SqlBagOStuff.php

Go to the documentation of this file.

<?php
use \MediaWiki\MediaWikiServices;
use \Wikimedia\WaitConditionLoop;

class SqlBagOStuff extends BagOStuff {
    protected $serverInfos;
    protected $serverTags;
    protected $numServers;
    protected $lastExpireAll = 0;
    protected $purgePeriod = 100;
    protected $shards = 1;
    protected $tableName = 'objectcache';
    protected $replicaOnly = false;
    protected $syncTimeout = 3;

    protected $separateMainLB;
    protected $conns;
    protected $connFailureTimes = [];
    protected $connFailureErrors = [];

    public function __construct( $params ) {
        parent::__construct( $params );

        $this->attrMap[self::ATTR_EMULATION] = self::QOS_EMULATION_SQL;
        $this->attrMap[self::ATTR_SYNCWRITES] = self::QOS_SYNCWRITES_NONE;

        if ( isset( $params['servers'] ) ) {
            $this->serverInfos = [];
            $this->serverTags = [];
            $this->numServers = count( $params['servers'] );
            $index = 0;
            foreach ( $params['servers'] as $tag => $info ) {
                $this->serverInfos[$index] = $info;
                if ( is_string( $tag ) ) {
                    $this->serverTags[$index] = $tag;
                } else {
                    $this->serverTags[$index] = isset( $info['host'] ) ? $info['host'] : "#$index";
                }
                ++$index;
            }
        } elseif ( isset( $params['server'] ) ) {
            $this->serverInfos = [ $params['server'] ];
            $this->numServers = count( $this->serverInfos );
        } else {
            // Default to using the main wiki's database servers
            $this->serverInfos = false;
            $this->numServers = 1;
            $this->attrMap[self::ATTR_SYNCWRITES] = self::QOS_SYNCWRITES_BE;
        }
        if ( isset( $params['purgePeriod'] ) ) {
            $this->purgePeriod = intval( $params['purgePeriod'] );
        }
        if ( isset( $params['tableName'] ) ) {
            $this->tableName = $params['tableName'];
        }
        if ( isset( $params['shards'] ) ) {
            $this->shards = intval( $params['shards'] );
        }
        if ( isset( $params['syncTimeout'] ) ) {
            $this->syncTimeout = $params['syncTimeout'];
        }
        $this->replicaOnly = !empty( $params['slaveOnly'] );
    }

    protected function getSeparateMainLB() {
        global $wgDBtype;

        if ( $wgDBtype === 'mysql' && $this->usesMainDB() ) {
            if ( !$this->separateMainLB ) {
                // We must keep a separate connection to MySQL in order to avoid deadlocks
                $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
                $this->separateMainLB = $lbFactory->newMainLB();
            }
            return $this->separateMainLB;
        } else {
            // However, SQLite has an opposite behavior. And PostgreSQL needs to know
            // if we are in transaction or not (@TODO: find some PostgreSQL work-around).
            return null;
        }
    }

    protected function getDB( $serverIndex ) {
        if ( !isset( $this->conns[$serverIndex] ) ) {
            if ( $serverIndex >= $this->numServers ) {
                throw new MWException( __METHOD__ . ": Invalid server index \"$serverIndex\"" );
            }

            # Don't keep timing out trying to connect for each call if the DB is down
            if ( isset( $this->connFailureErrors[$serverIndex] )
                && ( time() - $this->connFailureTimes[$serverIndex] ) < 60
            ) {
                throw $this->connFailureErrors[$serverIndex];
            }

            # If server connection info was given, use that
            if ( $this->serverInfos ) {
                $info = $this->serverInfos[$serverIndex];
                $type = isset( $info['type'] ) ? $info['type'] : 'mysql';
                $host = isset( $info['host'] ) ? $info['host'] : '[unknown]';
                $this->logger->debug( __CLASS__ . ": connecting to $host" );
                // Use a blank trx profiler to ignore expections as this is a cache
                $info['trxProfiler'] = new TransactionProfiler();
                $db = Database::factory( $type, $info );
                $db->clearFlag( DBO_TRX );
            } else {
                $index = $this->replicaOnly ? DB_REPLICA : DB_MASTER;
                if ( $this->getSeparateMainLB() ) {
                    $db = $this->getSeparateMainLB()->getConnection( $index );
                    $db->clearFlag( DBO_TRX ); // auto-commit mode
                } else {
                    $db = wfGetDB( $index );
                    // Can't mess with transaction rounds (DBO_TRX) :(
                }
            }
            $this->logger->debug( sprintf( "Connection %s will be used for SqlBagOStuff", $db ) );
            $this->conns[$serverIndex] = $db;
        }

        return $this->conns[$serverIndex];
    }

    protected function getTableByKey( $key ) {
        if ( $this->shards > 1 ) {
            $hash = hexdec( substr( md5( $key ), 0, 8 ) ) & 0x7fffffff;
            $tableIndex = $hash % $this->shards;
        } else {
            $tableIndex = 0;
        }
        if ( $this->numServers > 1 ) {
            $sortedServers = $this->serverTags;
            ArrayUtils::consistentHashSort( $sortedServers, $key );
            reset( $sortedServers );
            $serverIndex = key( $sortedServers );
        } else {
            $serverIndex = 0;
        }
        return [ $serverIndex, $this->getTableNameByShard( $tableIndex ) ];
    }

    protected function getTableNameByShard( $index ) {
        if ( $this->shards > 1 ) {
            $decimals = strlen( $this->shards - 1 );
            return $this->tableName .
                sprintf( "%0{$decimals}d", $index );
        } else {
            return $this->tableName;
        }
    }

    protected function doGet( $key, $flags = 0 ) {
        $casToken = null;

        return $this->getWithToken( $key, $casToken, $flags );
    }

    protected function getWithToken( $key, &$casToken, $flags = 0 ) {
        $values = $this->getMulti( [ $key ] );
        if ( array_key_exists( $key, $values ) ) {
            $casToken = $values[$key];
            return $values[$key];
        }
        return false;
    }

    public function getMulti( array $keys, $flags = 0 ) {
        $values = []; // array of (key => value)

        $keysByTable = [];
        foreach ( $keys as $key ) {
            list( $serverIndex, $tableName ) = $this->getTableByKey( $key );
            $keysByTable[$serverIndex][$tableName][] = $key;
        }

        $this->garbageCollect(); // expire old entries if any

        $dataRows = [];
        foreach ( $keysByTable as $serverIndex => $serverKeys ) {
            try {
                $db = $this->getDB( $serverIndex );
                foreach ( $serverKeys as $tableName => $tableKeys ) {
                    $res = $db->select( $tableName,
                        [ 'keyname', 'value', 'exptime' ],
                        [ 'keyname' => $tableKeys ],
                        __METHOD__,
                        // Approximate write-on-the-fly BagOStuff API via blocking.
                        // This approximation fails if a ROLLBACK happens (which is rare).
                        // We do not want to flush the TRX as that can break callers.
                        $db->trxLevel() ? [ 'LOCK IN SHARE MODE' ] : []
                    );
                    if ( $res === false ) {
                        continue;
                    }
                    foreach ( $res as $row ) {
                        $row->serverIndex = $serverIndex;
                        $row->tableName = $tableName;
                        $dataRows[$row->keyname] = $row;
                    }
                }
            } catch ( DBError $e ) {
                $this->handleReadError( $e, $serverIndex );
            }
        }

        foreach ( $keys as $key ) {
            if ( isset( $dataRows[$key] ) ) { // HIT?
                $row = $dataRows[$key];
                $this->debug( "get: retrieved data; expiry time is " . $row->exptime );
                $db = null;
                try {
                    $db = $this->getDB( $row->serverIndex );
                    if ( $this->isExpired( $db, $row->exptime ) ) { // MISS
                        $this->debug( "get: key has expired" );
                    } else { // HIT
                        $values[$key] = $this->unserialize( $db->decodeBlob( $row->value ) );
                    }
                } catch ( DBQueryError $e ) {
                    $this->handleWriteError( $e, $db, $row->serverIndex );
                }
            } else { // MISS
                $this->debug( 'get: no matching rows' );
            }
        }

        return $values;
    }

    public function setMulti( array $data, $expiry = 0 ) {
        $keysByTable = [];
        foreach ( $data as $key => $value ) {
            list( $serverIndex, $tableName ) = $this->getTableByKey( $key );
            $keysByTable[$serverIndex][$tableName][] = $key;
        }

        $this->garbageCollect(); // expire old entries if any

        $result = true;
        $exptime = (int)$expiry;
        foreach ( $keysByTable as $serverIndex => $serverKeys ) {
            $db = null;
            try {
                $db = $this->getDB( $serverIndex );
            } catch ( DBError $e ) {
                $this->handleWriteError( $e, $db, $serverIndex );
                $result = false;
                continue;
            }

            if ( $exptime < 0 ) {
                $exptime = 0;
            }

            if ( $exptime == 0 ) {
                $encExpiry = $this->getMaxDateTime( $db );
            } else {
                $exptime = $this->convertExpiry( $exptime );
                $encExpiry = $db->timestamp( $exptime );
            }
            foreach ( $serverKeys as $tableName => $tableKeys ) {
                $rows = [];
                foreach ( $tableKeys as $key ) {
                    $rows[] = [
                        'keyname' => $key,
                        'value' => $db->encodeBlob( $this->serialize( $data[$key] ) ),
                        'exptime' => $encExpiry,
                    ];
                }

                try {
                    $db->replace(
                        $tableName,
                        [ 'keyname' ],
                        $rows,
                        __METHOD__
                    );
                } catch ( DBError $e ) {
                    $this->handleWriteError( $e, $db, $serverIndex );
                    $result = false;
                }

            }

        }

        return $result;
    }

    public function set( $key, $value, $exptime = 0, $flags = 0 ) {
        $ok = $this->setMulti( [ $key => $value ], $exptime );
        if ( ( $flags & self::WRITE_SYNC ) == self::WRITE_SYNC ) {
            $ok = $this->waitForReplication() && $ok;
        }

        return $ok;
    }

    protected function cas( $casToken, $key, $value, $exptime = 0 ) {
        list( $serverIndex, $tableName ) = $this->getTableByKey( $key );
        $db = null;
        try {
            $db = $this->getDB( $serverIndex );
            $exptime = intval( $exptime );

            if ( $exptime < 0 ) {
                $exptime = 0;
            }

            if ( $exptime == 0 ) {
                $encExpiry = $this->getMaxDateTime( $db );
            } else {
                $exptime = $this->convertExpiry( $exptime );
                $encExpiry = $db->timestamp( $exptime );
            }
            // (bug 24425) use a replace if the db supports it instead of
            // delete/insert to avoid clashes with conflicting keynames
            $db->update(
                $tableName,
                [
                    'keyname' => $key,
                    'value' => $db->encodeBlob( $this->serialize( $value ) ),
                    'exptime' => $encExpiry
                ],
                [
                    'keyname' => $key,
                    'value' => $db->encodeBlob( $this->serialize( $casToken ) )
                ],
                __METHOD__
            );
        } catch ( DBQueryError $e ) {
            $this->handleWriteError( $e, $db, $serverIndex );

            return false;
        }

        return (bool)$db->affectedRows();
    }

    public function delete( $key ) {
        list( $serverIndex, $tableName ) = $this->getTableByKey( $key );
        $db = null;
        try {
            $db = $this->getDB( $serverIndex );
            $db->delete(
                $tableName,
                [ 'keyname' => $key ],
                __METHOD__ );
        } catch ( DBError $e ) {
            $this->handleWriteError( $e, $db, $serverIndex );
            return false;
        }

        return true;
    }

    public function incr( $key, $step = 1 ) {
        list( $serverIndex, $tableName ) = $this->getTableByKey( $key );
        $db = null;
        try {
            $db = $this->getDB( $serverIndex );
            $step = intval( $step );
            $row = $db->selectRow(
                $tableName,
                [ 'value', 'exptime' ],
                [ 'keyname' => $key ],
                __METHOD__,
                [ 'FOR UPDATE' ] );
            if ( $row === false ) {
                // Missing

                return null;
            }
            $db->delete( $tableName, [ 'keyname' => $key ], __METHOD__ );
            if ( $this->isExpired( $db, $row->exptime ) ) {
                // Expired, do not reinsert

                return null;
            }

            $oldValue = intval( $this->unserialize( $db->decodeBlob( $row->value ) ) );
            $newValue = $oldValue + $step;
            $db->insert( $tableName,
                [
                    'keyname' => $key,
                    'value' => $db->encodeBlob( $this->serialize( $newValue ) ),
                    'exptime' => $row->exptime
                ], __METHOD__, 'IGNORE' );

            if ( $db->affectedRows() == 0 ) {
                // Race condition. See bug 28611
                $newValue = null;
            }
        } catch ( DBError $e ) {
            $this->handleWriteError( $e, $db, $serverIndex );
            return null;
        }

        return $newValue;
    }

    public function merge( $key, callable $callback, $exptime = 0, $attempts = 10, $flags = 0 ) {
        $ok = $this->mergeViaCas( $key, $callback, $exptime, $attempts );
        if ( ( $flags & self::WRITE_SYNC ) == self::WRITE_SYNC ) {
            $ok = $this->waitForReplication() && $ok;
        }

        return $ok;
    }

    public function changeTTL( $key, $expiry = 0 ) {
        list( $serverIndex, $tableName ) = $this->getTableByKey( $key );
        $db = null;
        try {
            $db = $this->getDB( $serverIndex );
            $db->update(
                $tableName,
                [ 'exptime' => $db->timestamp( $this->convertExpiry( $expiry ) ) ],
                [ 'keyname' => $key, 'exptime > ' . $db->addQuotes( $db->timestamp( time() ) ) ],
                __METHOD__
            );
            if ( $db->affectedRows() == 0 ) {
                return false;
            }
        } catch ( DBError $e ) {
            $this->handleWriteError( $e, $db, $serverIndex );
            return false;
        }

        return true;
    }

    protected function isExpired( $db, $exptime ) {
        return $exptime != $this->getMaxDateTime( $db ) && wfTimestamp( TS_UNIX, $exptime ) < time();
    }

    protected function getMaxDateTime( $db ) {
        if ( time() > 0x7fffffff ) {
            return $db->timestamp( 1 << 62 );
        } else {
            return $db->timestamp( 0x7fffffff );
        }
    }

    protected function garbageCollect() {
        if ( !$this->purgePeriod || $this->replicaOnly ) {
            // Disabled
            return;
        }
        // Only purge on one in every $this->purgePeriod requests.
        if ( $this->purgePeriod !== 1 && mt_rand( 0, $this->purgePeriod - 1 ) ) {
            return;
        }
        $now = time();
        // Avoid repeating the delete within a few seconds
        if ( $now > ( $this->lastExpireAll + 1 ) ) {
            $this->lastExpireAll = $now;
            $this->expireAll();
        }
    }

    public function expireAll() {
        $this->deleteObjectsExpiringBefore( wfTimestampNow() );
    }

    public function deleteObjectsExpiringBefore( $timestamp, $progressCallback = false ) {
        for ( $serverIndex = 0; $serverIndex < $this->numServers; $serverIndex++ ) {
            $db = null;
            try {
                $db = $this->getDB( $serverIndex );
                $dbTimestamp = $db->timestamp( $timestamp );
                $totalSeconds = false;
                $baseConds = [ 'exptime < ' . $db->addQuotes( $dbTimestamp ) ];
                for ( $i = 0; $i < $this->shards; $i++ ) {
                    $maxExpTime = false;
                    while ( true ) {
                        $conds = $baseConds;
                        if ( $maxExpTime !== false ) {
                            $conds[] = 'exptime > ' . $db->addQuotes( $maxExpTime );
                        }
                        $rows = $db->select(
                            $this->getTableNameByShard( $i ),
                            [ 'keyname', 'exptime' ],
                            $conds,
                            __METHOD__,
                            [ 'LIMIT' => 100, 'ORDER BY' => 'exptime' ] );
                        if ( $rows === false || !$rows->numRows() ) {
                            break;
                        }
                        $keys = [];
                        $row = $rows->current();
                        $minExpTime = $row->exptime;
                        if ( $totalSeconds === false ) {
                            $totalSeconds = wfTimestamp( TS_UNIX, $timestamp )
                                - wfTimestamp( TS_UNIX, $minExpTime );
                        }
                        foreach ( $rows as $row ) {
                            $keys[] = $row->keyname;
                            $maxExpTime = $row->exptime;
                        }

                        $db->delete(
                            $this->getTableNameByShard( $i ),
                            [
                                'exptime >= ' . $db->addQuotes( $minExpTime ),
                                'exptime < ' . $db->addQuotes( $dbTimestamp ),
                                'keyname' => $keys
                            ],
                            __METHOD__ );

                        if ( $progressCallback ) {
                            if ( intval( $totalSeconds ) === 0 ) {
                                $percent = 0;
                            } else {
                                $remainingSeconds = wfTimestamp( TS_UNIX, $timestamp )
                                    - wfTimestamp( TS_UNIX, $maxExpTime );
                                if ( $remainingSeconds > $totalSeconds ) {
                                    $totalSeconds = $remainingSeconds;
                                }
                                $processedSeconds = $totalSeconds - $remainingSeconds;
                                $percent = ( $i + $processedSeconds / $totalSeconds )
                                    / $this->shards * 100;
                            }
                            $percent = ( $percent / $this->numServers )
                                + ( $serverIndex / $this->numServers * 100 );
                            call_user_func( $progressCallback, $percent );
                        }
                    }
                }
            } catch ( DBError $e ) {
                $this->handleWriteError( $e, $db, $serverIndex );
                return false;
            }
        }
        return true;
    }

    public function deleteAll() {
        for ( $serverIndex = 0; $serverIndex < $this->numServers; $serverIndex++ ) {
            $db = null;
            try {
                $db = $this->getDB( $serverIndex );
                for ( $i = 0; $i < $this->shards; $i++ ) {
                    $db->delete( $this->getTableNameByShard( $i ), '*', __METHOD__ );
                }
            } catch ( DBError $e ) {
                $this->handleWriteError( $e, $db, $serverIndex );
                return false;
            }
        }
        return true;
    }

    protected function serialize( &$data ) {
        $serial = serialize( $data );

        if ( function_exists( 'gzdeflate' ) ) {
            return gzdeflate( $serial );
        } else {
            return $serial;
        }
    }

    protected function unserialize( $serial ) {
        if ( function_exists( 'gzinflate' ) ) {
            MediaWiki\suppressWarnings();
            $decomp = gzinflate( $serial );
            MediaWiki\restoreWarnings();

            if ( false !== $decomp ) {
                $serial = $decomp;
            }
        }

        $ret = unserialize( $serial );

        return $ret;
    }

    protected function handleReadError( DBError $exception, $serverIndex ) {
        if ( $exception instanceof DBConnectionError ) {
            $this->markServerDown( $exception, $serverIndex );
        }
        $this->logger->error( "DBError: {$exception->getMessage()}" );
        if ( $exception instanceof DBConnectionError ) {
            $this->setLastError( BagOStuff::ERR_UNREACHABLE );
            $this->logger->debug( __METHOD__ . ": ignoring connection error" );
        } else {
            $this->setLastError( BagOStuff::ERR_UNEXPECTED );
            $this->logger->debug( __METHOD__ . ": ignoring query error" );
        }
    }

    protected function handleWriteError( DBError $exception, IDatabase $db = null, $serverIndex ) {
        if ( !$db ) {
            $this->markServerDown( $exception, $serverIndex );
        } elseif ( $db->wasReadOnlyError() ) {
            if ( $db->trxLevel() && $this->usesMainDB() ) {
                // Errors like deadlocks and connection drops already cause rollback.
                // For consistency, we have no choice but to throw an error and trigger
                // complete rollback if the main DB is also being used as the cache DB.
                throw $exception;
            }
        }

        $this->logger->error( "DBError: {$exception->getMessage()}" );
        if ( $exception instanceof DBConnectionError ) {
            $this->setLastError( BagOStuff::ERR_UNREACHABLE );
            $this->logger->debug( __METHOD__ . ": ignoring connection error" );
        } else {
            $this->setLastError( BagOStuff::ERR_UNEXPECTED );
            $this->logger->debug( __METHOD__ . ": ignoring query error" );
        }
    }

    protected function markServerDown( DBError $exception, $serverIndex ) {
        unset( $this->conns[$serverIndex] ); // bug T103435

        if ( isset( $this->connFailureTimes[$serverIndex] ) ) {
            if ( time() - $this->connFailureTimes[$serverIndex] >= 60 ) {
                unset( $this->connFailureTimes[$serverIndex] );
                unset( $this->connFailureErrors[$serverIndex] );
            } else {
                $this->logger->debug( __METHOD__ . ": Server #$serverIndex already down" );
                return;
            }
        }
        $now = time();
        $this->logger->info( __METHOD__ . ": Server #$serverIndex down until " . ( $now + 60 ) );
        $this->connFailureTimes[$serverIndex] = $now;
        $this->connFailureErrors[$serverIndex] = $exception;
    }

    public function createTables() {
        for ( $serverIndex = 0; $serverIndex < $this->numServers; $serverIndex++ ) {
            $db = $this->getDB( $serverIndex );
            if ( $db->getType() !== 'mysql' ) {
                throw new MWException( __METHOD__ . ' is not supported on this DB server' );
            }

            for ( $i = 0; $i < $this->shards; $i++ ) {
                $db->query(
                    'CREATE TABLE ' . $db->tableName( $this->getTableNameByShard( $i ) ) .
                    ' LIKE ' . $db->tableName( 'objectcache' ),
                    __METHOD__ );
            }
        }
    }

    protected function usesMainDB() {
        return !$this->serverInfos;
    }

    protected function waitForReplication() {
        if ( !$this->usesMainDB() ) {
            // Custom DB server list; probably doesn't use replication
            return true;
        }

        $lb = $this->getSeparateMainLB()
            ?: MediaWikiServices::getInstance()->getDBLoadBalancer();

        if ( $lb->getServerCount() <= 1 ) {
            return true; // no replica DBs
        }

        // Main LB is used; wait for any replica DBs to catch up
        $masterPos = $lb->getMasterPos();

        $loop = new WaitConditionLoop(
            function () use ( $lb, $masterPos ) {
                return $lb->waitForAll( $masterPos, 1 );
            },
            $this->syncTimeout,
            $this->busyCallbacks
        );

        return ( $loop->invoke() === $loop::CONDITION_REACHED );
    }
}