/** * 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 * . * * 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::kReplicationInitialSync #include "mongo/platform/basic.h" #include "mongo/db/repl/database_cloner.h" #include #include #include #include "mongo/client/remote_command_retry_scheduler.h" #include "mongo/db/catalog/collection_options.h" #include "mongo/db/commands/list_collections_filter.h" #include "mongo/db/repl/repl_server_parameters_gen.h" #include "mongo/db/repl/storage_interface.h" #include "mongo/stdx/functional.h" #include "mongo/util/assert_util.h" #include "mongo/util/destructor_guard.h" #include "mongo/util/fail_point_service.h" #include "mongo/util/log.h" #include "mongo/util/str.h" namespace mongo { namespace repl { // Failpoint which causes the initial sync function to hang before running listCollections. MONGO_FAIL_POINT_DEFINE(initialSyncHangBeforeListCollections); namespace { using LockGuard = stdx::lock_guard; using UniqueLock = stdx::unique_lock; using executor::RemoteCommandRequest; const char* kNameFieldName = "name"; const char* kOptionsFieldName = "options"; const char* kInfoFieldName = "info"; const char* kUUIDFieldName = "uuid"; // Failpoint which causes initial sync to hang right after listCollections, but before cloning // any colelctions in the 'database' database. MONGO_FAIL_POINT_DEFINE(initialSyncHangAfterListCollections); /** * Default listCollections predicate. */ bool acceptAllPred(const BSONObj&) { return true; } /** * Creates a listCollections command obj with an optional filter. */ BSONObj createListCollectionsCommandObject(const BSONObj& filter) { BSONObjBuilder output; output.append("listCollections", 1); if (!filter.isEmpty()) { output.append("filter", filter); } return output.obj(); } } // namespace DatabaseCloner::DatabaseCloner(executor::TaskExecutor* executor, ThreadPool* dbWorkThreadPool, const HostAndPort& source, const std::string& dbname, const BSONObj& listCollectionsFilter, const ListCollectionsPredicateFn& listCollectionsPred, StorageInterface* si, const CollectionCallbackFn& collWork, CallbackFn onCompletion) : _executor(executor), _dbWorkThreadPool(dbWorkThreadPool), _source(source), _dbname(dbname), _listCollectionsFilter( listCollectionsFilter.isEmpty() ? ListCollectionsFilter::makeTypeCollectionFilter() : ListCollectionsFilter::addTypeCollectionFilter(listCollectionsFilter)), _listCollectionsPredicate(listCollectionsPred ? listCollectionsPred : acceptAllPred), _storageInterface(si), _collectionWork(collWork), _onCompletion(std::move(onCompletion)), _listCollectionsFetcher(_executor, _source, _dbname, createListCollectionsCommandObject(_listCollectionsFilter), [=](const StatusWith& result, Fetcher::NextAction * nextAction, BSONObjBuilder * getMoreBob) { _listCollectionsCallback(result, nextAction, getMoreBob); }, ReadPreferenceSetting::secondaryPreferredMetadata(), RemoteCommandRequest::kNoTimeout /* find network timeout */, RemoteCommandRequest::kNoTimeout /* getMore network timeout */, RemoteCommandRetryScheduler::makeRetryPolicy( numInitialSyncListCollectionsAttempts.load(), executor::RemoteCommandRequest::kNoTimeout, RemoteCommandRetryScheduler::kAllRetriableErrors)), _startCollectionCloner([](CollectionCloner& cloner) { return cloner.startup(); }) { // Fetcher throws an exception on null executor. invariant(executor); uassert(ErrorCodes::BadValue, "db worker thread pool cannot be null", dbWorkThreadPool); uassert(ErrorCodes::BadValue, "empty database name", !dbname.empty()); uassert(ErrorCodes::BadValue, "storage interface cannot be null", si); uassert(ErrorCodes::BadValue, "collection callback function cannot be null", collWork); uassert(ErrorCodes::BadValue, "callback function cannot be null", _onCompletion); _stats.dbname = _dbname; } DatabaseCloner::~DatabaseCloner() { DESTRUCTOR_GUARD(shutdown(); join();); } const std::vector& DatabaseCloner::getCollectionInfos_forTest() const { LockGuard lk(_mutex); return _collectionInfos; } bool DatabaseCloner::isActive() const { LockGuard lk(_mutex); return _isActive_inlock(); } bool DatabaseCloner::_isActive_inlock() const { return State::kRunning == _state || State::kShuttingDown == _state; } bool DatabaseCloner::_isShuttingDown() const { LockGuard lk(_mutex); return State::kShuttingDown == _state; } Status DatabaseCloner::startup() noexcept { UniqueLock lk(_mutex); switch (_state) { case State::kPreStart: _state = State::kRunning; break; case State::kRunning: return Status(ErrorCodes::InternalError, "database cloner already started"); case State::kShuttingDown: return Status(ErrorCodes::ShutdownInProgress, "database cloner shutting down"); case State::kComplete: return Status(ErrorCodes::ShutdownInProgress, "database cloner completed"); } MONGO_FAIL_POINT_BLOCK(initialSyncHangBeforeListCollections, customArgs) { const auto& data = customArgs.getData(); const auto databaseElem = data["database"]; if (!databaseElem || databaseElem.checkAndGetStringData() == _dbname) { lk.unlock(); log() << "initial sync - initialSyncHangBeforeListCollections fail point " "enabled. Blocking until fail point is disabled."; while (MONGO_FAIL_POINT(initialSyncHangBeforeListCollections) && !_isShuttingDown()) { mongo::sleepsecs(1); } lk.lock(); } } _stats.start = _executor->now(); LOG(1) << "Scheduling listCollections call for database: " << _dbname; Status scheduleResult = _listCollectionsFetcher.schedule(); if (!scheduleResult.isOK()) { error() << "Error scheduling listCollections for database: " << _dbname << ", error:" << scheduleResult; _state = State::kComplete; return scheduleResult; } return Status::OK(); } void DatabaseCloner::shutdown() { stdx::lock_guard lock(_mutex); switch (_state) { case State::kPreStart: // Transition directly from PreStart to Complete if not started yet. _state = State::kComplete; return; case State::kRunning: _state = State::kShuttingDown; break; case State::kShuttingDown: case State::kComplete: // Nothing to do if we are already in ShuttingDown or Complete state. return; } for (auto&& collectionCloner : _collectionCloners) { collectionCloner.shutdown(); } _listCollectionsFetcher.shutdown(); } DatabaseCloner::Stats DatabaseCloner::getStats() const { LockGuard lk(_mutex); DatabaseCloner::Stats stats = _stats; for (auto&& collectionCloner : _collectionCloners) { stats.collectionStats.emplace_back(collectionCloner.getStats()); } return stats; } void DatabaseCloner::join() { UniqueLock lk(_mutex); _condition.wait(lk, [this]() { return !_isActive_inlock(); }); } void DatabaseCloner::setScheduleDbWorkFn_forTest(const ScheduleDbWorkFn& work) { LockGuard lk(_mutex); _scheduleDbWorkFn = work; } void DatabaseCloner::setStartCollectionClonerFn( const StartCollectionClonerFn& startCollectionCloner) { _startCollectionCloner = startCollectionCloner; } DatabaseCloner::State DatabaseCloner::getState_forTest() const { stdx::lock_guard lk(_mutex); return _state; } void DatabaseCloner::_listCollectionsCallback(const StatusWith& result, Fetcher::NextAction* nextAction, BSONObjBuilder* getMoreBob) { if (!result.isOK()) { _finishCallback(result.getStatus().withContext( str::stream() << "Error issuing listCollections on db '" << _dbname << "' (host:" << _source.toString() << ")")); return; } auto batchData(result.getValue()); auto&& documents = batchData.documents; UniqueLock lk(_mutex); // We may be called with multiple batches leading to a need to grow _collectionInfos. _collectionInfos.reserve(_collectionInfos.size() + documents.size()); std::copy_if(documents.begin(), documents.end(), std::back_inserter(_collectionInfos), _listCollectionsPredicate); _stats.collections += _collectionInfos.size(); // The fetcher will continue to call with kGetMore until an error or the last batch. if (*nextAction == Fetcher::NextAction::kGetMore) { invariant(getMoreBob); getMoreBob->append("getMore", batchData.cursorId); getMoreBob->append("collection", batchData.nss.coll()); return; } // Nothing to do for an empty database. if (_collectionInfos.empty()) { _finishCallback_inlock(lk, Status::OK()); return; } MONGO_FAIL_POINT_BLOCK(initialSyncHangAfterListCollections, options) { const BSONObj& data = options.getData(); if (data["database"].String() == _dbname) { log() << "initial sync - initialSyncHangAfterListCollections fail point " "enabled. Blocking until fail point is disabled."; while (MONGO_FAIL_POINT(initialSyncHangAfterListCollections)) { mongo::sleepsecs(1); } } } _collectionNamespaces.reserve(_collectionInfos.size()); std::set seen; for (auto&& info : _collectionInfos) { BSONElement nameElement = info.getField(kNameFieldName); if (nameElement.eoo()) { _finishCallback_inlock( lk, {ErrorCodes::FailedToParse, str::stream() << "collection info must contain '" << kNameFieldName << "' " << "field : " << info}); return; } if (nameElement.type() != mongo::String) { _finishCallback_inlock( lk, {ErrorCodes::TypeMismatch, str::stream() << "'" << kNameFieldName << "' field must be a string: " << info}); return; } const std::string collectionName = nameElement.String(); if (seen.find(collectionName) != seen.end()) { _finishCallback_inlock(lk, {ErrorCodes::Error(51005), str::stream() << "collection info contains duplicate collection name " << "'" << collectionName << "': " << info}); return; } BSONElement optionsElement = info.getField(kOptionsFieldName); if (optionsElement.eoo()) { _finishCallback_inlock( lk, {ErrorCodes::FailedToParse, str::stream() << "collection info must contain '" << kOptionsFieldName << "' " << "field : " << info}); return; } if (!optionsElement.isABSONObj()) { _finishCallback_inlock(lk, Status(ErrorCodes::TypeMismatch, str::stream() << "'" << kOptionsFieldName << "' field must be an object: " << info)); return; } const BSONObj optionsObj = optionsElement.Obj(); CollectionOptions options; auto parseStatus = options.parse(optionsObj, CollectionOptions::parseForStorage); if (!parseStatus.isOK()) { _finishCallback_inlock(lk, parseStatus); return; } BSONElement infoElement = info.getField(kInfoFieldName); if (infoElement.isABSONObj()) { BSONElement uuidElement = infoElement[kUUIDFieldName]; if (!uuidElement.eoo()) { auto res = CollectionUUID::parse(uuidElement); if (!res.isOK()) { _finishCallback_inlock(lk, res.getStatus()); return; } options.uuid = res.getValue(); } } // TODO(SERVER-27994): Ensure UUID present when FCV >= "3.6". seen.insert(collectionName); _collectionNamespaces.emplace_back(_dbname, collectionName); auto&& nss = *_collectionNamespaces.crbegin(); try { _collectionCloners.emplace_back( _executor, _dbWorkThreadPool, _source, nss, options, [=](const Status& status) { return _collectionClonerCallback(status, nss); }, _storageInterface, collectionClonerBatchSize); } catch (const AssertionException& ex) { _finishCallback_inlock(lk, ex.toStatus()); return; } } if (_scheduleDbWorkFn) { for (auto&& collectionCloner : _collectionCloners) { collectionCloner.setScheduleDbWorkFn_forTest(_scheduleDbWorkFn); } } // Start first collection cloner. _currentCollectionClonerIter = _collectionCloners.begin(); LOG(1) << " cloning collection " << _currentCollectionClonerIter->getSourceNamespace(); Status startStatus = _startCollectionCloner(*_currentCollectionClonerIter); if (!startStatus.isOK()) { LOG(1) << " failed to start collection cloning on " << _currentCollectionClonerIter->getSourceNamespace() << ": " << redact(startStatus); _finishCallback_inlock(lk, startStatus); return; } } void DatabaseCloner::_collectionClonerCallback(const Status& status, const NamespaceString& nss) { UniqueLock lk(_mutex); auto collStatus = Status::OK(); // Record failure, but do not return just yet, in case we want to do some logging. if (!status.isOK()) { collStatus = status.withContext( str::stream() << "Error cloning collection '" << nss.toString() << "'"); } // Forward collection cloner result to caller. lk.unlock(); _collectionWork(collStatus, nss); lk.lock(); // Failure to clone a collection will stop the database cloner from // cloning the rest of the collections in the listCollections result. if (!collStatus.isOK()) { Status failStatus = {ErrorCodes::InitialSyncFailure, collStatus.toString()}; _finishCallback_inlock(lk, failStatus); return; } ++_stats.clonedCollections; _currentCollectionClonerIter++; if (_currentCollectionClonerIter != _collectionCloners.end()) { Status startStatus = _startCollectionCloner(*_currentCollectionClonerIter); if (!startStatus.isOK()) { LOG(1) << " failed to start collection cloning on " << _currentCollectionClonerIter->getSourceNamespace() << ": " << redact(startStatus); _finishCallback_inlock(lk, startStatus); return; } return; } _finishCallback_inlock(lk, Status::OK()); } void DatabaseCloner::_finishCallback(const Status& status) { _onCompletion(status); LockGuard lk(_mutex); invariant(_state != State::kComplete); _state = State::kComplete; _condition.notify_all(); _stats.end = _executor->now(); LOG(1) << " database: " << _dbname << ", stats: " << _stats.toString(); } void DatabaseCloner::_finishCallback_inlock(UniqueLock& lk, const Status& status) { if (lk.owns_lock()) { lk.unlock(); } _finishCallback(status); } std::string DatabaseCloner::getDBName() const { return _dbname; } std::string DatabaseCloner::Stats::toString() const { return toBSON().toString(); } BSONObj DatabaseCloner::Stats::toBSON() const { BSONObjBuilder bob; bob.append("dbname", dbname); append(&bob); return bob.obj(); } void DatabaseCloner::Stats::append(BSONObjBuilder* builder) const { builder->appendNumber("collections", collections); builder->appendNumber("clonedCollections", clonedCollections); if (start != Date_t()) { builder->appendDate("start", start); if (end != Date_t()) { builder->appendDate("end", end); auto elapsed = end - start; long long elapsedMillis = duration_cast(elapsed).count(); builder->appendNumber("elapsedMillis", elapsedMillis); } } for (auto&& collection : collectionStats) { BSONObjBuilder collectionBuilder(builder->subobjStart(collection.ns)); collection.append(&collectionBuilder); collectionBuilder.doneFast(); } } std::ostream& operator<<(std::ostream& os, const DatabaseCloner::State& state) { switch (state) { case DatabaseCloner::State::kPreStart: return os << "PreStart"; case DatabaseCloner::State::kRunning: return os << "Running"; case DatabaseCloner::State::kShuttingDown: return os << "ShuttingDown"; case DatabaseCloner::State::kComplete: return os << "Complete"; } MONGO_UNREACHABLE; } } // namespace repl } // namespace mongo