summaryrefslogtreecommitdiff
path: root/src/mongo/db/repair.cpp
blob: 7332440372b963538fd7d6850c856a86b78ef047 (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
/**
 *    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 <algorithm>
#include <fmt/format.h>

#include "mongo/db/repair.h"

#include "mongo/base/status.h"
#include "mongo/base/string_data.h"
#include "mongo/bson/bson_validate.h"
#include "mongo/bson/bsonobjbuilder.h"
#include "mongo/db/catalog/collection.h"
#include "mongo/db/catalog/collection_catalog.h"
#include "mongo/db/catalog/collection_validation.h"
#include "mongo/db/catalog/database.h"
#include "mongo/db/catalog/database_holder.h"
#include "mongo/db/catalog/document_validation.h"
#include "mongo/db/catalog/index_key_validate.h"
#include "mongo/db/catalog/multi_index_block.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/db_raii.h"
#include "mongo/db/index/index_descriptor.h"
#include "mongo/db/index_builds_coordinator.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/query/query_knobs_gen.h"
#include "mongo/db/rebuild_indexes.h"
#include "mongo/db/repl_set_member_in_standalone_mode.h"
#include "mongo/db/storage/storage_engine.h"
#include "mongo/db/storage/storage_repair_observer.h"
#include "mongo/db/storage/storage_util.h"
#include "mongo/db/tenant_namespace.h"
#include "mongo/db/vector_clock.h"
#include "mongo/logv2/log.h"
#include "mongo/util/scopeguard.h"

namespace mongo {

using namespace fmt::literals;

Status rebuildIndexesForNamespace(OperationContext* opCtx,
                                  const NamespaceString& nss,
                                  StorageEngine* engine) {
    opCtx->checkForInterrupt();
    auto collection = CollectionCatalog::get(opCtx)->lookupCollectionByNamespace(opCtx, nss);
    auto swIndexNameObjs = getIndexNameObjs(collection);
    if (!swIndexNameObjs.isOK())
        return swIndexNameObjs.getStatus();

    std::vector<BSONObj> indexSpecs = swIndexNameObjs.getValue().second;
    Status status = rebuildIndexesOnCollection(opCtx, collection, indexSpecs, RepairData::kYes);
    if (!status.isOK())
        return status;

    engine->flushAllFiles(opCtx, /*callerHoldsReadLock*/ false);
    return Status::OK();
}

namespace {
Status dropUnfinishedIndexes(OperationContext* opCtx, Collection* collection) {
    std::vector<std::string> indexNames;
    collection->getAllIndexes(&indexNames);
    for (const auto& indexName : indexNames) {
        if (!collection->isIndexReady(indexName)) {
            LOGV2(3871400,
                  "Dropping unfinished index '{name}' after collection was modified by "
                  "repair",
                  "Dropping unfinished index after collection was modified by repair",
                  "index"_attr = indexName);

            WriteUnitOfWork wuow(opCtx);
            // There are no concurrent users of the index while --repair is running, so it is OK to
            // pass in a nullptr for the index 'ident', promising that the index is not in use.
            catalog::removeIndex(opCtx, indexName, collection, nullptr /*ident */);
            wuow.commit();

            StorageRepairObserver::get(opCtx->getServiceContext())
                ->invalidatingModification(str::stream()
                                           << "Dropped unfinished index '" << indexName << "' on "
                                           << collection->ns());
        }
    }
    return Status::OK();
}

Status repairCollections(OperationContext* opCtx,
                         StorageEngine* engine,
                         const std::string& dbName) {
    auto colls = CollectionCatalog::get(opCtx)->getAllCollectionNamesFromDb(opCtx, dbName);

    for (const auto& nss : colls) {
        auto status = repair::repairCollection(opCtx, engine, nss);
        if (!status.isOK()) {
            return status;
        }
    }
    return Status::OK();
}
}  // namespace

namespace repair {
Status repairDatabase(OperationContext* opCtx,
                      StorageEngine* engine,
                      const TenantDatabaseName& tenantDbName) {
    DisableDocumentValidation validationDisabler(opCtx);

    // We must hold some form of lock here
    invariant(opCtx->lockState()->isW());
    invariant(tenantDbName.dbName().find('.') == std::string::npos);

    LOGV2(21029, "repairDatabase", "db"_attr = tenantDbName);


    opCtx->checkForInterrupt();

    // Close the db and invalidate all current users and caches.
    auto databaseHolder = DatabaseHolder::get(opCtx);
    databaseHolder->close(opCtx, tenantDbName);

    // Reopening db is necessary for repairCollections.
    databaseHolder->openDb(opCtx, tenantDbName);

    auto status = repairCollections(opCtx, engine, tenantDbName.dbName());
    if (!status.isOK()) {
        LOGV2_FATAL_CONTINUE(21030,
                             "Failed to repair database {dbName}: {status_reason}",
                             "Failed to repair database",
                             "db"_attr = tenantDbName,
                             "error"_attr = status);
    }

    try {
        // Ensure that we don't trigger an exception when attempting to take locks.
        UninterruptibleLockGuard noInterrupt(opCtx->lockState());

        // Restore oplog Collection pointer cache.
        repl::acquireOplogCollectionForLogging(opCtx);
    } catch (...) {
        LOGV2_FATAL_CONTINUE(
            21031,
            "Unexpected exception encountered while reacquiring oplog collection after repair.");
        std::terminate();  // Logs additional info about the specific error.
    }

    return status;
}

Status repairCollection(OperationContext* opCtx,
                        StorageEngine* engine,
                        const NamespaceString& nss) {
    opCtx->checkForInterrupt();

    LOGV2(21027, "Repairing collection", "namespace"_attr = nss);

    Status status = Status::OK();
    {
        auto collection = CollectionCatalog::get(opCtx)->lookupCollectionByNamespace(opCtx, nss);
        status = engine->repairRecordStore(opCtx, collection->getCatalogId(), nss);
    }


    // Need to lookup from catalog again because the old collection object was invalidated by
    // repairRecordStore.
    auto collection = CollectionCatalog::get(opCtx)->lookupCollectionByNamespaceForMetadataWrite(
        opCtx, CollectionCatalog::LifetimeMode::kInplace, nss);

    // If data was modified during repairRecordStore, we know to rebuild indexes without needing
    // to run an expensive collection validation.
    if (status.code() == ErrorCodes::DataModifiedByRepair) {
        invariant(StorageRepairObserver::get(opCtx->getServiceContext())->isDataInvalidated(),
                  "Collection '{}' ({})"_format(collection->ns().toString(),
                                                collection->uuid().toString()));

        // If we are a replica set member in standalone mode and we have unfinished indexes,
        // drop them before rebuilding any completed indexes. Since we have already made
        // invalidating modifications to our data, it is safe to just drop the indexes entirely
        // to avoid the risk of the index rebuild failing.
        if (getReplSetMemberInStandaloneMode(opCtx->getServiceContext())) {
            if (auto status = dropUnfinishedIndexes(opCtx, collection); !status.isOK()) {
                return status;
            }
        }

        return rebuildIndexesForNamespace(opCtx, nss, engine);
    } else if (!status.isOK()) {
        return status;
    }

    // Run collection validation to avoid unecessarily rebuilding indexes on valid collections
    // with consistent indexes. Initialize the collection prior to validation.
    collection->init(opCtx);

    ValidateResults validateResults;
    BSONObjBuilder output;

    // Exclude full record store validation because we have already validated the underlying
    // record store in the call to repairRecordStore above.
    status =
        CollectionValidation::validate(opCtx,
                                       nss,
                                       CollectionValidation::ValidateMode::kForegroundFullIndexOnly,
                                       CollectionValidation::RepairMode::kFixErrors,
                                       &validateResults,
                                       &output);
    if (!status.isOK()) {
        return status;
    }

    BSONObjBuilder detailedResults;
    const bool debug = false;
    validateResults.appendToResultObj(&detailedResults, debug);

    LOGV2(21028,
          "Collection validation",
          "results"_attr = output.done(),
          "detailedResults"_attr = detailedResults.done());

    if (validateResults.repaired) {
        if (validateResults.valid) {
            LOGV2(4934000, "Validate successfully repaired all data", "collection"_attr = nss);
        } else {
            LOGV2(4934001, "Validate was unable to repair all data", "collection"_attr = nss);
        }
    } else {
        LOGV2(4934002, "Validate did not make any repairs", "collection"_attr = nss);
    }

    // If not valid, whether repair ran or not, indexes will need to be rebuilt.
    if (!validateResults.valid) {
        return rebuildIndexesForNamespace(opCtx, nss, engine);
    }
    return Status::OK();
}
}  // namespace repair

}  // namespace mongo