-
Notifications
You must be signed in to change notification settings - Fork 5.6k
/
mr.cpp
1913 lines (1629 loc) · 70.3 KB
/
mr.cpp
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
/**
* Copyright (C) 2012 10gen Inc.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License, version 3,
* as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
* As a special exception, the copyright holders give permission to link the
* code of portions of this program with the OpenSSL library under certain
* conditions as described in each individual source file and distribute
* linked combinations including the program with the OpenSSL library. You
* must comply with the GNU Affero General Public License in all respects for
* all of the code used other than as permitted herein. If you modify file(s)
* with this exception, you may extend this exception to your version of the
* file(s), but you are not obligated to do so. If you do not wish to do so,
* delete this exception statement from your version. If you delete this
* exception statement from all source files in the program, then also delete
* it in the license file.
*/
#define MONGO_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kCommand
#include "mongo/platform/basic.h"
#include "mongo/db/commands/mr.h"
#include "mongo/base/status_with.h"
#include "mongo/bson/util/builder.h"
#include "mongo/client/connpool.h"
#include "mongo/db/auth/authorization_session.h"
#include "mongo/db/bson/dotted_path_support.h"
#include "mongo/db/catalog/collection.h"
#include "mongo/db/catalog/collection_catalog_entry.h"
#include "mongo/db/catalog/database_holder.h"
#include "mongo/db/catalog/document_validation.h"
#include "mongo/db/catalog/index_catalog.h"
#include "mongo/db/catalog/index_key_validate.h"
#include "mongo/db/client.h"
#include "mongo/db/clientcursor.h"
#include "mongo/db/commands.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/db.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbhelpers.h"
#include "mongo/db/exec/working_set_common.h"
#include "mongo/db/index/index_descriptor.h"
#include "mongo/db/matcher/extensions_callback_real.h"
#include "mongo/db/op_observer.h"
#include "mongo/db/ops/insert.h"
#include "mongo/db/query/find_common.h"
#include "mongo/db/query/get_executor.h"
#include "mongo/db/query/plan_summary_stats.h"
#include "mongo/db/query/query_planner.h"
#include "mongo/db/repl/replication_coordinator.h"
#include "mongo/db/s/collection_metadata.h"
#include "mongo/db/s/collection_sharding_state.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/db/server_options.h"
#include "mongo/db/service_context.h"
#include "mongo/s/catalog_cache.h"
#include "mongo/s/client/parallel.h"
#include "mongo/s/client/shard_connection.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/grid.h"
#include "mongo/s/shard_key_pattern.h"
#include "mongo/s/stale_exception.h"
#include "mongo/scripting/engine.h"
#include "mongo/stdx/mutex.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
#include "mongo/util/scopeguard.h"
namespace mongo {
using std::set;
using std::shared_ptr;
using std::string;
using std::stringstream;
using std::unique_ptr;
using std::vector;
using IndexVersion = IndexDescriptor::IndexVersion;
namespace dps = ::mongo::dotted_path_support;
namespace mr {
AtomicUInt32 Config::JOB_NUMBER;
JSFunction::JSFunction(const std::string& type, const BSONElement& e) {
_type = type;
_code = e._asCode();
if (e.type() == CodeWScope)
_wantedScope = e.codeWScopeObject();
}
void JSFunction::init(State* state) {
_scope = state->scope();
verify(_scope);
_scope->init(&_wantedScope);
_func = _scope->createFunction(_code.c_str());
uassert(13598, str::stream() << "couldn't compile code for: " << _type, _func);
// install in JS scope so that it can be called in JS mode
_scope->setFunction(_type.c_str(), _code.c_str());
}
void JSMapper::init(State* state) {
_func.init(state);
_params = state->config().mapParams;
}
/**
* Applies the map function to an object, which should internally call emit()
*/
void JSMapper::map(const BSONObj& o) {
Scope* s = _func.scope();
verify(s);
if (s->invoke(_func.func(), &_params, &o, 0, true))
uasserted(9014, str::stream() << "map invoke failed: " << s->getError());
}
/**
* Applies the finalize function to a tuple obj (key, val)
* Returns tuple obj {_id: key, value: newval}
*/
BSONObj JSFinalizer::finalize(const BSONObj& o) {
Scope* s = _func.scope();
Scope::NoDBAccess no = s->disableDBAccess("can't access db inside finalize");
s->invokeSafe(_func.func(), &o, 0);
// don't want to use o.objsize() to size b
// since there are many cases where the point of finalize
// is converting many fields to 1
BSONObjBuilder b;
b.append(o.firstElement());
s->append(b, "value", "__returnValue");
return b.obj();
}
void JSReducer::init(State* state) {
_func.init(state);
}
/**
* Reduces a list of tuple objects (key, value) to a single tuple {"0": key, "1": value}
*/
BSONObj JSReducer::reduce(const BSONList& tuples) {
if (tuples.size() <= 1)
return tuples[0];
BSONObj key;
int endSizeEstimate = 16;
_reduce(tuples, key, endSizeEstimate);
BSONObjBuilder b(endSizeEstimate);
b.appendAs(key.firstElement(), "0");
_func.scope()->append(b, "1", "__returnValue");
return b.obj();
}
/**
* Reduces a list of tuple object (key, value) to a single tuple {_id: key, value: val}
* Also applies a finalizer method if present.
*/
BSONObj JSReducer::finalReduce(const BSONList& tuples, Finalizer* finalizer) {
BSONObj res;
BSONObj key;
if (tuples.size() == 1) {
// 1 obj, just use it
key = tuples[0];
BSONObjBuilder b(key.objsize());
BSONObjIterator it(key);
b.appendAs(it.next(), "_id");
b.appendAs(it.next(), "value");
res = b.obj();
} else {
// need to reduce
int endSizeEstimate = 16;
_reduce(tuples, key, endSizeEstimate);
BSONObjBuilder b(endSizeEstimate);
b.appendAs(key.firstElement(), "_id");
_func.scope()->append(b, "value", "__returnValue");
res = b.obj();
}
if (finalizer) {
res = finalizer->finalize(res);
}
return res;
}
/**
* actually applies a reduce, to a list of tuples (key, value).
* After the call, tuples will hold a single tuple {"0": key, "1": value}
*/
void JSReducer::_reduce(const BSONList& tuples, BSONObj& key, int& endSizeEstimate) {
uassert(10074, "need values", tuples.size());
int sizeEstimate = (tuples.size() * tuples.begin()->getField("value").size()) + 128;
// need to build the reduce args: ( key, [values] )
BSONObjBuilder reduceArgs(sizeEstimate);
std::unique_ptr<BSONArrayBuilder> valueBuilder;
unsigned n = 0;
for (; n < tuples.size(); n++) {
BSONObjIterator j(tuples[n]);
BSONElement keyE = j.next();
if (n == 0) {
reduceArgs.append(keyE);
key = keyE.wrap();
valueBuilder.reset(new BSONArrayBuilder(reduceArgs.subarrayStart("tuples")));
}
BSONElement ee = j.next();
uassert(13070, "value too large to reduce", ee.size() < (BSONObjMaxUserSize / 2));
// If adding this element to the array would cause it to be too large, break. The
// remainder of the tuples will be processed recursively at the end of this
// function.
if (valueBuilder->len() + ee.size() > BSONObjMaxUserSize) {
verify(n > 1); // if not, inf. loop
break;
}
valueBuilder->append(ee);
}
verify(valueBuilder);
valueBuilder->done();
BSONObj args = reduceArgs.obj();
Scope* s = _func.scope();
s->invokeSafe(_func.func(), &args, 0);
++numReduces;
if (s->type("__returnValue") == Array) {
uasserted(10075, "reduce -> multiple not supported yet");
return;
}
endSizeEstimate = key.objsize() + (args.objsize() / tuples.size());
if (n == tuples.size())
return;
// the input list was too large, add the rest of elmts to new tuples and reduce again
// note: would be better to use loop instead of recursion to avoid stack overflow
BSONList x;
for (; n < tuples.size(); n++) {
x.push_back(tuples[n]);
}
BSONObjBuilder temp(endSizeEstimate);
temp.append(key.firstElement());
s->append(temp, "1", "__returnValue");
x.push_back(temp.obj());
_reduce(x, key, endSizeEstimate);
}
Config::Config(const string& _dbname, const BSONObj& cmdObj) {
dbname = _dbname;
uassert(ErrorCodes::TypeMismatch,
str::stream() << "'mapReduce' must be of type String",
cmdObj.firstElement().type() == BSONType::String);
nss = NamespaceString(dbname, cmdObj.firstElement().valueStringData());
uassert(ErrorCodes::InvalidNamespace,
str::stream() << "Invalid namespace: " << nss.ns(),
nss.isValid());
verbose = cmdObj["verbose"].trueValue();
jsMode = cmdObj["jsMode"].trueValue();
splitInfo = 0;
if (cmdObj.hasField("splitInfo")) {
splitInfo = cmdObj["splitInfo"].Int();
}
jsMaxKeys = 500000;
reduceTriggerRatio = 10.0;
maxInMemSize = 500 * 1024;
uassert(13602, "outType is no longer a valid option", cmdObj["outType"].eoo());
outputOptions = parseOutputOptions(dbname, cmdObj);
shardedFirstPass = false;
if (cmdObj.hasField("shardedFirstPass") && cmdObj["shardedFirstPass"].trueValue()) {
massert(16054,
"shardedFirstPass should only use replace outType",
outputOptions.outType == REPLACE);
shardedFirstPass = true;
}
if (outputOptions.outType != INMEMORY) { // setup temp collection name
tempNamespace = NamespaceString(
outputOptions.outDB.empty() ? dbname : outputOptions.outDB,
str::stream() << "tmp.mr." << cmdObj.firstElement().valueStringData() << "_"
<< JOB_NUMBER.fetchAndAdd(1));
incLong = NamespaceString(str::stream() << tempNamespace.ns() << "_inc");
}
{
// scope and code
if (cmdObj["scope"].type() == Object)
scopeSetup = cmdObj["scope"].embeddedObjectUserCheck().getOwned();
mapper.reset(new JSMapper(cmdObj["map"]));
reducer.reset(new JSReducer(cmdObj["reduce"]));
if (cmdObj["finalize"].type() && cmdObj["finalize"].trueValue())
finalizer.reset(new JSFinalizer(cmdObj["finalize"]));
if (cmdObj["mapparams"].type() == Array) {
mapParams = cmdObj["mapparams"].embeddedObjectUserCheck().getOwned();
}
}
{
// query options
BSONElement q = cmdObj["query"];
if (q.type() == Object)
filter = q.embeddedObjectUserCheck();
else
uassert(13608, "query has to be blank or an Object", !q.trueValue());
BSONElement s = cmdObj["sort"];
if (s.type() == Object)
sort = s.embeddedObjectUserCheck();
else
uassert(13609, "sort has to be blank or an Object", !s.trueValue());
BSONElement collationElt = cmdObj["collation"];
if (collationElt.type() == Object)
collation = collationElt.embeddedObjectUserCheck();
else
uassert(40082,
str::stream()
<< "mapReduce 'collation' parameter must be of type Object but found type: "
<< typeName(collationElt.type()),
collationElt.eoo());
if (cmdObj["limit"].isNumber())
limit = cmdObj["limit"].numberLong();
else
limit = 0;
}
}
/**
* Clean up the temporary and incremental collections
*/
void State::dropTempCollections() {
// The cleanup handler should not be interruptible.
UninterruptibleLockGuard noInterrupt(_opCtx->lockState());
if (!_config.tempNamespace.isEmpty()) {
writeConflictRetry(_opCtx, "M/R dropTempCollections", _config.tempNamespace.ns(), [this] {
AutoGetDb autoDb(_opCtx, _config.tempNamespace.db(), MODE_X);
if (auto db = autoDb.getDb()) {
WriteUnitOfWork wunit(_opCtx);
uassert(
ErrorCodes::PrimarySteppedDown,
str::stream()
<< "no longer primary while dropping temporary collection for mapReduce: "
<< _config.tempNamespace.ns(),
repl::ReplicationCoordinator::get(_opCtx)->canAcceptWritesFor(
_opCtx, _config.tempNamespace));
uassertStatusOK(db->dropCollection(_opCtx, _config.tempNamespace.ns()));
wunit.commit();
}
});
// Always forget about temporary namespaces, so we don't cache lots of them
ShardConnection::forgetNS(_config.tempNamespace.ns());
}
if (_useIncremental && !_config.incLong.isEmpty()) {
// We don't want to log the deletion of incLong as it isn't replicated. While
// harmless, this would lead to a scary looking warning on the secondaries.
repl::UnreplicatedWritesBlock uwb(_opCtx);
writeConflictRetry(_opCtx, "M/R dropTempCollections", _config.incLong.ns(), [this] {
Lock::DBLock lk(_opCtx, _config.incLong.db(), MODE_X);
if (Database* db = dbHolder().get(_opCtx, _config.incLong.ns())) {
WriteUnitOfWork wunit(_opCtx);
uassertStatusOK(db->dropCollection(_opCtx, _config.incLong.ns()));
wunit.commit();
}
});
ShardConnection::forgetNS(_config.incLong.ns());
}
}
/**
* Create temporary collection, set up indexes
*/
void State::prepTempCollection() {
if (!_onDisk)
return;
dropTempCollections();
if (_useIncremental) {
// Create the inc collection and make sure we have index on "0" key.
// Intentionally not replicating the inc collection to secondaries.
repl::UnreplicatedWritesBlock uwb(_opCtx);
writeConflictRetry(_opCtx, "M/R prepTempCollection", _config.incLong.ns(), [this] {
OldClientWriteContext incCtx(_opCtx, _config.incLong.ns());
WriteUnitOfWork wuow(_opCtx);
Collection* incColl = incCtx.getCollection();
invariant(!incColl);
CollectionOptions options;
options.setNoIdIndex();
options.temp = true;
options.uuid.emplace(UUID::gen());
incColl = incCtx.db()->createCollection(
_opCtx, _config.incLong.ns(), options, false /* force no _id index */);
invariant(incColl);
auto rawIndexSpec =
BSON("key" << BSON("0" << 1) << "ns" << _config.incLong.ns() << "name"
<< "_temp_0");
auto indexSpec = uassertStatusOK(index_key_validate::validateIndexSpec(
_opCtx, rawIndexSpec, _config.incLong, serverGlobalParams.featureCompatibility));
Status status = incColl->getIndexCatalog()
->createIndexOnEmptyCollection(_opCtx, indexSpec)
.getStatus();
if (!status.isOK()) {
uasserted(17305,
str::stream() << "createIndex failed for mr incLong ns: "
<< _config.incLong.ns()
<< " err: "
<< status.code());
}
wuow.commit();
});
}
CollectionOptions finalOptions;
vector<BSONObj> indexesToInsert;
{
// copy indexes and collection options into temporary storage
OldClientWriteContext finalCtx(_opCtx, _config.outputOptions.finalNamespace.ns());
Collection* const finalColl = finalCtx.getCollection();
if (finalColl) {
finalOptions = finalColl->getCatalogEntry()->getCollectionOptions(_opCtx);
if (_config.finalOutputCollUUID) {
// The final output collection's UUID is passed from mongos if the final output
// collection is sharded. If a UUID was sent, ensure it matches what's on this
// shard.
uassert(ErrorCodes::InternalError,
str::stream()
<< "UUID sent by mongos for sharded final output collection "
<< _config.outputOptions.finalNamespace.ns()
<< " does not match UUID for the existing collection with that "
"name on this shard",
finalColl->uuid() == _config.finalOutputCollUUID);
}
IndexCatalog::IndexIterator ii =
finalColl->getIndexCatalog()->getIndexIterator(_opCtx, true);
// Iterate over finalColl's indexes.
while (ii.more()) {
IndexDescriptor* currIndex = ii.next();
BSONObjBuilder b;
b.append("ns", _config.tempNamespace.ns());
// Copy over contents of the index descriptor's infoObj.
BSONObjIterator j(currIndex->infoObj());
while (j.more()) {
BSONElement e = j.next();
if (str::equals(e.fieldName(), "_id") || str::equals(e.fieldName(), "ns"))
continue;
b.append(e);
}
indexesToInsert.push_back(b.obj());
}
}
}
writeConflictRetry(_opCtx, "M/R prepTempCollection", _config.tempNamespace.ns(), [&] {
// create temp collection and insert the indexes from temporary storage
OldClientWriteContext tempCtx(_opCtx, _config.tempNamespace.ns());
WriteUnitOfWork wuow(_opCtx);
uassert(
ErrorCodes::PrimarySteppedDown,
str::stream() << "no longer primary while creating temporary collection for mapReduce: "
<< _config.tempNamespace.ns(),
repl::ReplicationCoordinator::get(_opCtx)->canAcceptWritesFor(_opCtx,
_config.tempNamespace));
Collection* tempColl = tempCtx.getCollection();
invariant(!tempColl);
CollectionOptions options = finalOptions;
options.temp = true;
// If a UUID for the final output collection was sent by mongos (i.e., the final output
// collection is sharded), use the UUID mongos sent when creating the temp collection.
// When the temp collection is renamed to the final output collection, the UUID will be
// preserved.
options.uuid.emplace(_config.finalOutputCollUUID ? *_config.finalOutputCollUUID
: UUID::gen());
// Override createCollection's prohibition on creating new replicated collections without an
// _id index.
bool buildIdIndex = (options.autoIndexId == CollectionOptions::YES ||
options.autoIndexId == CollectionOptions::DEFAULT);
tempColl = tempCtx.db()->createCollection(
_opCtx, _config.tempNamespace.ns(), options, buildIdIndex);
for (vector<BSONObj>::iterator it = indexesToInsert.begin(); it != indexesToInsert.end();
++it) {
Status status =
tempColl->getIndexCatalog()->createIndexOnEmptyCollection(_opCtx, *it).getStatus();
if (!status.isOK()) {
if (status.code() == ErrorCodes::IndexAlreadyExists) {
continue;
}
uassertStatusOK(status);
}
// Log the createIndex operation.
auto uuid = tempColl->uuid();
getGlobalServiceContext()->getOpObserver()->onCreateIndex(
_opCtx, _config.tempNamespace, uuid, *it, false);
}
wuow.commit();
});
}
/**
* For inline mode, appends results to output object.
* Makes sure (key, value) tuple is formatted as {_id: key, value: val}
*/
void State::appendResults(BSONObjBuilder& final) {
if (_onDisk) {
if (!_config.outputOptions.outDB.empty()) {
BSONObjBuilder loc;
if (!_config.outputOptions.outDB.empty())
loc.append("db", _config.outputOptions.outDB);
if (!_config.outputOptions.collectionName.empty())
loc.append("collection", _config.outputOptions.collectionName);
final.append("result", loc.obj());
} else {
if (!_config.outputOptions.collectionName.empty())
final.append("result", _config.outputOptions.collectionName);
}
if (_config.splitInfo > 0) {
// add split points, used for shard
BSONObj res;
BSONObj idKey = BSON("_id" << 1);
if (!_db.runCommand("admin",
BSON("splitVector" << _config.outputOptions.finalNamespace.ns()
<< "keyPattern"
<< idKey
<< "maxChunkSizeBytes"
<< _config.splitInfo),
res)) {
uasserted(15921, str::stream() << "splitVector failed: " << res);
}
if (res.hasField("splitKeys"))
final.append(res.getField("splitKeys"));
}
return;
}
if (_jsMode) {
ScriptingFunction getResult = _scope->createFunction(
"var map = _mrMap;"
"var result = [];"
"for (key in map) {"
" result.push({_id: key, value: map[key]});"
"}"
"return result;");
_scope->invoke(getResult, 0, 0, 0, false);
BSONObj obj = _scope->getObject("__returnValue");
final.append("results", BSONArray(obj));
return;
}
uassert(13604, "too much data for in memory map/reduce", _size < BSONObjMaxUserSize);
BSONArrayBuilder b((int)(_size * 1.2)); // _size is data size, doesn't count overhead and keys
for (const auto& entry : *_temp) {
const BSONObj& key = entry.first;
const BSONList& all = entry.second;
verify(all.size() == 1);
BSONObjIterator vi(all[0]);
vi.next();
BSONObjBuilder temp(b.subobjStart());
temp.appendAs(key.firstElement(), "_id");
temp.appendAs(vi.next(), "value");
temp.done();
}
BSONArray res = b.arr();
final.append("results", res);
}
/**
* Does post processing on output collection.
* This may involve replacing, merging or reducing.
*/
long long State::postProcessCollection(OperationContext* opCtx,
CurOp* curOp,
ProgressMeterHolder& pm) {
if (_onDisk == false || _config.outputOptions.outType == Config::INMEMORY)
return numInMemKeys();
bool holdingGlobalLock = false;
if (_config.outputOptions.outNonAtomic)
return postProcessCollectionNonAtomic(opCtx, curOp, pm, holdingGlobalLock);
invariant(!opCtx->lockState()->isLocked());
// This must be global because we may write across different databases.
Lock::GlobalWrite lock(opCtx);
holdingGlobalLock = true;
return postProcessCollectionNonAtomic(opCtx, curOp, pm, holdingGlobalLock);
}
namespace {
// Runs a count against the namespace specified by 'ns'. If the caller holds the global write lock,
// then this function does not acquire any additional locks.
unsigned long long _collectionCount(OperationContext* opCtx,
const NamespaceString& nss,
bool callerHoldsGlobalLock) {
Collection* coll = nullptr;
boost::optional<AutoGetCollectionForReadCommand> ctx;
// If the global write lock is held, we must avoid using AutoGetCollectionForReadCommand as it
// may lead to deadlock when waiting for a majority snapshot to be committed. See SERVER-24596.
if (callerHoldsGlobalLock) {
Database* db = dbHolder().get(opCtx, nss.ns());
if (db) {
coll = db->getCollection(opCtx, nss);
}
} else {
ctx.emplace(opCtx, nss);
coll = ctx->getCollection();
}
return coll ? coll->numRecords(opCtx) : 0;
}
} // namespace
long long State::postProcessCollectionNonAtomic(OperationContext* opCtx,
CurOp* curOp,
ProgressMeterHolder& pm,
bool callerHoldsGlobalLock) {
if (_config.outputOptions.finalNamespace == _config.tempNamespace)
return _collectionCount(opCtx, _config.outputOptions.finalNamespace, callerHoldsGlobalLock);
if (_config.outputOptions.outType == Config::REPLACE ||
_collectionCount(opCtx, _config.outputOptions.finalNamespace, callerHoldsGlobalLock) == 0) {
// This must be global because we may write across different databases.
Lock::GlobalWrite lock(opCtx);
// replace: just rename from temp to final collection name, dropping previous collection
_db.dropCollection(_config.outputOptions.finalNamespace.ns());
BSONObj info;
if (!_db.runCommand("admin",
BSON("renameCollection" << _config.tempNamespace.ns() << "to"
<< _config.outputOptions.finalNamespace.ns()
<< "stayTemp"
<< _config.shardedFirstPass),
info)) {
uasserted(10076, str::stream() << "rename failed: " << info);
}
_db.dropCollection(_config.tempNamespace.ns());
} else if (_config.outputOptions.outType == Config::MERGE) {
// merge: upsert new docs into old collection
{
const auto count =
_collectionCount(opCtx, _config.tempNamespace, callerHoldsGlobalLock);
stdx::lock_guard<Client> lk(*opCtx->getClient());
curOp->setMessage_inlock(
"m/r: merge post processing", "M/R Merge Post Processing Progress", count);
}
unique_ptr<DBClientCursor> cursor = _db.query(_config.tempNamespace.ns(), BSONObj());
while (cursor->more()) {
Lock::DBLock lock(opCtx, _config.outputOptions.finalNamespace.db(), MODE_X);
BSONObj o = cursor->nextSafe();
Helpers::upsert(opCtx, _config.outputOptions.finalNamespace.ns(), o);
pm.hit();
}
_db.dropCollection(_config.tempNamespace.ns());
pm.finished();
} else if (_config.outputOptions.outType == Config::REDUCE) {
// reduce: apply reduce op on new result and existing one
BSONList values;
{
const auto count =
_collectionCount(opCtx, _config.tempNamespace, callerHoldsGlobalLock);
stdx::lock_guard<Client> lk(*opCtx->getClient());
curOp->setMessage_inlock(
"m/r: reduce post processing", "M/R Reduce Post Processing Progress", count);
}
unique_ptr<DBClientCursor> cursor = _db.query(_config.tempNamespace.ns(), BSONObj());
while (cursor->more()) {
// This must be global because we may write across different databases.
Lock::GlobalWrite lock(opCtx);
BSONObj temp = cursor->nextSafe();
BSONObj old;
bool found;
{
OldClientContext tx(opCtx, _config.outputOptions.finalNamespace.ns());
Collection* coll =
getCollectionOrUassert(opCtx, tx.db(), _config.outputOptions.finalNamespace);
found = Helpers::findOne(opCtx, coll, temp["_id"].wrap(), old, true);
}
if (found) {
// need to reduce
values.clear();
values.push_back(temp);
values.push_back(old);
Helpers::upsert(opCtx,
_config.outputOptions.finalNamespace.ns(),
_config.reducer->finalReduce(values, _config.finalizer.get()));
} else {
Helpers::upsert(opCtx, _config.outputOptions.finalNamespace.ns(), temp);
}
pm.hit();
}
pm.finished();
}
return _collectionCount(opCtx, _config.outputOptions.finalNamespace, callerHoldsGlobalLock);
}
/**
* Insert doc in collection. This should be replicated.
*/
void State::insert(const NamespaceString& nss, const BSONObj& o) {
verify(_onDisk);
writeConflictRetry(_opCtx, "M/R insert", nss.ns(), [this, &nss, &o] {
OldClientWriteContext ctx(_opCtx, nss.ns());
WriteUnitOfWork wuow(_opCtx);
uassert(
ErrorCodes::PrimarySteppedDown,
str::stream() << "no longer primary while inserting mapReduce result into collection: "
<< nss.ns()
<< ": "
<< redact(o),
repl::ReplicationCoordinator::get(_opCtx)->canAcceptWritesFor(_opCtx, nss));
Collection* coll = getCollectionOrUassert(_opCtx, ctx.db(), nss);
BSONObjBuilder b;
if (!o.hasField("_id")) {
b.appendOID("_id", NULL, true);
}
b.appendElements(o);
BSONObj bo = b.obj();
StatusWith<BSONObj> res = fixDocumentForInsert(_opCtx->getServiceContext(), bo);
uassertStatusOK(res.getStatus());
if (!res.getValue().isEmpty()) {
bo = res.getValue();
}
// TODO: Consider whether to pass OpDebug for stats tracking under SERVER-23261.
OpDebug* const nullOpDebug = nullptr;
uassertStatusOK(coll->insertDocument(_opCtx, InsertStatement(bo), nullOpDebug, true));
wuow.commit();
});
}
/**
* Insert doc into the inc collection. This should not be replicated.
*/
void State::_insertToInc(BSONObj& o) {
verify(_onDisk);
writeConflictRetry(_opCtx, "M/R insertToInc", _config.incLong.ns(), [this, &o] {
OldClientWriteContext ctx(_opCtx, _config.incLong.ns());
WriteUnitOfWork wuow(_opCtx);
Collection* coll = getCollectionOrUassert(_opCtx, ctx.db(), _config.incLong);
repl::UnreplicatedWritesBlock uwb(_opCtx);
// The documents inserted into the incremental collection are of the form
// {"0": <key>, "1": <value>}, so we cannot call fixDocumentForInsert(o) here because the
// check that the document has an "_id" field would fail. Instead, we directly verify that
// the size of the document to insert is smaller than 16MB.
if (o.objsize() > BSONObjMaxUserSize) {
uasserted(ErrorCodes::BadValue,
str::stream() << "object to insert too large for incremental collection"
<< ". size in bytes: "
<< o.objsize()
<< ", max size: "
<< BSONObjMaxUserSize);
}
// TODO: Consider whether to pass OpDebug for stats tracking under SERVER-23261.
OpDebug* const nullOpDebug = nullptr;
uassertStatusOK(coll->insertDocument(_opCtx, InsertStatement(o), nullOpDebug, true, false));
wuow.commit();
});
}
State::State(OperationContext* opCtx, const Config& c)
: _config(c),
_db(opCtx),
_useIncremental(true),
_opCtx(opCtx),
_size(0),
_dupCount(0),
_numEmits(0) {
_temp.reset(new InMemory());
_onDisk = _config.outputOptions.outType != Config::INMEMORY;
}
bool State::sourceExists() {
return _db.exists(_config.nss.ns());
}
State::~State() {
if (_onDisk) {
try {
dropTempCollections();
} catch (...) {
error() << "Unable to drop temporary collection created by mapReduce: "
<< _config.tempNamespace << ". This collection will be removed automatically "
"the next time the server starts up. "
<< exceptionToStatus();
}
}
if (_scope && !_scope->isKillPending() && _scope->getError().empty()) {
// cleanup js objects
try {
ScriptingFunction cleanup =
_scope->createFunction("delete _emitCt; delete _keyCt; delete _mrMap;");
_scope->invoke(cleanup, 0, 0, 0, true);
} catch (const DBException&) {
// not important because properties will be reset if scope is reused
LOG(1) << "MapReduce terminated during state destruction";
}
}
}
/**
* Initialize the mapreduce operation, creating the inc collection
*/
void State::init() {
// setup js
const string userToken =
AuthorizationSession::get(Client::getCurrent())->getAuthenticatedUserNamesToken();
_scope.reset(getGlobalScriptEngine()->newScopeForCurrentThread());
_scope->requireOwnedObjects();
_scope->registerOperation(_opCtx);
_scope->setLocalDB(_config.dbname);
_scope->loadStored(_opCtx, true);
if (!_config.scopeSetup.isEmpty())
_scope->init(&_config.scopeSetup);
_config.mapper->init(this);
_config.reducer->init(this);
if (_config.finalizer)
_config.finalizer->init(this);
_scope->setBoolean("_doFinal", _config.finalizer.get() != 0);
switchMode(_config.jsMode); // set up js-mode based on Config
// global JS map/reduce hashmap
// we use a standard JS object which means keys are only simple types
// we could also add a real hashmap from a library and object comparison methods
// for increased performance, we may want to look at v8 Harmony Map support
// _scope->setObject("_mrMap", BSONObj(), false);
ScriptingFunction init = _scope->createFunction(
"_emitCt = 0;"
"_keyCt = 0;"
"_dupCt = 0;"
"_redCt = 0;"
"if (typeof(_mrMap) === 'undefined') {"
" _mrMap = {};"
"}");
_scope->invoke(init, 0, 0, 0, true);
// js function to run reduce on all keys
// redfunc = _scope->createFunction("for (var key in hashmap) { print('Key is ' + key);
// list = hashmap[key]; ret = reduce(key, list); print('Value is ' + ret); };");
_reduceAll = _scope->createFunction(
"var map = _mrMap;"
"var list, ret;"
"for (var key in map) {"
" list = map[key];"
" if (list.length != 1) {"
" ret = _reduce(key, list);"
" map[key] = [ret];"
" ++_redCt;"
" }"
"}"
"_dupCt = 0;");
massert(16717, "error initializing JavaScript reduceAll function", _reduceAll != 0);
_reduceAndEmit = _scope->createFunction(
"var map = _mrMap;"
"var list, ret;"
"for (var key in map) {"
" list = map[key];"
" if (list.length == 1)"
" ret = list[0];"
" else {"
" ret = _reduce(key, list);"
" ++_redCt;"
" }"
" emit(key, ret);"
"}"
"delete _mrMap;");
massert(16718, "error initializing JavaScript reduce/emit function", _reduceAndEmit != 0);
_reduceAndFinalize = _scope->createFunction(
"var map = _mrMap;"
"var list, ret;"
"for (var key in map) {"
" list = map[key];"
" if (list.length == 1) {"
" if (!_doFinal) { continue; }"
" ret = list[0];"
" }"
" else {"
" ret = _reduce(key, list);"
" ++_redCt;"
" }"
" if (_doFinal)"
" ret = _finalize(key, ret);"
" map[key] = ret;"
"}");
massert(16719, "error creating JavaScript reduce/finalize function", _reduceAndFinalize != 0);
_reduceAndFinalizeAndInsert = _scope->createFunction(
"var map = _mrMap;"
"var list, ret;"
"for (var key in map) {"
" list = map[key];"
" if (list.length == 1)"
" ret = list[0];"
" else {"
" ret = _reduce(key, list);"
" ++_redCt;"
" }"
" if (_doFinal)"
" ret = _finalize(key, ret);"
" _nativeToTemp({_id: key, value: ret});"
"}");
massert(16720, "error initializing JavaScript functions", _reduceAndFinalizeAndInsert != 0);
}
void State::switchMode(bool jsMode) {
_jsMode = jsMode;
if (jsMode) {
// emit function that stays in JS
_scope->setFunction("emit",
"function(key, value) {"
" if (typeof(key) === 'object') {"
" _bailFromJS(key, value);"
" return;"
" }"
" ++_emitCt;"
" var map = _mrMap;"
" var list = map[key];"
" if (!list) {"
" ++_keyCt;"
" list = [];"
" map[key] = list;"
" }"
" else"
" ++_dupCt;"
" list.push(value);"
"}");
_scope->injectNative("_bailFromJS", _bailFromJS, this);
} else {