JobRunner.php

Go to the documentation of this file.

<?php
class JobRunner {
    protected $debug;

    public function setDebugHandler( $debug ) {
        $this->debug = $debug;
    }

    public function run( array $options ) {
        $response = array( 'jobs' => array(), 'reached' => 'none-ready' );

        $type = isset( $options['type'] ) ? $options['type'] : false;
        $maxJobs = isset( $options['maxJobs'] ) ? $options['maxJobs'] : false;
        $maxTime = isset( $options['maxTime'] ) ? $options['maxTime'] : false;
        $noThrottle = isset( $options['throttle'] ) && !$options['throttle'];

        $group = JobQueueGroup::singleton();
        // Handle any required periodic queue maintenance
        $count = $group->executeReadyPeriodicTasks();
        if ( $count > 0 ) {
            $this->runJobsLog( "Executed $count periodic queue task(s)." );
        }

        // Flush any pending DB writes for sanity
        wfGetLBFactory()->commitMasterChanges();

        // Some jobs types should not run until a certain timestamp
        $backoffs = array(); // map of (type => UNIX expiry)
        $backoffDeltas = array(); // map of (type => seconds)
        $wait = 'wait'; // block to read backoffs the first time

        $jobsRun = 0;
        $timeMsTotal = 0;
        $flags = JobQueueGroup::USE_CACHE;
        $startTime = microtime( true ); // time since jobs started running
        $lastTime = microtime( true ); // time since last slave check
        do {
            // Sync the persistent backoffs with concurrent runners
            $backoffs = $this->syncBackoffDeltas( $backoffs, $backoffDeltas, $wait );
            $blacklist = $noThrottle ? array() : array_keys( $backoffs );
            $wait = 'nowait'; // less important now

            if ( $type === false ) {
                $job = $group->pop( JobQueueGroup::TYPE_DEFAULT, $flags, $blacklist );
            } elseif ( in_array( $type, $blacklist ) ) {
                $job = false; // requested queue in backoff state
            } else {
                $job = $group->pop( $type ); // job from a single queue
            }

            if ( $job ) { // found a job
                $jType = $job->getType();

                // Back off of certain jobs for a while (for throttling and for errors)
                $ttw = $this->getBackoffTimeToWait( $job );
                if ( $ttw > 0 ) {
                    // Always add the delta for other runners in case the time running the
                    // job negated the backoff for each individually but not collectively.
                    $backoffDeltas[$jType] = isset( $backoffDeltas[$jType] )
                        ? $backoffDeltas[$jType] + $ttw
                        : $ttw;
                    $backoffs = $this->syncBackoffDeltas( $backoffs, $backoffDeltas, $wait );
                }

                $this->runJobsLog( $job->toString() . " STARTING" );

                // Run the job...
                wfProfileIn( __METHOD__ . '-' . get_class( $job ) );
                $jobStartTime = microtime( true );
                try {
                    ++$jobsRun;
                    $status = $job->run();
                    $error = $job->getLastError();
                    wfGetLBFactory()->commitMasterChanges();
                } catch ( MWException $e ) {
                    MWExceptionHandler::rollbackMasterChangesAndLog( $e );
                    $status = false;
                    $error = get_class( $e ) . ': ' . $e->getMessage();
                    MWExceptionHandler::logException( $e );
                }
                $timeMs = intval( ( microtime( true ) - $jobStartTime ) * 1000 );
                wfProfileOut( __METHOD__ . '-' . get_class( $job ) );
                $timeMsTotal += $timeMs;

                // Mark the job as done on success or when the job cannot be retried
                if ( $status !== false || !$job->allowRetries() ) {
                    $group->ack( $job ); // done
                }

                // Back off of certain jobs for a while (for throttling and for errors)
                if ( $status === false && mt_rand( 0, 49 ) == 0 ) {
                    $ttw = max( $ttw, 30 ); // too many errors
                    $backoffDeltas[$jType] = isset( $backoffDeltas[$jType] )
                        ? $backoffDeltas[$jType] + $ttw
                        : $ttw;
                }

                if ( $status === false ) {
                    $this->runJobsLog( $job->toString() . " t=$timeMs error={$error}" );
                } else {
                    $this->runJobsLog( $job->toString() . " t=$timeMs good" );
                }

                $response['jobs'][] = array(
                    'type'   => $jType,
                    'status' => ( $status === false ) ? 'failed' : 'ok',
                    'error'  => $error,
                    'time'   => $timeMs
                );

                // Break out if we hit the job count or wall time limits...
                if ( $maxJobs && $jobsRun >= $maxJobs ) {
                    $response['reached'] = 'job-limit';
                    break;
                } elseif ( $maxTime && ( microtime( true ) - $startTime ) > $maxTime ) {
                    $response['reached'] = 'time-limit';
                    break;
                }

                // Don't let any of the main DB slaves get backed up
                $timePassed = microtime( true ) - $lastTime;
                if ( $timePassed >= 5 || $timePassed < 0 ) {
                    wfWaitForSlaves( $lastTime );
                    $lastTime = microtime( true );
                }
                // Don't let any queue slaves/backups fall behind
                if ( $jobsRun > 0 && ( $jobsRun % 100 ) == 0 ) {
                    $group->waitForBackups();
                }

                // Bail if near-OOM instead of in a job
                $this->assertMemoryOK();
            }
        } while ( $job ); // stop when there are no jobs

        // Sync the persistent backoffs for the next runJobs.php pass
        if ( $backoffDeltas ) {
            $this->syncBackoffDeltas( $backoffs, $backoffDeltas, 'wait' );
        }

        $response['backoffs'] = $backoffs;
        $response['elapsed'] = $timeMsTotal;

        return $response;
    }

    private function getBackoffTimeToWait( Job $job ) {
        global $wgJobBackoffThrottling;

        if ( !isset( $wgJobBackoffThrottling[$job->getType()] ) ||
            $job instanceof DuplicateJob // no work was done
        ) {
            return 0; // not throttled
        }

        $itemsPerSecond = $wgJobBackoffThrottling[$job->getType()];
        if ( $itemsPerSecond <= 0 ) {
            return 0; // not throttled
        }

        $seconds = 0;
        if ( $job->workItemCount() > 0 ) {
            $exactSeconds = $job->workItemCount() / $itemsPerSecond;
            // use randomized rounding
            $seconds = floor( $exactSeconds );
            $remainder = $exactSeconds - $seconds;
            $seconds += ( mt_rand() / mt_getrandmax() < $remainder ) ? 1 : 0;
        }

        return (int)$seconds;
    }

    private function loadBackoffs( array $backoffs, $mode = 'wait' ) {
        $section = new ProfileSection( __METHOD__ );

        $file = wfTempDir() . '/mw-runJobs-backoffs.json';
        if ( is_file( $file ) ) {
            $noblock = ( $mode === 'nowait' ) ? LOCK_NB : 0;
            $handle = fopen( $file, 'rb' );
            if ( !flock( $handle, LOCK_SH | $noblock ) ) {
                fclose( $handle );
                return $backoffs; // don't wait on lock
            }
            $content = stream_get_contents( $handle );
            flock( $handle, LOCK_UN );
            fclose( $handle );
            $ctime = microtime( true );
            $cBackoffs = json_decode( $content, true ) ?: array();
            foreach ( $cBackoffs as $type => $timestamp ) {
                if ( $timestamp < $ctime ) {
                    unset( $cBackoffs[$type] );
                }
            }
        } else {
            $cBackoffs = array();
        }

        return $cBackoffs;
    }

    private function syncBackoffDeltas( array $backoffs, array &$deltas, $mode = 'wait' ) {
        $section = new ProfileSection( __METHOD__ );

        if ( !$deltas ) {
            return $this->loadBackoffs( $backoffs, $mode );
        }

        $noblock = ( $mode === 'nowait' ) ? LOCK_NB : 0;
        $file = wfTempDir() . '/mw-runJobs-backoffs.json';
        $handle = fopen( $file, 'wb+' );
        if ( !flock( $handle, LOCK_EX | $noblock ) ) {
            fclose( $handle );
            return $backoffs; // don't wait on lock
        }
        $ctime = microtime( true );
        $content = stream_get_contents( $handle );
        $cBackoffs = json_decode( $content, true ) ?: array();
        foreach ( $deltas as $type => $seconds ) {
            $cBackoffs[$type] = isset( $cBackoffs[$type] ) && $cBackoffs[$type] >= $ctime
                ? $cBackoffs[$type] + $seconds
                : $ctime + $seconds;
        }
        foreach ( $cBackoffs as $type => $timestamp ) {
            if ( $timestamp < $ctime ) {
                unset( $cBackoffs[$type] );
            }
        }
        ftruncate( $handle, 0 );
        fwrite( $handle, json_encode( $cBackoffs ) );
        flock( $handle, LOCK_UN );
        fclose( $handle );

        $deltas = array();

        return $cBackoffs;
    }

    private function assertMemoryOK() {
        static $maxBytes = null;
        if ( $maxBytes === null ) {
            $m = array();
            if ( preg_match( '!^(\d+)(k|m|g|)$!i', ini_get( 'memory_limit' ), $m ) ) {
                list( , $num, $unit ) = $m;
                $conv = array( 'g' => 1073741824, 'm' => 1048576, 'k' => 1024, '' => 1 );
                $maxBytes = $num * $conv[strtolower( $unit )];
            } else {
                $maxBytes = 0;
            }
        }
        $usedBytes = memory_get_usage();
        if ( $maxBytes && $usedBytes >= 0.95 * $maxBytes ) {
            throw new MWException( "Detected excessive memory usage ($usedBytes/$maxBytes)." );
        }
    }

    private function runJobsLog( $msg ) {
        if ( $this->debug ) {
            call_user_func_array( $this->debug, array( wfTimestamp( TS_DB ) . " $msg\n" ) );
        }
        wfDebugLog( 'runJobs', $msg );
    }
}