summaryrefslogtreecommitdiff
path: root/src/mongo/db/s/database_sharding_state.cpp
blob: 38c0db9159a8fee96f376cb82709908e4861a8c1 (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
/**
 *    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.
 */

#include "mongo/db/s/database_sharding_state.h"

#include <fmt/format.h>

#include "mongo/db/catalog_shard_feature_flag_gen.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/s/operation_sharding_state.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/logv2/log.h"
#include "mongo/s/stale_exception.h"
#include "mongo/stdx/unordered_map.h"
#include "mongo/util/fail_point.h"

#define MONGO_LOGV2_DEFAULT_COMPONENT ::mongo::logv2::LogComponent::kSharding

namespace mongo {
namespace {

class DatabaseShardingStateMap {
    DatabaseShardingStateMap& operator=(const DatabaseShardingStateMap&) = delete;
    DatabaseShardingStateMap(const DatabaseShardingStateMap&) = delete;

public:
    static const ServiceContext::Decoration<DatabaseShardingStateMap> get;

    DatabaseShardingStateMap() {}

    struct DSSAndLock {
        DSSAndLock(const DatabaseName& dbName)
            : dssMutex("DSSMutex::" + dbName.db()),
              dss(std::make_unique<DatabaseShardingState>(dbName)) {}

        const Lock::ResourceMutex dssMutex;
        std::unique_ptr<DatabaseShardingState> dss;
    };

    DSSAndLock* getOrCreate(const DatabaseName& dbName) {
        stdx::lock_guard<Latch> lg(_mutex);

        auto it = _databases.find(dbName);
        if (it == _databases.end()) {
            auto inserted = _databases.try_emplace(dbName, std::make_unique<DSSAndLock>(dbName));
            invariant(inserted.second);
            it = std::move(inserted.first);
        }

        return it->second.get();
    }

    std::vector<DatabaseName> getDatabaseNames() {
        stdx::lock_guard lg(_mutex);
        std::vector<DatabaseName> result;
        result.reserve(_databases.size());
        for (const auto& [dbName, _] : _databases) {
            result.emplace_back(dbName);
        }
        return result;
    }

private:
    Mutex _mutex = MONGO_MAKE_LATCH("DatabaseShardingStateMap::_mutex");

    // Entries of the _databases map must never be deleted or replaced. This is to guarantee that a
    // 'dbName' is always associated to the same 'ResourceMutex'.
    using DatabasesMap = stdx::unordered_map<DatabaseName, std::unique_ptr<DSSAndLock>>;
    DatabasesMap _databases;
};

const ServiceContext::Decoration<DatabaseShardingStateMap> DatabaseShardingStateMap::get =
    ServiceContext::declareDecoration<DatabaseShardingStateMap>();

}  // namespace

DatabaseShardingState::DatabaseShardingState(const DatabaseName& dbName) : _dbName(dbName) {}

DatabaseShardingState::ScopedExclusiveDatabaseShardingState::ScopedExclusiveDatabaseShardingState(
    Lock::ResourceLock lock, DatabaseShardingState* dss)
    : _lock(std::move(lock)), _dss(dss) {}

DatabaseShardingState::ScopedSharedDatabaseShardingState::ScopedSharedDatabaseShardingState(
    Lock::ResourceLock lock, DatabaseShardingState* dss)
    : DatabaseShardingState::ScopedExclusiveDatabaseShardingState(std::move(lock), dss) {}

DatabaseShardingState::ScopedExclusiveDatabaseShardingState DatabaseShardingState::acquireExclusive(
    OperationContext* opCtx, const DatabaseName& dbName) {

    DatabaseShardingStateMap::DSSAndLock* dssAndLock =
        DatabaseShardingStateMap::get(opCtx->getServiceContext()).getOrCreate(dbName);

    // First lock the RESOURCE_MUTEX associated to this dbName to guarantee stability of the
    // DatabaseShardingState pointer. After that, it is safe to get and store the
    // DatabaseShadingState*, as long as the RESOURCE_MUTEX is kept locked.
    Lock::ResourceLock lock(opCtx->lockState(), dssAndLock->dssMutex.getRid(), MODE_X);

    return ScopedExclusiveDatabaseShardingState(std::move(lock), dssAndLock->dss.get());
}

DatabaseShardingState::ScopedSharedDatabaseShardingState DatabaseShardingState::acquireShared(
    OperationContext* opCtx, const DatabaseName& dbName) {

    DatabaseShardingStateMap::DSSAndLock* dssAndLock =
        DatabaseShardingStateMap::get(opCtx->getServiceContext()).getOrCreate(dbName);

    // First lock the RESOURCE_MUTEX associated to this dbName to guarantee stability of the
    // DatabaseShardingState pointer. After that, it is safe to get and store the
    // DatabaseShadingState*, as long as the RESOURCE_MUTEX is kept locked.
    Lock::ResourceLock lock(opCtx->lockState(), dssAndLock->dssMutex.getRid(), MODE_IS);

    return ScopedSharedDatabaseShardingState(std::move(lock), dssAndLock->dss.get());
}

DatabaseShardingState::ScopedExclusiveDatabaseShardingState
DatabaseShardingState::assertDbLockedAndAcquireExclusive(OperationContext* opCtx,
                                                         const DatabaseName& dbName) {
    dassert(opCtx->lockState()->isDbLockedForMode(dbName, MODE_IS));
    return acquireExclusive(opCtx, dbName);
}

DatabaseShardingState::ScopedSharedDatabaseShardingState
DatabaseShardingState::assertDbLockedAndAcquireShared(OperationContext* opCtx,
                                                      const DatabaseName& dbName) {
    dassert(opCtx->lockState()->isDbLockedForMode(dbName, MODE_IS));
    return acquireShared(opCtx, dbName);
}

std::vector<DatabaseName> DatabaseShardingState::getDatabaseNames(OperationContext* opCtx) {
    auto& databasesMap = DatabaseShardingStateMap::get(opCtx->getServiceContext());
    return databasesMap.getDatabaseNames();
}

void DatabaseShardingState::assertMatchingDbVersion(OperationContext* opCtx,
                                                    const DatabaseName& dbName) {
    const auto receivedVersion =
        OperationShardingState::get(opCtx).getDbVersion(DatabaseNameUtil::serialize(dbName));
    if (!receivedVersion) {
        return;
    }

    assertMatchingDbVersion(opCtx, dbName, *receivedVersion);
}

void DatabaseShardingState::assertMatchingDbVersion(OperationContext* opCtx,
                                                    const DatabaseName& dbName,
                                                    const DatabaseVersion& receivedVersion) {
    const auto scopedDss = acquireShared(opCtx, dbName);

    {
        const auto critSecSignal = scopedDss->getCriticalSectionSignal(
            opCtx->lockState()->isWriteLocked() ? ShardingMigrationCriticalSection::kWrite
                                                : ShardingMigrationCriticalSection::kRead);
        const auto optCritSecReason = scopedDss->getCriticalSectionReason();

        uassert(
            StaleDbRoutingVersion(
                DatabaseNameUtil::serialize(dbName), receivedVersion, boost::none, critSecSignal),
            str::stream() << "The critical section for the database "
                          << dbName.toStringForErrorMsg()
                          << " is acquired with reason: " << scopedDss->getCriticalSectionReason(),
            !critSecSignal);
    }

    const auto wantedVersion = scopedDss->getDbVersion(opCtx);
    uassert(
        StaleDbRoutingVersion(DatabaseNameUtil::serialize(dbName), receivedVersion, boost::none),
        str::stream() << "No cached info for the database " << dbName.toStringForErrorMsg(),
        wantedVersion);

    uassert(
        StaleDbRoutingVersion(DatabaseNameUtil::serialize(dbName), receivedVersion, *wantedVersion),
        str::stream() << "Version mismatch for the database " << dbName.toStringForErrorMsg(),
        receivedVersion == *wantedVersion);
}

void DatabaseShardingState::assertIsPrimaryShardForDb(OperationContext* opCtx,
                                                      const DatabaseName& dbName) {
    if (dbName == DatabaseName::kConfig || dbName == DatabaseName::kAdmin) {
        uassert(7393700,
                "The config server is the primary shard for database: {}"_format(
                    dbName.toStringForErrorMsg()),
                serverGlobalParams.clusterRole.has(ClusterRole::ConfigServer));
        return;
    }

    auto expectedDbVersion =
        OperationShardingState::get(opCtx).getDbVersion(DatabaseNameUtil::serialize(dbName));

    uassert(ErrorCodes::IllegalOperation,
            str::stream() << "Received request without the version for the database "
                          << dbName.toStringForErrorMsg(),
            expectedDbVersion);

    Lock::DBLock dbLock(opCtx, dbName, MODE_IS);
    assertMatchingDbVersion(opCtx, dbName, *expectedDbVersion);

    const auto scopedDss = assertDbLockedAndAcquireShared(opCtx, dbName);
    const auto primaryShardId = scopedDss->_dbInfo->getPrimary();
    const auto thisShardId = ShardingState::get(opCtx)->shardId();
    uassert(ErrorCodes::IllegalOperation,
            str::stream() << "This is not the primary shard for the database "
                          << dbName.toStringForErrorMsg() << ". Expected: " << primaryShardId
                          << " Actual: " << thisShardId,
            primaryShardId == thisShardId);
}

void DatabaseShardingState::setDbInfo(OperationContext* opCtx, const DatabaseType& dbInfo) {
    invariant(opCtx->lockState()->isDbLockedForMode(_dbName, MODE_IX));

    LOGV2(7286900,
          "Setting this node's cached database info",
          logAttrs(_dbName),
          "dbVersion"_attr = dbInfo.getVersion());
    _dbInfo.emplace(dbInfo);
}

void DatabaseShardingState::clearDbInfo(OperationContext* opCtx, bool cancelOngoingRefresh) {
    invariant(opCtx->lockState()->isDbLockedForMode(_dbName, MODE_IX));

    if (cancelOngoingRefresh) {
        _cancelDbMetadataRefresh();
    }

    LOGV2(7286901, "Clearing this node's cached database info", logAttrs(_dbName));
    _dbInfo = boost::none;
}

boost::optional<DatabaseVersion> DatabaseShardingState::getDbVersion(
    OperationContext* opCtx) const {
    return _dbInfo ? boost::optional<DatabaseVersion>(_dbInfo->getVersion()) : boost::none;
}

void DatabaseShardingState::enterCriticalSectionCatchUpPhase(OperationContext* opCtx,
                                                             const BSONObj& reason) {
    _critSec.enterCriticalSectionCatchUpPhase(reason);

    _cancelDbMetadataRefresh();
}

void DatabaseShardingState::enterCriticalSectionCommitPhase(OperationContext* opCtx,
                                                            const BSONObj& reason) {
    _critSec.enterCriticalSectionCommitPhase(reason);
}

void DatabaseShardingState::exitCriticalSection(OperationContext* opCtx, const BSONObj& reason) {
    _critSec.exitCriticalSection(reason);
}

void DatabaseShardingState::exitCriticalSectionNoChecks(OperationContext* opCtx) {
    _critSec.exitCriticalSectionNoChecks();
}

void DatabaseShardingState::setMovePrimaryInProgress(OperationContext* opCtx) {
    invariant(opCtx->lockState()->isDbLockedForMode(_dbName, MODE_X));
    _movePrimaryInProgress = true;
}

void DatabaseShardingState::unsetMovePrimaryInProgress(OperationContext* opCtx) {
    invariant(opCtx->lockState()->isDbLockedForMode(_dbName, MODE_IX));
    _movePrimaryInProgress = false;
}

void DatabaseShardingState::setDbMetadataRefreshFuture(SharedSemiFuture<void> future,
                                                       CancellationSource cancellationSource) {
    invariant(!_dbMetadataRefresh);
    _dbMetadataRefresh.emplace(std::move(future), std::move(cancellationSource));
}

boost::optional<SharedSemiFuture<void>> DatabaseShardingState::getDbMetadataRefreshFuture() const {
    return _dbMetadataRefresh ? boost::optional<SharedSemiFuture<void>>(_dbMetadataRefresh->future)
                              : boost::none;
}

void DatabaseShardingState::resetDbMetadataRefreshFuture() {
    _dbMetadataRefresh = boost::none;
}

void DatabaseShardingState::_cancelDbMetadataRefresh() {
    if (_dbMetadataRefresh) {
        _dbMetadataRefresh->cancellationSource.cancel();
    }
}

boost::optional<bool> DatabaseShardingState::_isPrimaryShardForDb(OperationContext* opCtx) const {
    return _dbInfo
        ? boost::optional<bool>(_dbInfo->getPrimary() == ShardingState::get(opCtx)->shardId())
        : boost::none;
}

}  // namespace mongo