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
|
/**
* Copyright (C) 2020-present MongoDB, Inc.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the Server Side Public License, version 1,
* as published by MongoDB, Inc.
*
* 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
* Server Side Public License for more details.
*
* You should have received a copy of the Server Side Public License
* along with this program. If not, see
* <http://www.mongodb.com/licensing/server-side-public-license>.
*
* 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 Server Side 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_LOGV2_DEFAULT_COMPONENT ::mongo::logv2::LogComponent::kStorage
#include "mongo/platform/basic.h"
#include "mongo/db/startup_recovery.h"
#include "mongo/db/catalog/collection.h"
#include "mongo/db/catalog/create_collection.h"
#include "mongo/db/catalog/database.h"
#include "mongo/db/catalog/database_holder.h"
#include "mongo/db/catalog/multi_index_block.h"
#include "mongo/db/commands/feature_compatibility_version.h"
#include "mongo/db/commands/feature_compatibility_version_document_gen.h"
#include "mongo/db/commands/feature_compatibility_version_documentation.h"
#include "mongo/db/commands/feature_compatibility_version_parser.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/dbhelpers.h"
#include "mongo/db/index_builds_coordinator.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/rebuild_indexes.h"
#include "mongo/db/repair.h"
#include "mongo/db/repl/replication_coordinator.h"
#include "mongo/db/repl_set_member_in_standalone_mode.h"
#include "mongo/db/server_options.h"
#include "mongo/db/storage/storage_repair_observer.h"
#include "mongo/logv2/log.h"
#include "mongo/util/exit.h"
#include "mongo/util/fail_point.h"
#include "mongo/util/quick_exit.h"
#if !defined(_WIN32)
#include <sys/file.h>
#endif
namespace mongo {
namespace {
using startup_recovery::StartupRecoveryMode;
// Exit after repair has started, but before data is repaired.
MONGO_FAIL_POINT_DEFINE(exitBeforeDataRepair);
// Exit after repairing data, but before the replica set configuration is invalidated.
MONGO_FAIL_POINT_DEFINE(exitBeforeRepairInvalidatesConfig);
// Returns true if storage engine is writable.
bool isWriteableStorageEngine() {
return storageGlobalParams.engine != "devnull";
}
// Attempt to restore the featureCompatibilityVersion document if it is missing.
Status restoreMissingFeatureCompatibilityVersionDocument(OperationContext* opCtx) {
NamespaceString fcvNss(NamespaceString::kServerConfigurationNamespace);
// If the admin database, which contains the server configuration collection with the
// featureCompatibilityVersion document, does not exist, create it.
auto databaseHolder = DatabaseHolder::get(opCtx);
const TenantDatabaseName fcvTenantDbName(boost::none, fcvNss.db());
auto db = databaseHolder->getDb(opCtx, fcvTenantDbName);
if (!db) {
LOGV2(20998, "Re-creating admin database that was dropped.");
}
db = databaseHolder->openDb(opCtx, fcvTenantDbName);
invariant(db);
// If the server configuration collection, which contains the FCV document, does not exist, then
// create it.
auto catalog = CollectionCatalog::get(opCtx);
if (!catalog->lookupCollectionByNamespace(opCtx,
NamespaceString::kServerConfigurationNamespace)) {
// (Generic FCV reference): This FCV reference should exist across LTS binary versions.
LOGV2(4926905,
"Re-creating featureCompatibilityVersion document that was deleted. Creating new "
"document with last LTS version.",
"version"_attr = multiversion::toString(multiversion::GenericFCV::kLastLTS));
uassertStatusOK(
createCollection(opCtx, fcvNss.db().toString(), BSON("create" << fcvNss.coll())));
}
const CollectionPtr& fcvColl =
catalog->lookupCollectionByNamespace(opCtx, NamespaceString::kServerConfigurationNamespace);
invariant(fcvColl);
// Restore the featureCompatibilityVersion document if it is missing.
BSONObj featureCompatibilityVersion;
if (!Helpers::findOne(opCtx,
fcvColl,
BSON("_id" << multiversion::kParameterName),
featureCompatibilityVersion)) {
// (Generic FCV reference): This FCV reference should exist across LTS binary versions.
LOGV2(21000,
"Re-creating featureCompatibilityVersion document that was deleted. Creating new "
"document with version ",
"version"_attr = multiversion::toString(multiversion::GenericFCV::kLastLTS));
FeatureCompatibilityVersionDocument fcvDoc;
// (Generic FCV reference): This FCV reference should exist across LTS binary versions.
fcvDoc.setVersion(multiversion::GenericFCV::kLastLTS);
writeConflictRetry(opCtx, "insertFCVDocument", fcvNss.ns(), [&] {
WriteUnitOfWork wunit(opCtx);
OpDebug* const nullOpDebug = nullptr;
uassertStatusOK(fcvColl->insertDocument(
opCtx, InsertStatement(fcvDoc.toBSON()), nullOpDebug, false));
wunit.commit();
});
}
invariant(Helpers::findOne(
opCtx, fcvColl, BSON("_id" << multiversion::kParameterName), featureCompatibilityVersion));
return Status::OK();
}
/**
* Returns true if the collection associated with the given CollectionCatalogEntry has an index on
* the _id field
*/
bool checkIdIndexExists(OperationContext* opCtx, const CollectionPtr& coll) {
auto indexCount = coll->getTotalIndexCount();
auto indexNames = std::vector<std::string>(indexCount);
coll->getAllIndexes(&indexNames);
for (auto name : indexNames) {
if (name == "_id_") {
return true;
}
}
return false;
}
Status buildMissingIdIndex(OperationContext* opCtx, Collection* collection) {
LOGV2(4805002, "Building missing _id index", logAttrs(*collection));
MultiIndexBlock indexer;
ScopeGuard abortOnExit([&] {
CollectionWriter collWriter(collection);
indexer.abortIndexBuild(opCtx, collWriter, MultiIndexBlock::kNoopOnCleanUpFn);
});
const auto indexCatalog = collection->getIndexCatalog();
const auto idIndexSpec = indexCatalog->getDefaultIdIndexSpec(collection);
CollectionWriter collWriter(collection);
auto swSpecs = indexer.init(opCtx, collWriter, idIndexSpec, MultiIndexBlock::kNoopOnInitFn);
if (!swSpecs.isOK()) {
return swSpecs.getStatus();
}
auto status = indexer.insertAllDocumentsInCollection(opCtx, collection);
if (!status.isOK()) {
return status;
}
status = indexer.checkConstraints(opCtx, collection);
if (!status.isOK()) {
return status;
}
WriteUnitOfWork wuow(opCtx);
status = indexer.commit(
opCtx, collection, MultiIndexBlock::kNoopOnCreateEachFn, MultiIndexBlock::kNoopOnCommitFn);
wuow.commit();
abortOnExit.dismiss();
return status;
}
auto downgradeError =
Status{ErrorCodes::MustDowngrade,
str::stream() << "UPGRADE PROBLEM: The data files need to be fully upgraded to version "
"4.4 before attempting a binary upgrade; see "
<< feature_compatibility_version_documentation::kUpgradeLink
<< " for more details."};
/**
* Checks that all collections on a database have valid properties for this version of MongoDB.
*
* This validates that required collections have an _id index. If a collection is missing an _id
* index, this function will build it if EnsureIndexPolicy is kBuildMissing.
*
* Returns a MustDowngrade error if any index builds on the required _id field fail.
*/
enum class EnsureIndexPolicy { kBuildMissing, kError };
Status ensureCollectionProperties(OperationContext* opCtx,
Database* db,
EnsureIndexPolicy ensureIndexPolicy) {
auto catalog = CollectionCatalog::get(opCtx);
for (auto collIt = catalog->begin(opCtx, db->name()); collIt != catalog->end(opCtx); ++collIt) {
auto coll = *collIt;
if (!coll) {
break;
}
// All user-created replicated collections created since MongoDB 4.0 have _id indexes.
auto requiresIndex = coll->requiresIdIndex() && coll->ns().isReplicated();
const auto& collOptions = coll->getCollectionOptions();
auto hasAutoIndexIdField = collOptions.autoIndexId == CollectionOptions::YES;
// Even if the autoIndexId field is not YES, the collection may still have an _id index
// that was created manually by the user. Check the list of indexes to confirm index
// does not exist before attempting to build it or returning an error.
if (requiresIndex && !hasAutoIndexIdField && !checkIdIndexExists(opCtx, coll)) {
LOGV2(21001,
"collection {coll_ns} is missing an _id index",
"Collection is missing an _id index",
logAttrs(*coll.get()));
if (EnsureIndexPolicy::kBuildMissing == ensureIndexPolicy) {
auto status = buildMissingIdIndex(opCtx, collIt.getWritableCollection(opCtx));
if (!status.isOK()) {
LOGV2_ERROR(21021,
"could not build an _id index on collection {coll_ns}: {error}",
"Could not build an _id index on collection",
logAttrs(*coll.get()),
"error"_attr = status);
return downgradeError;
}
} else {
return downgradeError;
}
}
}
return Status::OK();
}
/**
* Opens each database and provides a callback on each one.
*/
template <typename Func>
void openDatabases(OperationContext* opCtx, const StorageEngine* storageEngine, Func&& onDatabase) {
invariant(opCtx->lockState()->isW());
auto databaseHolder = DatabaseHolder::get(opCtx);
auto tenantDbNames = storageEngine->listDatabases();
for (const auto& tenantDbName : tenantDbNames) {
LOGV2_DEBUG(21010, 1, " Opening database: {dbName}", "dbName"_attr = tenantDbName);
auto db = databaseHolder->openDb(opCtx, tenantDbName);
invariant(db);
onDatabase(db);
}
}
/**
* Returns 'true' if this server has a configuration document in local.system.replset.
*/
bool hasReplSetConfigDoc(OperationContext* opCtx) {
auto databaseHolder = DatabaseHolder::get(opCtx);
// We open the "local" database before reading to ensure the in-memory catalog entries for the
// 'kSystemReplSetNamespace' collection have been populated if the collection exists. If the
// "local" database doesn't exist at this point yet, then it will be created.
const auto nss = NamespaceString::kSystemReplSetNamespace;
const TenantDatabaseName tenantDbName(boost::none, nss.db());
databaseHolder->openDb(opCtx, tenantDbName);
BSONObj config;
return Helpers::getSingleton(opCtx, nss.ns().c_str(), config);
}
/**
* Check that the oplog is capped, and abort the process if it is not.
* Caller must lock DB before calling this function.
*/
void assertCappedOplog(OperationContext* opCtx, Database* db) {
const NamespaceString oplogNss(NamespaceString::kRsOplogNamespace);
invariant(opCtx->lockState()->isDbLockedForMode(oplogNss.db(), MODE_IS));
const CollectionPtr& oplogCollection =
CollectionCatalog::get(opCtx)->lookupCollectionByNamespace(opCtx, oplogNss);
if (oplogCollection && !oplogCollection->isCapped()) {
LOGV2_FATAL_NOTRACE(
40115,
"The oplog collection {oplogNamespace} is not capped; a capped oplog is a "
"requirement for replication to function.",
"The oplog collection is not capped; a capped oplog is a "
"requirement for replication to function.",
"oplogNamespace"_attr = oplogNss);
}
}
void clearTempFilesExceptForResumableBuilds(const std::vector<ResumeIndexInfo>& indexBuildsToResume,
const boost::filesystem::path& tempDir) {
StringSet resumableIndexFiles;
for (const auto& resumeInfo : indexBuildsToResume) {
const auto& indexes = resumeInfo.getIndexes();
for (const auto& index : indexes) {
boost::optional<StringData> indexFilename = index.getFileName();
if (indexFilename) {
resumableIndexFiles.insert(indexFilename->toString());
}
}
}
auto dirItr = boost::filesystem::directory_iterator(tempDir);
auto dirEnd = boost::filesystem::directory_iterator();
for (; dirItr != dirEnd; ++dirItr) {
auto curFilename = dirItr->path().filename().string();
if (!resumableIndexFiles.contains(curFilename)) {
boost::system::error_code ec;
boost::filesystem::remove(dirItr->path(), ec);
if (ec) {
LOGV2(5676601,
"Failed to clear temp directory file",
"filename"_attr = curFilename,
"error"_attr = ec.message());
}
}
}
}
void reconcileCatalogAndRebuildUnfinishedIndexes(
OperationContext* opCtx,
StorageEngine* storageEngine,
StorageEngine::LastShutdownState lastShutdownState) {
auto reconcileResult =
fassert(40593, storageEngine->reconcileCatalogAndIdents(opCtx, lastShutdownState));
auto tempDir = boost::filesystem::path(storageGlobalParams.dbpath).append("_tmp");
if (reconcileResult.indexBuildsToResume.empty() ||
lastShutdownState == StorageEngine::LastShutdownState::kUnclean) {
// If we did not find any index builds to resume or we are starting up after an unclean
// shutdown, nothing in the temp directory will be used. Thus, we can clear it completely.
LOGV2(5071100, "Clearing temp directory");
boost::system::error_code ec;
boost::filesystem::remove_all(tempDir, ec);
if (ec) {
LOGV2(5071101, "Failed to clear temp directory", "error"_attr = ec.message());
}
} else if (boost::filesystem::exists(tempDir)) {
// Clears the contents of the temp directory except for files for resumable builds.
LOGV2(5676600, "Clearing temp directory except for files for resumable builds");
clearTempFilesExceptForResumableBuilds(reconcileResult.indexBuildsToResume, tempDir);
}
// Determine which indexes need to be rebuilt. rebuildIndexesOnCollection() requires that all
// indexes on that collection are done at once, so we use a map to group them together.
StringMap<IndexNameObjs> nsToIndexNameObjMap;
auto catalog = CollectionCatalog::get(opCtx);
for (auto&& idxIdentifier : reconcileResult.indexesToRebuild) {
NamespaceString collNss = idxIdentifier.nss;
const std::string& indexName = idxIdentifier.indexName;
auto swIndexSpecs =
getIndexNameObjs(catalog->lookupCollectionByNamespace(opCtx, collNss),
[&indexName](const std::string& name) { return name == indexName; });
if (!swIndexSpecs.isOK() || swIndexSpecs.getValue().first.empty()) {
fassert(40590,
{ErrorCodes::InternalError,
str::stream() << "failed to get index spec for index " << indexName
<< " in collection " << collNss.toString()});
}
auto& indexesToRebuild = swIndexSpecs.getValue();
invariant(indexesToRebuild.first.size() == 1 && indexesToRebuild.second.size() == 1,
str::stream() << "Num Index Names: " << indexesToRebuild.first.size()
<< " Num Index Objects: " << indexesToRebuild.second.size());
auto& ino = nsToIndexNameObjMap[collNss.ns()];
ino.first.emplace_back(std::move(indexesToRebuild.first.back()));
ino.second.emplace_back(std::move(indexesToRebuild.second.back()));
}
for (const auto& entry : nsToIndexNameObjMap) {
NamespaceString collNss(entry.first);
auto collection = catalog->lookupCollectionByNamespace(opCtx, collNss);
for (const auto& indexName : entry.second.first) {
LOGV2(21004,
"Rebuilding index. Collection: {collNss} Index: {indexName}",
"Rebuilding index",
"namespace"_attr = collNss,
"index"_attr = indexName);
}
std::vector<BSONObj> indexSpecs = entry.second.second;
fassert(40592, rebuildIndexesOnCollection(opCtx, collection, indexSpecs, RepairData::kNo));
}
// Two-phase index builds depend on an eventually-replicated 'commitIndexBuild' oplog entry to
// complete. Therefore, when a replica set member is started in standalone mode, we cannot
// restart the index build because it will never complete.
if (getReplSetMemberInStandaloneMode(opCtx->getServiceContext())) {
LOGV2(21005, "Not restarting unfinished index builds because we are in standalone mode");
return;
}
// Once all unfinished indexes have been rebuilt, restart any unfinished index builds. This will
// not build any indexes to completion, but rather start the background thread to build the
// index, and wait for a replicated commit or abort oplog entry.
IndexBuildsCoordinator::get(opCtx)->restartIndexBuildsForRecovery(
opCtx, reconcileResult.indexBuildsToRestart, reconcileResult.indexBuildsToResume);
}
/**
* Sets the appropriate flag on the service context decorable 'replSetMemberInStandaloneMode' to
* 'true' if this is a replica set node running in standalone mode, otherwise 'false'.
*/
void setReplSetMemberInStandaloneMode(OperationContext* opCtx, StartupRecoveryMode mode) {
if (mode == StartupRecoveryMode::kReplicaSetMember) {
setReplSetMemberInStandaloneMode(opCtx->getServiceContext(), false);
return;
} else if (mode == StartupRecoveryMode::kReplicaSetMemberInStandalone) {
setReplSetMemberInStandaloneMode(opCtx->getServiceContext(), true);
return;
}
const bool usingReplication = repl::ReplicationCoordinator::get(opCtx)->isReplEnabled();
if (usingReplication) {
// Not in standalone mode.
setReplSetMemberInStandaloneMode(opCtx->getServiceContext(), false);
return;
}
invariant(opCtx->lockState()->isW());
CollectionPtr collection = CollectionCatalog::get(opCtx)->lookupCollectionByNamespace(
opCtx, NamespaceString::kSystemReplSetNamespace);
if (collection && !collection->isEmpty(opCtx)) {
setReplSetMemberInStandaloneMode(opCtx->getServiceContext(), true);
return;
}
setReplSetMemberInStandaloneMode(opCtx->getServiceContext(), false);
}
// Perform startup procedures for --repair mode.
void startupRepair(OperationContext* opCtx, StorageEngine* storageEngine) {
invariant(!storageGlobalParams.queryableBackupMode);
if (MONGO_unlikely(exitBeforeDataRepair.shouldFail())) {
LOGV2(21006, "Exiting because 'exitBeforeDataRepair' fail point was set.");
quickExit(EXIT_ABRUPT);
}
// Repair, restore, and initialize the featureCompatibilityVersion document before allowing
// repair to potentially rebuild indexes on the remaining collections. This ensures any
// FCV-dependent features are rebuilt properly. Note that we don't try to prevent
// repairDatabase from repairing this collection again, because it only consists of one
// document.
// If we fail to load the FCV document due to upgrade problems, we need to abort the repair in
// order to allow downgrading to older binary versions.
ScopeGuard abortRepairOnFCVErrors(
[&] { StorageRepairObserver::get(opCtx->getServiceContext())->onRepairDone(opCtx); });
auto catalog = CollectionCatalog::get(opCtx);
if (auto fcvColl = catalog->lookupCollectionByNamespace(
opCtx, NamespaceString::kServerConfigurationNamespace)) {
auto databaseHolder = DatabaseHolder::get(opCtx);
const TenantDatabaseName fcvTenantDbName(boost::none, fcvColl->ns().db());
databaseHolder->openDb(opCtx, fcvTenantDbName);
fassertNoTrace(4805000,
repair::repairCollection(
opCtx, storageEngine, NamespaceString::kServerConfigurationNamespace));
}
uassertStatusOK(restoreMissingFeatureCompatibilityVersionDocument(opCtx));
FeatureCompatibilityVersion::initializeForStartup(opCtx);
abortRepairOnFCVErrors.dismiss();
// The local database should be repaired before any other replicated collections so we know
// whether not to rebuild unfinished two-phase index builds if this is a replica set node
// running in standalone mode.
auto tenantDbNames = storageEngine->listDatabases();
if (auto it = std::find(tenantDbNames.begin(),
tenantDbNames.end(),
TenantDatabaseName(boost::none, NamespaceString::kLocalDb));
it != tenantDbNames.end()) {
fassertNoTrace(4805001, repair::repairDatabase(opCtx, storageEngine, *it));
// This must be set before rebuilding index builds on replicated collections.
setReplSetMemberInStandaloneMode(opCtx, StartupRecoveryMode::kAuto);
tenantDbNames.erase(it);
}
// Repair the remaining databases.
for (const auto& tenantDbName : tenantDbNames) {
fassertNoTrace(18506, repair::repairDatabase(opCtx, storageEngine, tenantDbName));
}
openDatabases(opCtx, storageEngine, [&](auto db) {
// Ensures all collections meet requirements such as having _id indexes, and corrects them
// if needed.
uassertStatusOK(ensureCollectionProperties(opCtx, db, EnsureIndexPolicy::kBuildMissing));
});
if (MONGO_unlikely(exitBeforeRepairInvalidatesConfig.shouldFail())) {
LOGV2(21008, "Exiting because 'exitBeforeRepairInvalidatesConfig' fail point was set.");
quickExit(EXIT_ABRUPT);
}
auto repairObserver = StorageRepairObserver::get(opCtx->getServiceContext());
repairObserver->onRepairDone(opCtx);
if (repairObserver->getModifications().size() > 0) {
const auto& mods = repairObserver->getModifications();
for (const auto& mod : mods) {
LOGV2_WARNING(21019, "repairModification", "description"_attr = mod.getDescription());
}
}
if (repairObserver->isDataInvalidated()) {
if (hasReplSetConfigDoc(opCtx)) {
LOGV2_WARNING(21020,
"WARNING: Repair may have modified replicated data. This node will no "
"longer be able to join a replica set without a full re-sync");
}
}
// There were modifications, but only benign ones.
if (repairObserver->getModifications().size() > 0 && !repairObserver->isDataInvalidated()) {
LOGV2(21009,
"Repair has made modifications to unreplicated data. The data is healthy and "
"the node is eligible to be returned to the replica set.");
}
}
// Perform routine startup recovery procedure.
void startupRecovery(OperationContext* opCtx,
StorageEngine* storageEngine,
StorageEngine::LastShutdownState lastShutdownState,
StartupRecoveryMode mode) {
invariant(!storageGlobalParams.repair);
// Determine whether this is a replica set node running in standalone mode. This must be set
// before determining whether to restart index builds.
setReplSetMemberInStandaloneMode(opCtx, mode);
// Initialize FCV before rebuilding indexes that may have features dependent on FCV.
FeatureCompatibilityVersion::initializeForStartup(opCtx);
// Drops abandoned idents. Rebuilds unfinished indexes and restarts incomplete two-phase
// index builds.
reconcileCatalogAndRebuildUnfinishedIndexes(opCtx, storageEngine, lastShutdownState);
const bool usingReplication = repl::ReplicationCoordinator::get(opCtx)->isReplEnabled();
// On replica set members we only clear temp collections on DBs other than "local" during
// promotion to primary. On secondaries, they are only cleared when the oplog tells them to. The
// local DB is special because it is not replicated. See SERVER-10927 for more details.
const bool shouldClearNonLocalTmpCollections =
!(hasReplSetConfigDoc(opCtx) || usingReplication);
openDatabases(opCtx, storageEngine, [&](auto db) {
auto dbName = db->name().dbName();
// Ensures all collections meet requirements such as having _id indexes, and corrects them
// if needed.
uassertStatusOK(ensureCollectionProperties(opCtx, db, EnsureIndexPolicy::kBuildMissing));
if (usingReplication) {
// We only care about _id indexes and drop-pending collections if we are in a replset.
db->checkForIdIndexesAndDropPendingCollections(opCtx);
// Ensure oplog is capped (mongodb does not guarantee order of inserts on noncapped
// collections)
if (dbName == NamespaceString::kLocalDb) {
assertCappedOplog(opCtx, db);
}
}
if (shouldClearNonLocalTmpCollections || dbName == NamespaceString::kLocalDb) {
db->clearTmpCollections(opCtx);
}
});
}
} // namespace
namespace startup_recovery {
/**
* Recovers or repairs all databases from a previous shutdown. May throw a MustDowngrade error
* if data files are incompatible with the current binary version.
*/
void repairAndRecoverDatabases(OperationContext* opCtx,
StorageEngine::LastShutdownState lastShutdownState) {
auto const storageEngine = opCtx->getServiceContext()->getStorageEngine();
Lock::GlobalWrite lk(opCtx);
// Use the BatchedCollectionCatalogWriter so all Collection writes to the in-memory catalog are
// done in a single copy-on-write of the catalog. This avoids quadratic behavior where we
// iterate over every collection and perform writes where the catalog would be copied every
// time.
BatchedCollectionCatalogWriter catalog(opCtx);
// Create the FCV document for the first time, if necessary. Replica set nodes only initialize
// the FCV when the replica set is first initiated or by data replication.
const bool usingReplication = repl::ReplicationCoordinator::get(opCtx)->isReplEnabled();
if (isWriteableStorageEngine() && !usingReplication) {
FeatureCompatibilityVersion::setIfCleanStartup(opCtx, repl::StorageInterface::get(opCtx));
}
if (storageGlobalParams.repair) {
startupRepair(opCtx, storageEngine);
} else {
startupRecovery(opCtx, storageEngine, lastShutdownState, StartupRecoveryMode::kAuto);
}
}
/**
* Runs startup recovery after system startup, either in replSet mode (will
* restart index builds) or replSet standalone mode (will not restart index builds). In no
* case will it create an FCV document nor run repair or read-only recovery.
*/
void runStartupRecoveryInMode(OperationContext* opCtx,
StorageEngine::LastShutdownState lastShutdownState,
StartupRecoveryMode mode) {
auto const storageEngine = opCtx->getServiceContext()->getStorageEngine();
Lock::GlobalWrite lk(opCtx);
invariant(isWriteableStorageEngine());
invariant(!storageGlobalParams.repair);
const bool usingReplication = repl::ReplicationCoordinator::get(opCtx)->isReplEnabled();
invariant(usingReplication);
invariant(mode == StartupRecoveryMode::kReplicaSetMember ||
mode == StartupRecoveryMode::kReplicaSetMemberInStandalone);
startupRecovery(opCtx, storageEngine, lastShutdownState, mode);
}
} // namespace startup_recovery
} // namespace mongo
|