FileBackendMultiWrite.php

Go to the documentation of this file.

<?php
use Wikimedia\Timestamp\ConvertibleTimestamp;
 
class FileBackendMultiWrite extends FileBackend {
    protected $backends = [];
 
    protected $masterIndex = -1;
    protected $readIndex = -1;
 
    protected $syncChecks = 0;
    protected $autoResync = false;
 
    protected $asyncWrites = false;
 
    private const CHECK_SIZE = 1;
    private const CHECK_TIME = 2;
    private const CHECK_SHA1 = 4;
 
    public function __construct( array $config ) {
        parent::__construct( $config );
        $this->syncChecks = $config['syncChecks'] ?? self::CHECK_SIZE;
        $this->autoResync = $config['autoResync'] ?? false;
        $this->asyncWrites = isset( $config['replication'] ) && $config['replication'] === 'async';
        // Construct backends here rather than via registration
        // to keep these backends hidden from outside the proxy.
        $namesUsed = [];
        foreach ( $config['backends'] as $index => $config ) {
            $name = $config['name'];
            if ( isset( $namesUsed[$name] ) ) { // don't break FileOp predicates
                throw new LogicException( "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
            unset( $config['lockManager'] ); // lock under proxy backend
            $config['domainId'] = $this->domainId; // use the proxy backend wiki ID
            if ( !empty( $config['isMultiMaster'] ) ) {
                if ( $this->masterIndex >= 0 ) {
                    throw new LogicException( 'More than one master backend defined.' );
                }
                $this->masterIndex = $index; // this is the "master"
                $config['fileJournal'] = $this->fileJournal; // log under proxy backend
            }
            if ( !empty( $config['readAffinity'] ) ) {
                $this->readIndex = $index; // prefer this for reads
            }
            // Create sub-backend object
            if ( !isset( $config['class'] ) ) {
                throw new InvalidArgumentException( '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 LogicException( 'No master backend defined.' );
        }
        if ( $this->readIndex < 0 ) {
            $this->readIndex = $this->masterIndex; // default
        }
    }
 
    final protected function doOperationsInternal( array $ops, array $opts ) {
        $status = $this->newStatus();
 
        $mbe = $this->backends[$this->masterIndex]; // convenience
 
        // Acquire any locks as needed
        $scopeLock = null;
        if ( empty( $opts['nonLocking'] ) ) {
            $scopeLock = $this->getScopedLocksForOps( $ops, $status );
            if ( !$status->isOK() ) {
                return $status; // abort
            }
        }
        // Get the list of paths to read/write
        $relevantPaths = $this->fileStoragePathsForOps( $ops );
        // Clear any cache entries (after locks acquired)
        $this->clearCache( $relevantPaths );
        $opts['preserveCache'] = true; // only locked files are cached
        // 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() ) {
            $this->logger->error(
                __METHOD__ . ": failed sync check: " . FormatJson::encode( $relevantPaths )
            );
            // Try to resync the clone backends to the master on the spot
            if (
                $this->autoResync === false ||
                !$this->resyncFiles( $relevantPaths, $this->autoResync )->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 everything didn't fail due to predicted errors. 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 ) {
                    continue; // done already
                }
 
                $realOps = $this->substOpBatchPaths( $ops, $backend );
                if ( $this->asyncWrites && !$this->hasVolatileSources( $ops ) ) {
                    // Bind $scopeLock to the callback to preserve locks
                    DeferredUpdates::addCallableUpdate(
                        function () use ( $backend, $realOps, $opts, $scopeLock, $relevantPaths ) {
                            $this->logger->error(
                                "'{$backend->getName()}' async replication; paths: " .
                                FormatJson::encode( $relevantPaths )
                            );
                            $backend->doOperations( $realOps, $opts );
                        }
                    );
                } else {
                    $this->logger->error(
                        "'{$backend->getName()}' sync replication; paths: " .
                        FormatJson::encode( $relevantPaths )
                    );
                    $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 = $this->newStatus();
        if ( $this->syncChecks == 0 || count( $this->backends ) <= 1 ) {
            return $status; // skip checks
        }
 
        // Preload all of the stat info in as few round trips as possible
        foreach ( $this->backends as $backend ) {
            $realPaths = $this->substPaths( $paths, $backend );
            $backend->preloadFileStat( [ 'srcs' => $realPaths, 'latest' => true ] );
        }
 
        foreach ( $paths as $path ) {
            $params = [ 'src' => $path, 'latest' => true ];
            // Get the state of the file on the master backend
            $masterBackend = $this->backends[$this->masterIndex];
            $masterParams = $this->substOpPaths( $params, $masterBackend );
            $masterStat = $masterBackend->getFileStat( $masterParams );
            if ( $masterStat === self::STAT_ERROR ) {
                $status->fatal( 'backend-fail-stat', $path );
                continue;
            }
            if ( $this->syncChecks & self::CHECK_SHA1 ) {
                $masterSha1 = $masterBackend->getFileSha1Base36( $masterParams );
                if ( ( $masterSha1 !== false ) !== (bool)$masterStat ) {
                    $status->fatal( 'backend-fail-hash', $path );
                    continue;
                }
            } else {
                $masterSha1 = null; // unused
            }
 
            // Check if all clone backends agree with the master...
            foreach ( $this->backends as $index => $cloneBackend ) {
                if ( $index === $this->masterIndex ) {
                    continue; // master
                }
 
                // Get the state of the file on the clone backend
                $cloneParams = $this->substOpPaths( $params, $cloneBackend );
                $cloneStat = $cloneBackend->getFileStat( $cloneParams );
 
                if ( $masterStat ) {
                    // File exists in the master backend
                    if ( !$cloneStat ) {
                        // File is missing from the clone backend
                        $status->fatal( 'backend-fail-synced', $path );
                    } elseif (
                        ( $this->syncChecks & self::CHECK_SIZE ) &&
                        $cloneStat['size'] !== $masterStat['size']
                    ) {
                        // File in the clone backend is different
                        $status->fatal( 'backend-fail-synced', $path );
                    } elseif (
                        ( $this->syncChecks & self::CHECK_TIME ) &&
                        abs(
                            ConvertibleTimestamp::convert( TS_UNIX, $masterStat['mtime'] ) -
                            ConvertibleTimestamp::convert( TS_UNIX, $cloneStat['mtime'] )
                        ) > 30
                    ) {
                        // File in the clone backend is significantly newer or older
                        $status->fatal( 'backend-fail-synced', $path );
                    } elseif (
                        ( $this->syncChecks & self::CHECK_SHA1 ) &&
                        $cloneBackend->getFileSha1Base36( $cloneParams ) !== $masterSha1
                    ) {
                        // File in the clone backend is different
                        $status->fatal( 'backend-fail-synced', $path );
                    }
                } else {
                    // File does not exist in the master backend
                    if ( $cloneStat ) {
                        // Stray file exists in the clone backend
                        $status->fatal( 'backend-fail-synced', $path );
                    }
                }
            }
        }
 
        return $status;
    }
 
    public function accessibilityCheck( array $paths ) {
        $status = $this->newStatus();
        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, $resyncMode = true ) {
        $status = $this->newStatus();
 
        $fname = __METHOD__;
        foreach ( $paths as $path ) {
            $params = [ 'src' => $path, 'latest' => true ];
            // Get the state of the file on the master backend
            $masterBackend = $this->backends[$this->masterIndex];
            $masterParams = $this->substOpPaths( $params, $masterBackend );
            $masterPath = $masterParams['src'];
            $masterStat = $masterBackend->getFileStat( $masterParams );
            if ( $masterStat === self::STAT_ERROR ) {
                $status->fatal( 'backend-fail-stat', $path );
                $this->logger->error( "$fname: file '$masterPath' is not available" );
                continue;
            }
            $masterSha1 = $masterBackend->getFileSha1Base36( $masterParams );
            if ( ( $masterSha1 !== false ) !== (bool)$masterStat ) {
                $status->fatal( 'backend-fail-hash', $path );
                $this->logger->error( "$fname: file '$masterPath' hash does not match stat" );
                continue;
            }
 
            // Check of all clone backends agree with the master...
            foreach ( $this->backends as $index => $cloneBackend ) {
                if ( $index === $this->masterIndex ) {
                    continue; // master
                }
 
                // Get the state of the file on the clone backend
                $cloneParams = $this->substOpPaths( $params, $cloneBackend );
                $clonePath = $cloneParams['src'];
                $cloneStat = $cloneBackend->getFileStat( $cloneParams );
                if ( $cloneStat === self::STAT_ERROR ) {
                    $status->fatal( 'backend-fail-stat', $path );
                    $this->logger->error( "$fname: file '$clonePath' is not available" );
                    continue;
                }
                $cloneSha1 = $cloneBackend->getFileSha1Base36( $cloneParams );
                if ( ( $cloneSha1 !== false ) !== (bool)$cloneStat ) {
                    $status->fatal( 'backend-fail-hash', $path );
                    $this->logger->error( "$fname: file '$clonePath' hash does not match stat" );
                    continue;
                }
 
                if ( $masterSha1 === $cloneSha1 ) {
                    // File is either the same in both backends or absent from both backends
                    $this->logger->debug( "$fname: file '$clonePath' matches '$masterPath'" );
                } elseif ( $masterSha1 !== false ) {
                    // File is either missing from or different in the clone backend
                    if (
                        $resyncMode === 'conservative' &&
                        $cloneStat &&
                        // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
                        $cloneStat['mtime'] > $masterStat['mtime']
                    ) {
                        // Do not replace files with older ones; reduces the risk of data loss
                        $status->fatal( 'backend-fail-synced', $path );
                    } else {
                        // Copy the master backend file to the clone backend in overwrite mode
                        $fsFile = $masterBackend->getLocalReference( $masterParams );
                        $status->merge( $cloneBackend->quickStore( [
                            'src' => $fsFile,
                            'dst' => $clonePath
                        ] ) );
                    }
                } elseif ( $masterStat === false ) {
                    // Stray file exists in the clone backend
                    if ( $resyncMode === 'conservative' ) {
                        // Do not delete stray files; reduces the risk of data loss
                        $status->fatal( 'backend-fail-synced', $path );
                    } else {
                        // Delete the stay file from the clone backend
                        $status->merge( $cloneBackend->quickDelete( [ 'src' => $clonePath ] ) );
                    }
                }
            }
        }
 
        if ( !$status->isOK() ) {
            $this->logger->error( "$fname: failed to resync: " . FormatJson::encode( $paths ) );
        }
 
        return $status;
    }
 
    protected function fileStoragePathsForOps( array $ops ) {
        $paths = [];
        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( [ '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 = []; // operations
        foreach ( $ops as $op ) {
            $newOp = $op; // operation
            foreach ( [ '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( [ $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, FileBackendStore $backend ) {
        return preg_replace(
            '!^mwstore://' . preg_quote( $backend->getName(), '!' ) . '/!',
            StringUtils::escapeRegexReplacement( "mwstore://{$this->name}/" ),
            $paths // string or array
        );
    }
 
    protected function hasVolatileSources( array $ops ) {
        foreach ( $ops as $op ) {
            if ( $op['op'] === 'store' && !isset( $op['srcRef'] ) ) {
                return true; // source file might be deleted anytime after do*Operations()
            }
        }
 
        return false;
    }
 
    protected function doQuickOperationsInternal( array $ops, array $opts ) {
        $status = $this->newStatus();
        // Do the operations on the master backend; setting StatusValue 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...
        foreach ( $this->backends as $index => $backend ) {
            if ( $index === $this->masterIndex ) {
                continue; // done already
            }
 
            $realOps = $this->substOpBatchPaths( $ops, $backend );
            if ( $this->asyncWrites && !$this->hasVolatileSources( $ops ) ) {
                DeferredUpdates::addCallableUpdate(
                    function () use ( $backend, $realOps ) {
                        $backend->doQuickOperations( $realOps );
                    }
                );
            } else {
                $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 doPrepare( array $params ) {
        return $this->doDirectoryOp( 'prepare', $params );
    }
 
    protected function doSecure( array $params ) {
        return $this->doDirectoryOp( 'secure', $params );
    }
 
    protected function doPublish( array $params ) {
        return $this->doDirectoryOp( 'publish', $params );
    }
 
    protected function doClean( array $params ) {
        return $this->doDirectoryOp( 'clean', $params );
    }
 
    protected function doDirectoryOp( $method, array $params ) {
        $status = $this->newStatus();
 
        $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
        $masterStatus = $this->backends[$this->masterIndex]->$method( $realParams );
        $status->merge( $masterStatus );
 
        foreach ( $this->backends as $index => $backend ) {
            if ( $index === $this->masterIndex ) {
                continue; // already done
            }
 
            $realParams = $this->substOpPaths( $params, $backend );
            if ( $this->asyncWrites ) {
                DeferredUpdates::addCallableUpdate(
                    function () use ( $backend, $method, $realParams ) {
                        $backend->$method( $realParams );
                    }
                );
            } else {
                $status->merge( $backend->$method( $realParams ) );
            }
        }
 
        return $status;
    }
 
    public function concatenate( array $params ) {
        $status = $this->newStatus();
        // We are writing to an FS file, so we don't need to do this per-backend
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        $status->merge( $this->backends[$index]->concatenate( $realParams ) );
 
        return $status;
    }
 
    public function fileExists( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->fileExists( $realParams );
    }
 
    public function getFileTimestamp( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->getFileTimestamp( $realParams );
    }
 
    public function getFileSize( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->getFileSize( $realParams );
    }
 
    public function getFileStat( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->getFileStat( $realParams );
    }
 
    public function getFileXAttributes( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->getFileXAttributes( $realParams );
    }
 
    public function getFileContentsMulti( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        $contentsM = $this->backends[$index]->getFileContentsMulti( $realParams );
 
        $contents = []; // (path => FSFile) mapping using the proxy backend's name
        foreach ( $contentsM as $path => $data ) {
            $contents[$this->unsubstPaths( $path, $this->backends[$index] )] = $data;
        }
 
        return $contents;
    }
 
    public function getFileSha1Base36( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->getFileSha1Base36( $realParams );
    }
 
    public function getFileProps( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->getFileProps( $realParams );
    }
 
    public function streamFile( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->streamFile( $realParams );
    }
 
    public function getLocalReferenceMulti( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        $fsFilesM = $this->backends[$index]->getLocalReferenceMulti( $realParams );
 
        $fsFiles = []; // (path => FSFile) mapping using the proxy backend's name
        foreach ( $fsFilesM as $path => $fsFile ) {
            $fsFiles[$this->unsubstPaths( $path, $this->backends[$index] )] = $fsFile;
        }
 
        return $fsFiles;
    }
 
    public function getLocalCopyMulti( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        $tempFilesM = $this->backends[$index]->getLocalCopyMulti( $realParams );
 
        $tempFiles = []; // (path => TempFSFile) mapping using the proxy backend's name
        foreach ( $tempFilesM as $path => $tempFile ) {
            $tempFiles[$this->unsubstPaths( $path, $this->backends[$index] )] = $tempFile;
        }
 
        return $tempFiles;
    }
 
    public function getFileHttpUrl( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->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 getFeatures() {
        return $this->backends[$this->masterIndex]->getFeatures();
    }
 
    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 preloadCache( array $paths ) {
        $realPaths = $this->substPaths( $paths, $this->backends[$this->readIndex] );
        $this->backends[$this->readIndex]->preloadCache( $realPaths );
    }
 
    public function preloadFileStat( array $params ) {
        $index = $this->getReadIndexFromParams( $params );
        $realParams = $this->substOpPaths( $params, $this->backends[$index] );
 
        return $this->backends[$index]->preloadFileStat( $realParams );
    }
 
    public function getScopedLocksForOps( array $ops, StatusValue $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 = [
            LockManager::LOCK_UW => $this->unsubstPaths(
                $paths[LockManager::LOCK_UW],
                $this->backends[$this->masterIndex]
            ),
            LockManager::LOCK_EX => $this->unsubstPaths(
                $paths[LockManager::LOCK_EX],
                $this->backends[$this->masterIndex]
            )
        ];
 
        // Actually acquire the locks
        return $this->getScopedFileLocks( $pbPaths, 'mixed', $status );
    }
 
    protected function getReadIndexFromParams( array $params ) {
        return !empty( $params['latest'] ) ? $this->masterIndex : $this->readIndex;
    }
}