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         /** @var array|null (path sha1 => stat) Pre-computed dst stat entries from listings */
  39         protected $statCache = null;
  40
  41         public function __construct() {
  42                 parent::__construct();
  43                 $this->addDescription( 'Copy files in one backend to another.' );
  44                 $this->addOption( 'src', 'Backend containing the source files', true, true );
  45                 $this->addOption( 'dst', 'Backend where files should be copied to', true, true );
  46                 $this->addOption( 'containers', 'Pipe separated list of containers', true, true );
  47                 $this->addOption( 'subdir', 'Only do items in this child directory', false, true );
  48                 $this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
  49                 $this->addOption( 'prestat', 'Stat the destination files first (try to use listings)' );
  50                 $this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
  51                 $this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
  52                 $this->addOption( 'syncviadelete', 'Delete destination files missing from source listing' );
  53                 $this->addOption( 'utf8only', 'Skip source files that do not have valid UTF-8 names' );
  54                 $this->setBatchSize( 50 );
  55         }
  56
  57         public function execute() {
  58                 $src = FileBackendGroup::singleton()->get( $this->getOption( 'src' ) );
  59                 $dst = FileBackendGroup::singleton()->get( $this->getOption( 'dst' ) );
  60                 $containers = explode( '|', $this->getOption( 'containers' ) );
  61                 $subDir = rtrim( $this->getOption( 'subdir', '' ), '/' );
  62
  63                 $rateFile = $this->getOption( 'ratefile' );
  64
  65                 foreach ( $containers as $container ) {
  66                         if ( $subDir != '' ) {
  67                                 $backendRel = "$container/$subDir";
  68                                 $this->output( "Doing container '$container', directory '$subDir'...\n" );
  69                         } else {
  70                                 $backendRel = $container;
  71                                 $this->output( "Doing container '$container'...\n" );
  72                         }
  73
  74                         if ( $this->hasOption( 'missingonly' ) ) {
  75                                 $this->output( "\tBuilding list of missing files..." );
  76                                 $srcPathsRel = $this->getListingDiffRel( $src, $dst, $backendRel );
  77                                 $this->output( count( $srcPathsRel ) . " file(s) need to be copied.\n" );
  78                         } else {
  79                                 $srcPathsRel = $src->getFileList( [
  80                                         'dir' => $src->getRootStoragePath() . "/$backendRel",
  81                                         'adviseStat' => true // avoid HEADs
  82                                 ] );
  83                                 if ( $srcPathsRel === null ) {
  84                                         $this->fatalError( "Could not list files in $container." );
  85                                 }
  86                         }
  87
  88                         if ( $this->getOption( 'prestat' ) && !$this->hasOption( 'missingonly' ) ) {
  89                                 // Build the stat cache for the destination files
  90                                 $this->output( "\tBuilding destination stat cache..." );
  91                                 $dstPathsRel = $dst->getFileList( [
  92                                         'dir' => $dst->getRootStoragePath() . "/$backendRel",
  93                                         'adviseStat' => true // avoid HEADs
  94                                 ] );
  95                                 if ( $dstPathsRel === null ) {
  96                                         $this->fatalError( "Could not list files in $container." );
  97                                 }
  98                                 $this->statCache = [];
  99                                 foreach ( $dstPathsRel as $dstPathRel ) {
 100                                         $path = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
 101                                         $this->statCache[sha1( $path )] = $dst->getFileStat( [ 'src' => $path ] );
 102                                 }
 103                                 $this->output( "done [" . count( $this->statCache ) . " file(s)]\n" );
 104                         }
 105
 106                         $this->output( "\tCopying file(s)...\n" );
 107                         $count = 0;
 108                         $batchPaths = [];
 109                         foreach ( $srcPathsRel as $srcPathRel ) {
 110                                 // Check up on the rate file periodically to adjust the concurrency
 111                                 if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
 112                                         $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
 113                                         $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
 114                                 }
 115                                 $batchPaths[$srcPathRel] = 1; // remove duplicates
 116                                 if ( count( $batchPaths ) >= $this->getBatchSize() ) {
 117                                         $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
 118                                         $batchPaths = []; // done
 119                                 }
 120                                 ++$count;
 121                         }
 122                         if ( count( $batchPaths ) ) { // left-overs
 123                                 $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
 124                                 $batchPaths = []; // done
 125                         }
 126                         $this->output( "\tCopied $count file(s).\n" );
 127
 128                         if ( $this->hasOption( 'syncviadelete' ) ) {
 129                                 $this->output( "\tBuilding list of excess destination files..." );
 130                                 $delPathsRel = $this->getListingDiffRel( $dst, $src, $backendRel );
 131                                 $this->output( count( $delPathsRel ) . " file(s) need to be deleted.\n" );
 132
 133                                 $this->output( "\tDeleting file(s)...\n" );
 134                                 $count = 0;
 135                                 $batchPaths = [];
 136                                 foreach ( $delPathsRel as $delPathRel ) {
 137                                         // Check up on the rate file periodically to adjust the concurrency
 138                                         if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
 139                                                 $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
 140                                                 $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
 141                                         }
 142                                         $batchPaths[$delPathRel] = 1; // remove duplicates
 143                                         if ( count( $batchPaths ) >= $this->getBatchSize() ) {
 144                                                 $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
 145                                                 $batchPaths = []; // done
 146                                         }
 147                                         ++$count;
 148                                 }
 149                                 if ( count( $batchPaths ) ) { // left-overs
 150                                         $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
 151                                         $batchPaths = []; // done
 152                                 }
 153
 154                                 $this->output( "\tDeleted $count file(s).\n" );
 155                         }
 156
 157                         if ( $subDir != '' ) {
 158                                 $this->output( "Finished container '$container', directory '$subDir'.\n" );
 159                         } else {
 160                                 $this->output( "Finished container '$container'.\n" );
 161                         }
 162                 }
 163
 164                 $this->output( "Done.\n" );
 165         }
 166
 167         /**
 168          * @param FileBackend $src
 169          * @param FileBackend $dst
 170          * @param string $backendRel
 171          * @return array (rel paths in $src minus those in $dst)
 172          */
 173         protected function getListingDiffRel( FileBackend $src, FileBackend $dst, $backendRel ) {
 174                 $srcPathsRel = $src->getFileList( [
 175                         'dir' => $src->getRootStoragePath() . "/$backendRel" ] );
 176                 if ( $srcPathsRel === null ) {
 177                         $this->fatalError( "Could not list files in source container." );
 178                 }
 179                 $dstPathsRel = $dst->getFileList( [
 180                         'dir' => $dst->getRootStoragePath() . "/$backendRel" ] );
 181                 if ( $dstPathsRel === null ) {
 182                         $this->fatalError( "Could not list files in destination container." );
 183                 }
 184                 // Get the list of destination files
 185                 $relFilesDstSha1 = [];
 186                 foreach ( $dstPathsRel as $dstPathRel ) {
 187                         $relFilesDstSha1[sha1( $dstPathRel )] = 1;
 188                 }
 189                 unset( $dstPathsRel ); // free
 190                 // Get the list of missing files
 191                 $missingPathsRel = [];
 192                 foreach ( $srcPathsRel as $srcPathRel ) {
 193                         if ( !isset( $relFilesDstSha1[sha1( $srcPathRel )] ) ) {
 194                                 $missingPathsRel[] = $srcPathRel;
 195                         }
 196                 }
 197                 unset( $srcPathsRel ); // free
 198
 199                 return $missingPathsRel;
 200         }
 201
 202         /**
 203          * @param array $srcPathsRel
 204          * @param string $backendRel
 205          * @param FileBackend $src
 206          * @param FileBackend $dst
 207          * @return void
 208          */
 209         protected function copyFileBatch(
 210                 array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst
 211         ) {
 212                 $ops = [];
 213                 $fsFiles = [];
 214                 $copiedRel = []; // for output message
 215                 $wikiId = $src->getWikiId();
 216
 217                 // Download the batch of source files into backend cache...
 218                 if ( $this->hasOption( 'missingonly' ) ) {
 219                         $srcPaths = [];
 220                         foreach ( $srcPathsRel as $srcPathRel ) {
 221                                 $srcPaths[] = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 222                         }
 223                         $t_start = microtime( true );
 224                         $fsFiles = $src->getLocalReferenceMulti( [ 'srcs' => $srcPaths, 'latest' => 1 ] );
 225                         $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 226                         $this->output( "\n\tDownloaded these file(s) [{$elapsed_ms}ms]:\n\t" .
 227                                 implode( "\n\t", $srcPaths ) . "\n\n" );
 228                 }
 229
 230                 // Determine what files need to be copied over...
 231                 foreach ( $srcPathsRel as $srcPathRel ) {
 232                         $srcPath = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 233                         $dstPath = $dst->getRootStoragePath() . "/$backendRel/$srcPathRel";
 234                         if ( $this->hasOption( 'utf8only' ) && !mb_check_encoding( $srcPath, 'UTF-8' ) ) {
 235                                 $this->error( "$wikiId: Detected illegal (non-UTF8) path for $srcPath." );
 236                                 continue;
 237                         } elseif ( !$this->hasOption( 'missingonly' )
 238                                 && $this->filesAreSame( $src, $dst, $srcPath, $dstPath )
 239                         ) {
 240                                 $this->output( "\tAlready have $srcPathRel.\n" );
 241                                 continue; // assume already copied...
 242                         }
 243                         $fsFile = array_key_exists( $srcPath, $fsFiles )
 244                                 ? $fsFiles[$srcPath]
 245                                 : $src->getLocalReference( [ 'src' => $srcPath, 'latest' => 1 ] );
 246                         if ( !$fsFile ) {
 247                                 $src->clearCache( [ $srcPath ] );
 248                                 if ( $src->fileExists( [ 'src' => $srcPath, 'latest' => 1 ] ) === false ) {
 249                                         $this->error( "$wikiId: File '$srcPath' was listed but does not exist." );
 250                                 } else {
 251                                         $this->error( "$wikiId: Could not get local copy of $srcPath." );
 252                                 }
 253                                 continue;
 254                         } elseif ( !$fsFile->exists() ) {
 255                                 // FSFileBackends just return the path for getLocalReference() and paths with
 256                                 // illegal slashes may get normalized to a different path. This can cause the
 257                                 // local reference to not exist...skip these broken files.
 258                                 $this->error( "$wikiId: Detected possible illegal path for $srcPath." );
 259                                 continue;
 260                         }
 261                         $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed
 262                         // Note: prepare() is usually fast for key/value backends
 263                         $status = $dst->prepare( [ 'dir' => dirname( $dstPath ), 'bypassReadOnly' => 1 ] );
 264                         if ( !$status->isOK() ) {
 265                                 $this->error( print_r( $status->getErrorsArray(), true ) );
 266                                 $this->fatalError( "$wikiId: Could not copy $srcPath to $dstPath." );
 267                         }
 268                         $ops[] = [ 'op' => 'store',
 269                                 'src' => $fsFile->getPath(), 'dst' => $dstPath, 'overwrite' => 1 ];
 270                         $copiedRel[] = $srcPathRel;
 271                 }
 272
 273                 // Copy in the batch of source files...
 274                 $t_start = microtime( true );
 275                 $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
 276                 if ( !$status->isOK() ) {
 277                         sleep( 10 ); // wait and retry copy again
 278                         $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
 279                 }
 280                 $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 281                 if ( !$status->isOK() ) {
 282                         $this->error( print_r( $status->getErrorsArray(), true ) );
 283                         $this->fatalError( "$wikiId: Could not copy file batch." );
 284                 } elseif ( count( $copiedRel ) ) {
 285                         $this->output( "\n\tCopied these file(s) [{$elapsed_ms}ms]:\n\t" .
 286                                 implode( "\n\t", $copiedRel ) . "\n\n" );
 287                 }
 288         }
 289
 290         /**
 291          * @param array $dstPathsRel
 292          * @param string $backendRel
 293          * @param FileBackend $dst
 294          * @return void
 295          */
 296         protected function delFileBatch(
 297                 array $dstPathsRel, $backendRel, FileBackend $dst
 298         ) {
 299                 $ops = [];
 300                 $deletedRel = []; // for output message
 301                 $wikiId = $dst->getWikiId();
 302
 303                 // Determine what files need to be copied over...
 304                 foreach ( $dstPathsRel as $dstPathRel ) {
 305                         $dstPath = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
 306                         $ops[] = [ 'op' => 'delete', 'src' => $dstPath ];
 307                         $deletedRel[] = $dstPathRel;
 308                 }
 309
 310                 // Delete the batch of source files...
 311                 $t_start = microtime( true );
 312                 $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
 313                 if ( !$status->isOK() ) {
 314                         sleep( 10 ); // wait and retry copy again
 315                         $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
 316                 }
 317                 $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 318                 if ( !$status->isOK() ) {
 319                         $this->error( print_r( $status->getErrorsArray(), true ) );
 320                         $this->fatalError( "$wikiId: Could not delete file batch." );
 321                 } elseif ( count( $deletedRel ) ) {
 322                         $this->output( "\n\tDeleted these file(s) [{$elapsed_ms}ms]:\n\t" .
 323                                 implode( "\n\t", $deletedRel ) . "\n\n" );
 324                 }
 325         }
 326
 327         /**
 328          * @param FileBackend $src
 329          * @param FileBackend $dst
 330          * @param string $sPath
 331          * @param string $dPath
 332          * @return bool
 333          */
 334         protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) {
 335                 $skipHash = $this->hasOption( 'skiphash' );
 336                 $srcStat = $src->getFileStat( [ 'src' => $sPath ] );
 337                 $dPathSha1 = sha1( $dPath );
 338                 if ( $this->statCache !== null ) {
 339                         // All dst files are already in stat cache
 340                         $dstStat = isset( $this->statCache[$dPathSha1] )
 341                                 ? $this->statCache[$dPathSha1]
 342                                 : false;
 343                 } else {
 344                         $dstStat = $dst->getFileStat( [ 'src' => $dPath ] );
 345                 }
 346                 // Initial fast checks to see if files are obviously different
 347                 $sameFast = (
 348                         is_array( $srcStat ) // sanity check that source exists
 349                         && is_array( $dstStat ) // dest exists
 350                         && $srcStat['size'] === $dstStat['size']
 351                 );
 352                 // More thorough checks against files
 353                 if ( !$sameFast ) {
 354                         $same = false; // no need to look farther
 355                 } elseif ( isset( $srcStat['md5'] ) && isset( $dstStat['md5'] ) ) {
 356                         // If MD5 was already in the stat info, just use it.
 357                         // This is useful as many objects stores can return this in object listing,
 358                         // so we can use it to avoid slow per-file HEADs.
 359                         $same = ( $srcStat['md5'] === $dstStat['md5'] );
 360                 } elseif ( $skipHash ) {
 361                         // This mode is good for copying to a backup location or resyncing clone
 362                         // backends in FileBackendMultiWrite (since they get writes second, they have
 363                         // higher timestamps). However, when copying the other way, this hits loads of
 364                         // false positives (possibly 100%) and wastes a bunch of time on GETs/PUTs.
 365                         $same = ( $srcStat['mtime'] <= $dstStat['mtime'] );
 366                 } else {
 367                         // This is the slowest method which does many per-file HEADs (unless an object
 368                         // store tracks SHA-1 in listings).
 369                         $same = ( $src->getFileSha1Base36( [ 'src' => $sPath, 'latest' => 1 ] )
 370                                 === $dst->getFileSha1Base36( [ 'src' => $dPath, 'latest' => 1 ] ) );
 371                 }
 372
 373                 return $same;
 374         }
 375 }
 376
 377 $maintClass = CopyFileBackend::class;
 378 require_once RUN_MAINTENANCE_IF_MAIN;