summaryrefslogtreecommitdiff
path: root/src/mongo/s/catalog_cache.cpp
blob: c74d3b2ac28cf60082e73fb3d5b5107e5fa40fc9 (plain)
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
/**
 *    Copyright (C) 2018-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::kSharding

#define LOGV2_FOR_CATALOG_REFRESH(ID, DLEVEL, MESSAGE, ...) \
    LOGV2_DEBUG_OPTIONS(                                    \
        ID, DLEVEL, {logv2::LogComponent::kShardingCatalogRefresh}, MESSAGE, ##__VA_ARGS__)

#include "mongo/platform/basic.h"

#include "mongo/s/catalog_cache.h"

#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/db/query/collation/collator_factory_interface.h"
#include "mongo/db/repl/optime_with.h"
#include "mongo/logv2/log.h"
#include "mongo/s/catalog/type_collection.h"
#include "mongo/s/catalog/type_database.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/database_version_helpers.h"
#include "mongo/s/grid.h"
#include "mongo/s/is_mongos.h"
#include "mongo/s/mongod_and_mongos_server_parameters_gen.h"
#include "mongo/s/stale_exception.h"
#include "mongo/util/concurrency/with_lock.h"
#include "mongo/util/scopeguard.h"
#include "mongo/util/timer.h"

namespace mongo {
namespace {

// How many times to try refreshing the routing info if the set of chunks loaded from the config
// server is found to be inconsistent.
const int kMaxInconsistentRoutingInfoRefreshAttempts = 3;

const int kDatabaseCacheSize = 10000;
const int kCollectionCacheSize = 10000;

const OperationContext::Decoration<bool> operationShouldBlockBehindCatalogCacheRefresh =
    OperationContext::declareDecoration<bool>();

const OperationContext::Decoration<bool> operationBlockedBehindCatalogCacheRefresh =
    OperationContext::declareDecoration<bool>();

}  // namespace

CatalogCache::CatalogCache(ServiceContext* const service, CatalogCacheLoader& cacheLoader)
    : _cacheLoader(cacheLoader),
      _executor(std::make_shared<ThreadPool>([] {
          ThreadPool::Options options;
          options.poolName = "CatalogCache";
          options.minThreads = 0;
          options.maxThreads = 6;
          return options;
      }())),
      _databaseCache(service, *_executor, _cacheLoader),
      _collectionCache(service, *_executor, _cacheLoader) {
    _executor->startup();
}

CatalogCache::~CatalogCache() {
    // The executor is used by the Database and Collection caches,
    // so it must be joined, before these caches are destroyed,
    // per the contract of ReadThroughCache.
    _executor->shutdown();
    _executor->join();
}

StatusWith<CachedDatabaseInfo> CatalogCache::getDatabase(OperationContext* opCtx,
                                                         StringData dbName,
                                                         bool allowLocks) {
    if (!allowLocks) {
        invariant(
            !opCtx->lockState() || !opCtx->lockState()->isLocked(),
            "Do not hold a lock while refreshing the catalog cache. Doing so would potentially "
            "hold the lock during a network call, and can lead to a deadlock as described in "
            "SERVER-37398.");
    }

    try {
        // TODO SERVER-49724: Make ReadThroughCache support StringData keys
        auto dbEntry =
            _databaseCache.acquire(opCtx, dbName.toString(), CacheCausalConsistency::kLatestKnown);
        uassert(ErrorCodes::NamespaceNotFound,
                str::stream() << "database " << dbName << " not found",
                dbEntry);

        return {CachedDatabaseInfo(*dbEntry)};
    } catch (const DBException& ex) {
        return ex.toStatus();
    }
}

StatusWith<ChunkManager> CatalogCache::_getCollectionRoutingInfoAt(
    OperationContext* opCtx,
    const NamespaceString& nss,
    boost::optional<Timestamp> atClusterTime,
    bool allowLocks) {
    if (!allowLocks) {
        invariant(!opCtx->lockState() || !opCtx->lockState()->isLocked(),
                  "Do not hold a lock while refreshing the catalog cache. Doing so would "
                  "potentially hold "
                  "the lock during a network call, and can lead to a deadlock as described in "
                  "SERVER-37398.");
    }

    try {
        const auto swDbInfo = getDatabase(opCtx, nss.db(), allowLocks);

        if (!swDbInfo.isOK()) {
            if (swDbInfo == ErrorCodes::NamespaceNotFound) {
                LOGV2_FOR_CATALOG_REFRESH(
                    4947103,
                    2,
                    "Invalidating cached collection entry because its database has been dropped",
                    "namespace"_attr = nss);
                invalidateCollectionEntry_LINEARIZABLE(nss);
            }
            return swDbInfo.getStatus();
        }

        const auto dbInfo = std::move(swDbInfo.getValue());

        const auto cacheConsistency = gEnableFinerGrainedCatalogCacheRefresh &&
                !operationShouldBlockBehindCatalogCacheRefresh(opCtx)
            ? CacheCausalConsistency::kLatestCached
            : CacheCausalConsistency::kLatestKnown;

        auto collEntryFuture = _collectionCache.acquireAsync(nss, cacheConsistency);

        // If the entry is in the cache return inmediately.
        if (collEntryFuture.isReady()) {
            setOperationShouldBlockBehindCatalogCacheRefresh(opCtx, false);
            return ChunkManager(dbInfo.primaryId(),
                                dbInfo.databaseVersion(),
                                collEntryFuture.get(opCtx),
                                atClusterTime);
        }

        if (allowLocks) {
            return Status{ErrorCodes::StaleShardVersion, "Routing info refresh did not complete"};
        }

        operationBlockedBehindCatalogCacheRefresh(opCtx) = true;

        size_t acquireTries = 0;
        Timer t;

        while (true) {
            try {
                auto collEntry = collEntryFuture.get(opCtx);
                _stats.totalRefreshWaitTimeMicros.addAndFetch(t.micros());

                setOperationShouldBlockBehindCatalogCacheRefresh(opCtx, false);

                return ChunkManager(dbInfo.primaryId(),
                                    dbInfo.databaseVersion(),
                                    std::move(collEntry),
                                    atClusterTime);
            } catch (ExceptionFor<ErrorCodes::ConflictingOperationInProgress>& ex) {
                _stats.totalRefreshWaitTimeMicros.addAndFetch(t.micros());
                acquireTries++;
                if (acquireTries == kMaxInconsistentRoutingInfoRefreshAttempts) {
                    return ex.toStatus();
                }
            }

            collEntryFuture = _collectionCache.acquireAsync(nss, cacheConsistency);
            t.reset();
        }
    } catch (const DBException& ex) {
        return ex.toStatus();
    }
}

StatusWith<ChunkManager> CatalogCache::getCollectionRoutingInfo(OperationContext* opCtx,
                                                                const NamespaceString& nss,
                                                                bool allowLocks) {
    return _getCollectionRoutingInfoAt(opCtx, nss, boost::none, allowLocks);
}

StatusWith<ChunkManager> CatalogCache::getCollectionRoutingInfoAt(OperationContext* opCtx,
                                                                  const NamespaceString& nss,
                                                                  Timestamp atClusterTime) {
    return _getCollectionRoutingInfoAt(opCtx, nss, atClusterTime, false);
}

StatusWith<CachedDatabaseInfo> CatalogCache::getDatabaseWithRefresh(OperationContext* opCtx,
                                                                    StringData dbName) {
    // TODO SERVER-49724: Make ReadThroughCache support StringData keys
    _databaseCache.invalidate(dbName.toString());
    return getDatabase(opCtx, dbName);
}

StatusWith<ChunkManager> CatalogCache::getCollectionRoutingInfoWithRefresh(
    OperationContext* opCtx, const NamespaceString& nss) {
    _collectionCache.invalidate(nss);
    setOperationShouldBlockBehindCatalogCacheRefresh(opCtx, true);
    return getCollectionRoutingInfo(opCtx, nss);
}

StatusWith<ChunkManager> CatalogCache::getShardedCollectionRoutingInfoWithRefresh(
    OperationContext* opCtx, const NamespaceString& nss) {
    auto routingInfoStatus = getCollectionRoutingInfoWithRefresh(opCtx, nss);
    if (routingInfoStatus.isOK() && !routingInfoStatus.getValue().isSharded()) {
        return {ErrorCodes::NamespaceNotSharded,
                str::stream() << "Collection " << nss.ns() << " is not sharded."};
    }
    return routingInfoStatus;
}

void CatalogCache::onStaleDatabaseVersion(const StringData dbName,
                                          const boost::optional<DatabaseVersion>& databaseVersion) {
    if (databaseVersion) {
        const auto version =
            ComparableDatabaseVersion::makeComparableDatabaseVersion(databaseVersion.get());
        LOGV2_FOR_CATALOG_REFRESH(4899101,
                                  2,
                                  "Registering new database version",
                                  "db"_attr = dbName,
                                  "version"_attr = version.toBSONForLogging());
        _databaseCache.advanceTimeInStore(dbName.toString(), version);
    } else {
        _databaseCache.invalidate(dbName.toString());
    }
}

void CatalogCache::setOperationShouldBlockBehindCatalogCacheRefresh(OperationContext* opCtx,
                                                                    bool shouldBlock) {
    if (gEnableFinerGrainedCatalogCacheRefresh) {
        operationShouldBlockBehindCatalogCacheRefresh(opCtx) = shouldBlock;
    }
}

void CatalogCache::invalidateShardOrEntireCollectionEntryForShardedCollection(
    const NamespaceString& nss,
    const boost::optional<ChunkVersion>& wantedVersion,
    const ShardId& shardId) {
    _stats.countStaleConfigErrors.addAndFetch(1);

    auto collectionEntry = _collectionCache.peekLatestCached(nss);

    const auto newChunkVersion = wantedVersion
        ? ComparableChunkVersion::makeComparableChunkVersion(*wantedVersion)
        : ComparableChunkVersion::makeComparableChunkVersionForForcedRefresh();

    const bool timeAdvanced = _collectionCache.advanceTimeInStore(nss, newChunkVersion);

    if (timeAdvanced && collectionEntry && collectionEntry->optRt) {
        // Shards marked stale will be reset on the next refresh.
        // We can mark the shard stale only if the time advanced, otherwise no refresh would happen
        // and the shard will remain marked stale.
        // Even if a concurrent refresh is happening this is still the old collectionEntry,
        // so it is safe to call setShardStale.
        collectionEntry->optRt->setShardStale(shardId);
    }
}

void CatalogCache::checkEpochOrThrow(const NamespaceString& nss,
                                     const ChunkVersion& targetCollectionVersion,
                                     const ShardId& shardId) {
    uassert(StaleConfigInfo(nss, targetCollectionVersion, boost::none, shardId),
            str::stream() << "could not act as router for " << nss.ns()
                          << ", no entry for database " << nss.db(),
            _databaseCache.peekLatestCached(nss.db().toString()));

    auto collectionValueHandle = _collectionCache.peekLatestCached(nss);
    uassert(StaleConfigInfo(nss, targetCollectionVersion, boost::none, shardId),
            str::stream() << "could not act as router for " << nss.ns()
                          << ", no entry for collection.",
            collectionValueHandle);

    uassert(StaleConfigInfo(nss, targetCollectionVersion, boost::none, shardId),
            str::stream() << "could not act as router for " << nss.ns() << ", wanted "
                          << targetCollectionVersion.toString()
                          << ", but found the collection was unsharded",
            collectionValueHandle->optRt);

    auto foundVersion = collectionValueHandle->optRt->getVersion();
    uassert(StaleConfigInfo(nss, targetCollectionVersion, foundVersion, shardId),
            str::stream() << "could not act as router for " << nss.ns() << ", wanted "
                          << targetCollectionVersion.toString() << ", but found "
                          << foundVersion.toString(),
            foundVersion.epoch() == targetCollectionVersion.epoch());
}

void CatalogCache::invalidateEntriesThatReferenceShard(const ShardId& shardId) {
    LOGV2_DEBUG(4997600,
                1,
                "Invalidating databases and collections referencing a specific shard",
                "shardId"_attr = shardId);

    _databaseCache.invalidateCachedValueIf(
        [&](const DatabaseType& dbt) { return dbt.getPrimary() == shardId; });

    // Invalidate collections which contain data on this shard.
    _collectionCache.invalidateCachedValueIf([&](const OptionalRoutingTableHistory& ort) {
        if (!ort.optRt)
            return false;
        const auto& rt = *ort.optRt;

        std::set<ShardId> shardIds;
        rt.getAllShardIds(&shardIds);

        LOGV2_DEBUG(22647,
                    3,
                    "Invalidating cached collection {namespace} that has data "
                    "on shard {shardId}",
                    "Invalidating cached collection",
                    "namespace"_attr = rt.nss(),
                    "shardId"_attr = shardId);
        return shardIds.find(shardId) != shardIds.end();
    });

    LOGV2(22648,
          "Finished invalidating databases and collections with data on shard: {shardId}",
          "Finished invalidating databases and collections that reference specific shard",
          "shardId"_attr = shardId);
}

void CatalogCache::purgeDatabase(StringData dbName) {
    _databaseCache.invalidate(dbName.toString());
    _collectionCache.invalidateKeyIf(
        [&](const NamespaceString& nss) { return nss.db() == dbName; });
}

void CatalogCache::purgeAllDatabases() {
    _databaseCache.invalidateAll();
    _collectionCache.invalidateAll();
}

void CatalogCache::report(BSONObjBuilder* builder) const {
    BSONObjBuilder cacheStatsBuilder(builder->subobjStart("catalogCache"));

    const size_t numDatabaseEntries = _databaseCache.getCacheInfo().size();
    const size_t numCollectionEntries = _collectionCache.getCacheInfo().size();

    cacheStatsBuilder.append("numDatabaseEntries", static_cast<long long>(numDatabaseEntries));
    cacheStatsBuilder.append("numCollectionEntries", static_cast<long long>(numCollectionEntries));

    _stats.report(&cacheStatsBuilder);
    _collectionCache.reportStats(&cacheStatsBuilder);
}

void CatalogCache::checkAndRecordOperationBlockedByRefresh(OperationContext* opCtx,
                                                           mongo::LogicalOp opType) {
    if (!isMongos() || !operationBlockedBehindCatalogCacheRefresh(opCtx)) {
        return;
    }

    auto& opsBlockedByRefresh = _stats.operationsBlockedByRefresh;

    opsBlockedByRefresh.countAllOperations.fetchAndAddRelaxed(1);

    switch (opType) {
        case LogicalOp::opInsert:
            opsBlockedByRefresh.countInserts.fetchAndAddRelaxed(1);
            break;
        case LogicalOp::opQuery:
            opsBlockedByRefresh.countQueries.fetchAndAddRelaxed(1);
            break;
        case LogicalOp::opUpdate:
            opsBlockedByRefresh.countUpdates.fetchAndAddRelaxed(1);
            break;
        case LogicalOp::opDelete:
            opsBlockedByRefresh.countDeletes.fetchAndAddRelaxed(1);
            break;
        case LogicalOp::opCommand:
            opsBlockedByRefresh.countCommands.fetchAndAddRelaxed(1);
            break;
        default:
            MONGO_UNREACHABLE;
    }
}

void CatalogCache::invalidateCollectionEntry_LINEARIZABLE(const NamespaceString& nss) {
    _collectionCache.invalidate(nss);
}

void CatalogCache::Stats::report(BSONObjBuilder* builder) const {
    builder->append("countStaleConfigErrors", countStaleConfigErrors.load());

    builder->append("totalRefreshWaitTimeMicros", totalRefreshWaitTimeMicros.load());

    if (isMongos()) {
        BSONObjBuilder operationsBlockedByRefreshBuilder(
            builder->subobjStart("operationsBlockedByRefresh"));

        operationsBlockedByRefreshBuilder.append(
            "countAllOperations", operationsBlockedByRefresh.countAllOperations.load());
        operationsBlockedByRefreshBuilder.append("countInserts",
                                                 operationsBlockedByRefresh.countInserts.load());
        operationsBlockedByRefreshBuilder.append("countQueries",
                                                 operationsBlockedByRefresh.countQueries.load());
        operationsBlockedByRefreshBuilder.append("countUpdates",
                                                 operationsBlockedByRefresh.countUpdates.load());
        operationsBlockedByRefreshBuilder.append("countDeletes",
                                                 operationsBlockedByRefresh.countDeletes.load());
        operationsBlockedByRefreshBuilder.append("countCommands",
                                                 operationsBlockedByRefresh.countCommands.load());

        operationsBlockedByRefreshBuilder.done();
    }
}

CatalogCache::DatabaseCache::DatabaseCache(ServiceContext* service,
                                           ThreadPoolInterface& threadPool,
                                           CatalogCacheLoader& catalogCacheLoader)
    : ReadThroughCache(_mutex,
                       service,
                       threadPool,
                       [this](OperationContext* opCtx,
                              const std::string& dbName,
                              const ValueHandle& db,
                              const ComparableDatabaseVersion& previousDbVersion) {
                           return _lookupDatabase(opCtx, dbName, previousDbVersion);
                       },
                       kDatabaseCacheSize),
      _catalogCacheLoader(catalogCacheLoader) {}

CatalogCache::DatabaseCache::LookupResult CatalogCache::DatabaseCache::_lookupDatabase(
    OperationContext* opCtx,
    const std::string& dbName,
    const ComparableDatabaseVersion& previousDbVersion) {
    // TODO (SERVER-34164): Track and increment stats for database refreshes

    LOGV2_FOR_CATALOG_REFRESH(24102, 2, "Refreshing cached database entry", "db"_attr = dbName);

    Timer t{};
    try {
        auto newDb = _catalogCacheLoader.getDatabase(dbName).get();
        uassertStatusOKWithContext(
            Grid::get(opCtx)->shardRegistry()->getShard(opCtx, newDb.getPrimary()),
            str::stream() << "The primary shard for database " << dbName << " does not exist");

        auto newDbVersion =
            ComparableDatabaseVersion::makeComparableDatabaseVersion(newDb.getVersion());
        LOGV2_FOR_CATALOG_REFRESH(24101,
                                  1,
                                  "Refreshed cached database entry",
                                  "db"_attr = dbName,
                                  "newDbVersion"_attr = newDbVersion.toBSONForLogging(),
                                  "oldDbVersion"_attr = previousDbVersion.toBSONForLogging(),
                                  "duration"_attr = Milliseconds(t.millis()));
        return CatalogCache::DatabaseCache::LookupResult(std::move(newDb), std::move(newDbVersion));
    } catch (const DBException& ex) {
        LOGV2_FOR_CATALOG_REFRESH(24100,
                                  1,
                                  "Error refreshing cached database entry",
                                  "db"_attr = dbName,
                                  "duration"_attr = Milliseconds(t.millis()),
                                  "error"_attr = redact(ex));
        if (ex.code() == ErrorCodes::NamespaceNotFound) {
            return CatalogCache::DatabaseCache::LookupResult(boost::none, previousDbVersion);
        }
        throw;
    }
}

CatalogCache::CollectionCache::CollectionCache(ServiceContext* service,
                                               ThreadPoolInterface& threadPool,
                                               CatalogCacheLoader& catalogCacheLoader)
    : ReadThroughCache(_mutex,
                       service,
                       threadPool,
                       [this](OperationContext* opCtx,
                              const NamespaceString& nss,
                              const ValueHandle& collectionHistory,
                              const ComparableChunkVersion& previousChunkVersion) {
                           return _lookupCollection(
                               opCtx, nss, collectionHistory, previousChunkVersion);
                       },
                       kCollectionCacheSize),
      _catalogCacheLoader(catalogCacheLoader) {}

void CatalogCache::CollectionCache::reportStats(BSONObjBuilder* builder) const {
    _stats.report(builder);
}

void CatalogCache::CollectionCache::_updateRefreshesStats(const bool isIncremental,
                                                          const bool add) {
    if (add) {
        if (isIncremental) {
            _stats.numActiveIncrementalRefreshes.addAndFetch(1);
            _stats.countIncrementalRefreshesStarted.addAndFetch(1);
        } else {
            _stats.numActiveFullRefreshes.addAndFetch(1);
            _stats.countFullRefreshesStarted.addAndFetch(1);
        }
    } else {
        if (isIncremental) {
            _stats.numActiveIncrementalRefreshes.subtractAndFetch(1);
        } else {
            _stats.numActiveFullRefreshes.subtractAndFetch(1);
        }
    }
}

void CatalogCache::CollectionCache::Stats::report(BSONObjBuilder* builder) const {
    builder->append("numActiveIncrementalRefreshes", numActiveIncrementalRefreshes.load());
    builder->append("countIncrementalRefreshesStarted", countIncrementalRefreshesStarted.load());

    builder->append("numActiveFullRefreshes", numActiveFullRefreshes.load());
    builder->append("countFullRefreshesStarted", countFullRefreshesStarted.load());

    builder->append("countFailedRefreshes", countFailedRefreshes.load());
}

CatalogCache::CollectionCache::LookupResult CatalogCache::CollectionCache::_lookupCollection(
    OperationContext* opCtx,
    const NamespaceString& nss,
    const RoutingTableHistoryValueHandle& existingHistory,
    const ComparableChunkVersion& previousVersion) {
    const bool isIncremental(existingHistory && existingHistory->optRt);
    _updateRefreshesStats(isIncremental, true);

    Timer t{};
    try {
        auto lookupVersion =
            isIncremental ? existingHistory->optRt->getVersion() : ChunkVersion::UNSHARDED();

        LOGV2_FOR_CATALOG_REFRESH(4619900,
                                  1,
                                  "Refreshing cached collection",
                                  "namespace"_attr = nss,
                                  "currentVersion"_attr = previousVersion.toBSONForLogging());

        auto collectionAndChunks = _catalogCacheLoader.getChunksSince(nss, lookupVersion).get();

        auto newRoutingHistory = [&] {
            // If we have routing info already and it's for the same collection epoch, we're
            // updating. Otherwise, we're making a whole new routing table.
            if (isIncremental &&
                existingHistory->optRt->getVersion().epoch() == collectionAndChunks.epoch) {
                return existingHistory->optRt->makeUpdated(collectionAndChunks.reshardingFields,
                                                           collectionAndChunks.changedChunks);
            }

            auto defaultCollator = [&]() -> std::unique_ptr<CollatorInterface> {
                if (!collectionAndChunks.defaultCollation.isEmpty()) {
                    // The collation should have been validated upon collection creation
                    return uassertStatusOK(
                        CollatorFactoryInterface::get(opCtx->getServiceContext())
                            ->makeFromBSON(collectionAndChunks.defaultCollation));
                }
                return nullptr;
            }();

            return RoutingTableHistory::makeNew(nss,
                                                collectionAndChunks.uuid,
                                                KeyPattern(collectionAndChunks.shardKeyPattern),
                                                std::move(defaultCollator),
                                                collectionAndChunks.shardKeyIsUnique,
                                                collectionAndChunks.epoch,
                                                std::move(collectionAndChunks.reshardingFields),
                                                collectionAndChunks.changedChunks);
        }();

        newRoutingHistory.setAllShardsRefreshed();

        // Check that the shards all match with what is on the config server
        std::set<ShardId> shardIds;
        newRoutingHistory.getAllShardIds(&shardIds);
        for (const auto& shardId : shardIds) {
            uassertStatusOKWithContext(Grid::get(opCtx)->shardRegistry()->getShard(opCtx, shardId),
                                       str::stream() << "Collection " << nss
                                                     << " references shard which does not exist");
        }

        const auto newVersion =
            ComparableChunkVersion::makeComparableChunkVersion(newRoutingHistory.getVersion());

        LOGV2_FOR_CATALOG_REFRESH(4619901,
                                  isIncremental || newVersion != previousVersion ? 0 : 1,
                                  "Refreshed cached collection",
                                  "namespace"_attr = nss,
                                  "newVersion"_attr = newVersion.toBSONForLogging(),
                                  "oldVersion"_attr = previousVersion.toBSONForLogging(),
                                  "duration"_attr = Milliseconds(t.millis()));
        _updateRefreshesStats(isIncremental, false);

        return LookupResult(OptionalRoutingTableHistory(std::move(newRoutingHistory)), newVersion);
    } catch (const DBException& ex) {
        _stats.countFailedRefreshes.addAndFetch(1);
        _updateRefreshesStats(isIncremental, false);

        if (ex.code() == ErrorCodes::NamespaceNotFound) {
            LOGV2_FOR_CATALOG_REFRESH(4619902,
                                      0,
                                      "Collection has found to be unsharded after refresh",
                                      "namespace"_attr = nss,
                                      "duration"_attr = Milliseconds(t.millis()));

            return LookupResult(
                OptionalRoutingTableHistory(),
                ComparableChunkVersion::makeComparableChunkVersion(ChunkVersion::UNSHARDED()));
        }

        LOGV2_FOR_CATALOG_REFRESH(4619903,
                                  0,
                                  "Error refreshing cached collection",
                                  "namespace"_attr = nss,
                                  "duration"_attr = Milliseconds(t.millis()),
                                  "error"_attr = redact(ex));

        throw;
    }
}

AtomicWord<uint64_t> ComparableDatabaseVersion::_uuidDisambiguatingSequenceNumSource{1ULL};

ComparableDatabaseVersion ComparableDatabaseVersion::makeComparableDatabaseVersion(
    const DatabaseVersion& version) {
    return ComparableDatabaseVersion(version, _uuidDisambiguatingSequenceNumSource.fetchAndAdd(1));
}

BSONObj ComparableDatabaseVersion::toBSONForLogging() const {
    BSONObjBuilder builder;
    if (_dbVersion)
        builder.append("dbVersion"_sd, _dbVersion->toBSON());
    else
        builder.append("dbVersion"_sd, "None");

    builder.append("uuidDisambiguatingSequenceNum"_sd,
                   static_cast<int64_t>(_uuidDisambiguatingSequenceNum));

    return builder.obj();
}


bool ComparableDatabaseVersion::operator==(const ComparableDatabaseVersion& other) const {
    if (!_dbVersion && !other._dbVersion)
        return true;  // Default constructed value
    if (_dbVersion.is_initialized() != other._dbVersion.is_initialized())
        return false;  // One side is default constructed value

    return sameUuid(other) && (_dbVersion->getLastMod() == other._dbVersion->getLastMod());
}

bool ComparableDatabaseVersion::operator<(const ComparableDatabaseVersion& other) const {
    if (!_dbVersion && !other._dbVersion)
        return false;  // Default constructed value

    if (_dbVersion && other._dbVersion && sameUuid(other)) {
        return _dbVersion->getLastMod() < other._dbVersion->getLastMod();
    } else {
        return _uuidDisambiguatingSequenceNum < other._uuidDisambiguatingSequenceNum;
    }
}

CachedDatabaseInfo::CachedDatabaseInfo(DatabaseType dbt) : _dbt(std::move(dbt)) {}

const ShardId& CachedDatabaseInfo::primaryId() const {
    return _dbt.getPrimary();
}

bool CachedDatabaseInfo::shardingEnabled() const {
    return _dbt.getSharded();
}

DatabaseVersion CachedDatabaseInfo::databaseVersion() const {
    return _dbt.getVersion();
}

}  // namespace mongo