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
|
/**
* 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/client/remote_command_targeter.h"
#include "mongo/db/auth/action_set.h"
#include "mongo/db/auth/action_type.h"
#include "mongo/db/auth/authorization_session.h"
#include "mongo/db/commands.h"
#include "mongo/db/repl/repl_client_info.h"
#include "mongo/db/s/active_migrations_registry.h"
#include "mongo/db/s/chunk_move_write_concern_options.h"
#include "mongo/db/s/migration_source_manager.h"
#include "mongo/db/s/move_timing_helper.h"
#include "mongo/db/s/sharding_state.h"
#include "mongo/db/s/sharding_statistics.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/grid.h"
#include "mongo/s/request_types/migration_secondary_throttle_options.h"
#include "mongo/s/request_types/move_chunk_request.h"
#include "mongo/util/concurrency/notification.h"
#include "mongo/util/concurrency/thread_pool.h"
#include "mongo/util/fail_point.h"
#include "mongo/util/log.h"
namespace mongo {
namespace {
/**
* If the specified status is not OK logs a warning and throws a DBException corresponding to the
* specified status.
*/
void uassertStatusOKWithWarning(const Status& status) {
if (!status.isOK()) {
warning() << "Chunk move failed" << causedBy(redact(status));
uassertStatusOK(status);
}
}
const WriteConcernOptions kMajorityWriteConcern(WriteConcernOptions::kMajority,
// Note: Even though we're setting UNSET here,
// kMajority implies JOURNAL if journaling is
// supported by mongod and
// writeConcernMajorityJournalDefault is set to true
// in the ReplSetConfig.
WriteConcernOptions::SyncMode::UNSET,
WriteConcernOptions::kWriteConcernTimeoutSharding);
// Tests can pause and resume moveChunk's progress at each step by enabling/disabling each failpoint
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep1);
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep2);
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep3);
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep4);
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep5);
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep6);
MONGO_FAIL_POINT_DEFINE(moveChunkHangAtStep7);
class MoveChunkCommand : public BasicCommand {
public:
MoveChunkCommand() : BasicCommand("moveChunk") {}
std::string help() const override {
return "should not be calling this directly";
}
AllowedOnSecondary secondaryAllowed(ServiceContext*) const override {
return AllowedOnSecondary::kNever;
}
bool adminOnly() const override {
return true;
}
bool supportsWriteConcern(const BSONObj& cmd) const override {
return true;
}
Status checkAuthForCommand(Client* client,
const std::string& dbname,
const BSONObj& cmdObj) const override {
if (!AuthorizationSession::get(client)->isAuthorizedForActionsOnResource(
ResourcePattern::forClusterResource(), ActionType::internal)) {
return Status(ErrorCodes::Unauthorized, "Unauthorized");
}
return Status::OK();
}
std::string parseNs(const std::string& dbname, const BSONObj& cmdObj) const override {
return CommandHelpers::parseNsFullyQualified(cmdObj);
}
bool run(OperationContext* opCtx,
const std::string& dbname,
const BSONObj& cmdObj,
BSONObjBuilder&) override {
auto shardingState = ShardingState::get(opCtx);
uassertStatusOK(shardingState->canAcceptShardedCommands());
const MoveChunkRequest moveChunkRequest = uassertStatusOK(
MoveChunkRequest::createFromCommand(NamespaceString(parseNs(dbname, cmdObj)), cmdObj));
// Make sure we're as up-to-date as possible with shard information. This catches the case
// where we might have changed a shard's host by removing/adding a shard with the same name.
Grid::get(opCtx)->shardRegistry()->reload(opCtx);
auto scopedMigration = uassertStatusOK(
ActiveMigrationsRegistry::get(opCtx).registerDonateChunk(moveChunkRequest));
// Check if there is an existing migration running and if so, join it
if (scopedMigration.mustExecute()) {
auto moveChunkComplete =
ExecutorFuture<void>(_getExecutor())
.then([moveChunkRequest,
scopedMigration = std::move(scopedMigration),
serviceContext = opCtx->getServiceContext()]() mutable {
ThreadClient tc("MoveChunk", serviceContext);
{
stdx::lock_guard<Client> lk(*tc.get());
tc->setSystemOperationKillable(lk);
}
auto uniqueOpCtx = Client::getCurrent()->makeOperationContext();
auto opCtx = uniqueOpCtx.get();
// Note: This internal authorization is tied to the lifetime of the client.
AuthorizationSession::get(opCtx->getClient())
->grantInternalAuthorization(opCtx->getClient());
Status status = {ErrorCodes::InternalError, "Uninitialized value"};
try {
_runImpl(opCtx, moveChunkRequest);
status = Status::OK();
} catch (const DBException& e) {
status = e.toStatus();
if (status.code() == ErrorCodes::LockTimeout) {
ShardingStatistics::get(opCtx)
.countDonorMoveChunkLockTimeout.addAndFetch(1);
}
} catch (const std::exception& e) {
scopedMigration.signalComplete(
{ErrorCodes::InternalError,
str::stream()
<< "Severe error occurred while running moveChunk command: "
<< e.what()});
throw;
}
scopedMigration.signalComplete(status);
uassertStatusOK(status);
});
moveChunkComplete.get(opCtx);
} else {
uassertStatusOK(scopedMigration.waitForCompletion(opCtx));
}
if (moveChunkRequest.getWaitForDelete()) {
// Ensure we capture the latest opTime in the system, since range deletion happens
// asynchronously with a different OperationContext. This must be done after the above
// join, because each caller must set the opTime to wait for writeConcern for on its own
// OperationContext.
// TODO (SERVER-30183): If this moveChunk joined an active moveChunk that did not have
// waitForDelete=true, the captured opTime may not reflect all the deletes.
auto& replClient = repl::ReplClientInfo::forClient(opCtx->getClient());
replClient.setLastOpToSystemLastOpTime(opCtx);
WriteConcernResult writeConcernResult;
writeConcernResult.wTimedOut = false;
Status majorityStatus = waitForWriteConcern(
opCtx, replClient.getLastOp(), kMajorityWriteConcern, &writeConcernResult);
if (!majorityStatus.isOK()) {
if (!writeConcernResult.wTimedOut) {
uassertStatusOK(majorityStatus);
}
return false;
}
}
return true;
}
private:
static void _runImpl(OperationContext* opCtx, const MoveChunkRequest& moveChunkRequest) {
const auto writeConcernForRangeDeleter =
uassertStatusOK(ChunkMoveWriteConcernOptions::getEffectiveWriteConcern(
opCtx, moveChunkRequest.getSecondaryThrottle()));
// Resolve the donor and recipient shards and their connection string
auto const shardRegistry = Grid::get(opCtx)->shardRegistry();
const auto donorConnStr =
uassertStatusOK(shardRegistry->getShard(opCtx, moveChunkRequest.getFromShardId()))
->getConnString();
const auto recipientHost = uassertStatusOK([&] {
auto recipientShard =
uassertStatusOK(shardRegistry->getShard(opCtx, moveChunkRequest.getToShardId()));
return recipientShard->getTargeter()->findHost(
opCtx, ReadPreferenceSetting{ReadPreference::PrimaryOnly});
}());
std::string unusedErrMsg;
MoveTimingHelper moveTimingHelper(opCtx,
"from",
moveChunkRequest.getNss().ns(),
moveChunkRequest.getMinKey(),
moveChunkRequest.getMaxKey(),
6, // Total number of steps
&unusedErrMsg,
moveChunkRequest.getToShardId(),
moveChunkRequest.getFromShardId());
moveTimingHelper.done(1);
moveChunkHangAtStep1.pauseWhileSet();
MigrationSourceManager migrationSourceManager(
opCtx, moveChunkRequest, donorConnStr, recipientHost);
moveTimingHelper.done(2);
moveChunkHangAtStep2.pauseWhileSet();
uassertStatusOKWithWarning(migrationSourceManager.startClone());
moveTimingHelper.done(3);
moveChunkHangAtStep3.pauseWhileSet();
uassertStatusOKWithWarning(migrationSourceManager.awaitToCatchUp());
moveTimingHelper.done(4);
moveChunkHangAtStep4.pauseWhileSet();
uassertStatusOKWithWarning(migrationSourceManager.enterCriticalSection());
uassertStatusOKWithWarning(migrationSourceManager.commitChunkOnRecipient());
moveTimingHelper.done(5);
moveChunkHangAtStep5.pauseWhileSet();
uassertStatusOKWithWarning(migrationSourceManager.commitChunkMetadataOnConfig());
moveTimingHelper.done(6);
moveChunkHangAtStep6.pauseWhileSet();
}
private:
// Returns a single-threaded executor to be used to run moveChunk commands. The executor is
// initialized on the first call to this function. Uses a shared_ptr because a shared_ptr is
// required to work with ExecutorFutures.
static std::shared_ptr<ThreadPool> _getExecutor() {
static Mutex mutex = MONGO_MAKE_LATCH("MoveChunkExecutor::_mutex");
static std::shared_ptr<ThreadPool> executor;
stdx::lock_guard<Latch> lg(mutex);
if (!executor) {
ThreadPool::Options options;
options.poolName = "MoveChunk";
options.minThreads = 0;
// We limit the size of the thread pool to a single thread because currently there can
// only be one moveChunk operation on a shard at a time.
options.maxThreads = 1;
executor = std::make_shared<ThreadPool>(std::move(options));
executor->startup();
}
return executor;
}
} moveChunkCmd;
} // namespace
} // namespace mongo
|