maintenance/copyFileBackend.php

   1 <?php
   2 /**
   3  * Copy all files in some containers of one backend to another.
   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 Maintenance
  22  */
  23
  24 require_once __DIR__ . '/Maintenance.php';
  25
  26 /**
  27  * Copy all files in one container of one backend to another.
  28  *
  29  * This can also be used to re-shard the files for one backend using the
  30  * config of second backend. The second backend should have the same config
  31  * as the first, except for it having a different name and different sharding
  32  * configuration. The backend should be made read-only while this runs.
  33  * After this script finishes, the old files in the containers can be deleted.
  34  *
  35  * @ingroup Maintenance
  36  */
  37 class CopyFileBackend extends Maintenance {
  38         protected $statCache = array();
  39
  40         public function __construct() {
  41                 parent::__construct();
  42                 $this->mDescription = "Copy files in one backend to another.";
  43                 $this->addOption( 'src', 'Backend containing the source files', true, true );
  44                 $this->addOption( 'dst', 'Backend where files should be copied to', true, true );
  45                 $this->addOption( 'containers', 'Pipe separated list of containers', true, true );
  46                 $this->addOption( 'subdir', 'Only do items in this child directory', false, true );
  47                 $this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
  48                 $this->addOption( 'prestat', 'Stat the destination files first (try to use listings)' );
  49                 $this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
  50                 $this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
  51                 $this->addOption( 'syncviadelete', 'Delete destination files missing from source listing' );
  52                 $this->addOption( 'utf8only', 'Skip source files that do not have valid UTF-8 names' );
  53                 $this->setBatchSize( 50 );
  54         }
  55
  56         public function execute() {
  57                 $src = FileBackendGroup::singleton()->get( $this->getOption( 'src' ) );
  58                 $dst = FileBackendGroup::singleton()->get( $this->getOption( 'dst' ) );
  59                 $containers = explode( '|', $this->getOption( 'containers' ) );
  60                 $subDir = rtrim( $this->getOption( 'subdir', '' ), '/' );
  61
  62                 $rateFile = $this->getOption( 'ratefile' );
  63
  64                 if ( $this->hasOption( 'utf8only' ) && !extension_loaded( 'mbstring' ) ) {
  65                         $this->error( "Cannot check for UTF-8, mbstring extension missing.", 1 ); // die
  66                 }
  67
  68                 foreach ( $containers as $container ) {
  69                         if ( $subDir != '' ) {
  70                                 $backendRel = "$container/$subDir";
  71                                 $this->output( "Doing container '$container', directory '$subDir'...\n" );
  72                         } else {
  73                                 $backendRel = $container;
  74                                 $this->output( "Doing container '$container'...\n" );
  75                         }
  76
  77                         if ( $this->hasOption( 'missingonly' ) ) {
  78                                 $this->output( "\tBuilding list of missing files..." );
  79                                 $srcPathsRel = $this->getListingDiffRel( $src, $dst, $backendRel );
  80                                 $this->output( count( $srcPathsRel ) . " file(s) need to be copied.\n" );
  81                         } else {
  82                                 $srcPathsRel = $src->getFileList( array(
  83                                         'dir' => $src->getRootStoragePath() . "/$backendRel",
  84                                         'adviseStat' => true // avoid HEADs
  85                                 ) );
  86                                 if ( $srcPathsRel === null ) {
  87                                         $this->error( "Could not list files in $container.", 1 ); // die
  88                                 }
  89                         }
  90
  91                         if ( $this->getOption( 'prestat' ) && !$this->hasOption( 'missingonly' ) ) {
  92                                 // Build the stat cache for the destination files
  93                                 $this->output( "\tBuilding destination stat cache..." );
  94                                 $dstPathsRel = $dst->getFileList( array(
  95                                         'dir' => $dst->getRootStoragePath() . "/$backendRel",
  96                                         'adviseStat' => true // avoid HEADs
  97                                 ) );
  98                                 if ( $dstPathsRel === null ) {
  99                                         $this->error( "Could not list files in $container.", 1 ); // die
 100                                 }
 101                                 $this->statCache = array(); // clear
 102                                 foreach ( $dstPathsRel as $dstPathRel ) {
 103                                         $path = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
 104                                         $this->statCache[sha1( $path )] = $dst->getFileStat( array( 'src' => $path ) );
 105                                 }
 106                                 $this->output( "done [" . count( $this->statCache ) . " file(s)]\n" );
 107                         }
 108
 109                         $this->output( "\tCopying file(s)...\n" );
 110                         $count = 0;
 111                         $batchPaths = array();
 112                         foreach ( $srcPathsRel as $srcPathRel ) {
 113                                 // Check up on the rate file periodically to adjust the concurrency
 114                                 if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
 115                                         $this->mBatchSize = max( 1, (int)file_get_contents( $rateFile ) );
 116                                         $this->output( "\tBatch size is now {$this->mBatchSize}.\n" );
 117                                 }
 118                                 $batchPaths[$srcPathRel] = 1; // remove duplicates
 119                                 if ( count( $batchPaths ) >= $this->mBatchSize ) {
 120                                         $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
 121                                         $batchPaths = array(); // done
 122                                 }
 123                                 ++$count;
 124                         }
 125                         if ( count( $batchPaths ) ) { // left-overs
 126                                 $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
 127                                 $batchPaths = array(); // done
 128                         }
 129                         $this->output( "\tCopied $count file(s).\n" );
 130
 131                         if ( $this->hasOption( 'syncviadelete' ) ) {
 132                                 $this->output( "\tBuilding list of excess destination files..." );
 133                                 $delPathsRel = $this->getListingDiffRel( $dst, $src, $backendRel );
 134                                 $this->output( count( $delPathsRel ) . " file(s) need to be deleted.\n" );
 135
 136                                 $this->output( "\tDeleting file(s)...\n" );
 137                                 $count = 0;
 138                                 $batchPaths = array();
 139                                 foreach ( $delPathsRel as $delPathRel ) {
 140                                         // Check up on the rate file periodically to adjust the concurrency
 141                                         if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
 142                                                 $this->mBatchSize = max( 1, (int)file_get_contents( $rateFile ) );
 143                                                 $this->output( "\tBatch size is now {$this->mBatchSize}.\n" );
 144                                         }
 145                                         $batchPaths[$delPathRel] = 1; // remove duplicates
 146                                         if ( count( $batchPaths ) >= $this->mBatchSize ) {
 147                                                 $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
 148                                                 $batchPaths = array(); // done
 149                                         }
 150                                         ++$count;
 151                                 }
 152                                 if ( count( $batchPaths ) ) { // left-overs
 153                                         $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
 154                                         $batchPaths = array(); // done
 155                                 }
 156
 157                                 $this->output( "\tDeleted $count file(s).\n" );
 158                         }
 159
 160                         if ( $subDir != '' ) {
 161                                 $this->output( "Finished container '$container', directory '$subDir'.\n" );
 162                         } else {
 163                                 $this->output( "Finished container '$container'.\n" );
 164                         }
 165                 }
 166
 167                 $this->output( "Done.\n" );
 168         }
 169
 170         /**
 171          * @param FileBackend $src
 172          * @param FileBackend $dst
 173          * @param string $backendRel
 174          * @return array (rel paths in $src minus those in $dst)
 175          */
 176         protected function getListingDiffRel( FileBackend $src, FileBackend $dst, $backendRel ) {
 177                 $srcPathsRel = $src->getFileList( array(
 178                         'dir' => $src->getRootStoragePath() . "/$backendRel" ) );
 179                 if ( $srcPathsRel === null ) {
 180                         $this->error( "Could not list files in source container.", 1 ); // die
 181                 }
 182                 $dstPathsRel = $dst->getFileList( array(
 183                         'dir' => $dst->getRootStoragePath() . "/$backendRel" ) );
 184                 if ( $dstPathsRel === null ) {
 185                         $this->error( "Could not list files in destination container.", 1 ); // die
 186                 }
 187                 // Get the list of destination files
 188                 $relFilesDstSha1 = array();
 189                 foreach ( $dstPathsRel as $dstPathRel ) {
 190                         $relFilesDstSha1[sha1( $dstPathRel )] = 1;
 191                 }
 192                 unset( $dstPathsRel ); // free
 193                 // Get the list of missing files
 194                 $missingPathsRel = array();
 195                 foreach ( $srcPathsRel as $srcPathRel ) {
 196                         if ( !isset( $relFilesDstSha1[sha1( $srcPathRel )] ) ) {
 197                                 $missingPathsRel[] = $srcPathRel;
 198                         }
 199                 }
 200                 unset( $srcPathsRel ); // free
 201
 202                 return $missingPathsRel;
 203         }
 204
 205         /**
 206          * @param array $srcPathsRel
 207          * @param string $backendRel
 208          * @param FileBackend $src
 209          * @param FileBackend $dst
 210          * @return void
 211          */
 212         protected function copyFileBatch(
 213                 array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst
 214         ) {
 215                 $ops = array();
 216                 $fsFiles = array();
 217                 $copiedRel = array(); // for output message
 218                 $wikiId = $src->getWikiId();
 219
 220                 // Download the batch of source files into backend cache...
 221                 if ( $this->hasOption( 'missingonly' ) ) {
 222                         $srcPaths = array();
 223                         foreach ( $srcPathsRel as $srcPathRel ) {
 224                                 $srcPaths[] = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 225                         }
 226                         $t_start = microtime( true );
 227                         $fsFiles = $src->getLocalReferenceMulti( array( 'srcs' => $srcPaths, 'latest' => 1 ) );
 228                         $ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 229                         $this->output( "\n\tDownloaded these file(s) [{$ellapsed_ms}ms]:\n\t" .
 230                                 implode( "\n\t", $srcPaths ) . "\n\n" );
 231                 }
 232
 233                 // Determine what files need to be copied over...
 234                 foreach ( $srcPathsRel as $srcPathRel ) {
 235                         $srcPath = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 236                         $dstPath = $dst->getRootStoragePath() . "/$backendRel/$srcPathRel";
 237                         if ( $this->hasOption( 'utf8only' ) && !mb_check_encoding( $srcPath, 'UTF-8' ) ) {
 238                                 $this->error( "$wikiId: Detected illegal (non-UTF8) path for $srcPath." );
 239                                 continue;
 240                         } elseif ( !$this->hasOption( 'missingonly' )
 241                                 && $this->filesAreSame( $src, $dst, $srcPath, $dstPath ) )
 242                         {
 243                                 $this->output( "\tAlready have $srcPathRel.\n" );
 244                                 continue; // assume already copied...
 245                         }
 246                         $fsFile = array_key_exists( $srcPath, $fsFiles )
 247                                 ? $fsFiles[$srcPath]
 248                                 : $src->getLocalReference( array( 'src' => $srcPath, 'latest' => 1 ) );
 249                         if ( !$fsFile ) {
 250                                 $src->clearCache( array( $srcPath ) );
 251                                 if ( $src->fileExists( array( 'src' => $srcPath, 'latest' => 1 ) ) === false ) {
 252                                         $this->error( "$wikiId: File '$srcPath' was listed but does not exist." );
 253                                 } else {
 254                                         $this->error( "$wikiId: Could not get local copy of $srcPath." );
 255                                 }
 256                                 continue;
 257                         } elseif ( !$fsFile->exists() ) {
 258                                 // FSFileBackends just return the path for getLocalReference() and paths with
 259                                 // illegal slashes may get normalized to a different path. This can cause the
 260                                 // local reference to not exist...skip these broken files.
 261                                 $this->error( "$wikiId: Detected possible illegal path for $srcPath." );
 262                                 continue;
 263                         }
 264                         $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed
 265                         // Note: prepare() is usually fast for key/value backends
 266                         $status = $dst->prepare( array( 'dir' => dirname( $dstPath ), 'bypassReadOnly' => 1 ) );
 267                         if ( !$status->isOK() ) {
 268                                 $this->error( print_r( $status->getErrorsArray(), true ) );
 269                                 $this->error( "$wikiId: Could not copy $srcPath to $dstPath.", 1 ); // die
 270                         }
 271                         $ops[] = array( 'op' => 'store',
 272                                 'src' => $fsFile->getPath(), 'dst' => $dstPath, 'overwrite' => 1 );
 273                         $copiedRel[] = $srcPathRel;
 274                 }
 275
 276                 // Copy in the batch of source files...
 277                 $t_start = microtime( true );
 278                 $status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) );
 279                 if ( !$status->isOK() ) {
 280                         sleep( 10 ); // wait and retry copy again
 281                         $status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) );
 282                 }
 283                 $ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 284                 if ( !$status->isOK() ) {
 285                         $this->error( print_r( $status->getErrorsArray(), true ) );
 286                         $this->error( "$wikiId: Could not copy file batch.", 1 ); // die
 287                 } elseif ( count( $copiedRel ) ) {
 288                         $this->output( "\n\tCopied these file(s) [{$ellapsed_ms}ms]:\n\t" .
 289                                 implode( "\n\t", $copiedRel ) . "\n\n" );
 290                 }
 291         }
 292
 293         /**
 294          * @param array $dstPathsRel
 295          * @param string $backendRel
 296          * @param FileBackend $dst
 297          * @return void
 298          */
 299         protected function delFileBatch(
 300                 array $dstPathsRel, $backendRel, FileBackend $dst
 301         ) {
 302                 $ops = array();
 303                 $deletedRel = array(); // for output message
 304                 $wikiId = $dst->getWikiId();
 305
 306                 // Determine what files need to be copied over...
 307                 foreach ( $dstPathsRel as $dstPathRel ) {
 308                         $dstPath = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
 309                         $ops[] = array( 'op' => 'delete', 'src' => $dstPath );
 310                         $deletedRel[] = $dstPathRel;
 311                 }
 312
 313                 // Delete the batch of source files...
 314                 $t_start = microtime( true );
 315                 $status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) );
 316                 if ( !$status->isOK() ) {
 317                         sleep( 10 ); // wait and retry copy again
 318                         $status = $dst->doQuickOperations( $ops, array( 'bypassReadOnly' => 1 ) );
 319                 }
 320                 $ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 321                 if ( !$status->isOK() ) {
 322                         $this->error( print_r( $status->getErrorsArray(), true ) );
 323                         $this->error( "$wikiId: Could not delete file batch.", 1 ); // die
 324                 } elseif ( count( $deletedRel ) ) {
 325                         $this->output( "\n\tDeleted these file(s) [{$ellapsed_ms}ms]:\n\t" .
 326                                 implode( "\n\t", $deletedRel ) . "\n\n" );
 327                 }
 328         }
 329
 330         /**
 331          * @param FileBackend $src
 332          * @param FileBackend $dst
 333          * @param string $sPath
 334          * @param string $dPath
 335          * @return bool
 336          */
 337         protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) {
 338                 $skipHash = $this->hasOption( 'skiphash' );
 339                 $srcStat = $src->getFileStat( array( 'src' => $sPath ) );
 340                 $dPathSha1 = sha1( $dPath );
 341                 $dstStat = isset( $this->statCache[$dPathSha1] )
 342                         ? $this->statCache[$dPathSha1]
 343                         : $dst->getFileStat( array( 'src' => $dPath ) );
 344                 return (
 345                         is_array( $srcStat ) // sanity check that source exists
 346                         && is_array( $dstStat ) // dest exists
 347                         && $srcStat['size'] === $dstStat['size']
 348                         && ( !$skipHash || $srcStat['mtime'] <= $dstStat['mtime'] )
 349                         && ( $skipHash || $src->getFileSha1Base36( array( 'src' => $sPath, 'latest' => 1 ) )
 350                                 === $dst->getFileSha1Base36( array( 'src' => $dPath, 'latest' => 1 ) )
 351                         )
 352                 );
 353         }
 354 }
 355
 356 $maintClass = 'CopyFileBackend';
 357 require_once RUN_MAINTENANCE_IF_MAIN;