summaryrefslogtreecommitdiff
path: root/src/mongo/db/index/index_build_interceptor.h
blob: d4081c3be924e18059381fbc33c8ceec26043b83 (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
/**
 *    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.
 */

#pragma once

#include <memory>

#include "mongo/db/index/duplicate_key_tracker.h"
#include "mongo/db/index/index_access_method.h"
#include "mongo/db/index/multikey_paths.h"
#include "mongo/db/index/skipped_record_tracker.h"
#include "mongo/db/namespace_string.h"
#include "mongo/db/storage/temporary_record_store.h"
#include "mongo/platform/atomic_word.h"

namespace mongo {

class BSONObj;
class IndexAccessMethod;
class OperationContext;

class IndexBuildInterceptor {
public:
    /**
     * Determines if we will yield locks while draining the side tables.
     */
    enum class DrainYieldPolicy { kNoYield, kYield };

    enum class Op { kInsert, kDelete };

    /**
     * Indicates whether this interceptor will allow tracking of documents skipped due to key
     * generation errors. When 'kTrack', a SkippedRecordTracker is created.
     */
    enum class TrackSkippedRecords { kNoTrack, kTrack };

    static bool typeCanFastpathMultikeyUpdates(IndexType type);

    /**
     * Creates a temporary table for writes during an index build. Additionally creates a temporary
     * table to store any duplicate key constraint violations found during the build, if the index
     * being built has uniqueness constraints.
     *
     * deleteTemporaryTable() must be called before destruction to delete the temporary tables.
     */
    IndexBuildInterceptor(OperationContext* opCtx,
                          IndexCatalogEntry* entry,
                          TrackSkippedRecords trackSkippedRecords);

    /**
     * Deletes the temporary side writes and duplicate key constraint violations tables. Must be
     * called before object destruction.
     */
    void deleteTemporaryTables(OperationContext* opCtx);

    /**
     * Client writes that are concurrent with an index build will have their index updates written
     * to a temporary table. After the index table scan is complete, these updates will be applied
     * to the underlying index table.
     *
     * On success, `numKeysOut` if non-null will contain the number of keys added or removed.
     */
    Status sideWrite(OperationContext* opCtx,
                     const std::vector<KeyString::Value>& keys,
                     const KeyStringSet& multikeyMetadataKeys,
                     const MultikeyPaths& multikeyPaths,
                     RecordId loc,
                     Op op,
                     int64_t* const numKeysOut);

    /**
     * Given a set of duplicate keys, record the keys for later verification by a call to
     * checkDuplicateKeyConstraints();
     */
    Status recordDuplicateKeys(OperationContext* opCtx, const std::vector<BSONObj>& keys);

    /**
     * Returns Status::OK if all previously recorded duplicate key constraint violations have been
     * resolved for the index. Returns a DuplicateKey error if there are still duplicate key
     * constraint violations on the index.
     */
    Status checkDuplicateKeyConstraints(OperationContext* opCtx) const;


    /**
     * Performs a resumable scan on the side writes table, and either inserts or removes each key
     * from the underlying IndexAccessMethod. This will only insert as many records as are visible
     * in the current snapshot.
     *
     * This is resumable, so subsequent calls will start the scan at the record immediately
     * following the last inserted record from a previous call to drainWritesIntoIndex.
     *
     * TODO (SERVER-40894): Implement draining while reading at a timestamp. The following comment
     * does not apply.
     * When 'readSource' is not kUnset, perform the drain by reading at the timestamp described by
     * the ReadSource. This will always reset the ReadSource to its original value before returning.
     * The drain otherwise reads at the pre-existing ReadSource on the RecoveryUnit. This may be
     * necessary by callers that can only guarantee consistency of data up to a certain point in
     * time.
     */
    Status drainWritesIntoIndex(OperationContext* opCtx,
                                const InsertDeleteOptions& options,
                                RecoveryUnit::ReadSource readSource,
                                DrainYieldPolicy drainYieldPolicy);

    SkippedRecordTracker* getSkippedRecordTracker() {
        return _skippedRecordTracker.get();
    }

    const SkippedRecordTracker* getSkippedRecordTracker() const {
        return _skippedRecordTracker.get();
    }

    /**
     * Tries to index previously skipped records. For each record, if the new indexing attempt is
     * successful, keys are written directly to the index. Unsuccessful key generation or writes
     * will return errors.
     */
    Status retrySkippedRecords(OperationContext* opCtx, const Collection* collection);

    /**
     * Returns 'true' if there are no visible records remaining to be applied from the side writes
     * table. Ensure that this returns 'true' when an index build is completed.
     */
    bool areAllWritesApplied(OperationContext* opCtx) const;

    /**
     * Returns true if all recorded duplicate key constraint violations have been checked.
     */
    bool areAllConstraintsChecked(OperationContext* opCtx) const;


    /**
     * When an index builder wants to commit, use this to retrieve any recorded multikey paths
     * that were tracked during the build.
     */
    boost::optional<MultikeyPaths> getMultikeyPaths() const;

    const std::string& getSideWritesTableIdent() const;

    const std::string& getConstraintViolationsTableIdent() const;

private:
    using SideWriteRecord = std::pair<RecordId, BSONObj>;

    Status _applyWrite(OperationContext* opCtx,
                       const BSONObj& doc,
                       const InsertDeleteOptions& options,
                       int64_t* const keysInserted,
                       int64_t* const keysDeleted);

    /**
     * Yield lock manager locks and abandon the current storage engine snapshot.
     */
    void _yield(OperationContext* opCtx);

    // The entry for the index that is being built.
    IndexCatalogEntry* _indexCatalogEntry;

    // This temporary record store records intercepted keys that will be written into the index by
    // calling drainWritesIntoIndex(). It is owned by the interceptor and dropped along with it.
    std::unique_ptr<TemporaryRecordStore> _sideWritesTable;

    // Records RecordIds that have been skipped due to indexing errors.
    std::unique_ptr<SkippedRecordTracker> _skippedRecordTracker;

    std::unique_ptr<DuplicateKeyTracker> _duplicateKeyTracker;

    int64_t _numApplied{0};

    // This allows the counter to be used in a RecoveryUnit rollback handler where the
    // IndexBuildInterceptor is no longer available (e.g. due to index build cleanup). If there are
    // additional fields that have to be referenced in commit/rollback handlers, this counter should
    // be moved to a new IndexBuildsInterceptor::InternalState structure that will be managed as a
    // shared resource.
    std::shared_ptr<AtomicWord<long long>> _sideWritesCounter;

    mutable Mutex _multikeyPathMutex =
        MONGO_MAKE_LATCH("IndexBuildInterceptor::_multikeyPathMutex");
    boost::optional<MultikeyPaths> _multikeyPaths;
};

}  // namespace mongo