maintenance/storage/checkStorage.php

   1 <?php
   2 /**
   3  * Fsck for MediaWiki
   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 ExternalStorage
  22  */
  23
  24 use MediaWiki\MediaWikiServices;
  25
  26 if ( !defined( 'MEDIAWIKI' ) ) {
  27         $optionsWithoutArgs = [ 'fix' ];
  28         require_once __DIR__ . '/../commandLine.inc';
  29
  30         $cs = new CheckStorage;
  31         $fix = isset( $options['fix'] );
  32         if ( isset( $args[0] ) ) {
  33                 $xml = $args[0];
  34         } else {
  35                 $xml = false;
  36         }
  37         $cs->check( $fix, $xml );
  38 }
  39
  40 // ----------------------------------------------------------------------------------
  41
  42 /**
  43  * Maintenance script to do various checks on external storage.
  44  *
  45  * @fixme this should extend the base Maintenance class
  46  * @ingroup Maintenance ExternalStorage
  47  */
  48 class CheckStorage {
  49         const CONCAT_HEADER = 'O:27:"concatenatedgziphistoryblob"';
  50         public $oldIdMap, $errors;
  51         public $dbStore = null;
  52
  53         public $errorDescriptions = [
  54                 'restore text' => 'Damaged text, need to be restored from a backup',
  55                 'restore revision' => 'Damaged revision row, need to be restored from a backup',
  56                 'unfixable' => 'Unexpected errors with no automated fixing method',
  57                 'fixed' => 'Errors already fixed',
  58                 'fixable' => 'Errors which would already be fixed if --fix was specified',
  59         ];
  60
  61         function check( $fix = false, $xml = '' ) {
  62                 $dbr = wfGetDB( DB_REPLICA );
  63                 if ( $fix ) {
  64                         print "Checking, will fix errors if possible...\n";
  65                 } else {
  66                         print "Checking...\n";
  67                 }
  68                 $maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, __METHOD__ );
  69                 $chunkSize = 1000;
  70                 $flagStats = [];
  71                 $objectStats = [];
  72                 $knownFlags = [ 'external', 'gzip', 'object', 'utf-8' ];
  73                 $this->errors = [
  74                         'restore text' => [],
  75                         'restore revision' => [],
  76                         'unfixable' => [],
  77                         'fixed' => [],
  78                         'fixable' => [],
  79                 ];
  80
  81                 for ( $chunkStart = 1; $chunkStart < $maxRevId; $chunkStart += $chunkSize ) {
  82                         $chunkEnd = $chunkStart + $chunkSize - 1;
  83                         // print "$chunkStart of $maxRevId\n";
  84
  85                         // Fetch revision rows
  86                         $this->oldIdMap = [];
  87                         $dbr->ping();
  88                         $res = $dbr->select( 'revision', [ 'rev_id', 'rev_text_id' ],
  89                                 [ "rev_id BETWEEN $chunkStart AND $chunkEnd" ], __METHOD__ );
  90                         foreach ( $res as $row ) {
  91                                 $this->oldIdMap[$row->rev_id] = $row->rev_text_id;
  92                         }
  93                         $dbr->freeResult( $res );
  94
  95                         if ( !count( $this->oldIdMap ) ) {
  96                                 continue;
  97                         }
  98
  99                         // Fetch old_flags
 100                         $missingTextRows = array_flip( $this->oldIdMap );
 101                         $externalRevs = [];
 102                         $objectRevs = [];
 103                         $res = $dbr->select( 'text', [ 'old_id', 'old_flags' ],
 104                                 'old_id IN (' . implode( ',', $this->oldIdMap ) . ')', __METHOD__ );
 105                         foreach ( $res as $row ) {
 106                                 /**
 107                                  * @var $flags int
 108                                  */
 109                                 $flags = $row->old_flags;
 110                                 $id = $row->old_id;
 111
 112                                 // Create flagStats row if it doesn't exist
 113                                 $flagStats = $flagStats + [ $flags => 0 ];
 114                                 // Increment counter
 115                                 $flagStats[$flags]++;
 116
 117                                 // Not missing
 118                                 unset( $missingTextRows[$row->old_id] );
 119
 120                                 // Check for external or object
 121                                 if ( $flags == '' ) {
 122                                         $flagArray = [];
 123                                 } else {
 124                                         $flagArray = explode( ',', $flags );
 125                                 }
 126                                 if ( in_array( 'external', $flagArray ) ) {
 127                                         $externalRevs[] = $id;
 128                                 } elseif ( in_array( 'object', $flagArray ) ) {
 129                                         $objectRevs[] = $id;
 130                                 }
 131
 132                                 // Check for unrecognised flags
 133                                 if ( $flags == '0' ) {
 134                                         // This is a known bug from 2004
 135                                         // It's safe to just erase the old_flags field
 136                                         if ( $fix ) {
 137                                                 $this->error( 'fixed', "Warning: old_flags set to 0", $id );
 138                                                 $dbw = wfGetDB( DB_MASTER );
 139                                                 $dbw->ping();
 140                                                 $dbw->update( 'text', [ 'old_flags' => '' ],
 141                                                         [ 'old_id' => $id ], __METHOD__ );
 142                                                 echo "Fixed\n";
 143                                         } else {
 144                                                 $this->error( 'fixable', "Warning: old_flags set to 0", $id );
 145                                         }
 146                                 } elseif ( count( array_diff( $flagArray, $knownFlags ) ) ) {
 147                                         $this->error( 'unfixable', "Error: invalid flags field \"$flags\"", $id );
 148                                 }
 149                         }
 150                         $dbr->freeResult( $res );
 151
 152                         // Output errors for any missing text rows
 153                         foreach ( $missingTextRows as $oldId => $revId ) {
 154                                 $this->error( 'restore revision', "Error: missing text row", $oldId );
 155                         }
 156
 157                         // Verify external revisions
 158                         $externalConcatBlobs = [];
 159                         $externalNormalBlobs = [];
 160                         if ( count( $externalRevs ) ) {
 161                                 $res = $dbr->select( 'text', [ 'old_id', 'old_flags', 'old_text' ],
 162                                         [ 'old_id IN (' . implode( ',', $externalRevs ) . ')' ], __METHOD__ );
 163                                 foreach ( $res as $row ) {
 164                                         $urlParts = explode( '://', $row->old_text, 2 );
 165                                         if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {
 166                                                 $this->error( 'restore text', "Error: invalid URL \"{$row->old_text}\"", $row->old_id );
 167                                                 continue;
 168                                         }
 169                                         list( $proto, ) = $urlParts;
 170                                         if ( $proto != 'DB' ) {
 171                                                 $this->error( 'restore text', "Error: invalid external protocol \"$proto\"", $row->old_id );
 172                                                 continue;
 173                                         }
 174                                         $path = explode( '/', $row->old_text );
 175                                         $cluster = $path[2];
 176                                         $id = $path[3];
 177                                         if ( isset( $path[4] ) ) {
 178                                                 $externalConcatBlobs[$cluster][$id][] = $row->old_id;
 179                                         } else {
 180                                                 $externalNormalBlobs[$cluster][$id][] = $row->old_id;
 181                                         }
 182                                 }
 183                                 $dbr->freeResult( $res );
 184                         }
 185
 186                         // Check external concat blobs for the right header
 187                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 188
 189                         // Check external normal blobs for existence
 190                         if ( count( $externalNormalBlobs ) ) {
 191                                 if ( is_null( $this->dbStore ) ) {
 192                                         $this->dbStore = new ExternalStoreDB;
 193                                 }
 194                                 foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {
 195                                         $blobIds = array_keys( $xBlobIds );
 196                                         $extDb =& $this->dbStore->getSlave( $cluster );
 197                                         $blobsTable = $this->dbStore->getTable( $extDb );
 198                                         $res = $extDb->select( $blobsTable,
 199                                                 [ 'blob_id' ],
 200                                                 [ 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ], __METHOD__ );
 201                                         foreach ( $res as $row ) {
 202                                                 unset( $xBlobIds[$row->blob_id] );
 203                                         }
 204                                         $extDb->freeResult( $res );
 205                                         // Print errors for missing blobs rows
 206                                         foreach ( $xBlobIds as $blobId => $oldId ) {
 207                                                 $this->error( 'restore text', "Error: missing target $blobId for one-part ES URL", $oldId );
 208                                         }
 209                                 }
 210                         }
 211
 212                         // Check local objects
 213                         $dbr->ping();
 214                         $concatBlobs = [];
 215                         $curIds = [];
 216                         if ( count( $objectRevs ) ) {
 217                                 $headerLength = 300;
 218                                 $res = $dbr->select(
 219                                         'text',
 220                                         [ 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ],
 221                                         [ 'old_id IN (' . implode( ',', $objectRevs ) . ')' ],
 222                                         __METHOD__
 223                                 );
 224                                 foreach ( $res as $row ) {
 225                                         $oldId = $row->old_id;
 226                                         $matches = [];
 227                                         if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {
 228                                                 $this->error( 'restore text', "Error: invalid object header", $oldId );
 229                                                 continue;
 230                                         }
 231
 232                                         $className = strtolower( $matches[2] );
 233                                         if ( strlen( $className ) != $matches[1] ) {
 234                                                 $this->error(
 235                                                         'restore text',
 236                                                         "Error: invalid object header, wrong class name length",
 237                                                         $oldId
 238                                                 );
 239                                                 continue;
 240                                         }
 241
 242                                         $objectStats = $objectStats + [ $className => 0 ];
 243                                         $objectStats[$className]++;
 244
 245                                         switch ( $className ) {
 246                                                 case 'concatenatedgziphistoryblob':
 247                                                         // Good
 248                                                         break;
 249                                                 case 'historyblobstub':
 250                                                 case 'historyblobcurstub':
 251                                                         if ( strlen( $row->header ) == $headerLength ) {
 252                                                                 $this->error( 'unfixable', "Error: overlong stub header", $oldId );
 253                                                                 continue;
 254                                                         }
 255                                                         $stubObj = unserialize( $row->header );
 256                                                         if ( !is_object( $stubObj ) ) {
 257                                                                 $this->error( 'restore text', "Error: unable to unserialize stub object", $oldId );
 258                                                                 continue;
 259                                                         }
 260                                                         if ( $className == 'historyblobstub' ) {
 261                                                                 $concatBlobs[$stubObj->mOldId][] = $oldId;
 262                                                         } else {
 263                                                                 $curIds[$stubObj->mCurId][] = $oldId;
 264                                                         }
 265                                                         break;
 266                                                 default:
 267                                                         $this->error( 'unfixable', "Error: unrecognised object class \"$className\"", $oldId );
 268                                         }
 269                                 }
 270                                 $dbr->freeResult( $res );
 271                         }
 272
 273                         // Check local concat blob validity
 274                         $externalConcatBlobs = [];
 275                         if ( count( $concatBlobs ) ) {
 276                                 $headerLength = 300;
 277                                 $res = $dbr->select(
 278                                         'text',
 279                                         [ 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ],
 280                                         [ 'old_id IN (' . implode( ',', array_keys( $concatBlobs ) ) . ')' ],
 281                                         __METHOD__
 282                                 );
 283                                 foreach ( $res as $row ) {
 284                                         $flags = explode( ',', $row->old_flags );
 285                                         if ( in_array( 'external', $flags ) ) {
 286                                                 // Concat blob is in external storage?
 287                                                 if ( in_array( 'object', $flags ) ) {
 288                                                         $urlParts = explode( '/', $row->header );
 289                                                         if ( $urlParts[0] != 'DB:' ) {
 290                                                                 $this->error(
 291                                                                         'unfixable',
 292                                                                         "Error: unrecognised external storage type \"{$urlParts[0]}",
 293                                                                         $row->old_id
 294                                                                 );
 295                                                         } else {
 296                                                                 $cluster = $urlParts[2];
 297                                                                 $id = $urlParts[3];
 298                                                                 if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {
 299                                                                         $externalConcatBlobs[$cluster][$id] = [];
 300                                                                 }
 301                                                                 $externalConcatBlobs[$cluster][$id] = array_merge(
 302                                                                         $externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]
 303                                                                 );
 304                                                         }
 305                                                 } else {
 306                                                         $this->error(
 307                                                                 'unfixable',
 308                                                                 "Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",
 309                                                                 $concatBlobs[$row->old_id] );
 310                                                 }
 311                                         } elseif ( strcasecmp(
 312                                                 substr( $row->header, 0, strlen( self::CONCAT_HEADER ) ),
 313                                                 self::CONCAT_HEADER
 314                                         ) ) {
 315                                                 $this->error(
 316                                                         'restore text',
 317                                                         "Error: Incorrect object header for concat bulk row {$row->old_id}",
 318                                                         $concatBlobs[$row->old_id]
 319                                                 );
 320                                         } # else good
 321
 322                                         unset( $concatBlobs[$row->old_id] );
 323                                 }
 324                                 $dbr->freeResult( $res );
 325                         }
 326
 327                         // Check targets of unresolved stubs
 328                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 329                         // next chunk
 330                 }
 331
 332                 print "\n\nErrors:\n";
 333                 foreach ( $this->errors as $name => $errors ) {
 334                         if ( count( $errors ) ) {
 335                                 $description = $this->errorDescriptions[$name];
 336                                 echo "$description: " . implode( ',', array_keys( $errors ) ) . "\n";
 337                         }
 338                 }
 339
 340                 if ( count( $this->errors['restore text'] ) && $fix ) {
 341                         if ( (string)$xml !== '' ) {
 342                                 $this->restoreText( array_keys( $this->errors['restore text'] ), $xml );
 343                         } else {
 344                                 echo "Can't fix text, no XML backup specified\n";
 345                         }
 346                 }
 347
 348                 print "\nFlag statistics:\n";
 349                 $total = array_sum( $flagStats );
 350                 foreach ( $flagStats as $flag => $count ) {
 351                         printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );
 352                 }
 353                 print "\nLocal object statistics:\n";
 354                 $total = array_sum( $objectStats );
 355                 foreach ( $objectStats as $className => $count ) {
 356                         printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );
 357                 }
 358         }
 359
 360         function error( $type, $msg, $ids ) {
 361                 if ( is_array( $ids ) && count( $ids ) == 1 ) {
 362                         $ids = reset( $ids );
 363                 }
 364                 if ( is_array( $ids ) ) {
 365                         $revIds = [];
 366                         foreach ( $ids as $id ) {
 367                                 $revIds = array_merge( $revIds, array_keys( $this->oldIdMap, $id ) );
 368                         }
 369                         print "$msg in text rows " . implode( ', ', $ids ) .
 370                                 ", revisions " . implode( ', ', $revIds ) . "\n";
 371                 } else {
 372                         $id = $ids;
 373                         $revIds = array_keys( $this->oldIdMap, $id );
 374                         if ( count( $revIds ) == 1 ) {
 375                                 print "$msg in old_id $id, rev_id {$revIds[0]}\n";
 376                         } else {
 377                                 print "$msg in old_id $id, revisions " . implode( ', ', $revIds ) . "\n";
 378                         }
 379                 }
 380                 $this->errors[$type] = $this->errors[$type] + array_flip( $revIds );
 381         }
 382
 383         function checkExternalConcatBlobs( $externalConcatBlobs ) {
 384                 if ( !count( $externalConcatBlobs ) ) {
 385                         return;
 386                 }
 387
 388                 if ( is_null( $this->dbStore ) ) {
 389                         $this->dbStore = new ExternalStoreDB;
 390                 }
 391
 392                 foreach ( $externalConcatBlobs as $cluster => $oldIds ) {
 393                         $blobIds = array_keys( $oldIds );
 394                         $extDb =& $this->dbStore->getSlave( $cluster );
 395                         $blobsTable = $this->dbStore->getTable( $extDb );
 396                         $headerLength = strlen( self::CONCAT_HEADER );
 397                         $res = $extDb->select( $blobsTable,
 398                                 [ 'blob_id', "LEFT(blob_text, $headerLength) AS header" ],
 399                                 [ 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ], __METHOD__ );
 400                         foreach ( $res as $row ) {
 401                                 if ( strcasecmp( $row->header, self::CONCAT_HEADER ) ) {
 402                                         $this->error(
 403                                                 'restore text',
 404                                                 "Error: invalid header on target $cluster/{$row->blob_id} of two-part ES URL",
 405                                                 $oldIds[$row->blob_id]
 406                                         );
 407                                 }
 408                                 unset( $oldIds[$row->blob_id] );
 409                         }
 410                         $extDb->freeResult( $res );
 411
 412                         // Print errors for missing blobs rows
 413                         foreach ( $oldIds as $blobId => $oldIds2 ) {
 414                                 $this->error(
 415                                         'restore text',
 416                                         "Error: missing target $cluster/$blobId for two-part ES URL",
 417                                         $oldIds2
 418                                 );
 419                         }
 420                 }
 421         }
 422
 423         function restoreText( $revIds, $xml ) {
 424                 global $wgDBname;
 425                 $tmpDir = wfTempDir();
 426
 427                 if ( !count( $revIds ) ) {
 428                         return;
 429                 }
 430
 431                 print "Restoring text from XML backup...\n";
 432
 433                 $revFileName = "$tmpDir/broken-revlist-$wgDBname";
 434                 $filteredXmlFileName = "$tmpDir/filtered-$wgDBname.xml";
 435
 436                 // Write revision list
 437                 if ( !file_put_contents( $revFileName, implode( "\n", $revIds ) ) ) {
 438                         echo "Error writing revision list, can't restore text\n";
 439
 440                         return;
 441                 }
 442
 443                 // Run mwdumper
 444                 echo "Filtering XML dump...\n";
 445                 $exitStatus = 0;
 446                 passthru( 'mwdumper ' .
 447                         wfEscapeShellArg(
 448                                 "--output=file:$filteredXmlFileName",
 449                                 "--filter=revlist:$revFileName",
 450                                 $xml
 451                         ), $exitStatus
 452                 );
 453
 454                 if ( $exitStatus ) {
 455                         echo "mwdumper died with exit status $exitStatus\n";
 456
 457                         return;
 458                 }
 459
 460                 $file = fopen( $filteredXmlFileName, 'r' );
 461                 if ( !$file ) {
 462                         echo "Unable to open filtered XML file\n";
 463
 464                         return;
 465                 }
 466
 467                 $dbr = wfGetDB( DB_REPLICA );
 468                 $dbw = wfGetDB( DB_MASTER );
 469                 $dbr->ping();
 470                 $dbw->ping();
 471
 472                 $source = new ImportStreamSource( $file );
 473                 $importer = new WikiImporter(
 474                         $source,
 475                         MediaWikiServices::getInstance()->getMainConfig()
 476                 );
 477                 $importer->setRevisionCallback( [ $this, 'importRevision' ] );
 478                 $importer->doImport();
 479         }
 480
 481         function importRevision( &$revision, &$importer ) {
 482                 $id = $revision->getID();
 483                 $content = $revision->getContent( Revision::RAW );
 484                 $id = $id ? $id : '';
 485
 486                 if ( $content === null ) {
 487                         echo "Revision $id is broken, we have no content available\n";
 488
 489                         return;
 490                 }
 491
 492                 $text = $content->serialize();
 493                 if ( $text === '' ) {
 494                         // This is what happens if the revision was broken at the time the
 495                         // dump was made. Unfortunately, it also happens if the revision was
 496                         // legitimately blank, so there's no way to tell the difference. To
 497                         // be safe, we'll skip it and leave it broken
 498
 499                         echo "Revision $id is blank in the dump, may have been broken before export\n";
 500
 501                         return;
 502                 }
 503
 504                 if ( !$id ) {
 505                         // No ID, can't import
 506                         echo "No id tag in revision, can't import\n";
 507
 508                         return;
 509                 }
 510
 511                 // Find text row again
 512                 $dbr = wfGetDB( DB_REPLICA );
 513                 $oldId = $dbr->selectField( 'revision', 'rev_text_id', [ 'rev_id' => $id ], __METHOD__ );
 514                 if ( !$oldId ) {
 515                         echo "Missing revision row for rev_id $id\n";
 516
 517                         return;
 518                 }
 519
 520                 // Compress the text
 521                 $flags = Revision::compressRevisionText( $text );
 522
 523                 // Update the text row
 524                 $dbw = wfGetDB( DB_MASTER );
 525                 $dbw->update( 'text',
 526                         [ 'old_flags' => $flags, 'old_text' => $text ],
 527                         [ 'old_id' => $oldId ],
 528                         __METHOD__, [ 'LIMIT' => 1 ]
 529                 );
 530
 531                 // Remove it from the unfixed list and add it to the fixed list
 532                 unset( $this->errors['restore text'][$id] );
 533                 $this->errors['fixed'][$id] = true;
 534         }
 535 }