summaryrefslogtreecommitdiff
path: root/src/mongo/db/logical_session_cache_impl.cpp
blob: 2095595eb5f3eb13663610d2f96d2c857479b45c (plain)
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
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
/**
 *    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::kControl

#include "mongo/platform/basic.h"

#include "mongo/db/logical_session_cache_impl.h"

#include "mongo/db/logical_session_id.h"
#include "mongo/db/logical_session_id_helpers.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/s/operation_sharding_state.h"
#include "mongo/db/service_context.h"
#include "mongo/platform/atomic_word.h"
#include "mongo/util/duration.h"
#include "mongo/util/log.h"
#include "mongo/util/periodic_runner.h"
#include "mongo/util/scopeguard.h"

namespace mongo {

namespace {

void clearShardingOperationFailedStatus(OperationContext* opCtx) {
    // We do not intend to immediately act upon sharding errors if we receive them during sessions
    // collection operations. We will instead attempt the same operations during the next refresh
    // cycle.
    OperationShardingState::get(opCtx).resetShardingOperationFailedStatus();
}

}  // namespace

LogicalSessionCacheImpl::LogicalSessionCacheImpl(std::unique_ptr<ServiceLiaison> service,
                                                 std::shared_ptr<SessionsCollection> collection,
                                                 ReapSessionsOlderThanFn reapSessionsOlderThanFn)
    : _service(std::move(service)),
      _sessionsColl(std::move(collection)),
      _reapSessionsOlderThanFn(std::move(reapSessionsOlderThanFn)) {
    _stats.setLastSessionsCollectionJobTimestamp(now());
    _stats.setLastTransactionReaperJobTimestamp(now());

    if (!disableLogicalSessionCacheRefresh) {
        _service->scheduleJob({"LogicalSessionCacheRefresh",
                               [this](Client* client) { _periodicRefresh(client); },
                               Milliseconds(logicalSessionRefreshMillis)});

        _service->scheduleJob({"LogicalSessionCacheReap",
                               [this](Client* client) { _periodicReap(client); },
                               Milliseconds(logicalSessionRefreshMillis)});
    }
}

LogicalSessionCacheImpl::~LogicalSessionCacheImpl() {
    joinOnShutDown();
}

void LogicalSessionCacheImpl::joinOnShutDown() {
    _service->join();
}

Status LogicalSessionCacheImpl::promote(LogicalSessionId lsid) {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    auto it = _activeSessions.find(lsid);
    if (it == _activeSessions.end()) {
        return {ErrorCodes::NoSuchSession, "no matching session record found in the cache"};
    }

    return Status::OK();
}

Status LogicalSessionCacheImpl::startSession(OperationContext* opCtx, LogicalSessionRecord record) {
    // Add the new record to our local cache. We will insert it into the sessions collection
    // the next time _refresh is called. If there is already a record in the cache for this
    // session, we'll just write over it with our newer, more recent one.
    return _addToCache(record);
}

Status LogicalSessionCacheImpl::refreshSessions(OperationContext* opCtx,
                                                const RefreshSessionsCmdFromClient& cmd) {
    // Update the timestamps of all these records in our cache.
    auto sessions = makeLogicalSessionIds(cmd.getRefreshSessions(), opCtx);
    for (const auto& lsid : sessions) {
        if (!promote(lsid).isOK()) {
            // This is a new record, insert it.
            auto addToCacheStatus = _addToCache(makeLogicalSessionRecord(opCtx, lsid, now()));
            if (!addToCacheStatus.isOK()) {
                return addToCacheStatus;
            }
        }
    }

    return Status::OK();
}

Status LogicalSessionCacheImpl::refreshSessions(OperationContext* opCtx,
                                                const RefreshSessionsCmdFromClusterMember& cmd) {
    // Update the timestamps of all these records in our cache.
    auto records = cmd.getRefreshSessionsInternal();
    for (const auto& record : records) {
        if (!promote(record.getId()).isOK()) {
            // This is a new record, insert it.
            auto addToCacheStatus = _addToCache(record);
            if (!addToCacheStatus.isOK()) {
                return addToCacheStatus;
            }
        }
    }

    return Status::OK();
}

Status LogicalSessionCacheImpl::vivify(OperationContext* opCtx, const LogicalSessionId& lsid) {
    if (!promote(lsid).isOK()) {
        return startSession(opCtx, makeLogicalSessionRecord(opCtx, lsid, now()));
    }
    return Status::OK();
}

Status LogicalSessionCacheImpl::refreshNow(Client* client) {
    try {
        _refresh(client);
    } catch (...) {
        return exceptionToStatus();
    }
    return Status::OK();
}

Status LogicalSessionCacheImpl::reapNow(Client* client) {
    return _reap(client);
}

Date_t LogicalSessionCacheImpl::now() {
    return _service->now();
}

size_t LogicalSessionCacheImpl::size() {
    stdx::lock_guard<stdx::mutex> lock(_mutex);
    return _activeSessions.size();
}

void LogicalSessionCacheImpl::_periodicRefresh(Client* client) {
    try {
        _refresh(client);
    } catch (...) {
        log() << "Failed to refresh session cache: " << exceptionToStatus();
    }
}

void LogicalSessionCacheImpl::_periodicReap(Client* client) {
    auto res = _reap(client);
    if (!res.isOK()) {
        log() << "Failed to reap transaction table: " << res;
    }

    return;
}

Status LogicalSessionCacheImpl::_reap(Client* client) {
    // Take the lock to update some stats.
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);

        // Clear the last set of stats for our new run.
        _stats.setLastTransactionReaperJobDurationMillis(0);
        _stats.setLastTransactionReaperJobEntriesCleanedUp(0);

        // Start the new run.
        _stats.setLastTransactionReaperJobTimestamp(now());
        _stats.setTransactionReaperJobCount(_stats.getTransactionReaperJobCount() + 1);
    }

    boost::optional<ServiceContext::UniqueOperationContext> uniqueCtx;
    auto* const opCtx = [&] {
        if (client->getOperationContext()) {
            return client->getOperationContext();
        }

        uniqueCtx.emplace(client->makeOperationContext());
        return uniqueCtx->get();
    }();

    int numReaped = 0;

    try {
        ON_BLOCK_EXIT([&opCtx] { clearShardingOperationFailedStatus(opCtx); });

        auto existsStatus = _sessionsColl->checkSessionsCollectionExists(opCtx);
        if (!existsStatus.isOK()) {
            StringData notSetUpWarning =
                "Sessions collection is not set up; "
                "waiting until next sessions reap interval";
            if (existsStatus.code() != ErrorCodes::NamespaceNotFound ||
                existsStatus.code() != ErrorCodes::NamespaceNotSharded) {
                log() << notSetUpWarning << ": " << existsStatus.reason();
            } else {
                log() << notSetUpWarning;
            }

            return Status::OK();
        }

        numReaped =
            _reapSessionsOlderThanFn(opCtx,
                                     *_sessionsColl,
                                     opCtx->getServiceContext()->getFastClockSource()->now() -
                                         Minutes(gTransactionRecordMinimumLifetimeMinutes));
    } catch (const DBException& ex) {
        {
            stdx::lock_guard<stdx::mutex> lk(_mutex);
            auto millis = now() - _stats.getLastTransactionReaperJobTimestamp();
            _stats.setLastTransactionReaperJobDurationMillis(millis.count());
        }

        return ex.toStatus();
    }

    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        auto millis = now() - _stats.getLastTransactionReaperJobTimestamp();
        _stats.setLastTransactionReaperJobDurationMillis(millis.count());
        _stats.setLastTransactionReaperJobEntriesCleanedUp(numReaped);
    }

    return Status::OK();
}

void LogicalSessionCacheImpl::_refresh(Client* client) {
    // Stats for serverStatus:
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);

        // Clear the refresh-related stats with the beginning of our run.
        _stats.setLastSessionsCollectionJobDurationMillis(0);
        _stats.setLastSessionsCollectionJobEntriesRefreshed(0);
        _stats.setLastSessionsCollectionJobEntriesEnded(0);
        _stats.setLastSessionsCollectionJobCursorsClosed(0);

        // Start the new run.
        _stats.setLastSessionsCollectionJobTimestamp(now());
        _stats.setSessionsCollectionJobCount(_stats.getSessionsCollectionJobCount() + 1);
    }

    // This will finish timing _refresh for our stats no matter when we return.
    const auto timeRefreshJob = makeGuard([this] {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        auto millis = now() - _stats.getLastSessionsCollectionJobTimestamp();
        _stats.setLastSessionsCollectionJobDurationMillis(millis.count());
    });

    // get or make an opCtx
    boost::optional<ServiceContext::UniqueOperationContext> uniqueCtx;
    auto* const opCtx = [&client, &uniqueCtx] {
        if (client->getOperationContext()) {
            return client->getOperationContext();
        }

        uniqueCtx.emplace(client->makeOperationContext());
        return uniqueCtx->get();
    }();

    ON_BLOCK_EXIT([&opCtx] { clearShardingOperationFailedStatus(opCtx); });

    auto setupStatus = _sessionsColl->setupSessionsCollection(opCtx);

    if (!setupStatus.isOK()) {
        log() << "Sessions collection is not set up; "
              << "waiting until next sessions refresh interval: " << setupStatus.reason();
        return;
    }

    LogicalSessionIdSet staleSessions;
    LogicalSessionIdSet explicitlyEndingSessions;
    LogicalSessionIdMap<LogicalSessionRecord> activeSessions;

    {
        using std::swap;
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        swap(explicitlyEndingSessions, _endingSessions);
        swap(activeSessions, _activeSessions);
    }

    // Create guards that in the case of a exception replace the ending or active sessions that
    // swapped out of LogicalSessionCache, and merges in any records that had been added since we
    // swapped them out.
    auto backSwap = [this](auto& member, auto& temp) {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        using std::swap;
        swap(member, temp);
        for (const auto& it : temp) {
            member.emplace(it);
        }
    };
    auto activeSessionsBackSwapper = makeGuard([&] { backSwap(_activeSessions, activeSessions); });
    auto explicitlyEndingBackSwaper =
        makeGuard([&] { backSwap(_endingSessions, explicitlyEndingSessions); });

    // remove all explicitlyEndingSessions from activeSessions
    for (const auto& lsid : explicitlyEndingSessions) {
        activeSessions.erase(lsid);
    }

    // Refresh all recently active sessions as well as for sessions attached to running ops
    LogicalSessionRecordSet activeSessionRecords;

    auto runningOpSessions = _service->getActiveOpSessions();

    for (const auto& it : runningOpSessions) {
        // if a running op is the cause of an upsert, we won't have a user name for the record
        if (explicitlyEndingSessions.count(it) > 0) {
            continue;
        }
        activeSessionRecords.insert(makeLogicalSessionRecord(it, now()));
    }
    for (const auto& it : activeSessions) {
        activeSessionRecords.insert(it.second);
    }

    // Refresh the active sessions in the sessions collection.
    uassertStatusOK(_sessionsColl->refreshSessions(opCtx, activeSessionRecords));
    activeSessionsBackSwapper.dismiss();
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        _stats.setLastSessionsCollectionJobEntriesRefreshed(activeSessionRecords.size());
    }

    // Remove the ending sessions from the sessions collection.
    uassertStatusOK(_sessionsColl->removeRecords(opCtx, explicitlyEndingSessions));
    explicitlyEndingBackSwaper.dismiss();
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        _stats.setLastSessionsCollectionJobEntriesEnded(explicitlyEndingSessions.size());
    }

    // Find which running, but not recently active sessions, are expired, and add them
    // to the list of sessions to kill cursors for

    KillAllSessionsByPatternSet patterns;

    auto openCursorSessions = _service->getOpenCursorSessions(opCtx);
    // Exclude sessions added to _activeSessions from the openCursorSession to avoid race between
    // killing cursors on the removed sessions and creating sessions.
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);

        for (const auto& it : _activeSessions) {
            auto newSessionIt = openCursorSessions.find(it.first);
            if (newSessionIt != openCursorSessions.end()) {
                openCursorSessions.erase(newSessionIt);
            }
        }
    }

    // think about pruning ending and active out of openCursorSessions
    auto statusAndRemovedSessions = _sessionsColl->findRemovedSessions(opCtx, openCursorSessions);

    if (statusAndRemovedSessions.isOK()) {
        auto removedSessions = statusAndRemovedSessions.getValue();
        for (const auto& lsid : removedSessions) {
            patterns.emplace(makeKillAllSessionsByPattern(opCtx, lsid));
        }
    } else {
        // Ignore errors.
    }

    // Add all of the explicitly ended sessions to the list of sessions to kill cursors for.
    for (const auto& lsid : explicitlyEndingSessions) {
        patterns.emplace(makeKillAllSessionsByPattern(opCtx, lsid));
    }

    SessionKiller::Matcher matcher(std::move(patterns));
    auto killRes = _service->killCursorsWithMatchingSessions(opCtx, std::move(matcher));
    {
        stdx::lock_guard<stdx::mutex> lk(_mutex);
        _stats.setLastSessionsCollectionJobCursorsClosed(killRes.second);
    }
}

void LogicalSessionCacheImpl::endSessions(const LogicalSessionIdSet& sessions) {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    _endingSessions.insert(begin(sessions), end(sessions));
}

LogicalSessionCacheStats LogicalSessionCacheImpl::getStats() {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    _stats.setActiveSessionsCount(_activeSessions.size());
    return _stats;
}

Status LogicalSessionCacheImpl::_addToCache(LogicalSessionRecord record) {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    if (_activeSessions.size() >= static_cast<size_t>(maxSessions)) {
        return {ErrorCodes::TooManyLogicalSessions, "cannot add session into the cache"};
    }

    _activeSessions.insert(std::make_pair(record.getId(), record));
    return Status::OK();
}

std::vector<LogicalSessionId> LogicalSessionCacheImpl::listIds() const {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    std::vector<LogicalSessionId> ret;
    ret.reserve(_activeSessions.size());
    for (const auto& id : _activeSessions) {
        ret.push_back(id.first);
    }
    return ret;
}

std::vector<LogicalSessionId> LogicalSessionCacheImpl::listIds(
    const std::vector<SHA256Block>& userDigests) const {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    std::vector<LogicalSessionId> ret;
    for (const auto& it : _activeSessions) {
        if (std::find(userDigests.cbegin(), userDigests.cend(), it.first.getUid()) !=
            userDigests.cend()) {
            ret.push_back(it.first);
        }
    }
    return ret;
}

boost::optional<LogicalSessionRecord> LogicalSessionCacheImpl::peekCached(
    const LogicalSessionId& id) const {
    stdx::lock_guard<stdx::mutex> lk(_mutex);
    const auto it = _activeSessions.find(id);
    if (it == _activeSessions.end()) {
        return boost::none;
    }
    return it->second;
}

}  // namespace mongo