includes/filerepo/backend/FileBackendMultiWrite.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup FileBackend
   5  * @author Aaron Schulz
   6  */
   7
   8 /**
   9  * @brief Proxy backend that mirrors writes to several internal backends.
  10  *
  11  * This class defines a multi-write backend. Multiple backends can be
  12  * registered to this proxy backend and it will act as a single backend.
  13  * Use this when all access to those backends is through this proxy backend.
  14  * At least one of the backends must be declared the "master" backend.
  15  *
  16  * Only use this class when transitioning from one storage system to another.
  17  *
  18  * Read operations are only done on the 'master' backend for consistency.
  19  * Write operations are performed on all backends, in the order defined.
  20  * If an operation fails on one backend it will be rolled back from the others.
  21  *
  22  * @ingroup FileBackend
  23  * @since 1.19
  24  */
  25 class FileBackendMultiWrite extends FileBackend {
  26         /** @var Array Prioritized list of FileBackendStore objects */
  27         protected $backends = array(); // array of (backend index => backends)
  28         protected $masterIndex = -1; // integer; index of master backend
  29         protected $syncChecks = 0; // integer bitfield
  30
  31         /* Possible internal backend consistency checks */
  32         const CHECK_SIZE = 1;
  33         const CHECK_TIME = 2;
  34
  35         /**
  36          * Construct a proxy backend that consists of several internal backends.
  37          * Additional $config params include:
  38          *     'backends'    : Array of backend config and multi-backend settings.
  39          *                     Each value is the config used in the constructor of a
  40          *                     FileBackendStore class, but with these additional settings:
  41          *                         'class'         : The name of the backend class
  42          *                         'isMultiMaster' : This must be set for one backend.
  43          *     'syncChecks'  : Integer bitfield of internal backend sync checks to perform.
  44          *                     Possible bits include self::CHECK_SIZE and self::CHECK_TIME.
  45          *                     The checks are done before allowing any file operations.
  46          * @param $config Array
  47          */
  48         public function __construct( array $config ) {
  49                 parent::__construct( $config );
  50                 $namesUsed = array();
  51                 // Construct backends here rather than via registration
  52                 // to keep these backends hidden from outside the proxy.
  53                 foreach ( $config['backends'] as $index => $config ) {
  54                         $name = $config['name'];
  55                         if ( isset( $namesUsed[$name] ) ) { // don't break FileOp predicates
  56                                 throw new MWException( "Two or more backends defined with the name $name." );
  57                         }
  58                         $namesUsed[$name] = 1;
  59                         if ( !isset( $config['class'] ) ) {
  60                                 throw new MWException( 'No class given for a backend config.' );
  61                         }
  62                         $class = $config['class'];
  63                         $this->backends[$index] = new $class( $config );
  64                         if ( !empty( $config['isMultiMaster'] ) ) {
  65                                 if ( $this->masterIndex >= 0 ) {
  66                                         throw new MWException( 'More than one master backend defined.' );
  67                                 }
  68                                 $this->masterIndex = $index;
  69                         }
  70                 }
  71                 if ( $this->masterIndex < 0 ) { // need backends and must have a master
  72                         throw new MWException( 'No master backend defined.' );
  73                 }
  74                 $this->syncChecks = isset( $config['syncChecks'] )
  75                         ? $config['syncChecks']
  76                         : self::CHECK_SIZE;
  77         }
  78
  79         /**
  80          * @see FileBackend::doOperationsInternal()
  81          * @return Status
  82          */
  83         final protected function doOperationsInternal( array $ops, array $opts ) {
  84                 $status = Status::newGood();
  85
  86                 $performOps = array(); // list of FileOp objects
  87                 $filesRead = array(); // storage paths read from
  88                 $filesChanged = array(); // storage paths written to
  89                 // Build up a list of FileOps. The list will have all the ops
  90                 // for one backend, then all the ops for the next, and so on.
  91                 // These batches of ops are all part of a continuous array.
  92                 // Also build up a list of files read/changed...
  93                 foreach ( $this->backends as $index => $backend ) {
  94                         $backendOps = $this->substOpBatchPaths( $ops, $backend );
  95                         // Add on the operation batch for this backend
  96                         $performOps = array_merge( $performOps, $backend->getOperations( $backendOps ) );
  97                         if ( $index == 0 ) { // first batch
  98                                 // Get the files used for these operations. Each backend has a batch of
  99                                 // the same operations, so we only need to get them from the first batch.
 100                                 foreach ( $performOps as $fileOp ) {
 101                                         $filesRead = array_merge( $filesRead, $fileOp->storagePathsRead() );
 102                                         $filesChanged = array_merge( $filesChanged, $fileOp->storagePathsChanged() );
 103                                 }
 104                                 // Get the paths under the proxy backend's name
 105                                 $filesRead = $this->unsubstPaths( $filesRead );
 106                                 $filesChanged = $this->unsubstPaths( $filesChanged );
 107                         }
 108                 }
 109
 110                 // Try to lock those files for the scope of this function...
 111                 if ( empty( $opts['nonLocking'] ) ) {
 112                         $filesLockSh = array_diff( $filesRead, $filesChanged ); // optimization
 113                         $filesLockEx = $filesChanged;
 114                         // Get a shared lock on the parent directory of each path changed
 115                         $filesLockSh = array_merge( $filesLockSh, array_map( 'dirname', $filesLockEx ) );
 116                         // Try to lock those files for the scope of this function...
 117                         $scopeLockS = $this->getScopedFileLocks( $filesLockSh, LockManager::LOCK_UW, $status );
 118                         $scopeLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 119                         if ( !$status->isOK() ) {
 120                                 return $status; // abort
 121                         }
 122                 }
 123
 124                 // Clear any cache entries (after locks acquired)
 125                 $this->clearCache();
 126
 127                 // Do a consistency check to see if the backends agree
 128                 if ( count( $this->backends ) > 1 ) {
 129                         $status->merge( $this->consistencyCheck( array_merge( $filesRead, $filesChanged ) ) );
 130                         if ( !$status->isOK() ) {
 131                                 return $status; // abort
 132                         }
 133                 }
 134
 135                 // Actually attempt the operation batch...
 136                 $subStatus = FileOp::attemptBatch( $performOps, $opts, $this->fileJournal );
 137
 138                 $success = array();
 139                 $failCount = 0;
 140                 $successCount = 0;
 141                 // Make 'success', 'successCount', and 'failCount' fields reflect
 142                 // the overall operation, rather than all the batches for each backend.
 143                 // Do this by only using success values from the master backend's batch.
 144                 $batchStart = $this->masterIndex * count( $ops );
 145                 $batchEnd = $batchStart + count( $ops ) - 1;
 146                 for ( $i = $batchStart; $i <= $batchEnd; $i++ ) {
 147                         if ( !isset( $subStatus->success[$i] ) ) {
 148                                 break; // failed out before trying this op
 149                         } elseif ( $subStatus->success[$i] ) {
 150                                 ++$successCount;
 151                         } else {
 152                                 ++$failCount;
 153                         }
 154                         $success[] = $subStatus->success[$i];
 155                 }
 156                 $subStatus->success = $success;
 157                 $subStatus->successCount = $successCount;
 158                 $subStatus->failCount = $failCount;
 159
 160                 // Merge errors into status fields
 161                 $status->merge( $subStatus );
 162                 $status->success = $subStatus->success; // not done in merge()
 163
 164                 return $status;
 165         }
 166
 167         /**
 168          * Check that a set of files are consistent across all internal backends
 169          *
 170          * @param $paths Array
 171          * @return Status
 172          */
 173         public function consistencyCheck( array $paths ) {
 174                 $status = Status::newGood();
 175                 if ( $this->syncChecks == 0 ) {
 176                         return $status; // skip checks
 177                 }
 178
 179                 $mBackend = $this->backends[$this->masterIndex];
 180                 foreach ( array_unique( $paths ) as $path ) {
 181                         $params = array( 'src' => $path, 'latest' => true );
 182                         // Stat the file on the 'master' backend
 183                         $mStat = $mBackend->getFileStat( $this->substOpPaths( $params, $mBackend ) );
 184                         // Check of all clone backends agree with the master...
 185                         foreach ( $this->backends as $index => $cBackend ) {
 186                                 if ( $index === $this->masterIndex ) {
 187                                         continue; // master
 188                                 }
 189                                 $cStat = $cBackend->getFileStat( $this->substOpPaths( $params, $cBackend ) );
 190                                 if ( $mStat ) { // file is in master
 191                                         if ( !$cStat ) { // file should exist
 192                                                 $status->fatal( 'backend-fail-synced', $path );
 193                                                 continue;
 194                                         }
 195                                         if ( $this->syncChecks & self::CHECK_SIZE ) {
 196                                                 if ( $cStat['size'] != $mStat['size'] ) { // wrong size
 197                                                         $status->fatal( 'backend-fail-synced', $path );
 198                                                         continue;
 199                                                 }
 200                                         }
 201                                         if ( $this->syncChecks & self::CHECK_TIME ) {
 202                                                 $mTs = wfTimestamp( TS_UNIX, $mStat['mtime'] );
 203                                                 $cTs = wfTimestamp( TS_UNIX, $cStat['mtime'] );
 204                                                 if ( abs( $mTs - $cTs ) > 30 ) { // outdated file somewhere
 205                                                         $status->fatal( 'backend-fail-synced', $path );
 206                                                         continue;
 207                                                 }
 208                                         }
 209                                 } else { // file is not in master
 210                                         if ( $cStat ) { // file should not exist
 211                                                 $status->fatal( 'backend-fail-synced', $path );
 212                                         }
 213                                 }
 214                         }
 215                 }
 216
 217                 return $status;
 218         }
 219
 220         /**
 221          * Substitute the backend name in storage path parameters
 222          * for a set of operations with that of a given internal backend.
 223          *
 224          * @param $ops Array List of file operation arrays
 225          * @param $backend FileBackendStore
 226          * @return Array
 227          */
 228         protected function substOpBatchPaths( array $ops, FileBackendStore $backend ) {
 229                 $newOps = array(); // operations
 230                 foreach ( $ops as $op ) {
 231                         $newOp = $op; // operation
 232                         foreach ( array( 'src', 'srcs', 'dst', 'dir' ) as $par ) {
 233                                 if ( isset( $newOp[$par] ) ) { // string or array
 234                                         $newOp[$par] = $this->substPaths( $newOp[$par], $backend );
 235                                 }
 236                         }
 237                         $newOps[] = $newOp;
 238                 }
 239                 return $newOps;
 240         }
 241
 242         /**
 243          * Same as substOpBatchPaths() but for a single operation
 244          *
 245          * @param $op File operation array
 246          * @param $backend FileBackendStore
 247          * @return Array
 248          */
 249         protected function substOpPaths( array $ops, FileBackendStore $backend ) {
 250                 $newOps = $this->substOpBatchPaths( array( $ops ), $backend );
 251                 return $newOps[0];
 252         }
 253
 254         /**
 255          * Substitute the backend of storage paths with an internal backend's name
 256          *
 257          * @param $paths Array|string List of paths or single string path
 258          * @param $backend FileBackendStore
 259          * @return Array|string
 260          */
 261         protected function substPaths( $paths, FileBackendStore $backend ) {
 262                 return preg_replace(
 263                         '!^mwstore://' . preg_quote( $this->name ) . '/!',
 264                         StringUtils::escapeRegexReplacement( "mwstore://{$backend->getName()}/" ),
 265                         $paths // string or array
 266                 );
 267         }
 268
 269         /**
 270          * Substitute the backend of internal storage paths with the proxy backend's name
 271          *
 272          * @param $paths Array|string List of paths or single string path
 273          * @return Array|string
 274          */
 275         protected function unsubstPaths( $paths ) {
 276                 return preg_replace(
 277                         '!^mwstore://([^/]+)!',
 278                         StringUtils::escapeRegexReplacement( "mwstore://{$this->name}" ),
 279                         $paths // string or array
 280                 );
 281         }
 282
 283         /**
 284          * @see FileBackend::doPrepare()
 285          * @return Status
 286          */
 287         protected function doPrepare( array $params ) {
 288                 $status = Status::newGood();
 289                 foreach ( $this->backends as $backend ) {
 290                         $realParams = $this->substOpPaths( $params, $backend );
 291                         $status->merge( $backend->doPrepare( $realParams ) );
 292                 }
 293                 return $status;
 294         }
 295
 296         /**
 297          * @see FileBackend::doSecure()
 298          * @return Status
 299          */
 300         protected function doSecure( array $params ) {
 301                 $status = Status::newGood();
 302                 foreach ( $this->backends as $backend ) {
 303                         $realParams = $this->substOpPaths( $params, $backend );
 304                         $status->merge( $backend->doSecure( $realParams ) );
 305                 }
 306                 return $status;
 307         }
 308
 309         /**
 310          * @see FileBackend::doClean()
 311          * @return Status
 312          */
 313         protected function doClean( array $params ) {
 314                 $status = Status::newGood();
 315                 foreach ( $this->backends as $backend ) {
 316                         $realParams = $this->substOpPaths( $params, $backend );
 317                         $status->merge( $backend->doClean( $realParams ) );
 318                 }
 319                 return $status;
 320         }
 321
 322         /**
 323          * @see FileBackend::getFileList()
 324          */
 325         public function concatenate( array $params ) {
 326                 // We are writing to an FS file, so we don't need to do this per-backend
 327                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 328                 return $this->backends[$this->masterIndex]->concatenate( $realParams );
 329         }
 330
 331         /**
 332          * @see FileBackend::fileExists()
 333          */
 334         public function fileExists( array $params ) {
 335                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 336                 return $this->backends[$this->masterIndex]->fileExists( $realParams );
 337         }
 338
 339         /**
 340          * @see FileBackend::getFileTimestamp()
 341          */
 342         public function getFileTimestamp( array $params ) {
 343                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 344                 return $this->backends[$this->masterIndex]->getFileTimestamp( $realParams );
 345         }
 346
 347         /**
 348          * @see FileBackend::getFileSize()
 349          */
 350         public function getFileSize( array $params ) {
 351                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 352                 return $this->backends[$this->masterIndex]->getFileSize( $realParams );
 353         }
 354
 355         /**
 356          * @see FileBackend::getFileStat()
 357          */
 358         public function getFileStat( array $params ) {
 359                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 360                 return $this->backends[$this->masterIndex]->getFileStat( $realParams );
 361         }
 362
 363         /**
 364          * @see FileBackend::getFileContents()
 365          */
 366         public function getFileContents( array $params ) {
 367                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 368                 return $this->backends[$this->masterIndex]->getFileContents( $realParams );
 369         }
 370
 371         /**
 372          * @see FileBackend::getFileSha1Base36()
 373          */
 374         public function getFileSha1Base36( array $params ) {
 375                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 376                 return $this->backends[$this->masterIndex]->getFileSha1Base36( $realParams );
 377         }
 378
 379         /**
 380          * @see FileBackend::getFileProps()
 381          */
 382         public function getFileProps( array $params ) {
 383                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 384                 return $this->backends[$this->masterIndex]->getFileProps( $realParams );
 385         }
 386
 387         /**
 388          * @see FileBackend::streamFile()
 389          */
 390         public function streamFile( array $params ) {
 391                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 392                 return $this->backends[$this->masterIndex]->streamFile( $realParams );
 393         }
 394
 395         /**
 396          * @see FileBackend::getLocalReference()
 397          */
 398         public function getLocalReference( array $params ) {
 399                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 400                 return $this->backends[$this->masterIndex]->getLocalReference( $realParams );
 401         }
 402
 403         /**
 404          * @see FileBackend::getLocalCopy()
 405          */
 406         public function getLocalCopy( array $params ) {
 407                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 408                 return $this->backends[$this->masterIndex]->getLocalCopy( $realParams );
 409         }
 410
 411         /**
 412          * @see FileBackend::getFileList()
 413          */
 414         public function getFileList( array $params ) {
 415                 $realParams = $this->substOpPaths( $params, $this->backends[$this->masterIndex] );
 416                 return $this->backends[$this->masterIndex]->getFileList( $realParams );
 417         }
 418
 419         /**
 420          * @see FileBackend::clearCache()
 421          */
 422         public function clearCache( array $paths = null ) {
 423                 foreach ( $this->backends as $backend ) {
 424                         $realPaths = is_array( $paths ) ? $this->substPaths( $paths, $backend ) : null;
 425                         $backend->clearCache( $realPaths );
 426                 }
 427         }
 428 }