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
|
/**
* Copyright (C) 2017 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/s/catalog/sharding_catalog_manager_impl.h"
#include "mongo/base/status_with.h"
#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/bson/util/bson_extract.h"
#include "mongo/client/connection_string.h"
#include "mongo/client/read_preference.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbdirectclient.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/operation_context.h"
#include "mongo/rpc/get_status_from_command_result.h"
#include "mongo/s/catalog/sharding_catalog_client.h"
#include "mongo/s/catalog/type_chunk.h"
#include "mongo/s/client/shard.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/grid.h"
#include "mongo/s/shard_key_pattern.h"
#include "mongo/util/fail_point_service.h"
#include "mongo/util/log.h"
#include "mongo/util/mongoutils/str.h"
namespace mongo {
namespace {
MONGO_FP_DECLARE(migrationCommitVersionError);
/**
* Append min, max and version information from chunk to the buffer for logChange purposes.
*/
void appendShortVersion(BufBuilder* b, const ChunkType& chunk) {
BSONObjBuilder bb(*b);
bb.append(ChunkType::min(), chunk.getMin());
bb.append(ChunkType::max(), chunk.getMax());
if (chunk.isVersionSet())
chunk.getVersion().addToBSON(bb, ChunkType::DEPRECATED_lastmod());
bb.done();
}
BSONArray buildMergeChunksApplyOpsUpdates(const std::vector<ChunkType>& chunksToMerge,
const ChunkVersion& mergeVersion) {
BSONArrayBuilder updates;
// Build an update operation to expand the first chunk into the newly merged chunk
{
BSONObjBuilder op;
op.append("op", "u");
op.appendBool("b", false); // no upsert
op.append("ns", ChunkType::ConfigNS);
// expand first chunk into newly merged chunk
ChunkType mergedChunk(chunksToMerge.front());
mergedChunk.setMax(chunksToMerge.back().getMax());
// fill in additional details for sending through applyOps
mergedChunk.setVersion(mergeVersion);
// add the new chunk information as the update object
op.append("o", mergedChunk.toConfigBSON());
// query object
op.append("o2", BSON(ChunkType::name(mergedChunk.getName())));
updates.append(op.obj());
}
// Build update operations to delete the rest of the chunks to be merged. Remember not
// to delete the first chunk we're expanding
for (size_t i = 1; i < chunksToMerge.size(); ++i) {
BSONObjBuilder op;
op.append("op", "d");
op.append("ns", ChunkType::ConfigNS);
op.append("o", BSON(ChunkType::name(chunksToMerge[i].getName())));
updates.append(op.obj());
}
return updates.arr();
}
BSONArray buildMergeChunksApplyOpsPrecond(const std::vector<ChunkType>& chunksToMerge,
const ChunkVersion& collVersion) {
BSONArrayBuilder preCond;
for (auto chunk : chunksToMerge) {
BSONObjBuilder b;
b.append("ns", ChunkType::ConfigNS);
b.append(
"q",
BSON("query" << BSON(ChunkType::ns(chunk.getNS()) << ChunkType::min(chunk.getMin())
<< ChunkType::max(chunk.getMax()))
<< "orderby"
<< BSON(ChunkType::DEPRECATED_lastmod() << -1)));
b.append("res",
BSON(ChunkType::DEPRECATED_epoch(collVersion.epoch())
<< ChunkType::shard(chunk.getShard().toString())));
preCond.append(b.obj());
}
return preCond.arr();
}
/**
* Checks that the epoch in the version the shard sent with the command matches the epoch of the
* collection version found on the config server. It is possible for a migration to end up running
* partly without the protection of the distributed lock. This function checks that the collection
* has not been dropped and recreated since the migration began, unbeknown to the shard when the
* command was sent.
*/
Status checkCollectionVersionEpoch(OperationContext* opCtx,
const NamespaceString& nss,
const ChunkType& aChunk,
const OID& collectionEpoch) {
auto findResponseWith =
Grid::get(opCtx)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
repl::ReadConcernLevel::kLocalReadConcern,
NamespaceString(ChunkType::ConfigNS),
BSON(ChunkType::ns() << nss.ns()),
BSONObj(),
1);
if (!findResponseWith.isOK()) {
return findResponseWith.getStatus();
}
if (MONGO_FAIL_POINT(migrationCommitVersionError)) {
uassert(ErrorCodes::StaleEpoch,
"failpoint 'migrationCommitVersionError' generated error",
false);
}
if (findResponseWith.getValue().docs.empty()) {
return Status(
ErrorCodes::IncompatibleShardingMetadata,
str::stream()
<< "Could not find any chunks for collection '"
<< nss.ns()
<< "'. The collection has been dropped since the migration began. Aborting"
" migration commit for chunk ("
<< redact(aChunk.getRange().toString())
<< ").");
}
auto chunkWith = ChunkType::fromConfigBSON(findResponseWith.getValue().docs.front());
if (!chunkWith.isOK()) {
return chunkWith.getStatus();
} else if (chunkWith.getValue().getVersion().epoch() != collectionEpoch) {
return Status(ErrorCodes::StaleEpoch,
str::stream() << "The collection '" << nss.ns()
<< "' has been dropped and recreated since the migration began."
" The config server's collection version epoch is now '"
<< chunkWith.getValue().getVersion().epoch().toString()
<< "', but the shard's is "
<< collectionEpoch.toString()
<< "'. Aborting migration commit for chunk ("
<< redact(aChunk.getRange().toString())
<< ").");
}
return Status::OK();
}
Status checkChunkIsOnShard(OperationContext* opCtx,
const NamespaceString& nss,
const BSONObj& min,
const BSONObj& max,
const ShardId& shard) {
BSONObj chunkQuery =
BSON(ChunkType::ns() << nss.ns() << ChunkType::min() << min << ChunkType::max() << max
<< ChunkType::shard()
<< shard);
// Must use local read concern because we're going to perform subsequent writes.
auto findResponseWith =
Grid::get(opCtx)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
repl::ReadConcernLevel::kLocalReadConcern,
NamespaceString(ChunkType::ConfigNS),
chunkQuery,
BSONObj(),
1);
if (!findResponseWith.isOK()) {
return findResponseWith.getStatus();
}
if (findResponseWith.getValue().docs.empty()) {
return {ErrorCodes::Error(40165),
str::stream()
<< "Could not find the chunk ("
<< chunkQuery.toString()
<< ") on the shard. Cannot execute the migration commit with invalid chunks."};
}
return Status::OK();
}
BSONObj makeCommitChunkApplyOpsCommand(const NamespaceString& nss,
const ChunkType& migratedChunk,
const boost::optional<ChunkType>& controlChunk,
StringData fromShard,
StringData toShard) {
// Update migratedChunk's version and shard.
BSONArrayBuilder updates;
{
BSONObjBuilder op;
op.append("op", "u");
op.appendBool("b", false); // No upserting
op.append("ns", ChunkType::ConfigNS);
BSONObjBuilder n(op.subobjStart("o"));
n.append(ChunkType::name(), ChunkType::genID(nss.ns(), migratedChunk.getMin()));
migratedChunk.getVersion().addToBSON(n, ChunkType::DEPRECATED_lastmod());
n.append(ChunkType::ns(), nss.ns());
n.append(ChunkType::min(), migratedChunk.getMin());
n.append(ChunkType::max(), migratedChunk.getMax());
n.append(ChunkType::shard(), toShard);
n.done();
BSONObjBuilder q(op.subobjStart("o2"));
q.append(ChunkType::name(), ChunkType::genID(nss.ns(), migratedChunk.getMin()));
q.done();
updates.append(op.obj());
}
// If we have a controlChunk, update its chunk version.
if (controlChunk) {
BSONObjBuilder op;
op.append("op", "u");
op.appendBool("b", false);
op.append("ns", ChunkType::ConfigNS);
BSONObjBuilder n(op.subobjStart("o"));
n.append(ChunkType::name(), ChunkType::genID(nss.ns(), controlChunk->getMin()));
controlChunk->getVersion().addToBSON(n, ChunkType::DEPRECATED_lastmod());
n.append(ChunkType::ns(), nss.ns());
n.append(ChunkType::min(), controlChunk->getMin());
n.append(ChunkType::max(), controlChunk->getMax());
n.append(ChunkType::shard(), fromShard);
n.done();
BSONObjBuilder q(op.subobjStart("o2"));
q.append(ChunkType::name(), ChunkType::genID(nss.ns(), controlChunk->getMin()));
q.done();
updates.append(op.obj());
}
// Do not give applyOps a write concern. If applyOps tries to wait for replication, it will fail
// because of the GlobalWrite lock CommitChunkMigration already holds. Replication will not be
// able to take the lock it requires.
return BSON("applyOps" << updates.arr());
}
} // namespace
Status ShardingCatalogManagerImpl::commitChunkSplit(OperationContext* opCtx,
const NamespaceString& ns,
const OID& requestEpoch,
const ChunkRange& range,
const std::vector<BSONObj>& splitPoints,
const std::string& shardName) {
// Take _kChunkOpLock in exclusive mode to prevent concurrent chunk splits, merges, and
// migrations
// TODO(SERVER-25359): Replace with a collection-specific lock map to allow splits/merges/
// move chunks on different collections to proceed in parallel
Lock::ExclusiveLock lk(opCtx->lockState(), _kChunkOpLock);
// Get the chunk with highest version for this namespace
auto findStatus = Grid::get(opCtx)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
repl::ReadConcernLevel::kLocalReadConcern,
NamespaceString(ChunkType::ConfigNS),
BSON("ns" << ns.ns()),
BSON(ChunkType::DEPRECATED_lastmod << -1),
1);
if (!findStatus.isOK()) {
return findStatus.getStatus();
}
const auto& chunksVector = findStatus.getValue().docs;
if (chunksVector.empty())
return {ErrorCodes::IllegalOperation,
"collection does not exist, isn't sharded, or has no chunks"};
ChunkVersion collVersion =
ChunkVersion::fromBSON(chunksVector.front(), ChunkType::DEPRECATED_lastmod());
// Return an error if epoch of chunk does not match epoch of request
if (collVersion.epoch() != requestEpoch) {
return {ErrorCodes::StaleEpoch,
"epoch of chunk does not match epoch of request. This most likely means "
"that the collection was dropped and re-created."};
}
std::vector<ChunkType> newChunks;
ChunkVersion currentMaxVersion = collVersion;
auto startKey = range.getMin();
auto newChunkBounds(splitPoints);
newChunkBounds.push_back(range.getMax());
BSONArrayBuilder updates;
for (const auto& endKey : newChunkBounds) {
// Verify the split points are all within the chunk
if (endKey.woCompare(range.getMax()) != 0 && !range.containsKey(endKey)) {
return {ErrorCodes::InvalidOptions,
str::stream() << "Split key " << endKey << " not contained within chunk "
<< range.toString()};
}
// Verify the split points came in increasing order
if (endKey.woCompare(startKey) < 0) {
return {
ErrorCodes::InvalidOptions,
str::stream() << "Split keys must be specified in strictly increasing order. Key "
<< endKey
<< " was specified after "
<< startKey
<< "."};
}
// Verify that splitPoints are not repeated
if (endKey.woCompare(startKey) == 0) {
return {ErrorCodes::InvalidOptions,
str::stream() << "Split on lower bound of chunk "
<< ChunkRange(startKey, endKey).toString()
<< "is not allowed"};
}
// verify that splits don't create too-big shard keys
Status shardKeyStatus = ShardKeyPattern::checkShardKeySize(endKey);
if (!shardKeyStatus.isOK()) {
return shardKeyStatus;
}
// splits only update the 'minor' portion of version
currentMaxVersion.incMinor();
// build an update operation against the chunks collection of the config database
// with upsert true
BSONObjBuilder op;
op.append("op", "u");
op.appendBool("b", true);
op.append("ns", ChunkType::ConfigNS);
// add the modified (new) chunk information as the update object
BSONObjBuilder n(op.subobjStart("o"));
n.append(ChunkType::name(), ChunkType::genID(ns.ns(), startKey));
currentMaxVersion.addToBSON(n, ChunkType::DEPRECATED_lastmod());
n.append(ChunkType::ns(), ns.ns());
n.append(ChunkType::min(), startKey);
n.append(ChunkType::max(), endKey);
n.append(ChunkType::shard(), shardName);
n.done();
// add the chunk's _id as the query part of the update statement
BSONObjBuilder q(op.subobjStart("o2"));
q.append(ChunkType::name(), ChunkType::genID(ns.ns(), startKey));
q.done();
updates.append(op.obj());
// remember this chunk info for logging later
ChunkType chunk;
chunk.setMin(startKey);
chunk.setMax(endKey);
chunk.setVersion(currentMaxVersion);
newChunks.push_back(std::move(chunk));
startKey = endKey;
}
BSONArrayBuilder preCond;
{
BSONObjBuilder b;
b.append("ns", ChunkType::ConfigNS);
b.append("q",
BSON("query" << BSON(ChunkType::ns(ns.ns()) << ChunkType::min() << range.getMin()
<< ChunkType::max()
<< range.getMax())
<< "orderby"
<< BSON(ChunkType::DEPRECATED_lastmod() << -1)));
{
BSONObjBuilder bb(b.subobjStart("res"));
bb.append(ChunkType::DEPRECATED_epoch(), requestEpoch);
bb.append(ChunkType::shard(), shardName);
}
preCond.append(b.obj());
}
// apply the batch of updates to remote and local metadata
Status applyOpsStatus = Grid::get(opCtx)->catalogClient(opCtx)->applyChunkOpsDeprecated(
opCtx,
updates.arr(),
preCond.arr(),
ns.ns(),
currentMaxVersion,
WriteConcernOptions(),
repl::ReadConcernLevel::kLocalReadConcern);
if (!applyOpsStatus.isOK()) {
return applyOpsStatus;
}
// log changes
BSONObjBuilder logDetail;
{
BSONObjBuilder b(logDetail.subobjStart("before"));
b.append(ChunkType::min(), range.getMin());
b.append(ChunkType::max(), range.getMax());
collVersion.addToBSON(b, ChunkType::DEPRECATED_lastmod());
}
if (newChunks.size() == 2) {
appendShortVersion(&logDetail.subobjStart("left"), newChunks[0]);
appendShortVersion(&logDetail.subobjStart("right"), newChunks[1]);
Grid::get(opCtx)->catalogClient(opCtx)->logChange(
opCtx, "split", ns.ns(), logDetail.obj(), WriteConcernOptions());
} else {
BSONObj beforeDetailObj = logDetail.obj();
BSONObj firstDetailObj = beforeDetailObj.getOwned();
const int newChunksSize = newChunks.size();
for (int i = 0; i < newChunksSize; i++) {
BSONObjBuilder chunkDetail;
chunkDetail.appendElements(beforeDetailObj);
chunkDetail.append("number", i + 1);
chunkDetail.append("of", newChunksSize);
appendShortVersion(&chunkDetail.subobjStart("chunk"), newChunks[i]);
Grid::get(opCtx)->catalogClient(opCtx)->logChange(
opCtx, "multi-split", ns.ns(), chunkDetail.obj(), WriteConcernOptions());
}
}
return applyOpsStatus;
}
Status ShardingCatalogManagerImpl::commitChunkMerge(OperationContext* opCtx,
const NamespaceString& ns,
const OID& requestEpoch,
const std::vector<BSONObj>& chunkBoundaries,
const std::string& shardName) {
// This method must never be called with empty chunks to merge
invariant(!chunkBoundaries.empty());
// Take _kChunkOpLock in exclusive mode to prevent concurrent chunk splits, merges, and
// migrations
// TODO(SERVER-25359): Replace with a collection-specific lock map to allow splits/merges/
// move chunks on different collections to proceed in parallel
Lock::ExclusiveLock lk(opCtx->lockState(), _kChunkOpLock);
// Get the chunk with the highest version for this namespace
auto findStatus = Grid::get(opCtx)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
repl::ReadConcernLevel::kLocalReadConcern,
NamespaceString(ChunkType::ConfigNS),
BSON("ns" << ns.ns()),
BSON(ChunkType::DEPRECATED_lastmod << -1),
1);
if (!findStatus.isOK()) {
return findStatus.getStatus();
}
const auto& chunksVector = findStatus.getValue().docs;
if (chunksVector.empty())
return {ErrorCodes::IllegalOperation,
"collection does not exist, isn't sharded, or has no chunks"};
ChunkVersion collVersion =
ChunkVersion::fromBSON(chunksVector.front(), ChunkType::DEPRECATED_lastmod());
// Return an error if epoch of chunk does not match epoch of request
if (collVersion.epoch() != requestEpoch) {
return {ErrorCodes::StaleEpoch,
"epoch of chunk does not match epoch of request. This most likely means "
"that the collection was dropped and re-created."};
}
// Build chunks to be merged
std::vector<ChunkType> chunksToMerge;
ChunkType itChunk;
itChunk.setMax(chunkBoundaries.front());
itChunk.setNS(ns.ns());
itChunk.setShard(shardName);
// Do not use the first chunk boundary as a max bound while building chunks
for (size_t i = 1; i < chunkBoundaries.size(); ++i) {
itChunk.setMin(itChunk.getMax());
// Ensure the chunk boundaries are strictly increasing
if (chunkBoundaries[i].woCompare(itChunk.getMin()) <= 0) {
return {
ErrorCodes::InvalidOptions,
str::stream()
<< "Chunk boundaries must be specified in strictly increasing order. Boundary "
<< chunkBoundaries[i]
<< " was specified after "
<< itChunk.getMin()
<< "."};
}
itChunk.setMax(chunkBoundaries[i]);
chunksToMerge.push_back(itChunk);
}
ChunkVersion mergeVersion = collVersion;
mergeVersion.incMinor();
auto updates = buildMergeChunksApplyOpsUpdates(chunksToMerge, mergeVersion);
auto preCond = buildMergeChunksApplyOpsPrecond(chunksToMerge, collVersion);
// apply the batch of updates to remote and local metadata
Status applyOpsStatus = Grid::get(opCtx)->catalogClient(opCtx)->applyChunkOpsDeprecated(
opCtx,
updates,
preCond,
ns.ns(),
mergeVersion,
WriteConcernOptions(),
repl::ReadConcernLevel::kLocalReadConcern);
if (!applyOpsStatus.isOK()) {
return applyOpsStatus;
}
// log changes
BSONObjBuilder logDetail;
{
BSONArrayBuilder b(logDetail.subarrayStart("merged"));
for (auto chunkToMerge : chunksToMerge) {
b.append(chunkToMerge.toConfigBSON());
}
}
collVersion.addToBSON(logDetail, "prevShardVersion");
mergeVersion.addToBSON(logDetail, "mergedVersion");
Grid::get(opCtx)->catalogClient(opCtx)->logChange(
opCtx, "merge", ns.ns(), logDetail.obj(), WriteConcernOptions());
return applyOpsStatus;
}
StatusWith<BSONObj> ShardingCatalogManagerImpl::commitChunkMigration(
OperationContext* opCtx,
const NamespaceString& nss,
const ChunkType& migratedChunk,
const boost::optional<ChunkType>& controlChunk,
const OID& collectionEpoch,
const ShardId& fromShard,
const ShardId& toShard) {
// Take _kChunkOpLock in exclusive mode to prevent concurrent chunk splits, merges, and
// migrations.
//
// ConfigSvrCommitChunkMigration commands must be run serially because the new ChunkVersions
// for migrated chunks are generated within the command and must be committed to the database
// before another chunk commit generates new ChunkVersions in the same manner.
//
// TODO(SERVER-25359): Replace with a collection-specific lock map to allow splits/merges/
// move chunks on different collections to proceed in parallel.
// (Note: This is not needed while we have a global lock, taken here only for consistency.)
Lock::ExclusiveLock lk(opCtx->lockState(), _kChunkOpLock);
// Ensure that the epoch passed in still matches the real state of the database.
auto epochCheck = checkCollectionVersionEpoch(opCtx, nss, migratedChunk, collectionEpoch);
if (!epochCheck.isOK()) {
return epochCheck;
}
// Check that migratedChunk and controlChunk are where they should be, on fromShard.
auto migratedOnShard =
checkChunkIsOnShard(opCtx, nss, migratedChunk.getMin(), migratedChunk.getMax(), fromShard);
if (!migratedOnShard.isOK()) {
return migratedOnShard;
}
if (controlChunk) {
auto controlOnShard = checkChunkIsOnShard(
opCtx, nss, controlChunk->getMin(), controlChunk->getMax(), fromShard);
if (!controlOnShard.isOK()) {
return controlOnShard;
}
}
// Must use local read concern because we will perform subsequent writes.
auto findResponse = Grid::get(opCtx)->shardRegistry()->getConfigShard()->exhaustiveFindOnConfig(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
repl::ReadConcernLevel::kLocalReadConcern,
NamespaceString(ChunkType::ConfigNS),
BSON("ns" << nss.ns()),
BSON(ChunkType::DEPRECATED_lastmod << -1),
1);
if (!findResponse.isOK()) {
return findResponse.getStatus();
}
std::vector<BSONObj> chunksVector = std::move(findResponse.getValue().docs);
if (chunksVector.empty()) {
return Status(ErrorCodes::Error(40164),
str::stream() << "Tried to find max chunk version for collection '"
<< nss.ns()
<< ", but found no chunks");
}
ChunkVersion currentMaxVersion =
ChunkVersion::fromBSON(chunksVector.front(), ChunkType::DEPRECATED_lastmod());
// Use the incremented major version of the result returned.
// Generate the new versions of migratedChunk and controlChunk.
// Migrating chunk's minor version will be 0.
ChunkType newMigratedChunk = migratedChunk;
newMigratedChunk.setVersion(
ChunkVersion(currentMaxVersion.majorVersion() + 1, 0, currentMaxVersion.epoch()));
// Control chunk's minor version will be 1 (if control chunk is present).
boost::optional<ChunkType> newControlChunk = boost::none;
if (controlChunk) {
newControlChunk = controlChunk.get();
newControlChunk->setVersion(
ChunkVersion(currentMaxVersion.majorVersion() + 1, 1, currentMaxVersion.epoch()));
}
auto command = makeCommitChunkApplyOpsCommand(
nss, newMigratedChunk, newControlChunk, fromShard.toString(), toShard.toString());
StatusWith<Shard::CommandResponse> applyOpsCommandResponse =
Grid::get(opCtx)->shardRegistry()->getConfigShard()->runCommandWithFixedRetryAttempts(
opCtx,
ReadPreferenceSetting{ReadPreference::PrimaryOnly},
nss.db().toString(),
command,
Shard::RetryPolicy::kIdempotent);
if (!applyOpsCommandResponse.isOK()) {
return applyOpsCommandResponse.getStatus();
}
if (!applyOpsCommandResponse.getValue().commandStatus.isOK()) {
return applyOpsCommandResponse.getValue().commandStatus;
}
BSONObjBuilder result;
newMigratedChunk.getVersion().appendWithFieldForCommands(&result, "migratedChunkVersion");
if (controlChunk) {
newControlChunk->getVersion().appendWithFieldForCommands(&result, "controlChunkVersion");
}
return result.obj();
}
} // namespace mongo
|