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 use MediaWiki\Maintenance\Maintenance;
  25 use Wikimedia\FileBackend\FileBackend;
  26
  27 // @codeCoverageIgnoreStart
  28 require_once __DIR__ . '/Maintenance.php';
  29 // @codeCoverageIgnoreEnd
  30
  31 /**
  32  * Copy all files in one container of one backend to another.
  33  *
  34  * This can also be used to re-shard the files for one backend using the
  35  * config of second backend. The second backend should have the same config
  36  * as the first, except for it having a different name and different sharding
  37  * configuration. The backend should be made read-only while this runs.
  38  * After this script finishes, the old files in the containers can be deleted.
  39  *
  40  * @ingroup Maintenance
  41  */
  42 class CopyFileBackend extends Maintenance {
  43         /** @var array|null (path sha1 => stat) Pre-computed dst stat entries from listings */
  44         protected $statCache = null;
  45
  46         public function __construct() {
  47                 parent::__construct();
  48                 $this->addDescription( 'Copy files in one backend to another.' );
  49                 $this->addOption( 'src', 'Backend containing the source files', true, true );
  50                 $this->addOption( 'dst', 'Backend where files should be copied to', true, true );
  51                 $this->addOption( 'containers', 'Pipe separated list of containers', true, true );
  52                 $this->addOption( 'subdir', 'Only do items in this child directory', false, true );
  53                 $this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
  54                 $this->addOption( 'prestat', 'Stat the destination files first (try to use listings)' );
  55                 $this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
  56                 $this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
  57                 $this->addOption( 'syncviadelete', 'Delete destination files missing from source listing' );
  58                 $this->addOption( 'utf8only', 'Skip source files that do not have valid UTF-8 names' );
  59                 $this->setBatchSize( 50 );
  60         }
  61
  62         public function execute() {
  63                 $backendGroup = $this->getServiceContainer()->getFileBackendGroup();
  64                 $src = $backendGroup->get( $this->getOption( 'src' ) );
  65                 $dst = $backendGroup->get( $this->getOption( 'dst' ) );
  66                 $containers = explode( '|', $this->getOption( 'containers' ) );
  67                 $subDir = rtrim( $this->getOption( 'subdir', '' ), '/' );
  68
  69                 $rateFile = $this->getOption( 'ratefile' );
  70
  71                 foreach ( $containers as $container ) {
  72                         if ( $subDir != '' ) {
  73                                 $backendRel = "$container/$subDir";
  74                                 $this->output( "Doing container '$container', directory '$subDir'...\n" );
  75                         } else {
  76                                 $backendRel = $container;
  77                                 $this->output( "Doing container '$container'...\n" );
  78                         }
  79
  80                         if ( $this->hasOption( 'missingonly' ) ) {
  81                                 $this->output( "\tBuilding list of missing files..." );
  82                                 $srcPathsRel = $this->getListingDiffRel( $src, $dst, $backendRel );
  83                                 $this->output( count( $srcPathsRel ) . " file(s) need to be copied.\n" );
  84                         } else {
  85                                 $srcPathsRel = $src->getFileList( [
  86                                         'dir' => $src->getRootStoragePath() . "/$backendRel",
  87                                         'adviseStat' => true // avoid HEADs
  88                                 ] );
  89                                 if ( $srcPathsRel === null ) {
  90                                         $this->fatalError( "Could not list files in $container." );
  91                                 }
  92                         }
  93
  94                         if ( $this->getOption( 'prestat' ) && !$this->hasOption( 'missingonly' ) ) {
  95                                 // Build the stat cache for the destination files
  96                                 $this->output( "\tBuilding destination stat cache..." );
  97                                 $dstPathsRel = $dst->getFileList( [
  98                                         'dir' => $dst->getRootStoragePath() . "/$backendRel",
  99                                         'adviseStat' => true // avoid HEADs
 100                                 ] );
 101                                 if ( $dstPathsRel === null ) {
 102                                         $this->fatalError( "Could not list files in $container." );
 103                                 }
 104                                 $this->statCache = [];
 105                                 foreach ( $dstPathsRel as $dstPathRel ) {
 106                                         $path = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
 107                                         $this->statCache[sha1( $path )] = $dst->getFileStat( [ 'src' => $path ] );
 108                                 }
 109                                 $this->output( "done [" . count( $this->statCache ) . " file(s)]\n" );
 110                         }
 111
 112                         $this->output( "\tCopying file(s)...\n" );
 113                         $count = 0;
 114                         $batchPaths = [];
 115                         foreach ( $srcPathsRel as $srcPathRel ) {
 116                                 // Check up on the rate file periodically to adjust the concurrency
 117                                 if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
 118                                         $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
 119                                         $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
 120                                 }
 121                                 $batchPaths[$srcPathRel] = 1; // remove duplicates
 122                                 if ( count( $batchPaths ) >= $this->getBatchSize() ) {
 123                                         $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
 124                                         $batchPaths = []; // done
 125                                 }
 126                                 ++$count;
 127                         }
 128                         if ( count( $batchPaths ) ) { // left-overs
 129                                 $this->copyFileBatch( array_keys( $batchPaths ), $backendRel, $src, $dst );
 130                         }
 131                         $this->output( "\tCopied $count file(s).\n" );
 132
 133                         if ( $this->hasOption( 'syncviadelete' ) ) {
 134                                 $this->output( "\tBuilding list of excess destination files..." );
 135                                 $delPathsRel = $this->getListingDiffRel( $dst, $src, $backendRel );
 136                                 $this->output( count( $delPathsRel ) . " file(s) need to be deleted.\n" );
 137
 138                                 $this->output( "\tDeleting file(s)...\n" );
 139                                 $count = 0;
 140                                 $batchPaths = [];
 141                                 foreach ( $delPathsRel as $delPathRel ) {
 142                                         // Check up on the rate file periodically to adjust the concurrency
 143                                         if ( $rateFile && ( !$count || ( $count % 500 ) == 0 ) ) {
 144                                                 $this->setBatchSize( max( 1, (int)file_get_contents( $rateFile ) ) );
 145                                                 $this->output( "\tBatch size is now {$this->getBatchSize()}.\n" );
 146                                         }
 147                                         $batchPaths[$delPathRel] = 1; // remove duplicates
 148                                         if ( count( $batchPaths ) >= $this->getBatchSize() ) {
 149                                                 $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
 150                                                 $batchPaths = []; // done
 151                                         }
 152                                         ++$count;
 153                                 }
 154                                 if ( count( $batchPaths ) ) { // left-overs
 155                                         $this->delFileBatch( array_keys( $batchPaths ), $backendRel, $dst );
 156                                 }
 157
 158                                 $this->output( "\tDeleted $count file(s).\n" );
 159                         }
 160
 161                         if ( $subDir != '' ) {
 162                                 $this->output( "Finished container '$container', directory '$subDir'.\n" );
 163                         } else {
 164                                 $this->output( "Finished container '$container'.\n" );
 165                         }
 166                 }
 167
 168                 $this->output( "Done.\n" );
 169         }
 170
 171         /**
 172          * @param FileBackend $src
 173          * @param FileBackend $dst
 174          * @param string $backendRel
 175          * @return string[] (rel paths in $src minus those in $dst)
 176          */
 177         protected function getListingDiffRel( FileBackend $src, FileBackend $dst, $backendRel ) {
 178                 $srcPathsRel = $src->getFileList( [
 179                         'dir' => $src->getRootStoragePath() . "/$backendRel" ] );
 180                 if ( $srcPathsRel === null ) {
 181                         $this->fatalError( "Could not list files in source container." );
 182                 }
 183                 $dstPathsRel = $dst->getFileList( [
 184                         'dir' => $dst->getRootStoragePath() . "/$backendRel" ] );
 185                 if ( $dstPathsRel === null ) {
 186                         $this->fatalError( "Could not list files in destination container." );
 187                 }
 188                 // Get the list of destination files
 189                 $relFilesDstSha1 = [];
 190                 foreach ( $dstPathsRel as $dstPathRel ) {
 191                         $relFilesDstSha1[sha1( $dstPathRel )] = 1;
 192                 }
 193                 unset( $dstPathsRel ); // free
 194                 // Get the list of missing files
 195                 $missingPathsRel = [];
 196                 foreach ( $srcPathsRel as $srcPathRel ) {
 197                         if ( !isset( $relFilesDstSha1[sha1( $srcPathRel )] ) ) {
 198                                 $missingPathsRel[] = $srcPathRel;
 199                         }
 200                 }
 201                 unset( $srcPathsRel ); // free
 202
 203                 return $missingPathsRel;
 204         }
 205
 206         /**
 207          * @param string[] $srcPathsRel
 208          * @param string $backendRel
 209          * @param FileBackend $src
 210          * @param FileBackend $dst
 211          * @return void
 212          */
 213         protected function copyFileBatch(
 214                 array $srcPathsRel, $backendRel, FileBackend $src, FileBackend $dst
 215         ) {
 216                 $ops = [];
 217                 $fsFiles = [];
 218                 $copiedRel = []; // for output message
 219                 $domainId = $src->getDomainId();
 220
 221                 // Download the batch of source files into backend cache...
 222                 if ( $this->hasOption( 'missingonly' ) ) {
 223                         $srcPaths = [];
 224                         foreach ( $srcPathsRel as $srcPathRel ) {
 225                                 $srcPaths[] = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 226                         }
 227                         $t_start = microtime( true );
 228                         $fsFiles = $src->getLocalReferenceMulti( [ 'srcs' => $srcPaths, 'latest' => 1 ] );
 229                         $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 230                         $this->output( "\n\tDownloaded these file(s) [{$elapsed_ms}ms]:\n\t" .
 231                                 implode( "\n\t", $srcPaths ) . "\n\n" );
 232                 }
 233
 234                 // Determine what files need to be copied over...
 235                 foreach ( $srcPathsRel as $srcPathRel ) {
 236                         $srcPath = $src->getRootStoragePath() . "/$backendRel/$srcPathRel";
 237                         $dstPath = $dst->getRootStoragePath() . "/$backendRel/$srcPathRel";
 238                         if ( $this->hasOption( 'utf8only' ) && !mb_check_encoding( $srcPath, 'UTF-8' ) ) {
 239                                 $this->error( "$domainId: Detected illegal (non-UTF8) path for $srcPath." );
 240                                 continue;
 241                         } elseif ( !$this->hasOption( 'missingonly' )
 242                                 && $this->filesAreSame( $src, $dst, $srcPath, $dstPath )
 243                         ) {
 244                                 $this->output( "\tAlready have $srcPathRel.\n" );
 245                                 continue; // assume already copied...
 246                         }
 247                         $fsFile = array_key_exists( $srcPath, $fsFiles )
 248                                 ? $fsFiles[$srcPath]
 249                                 : $src->getLocalReference( [ 'src' => $srcPath, 'latest' => 1 ] );
 250                         if ( !$fsFile ) {
 251                                 $src->clearCache( [ $srcPath ] );
 252                                 if ( $src->fileExists( [ 'src' => $srcPath, 'latest' => 1 ] ) === false ) {
 253                                         $this->error( "$domainId: File '$srcPath' was listed but does not exist." );
 254                                 } else {
 255                                         $this->error( "$domainId: Could not get local copy of $srcPath." );
 256                                 }
 257                                 continue;
 258                         } elseif ( !$fsFile->exists() ) {
 259                                 // FSFileBackends just return the path for getLocalReference() and paths with
 260                                 // illegal slashes may get normalized to a different path. This can cause the
 261                                 // local reference to not exist...skip these broken files.
 262                                 $this->error( "$domainId: Detected possible illegal path for $srcPath." );
 263                                 continue;
 264                         }
 265                         $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed
 266                         // Note: prepare() is usually fast for key/value backends
 267                         $status = $dst->prepare( [ 'dir' => dirname( $dstPath ), 'bypassReadOnly' => true ] );
 268                         if ( !$status->isOK() ) {
 269                                 $this->error( $status );
 270                                 $this->fatalError( "$domainId: Could not copy $srcPath to $dstPath." );
 271                         }
 272                         $ops[] = [ 'op' => 'store',
 273                                 'src' => $fsFile->getPath(), 'dst' => $dstPath, 'overwrite' => true ];
 274                         $copiedRel[] = $srcPathRel;
 275                 }
 276
 277                 // Copy in the batch of source files...
 278                 $t_start = microtime( true );
 279                 $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
 280                 if ( !$status->isOK() ) {
 281                         sleep( 10 ); // wait and retry copy again
 282                         $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
 283                 }
 284                 $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 285                 if ( !$status->isOK() ) {
 286                         $this->error( $status );
 287                         $this->fatalError( "$domainId: Could not copy file batch." );
 288                 } elseif ( count( $copiedRel ) ) {
 289                         $this->output( "\n\tCopied these file(s) [{$elapsed_ms}ms]:\n\t" .
 290                                 implode( "\n\t", $copiedRel ) . "\n\n" );
 291                 }
 292         }
 293
 294         /**
 295          * @param string[] $dstPathsRel
 296          * @param string $backendRel
 297          * @param FileBackend $dst
 298          * @return void
 299          */
 300         protected function delFileBatch(
 301                 array $dstPathsRel, $backendRel, FileBackend $dst
 302         ) {
 303                 $ops = [];
 304                 $deletedRel = []; // for output message
 305                 $domainId = $dst->getDomainId();
 306
 307                 // Determine what files need to be copied over...
 308                 foreach ( $dstPathsRel as $dstPathRel ) {
 309                         $dstPath = $dst->getRootStoragePath() . "/$backendRel/$dstPathRel";
 310                         $ops[] = [ 'op' => 'delete', 'src' => $dstPath ];
 311                         $deletedRel[] = $dstPathRel;
 312                 }
 313
 314                 // Delete the batch of source files...
 315                 $t_start = microtime( true );
 316                 $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
 317                 if ( !$status->isOK() ) {
 318                         sleep( 10 ); // wait and retry copy again
 319                         $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => true ] );
 320                 }
 321                 $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 322                 if ( !$status->isOK() ) {
 323                         $this->error( $status );
 324                         $this->fatalError( "$domainId: Could not delete file batch." );
 325                 } elseif ( count( $deletedRel ) ) {
 326                         $this->output( "\n\tDeleted these file(s) [{$elapsed_ms}ms]:\n\t" .
 327                                 implode( "\n\t", $deletedRel ) . "\n\n" );
 328                 }
 329         }
 330
 331         /**
 332          * @param FileBackend $src
 333          * @param FileBackend $dst
 334          * @param string $sPath
 335          * @param string $dPath
 336          * @return bool
 337          */
 338         protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) {
 339                 $skipHash = $this->hasOption( 'skiphash' );
 340                 $srcStat = $src->getFileStat( [ 'src' => $sPath ] );
 341                 $dPathSha1 = sha1( $dPath );
 342                 if ( $this->statCache !== null ) {
 343                         // All dst files are already in stat cache
 344                         $dstStat = $this->statCache[$dPathSha1] ?? false;
 345                 } else {
 346                         $dstStat = $dst->getFileStat( [ 'src' => $dPath ] );
 347                 }
 348                 // Initial fast checks to see if files are obviously different
 349                 $sameFast = (
 350                         is_array( $srcStat )
 351                         && is_array( $dstStat ) // dest exists
 352                         && $srcStat['size'] === $dstStat['size']
 353                 );
 354                 // More thorough checks against files
 355                 if ( !$sameFast ) {
 356                         $same = false; // no need to look farther
 357                 } elseif ( isset( $srcStat['md5'] ) && isset( $dstStat['md5'] ) ) {
 358                         // If MD5 was already in the stat info, just use it.
 359                         // This is useful as many objects stores can return this in object listing,
 360                         // so we can use it to avoid slow per-file HEADs.
 361                         $same = ( $srcStat['md5'] === $dstStat['md5'] );
 362                 } elseif ( $skipHash ) {
 363                         // This mode is good for copying to a backup location or resyncing clone
 364                         // backends in FileBackendMultiWrite (since they get writes second, they have
 365                         // higher timestamps). However, when copying the other way, this hits loads of
 366                         // false positives (possibly 100%) and wastes a bunch of time on GETs/PUTs.
 367                         // @phan-suppress-next-line PhanTypeArraySuspiciousNullable
 368                         $same = ( $srcStat['mtime'] <= $dstStat['mtime'] );
 369                 } else {
 370                         // This is the slowest method which does many per-file HEADs (unless an object
 371                         // store tracks SHA-1 in listings).
 372                         $same = ( $src->getFileSha1Base36( [ 'src' => $sPath, 'latest' => 1 ] )
 373                                 === $dst->getFileSha1Base36( [ 'src' => $dPath, 'latest' => 1 ] ) );
 374                 }
 375
 376                 return $same;
 377         }
 378 }
 379
 380 // @codeCoverageIgnoreStart
 381 $maintClass = CopyFileBackend::class;
 382 require_once RUN_MAINTENANCE_IF_MAIN;
 383 // @codeCoverageIgnoreEnd