CdnCacheUpdate.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use MediaWiki\Page\PageReference;
use Wikimedia\Assert\Assert;
use Wikimedia\IPUtils;
 
class CdnCacheUpdate implements DeferrableUpdate, MergeableUpdate {
    private $urlTuples = [];
    private $pageTuples = [];
 
    private const MAX_REBOUND_DELAY = 300;
 
    public function __construct( array $targets, array $options = [] ) {
        $delay = min(
            (int)max( $options['reboundDelay'] ?? 0, 0 ),
            self::MAX_REBOUND_DELAY
        );
 
        foreach ( $targets as $target ) {
            if ( $target instanceof PageReference ) {
                $this->pageTuples[] = [ $target, $delay ];
            } else {
                $this->urlTuples[] = [ $target, $delay ];
            }
        }
    }
 
    public function merge( MergeableUpdate $update ) {
        Assert::parameterType( __CLASS__, $update, '$update' );
        '@phan-var self $update';
 
        $this->urlTuples = array_merge( $this->urlTuples, $update->urlTuples );
        $this->pageTuples = array_merge( $this->pageTuples, $update->pageTuples );
    }
 
    public static function newFromTitles( $pages, $urls = [] ) {
        return new CdnCacheUpdate( array_merge( $pages, $urls ) );
    }
 
    public function doUpdate() {
        // Resolve the final list of URLs just before purging them (T240083)
        $reboundDelayByUrl = $this->resolveReboundDelayByUrl();
 
        // Send the immediate purges to CDN
        self::purge( array_keys( $reboundDelayByUrl ) );
        $immediatePurgeTimestamp = time();
 
        // Get the URLs that need rebound purges, grouped by seconds of purge delay
        $urlsWithReboundByDelay = [];
        foreach ( $reboundDelayByUrl as $url => $delay ) {
            if ( $delay > 0 ) {
                $urlsWithReboundByDelay[$delay][] = $url;
            }
        }
        // Enqueue delayed purge jobs for these URLs (usually only one job)
        $jobs = [];
        foreach ( $urlsWithReboundByDelay as $delay => $urls ) {
            $jobs[] = new CdnPurgeJob( [
                'urls' => $urls,
                'jobReleaseTimestamp' => $immediatePurgeTimestamp + $delay
            ] );
        }
        JobQueueGroup::singleton()->lazyPush( $jobs );
    }
 
    public static function purge( array $urls ) {
        $cdnServers = MediaWikiServices::getInstance()->getMainConfig()->get( 'CdnServers' );
        $htcpRouting = MediaWikiServices::getInstance()->getMainConfig()->get( 'HTCPRouting' );
        if ( !$urls ) {
            return;
        }
 
        // Remove duplicate URLs from list
        $urls = array_unique( $urls );
 
        wfDebugLog( 'squid', __METHOD__ . ': ' . implode( ' ', $urls ) );
 
        // Reliably broadcast the purge to all edge nodes
        $ts = microtime( true );
        $relayerGroup = MediaWikiServices::getInstance()->getEventRelayerGroup();
        $relayerGroup->getRelayer( 'cdn-url-purges' )->notifyMulti(
            'cdn-url-purges',
            array_map(
                static function ( $url ) use ( $ts ) {
                    return [
                        'url' => $url,
                        'timestamp' => $ts,
                    ];
                },
                $urls
            )
        );
 
        // Send lossy UDP broadcasting if enabled
        if ( $htcpRouting ) {
            self::HTCPPurge( $urls );
        }
 
        // Do direct server purges if enabled (this does not scale very well)
        if ( $cdnServers ) {
            self::naivePurge( $urls );
        }
    }
 
    public function getUrls() {
        return array_keys( $this->resolveReboundDelayByUrl() );
    }
 
    private function resolveReboundDelayByUrl() {
        $services = MediaWikiServices::getInstance();
        // Avoid multiple queries for HtmlCacheUpdater::getUrls() call
        $lb = $services->getLinkBatchFactory()->newLinkBatch();
        foreach ( $this->pageTuples as list( $page, $delay ) ) {
            $lb->addObj( $page );
        }
        $lb->execute();
 
        $reboundDelayByUrl = [];
 
        // Resolve the titles into CDN URLs
        $htmlCacheUpdater = $services->getHtmlCacheUpdater();
        foreach ( $this->pageTuples as list( $page, $delay ) ) {
            foreach ( $htmlCacheUpdater->getUrls( $page ) as $url ) {
                // Use the highest rebound for duplicate URLs in order to handle the most lag
                $reboundDelayByUrl[$url] = max( $reboundDelayByUrl[$url] ?? 0, $delay );
            }
        }
 
        foreach ( $this->urlTuples as list( $url, $delay ) ) {
            // Use the highest rebound for duplicate URLs in order to handle the most lag
            $reboundDelayByUrl[$url] = max( $reboundDelayByUrl[$url] ?? 0, $delay );
        }
 
        return $reboundDelayByUrl;
    }
 
    private static function HTCPPurge( array $urls ) {
        $htcpRouting = MediaWikiServices::getInstance()->getMainConfig()->get( 'HTCPRouting' );
        $htcpMulticastTTL = MediaWikiServices::getInstance()->getMainConfig()->get( 'HTCPMulticastTTL' );
        // HTCP CLR operation
        $htcpOpCLR = 4;
 
        // @todo FIXME: PHP doesn't support these socket constants (include/linux/in.h)
        if ( !defined( "IPPROTO_IP" ) ) {
            define( "IPPROTO_IP", 0 );
            define( "IP_MULTICAST_LOOP", 34 );
            define( "IP_MULTICAST_TTL", 33 );
        }
 
        // pfsockopen doesn't work because we need set_sock_opt
        $conn = socket_create( AF_INET, SOCK_DGRAM, SOL_UDP );
        if ( !$conn ) {
            $errstr = socket_strerror( socket_last_error() );
            wfDebugLog( 'squid', __METHOD__ .
                ": Error opening UDP socket: $errstr" );
 
            return;
        }
 
        // Set socket options
        socket_set_option( $conn, IPPROTO_IP, IP_MULTICAST_LOOP, 0 );
        if ( $htcpMulticastTTL != 1 ) {
            // Set multicast time to live (hop count) option on socket
            socket_set_option( $conn, IPPROTO_IP, IP_MULTICAST_TTL,
                $htcpMulticastTTL );
        }
 
        // Get sequential trx IDs for packet loss counting
        $idGenerator = MediaWikiServices::getInstance()->getGlobalIdGenerator();
        $ids = $idGenerator->newSequentialPerNodeIDs(
            'squidhtcppurge',
            32,
            count( $urls )
        );
 
        foreach ( $urls as $url ) {
            if ( !is_string( $url ) ) {
                throw new MWException( 'Bad purge URL' );
            }
            $url = self::expand( $url );
            $conf = self::getRuleForURL( $url, $htcpRouting );
            if ( !$conf ) {
                wfDebugLog( 'squid', __METHOD__ .
                    "No HTCP rule configured for URL {$url} , skipping" );
                continue;
            }
 
            if ( isset( $conf['host'] ) && isset( $conf['port'] ) ) {
                // Normalize single entries
                $conf = [ $conf ];
            }
            foreach ( $conf as $subconf ) {
                if ( !isset( $subconf['host'] ) || !isset( $subconf['port'] ) ) {
                    throw new MWException( "Invalid HTCP rule for URL $url\n" );
                }
            }
 
            // Construct a minimal HTCP request diagram
            // as per RFC 2756
            // Opcode 'CLR', no response desired, no auth
            $htcpTransID = current( $ids );
            next( $ids );
 
            $htcpSpecifier = pack( 'na4na*na8n',
                4, 'HEAD', strlen( $url ), $url,
                8, 'HTTP/1.0', 0 );
 
            $htcpDataLen = 8 + 2 + strlen( $htcpSpecifier );
            $htcpLen = 4 + $htcpDataLen + 2;
 
            // Note! Squid gets the bit order of the first
            // word wrong, wrt the RFC. Apparently no other
            // implementation exists, so adapt to Squid
            $htcpPacket = pack( 'nxxnCxNxxa*n',
                $htcpLen, $htcpDataLen, $htcpOpCLR,
                $htcpTransID, $htcpSpecifier, 2 );
 
            wfDebugLog( 'squid', __METHOD__ .
                "Purging URL $url via HTCP" );
            foreach ( $conf as $subconf ) {
                socket_sendto( $conn, $htcpPacket, $htcpLen, 0,
                    $subconf['host'], $subconf['port'] );
            }
        }
    }
 
    private static function naivePurge( array $urls ) {
        $cdnServers = MediaWikiServices::getInstance()->getMainConfig()->get( 'CdnServers' );
 
        $reqs = [];
        foreach ( $urls as $url ) {
            $url = self::expand( $url );
            $urlInfo = wfParseUrl( $url );
            $urlHost = strlen( $urlInfo['port'] ?? null )
                ? IPUtils::combineHostAndPort( $urlInfo['host'], $urlInfo['port'] )
                : $urlInfo['host'];
            $baseReq = [
                'method' => 'PURGE',
                'url' => $url,
                'headers' => [
                    'Host' => $urlHost,
                    'Connection' => 'Keep-Alive',
                    'Proxy-Connection' => 'Keep-Alive',
                    'User-Agent' => 'MediaWiki/' . MW_VERSION . ' ' . __CLASS__
                ]
            ];
            foreach ( $cdnServers as $server ) {
                $reqs[] = ( $baseReq + [ 'proxy' => $server ] );
            }
        }
 
        $http = MediaWikiServices::getInstance()->getHttpRequestFactory()
            ->createMultiClient( [ 'maxConnsPerHost' => 8, 'usePipelining' => true ] );
        $http->runMulti( $reqs );
    }
 
    private static function expand( $url ) {
        return wfExpandUrl( $url, PROTO_INTERNAL );
    }
 
    private static function getRuleForURL( $url, $rules ) {
        foreach ( $rules as $regex => $routing ) {
            if ( $regex === '' || preg_match( $regex, $url ) ) {
                return $routing;
            }
        }
 
        return false;
    }
}