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
|
/**
* 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.
*/
#pragma once
#include <string>
#include <vector>
#include "mongo/db/concurrency/d_concurrency.h"
#include "mongo/db/repl/abstract_async_component.h"
#include "mongo/db/repl/oplog.h"
#include "mongo/db/repl/oplog_buffer.h"
#include "mongo/db/repl/oplog_entry.h"
#include "mongo/db/repl/tenant_oplog_batcher.h"
#include "mongo/db/serverless/serverless_types_gen.h"
#include "mongo/util/future.h"
namespace mongo {
class ThreadPool;
namespace repl {
/**
* This class reads oplog entries from a tenant migration, then applies those entries to the
* (real) oplog, then writes out no-op entries corresponding to the original oplog entries
* from the oplog buffer. Applier will not apply, but will write no-op entries for,
* entries before the applyFromOpTime.
*
*/
class TenantOplogApplier : public AbstractAsyncComponent,
public std::enable_shared_from_this<TenantOplogApplier> {
public:
struct OpTimePair {
OpTimePair() = default;
OpTimePair(OpTime in_donorOpTime, OpTime in_recipientOpTime)
: donorOpTime(in_donorOpTime), recipientOpTime(in_recipientOpTime) {}
bool operator<(const OpTimePair& other) const {
if (donorOpTime == other.donorOpTime)
return recipientOpTime < other.recipientOpTime;
return donorOpTime < other.donorOpTime;
}
std::string toString() const {
return BSON("donorOpTime" << donorOpTime << "recipientOpTime" << recipientOpTime)
.toString();
}
OpTime donorOpTime;
OpTime recipientOpTime;
};
TenantOplogApplier(const UUID& migrationUuid,
const MigrationProtocolEnum& protocol,
boost::optional<std::string> tenantId,
OpTime StartApplyingAfterOpTime,
RandomAccessOplogBuffer* oplogBuffer,
std::shared_ptr<executor::TaskExecutor> executor,
ThreadPool* writerPool,
Timestamp resumeBatchingTs = Timestamp());
virtual ~TenantOplogApplier();
/**
* Return a future which will be notified when that optime has been reached. Future will
* contain donor and recipient optime of last oplog entry in batch where donor optime is greater
* than passed-in time. To be noted, recipient optime returned in the future can be null if the
* tenant oplog applier has never applied any tenant oplog entries (i.e., non resume token no-op
* entries) till that batch.
*/
SemiFuture<OpTimePair> getNotificationForOpTime(OpTime donorOpTime);
size_t getNumOpsApplied() {
stdx::lock_guard lk(_mutex);
return _numOpsApplied;
}
/**
* This should only be called once before the applier starts.
*/
void setCloneFinishedRecipientOpTime(OpTime cloneFinishedRecipientOpTime);
/**
* Returns the optime the applier will start applying from.
*/
OpTime getStartApplyingAfterOpTime() const;
/**
* Returns the timestamp the applier will resume batching from.
*/
Timestamp getResumeBatchingTs() const;
private:
void _doStartup_inlock() final;
void _doShutdown_inlock() noexcept final;
void _preJoin() noexcept final;
void _finishShutdown(WithLock lk, Status status);
void _applyLoop(TenantOplogBatch batch);
bool _shouldStopApplying(Status status);
// Indicates an oplog entry should be ignored and not applied.
bool _shouldIgnore(const OplogEntry& entry);
void _applyOplogBatch(TenantOplogBatch* batch);
Status _applyOplogBatchPerWorker(std::vector<ApplierOperation>* ops);
void _checkNsAndUuidsBelongToTenant(OperationContext* opCtx, const TenantOplogBatch& batch);
OpTimePair _writeNoOpEntries(OperationContext* opCtx, const TenantOplogBatch& batch);
using TenantNoOpEntry = std::pair<ApplierOperation, std::vector<OplogSlot>::iterator>;
void _writeNoOpsForRange(OpObserver* opObserver,
std::vector<TenantNoOpEntry>::const_iterator begin,
std::vector<TenantNoOpEntry>::const_iterator end);
void _writeSessionNoOpsForRange(std::vector<TenantNoOpEntry>::const_iterator begin,
std::vector<TenantNoOpEntry>::const_iterator end);
Status _applyOplogEntryOrGroupedInserts(OperationContext* opCtx,
const OplogEntryOrGroupedInserts& entryOrGroupedInserts,
OplogApplication::Mode oplogApplicationMode,
bool isDataConsistent);
std::vector<std::vector<ApplierOperation>> _fillWriterVectors(OperationContext* opCtx,
TenantOplogBatch* batch);
/**
* Acquires Intent Exclusive (IX) lock for each tenant referred to by oplog entries [entryBegin;
* entryEnd) and returns lock objects.
*/
std::vector<Lock::TenantLock> _acquireIntentExclusiveTenantLocks(
OperationContext* opCtx,
std::vector<TenantNoOpEntry>::const_iterator entryBegin,
std::vector<TenantNoOpEntry>::const_iterator entryEnd) const;
/**
* Sets the _finalStatus to the new status if and only if the old status is "OK".
*/
void _setFinalStatusIfOk(WithLock, Status newStatus);
Mutex* _getMutex() noexcept final {
return &_mutex;
}
Mutex _mutex = MONGO_MAKE_LATCH("TenantOplogApplier::_mutex");
// All member variables are labeled with one of the following codes indicating the
// synchronization rules for accessing them.
//
// (R) Read-only in concurrent operation; no synchronization required.
// (S) Self-synchronizing; access according to class's own rules.
// (M) Reads and writes guarded by _mutex
// (X) Access only allowed from the main flow of control called from run() or constructor.
// Handles consuming oplog entries from the OplogBuffer for oplog application.
std::shared_ptr<TenantOplogBatcher> _oplogBatcher; // (R)
const UUID _migrationUuid; // (R)
const MigrationProtocolEnum _protocol; // (R)
// For multi-tenant migration protocol, _tenantId is set.
// But, for shard merge protcol, _tenantId is empty.
const boost::optional<std::string> _tenantId; // (R)
const OpTime _startApplyingAfterOpTime; // (R)
RandomAccessOplogBuffer* _oplogBuffer; // (R)
std::shared_ptr<executor::TaskExecutor> _executor; // (R)
// All no-op entries written by this tenant migration should have OpTime greater than this
// OpTime.
OpTime _cloneFinishedRecipientOpTime = OpTime(); // (R)
// Keeps track of last applied donor and recipient optimes by the tenant oplog applier.
// This gets updated only on batch boundaries.
OpTimePair _lastAppliedOpTimesUpToLastBatch; // (M)
// Pool of worker threads for writing ops to the databases.
// Not owned by us.
ThreadPool* const _writerPool; // (S)
// The timestamp to resume batching from. A null timestamp indicates that the oplog applier
// is starting fresh (not a retry), and will start batching from the beginning of the oplog
// buffer.
const Timestamp _resumeBatchingTs; // (R)
std::map<OpTime, SharedPromise<OpTimePair>> _opTimeNotificationList; // (M)
Status _finalStatus = Status::OK(); // (M)
stdx::unordered_set<UUID, UUID::Hash> _knownGoodUuids; // (X)
bool _applyLoopApplyingBatch = false; // (M)
size_t _numOpsApplied{0}; // (M)
};
/**
* Creates the default thread pool for writer tasks.
*/
std::unique_ptr<ThreadPool> makeTenantMigrationWriterPool();
std::unique_ptr<ThreadPool> makeTenantMigrationWriterPool(int threadCount);
} // namespace repl
} // namespace mongo
|
