includes/filerepo/backend/FileBackendMultiWrite.php

   1 <?php
   2 /**
   3  * Proxy backend that mirrors writes to several internal backends.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup FileBackend
  22  * @author Aaron Schulz
  23  */
  24
  25 /**
  26  * @brief Proxy backend that mirrors writes to several internal backends.
  27  *
  28  * This class defines a multi-write backend. Multiple backends can be
  29  * registered to this proxy backend and it will act as a single backend.
  30  * Use this when all access to those backends is through this proxy backend.
  31  * At least one of the backends must be declared the "master" backend.
  32  *
  33  * Only use this class when transitioning from one storage system to another.
  34  *
  35  * Read operations are only done on the 'master' backend for consistency.
  36  * Write operations are performed on all backends, in the order defined.
  37  * If an operation fails on one backend it will be rolled back from the others.
  38  *
  39  * @ingroup FileBackend
  40  * @since 1.19
  41  */
  42 class FileBackendMultiWrite extends FileBackend {
  43         /** @var Array Prioritized list of FileBackendStore objects */
  44         protected $backends = array(); // array of (backend index => backends)
  45         protected $masterIndex = -1; // integer; index of master backend
  46         protected $syncChecks = 0; // integer bitfield
  47
  48         /* Possible internal backend consistency checks */
  49         const CHECK_SIZE = 1;
  50         const CHECK_TIME = 2;
  51         const CHECK_SHA1 = 4;
  52
  53         /**
  54          * Construct a proxy backend that consists of several internal backends.
  55          * Locking, journaling, and read-only checks are handled by the proxy backend.
  56          *
  57          * Additional $config params include:
  58          *     'backends'    : Array of backend config and multi-backend settings.
  59          *                     Each value is the config used in the constructor of a
  60          *                     FileBackendStore class, but with these additional settings:
  61          *                         'class'         : The name of the backend class
  62          *                         'isMultiMaster' : This must be set for one backend.
  63          *                         'template:      : If given a backend name, this will use
  64          *                                           the config of that backend as a template.
  65          *                                           Values specified here take precedence.
  66          *     'syncChecks'  : Integer bitfield of internal backend sync checks to perform.
  67          *                     Possible bits include the FileBackendMultiWrite::CHECK_* constants.
  68          *                     There are constants for SIZE, TIME, and SHA1.
  69          *                     The checks are done before allowing any file operations.
  70          * @param $config Array
  71          * @throws MWException
  72          */
  73         public function __construct( array $config ) {
  74                 parent::__construct( $config );
  75                 $namesUsed = array();
  76                 // Construct backends here rather than via registration
  77                 // to keep these backends hidden from outside the proxy.
  78                 foreach ( $config['backends'] as $index => $config ) {
  79                         if ( isset( $config['template'] ) ) {
  80                                 // Config is just a modified version of a registered backend's.
  81                                 // This should only be used when that config is used only be this backend.
  82                                 $config = $config + FileBackendGroup::singleton()->config( $config['template'] );
  83                         }
  84                         $name = $config['name'];
  85                         if ( isset( $namesUsed[$name] ) ) { // don't break FileOp predicates
  86                                 throw new MWException( "Two or more backends defined with the name $name." );
  87                         }
  88                         $namesUsed[$name] = 1;
  89                         // Alter certain sub-backend settings for sanity
  90                         unset( $config['readOnly'] ); // use proxy backend setting
  91                         unset( $config['fileJournal'] ); // use proxy backend journal
  92                         $config['lockManager'] = 'nullLockManager'; // lock under proxy backend
  93                         if ( !empty( $config['isMultiMaster'] ) ) {
  94                                 if ( $this->masterIndex >= 0 ) {
  95                                         throw new MWException( 'More than one master backend defined.' );
  96                                 }
  97                                 $this->masterIndex = $index; // this is the "master"
  98                                 $config['fileJournal'] = $this->fileJournal; // log under proxy backend
  99                         }
 100                         // Create sub-backend object
 101                         if ( !isset( $config['class'] ) ) {
 102                                 throw new MWException( 'No class given for a backend config.' );
 103                         }
 104                         $class = $config['class'];
 105                         $this->backends[$index] = new $class( $config );
 106                 }
 107                 if ( $this->masterIndex < 0 ) { // need backends and must have a master
 108                         throw new MWException( 'No master backend defined.' );
 109                 }
 110                 $this->syncChecks = isset( $config['syncChecks'] )
 111                         ? $config['syncChecks']
 112                         : self::CHECK_SIZE;
 113         }
 114
 115         /**
 116          * @see FileBackend::doOperationsInternal()
 117          * @return Status
 118          */
 119         final protected function doOperationsInternal( array $ops, array $opts ) {
 120                 $status = Status::newGood();
 121
 122                 $mbe = $this->backends[$this->masterIndex]; // convenience
 123
 124                 // Get the paths to lock from the master backend
 125                 $realOps = $this->substOpBatchPaths( $ops, $mbe );
 126                 $paths = $mbe->getPathsToLockForOpsInternal( $mbe->getOperationsInternal( $realOps ) );
 127                 // Get the paths under the proxy backend's name
 128                 $paths['sh'] = $this->unsubstPaths( $paths['sh'] );
 129                 $paths['ex'] = $this->unsubstPaths( $paths['ex'] );
 130                 // Try to lock those files for the scope of this function...
 131                 if ( empty( $opts['nonLocking'] ) ) {
 132                         // Try to lock those files for the scope of this function...
 133                         $scopeLockS = $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status );
 134                         $scopeLockE = $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status );
 135                         if ( !$status->isOK() ) {
 136                                 return $status; // abort
 137                         }
 138                 }
 139                 // Clear any cache entries (after locks acquired)
 140                 $this->clearCache();
 141                 // Do a consistency check to see if the backends agree
 142                 $status->merge( $this->consistencyCheck( array_merge( $paths['sh'], $paths['ex'] ) ) );
 143                 if ( !$status->isOK() ) {
 144                         return $status; // abort
 145                 }
 146                 // Actually attempt the operation batch on the master backend...
 147                 $masterStatus = $mbe->doOperations( $realOps, $opts );
 148                 $status->merge( $masterStatus );
 149                 // Propagate the operations to the clone backends...
 150                 foreach ( $this->backends as $index => $backend ) {
 151                         if ( $index !== $this->masterIndex ) { // not done already
 152                                 $realOps = $this->substOpBatchPaths( $ops, $backend );
 153                                 $status->merge( $backend->doOperations( $realOps, $opts ) );
 154                         }
 155                 }
 156                 // Make 'success', 'successCount', and 'failCount' fields reflect
 157                 // the overall operation, rather than all the batches for each backend.
 158                 // Do this by only using success values from the master backend's batch.
 159                 $status->success = $masterStatus->success;
 160                 $status->successCount = $masterStatus->successCount;
 161                 $status->failCount = $masterStatus->failCount;
 162
 163                 return $status;
 164         }
 165
 166         /**
 167          * Check that a set of files are consistent across all internal backends
 168          *
 169          * @param $paths Array
 170          * @return Status
 171          */
 172         public function consistencyCheck( array $paths ) {
 173                 $status = Status::newGood();
 174                 if ( $this->syncChecks == 0 || count( $this->backends ) <= 1 ) {
 175                         return $status; // skip checks
 176                 }
 177
 178                 $mBackend = $this->backends[$this->masterIndex];
 179                 foreach ( array_unique( $paths ) as $path ) {
 180                         $params = array( 'src' => $path, 'latest' => true );
 181                         $mParams = $this->substOpPaths( $params, $mBackend );
 182                         // Stat the file on the 'master' backend
 183                         $mStat = $mBackend->getFileStat( $mParams );
 184                         if ( $this->syncChecks & self::CHECK_SHA1 ) {
 185                                 $mSha1 = $mBackend->getFileSha1( $mParams );
 186                         } else {
 187                                 $mSha1 = false;
 188                         }
 189                         $mUsable = $mBackend->isPathUsableInternal( $mParams['src'] );
 190                         // Check of all clone backends agree with the master...
 191                         foreach ( $this->backends as $index => $cBackend ) {
 192                                 if ( $index === $this->masterIndex ) {
 193                                         continue; // master
 194                                 }
 195                                 $cParams = $this->substOpPaths( $params, $cBackend );
 196                                 $cStat = $cBackend->getFileStat( $cParams );
 197                                 if ( $mStat ) { // file is in master
 198                                         if ( !$cStat ) { // file should exist
 199                                                 $status->fatal( 'backend-fail-synced', $path );
 200                                                 continue;
 201                                         }
 202                                         if ( $this->syncChecks & self::CHECK_SIZE ) {
 203                                                 if ( $cStat['size'] != $mStat['size'] ) { // wrong size
 204                                                         $status->fatal( 'backend-fail-synced', $path );
 205                                                         continue;
 206                                                 }
 207                                         }
 208                                         if ( $this->syncChecks & self::CHECK_TIME ) {
 209                                                 $mTs = wfTimestamp( TS_UNIX, $mStat['mtime'] );
 210                                                 $cTs = wfTimestamp( TS_UNIX, $cStat['mtime'] );
 211                                                 if ( abs( $mTs - $cTs ) > 30 ) { // outdated file somewhere
 212                                                         $status->fatal( 'backend-fail-synced', $path );
 213                                                         continue;
 214                                                 }
 215                                         }
 216                                         if ( $this->syncChecks & self::CHECK_SHA1 ) {
 217                                                 if ( $cBackend->getFileSha1( $cParams ) !== $mSha1 ) { // wrong SHA1
 218                                                         $status->fatal( 'backend-fail-synced', $path );
 219                                                         continue;
 220                                                 }
 221                                         }
 222                                 } else { // file is not in master
 223                                         if ( $cStat ) { // file should not exist
 224                                                 $status->fatal( 'backend-fail-synced', $path );
 225                                         }
 226                                 }
 227                                 if ( $mUsable !== $cBackend->isPathUsableInternal( $cParams['src'] ) ) {
 228                                         $status->fatal( 'backend-fail-synced', $path );
 229                                 }
 230                         }
 231                 }
 232
 233                 return $status;
 234         }
 235
 236         /**
 237          * Substitute the backend name in storage path parameters
 238          * for a set of operations with that of a given internal backend.
 239          *
 240          * @param $ops Array List of file operation arrays
 241          * @param $backend FileBackendStore
 242          * @return Array
 243          */
 244         protected function substOpBatchPaths( array $ops, FileBackendStore $backend ) {
 245                 $newOps = array(); // operations
 246                 foreach ( $ops as $op ) {
 247                         $newOp = $op; // operation
 248                         foreach ( array( 'src', 'srcs', 'dst', 'dir' ) as $par ) {
 249                                 if ( isset( $newOp[$par] ) ) { // string or array
 250                                         $newOp[$par] = $this->substPaths( $newOp[$par], $backend );
 251                                 }
 252                         }
 253                         $newOps[] = $newOp;
 254                 }
 255                 return $newOps;
 256         }
 257
 258         /**
 259          * Same as substOpBatchPaths() but for a single operation
 260          *
 261          * @param $ops array File operation array
 262          * @param $backend FileBackendStore
 263          * @return Array
 264          */
 265         protected function substOpPaths( array $ops, FileBackendStore $backend ) {
 266                 $newOps = $this->substOpBatchPaths( array( $ops ), $backend );
 267                 return $newOps[0];
 268         }
 269
 270         /**
 271          * Substitute the backend of storage paths with an internal backend's name
 272          *
 273          * @param $paths Array|string List of paths or single string path
 274          * @param $backend FileBackendStore
 275          * @return Array|string
 276          */
 277         protected function substPaths( $paths, FileBackendStore $backend ) {
 278                 return preg_replace(
 279                         '!^mwstore://' . preg_quote( $this->name ) . '/!',
 280                         StringUtils::escapeRegexReplacement( "mwstore://{$backend->getName()}/" ),
 281                         $paths // string or array
 282                 );
 283         }
 284
 285         /**
 286          * Substitute the backend of internal storage paths with the proxy backend's name
 287          *
 288          * @param $paths Array|string List of paths or single string path
 289          * @return Array|string
 290          */
 291         protected function unsubstPaths( $paths ) {
 292                 return preg_replace(
 293                         '!^mwstore://([^/]+)!',
 294                         StringUtils::escapeRegexReplacement( "mwstore://{$this->name}" ),
 295                         $paths // string or array
 296                 );
 297         }
 298
 299         /**
 300          * @see FileBackend::doQuickOperationsInternal()
 301          * @return Status
 302          */
 303         protected function doQuickOperationsInternal( array $ops ) {
 304                 $status = Status::newGood();
 305                 // Do the operations on the master backend; setting Status fields...
 306                 $realOps = $this->substOpBatchPaths( $ops, $this->backends[$this->masterIndex] );
 307                 $masterStatus = $this->backends[$this->masterIndex]->doQuickOperations( $realOps );
 308                 $status->merge( $masterStatus );
 309                 // Propagate the operations to the clone backends...
 310                 foreach ( $this->backends as $index => $backend ) {
 311                         if ( $index !== $this->masterIndex ) { // not done already
 312                                 $realOps = $this->substOpBatchPaths( $ops, $backend );
 313                                 $status->merge( $backend->doQuickOperations( $realOps ) );
 314                         }
 315                 }
 316                 // Make 'success', 'successCount', and 'failCount' fields reflect
 317                 // the overall operation, rather than all the batches for each backend.
 318                 // Do this by only using success values from the master backend's batch.
 319                 $status->success = $masterStatus->success;
 320                 $status->successCount = $masterStatus->successCount;
 321                 $status->failCount = $masterStatus->failCount;
 322                 return $status;
 323         }
 324
 325         /**
 326          * @see FileBackend::doPrepare()
 327          * @return Status
 328          */
 329         protected function doPrepare( array $params ) {
 330                 $status = Status::newGood();
 331                 foreach ( $this->backends as $backend ) {
 332                         $realParams = $this->substOpPaths( $params, $backend );
 333                         $status->merge( $backend->doPrepare( $realParams ) );
 334                 }
 335                 return $status;
 336         }
 337
 338         /**
 339          * @see FileBackend::doSecure()
 340          * @param $params array
 341          * @return Status
 342          */
 343         protected function doSecure( array $params ) {
 344                 $status = Status::newGood();
 345                 foreach ( $this->backends as $backend ) {
 346                         $realParams = $this->substOpPaths( $params, $backend );
 347                         $status->merge( $backend->doSecure( $realParams ) );
 348                 }
 349                 return $status;
 350         }
 351
 352         /**
 353          * @see FileBackend::doPublish()
 354          * @param $params array
 355          * @return Status
 356          */
 357         protected function doPublish( array $params ) {
 358                 $status = Status::newGood();
 359                 foreach ( $this->backends as $backend ) {
 360                         $realParams = $this->substOpPaths( $params, $backend );
 361                         $status->merge( $backend->doPublish( $realParams ) );
 362                 }
 363                 return $status;
 364         }
 365
 366         /**
 367          * @see FileBackend::doClean()
 368          * @param $params array
 369          * @return Status
 370          */
 371         protected function doClean( array $params ) {
 372                 $status = Status::newGood();
 373                 foreach ( $this->backends as $backend ) {
 374                         $realParams = $this->substOpPaths( $params, $backend );
 375                         $status->merge( $backend->doClean( $realParams ) );
 376                 }
 377                 return $status;
 378         }
 379
 380         /**
 381          * @see FileBackend::concatenate()
 382          * @param $params array
 383          * @return Status
 384          */
 385         public function concatenate( array $params ) {
 386                 // We are writing to an FS file, so we don't need to do this per-backend
 387                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 388                 return $this->backends[$this->masterIndex]->concatenate( $realParams );
 389         }
 390
 391         /**
 392          * @see FileBackend::fileExists()
 393          * @param $params array
 394          */
 395         public function fileExists( array $params ) {
 396                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 397                 return $this->backends[$this->masterIndex]->fileExists( $realParams );
 398         }
 399
 400         /**
 401          * @see FileBackend::getFileTimestamp()
 402          * @param $params array
 403          * @return bool|string
 404          */
 405         public function getFileTimestamp( array $params ) {
 406                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 407                 return $this->backends[$this->masterIndex]->getFileTimestamp( $realParams );
 408         }
 409
 410         /**
 411          * @see FileBackend::getFileSize()
 412          * @param $params array
 413          * @return bool|int
 414          */
 415         public function getFileSize( array $params ) {
 416                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 417                 return $this->backends[$this->masterIndex]->getFileSize( $realParams );
 418         }
 419
 420         /**
 421          * @see FileBackend::getFileStat()
 422          * @param $params array
 423          * @return Array|bool|null
 424          */
 425         public function getFileStat( array $params ) {
 426                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 427                 return $this->backends[$this->masterIndex]->getFileStat( $realParams );
 428         }
 429
 430         /**
 431          * @see FileBackend::getFileContents()
 432          * @param $params array
 433          * @return bool|string
 434          */
 435         public function getFileContents( array $params ) {
 436                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 437                 return $this->backends[$this->masterIndex]->getFileContents( $realParams );
 438         }
 439
 440         /**
 441          * @see FileBackend::getFileSha1Base36()
 442          * @param $params array
 443          * @return bool|string
 444          */
 445         public function getFileSha1Base36( array $params ) {
 446                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 447                 return $this->backends[$this->masterIndex]->getFileSha1Base36( $realParams );
 448         }
 449
 450         /**
 451          * @see FileBackend::getFileProps()
 452          * @param $params array
 453          * @return Array
 454          */
 455         public function getFileProps( array $params ) {
 456                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 457                 return $this->backends[$this->masterIndex]->getFileProps( $realParams );
 458         }
 459
 460         /**
 461          * @see FileBackend::streamFile()
 462          * @param $params array
 463          * @return \Status
 464          */
 465         public function streamFile( array $params ) {
 466                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 467                 return $this->backends[$this->masterIndex]->streamFile( $realParams );
 468         }
 469
 470         /**
 471          * @see FileBackend::getLocalReference()
 472          * @param $params array
 473          * @return FSFile|null
 474          */
 475         public function getLocalReference( array $params ) {
 476                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 477                 return $this->backends[$this->masterIndex]->getLocalReference( $realParams );
 478         }
 479
 480         /**
 481          * @see FileBackend::getLocalCopy()
 482          * @param $params array
 483          * @return null|TempFSFile
 484          */
 485         public function getLocalCopy( array $params ) {
 486                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 487                 return $this->backends[$this->masterIndex]->getLocalCopy( $realParams );
 488         }
 489
 490         /**
 491          * @see FileBackend::directoryExists()
 492          * @param $params array
 493          * @return bool|null
 494          */
 495         public function directoryExists( array $params ) {
 496                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 497                 return $this->backends[$this->masterIndex]->directoryExists( $realParams );
 498         }
 499
 500         /**
 501          * @see FileBackend::getSubdirectoryList()
 502          * @param $params array
 503          * @return Array|null|Traversable
 504          */
 505         public function getDirectoryList( array $params ) {
 506                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 507                 return $this->backends[$this->masterIndex]->getDirectoryList( $realParams );
 508         }
 509
 510         /**
 511          * @see FileBackend::getFileList()
 512          * @param $params array
 513          * @return Array|null|\Traversable
 514          */
 515         public function getFileList( array $params ) {
 516                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 517                 return $this->backends[$this->masterIndex]->getFileList( $realParams );
 518         }
 519
 520         /**
 521          * @see FileBackend::clearCache()
 522          */
 523         public function clearCache( array $paths = null ) {
 524                 foreach ( $this->backends as $backend ) {
 525                         $realPaths = is_array( $paths ) ? $this->substPaths( $paths, $backend ) : null;
 526                         $backend->clearCache( $realPaths );
 527                 }
 528         }
 529
 530         /**
 531          * @see FileBackend::getScopedLocksForOps()
 532          */
 533         public function getScopedLocksForOps( array $ops, Status $status ) {
 534                 $fileOps = $this->backends[$this->masterIndex]->getOperationsInternal( $ops );
 535                 // Get the paths to lock from the master backend
 536                 $paths = $this->backends[$this->masterIndex]->getPathsToLockForOpsInternal( $fileOps );
 537                 // Get the paths under the proxy backend's name
 538                 $paths['sh'] = $this->unsubstPaths( $paths['sh'] );
 539                 $paths['ex'] = $this->unsubstPaths( $paths['ex'] );
 540                 return array(
 541                         $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status ),
 542                         $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status )
 543                 );
 544         }
 545 }