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
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
|
/**
* Copyright (C) 2013 MongoDB Inc.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License, version 3,
* as published by the Free Software Foundation.
*
* 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
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
* 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 GNU Affero General 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/base/error_codes.h"
#include "mongo/base/owned_pointer_vector.h"
#include "mongo/client/remote_command_targeter.h"
#include "mongo/db/client.h"
#include "mongo/db/client.h"
#include "mongo/db/commands.h"
#include "mongo/db/commands/write_commands/write_commands_common.h"
#include "mongo/db/lasterror.h"
#include "mongo/db/stats/counters.h"
#include "mongo/db/stats/counters.h"
#include "mongo/s/chunk_manager_targeter.h"
#include "mongo/s/client/dbclient_multi_command.h"
#include "mongo/s/client/shard_registry.h"
#include "mongo/s/cluster_last_error_info.h"
#include "mongo/s/commands/cluster_explain.h"
#include "mongo/s/commands/cluster_write.h"
#include "mongo/s/grid.h"
#include "mongo/s/write_ops/batch_upconvert.h"
#include "mongo/s/write_ops/batched_command_request.h"
#include "mongo/s/write_ops/batched_command_response.h"
#include "mongo/util/timer.h"
namespace mongo {
using std::string;
using std::stringstream;
using std::vector;
namespace {
/**
* Base class for mongos write commands. Cluster write commands support batch writes and write
* concern, and return per-item error information. All cluster write commands use the entry
* point ClusterWriteCmd::run().
*
* Batch execution (targeting and dispatching) is performed by the BatchWriteExec class.
*/
class ClusterWriteCmd : public Command {
public:
virtual ~ClusterWriteCmd() {}
virtual bool slaveOk() const {
return false;
}
virtual bool supportsWriteConcern(const BSONObj& cmd) const override {
return true;
}
virtual Status checkAuthForCommand(Client* client,
const std::string& dbname,
const BSONObj& cmdObj) {
Status status = auth::checkAuthForWriteCommand(AuthorizationSession::get(client),
_writeType,
NamespaceString(parseNs(dbname, cmdObj)),
cmdObj);
// TODO: Remove this when we standardize GLE reporting from commands
if (!status.isOK()) {
LastError::get(client).setLastError(status.code(), status.reason());
}
return status;
}
virtual Status explain(OperationContext* txn,
const std::string& dbname,
const BSONObj& cmdObj,
ExplainCommon::Verbosity verbosity,
const rpc::ServerSelectionMetadata& serverSelectionMetadata,
BSONObjBuilder* out) const {
BatchedCommandRequest request(_writeType);
string errMsg;
if (!request.parseBSON(dbname, cmdObj, &errMsg) || !request.isValid(&errMsg)) {
return Status(ErrorCodes::FailedToParse, errMsg);
}
// We can only explain write batches of size 1.
if (request.sizeWriteOps() != 1U) {
return Status(ErrorCodes::InvalidLength, "explained write batches must be of size 1");
}
BSONObjBuilder explainCmdBob;
int options = 0;
ClusterExplain::wrapAsExplain(
cmdObj, verbosity, serverSelectionMetadata, &explainCmdBob, &options);
// We will time how long it takes to run the commands on the shards.
Timer timer;
// Target the command to the shards based on the singleton batch item.
BatchItemRef targetingBatchItem(&request, 0);
vector<Strategy::CommandResult> shardResults;
Status status =
_commandOpWrite(txn, dbname, explainCmdBob.obj(), targetingBatchItem, &shardResults);
if (!status.isOK()) {
return status;
}
return ClusterExplain::buildExplainResult(
txn, shardResults, ClusterExplain::kWriteOnShards, timer.millis(), out);
}
virtual bool run(OperationContext* txn,
const string& dbname,
BSONObj& cmdObj,
int options,
string& errmsg,
BSONObjBuilder& result) {
BatchedCommandRequest request(_writeType);
BatchedCommandResponse response;
ClusterWriter writer(true, 0);
LastError* cmdLastError = &LastError::get(cc());
{
// Disable the last error object for the duration of the write
LastError::Disabled disableLastError(cmdLastError);
// TODO: if we do namespace parsing, push this to the type
if (!request.parseBSON(dbname, cmdObj, &errmsg) || !request.isValid(&errmsg)) {
// Batch parse failure
response.setOk(false);
response.setErrCode(ErrorCodes::FailedToParse);
response.setErrMessage(errmsg);
} else {
writer.write(txn, request, &response);
}
dassert(response.isValid(NULL));
}
// Populate the lastError object based on the write response
cmdLastError->reset();
batchErrorToLastError(request, response, cmdLastError);
size_t numAttempts;
if (!response.getOk()) {
numAttempts = 0;
} else if (request.getOrdered() && response.isErrDetailsSet()) {
// Add one failed attempt
numAttempts = response.getErrDetailsAt(0)->getIndex() + 1;
} else {
numAttempts = request.sizeWriteOps();
}
// TODO: increase opcounters by more than one
if (_writeType == BatchedCommandRequest::BatchType_Insert) {
for (size_t i = 0; i < numAttempts; ++i) {
globalOpCounters.gotInsert();
}
} else if (_writeType == BatchedCommandRequest::BatchType_Update) {
for (size_t i = 0; i < numAttempts; ++i) {
globalOpCounters.gotUpdate();
}
} else if (_writeType == BatchedCommandRequest::BatchType_Delete) {
for (size_t i = 0; i < numAttempts; ++i) {
globalOpCounters.gotDelete();
}
}
// Save the last opTimes written on each shard for this client, to allow GLE to work
if (haveClient()) {
ClusterLastErrorInfo::get(cc()).addHostOpTimes(writer.getStats().getWriteOpTimes());
}
// TODO
// There's a pending issue about how to report response here. If we use
// the command infra-structure, we should reuse the 'errmsg' field. But
// we have already filed that message inside the BatchCommandResponse.
// return response.getOk();
result.appendElements(response.toBSON());
return true;
}
protected:
/**
* Instantiates a command that can be invoked by "name", which will be capable of issuing
* write batches of type "writeType", and will require privilege "action" to run.
*/
ClusterWriteCmd(StringData name, BatchedCommandRequest::BatchType writeType)
: Command(name), _writeType(writeType) {}
private:
// Type of batch (e.g. insert, update).
const BatchedCommandRequest::BatchType _writeType;
/**
* Executes a write command against a particular database, and targets the command based on
* a write operation.
*
* Does *not* retry or retarget if the metadata is stale.
*/
static Status _commandOpWrite(OperationContext* txn,
const std::string& dbName,
const BSONObj& command,
BatchItemRef targetingBatchItem,
std::vector<Strategy::CommandResult>* results) {
// Note that this implementation will not handle targeting retries and does not completely
// emulate write behavior
TargeterStats stats;
ChunkManagerTargeter targeter(
NamespaceString(targetingBatchItem.getRequest()->getTargetingNS()), &stats);
Status status = targeter.init(txn);
if (!status.isOK())
return status;
OwnedPointerVector<ShardEndpoint> endpointsOwned;
vector<ShardEndpoint*>& endpoints = endpointsOwned.mutableVector();
if (targetingBatchItem.getOpType() == BatchedCommandRequest::BatchType_Insert) {
ShardEndpoint* endpoint;
Status status = targeter.targetInsert(txn, targetingBatchItem.getDocument(), &endpoint);
if (!status.isOK())
return status;
endpoints.push_back(endpoint);
} else if (targetingBatchItem.getOpType() == BatchedCommandRequest::BatchType_Update) {
Status status = targeter.targetUpdate(txn, *targetingBatchItem.getUpdate(), &endpoints);
if (!status.isOK())
return status;
} else {
invariant(targetingBatchItem.getOpType() == BatchedCommandRequest::BatchType_Delete);
Status status = targeter.targetDelete(txn, *targetingBatchItem.getDelete(), &endpoints);
if (!status.isOK())
return status;
}
DBClientMultiCommand dispatcher;
// Assemble requests
for (vector<ShardEndpoint*>::const_iterator it = endpoints.begin(); it != endpoints.end();
++it) {
const ShardEndpoint* endpoint = *it;
const ReadPreferenceSetting readPref(ReadPreference::PrimaryOnly, TagSet());
auto shardStatus = grid.shardRegistry()->getShard(txn, endpoint->shardName);
if (!shardStatus.isOK()) {
return shardStatus.getStatus();
}
auto swHostAndPort = shardStatus.getValue()->getTargeter()->findHostNoWait(readPref);
if (!swHostAndPort.isOK()) {
return swHostAndPort.getStatus();
}
ConnectionString host(swHostAndPort.getValue());
dispatcher.addCommand(host, dbName, command);
}
// Errors reported when recv'ing responses
dispatcher.sendAll();
Status dispatchStatus = Status::OK();
// Recv responses
while (dispatcher.numPending() > 0) {
ConnectionString host;
RawBSONSerializable response;
Status status = dispatcher.recvAny(&host, &response);
if (!status.isOK()) {
// We always need to recv() all the sent operations
dispatchStatus = status;
continue;
}
Strategy::CommandResult result;
result.target = host;
{
auto shardStatus = grid.shardRegistry()->getShard(txn, host.toString());
if (!shardStatus.isOK()) {
return shardStatus.getStatus();
}
result.shardTargetId = shardStatus.getValue()->getId();
}
result.result = response.toBSON();
results->push_back(result);
}
return dispatchStatus;
}
};
class ClusterCmdInsert : public ClusterWriteCmd {
public:
ClusterCmdInsert() : ClusterWriteCmd("insert", BatchedCommandRequest::BatchType_Insert) {}
void help(stringstream& help) const {
help << "insert documents";
}
} clusterInsertCmd;
class ClusterCmdUpdate : public ClusterWriteCmd {
public:
ClusterCmdUpdate() : ClusterWriteCmd("update", BatchedCommandRequest::BatchType_Update) {}
void help(stringstream& help) const {
help << "update documents";
}
} clusterUpdateCmd;
class ClusterCmdDelete : public ClusterWriteCmd {
public:
ClusterCmdDelete() : ClusterWriteCmd("delete", BatchedCommandRequest::BatchType_Delete) {}
void help(stringstream& help) const {
help << "delete documents";
}
} clusterDeleteCmd;
} // namespace
} // namespace mongo
|