Lock::CollectionLock> _collLock;
// Context object on the target database. Must appear after writeLock, so that it is
// destroyed in proper order.
scoped_ptr<Client::Context> _context;
// Target collection.
Collection* _collection;
};
void WriteBatchExecutor::bulkExecute( const BatchedCommandRequest& request,
std::vector<BatchedUpsertDetail*>* upsertedIds,
std::vector<WriteErrorDetail*>* errors ) {
if ( request.getBatchType() == BatchedCommandRequest::BatchType_Insert ) {
execInserts( request, errors );
}
else if ( request.getBatchType() == BatchedCommandRequest::BatchType_Update ) {
for ( size_t i = 0; i < request.sizeWriteOps(); i++ ) {
WriteErrorDetail* error = NULL;
BSONObj upsertedId;
execUpdate( BatchItemRef( &request, i ), &upsertedId, &error );
if ( !upsertedId.isEmpty() ) {
BatchedUpsertDetail* batchUpsertedId = new BatchedUpsertDetail;
batchUpsertedId->setIndex( i );
batchUpsertedId->setUpsertedID( upsertedId );
upsertedIds->push_back( batchUpsertedId );
}
if ( error ) {
errors->push_back( error );
if ( request.getOrdered() )
break;
}
}
}
else {
dassert( request.getBatchType() == BatchedCommandRequest::BatchType_Delete );
for ( size_t i = 0; i < request.sizeWriteOps(); i++ ) {
WriteErrorDetail* error = NULL;
execRemove( BatchItemRef( &request, i ), &error );
if ( error ) {
errors->push_back( error );
if ( request.getOrdered() )
break;
}
}
}
// Fill in stale version errors for unordered batches (update/delete can't do this on own)
if ( !errors->empty() && !request.getOrdered() ) {
const WriteErrorDetail* finalError = errors->back();
if ( finalError->getErrCode() == ErrorCodes::StaleShardVersion ) {
for ( size_t i = finalError->getIndex() + 1; i < request.sizeWriteOps(); i++ ) {
WriteErrorDetail* dupStaleError = new WriteErrorDetail;
finalError->cloneTo( dupStaleError );
errors->push_back( dupStaleError );
}
}
}
}
// Goes over the request and preprocesses normalized versions of all the inserts in the request
static void normalizeInserts( const BatchedCommandRequest& request,
vector<StatusWith<BSONObj> >* normalizedInserts ) {
normalizedInserts->reserve(request.sizeWriteOps());
for ( size_t i = 0; i < request.sizeWriteOps(); ++i ) {
BSONObj insertDoc = request.getInsertRequest()->getDocumentsAt( i );
StatusWith<BSONObj> normalInsert = fixDocumentForInsert( insertDoc );
normalizedInserts->push_back( normalInsert );
if ( request.getOrdered() && !normalInsert.isOK() )
break;
}
}
void WriteBatchExecutor::execInserts( const BatchedCommandRequest& request,
std::vector<WriteErrorDetail*>* errors ) {
// Theory of operation:
//
// Instantiates an ExecInsertsState, which represents all of the state involved in the batch
// insert execution algorithm. Most importantly, encapsulates the lock state.
//
// Every iteration of the loop in execInserts() processes one document insertion, by calling
// insertOne() exactly once for a given value of state.currIndex.
//
// If the ExecInsertsState indicates that the requisite write locks are not held, insertOne
// acquires them and performs lock-acquisition-time checks. However, on non-error
// execution, it does not release the locks. Therefore, the yielding logic in the while
// loop in execInserts() is solely responsible for lock release in the non-error case.
//
// Internally, insertOne loops performing the single insert until it completes without a
// PageFaultException, or until it fails with some kind of error. Errors are mostly
// propagated via the request->error field, but DBExceptions or std::exceptions may escape,
// particularly on operation interruption. These kinds of errors necessarily prevent
// further insertOne calls, and stop the batch. As a result, the only expected source of
// such exceptions are interruptions.
ExecInsertsState state(_txn, &request);
normalizeInserts(request, &state.normalizedInserts);
// Yield frequency is based on the same constants used by PlanYieldPolicy.
ElapsedTracker elapsedTracker(internalQueryExecYieldIterations,
internalQueryExecYieldPeriodMS);
for (state.currIndex = 0;
state.currIndex < state.request->sizeWriteOps();
++state.currIndex) {
if (elapsedTracker.intervalHasElapsed()) {
// Yield between inserts.
if (state.hasLock()) {
// Release our locks. They get reacquired when insertOne() calls
// ExecInsertsState::lockAndCheck(). Since the lock manager guarantees FIFO
// queues waiting on locks, there is no need to explicitly sleep or give up
// control of the processor here.
state.unlock();
// This releases any storage engine held locks/snapshots.
_txn->recoveryUnit()->commitAndRestart();
}
_txn->checkForInterrupt();
elapsedTracker.resetLastTime();
}
WriteErrorDetail* error = NULL;
execOneInsert(&state, &error);
if (error) {
errors->push_back(error);
error->setIndex(state.currIndex);
if (request.getOrdered())
return;
}
}
}
void WriteBatchExecutor::execUpdate( const BatchItemRef& updateItem,
BSONObj* upsertedId,
WriteErrorDetail** error ) {
// BEGIN CURRENT OP
CurOp currentOp( _txn->getClient(), _txn->getClient()->curop() );
beginCurrentOp( ¤tOp, _txn->getClient(), updateItem );
incOpStats( updateItem );
WriteOpResult result;
multiUpdate( _txn, updateItem, &result );
if ( !result.getStats().upsertedID.isEmpty() ) {
*upsertedId = result.getStats().upsertedID;
}
// END CURRENT OP
incWriteStats( updateItem, result.getStats(), result.getError(), ¤tOp );
finishCurrentOp( _txn, ¤tOp, result.getError() );
// End current transaction and release snapshot.
_txn->recoveryUnit()->commitAndRestart();
if ( result.getError() ) {
result.getError()->setIndex( updateItem.getItemIndex() );
*error = result.releaseError();
}
}
void WriteBatchExecutor::execRemove( const BatchItemRef& removeItem,
WriteErrorDetail** error ) {
// Removes are similar to updates, but page faults are handled externally
// BEGIN CURRENT OP
CurOp currentOp( _txn->getClient(), _txn->getClient()->curop() );
beginCurrentOp( ¤tOp, _txn->getClient(), removeItem );
incOpStats( removeItem );
WriteOpResult result;
multiRemove( _txn, removeItem, &result );
// END CURRENT OP
incWriteStats( removeItem, result.getStats(), result.getError(), ¤tOp );
finishCurrentOp( _txn, ¤tOp, result.getError() );
// End current transaction and release snapshot.
_txn->recoveryUnit()->commitAndRestart();
if ( result.getError() ) {
result.getError()->setIndex( removeItem.getItemIndex() );
*error = result.releaseError();
}
}
//
// IN-DB-LOCK CORE OPERATIONS
//
WriteBatchExecutor::ExecInsertsState::ExecInsertsState(OperationContext* txn,
const BatchedCommandRequest* aRequest) :
txn(txn),
request(aRequest),
currIndex(0),
_transaction(txn, MODE_IX),
_collection(NULL) {
}
bool WriteBatchExecutor::ExecInsertsState::_lockAndCheckImpl(WriteOpResult* result,
bool intentLock) {
if (hasLock()) {
// TODO: Client::Context legacy, needs to be removed
txn->getCurOp()->enter(_context->ns(),
_context->db() ? _context->db()->getProfilingLevel() : 0);
return true;
}
if (request->isInsertIndexRequest())
intentLock = false; // can't build indexes in intent mode
invariant(!_context.get());
const NamespaceString& nss = request->getNSS();
_collLock.reset(); // give up locks if any
_writeLock.reset();
_writeLock.reset(new Lock::DBLock(txn->lockState(),
nss.db(),
intentLock ? MODE_IX : MODE_X));
if (intentLock && dbHolder().get(txn, nss.db()) == NULL) {
// Ensure exclusive lock in case the database doesn't yet exist
_writeLock.reset();
_writeLock.reset(new Lock::DBLock(txn->lockState(),
nss.db(),
MODE_X));
intentLock = false;
}
_collLock.reset(new Lock::CollectionLock(txn->lockState(),
nss.ns(),
intentLock ? MODE_IX : MODE_X));
if (!checkIsMasterForDatabase(nss, result)) {
return false;
}
if (!checkShardVersion(txn, &shardingState, *request, result)) {
return false;
}
if (!checkIndexConstraints(txn, &shardingState, *request, result)) {
return false;
}
_context.reset();
_context.reset(new Client::Context(txn, nss, false));
Database* database = _context->db();
dassert(database);
_collection = database->getCollection(request->getTargetingNS());
if (!_collection) {
if (intentLock) {
// try again with full X lock.
unlock();
return _lockAndCheckImpl(result, false);
}
WriteUnitOfWork wunit (txn);
// Implicitly create if it doesn't exist
_collection = database->createCollection(txn, request->getTargetingNS());
if (!_collection) {
result->setError(
toWriteError(Status(ErrorCodes::InternalError,
"could not create collection " +
request->getTargetingNS())));
return false;
}
repl::logOp(txn,
"c",
(database->name() + ".$cmd").c_str(),
BSON("create" << nsToCollectionSubstring(request->getTargetingNS())));
wunit.commit();
}
return true;
}
bool WriteBatchExecutor::ExecInsertsState::lockAndCheck(WriteOpResult* result) {
if (_lockAndCheckImpl(result))
return true;
unlock();
return false;
}
void WriteBatchExecutor::ExecInsertsState::unlock() {
_collection = NULL;
_context.reset();
_collLock.reset();
_writeLock.reset();
}
static void insertOne(WriteBatchExecutor::ExecInsertsState* state, WriteOpResult* result) {
invariant(state->currIndex < state->normalizedInserts.size());
const StatusWith<BSONObj>& normalizedInsert(state->normalizedInserts[state->currIndex]);
if (!normalizedInsert.isOK()) {
result->setError(toWriteError(normalizedInsert.getStatus()));
return;
}
const BSONObj& insertDoc = normalizedInsert.getValue().isEmpty() ?
state->request->getInsertRequest()->getDocumentsAt( state->currIndex ) :
normalizedInsert.getValue();
try {
if (!state->request->isInsertIndexRequest()) {
if (state->lockAndCheck(result)) {
singleInsert(state->txn, insertDoc, state->getCollection(), result);
}
}
else {
singleCreateIndex(state->txn, insertDoc, result);
}
}
catch (const DBException& ex) {
Status status(ex.toStatus());
if (ErrorCodes::isInterruption(status.code()))
throw;
result->setError(toWriteError(status));
}
// Errors release the write lock, as a matter of policy.
if (result->getError()) {
state->txn->recoveryUnit()->commitAndRestart();
state->unlock();
}
}
void WriteBatchExecutor::execOneInsert(ExecInsertsState* state, WriteErrorDetail** error) {
BatchItemRef currInsertItem(state->request, state->currIndex);
CurOp currentOp( _txn->getClient(), _txn->getClient()->curop() );
beginCurrentOp( ¤tOp, _txn->getClient(), currInsertItem );
incOpStats(currInsertItem);
WriteOpResult result;
insertOne(state, &result);
incWriteStats(currInsertItem,
result.getStats(),
result.getError(),
¤tOp);
finishCurrentOp(_txn, ¤tOp, result.getError());
if (result.getError()) {
*error = result.releaseError();
}
}
/**
* Perform a single insert into a collection. Requires the insert be preprocessed and the
* collection already has been created.
*
* Might fault or error, otherwise populates the result.
*/
static void singleInsert( OperationContext* txn,
const BSONObj& docToInsert,
Collection* collection,
WriteOpResult* result ) {
const string& insertNS = collection->ns().ns();
invariant(txn->lockState()->isCollectionLockedForMode(insertNS, MODE_IX));
WriteUnitOfWork wunit(txn);
StatusWith<RecordId> status = collection->insertDocument( txn, docToInsert, true );
if ( !status.isOK() ) {
result->setError(toWriteError(status.getStatus()));
}
else {
repl::logOp( txn, "i", insertNS.c_str(), docToInsert );
result->getStats().n = 1;
wunit.commit();
}
}
/**
* Perform a single index creation on a collection. Requires the index descriptor be
* preprocessed.
*
* Might fault or error, otherwise populates the result.
*/
static void singleCreateIndex(OperationContext* txn,
const BSONObj& indexDesc,
WriteOpResult* result) {
BSONElement nsElement = indexDesc["ns"];
uassert(ErrorCodes::NoSuchKey,
"Missing \"ns\" field in index description",
!nsElement.eoo());
uassert(ErrorCodes::TypeMismatch,
str::stream() << "Expected \"ns\" field of index description to be a " "string, "
"but found a " << typeName(nsElement.type()),
nsElement.type() == String);
const NamespaceString ns(nsElement.valueStringData());
BSONObjBuilder cmdBuilder;
cmdBuilder << "createIndexes" << ns.coll();
cmdBuilder << "indexes" << BSON_ARRAY(indexDesc);
BSONObj cmd = cmdBuilder.done();
Command* createIndexesCmd = Command::findCommand("createIndexes");
invariant(createIndexesCmd);
std::string errmsg;
BSONObjBuilder resultBuilder;
const bool success = createIndexesCmd->run(
txn,
ns.db().toString(),
cmd,
0,
errmsg,
resultBuilder,
false /* fromrepl */);
Command::appendCommandStatus(resultBuilder, success, errmsg);
BSONObj cmdResult = resultBuilder.done();
uassertStatusOK(Command::getStatusFromCommandResult(cmdResult));
const long long numIndexesBefore = cmdResult["numIndexesBefore"].safeNumberLong();
const long long numIndexesAfter = cmdResult["numIndexesAfter"].safeNumberLong();
if (numIndexesAfter - numIndexesBefore == 1) {
result->getStats().n = 1;
}
else if (numIndexesAfter != 0 && numIndexesAfter != numIndexesBefore) {
severe() <<
"Created multiple indexes while attempting to create only 1; numIndexesBefore = " <<
numIndexesBefore << "; numIndexesAfter = " << numIndexesAfter;
fassertFailed(28547);
}
else {
result->getStats().n = 0;
}
}
static void multiUpdate( OperationContext* txn,
const BatchItemRef& updateItem,
WriteOpResult* result ) {
const NamespaceString nsString(updateItem.getRequest()->getNS());
const bool isMulti = updateItem.getUpdate()->getMulti();
UpdateRequest request(nsString);
request.setQuery(updateItem.getUpdate()->getQuery());
request.setUpdates(updateItem.getUpdate()->getUpdateExpr());
request.setMulti(isMulti);
request.setUpsert(updateItem.getUpdate()->getUpsert());
request.setUpdateOpLog(true);
UpdateLifecycleImpl updateLifecycle(true, request.getNamespaceString());
request.setLifecycle(&updateLifecycle);
// Updates from the write commands path can yield.
request.setYieldPolicy(PlanExecutor::YIELD_AUTO);
int attempt = 0;
bool createCollection = false;
for ( int fakeLoop = 0; fakeLoop < 1; fakeLoop++ ) {
ParsedUpdate parsedUpdate(txn, &request);
Status status = parsedUpdate.parseRequest();
if (!status.isOK()) {
result->setError(toWriteError(status));
return;
}
if ( createCollection ) {
ScopedTransaction transaction(txn, MODE_IX);
Lock::DBLock lk(txn->lockState(), nsString.db(), MODE_X);
Client::Context ctx(txn, nsString.ns(), false /* don't check version */);
Database* db = ctx.db();
if ( db->getCollection( nsString.ns() ) ) {
// someone else beat us to it
}
else {
WriteUnitOfWork wuow(txn);
uassertStatusOK( userCreateNS( txn, db,
nsString.ns(), BSONObj(),
!request.isFromReplication() ) );
wuow.commit();
}
}
///////////////////////////////////////////
ScopedTransaction transaction(txn, MODE_IX);
Lock::DBLock dbLock(txn->lockState(), nsString.db(), MODE_IX);
Lock::CollectionLock colLock(txn->lockState(),
nsString.ns(),
MODE_IX);
///////////////////////////////////////////
if (!checkShardVersion(txn, &shardingState, *updateItem.getRequest(), result))
return;
Database* const db = dbHolder().get(txn, nsString.db());
if (db == NULL) {
if (createCollection) {
// we raced with some, accept defeat
result->getStats().nModified = 0;
result->getStats().n = 0;
return;
}
// Database not yet created
if (!request.isUpsert()) {
// not an upsert, no database, nothing to do
result->getStats().nModified = 0;
result->getStats().n = 0;
return;
}
// upsert, don't try to get a context as no MODE_X lock is held
fakeLoop = -1;
createCollection = true;
continue;
}
Client::Context ctx(txn, nsString.ns(), false /* don't check version */);
Collection* collection = db->getCollection(nsString.ns());
if ( collection == NULL ) {
if ( createCollection ) {
// we raced with some, accept defeat
result->getStats().nModified = 0;
result->getStats().n = 0;
return;
}
if ( !request.isUpsert() ) {
// not an upsert, no collection, nothing to do
result->getStats().nModified = 0;
result->getStats().n = 0;
return;
}
// upsert, mark that we should create collection
fakeLoop = -1;
createCollection = true;
continue;
}
OpDebug* debug = &txn->getCurOp()->debug();
try {
invariant(collection);
PlanExecutor* rawExec;
uassertStatusOK(getExecutorUpdate(txn, collection, &parsedUpdate, debug, &rawExec));
boost::scoped_ptr<PlanExecutor> exec(rawExec);
uassertStatusOK(exec->executePlan());
UpdateResult res = UpdateStage::makeUpdateResult(exec.get(), debug);
const long long numDocsModified = res.numDocsModified;
const long long numMatched = res.numMatched;
const BSONObj resUpsertedID = res.upserted;
// We have an _id from an insert
const bool didInsert = !resUpsertedID.isEmpty();
result->getStats().nModified = didInsert ? 0 : numDocsModified;
result->getStats().n = didInsert ? 1 : numMatched;
result->getStats().upsertedID = resUpsertedID;
}
catch ( const WriteConflictException& dle ) {
debug->writeConflicts++;
if ( isMulti ) {
log() << "Had WriteConflict during multi update, aborting";
throw;
}
createCollection = false;
// RESTART LOOP
fakeLoop = -1;
txn->recoveryUnit()->commitAndRestart();
WriteConflictException::logAndBackoff( attempt++, "update", nsString.ns() );
}
catch (const DBException& ex) {
Status status = ex.toStatus();
if (ErrorCodes::isInterruption(status.code())) {
throw;
}
result->setError(toWriteError(status));
}
}
}
/**
* Perform a remove operation, which might remove multiple documents. Dispatches to remove code
* currently to do most of this.
*
* Might fault or error, otherwise populates the result.
*/
static void multiRemove( OperationContext* txn,
const BatchItemRef& removeItem,
WriteOpResult* result ) {
const NamespaceString& nss = removeItem.getRequest()->getNSS();
DeleteRequest request(nss);
request.setQuery( removeItem.getDelete()->getQuery() );
request.setMulti( removeItem.getDelete()->getLimit() != 1 );
request.setUpdateOpLog(true);
request.setGod( false );
// Deletes running through the write commands path can yield.
request.setYieldPolicy(PlanExecutor::YIELD_AUTO);
int attempt = 1;
while ( 1 ) {
try {
ParsedDelete parsedDelete(txn, &request);
Status status = parsedDelete.parseRequest();
if (!status.isOK()) {
result->setError(toWriteError(status));
return;
}
ScopedTransaction scopedXact(txn, MODE_IX);
AutoGetDb autoDb(txn, nss.db(), MODE_IX);
if (!autoDb.getDb()) {
break;
}
Lock::CollectionLock collLock(txn->lockState(), nss.ns(), MODE_IX);
// Check version once we're locked
if (!checkShardVersion(txn, &shardingState, *removeItem.getRequest(), result)) {
// Version error
return;
}
// Context once we're locked, to set more details in currentOp()
// TODO: better constructor?
Client::Context ctx(txn, nss.ns(), false /* don't check version */);
PlanExecutor* rawExec;
uassertStatusOK(getExecutorDelete(txn,
ctx.db()->getCollection(nss),
&parsedDelete,
&rawExec));
boost::scoped_ptr<PlanExecutor> exec(rawExec);
// Execute the delete and retrieve the number deleted.
uassertStatusOK(exec->executePlan());
result->getStats().n = DeleteStage::getNumDeleted(exec.get());
break;
}
catch ( const WriteConflictException& dle ) {
txn->getCurOp()->debug().writeConflicts++;
WriteConflictException::logAndBackoff( attempt++, "delete", nss.ns() );
}
catch ( const DBException& ex ) {
Status status = ex.toStatus();
if (ErrorCodes::isInterruption(status.code())) {
throw;
}
result->setError(toWriteError(status));
return;
}
}
}
} // namespace mongo
|