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
|
/**
* 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.
*/
#include "mongo/platform/basic.h"
#include <string>
#include <boost/filesystem/operations.hpp>
#include "mongo/db/storage/storage_util.h"
#include "mongo/db/catalog/collection_catalog.h"
#include "mongo/db/storage/durable_catalog.h"
#include "mongo/db/storage/ident.h"
#include "mongo/db/storage/kv/kv_engine.h"
#include "mongo/db/storage/storage_engine.h"
#include "mongo/db/storage/storage_parameters_gen.h"
#include "mongo/logv2/log.h"
#define MONGO_LOGV2_DEFAULT_COMPONENT ::mongo::logv2::LogComponent::kStorage
namespace mongo {
namespace catalog {
namespace {
auto removeEmptyDirectory =
[](ServiceContext* svcCtx, StorageEngine* storageEngine, const NamespaceString& ns) {
// Nothing to do if not using directoryperdb or there are still collections in the database.
// If we don't support supportsPendingDrops then this is executing before the collection is
// removed from the catalog. In that case, just blindly attempt to delete the directory, it
// will only succeed if it is empty which is the behavior we want.
auto collectionCatalog = CollectionCatalog::latest(svcCtx);
const DatabaseName& dbName = ns.dbName();
if (!storageEngine->isUsingDirectoryPerDb() ||
(storageEngine->supportsPendingDrops() && !collectionCatalog->range(dbName).empty())) {
return;
}
boost::system::error_code ec;
boost::filesystem::remove(storageEngine->getFilesystemPathForDb(dbName), ec);
if (!ec) {
LOGV2(4888200, "Removed empty database directory", logAttrs(dbName));
} else if (collectionCatalog->range(dbName).empty()) {
// It is possible for a new collection to be created in the database between when we
// check whether the database is empty and actually attempting to remove the directory.
// In this case, don't log that the removal failed because it is expected. However,
// since we attempt to remove the directory for both the collection and index ident
// drops, once the database is empty it will be still logged until the final of these
// ident drops occurs.
LOGV2_DEBUG(4888201,
1,
"Failed to remove database directory",
logAttrs(dbName),
"error"_attr = ec.message());
}
};
} // namespace
void removeIndex(OperationContext* opCtx,
StringData indexName,
Collection* collection,
std::shared_ptr<IndexCatalogEntry> entry,
DataRemoval dataRemoval) {
auto durableCatalog = DurableCatalog::get(opCtx);
std::shared_ptr<Ident> ident = [&]() -> std::shared_ptr<Ident> {
if (!entry) {
return nullptr;
}
return entry->getSharedIdent();
}();
// If 'ident' is a nullptr, then there is no in-memory state. In that case, create an otherwise
// unreferenced Ident for the ident reaper to use: the reaper will not need to wait for existing
// users to finish.
if (!ident) {
ident = std::make_shared<Ident>(
durableCatalog->getIndexIdent(opCtx, collection->getCatalogId(), indexName));
}
// Run the first phase of drop to remove the catalog entry.
collection->removeIndex(opCtx, indexName);
// The OperationContext may not be valid when the RecoveryUnit executes the onCommit handlers.
// Therefore, anything that would normally be fetched from the opCtx must be passed in
// separately to the onCommit handler below.
//
// Index creation (and deletion) are allowed in multi-document transactions that use the same
// RecoveryUnit throughout but not the same OperationContext.
auto recoveryUnit = opCtx->recoveryUnit();
auto storageEngine = opCtx->getServiceContext()->getStorageEngine();
const bool isTwoPhaseDrop =
storageEngine->supportsPendingDrops() && dataRemoval == DataRemoval::kTwoPhase;
if (isTwoPhaseDrop) {
invariant(entry);
CollectionCatalog::get(opCtx)->dropIndex(
opCtx, collection->ns(), entry, /*isDropPending=*/true);
}
// Schedule the second phase of drop to delete the data when it is no longer in use, if the
// first phase is successfully committed.
opCtx->recoveryUnit()->onCommit(
[svcCtx = opCtx->getServiceContext(),
recoveryUnit,
storageEngine,
uuid = collection->uuid(),
nss = collection->ns(),
indexNameStr = indexName.toString(),
ident,
isTwoPhaseDrop](OperationContext*, boost::optional<Timestamp> commitTimestamp) {
StorageEngine::DropIdentCallback onDrop =
[svcCtx, storageEngine, nss, ident = ident->getIdent(), isTwoPhaseDrop] {
removeEmptyDirectory(svcCtx, storageEngine, nss);
if (isTwoPhaseDrop) {
CollectionCatalog::write(svcCtx, [&](CollectionCatalog& catalog) {
catalog.notifyIdentDropped(ident);
});
}
};
if (isTwoPhaseDrop) {
if (!commitTimestamp) {
// Standalone mode will not provide a timestamp.
commitTimestamp = Timestamp::min();
}
LOGV2(22206,
"Deferring table drop for index",
"index"_attr = indexNameStr,
logAttrs(nss),
"uuid"_attr = uuid,
"ident"_attr = ident->getIdent(),
"commitTimestamp"_attr = commitTimestamp);
storageEngine->addDropPendingIdent(*commitTimestamp, ident, std::move(onDrop));
} else {
LOGV2(6361201,
"Completing drop for index table immediately",
"ident"_attr = ident->getIdent(),
"index"_attr = indexNameStr,
logAttrs(nss));
// Intentionally ignoring failure here. Since we've removed the metadata pointing to
// the collection, we should never see it again anyway.
storageEngine->getEngine()
->dropIdent(recoveryUnit, ident->getIdent(), std::move(onDrop))
.ignore();
}
});
}
Status dropCollection(OperationContext* opCtx,
const NamespaceString& nss,
RecordId collectionCatalogId,
std::shared_ptr<Ident> ident) {
invariant(ident);
// Run the first phase of drop to remove the catalog entry.
Status status = DurableCatalog::get(opCtx)->dropCollection(opCtx, collectionCatalogId);
if (!status.isOK()) {
return status;
}
// The OperationContext may not be valid when the RecoveryUnit executes the onCommit handlers.
// Therefore, anything that would normally be fetched from the opCtx must be passed in
// separately to the onCommit handler below.
//
// Create (and drop) collection are allowed in multi-document transactions that use the same
// RecoveryUnit throughout but not the same OperationContext.
auto recoveryUnit = opCtx->recoveryUnit();
auto storageEngine = opCtx->getServiceContext()->getStorageEngine();
// Schedule the second phase of drop to delete the data when it is no longer in use, if the
// first phase is successfully committed.
opCtx->recoveryUnit()->onCommit(
[svcCtx = opCtx->getServiceContext(), recoveryUnit, storageEngine, nss, ident](
OperationContext*, boost::optional<Timestamp> commitTimestamp) {
StorageEngine::DropIdentCallback onDrop =
[svcCtx, storageEngine, nss, ident = ident->getIdent()] {
removeEmptyDirectory(svcCtx, storageEngine, nss);
if (storageEngine->supportsPendingDrops()) {
CollectionCatalog::write(svcCtx, [&](CollectionCatalog& catalog) {
catalog.notifyIdentDropped(ident);
});
}
};
if (storageEngine->supportsPendingDrops()) {
if (!commitTimestamp) {
// Standalone mode will not provide a timestamp.
commitTimestamp = Timestamp::min();
}
LOGV2(22214,
"Deferring table drop for collection",
logAttrs(nss),
"ident"_attr = ident->getIdent(),
"commitTimestamp"_attr = commitTimestamp);
storageEngine->addDropPendingIdent(*commitTimestamp, ident, std::move(onDrop));
} else {
// Intentionally ignoring failure here. Since we've removed the metadata pointing to
// the collection, we should never see it again anyway.
storageEngine->getEngine()
->dropIdent(recoveryUnit, ident->getIdent(), std::move(onDrop))
.ignore();
}
});
return Status::OK();
}
} // namespace catalog
} // namespace mongo
|
