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
|
/**
* 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::kSharding
#include "mongo/platform/basic.h"
#include "mongo/db/s/collection_sharding_state.h"
#include "mongo/db/repl/read_concern_args.h"
#include "mongo/db/s/operation_sharding_state.h"
#include "mongo/db/s/sharded_connection_info.h"
#include "mongo/s/stale_exception.h"
#include "mongo/util/fail_point_service.h"
#include "mongo/util/log.h"
#include "mongo/util/string_map.h"
namespace mongo {
MONGO_FAIL_POINT_DEFINE(useFCV44CheckShardVersionProtocol);
namespace {
class CollectionShardingStateMap {
CollectionShardingStateMap(const CollectionShardingStateMap&) = delete;
CollectionShardingStateMap& operator=(const CollectionShardingStateMap&) = delete;
public:
static const ServiceContext::Decoration<boost::optional<CollectionShardingStateMap>> get;
CollectionShardingStateMap(std::unique_ptr<CollectionShardingStateFactory> factory)
: _factory(std::move(factory)) {}
CollectionShardingState& getOrCreate(const NamespaceString& nss) {
stdx::lock_guard<stdx::mutex> lg(_mutex);
auto it = _collections.find(nss.ns());
if (it == _collections.end()) {
auto inserted = _collections.try_emplace(nss.ns(), _factory->make(nss));
invariant(inserted.second);
it = std::move(inserted.first);
}
return *it->second;
}
void report(OperationContext* opCtx, BSONObjBuilder* builder) {
BSONObjBuilder versionB(builder->subobjStart("versions"));
{
stdx::lock_guard<stdx::mutex> lg(_mutex);
for (auto& coll : _collections) {
const auto optMetadata = coll.second->getCurrentMetadataIfKnown();
if (optMetadata) {
const auto& metadata = *optMetadata;
versionB.appendTimestamp(coll.first, metadata->getShardVersion().toLong());
}
}
}
versionB.done();
}
private:
using CollectionsMap = StringMap<std::shared_ptr<CollectionShardingState>>;
std::unique_ptr<CollectionShardingStateFactory> _factory;
stdx::mutex _mutex;
CollectionsMap _collections;
};
const ServiceContext::Decoration<boost::optional<CollectionShardingStateMap>>
CollectionShardingStateMap::get =
ServiceContext::declareDecoration<boost::optional<CollectionShardingStateMap>>();
class UnshardedCollection : public ScopedCollectionMetadata::Impl {
public:
UnshardedCollection() = default;
const CollectionMetadata& get() override {
return _metadata;
}
private:
CollectionMetadata _metadata;
};
const auto kUnshardedCollection = std::make_shared<UnshardedCollection>();
boost::optional<ChunkVersion> getOperationReceivedVersion(OperationContext* opCtx,
const NamespaceString& nss) {
auto& oss = OperationShardingState::get(opCtx);
// If there is a version attached to the OperationContext, use it as the received version,
// otherwise get the received version from the ShardedConnectionInfo
if (oss.hasShardVersion()) {
return oss.getShardVersion(nss);
} else if (auto const info = ShardedConnectionInfo::get(opCtx->getClient(), false)) {
auto connectionShardVersion = info->getVersion(nss.ns());
// For backwards compatibility with map/reduce, which can access up to 2 sharded collections
// in a single call, the lack of version for a namespace on the collection must be treated
// as UNSHARDED
return connectionShardVersion.value_or(ChunkVersion::UNSHARDED());
}
// There is no shard version information on either 'opCtx' or 'client'. This means that the
// operation represented by 'opCtx' is unversioned, and the shard version is always OK for
// unversioned operations
return boost::none;
}
} // namespace
CollectionShardingState::CollectionShardingState(NamespaceString nss)
: _stateChangeMutex(nss.toString()), _nss(std::move(nss)) {}
CollectionShardingState* CollectionShardingState::get(OperationContext* opCtx,
const NamespaceString& nss) {
// Collection lock must be held to have a reference to the collection's sharding state
dassert(opCtx->lockState()->isCollectionLockedForMode(nss, MODE_IS));
auto& collectionsMap = CollectionShardingStateMap::get(opCtx->getServiceContext());
return &collectionsMap->getOrCreate(nss);
}
CollectionShardingState* CollectionShardingState::get_UNSAFE(ServiceContext* svcCtx,
const NamespaceString& nss) {
auto& collectionsMap = CollectionShardingStateMap::get(svcCtx);
return &collectionsMap->getOrCreate(nss);
}
void CollectionShardingState::report(OperationContext* opCtx, BSONObjBuilder* builder) {
auto& collectionsMap = CollectionShardingStateMap::get(opCtx->getServiceContext());
collectionsMap->report(opCtx, builder);
}
ScopedCollectionMetadata CollectionShardingState::getOrphansFilter(OperationContext* opCtx,
bool isCollection) {
const auto atClusterTime = repl::ReadConcernArgs::get(opCtx).getArgsAtClusterTime();
auto optMetadata = _getMetadataWithVersionCheckAt(opCtx, atClusterTime, isCollection);
if (!optMetadata)
return {kUnshardedCollection};
return {std::move(*optMetadata)};
}
ScopedCollectionMetadata CollectionShardingState::getCurrentMetadata() {
auto optMetadata = _getMetadata(boost::none);
if (!optMetadata)
return {kUnshardedCollection};
return {std::move(*optMetadata)};
}
boost::optional<ScopedCollectionMetadata> CollectionShardingState::getCurrentMetadataIfKnown() {
return _getMetadata(boost::none);
}
boost::optional<ChunkVersion> CollectionShardingState::getCurrentShardVersionIfKnown() {
const auto optMetadata = _getMetadata(boost::none);
if (!optMetadata)
return boost::none;
const auto& metadata = *optMetadata;
if (!metadata->isSharded())
return ChunkVersion::UNSHARDED();
return metadata->getCollVersion();
}
void CollectionShardingState::checkShardVersionOrThrow(OperationContext* opCtx, bool isCollection) {
(void)_getMetadataWithVersionCheckAt(opCtx, boost::none, isCollection);
}
boost::optional<ScopedCollectionMetadata> CollectionShardingState::_getMetadataWithVersionCheckAt(
OperationContext* opCtx,
const boost::optional<mongo::LogicalTime>& atClusterTime,
bool isCollection) {
const auto optReceivedShardVersion = getOperationReceivedVersion(opCtx, _nss);
if (!optReceivedShardVersion) {
return boost::none;
}
const auto& receivedShardVersion = *optReceivedShardVersion;
if (ChunkVersion::isIgnoredVersion(receivedShardVersion)) {
return boost::none;
}
// An operation with read concern 'available' should never have shardVersion set.
invariant(repl::ReadConcernArgs::get(opCtx).getLevel() !=
repl::ReadConcernLevel::kAvailableReadConcern);
auto csrLock = CSRLock::lockShared(opCtx, this);
auto metadata = _getMetadata(atClusterTime);
auto wantedShardVersion = ChunkVersion::UNSHARDED();
if (MONGO_FAIL_POINT(useFCV44CheckShardVersionProtocol)) {
LOG(0) << "Received shardVersion: " << receivedShardVersion << " for " << _nss.ns();
if (isCollection) {
LOG(0) << "Namespace " << _nss.ns() << " is collection, "
<< (metadata ? "have shardVersion cached" : "don't know shardVersion");
uassert(StaleConfigInfo(_nss, receivedShardVersion, wantedShardVersion),
"don't know shardVersion",
metadata);
wantedShardVersion = (*metadata)->getShardVersion();
}
LOG(0) << "Wanted shardVersion: " << wantedShardVersion << " for " << _nss.ns();
} else {
if (metadata && (*metadata)->isSharded()) {
wantedShardVersion = (*metadata)->getShardVersion();
}
}
auto criticalSectionSignal = [&] {
return _critSec.getSignal(opCtx->lockState()->isWriteLocked()
? ShardingMigrationCriticalSection::kWrite
: ShardingMigrationCriticalSection::kRead);
}();
if (criticalSectionSignal) {
// Set migration critical section on operation sharding state: operation will wait for the
// migration to finish before returning failure and retrying.
auto& oss = OperationShardingState::get(opCtx);
oss.setMigrationCriticalSectionSignal(criticalSectionSignal);
uasserted(StaleConfigInfo(_nss, receivedShardVersion, wantedShardVersion),
str::stream() << "migration commit in progress for " << _nss.ns());
}
if (receivedShardVersion.isWriteCompatibleWith(wantedShardVersion)) {
return metadata;
}
//
// Figure out exactly why not compatible, send appropriate error message
// The versions themselves are returned in the error, so not needed in messages here
//
StaleConfigInfo sci(_nss, receivedShardVersion, wantedShardVersion);
uassert(std::move(sci),
str::stream() << "epoch mismatch detected for " << _nss.ns() << ", "
<< "the collection may have been dropped and recreated",
wantedShardVersion.epoch() == receivedShardVersion.epoch());
if (!wantedShardVersion.isSet() && receivedShardVersion.isSet()) {
uasserted(std::move(sci),
str::stream() << "this shard no longer contains chunks for " << _nss.ns() << ", "
<< "the collection may have been dropped");
}
if (wantedShardVersion.isSet() && !receivedShardVersion.isSet()) {
uasserted(std::move(sci),
str::stream() << "this shard contains chunks for " << _nss.ns() << ", "
<< "but the client expects unsharded collection");
}
if (wantedShardVersion.majorVersion() != receivedShardVersion.majorVersion()) {
// Could be > or < - wanted is > if this is the source of a migration, wanted < if this is
// the target of a migration
uasserted(std::move(sci), str::stream() << "version mismatch detected for " << _nss.ns());
}
// Those are all the reasons the versions can mismatch
MONGO_UNREACHABLE;
}
void CollectionShardingState::enterCriticalSectionCatchUpPhase(OperationContext* opCtx, CSRLock&) {
invariant(opCtx->lockState()->isCollectionLockedForMode(_nss, MODE_X));
_critSec.enterCriticalSectionCatchUpPhase();
}
void CollectionShardingState::enterCriticalSectionCommitPhase(OperationContext* opCtx, CSRLock&) {
invariant(opCtx->lockState()->isCollectionLockedForMode(_nss, MODE_X));
_critSec.enterCriticalSectionCommitPhase();
}
void CollectionShardingState::exitCriticalSection(OperationContext* opCtx, CSRLock&) {
invariant(opCtx->lockState()->isCollectionLockedForMode(_nss, MODE_IX));
_critSec.exitCriticalSection();
}
void CollectionShardingStateFactory::set(ServiceContext* service,
std::unique_ptr<CollectionShardingStateFactory> factory) {
auto& collectionsMap = CollectionShardingStateMap::get(service);
invariant(!collectionsMap);
invariant(factory);
collectionsMap.emplace(std::move(factory));
}
void CollectionShardingStateFactory::clear(ServiceContext* service) {
auto& collectionsMap = CollectionShardingStateMap::get(service);
collectionsMap.reset();
}
} // namespace mongo
|