DeferredUpdates.php

Go to the documentation of this file.

<?php
use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface;
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
use Psr\Log\LoggerInterface;
use Wikimedia\Rdbms\DBTransactionError;
use Wikimedia\Rdbms\IDatabase;
use Wikimedia\Rdbms\ILBFactory;
use Wikimedia\Rdbms\LBFactory;
use Wikimedia\Rdbms\LoadBalancer;
 
class DeferredUpdates {
    private static $scopeStack;
 
    public const ALL = 0;
    public const PRESEND = 1;
    public const POSTSEND = 2;
 
    public const STAGES = [ self::PRESEND, self::POSTSEND ];
 
    private const BIG_QUEUE_SIZE = 100;
 
    public static function addUpdate( DeferrableUpdate $update, $stage = self::POSTSEND ) {
        global $wgCommandLineMode;
 
        self::getScopeStack()->current()->addUpdate( $update, $stage );
        // If CLI mode is active and no RDBMs transaction round is in the way, then run all
        // the pending updates now. This is needed for scripts that never, or rarely, use the
        // RDBMs layer, but that do modify systems via deferred updates. This logic avoids
        // excessive pending update queue sizes when long-running scripts never trigger the
        // basic RDBMs hooks for running pending updates.
        if ( $wgCommandLineMode ) {
            self::tryOpportunisticExecute( 'run' );
        }
    }
 
    public static function addCallableUpdate( $callable, $stage = self::POSTSEND, $dbw = null ) {
        self::addUpdate( new MWCallableUpdate( $callable, wfGetCaller(), $dbw ), $stage );
    }
 
    public static function doUpdates( $mode = 'run', $stage = self::ALL ) {
        $services = MediaWikiServices::getInstance();
        $stats = $services->getStatsdDataFactory();
        $lbf = $services->getDBLoadBalancerFactory();
        $logger = LoggerFactory::getInstance( 'DeferredUpdates' );
        $httpMethod = $services->getMainConfig()->get( 'CommandLineMode' )
            ? 'cli'
            : strtolower( RequestContext::getMain()->getRequest()->getMethod() );
 
        $guiError = null;
        $exception = null;
 
        $scope = self::getScopeStack()->current();
 
        // T249069: recursion is not possible once explicit transaction rounds are involved
        $activeUpdate = $scope->getActiveUpdate();
        if ( $activeUpdate ) {
            $class = get_class( $activeUpdate );
            if ( !( $activeUpdate instanceof TransactionRoundAwareUpdate ) ) {
                throw new LogicException(
                    __METHOD__ . ": reached from $class, which is not TransactionRoundAwareUpdate"
                );
            }
            if ( $activeUpdate->getTransactionRoundRequirement() !== $activeUpdate::TRX_ROUND_ABSENT ) {
                throw new LogicException(
                    __METHOD__ . ": reached from $class, which does not specify TRX_ROUND_ABSENT"
                );
            }
        }
 
        $scope->processUpdates(
            $stage,
            function ( DeferrableUpdate $update, $activeStage )
                use ( $mode, $lbf, $logger, $stats, $httpMethod, &$guiError, &$exception )
            {
                // If applicable, just enqueue the update as a job in the job queue system
                if ( $mode === 'enqueue' && $update instanceof EnqueueableDataUpdate ) {
                    self::jobify( $update, $lbf, $logger, $stats, $httpMethod );
 
                    return;
                }
 
                // Otherwise, run the update....
                $scopeStack = self::getScopeStack();
                $childScope = $scopeStack->descend( $activeStage, $update );
                try {
                    $e = self::run( $update, $lbf, $logger, $stats, $httpMethod );
                    $guiError = $guiError ?: ( $e instanceof ErrorPageError ? $e : null );
                    $exception = $exception ?: $e;
                    // Any addUpdate() calls between descend() and ascend() used the sub-queue.
                    // In rare cases, DeferrableUpdate::doUpdates() will process them by calling
                    // doUpdates() itself. In any case, process remaining updates in the subqueue.
                    // them, enqueueing them, or transferring them to the parent scope
                    // queues as appropriate...
                    $childScope->processUpdates(
                        $activeStage,
                        function ( DeferrableUpdate $subUpdate )
                            use ( $lbf, $logger, $stats, $httpMethod, &$guiError, &$exception )
                        {
                            $e = self::run( $subUpdate, $lbf, $logger, $stats, $httpMethod );
                            $guiError = $guiError ?: ( $e instanceof ErrorPageError ? $e : null );
                            $exception = $exception ?: $e;
                        }
                    );
                } finally {
                    $scopeStack->ascend();
                }
            }
        );
 
        // VW-style hack to work around T190178, so we can make sure
        // PageMetaDataUpdater doesn't throw exceptions.
        if ( $exception && defined( 'MW_PHPUNIT_TEST' ) ) {
            throw $exception;
        }
 
        // Throw the first of any GUI errors as long as the context is HTTP pre-send. However,
        // callers should check permissions *before* enqueueing updates. If the main transaction
        // round actions succeed but some deferred updates fail due to permissions errors then
        // there is a risk that some secondary data was not properly updated.
        if ( $guiError && $stage === self::PRESEND && !headers_sent() ) {
            throw $guiError;
        }
    }
 
    public static function tryOpportunisticExecute( $mode = 'run' ) {
        // Leave execution up to the current loop if an update is already in progress
        if ( self::getRecursiveExecutionStackDepth() ) {
            return false;
        }
 
        // Run the updates for this context if they will have outer transaction scope
        if ( !self::areDatabaseTransactionsActive() ) {
            self::doUpdates( $mode, self::ALL );
 
            return true;
        }
 
        if ( self::pendingUpdatesCount() >= self::BIG_QUEUE_SIZE ) {
            // There are a large number of pending updates and none of them can run yet.
            // The odds of losing updates due to an error increase when executing long queues
            // and when large amounts of time pass while tasks are queued. Mitigate this by
            // trying to migrate updates to the job queue system (where applicable).
            self::getScopeStack()->current()->consumeMatchingUpdates(
                self::ALL,
                EnqueueableDataUpdate::class,
                static function ( EnqueueableDataUpdate $update ) {
                    $spec = $update->getAsJobSpecification();
                    JobQueueGroup::singleton( $spec['domain'] )->push( $spec['job'] );
                }
            );
        }
 
        return false;
    }
 
    public static function pendingUpdatesCount() {
        return self::getScopeStack()->current()->pendingUpdatesCount();
    }
 
    public static function getPendingUpdates( $stage = self::ALL ) {
        return self::getScopeStack()->current()->getPendingUpdates( $stage );
    }
 
    public static function clearPendingUpdates() {
        self::getScopeStack()->current()->clearPendingUpdates();
    }
 
    public static function getRecursiveExecutionStackDepth() {
        return self::getScopeStack()->getRecursiveDepth();
    }
 
    private static function run(
        DeferrableUpdate $update,
        ILBFactory $lbFactory,
        LoggerInterface $logger,
        StatsdDataFactoryInterface $stats,
        $httpMethod
    ): ?Throwable {
        $suffix = ( $update instanceof DeferrableCallback ) ? "_{$update->getOrigin()}" : '';
        $type = get_class( $update ) . $suffix;
        $stats->increment( "deferred_updates.$httpMethod.$type" );
 
        $updateId = spl_object_id( $update );
        $logger->debug( __METHOD__ . ": started $type #$updateId" );
        $startTime = microtime( true );
        $e = null;
        try {
            self::attemptUpdate( $update, $lbFactory );
 
            return null;
        } catch ( Throwable $e ) {
        } finally {
            $executionTime = microtime( true ) - $startTime;
            $logger->debug( __METHOD__ . ": ended $type #$updateId, processing time: $executionTime" );
        }
 
        MWExceptionHandler::logException( $e );
        $logger->error(
            "Deferred update '{deferred_type}' failed to run.",
            [
                'deferred_type' => $type,
                'exception' => $e,
            ]
        );
 
        $lbFactory->rollbackPrimaryChanges( __METHOD__ );
 
        // Try to push the update as a job so it can run later if possible
        if ( $update instanceof EnqueueableDataUpdate ) {
            $jobEx = null;
            try {
                $spec = $update->getAsJobSpecification();
                JobQueueGroup::singleton( $spec['domain'] )->push( $spec['job'] );
 
                return $e;
            } catch ( Throwable $jobEx ) {
            }
 
            MWExceptionHandler::logException( $jobEx );
            $logger->error(
                "Deferred update '{deferred_type}' failed to enqueue as a job.",
                [
                    'deferred_type' => $type,
                    'exception' => $jobEx,
                ]
            );
 
            $lbFactory->rollbackPrimaryChanges( __METHOD__ );
        }
 
        return $e;
    }
 
    private static function jobify(
        EnqueueableDataUpdate $update,
        LBFactory $lbFactory,
        LoggerInterface $logger,
        StatsdDataFactoryInterface $stats,
        $httpMethod
    ) {
        $type = get_class( $update );
        $stats->increment( "deferred_updates.$httpMethod.$type" );
 
        $jobEx = null;
        try {
            $spec = $update->getAsJobSpecification();
            JobQueueGroup::singleton( $spec['domain'] )->push( $spec['job'] );
 
            return;
        } catch ( Throwable $jobEx ) {
        }
 
        MWExceptionHandler::logException( $jobEx );
        $logger->error(
            "Deferred update '$type' failed to enqueue as a job.",
            [
                'deferred_type' => $type,
                'exception' => $jobEx,
            ]
        );
 
        $lbFactory->rollbackPrimaryChanges( __METHOD__ );
    }
 
    public static function attemptUpdate( DeferrableUpdate $update, ILBFactory $lbFactory ) {
        $ticket = $lbFactory->getEmptyTransactionTicket( __METHOD__ );
        if ( !$ticket || $lbFactory->hasTransactionRound() ) {
            throw new DBTransactionError( null, "A database transaction round is pending." );
        }
 
        if ( $update instanceof DataUpdate ) {
            $update->setTransactionTicket( $ticket );
        }
 
        // Designate $update::doUpdate() as the write round owner
        $fnameTrxOwner = ( $update instanceof DeferrableCallback )
            ? $update->getOrigin()
            : get_class( $update ) . '::doUpdate';
        // Determine whether the write round will be explicit or implicit
        $useExplicitTrxRound = !(
            $update instanceof TransactionRoundAwareUpdate &&
            $update->getTransactionRoundRequirement() == $update::TRX_ROUND_ABSENT
        );
 
        // Flush any pending changes left over from an implicit transaction round
        if ( $useExplicitTrxRound ) {
            $lbFactory->beginPrimaryChanges( $fnameTrxOwner ); // new explicit round
        } else {
            $lbFactory->commitPrimaryChanges( $fnameTrxOwner ); // new implicit round
        }
        // Run the update after any stale primary DB view snapshots have been flushed
        $update->doUpdate();
        // Commit any pending changes from the explicit or implicit transaction round
        $lbFactory->commitPrimaryChanges( $fnameTrxOwner );
    }
 
    private static function areDatabaseTransactionsActive() {
        $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        if ( $lbFactory->hasTransactionRound() || !$lbFactory->isReadyForRoundOperations() ) {
            return true;
        }
 
        $connsBusy = false;
        $lbFactory->forEachLB( static function ( LoadBalancer $lb ) use ( &$connsBusy ) {
            $lb->forEachOpenPrimaryConnection( static function ( IDatabase $conn ) use ( &$connsBusy ) {
                if ( $conn->writesOrCallbacksPending() || $conn->explicitTrxActive() ) {
                    $connsBusy = true;
                }
            } );
        } );
 
        return $connsBusy;
    }
 
    private static function getScopeStack() {
        if ( self::$scopeStack === null ) {
            self::$scopeStack = new DeferredUpdatesScopeStack();
        }
 
        return self::$scopeStack;
    }
}