- Generated on Wed Jun 25 2014 21:02:16 for MediaWiki by
1.7.6.1
|
MediaWiki
REL1_21
|
00001 <?php 00024 require_once( __DIR__ . '/Maintenance.php' ); 00025 00032 class SyncFileBackend extends Maintenance { 00033 public function __construct() { 00034 parent::__construct(); 00035 $this->mDescription = "Sync one file backend with another using the journal"; 00036 $this->addOption( 'src', 'Name of backend to sync from', true, true ); 00037 $this->addOption( 'dst', 'Name of destination backend to sync', false, true ); 00038 $this->addOption( 'start', 'Starting journal ID', false, true ); 00039 $this->addOption( 'end', 'Ending journal ID', false, true ); 00040 $this->addOption( 'posdir', 'Directory to read/record journal positions', false, true ); 00041 $this->addOption( 'posdump', 'Just dump current journal position into the position dir.' ); 00042 $this->addOption( 'postime', 'For position dumps, get the ID at this time', false, true ); 00043 $this->addOption( 'verbose', 'Verbose mode', false, false, 'v' ); 00044 $this->setBatchSize( 50 ); 00045 } 00046 00047 public function execute() { 00048 $src = FileBackendGroup::singleton()->get( $this->getOption( 'src' ) ); 00049 00050 $posDir = $this->getOption( 'posdir' ); 00051 $posFile = $posDir ? $posDir . '/' . wfWikiID() : false; 00052 00053 if ( $this->hasOption( 'posdump' ) ) { 00054 // Just dump the current position into the specified position dir 00055 if ( !$this->hasOption( 'posdir' ) ) { 00056 $this->error( "Param posdir required!", 1 ); 00057 } 00058 if ( $this->hasOption( 'postime' ) ) { 00059 $id = (int)$src->getJournal()->getPositionAtTime( $this->getOption( 'postime' ) ); 00060 $this->output( "Requested journal position is $id.\n" ); 00061 } else { 00062 $id = (int)$src->getJournal()->getCurrentPosition(); 00063 $this->output( "Current journal position is $id.\n" ); 00064 } 00065 if ( file_put_contents( $posFile, $id, LOCK_EX ) !== false ) { 00066 $this->output( "Saved journal position file.\n" ); 00067 } else { 00068 $this->output( "Could not save journal position file.\n" ); 00069 } 00070 if ( $this->isQuiet() ) { 00071 print $id; // give a single machine-readable number 00072 } 00073 return; 00074 } 00075 00076 if ( !$this->hasOption( 'dst' ) ) { 00077 $this->error( "Param dst required!", 1 ); 00078 } 00079 $dst = FileBackendGroup::singleton()->get( $this->getOption( 'dst' ) ); 00080 00081 $start = $this->getOption( 'start', 0 ); 00082 if ( !$start && $posFile && is_dir( $posDir ) ) { 00083 $start = is_file( $posFile ) 00084 ? (int)trim( file_get_contents( $posFile ) ) 00085 : 0; 00086 ++$start; // we already did this ID, start with the next one 00087 $startFromPosFile = true; 00088 } else { 00089 $startFromPosFile = false; 00090 } 00091 $end = $this->getOption( 'end', INF ); 00092 00093 $this->output( "Synchronizing backend '{$dst->getName()}' to '{$src->getName()}'...\n" ); 00094 $this->output( "Starting journal position is $start.\n" ); 00095 if ( is_finite( $end ) ) { 00096 $this->output( "Ending journal position is $end.\n" ); 00097 } 00098 00099 // Periodically update the position file 00100 $callback = function( $pos ) use ( $startFromPosFile, $posFile, $start ) { 00101 if ( $startFromPosFile && $pos >= $start ) { // successfully advanced 00102 file_put_contents( $posFile, $pos, LOCK_EX ); 00103 } 00104 }; 00105 00106 // Actually sync the dest backend with the reference backend 00107 $lastOKPos = $this->syncBackends( $src, $dst, $start, $end, $callback ); 00108 00109 // Update the sync position file 00110 if ( $startFromPosFile && $lastOKPos >= $start ) { // successfully advanced 00111 if ( file_put_contents( $posFile, $lastOKPos, LOCK_EX ) !== false ) { 00112 $this->output( "Updated journal position file.\n" ); 00113 } else { 00114 $this->output( "Could not update journal position file.\n" ); 00115 } 00116 } 00117 00118 if ( $lastOKPos === false ) { 00119 if ( !$start ) { 00120 $this->output( "No journal entries found.\n" ); 00121 } else { 00122 $this->output( "No new journal entries found.\n" ); 00123 } 00124 } else { 00125 $this->output( "Stopped synchronization at journal position $lastOKPos.\n" ); 00126 } 00127 00128 if ( $this->isQuiet() ) { 00129 print $lastOKPos; // give a single machine-readable number 00130 } 00131 } 00132 00144 protected function syncBackends( 00145 FileBackend $src, FileBackend $dst, $start, $end, Closure $callback 00146 ) { 00147 $lastOKPos = 0; // failed 00148 $first = true; // first batch 00149 00150 if ( $start > $end ) { // sanity 00151 $this->error( "Error: given starting ID greater than ending ID.", 1 ); 00152 } 00153 00154 do { 00155 $limit = min( $this->mBatchSize, $end - $start + 1 ); // don't go pass ending ID 00156 $this->output( "Doing id $start to " . ( $start + $limit - 1 ) . "...\n" ); 00157 00158 $entries = $src->getJournal()->getChangeEntries( $start, $limit, $next ); 00159 $start = $next; // start where we left off next time 00160 if ( $first && !count( $entries ) ) { 00161 return false; // nothing to do 00162 } 00163 $first = false; 00164 00165 $lastPosInBatch = 0; 00166 $pathsInBatch = array(); // changed paths 00167 foreach ( $entries as $entry ) { 00168 if ( $entry['op'] !== 'null' ) { // null ops are just for reference 00169 $pathsInBatch[$entry['path']] = 1; // remove duplicates 00170 } 00171 $lastPosInBatch = $entry['id']; 00172 } 00173 00174 $status = $this->syncFileBatch( array_keys( $pathsInBatch ), $src, $dst ); 00175 if ( $status->isOK() ) { 00176 $lastOKPos = max( $lastOKPos, $lastPosInBatch ); 00177 $callback( $lastOKPos ); // update position file 00178 } else { 00179 $this->error( print_r( $status->getErrorsArray(), true ) ); 00180 break; // no gaps; everything up to $lastPos must be OK 00181 } 00182 00183 if ( !$start ) { 00184 $this->output( "End of journal entries.\n" ); 00185 } 00186 } while ( $start && $start <= $end ); 00187 00188 return $lastOKPos; 00189 } 00190 00199 protected function syncFileBatch( array $paths, FileBackend $src, FileBackend $dst ) { 00200 $status = Status::newGood(); 00201 if ( !count( $paths ) ) { 00202 return $status; // nothing to do 00203 } 00204 00205 // Source: convert internal backend names (FileBackendMultiWrite) to the public one 00206 $sPaths = $this->replaceNamePaths( $paths, $src ); 00207 // Destination: get corresponding path name 00208 $dPaths = $this->replaceNamePaths( $paths, $dst ); 00209 00210 // Lock the live backend paths from modification 00211 $sLock = $src->getScopedFileLocks( $sPaths, LockManager::LOCK_UW, $status ); 00212 $eLock = $dst->getScopedFileLocks( $dPaths, LockManager::LOCK_EX, $status ); 00213 if ( !$status->isOK() ) { 00214 return $status; 00215 } 00216 00217 $ops = array(); 00218 $fsFiles = array(); 00219 foreach ( $sPaths as $i => $sPath ) { 00220 $dPath = $dPaths[$i]; // destination 00221 $sExists = $src->fileExists( array( 'src' => $sPath, 'latest' => 1 ) ); 00222 if ( $sExists === true ) { // exists in source 00223 if ( $this->filesAreSame( $src, $dst, $sPath, $dPath ) ) { 00224 continue; // avoid local copies for non-FS backends 00225 } 00226 // Note: getLocalReference() is fast for FS backends 00227 $fsFile = $src->getLocalReference( array( 'src' => $sPath, 'latest' => 1 ) ); 00228 if ( !$fsFile ) { 00229 $this->error( "Unable to sync '$dPath': could not get local copy." ); 00230 $status->fatal( 'backend-fail-internal', $src->getName() ); 00231 return $status; 00232 } 00233 $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed 00234 // Note: prepare() is usually fast for key/value backends 00235 $status->merge( $dst->prepare( array( 00236 'dir' => dirname( $dPath ), 'bypassReadOnly' => 1 ) ) ); 00237 if ( !$status->isOK() ) { 00238 return $status; 00239 } 00240 $ops[] = array( 'op' => 'store', 00241 'src' => $fsFile->getPath(), 'dst' => $dPath, 'overwrite' => 1 ); 00242 } elseif ( $sExists === false ) { // does not exist in source 00243 $ops[] = array( 'op' => 'delete', 'src' => $dPath, 'ignoreMissingSource' => 1 ); 00244 } else { // error 00245 $this->error( "Unable to sync '$dPath': could not stat file." ); 00246 $status->fatal( 'backend-fail-internal', $src->getName() ); 00247 return $status; 00248 } 00249 } 00250 00251 $t_start = microtime( true ); 00252 $status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) ); 00253 if ( !$status->isOK() ) { 00254 sleep( 10 ); // wait and retry copy again 00255 $status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) ); 00256 } 00257 $ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 ); 00258 if ( $status->isOK() && $this->getOption( 'verbose' ) ) { 00259 $this->output( "Synchronized these file(s) [{$ellapsed_ms}ms]:\n" . 00260 implode( "\n", $dPaths ) . "\n" ); 00261 } 00262 00263 return $status; 00264 } 00265 00272 protected function replaceNamePaths( $paths, FileBackend $backend ) { 00273 return preg_replace( 00274 '!^mwstore://([^/]+)!', 00275 StringUtils::escapeRegexReplacement( "mwstore://" . $backend->getName() ), 00276 $paths // string or array 00277 ); 00278 } 00279 00280 protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) { 00281 return ( 00282 ( $src->getFileSize( array( 'src' => $sPath ) ) 00283 === $dst->getFileSize( array( 'src' => $dPath ) ) // short-circuit 00284 ) && ( $src->getFileSha1Base36( array( 'src' => $sPath ) ) 00285 === $dst->getFileSha1Base36( array( 'src' => $dPath ) ) 00286 ) 00287 ); 00288 } 00289 } 00290 00291 $maintClass = "SyncFileBackend"; 00292 require_once( RUN_MAINTENANCE_IF_MAIN );
1.7.6.1