summaryrefslogtreecommitdiff
path: root/src/mongo/db/s/resharding/resharding_oplog_batch_preparer.cpp
blob: ca382ace57c5e64d5b51b61f61435734bc09263f (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
/**
 *    Copyright (C) 2021-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/platform/basic.h"

#include "mongo/db/s/resharding/resharding_oplog_batch_preparer.h"

#include <third_party/murmurhash3/MurmurHash3.h>

#include "mongo/bson/bsonelement_comparator.h"
#include "mongo/db/ops/write_ops_retryability.h"
#include "mongo/db/query/collation/collator_interface.h"
#include "mongo/db/repl/apply_ops.h"
#include "mongo/db/s/resharding/resharding_server_parameters_gen.h"
#include "mongo/db/session/logical_session_id.h"
#include "mongo/logv2/redaction.h"
#include "mongo/util/assert_util.h"
#include "mongo/util/str.h"

namespace mongo {

namespace {

/**
 * Return true if we need to update config.transactions collection for this oplog entry.
 */
bool shouldUpdateTxnTable(const repl::OplogEntry& op) {
    if (op.getCommandType() == repl::OplogEntry::CommandType::kAbortTransaction) {
        return true;
    }

    if (!op.getSessionId()) {
        return false;
    }

    if (!op.getTxnNumber()) {
        return false;
    }

    if (op.getCommandType() == repl::OplogEntry::CommandType::kApplyOps) {
        // This applyOps oplog entry is guaranteed to correspond to a committed transaction since
        // the resharding aggregation pipeline does not output applyOps oplog entries for aborted
        // transactions (i.e. it only outputs the abortTransaction oplog entry).

        if (isInternalSessionForRetryableWrite(*op.getSessionId())) {
            // For a retryable internal transaction, we need to update the config.transactions
            // collection upon writing the noop oplog entries for retryable operations contained
            // within each applyOps oplog entry.
            return true;
        }

        // The resharding aggregation pipeline also does not output the commitTransaction oplog
        // entry so for a non-retryable transaction, we need to the update to the
        // config.transactions collection upon seeing the final applyOps oplog entry.
        return !op.isPartialTransaction();
    }

    return false;
}

}  // anonymous namespace

using WriterVectors = ReshardingOplogBatchPreparer::WriterVectors;

ReshardingOplogBatchPreparer::ReshardingOplogBatchPreparer(
    std::unique_ptr<CollatorInterface> defaultCollator)
    : _defaultCollator(std::move(defaultCollator)) {}

void ReshardingOplogBatchPreparer::throwIfUnsupportedCommandOp(const OplogEntry& op) {
    invariant(op.isCommand());
    switch (op.getCommandType()) {
        case OplogEntry::CommandType::kApplyOps:
        case OplogEntry::CommandType::kCommitTransaction:
        case OplogEntry::CommandType::kAbortTransaction:
            return;

        case OplogEntry::CommandType::kDrop:
            uasserted(ErrorCodes::OplogOperationUnsupported,
                      str::stream() << "Received drop command for resharding's source collection: "
                                    << redact(op.toBSONForLogging()));

        default:
            uasserted(ErrorCodes::OplogOperationUnsupported,
                      str::stream() << "Command not supported during resharding: "
                                    << redact(op.toBSONForLogging()));
    }
}

WriterVectors ReshardingOplogBatchPreparer::makeCrudOpWriterVectors(
    const OplogBatchToPrepare& batch, std::list<OplogEntry>& derivedOps) const {
    invariant(derivedOps.empty());

    auto writerVectors = _makeEmptyWriterVectors();

    for (const auto& op : batch) {
        if (op.isCrudOpType()) {
            _appendCrudOpToWriterVector(&op, writerVectors);
        } else if (op.isCommand()) {
            throwIfUnsupportedCommandOp(op);

            if (op.getCommandType() != OplogEntry::CommandType::kApplyOps) {
                continue;
            }

            auto applyOpsInfo = repl::ApplyOpsCommandInfo::parse(op.getObject());
            uassert(
                ErrorCodes::OplogOperationUnsupported,
                str::stream() << "Commands within applyOps are not supported during resharding: "
                              << redact(op.toBSONForLogging()),
                applyOpsInfo.areOpsCrudOnly());

            auto unrolledOp =
                uassertStatusOK(repl::MutableOplogEntry::parse(op.getEntry().toBSON()));

            for (const auto& innerOp : applyOpsInfo.getOperations()) {
                unrolledOp.setDurableReplOperation(repl::DurableReplOperation::parse(
                    IDLParserContext{
                        "ReshardingOplogBatchPreparer::makeCrudOpWriterVectors innerOp"},
                    innerOp));

                if (isWouldChangeOwningShardSentinelOplogEntry(unrolledOp)) {
                    continue;
                }

                // There isn't a direct way to convert from a MutableOplogEntry to a
                // DurableOplogEntry or OplogEntry. We serialize the unrolledOp to have it get
                // re-parsed into an OplogEntry.
                auto& derivedOp = derivedOps.emplace_back(unrolledOp.toBSON());
                invariant(derivedOp.isCrudOpType());

                // `&derivedOp` is guaranteed to remain stable while we append more derived oplog
                // entries because `derivedOps` is a std::list.
                _appendCrudOpToWriterVector(&derivedOp, writerVectors);
            }
        } else {
            invariant(repl::OpTypeEnum::kNoop == op.getOpType());
        }
    }

    return writerVectors;
}

WriterVectors ReshardingOplogBatchPreparer::makeSessionOpWriterVectors(
    const OplogBatchToPrepare& batch, std::list<OplogEntry>& derivedOps) const {
    auto writerVectors = _makeEmptyWriterVectors();

    struct SessionOpsList {
        TxnNumber txnNum = kUninitializedTxnNumber;
        std::vector<const OplogEntry*> ops;
    };

    LogicalSessionIdMap<SessionOpsList> sessionTracker;

    auto updateSessionTracker = [&](const OplogEntry* op) {
        if (const auto& lsid = op->getSessionId()) {
            uassert(4990700,
                    str::stream() << "Missing txnNumber for oplog entry with lsid: "
                                  << redact(op->toBSONForLogging()),
                    op->getTxnNumber());

            auto txnNumber = *op->getTxnNumber();

            auto& retryableOpList = sessionTracker[*lsid];
            if (txnNumber == retryableOpList.txnNum) {
                retryableOpList.ops.emplace_back(op);
            } else if (txnNumber > retryableOpList.txnNum) {
                retryableOpList.ops = {op};
                retryableOpList.txnNum = txnNumber;
            } else {
                uasserted(4990401,
                          str::stream() << "Encountered out of order txnNumbers; batch had "
                                        << redact(op->toBSONForLogging()) << " after "
                                        << redact(retryableOpList.ops.back()->toBSONForLogging()));
            }
        }
    };

    for (auto& op : batch) {
        if (op.isCrudOpType()) {
            updateSessionTracker(&op);
        } else if (op.isCommand()) {
            throwIfUnsupportedCommandOp(op);

            if (!shouldUpdateTxnTable(op)) {
                continue;
            }

            auto sessionId = *op.getSessionId();

            if (isInternalSessionForRetryableWrite(sessionId) &&
                op.getCommandType() == OplogEntry::CommandType::kApplyOps) {
                // Derive retryable write CRUD oplog entries from this retryable internal
                // transaction applyOps oplog entry.

                auto applyOpsInfo = repl::ApplyOpsCommandInfo::parse(op.getObject());
                uassert(ErrorCodes::OplogOperationUnsupported,
                        str::stream()
                            << "Commands within applyOps are not supported during resharding: "
                            << redact(op.toBSONForLogging()),
                        applyOpsInfo.areOpsCrudOnly());

                auto unrolledOp =
                    uassertStatusOK(repl::MutableOplogEntry::parse(op.getEntry().toBSON()));
                unrolledOp.setSessionId(*getParentSessionId(sessionId));
                unrolledOp.setTxnNumber(*sessionId.getTxnNumber());

                for (const auto& innerOp : applyOpsInfo.getOperations()) {
                    auto replOp = repl::ReplOperation::parse(
                        IDLParserContext{
                            "ReshardingOplogBatchPreparer::makeSessionOpWriterVectors innerOp"},
                        innerOp);
                    if (replOp.getStatementIds().empty()) {
                        // Skip this operation since it is not retryable.
                        continue;
                    }
                    unrolledOp.setDurableReplOperation(replOp);

                    // There isn't a direct way to convert from a MutableOplogEntry to a
                    // DurableOplogEntry or OplogEntry. We serialize the unrolledOp to have it get
                    // re-parsed into an OplogEntry.
                    auto& derivedOp = derivedOps.emplace_back(unrolledOp.toBSON());
                    invariant(derivedOp.isCrudOpType() ||
                              isWouldChangeOwningShardSentinelOplogEntry(unrolledOp));

                    // `&derivedOp` is guaranteed to remain stable while we append more derived
                    // oplog entries because `derivedOps` is a std::list.
                    updateSessionTracker(&derivedOp);
                }
            } else {
                updateSessionTracker(&op);
            }
        } else {
            invariant(repl::OpTypeEnum::kNoop == op.getOpType());
        }
    }

    for (auto& [lsid, opList] : sessionTracker) {
        for (auto& op : opList.ops) {
            _appendSessionOpToWriterVector(lsid, op, writerVectors);
        }
    }

    return writerVectors;
}

WriterVectors ReshardingOplogBatchPreparer::_makeEmptyWriterVectors() const {
    return WriterVectors(size_t(resharding::gReshardingOplogBatchTaskCount.load()));
}

void ReshardingOplogBatchPreparer::_appendCrudOpToWriterVector(const OplogEntry* op,
                                                               WriterVectors& writerVectors) const {
    BSONElementComparator elementHasher{BSONElementComparator::FieldNamesMode::kIgnore,
                                        _defaultCollator.get()};

    const size_t idHash = elementHasher.hash(op->getIdElement());

    uint32_t hash = 0;
    MurmurHash3_x86_32(&idHash, sizeof(idHash), hash, &hash);

    _appendOpToWriterVector(hash, op, writerVectors);
}

void ReshardingOplogBatchPreparer::_appendSessionOpToWriterVector(
    const LogicalSessionId& lsid, const OplogEntry* op, WriterVectors& writerVectors) const {
    LogicalSessionIdHash lsidHasher;
    _appendOpToWriterVector(lsidHasher(lsid), op, writerVectors);
}

void ReshardingOplogBatchPreparer::_appendOpToWriterVector(std::uint32_t hash,
                                                           const OplogEntry* op,
                                                           WriterVectors& writerVectors) const {
    auto& writer = writerVectors[hash % writerVectors.size()];
    if (writer.empty()) {
        // Skip a few growth rounds in anticipation that we'll be appending more.
        writer.reserve(8U);
    }
    writer.emplace_back(op);
}

}  // namespace mongo