FileBackendMultiWrite.php

Go to the documentation of this file.

<?php
class FileBackendMultiWrite extends FileBackend {
    protected $backends = array(); // array of (backend index => backends)
    protected $masterIndex = -1; // integer; index of master backend
    protected $syncChecks = 0; // integer; bitfield
    protected $autoResync = false; // boolean

    protected $noPushDirConts = array();
    protected $noPushQuickOps = false; // boolean

    /* Possible internal backend consistency checks */
    const CHECK_SIZE = 1;
    const CHECK_TIME = 2;
    const CHECK_SHA1 = 4;

    public function __construct( array $config ) {
        parent::__construct( $config );
        $this->syncChecks = isset( $config['syncChecks'] )
            ? $config['syncChecks']
            : self::CHECK_SIZE;
        $this->autoResync = isset( $config['autoResync'] )
            ? $config['autoResync']
            : false;
        $this->noPushQuickOps = isset( $config['noPushQuickOps'] )
            ? $config['noPushQuickOps']
            : false;
        $this->noPushDirConts = isset( $config['noPushDirConts'] )
            ? $config['noPushDirConts']
            : array();
        // Construct backends here rather than via registration
        // to keep these backends hidden from outside the proxy.
        $namesUsed = array();
        foreach ( $config['backends'] as $index => $config ) {
            if ( isset( $config['template'] ) ) {
                // Config is just a modified version of a registered backend's.
                // This should only be used when that config is used only by this backend.
                $config = $config + FileBackendGroup::singleton()->config( $config['template'] );
            }
            $name = $config['name'];
            if ( isset( $namesUsed[$name] ) ) { // don't break FileOp predicates
                throw new MWException( "Two or more backends defined with the name $name." );
            }
            $namesUsed[$name] = 1;
            // Alter certain sub-backend settings for sanity
            unset( $config['readOnly'] ); // use proxy backend setting
            unset( $config['fileJournal'] ); // use proxy backend journal
            $config['wikiId'] = $this->wikiId; // use the proxy backend wiki ID
            $config['lockManager'] = 'nullLockManager'; // lock under proxy backend
            if ( !empty( $config['isMultiMaster'] ) ) {
                if ( $this->masterIndex >= 0 ) {
                    throw new MWException( 'More than one master backend defined.' );
                }
                $this->masterIndex = $index; // this is the "master"
                $config['fileJournal'] = $this->fileJournal; // log under proxy backend
            }
            // Create sub-backend object
            if ( !isset( $config['class'] ) ) {
                throw new MWException( 'No class given for a backend config.' );
            }
            $class = $config['class'];
            $this->backends[$index] = new $class( $config );
        }
        if ( $this->masterIndex < 0 ) { // need backends and must have a master
            throw new MWException( 'No master backend defined.' );
        }
    }

    final protected function doOperationsInternal( array $ops, array $opts ) {
        $status = Status::newGood();

        $mbe = $this->backends[$this->masterIndex]; // convenience

        // Try to lock those files for the scope of this function...
        if ( empty( $opts['nonLocking'] ) ) {
            // Try to lock those files for the scope of this function...
            $scopeLock = $this->getScopedLocksForOps( $ops, $status );
            if ( !$status->isOK() ) {
                return $status; // abort
            }
        }
        // Clear any cache entries (after locks acquired)
        $this->clearCache();
        $opts['preserveCache'] = true; // only locked files are cached
        // Get the list of paths to read/write...
        $relevantPaths = $this->fileStoragePathsForOps( $ops );
        // Check if the paths are valid and accessible on all backends...
        $status->merge( $this->accessibilityCheck( $relevantPaths ) );
        if ( !$status->isOK() ) {
            return $status; // abort
        }
        // Do a consistency check to see if the backends are consistent...
        $syncStatus = $this->consistencyCheck( $relevantPaths );
        if ( !$syncStatus->isOK() ) {
            wfDebugLog( 'FileOperation', get_class( $this ) .
                " failed sync check: " . FormatJson::encode( $relevantPaths ) );
            // Try to resync the clone backends to the master on the spot...
            if ( !$this->autoResync || !$this->resyncFiles( $relevantPaths )->isOK() ) {
                $status->merge( $syncStatus );
                return $status; // abort
            }
        }
        // Actually attempt the operation batch on the master backend...
        $realOps = $this->substOpBatchPaths( $ops, $mbe );
        $masterStatus = $mbe->doOperations( $realOps, $opts );
        $status->merge( $masterStatus );
        // Propagate the operations to the clone backends if there were no unexpected errors
        // and if there were either no expected errors or if the 'force' option was used.
        // However, if nothing succeeded at all, then don't replicate any of the operations.
        // If $ops only had one operation, this might avoid backend sync inconsistencies.
        if ( $masterStatus->isOK() && $masterStatus->successCount > 0 ) {
            foreach ( $this->backends as $index => $backend ) {
                if ( $index !== $this->masterIndex ) { // not done already
                    $realOps = $this->substOpBatchPaths( $ops, $backend );
                    $status->merge( $backend->doOperations( $realOps, $opts ) );
                }
            }
        }
        // Make 'success', 'successCount', and 'failCount' fields reflect
        // the overall operation, rather than all the batches for each backend.
        // Do this by only using success values from the master backend's batch.
        $status->success = $masterStatus->success;
        $status->successCount = $masterStatus->successCount;
        $status->failCount = $masterStatus->failCount;

        return $status;
    }

    public function consistencyCheck( array $paths ) {
        $status = Status::newGood();
        if ( $this->syncChecks == 0 || count( $this->backends ) <= 1 ) {
            return $status; // skip checks
        }

        $mBackend = $this->backends[$this->masterIndex];
        foreach ( $paths as $path ) {
            $params = array( 'src' => $path, 'latest' => true );
            $mParams = $this->substOpPaths( $params, $mBackend );
            // Stat the file on the 'master' backend
            $mStat = $mBackend->getFileStat( $mParams );
            if ( $this->syncChecks & self::CHECK_SHA1 ) {
                $mSha1 = $mBackend->getFileSha1Base36( $mParams );
            } else {
                $mSha1 = false;
            }
            // Check if all clone backends agree with the master...
            foreach ( $this->backends as $index => $cBackend ) {
                if ( $index === $this->masterIndex ) {
                    continue; // master
                }
                $cParams = $this->substOpPaths( $params, $cBackend );
                $cStat = $cBackend->getFileStat( $cParams );
                if ( $mStat ) { // file is in master
                    if ( !$cStat ) { // file should exist
                        $status->fatal( 'backend-fail-synced', $path );
                        continue;
                    }
                    if ( $this->syncChecks & self::CHECK_SIZE ) {
                        if ( $cStat['size'] != $mStat['size'] ) { // wrong size
                            $status->fatal( 'backend-fail-synced', $path );
                            continue;
                        }
                    }
                    if ( $this->syncChecks & self::CHECK_TIME ) {
                        $mTs = wfTimestamp( TS_UNIX, $mStat['mtime'] );
                        $cTs = wfTimestamp( TS_UNIX, $cStat['mtime'] );
                        if ( abs( $mTs - $cTs ) > 30 ) { // outdated file somewhere
                            $status->fatal( 'backend-fail-synced', $path );
                            continue;
                        }
                    }
                    if ( $this->syncChecks & self::CHECK_SHA1 ) {
                        if ( $cBackend->getFileSha1Base36( $cParams ) !== $mSha1 ) { // wrong SHA1
                            $status->fatal( 'backend-fail-synced', $path );
                            continue;
                        }
                    }
                } else { // file is not in master
                    if ( $cStat ) { // file should not exist
                        $status->fatal( 'backend-fail-synced', $path );
                    }
                }
            }
        }

        return $status;
    }

    public function accessibilityCheck( array $paths ) {
        $status = Status::newGood();
        if ( count( $this->backends ) <= 1 ) {
            return $status; // skip checks
        }

        foreach ( $paths as $path ) {
            foreach ( $this->backends as $backend ) {
                $realPath = $this->substPaths( $path, $backend );
                if ( !$backend->isPathUsableInternal( $realPath ) ) {
                    $status->fatal( 'backend-fail-usable', $path );
                }
            }
        }

        return $status;
    }

    public function resyncFiles( array $paths ) {
        $status = Status::newGood();

        $mBackend = $this->backends[$this->masterIndex];
        foreach ( $paths as $path ) {
            $mPath = $this->substPaths( $path, $mBackend );
            $mSha1 = $mBackend->getFileSha1Base36( array( 'src' => $mPath, 'latest' => true ) );
            $mStat = $mBackend->getFileStat( array( 'src' => $mPath, 'latest' => true ) );
            if ( $mStat === null || ( $mSha1 !== false && !$mStat ) ) { // sanity
                $status->fatal( 'backend-fail-internal', $this->name );
                continue; // file is not available on the master backend...
            }
            // Check of all clone backends agree with the master...
            foreach ( $this->backends as $index => $cBackend ) {
                if ( $index === $this->masterIndex ) {
                    continue; // master
                }
                $cPath = $this->substPaths( $path, $cBackend );
                $cSha1 = $cBackend->getFileSha1Base36( array( 'src' => $cPath, 'latest' => true ) );
                $cStat = $cBackend->getFileStat( array( 'src' => $cPath, 'latest' => true ) );
                if ( $cStat === null || ( $cSha1 !== false && !$cStat ) ) { // sanity
                    $status->fatal( 'backend-fail-internal', $cBackend->getName() );
                    continue; // file is not available on the clone backend...
                }
                if ( $mSha1 === $cSha1 ) {
                    // already synced; nothing to do
                } elseif ( $mSha1 !== false ) { // file is in master
                    if ( $this->autoResync === 'conservative'
                        && $cStat && $cStat['mtime'] > $mStat['mtime'] )
                    {
                        $status->fatal( 'backend-fail-synced', $path );
                        continue; // don't rollback data
                    }
                    $fsFile = $mBackend->getLocalReference(
                        array( 'src' => $mPath, 'latest' => true ) );
                    $status->merge( $cBackend->quickStore(
                        array( 'src' => $fsFile->getPath(), 'dst' => $cPath )
                    ) );
                } elseif ( $mStat === false ) { // file is not in master
                    if ( $this->autoResync === 'conservative' ) {
                        $status->fatal( 'backend-fail-synced', $path );
                        continue; // don't delete data
                    }
                    $status->merge( $cBackend->quickDelete( array( 'src' => $cPath ) ) );
                }
            }
        }

        return $status;
    }

    protected function fileStoragePathsForOps( array $ops ) {
        $paths = array();
        foreach ( $ops as $op ) {
            if ( isset( $op['src'] ) ) {
                // For things like copy/move/delete with "ignoreMissingSource" and there
                // is no source file, nothing should happen and there should be no errors.
                if ( empty( $op['ignoreMissingSource'] )
                    || $this->fileExists( array( 'src' => $op['src'] ) ) )
                {
                    $paths[] = $op['src'];
                }
            }
            if ( isset( $op['srcs'] ) ) {
                $paths = array_merge( $paths, $op['srcs'] );
            }
            if ( isset( $op['dst'] ) ) {
                $paths[] = $op['dst'];
            }
        }
        return array_values( array_unique( array_filter( $paths, 'FileBackend::isStoragePath' ) ) );
    }

    protected function substOpBatchPaths( array $ops, FileBackendStore $backend ) {
        $newOps = array(); // operations
        foreach ( $ops as $op ) {
            $newOp = $op; // operation
            foreach ( array( 'src', 'srcs', 'dst', 'dir' ) as $par ) {
                if ( isset( $newOp[$par] ) ) { // string or array
                    $newOp[$par] = $this->substPaths( $newOp[$par], $backend );
                }
            }
            $newOps[] = $newOp;
        }
        return $newOps;
    }

    protected function substOpPaths( array $ops, FileBackendStore $backend ) {
        $newOps = $this->substOpBatchPaths( array( $ops ), $backend );
        return $newOps[0];
    }

    protected function substPaths( $paths, FileBackendStore $backend ) {
        return preg_replace(
            '!^mwstore://' . preg_quote( $this->name ) . '/!',
            StringUtils::escapeRegexReplacement( "mwstore://{$backend->getName()}/" ),
            $paths // string or array
        );
    }

    protected function unsubstPaths( $paths ) {
        return preg_replace(
            '!^mwstore://([^/]+)!',
            StringUtils::escapeRegexReplacement( "mwstore://{$this->name}" ),
            $paths // string or array
        );
    }

    protected function doQuickOperationsInternal( array $ops ) {
        $status = Status::newGood();
        // Do the operations on the master backend; setting Status fields...
        $realOps = $this->substOpBatchPaths( $ops, $this->backends[$this->masterIndex] );
        $masterStatus = $this->backends[$this->masterIndex]->doQuickOperations( $realOps );
        $status->merge( $masterStatus );
        // Propagate the operations to the clone backends...
        if ( !$this->noPushQuickOps ) {
            foreach ( $this->backends as $index => $backend ) {
                if ( $index !== $this->masterIndex ) { // not done already
                    $realOps = $this->substOpBatchPaths( $ops, $backend );
                    $status->merge( $backend->doQuickOperations( $realOps ) );
                }
            }
        }
        // Make 'success', 'successCount', and 'failCount' fields reflect
        // the overall operation, rather than all the batches for each backend.
        // Do this by only using success values from the master backend's batch.
        $status->success = $masterStatus->success;
        $status->successCount = $masterStatus->successCount;
        $status->failCount = $masterStatus->failCount;
        return $status;
    }

    protected function replicateContainerDirChanges( $path ) {
        list( , $shortCont, ) = self::splitStoragePath( $path );
        return !in_array( $shortCont, $this->noPushDirConts );
    }

    protected function doPrepare( array $params ) {
        $status = Status::newGood();
        $replicate = $this->replicateContainerDirChanges( $params['dir'] );
        foreach ( $this->backends as $index => $backend ) {
            if ( $replicate || $index == $this->masterIndex ) {
                $realParams = $this->substOpPaths( $params, $backend );
                $status->merge( $backend->doPrepare( $realParams ) );
            }
        }
        return $status;
    }

    protected function doSecure( array $params ) {
        $status = Status::newGood();
        $replicate = $this->replicateContainerDirChanges( $params['dir'] );
        foreach ( $this->backends as $index => $backend ) {
            if ( $replicate || $index == $this->masterIndex ) {
                $realParams = $this->substOpPaths( $params, $backend );
                $status->merge( $backend->doSecure( $realParams ) );
            }
        }
        return $status;
    }

    protected function doPublish( array $params ) {
        $status = Status::newGood();
        $replicate = $this->replicateContainerDirChanges( $params['dir'] );
        foreach ( $this->backends as $index => $backend ) {
            if ( $replicate || $index == $this->masterIndex ) {
                $realParams = $this->substOpPaths( $params, $backend );
                $status->merge( $backend->doPublish( $realParams ) );
            }
        }
        return $status;
    }

    protected function doClean( array $params ) {
        $status = Status::newGood();
        $replicate = $this->replicateContainerDirChanges( $params['dir'] );
        foreach ( $this->backends as $index => $backend ) {
            if ( $replicate || $index == $this->masterIndex ) {
                $realParams = $this->substOpPaths( $params, $backend );
                $status->merge( $backend->doClean( $realParams ) );
            }
        }
        return $status;
    }

    public function concatenate( array $params ) {
        // We are writing to an FS file, so we don't need to do this per-backend
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->concatenate( $realParams );
    }

    public function fileExists( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->fileExists( $realParams );
    }

    public function getFileTimestamp( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileTimestamp( $realParams );
    }

    public function getFileSize( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileSize( $realParams );
    }

    public function getFileStat( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileStat( $realParams );
    }

    public function getFileContentsMulti( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        $contentsM = $this->backends[$this->masterIndex]->getFileContentsMulti( $realParams );

        $contents = array(); // (path => FSFile) mapping using the proxy backend's name
        foreach ( $contentsM as $path => $data ) {
            $contents[$this->unsubstPaths( $path )] = $data;
        }
        return $contents;
    }

    public function getFileSha1Base36( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileSha1Base36( $realParams );
    }

    public function getFileProps( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileProps( $realParams );
    }

    public function streamFile( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->streamFile( $realParams );
    }

    public function getLocalReferenceMulti( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        $fsFilesM = $this->backends[$this->masterIndex]->getLocalReferenceMulti( $realParams );

        $fsFiles = array(); // (path => FSFile) mapping using the proxy backend's name
        foreach ( $fsFilesM as $path => $fsFile ) {
            $fsFiles[$this->unsubstPaths( $path )] = $fsFile;
        }
        return $fsFiles;
    }

    public function getLocalCopyMulti( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        $tempFilesM = $this->backends[$this->masterIndex]->getLocalCopyMulti( $realParams );

        $tempFiles = array(); // (path => TempFSFile) mapping using the proxy backend's name
        foreach ( $tempFilesM as $path => $tempFile ) {
            $tempFiles[$this->unsubstPaths( $path )] = $tempFile;
        }
        return $tempFiles;
    }

    public function getFileHttpUrl( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileHttpUrl( $realParams );
    }

    public function directoryExists( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->directoryExists( $realParams );
    }

    public function getDirectoryList( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getDirectoryList( $realParams );
    }

    public function getFileList( array $params ) {
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        return $this->backends[$this->masterIndex]->getFileList( $realParams );
    }

    public function clearCache( array $paths = null ) {
        foreach ( $this->backends as $backend ) {
            $realPaths = is_array( $paths ) ? $this->substPaths( $paths, $backend ) : null;
            $backend->clearCache( $realPaths );
        }
    }

    public function getScopedLocksForOps( array $ops, Status $status ) {
        $realOps = $this->substOpBatchPaths( $ops, $this->backends[$this->masterIndex] );
        $fileOps = $this->backends[$this->masterIndex]->getOperationsInternal( $realOps );
        // Get the paths to lock from the master backend
        $paths = $this->backends[$this->masterIndex]->getPathsToLockForOpsInternal( $fileOps );
        // Get the paths under the proxy backend's name
        $pbPaths = array(
            LockManager::LOCK_UW => $this->unsubstPaths( $paths[LockManager::LOCK_UW] ),
            LockManager::LOCK_EX => $this->unsubstPaths( $paths[LockManager::LOCK_EX] )
        );
        // Actually acquire the locks
        return array( $this->getScopedFileLocks( $pbPaths, 'mixed', $status ) );
    }
}