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
|
/**
* 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_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kStorage
#include "mongo/platform/basic.h"
#include "mongo/db/catalog/index_builds_manager.h"
#include "mongo/db/catalog/collection.h"
#include "mongo/db/catalog/collection_catalog.h"
#include "mongo/db/catalog/index_catalog.h"
#include "mongo/db/catalog/index_timestamp_helper.h"
#include "mongo/db/concurrency/write_conflict_exception.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/storage/write_unit_of_work.h"
#include "mongo/util/assert_util.h"
#include "mongo/util/log.h"
#include "mongo/util/str.h"
namespace mongo {
namespace {
/**
* Returns basic info on index builders.
*/
std::string toSummary(const std::map<UUID, std::shared_ptr<MultiIndexBlock>>& builders) {
str::stream ss;
ss << "Number of builders: " << builders.size() << ": [";
bool first = true;
for (const auto& pair : builders) {
if (!first) {
ss << ", ";
}
ss << pair.first;
first = false;
}
ss << "]";
return ss;
}
} // namespace
IndexBuildsManager::SetupOptions::SetupOptions() = default;
IndexBuildsManager::~IndexBuildsManager() {
invariant(_builders.empty(),
str::stream() << "Index builds still active: " << toSummary(_builders));
}
Status IndexBuildsManager::setUpIndexBuild(OperationContext* opCtx,
Collection* collection,
const std::vector<BSONObj>& specs,
const UUID& buildUUID,
OnInitFn onInit,
SetupOptions options) {
_registerIndexBuild(buildUUID);
const auto& nss = collection->ns();
invariant(opCtx->lockState()->isCollectionLockedForMode(nss, MODE_X),
str::stream() << "Unable to set up index build " << buildUUID << ": collection "
<< nss.ns() << " is not locked in exclusive mode.");
auto builder = _getBuilder(buildUUID);
if (options.protocol == IndexBuildProtocol::kTwoPhase) {
builder->setTwoPhaseBuildUUID(buildUUID);
}
// Ignore uniqueness constraint violations when relaxed (on secondaries). Secondaries can
// complete index builds in the middle of batches, which creates the potential for finding
// duplicate key violations where there otherwise would be none at consistent states.
if (options.indexConstraints == IndexConstraints::kRelax) {
builder->ignoreUniqueConstraint();
}
std::vector<BSONObj> indexes;
try {
indexes = writeConflictRetry(opCtx, "IndexBuildsManager::setUpIndexBuild", nss.ns(), [&]() {
return uassertStatusOK(builder->init(opCtx, collection, specs, onInit));
});
} catch (const DBException& ex) {
return ex.toStatus();
}
log() << "Index build initialized: " << buildUUID << ": " << nss << " (" << collection->uuid()
<< " ): indexes: " << indexes.size();
return Status::OK();
}
Status IndexBuildsManager::startBuildingIndex(OperationContext* opCtx,
Collection* collection,
const UUID& buildUUID) {
auto builder = _getBuilder(buildUUID);
return builder->insertAllDocumentsInCollection(opCtx, collection);
}
StatusWith<std::pair<long long, long long>> IndexBuildsManager::startBuildingIndexForRecovery(
OperationContext* opCtx, NamespaceString ns, const UUID& buildUUID, RepairData repair) {
auto builder = _getBuilder(buildUUID);
auto coll = CollectionCatalog::get(opCtx).lookupCollectionByNamespace(opCtx, ns);
auto rs = coll ? coll->getRecordStore() : nullptr;
// Iterate all records in the collection. Validate the records and index them
// if they are valid. Delete them (if in repair mode), or crash, if they are not valid.
long long numRecords = 0;
long long dataSize = 0;
auto cursor = rs->getCursor(opCtx);
auto record = cursor->next();
while (record) {
opCtx->checkForInterrupt();
// Cursor is left one past the end of the batch inside writeConflictRetry
auto beginBatchId = record->id;
Status status = writeConflictRetry(opCtx, "repairDatabase", ns.ns(), [&] {
// In the case of WCE in a partial batch, we need to go back to the beginning
if (!record || (beginBatchId != record->id)) {
record = cursor->seekExact(beginBatchId);
}
WriteUnitOfWork wunit(opCtx);
for (int i = 0; record && i < internalInsertMaxBatchSize.load(); i++) {
RecordId id = record->id;
RecordData& data = record->data;
// Use the latest BSON validation version. We retain decimal data when repairing
// database even if decimal is disabled.
auto validStatus = validateBSON(data.data(), data.size(), BSONVersion::kLatest);
if (!validStatus.isOK()) {
if (repair == RepairData::kNo) {
severe() << "Invalid BSON detected at " << id << ": "
<< redact(validStatus);
fassertFailed(31396);
}
warning() << "Invalid BSON detected at " << id << ": " << redact(validStatus)
<< ". Deleting.";
rs->deleteRecord(opCtx, id);
} else {
numRecords++;
dataSize += data.size();
auto insertStatus = builder->insert(opCtx, data.releaseToBson(), id);
if (!insertStatus.isOK()) {
return insertStatus;
}
}
record = cursor->next();
}
// Time to yield; make a safe copy of the current record before releasing our cursor.
if (record)
record->data.makeOwned();
cursor->save(); // Can't fail per API definition
// When this exits via success or WCE, we need to restore the cursor
ON_BLOCK_EXIT([opCtx, ns, &cursor]() {
// restore CAN throw WCE per API
writeConflictRetry(
opCtx, "retryRestoreCursor", ns.ns(), [&cursor] { cursor->restore(); });
});
wunit.commit();
return Status::OK();
});
if (!status.isOK()) {
return status;
}
}
Status status = builder->dumpInsertsFromBulk(opCtx);
if (!status.isOK()) {
return status;
}
return std::make_pair(numRecords, dataSize);
}
Status IndexBuildsManager::drainBackgroundWrites(
OperationContext* opCtx,
const UUID& buildUUID,
RecoveryUnit::ReadSource readSource,
IndexBuildInterceptor::DrainYieldPolicy drainYieldPolicy) {
auto builder = _getBuilder(buildUUID);
return builder->drainBackgroundWrites(opCtx, readSource, drainYieldPolicy);
}
Status IndexBuildsManager::retrySkippedRecords(OperationContext* opCtx,
const UUID& buildUUID,
Collection* collection) {
auto builder = _getBuilder(buildUUID);
return builder->retrySkippedRecords(opCtx, collection);
}
Status IndexBuildsManager::checkIndexConstraintViolations(OperationContext* opCtx,
const UUID& buildUUID) {
auto builder = _getBuilder(buildUUID);
return builder->checkConstraints(opCtx);
}
Status IndexBuildsManager::commitIndexBuild(OperationContext* opCtx,
Collection* collection,
const NamespaceString& nss,
const UUID& buildUUID,
MultiIndexBlock::OnCreateEachFn onCreateEachFn,
MultiIndexBlock::OnCommitFn onCommitFn) {
auto builder = _getBuilder(buildUUID);
return writeConflictRetry(
opCtx,
"IndexBuildsManager::commitIndexBuild",
nss.ns(),
[builder, opCtx, collection, nss, &onCreateEachFn, &onCommitFn] {
WriteUnitOfWork wunit(opCtx);
auto status = builder->commit(opCtx, collection, onCreateEachFn, onCommitFn);
if (!status.isOK()) {
return status;
}
// Eventually, we will obtain the timestamp for completing the index build from the
// commitIndexBuild oplog entry.
// The current logic for timestamping index completion is consistent with the
// IndexBuilder. See SERVER-38986 and SERVER-34896.
IndexTimestampHelper::setGhostCommitTimestampForCatalogWrite(opCtx, nss);
wunit.commit();
return Status::OK();
});
}
bool IndexBuildsManager::abortIndexBuild(const UUID& buildUUID, const std::string& reason) {
stdx::unique_lock<Latch> lk(_mutex);
auto builderIt = _builders.find(buildUUID);
if (builderIt == _builders.end()) {
return false;
}
std::shared_ptr<MultiIndexBlock> builder = builderIt->second;
lk.unlock();
builder->abort(reason);
return true;
}
bool IndexBuildsManager::abortIndexBuildWithoutCleanup(OperationContext* opCtx,
const UUID& buildUUID,
const std::string& reason) {
stdx::unique_lock<Latch> lk(_mutex);
auto builderIt = _builders.find(buildUUID);
if (builderIt == _builders.end()) {
return false;
}
log() << "Index build aborted without cleanup: " << buildUUID << ": " << reason;
std::shared_ptr<MultiIndexBlock> builder = builderIt->second;
lk.unlock();
builder->abortWithoutCleanup(opCtx);
return true;
}
void IndexBuildsManager::tearDownIndexBuild(OperationContext* opCtx,
Collection* collection,
const UUID& buildUUID,
OnCleanUpFn onCleanUpFn) {
// TODO verify that the index builder is in a finished state before allowing its destruction.
auto builder = _getBuilder(buildUUID);
builder->cleanUpAfterBuild(opCtx, collection, onCleanUpFn);
_unregisterIndexBuild(buildUUID);
}
bool IndexBuildsManager::isBackgroundBuilding(const UUID& buildUUID) {
auto builder = _getBuilder(buildUUID);
return builder->isBackgroundBuilding();
}
void IndexBuildsManager::verifyNoIndexBuilds_forTestOnly() {
invariant(_builders.empty());
}
void IndexBuildsManager::_registerIndexBuild(UUID buildUUID) {
stdx::unique_lock<Latch> lk(_mutex);
std::shared_ptr<MultiIndexBlock> mib = std::make_shared<MultiIndexBlock>();
invariant(_builders.insert(std::make_pair(buildUUID, mib)).second);
}
void IndexBuildsManager::_unregisterIndexBuild(const UUID& buildUUID) {
stdx::unique_lock<Latch> lk(_mutex);
auto builderIt = _builders.find(buildUUID);
invariant(builderIt != _builders.end());
_builders.erase(builderIt);
}
std::shared_ptr<MultiIndexBlock> IndexBuildsManager::_getBuilder(const UUID& buildUUID) {
stdx::unique_lock<Latch> lk(_mutex);
auto builderIt = _builders.find(buildUUID);
invariant(builderIt != _builders.end());
return builderIt->second;
}
} // namespace mongo
|