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
|
/**
* 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.
*/
#pragma once
#include "mongo/db/concurrency/exception_util.h"
#include "mongo/db/curop.h"
#include "mongo/db/prepare_conflict_tracker.h"
#include "mongo/db/storage/wiredtiger/wiredtiger_record_store.h"
#include "mongo/db/storage/wiredtiger/wiredtiger_recovery_unit.h"
#include "mongo/util/fail_point.h"
namespace mongo {
// When set, WT_ROLLBACK is returned in place of retrying on WT_PREPARE_CONFLICT errors.
extern FailPoint WTSkipPrepareConflictRetries;
extern FailPoint WTPrintPrepareConflictLog;
/**
* Logs a message with the number of prepare conflict retry attempts.
*/
void wiredTigerPrepareConflictLog(int attempt);
/**
* Logs a message to confirm we've hit the WTPrintPrepareConflictLog fail point.
*/
void wiredTigerPrepareConflictFailPointLog();
void wiredTigerPrepareConflictOplogResourceLog();
/**
* Runs the argument function f as many times as needed for f to return an error other than
* WT_PREPARE_CONFLICT. Each time f returns WT_PREPARE_CONFLICT we wait until the current unit of
* work commits or aborts, and then try f again. Imposes no upper limit on the number of times to
* re-try f, so any required timeout behavior must be enforced within f.
* The function f must return a WiredTiger error code.
*/
template <typename F>
int wiredTigerPrepareConflictRetry(OperationContext* opCtx, F&& f) {
invariant(opCtx);
int ret = WT_READ_CHECK(f());
if (ret != WT_PREPARE_CONFLICT)
return ret;
if (opCtx->recoveryUnit()->isTimestamped()) {
// This transaction is holding a resource in the form of an oplog slot. Committed
// transactions that get a later oplog slot will be unable to replicate until this resource
// is released (in the form of this transaction committing or aborting). For this case, we
// choose to abort our transaction and retry instead of blocking. It's possible that we can
// be blocking on a prepared update that requires replication to make progress, creating a
// stall in the MDB cluster.
wiredTigerPrepareConflictOplogResourceLog();
throwWriteConflictException("Holding a resource (oplog slot).");
}
auto recoveryUnit = WiredTigerRecoveryUnit::get(opCtx);
int attempts = 1;
// If we return from this function, we have either returned successfully or we've returned an
// error other than WT_PREPARE_CONFLICT. Reset PrepareConflictTracker accordingly.
ON_BLOCK_EXIT([opCtx] { PrepareConflictTracker::get(opCtx).endPrepareConflict(opCtx); });
PrepareConflictTracker::get(opCtx).beginPrepareConflict(opCtx);
auto client = opCtx->getClient();
if (client->isFromSystemConnection()) {
// System (internal) connections that hit a prepare conflict should be killable to prevent
// deadlocks with prepared transactions on replica set step up and step down.
stdx::lock_guard<Client> lk(*client);
invariant(client->canKillSystemOperationInStepdown(lk));
}
// It is contradictory to be running into a prepare conflict when we are ignoring interruptions,
// particularly when running code inside an
// OperationContext::runWithoutInterruptionExceptAtGlobalShutdown block. Operations executed in
// this way are expected to be set to ignore prepare conflicts.
invariant(!opCtx->isIgnoringInterrupts());
if (MONGO_unlikely(WTPrintPrepareConflictLog.shouldFail())) {
wiredTigerPrepareConflictFailPointLog();
}
CurOp::get(opCtx)->debug().additiveMetrics.incrementPrepareReadConflicts(1);
wiredTigerPrepareConflictLog(attempts);
const auto lockerInfo = opCtx->lockState()->getLockerInfo(boost::none);
invariant(lockerInfo);
for (const auto& lock : lockerInfo->locks) {
const auto type = lock.resourceId.getType();
// If a user operation on secondaries acquires a lock in MODE_S and then blocks on a prepare
// conflict with a prepared transaction, deadlock will occur at the commit time of the
// prepared transaction when it attempts to reacquire (since locks were yielded on
// secondaries) an IX lock that conflicts with the MODE_S lock held by the user operation.
// User operations that acquire MODE_X locks and block on prepare conflicts could lead to
// the same problem. However, user operations on secondaries should never hold MODE_X locks.
// Since prepared transactions will not reacquire RESOURCE_MUTEX / RESOURCE_METADATA locks
// at commit time, these lock types are safe. Therefore, invariant here that we do not get a
// prepare conflict while holding a global, database, or collection MODE_S lock (or MODE_X
// lock for completeness).
if (type == RESOURCE_GLOBAL || type == RESOURCE_DATABASE || type == RESOURCE_COLLECTION)
invariant(lock.mode != MODE_S && lock.mode != MODE_X,
str::stream() << lock.resourceId.toString() << " in " << modeName(lock.mode));
}
if (MONGO_unlikely(WTSkipPrepareConflictRetries.shouldFail())) {
// Callers of wiredTigerPrepareConflictRetry() should eventually call wtRCToStatus() via
// invariantWTOK() and have the WT_ROLLBACK error bubble up as a WriteConflictException.
// Enabling the "skipWriteConflictRetries" failpoint in conjunction with the
// "WTSkipPrepareConflictRetries" failpoint prevents the higher layers from retrying the
// entire operation.
return WT_ROLLBACK;
}
while (true) {
attempts++;
auto lastCount = recoveryUnit->getSessionCache()->getPrepareCommitOrAbortCount();
ret = WT_READ_CHECK(f());
if (ret != WT_PREPARE_CONFLICT)
return ret;
CurOp::get(opCtx)->debug().additiveMetrics.incrementPrepareReadConflicts(1);
wiredTigerPrepareConflictLog(attempts);
// Wait on the session cache to signal that a unit of work has been committed or aborted.
recoveryUnit->getSessionCache()->waitUntilPreparedUnitOfWorkCommitsOrAborts(opCtx,
lastCount);
}
}
} // namespace mongo
|
