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 use MediaWiki\Shell\Shell;
  26
  27 if ( !defined( 'MEDIAWIKI' ) ) {
  28         $optionsWithoutArgs = [ 'fix' ];
  29         require_once __DIR__ . '/../commandLine.inc';
  30
  31         $cs = new CheckStorage;
  32         $fix = isset( $options['fix'] );
  33         $xml = $args[0] ?? false;
  34         $cs->check( $fix, $xml );
  35 }
  36
  37 // ----------------------------------------------------------------------------------
  38
  39 /**
  40  * Maintenance script to do various checks on external storage.
  41  *
  42  * @fixme this should extend the base Maintenance class
  43  * @ingroup Maintenance ExternalStorage
  44  */
  45 class CheckStorage {
  46         const CONCAT_HEADER = 'O:27:"concatenatedgziphistoryblob"';
  47         public $oldIdMap, $errors;
  48         /** @var ExternalStoreDB */
  49         public $dbStore = null;
  50
  51         public $errorDescriptions = [
  52                 'restore text' => 'Damaged text, need to be restored from a backup',
  53                 'restore revision' => 'Damaged revision row, need to be restored from a backup',
  54                 'unfixable' => 'Unexpected errors with no automated fixing method',
  55                 'fixed' => 'Errors already fixed',
  56                 'fixable' => 'Errors which would already be fixed if --fix was specified',
  57         ];
  58
  59         function check( $fix = false, $xml = '' ) {
  60                 global $wgMultiContentRevisionSchemaMigrationStage;
  61
  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)', '', __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                         $this->oldIdMap = [];
  86                         $dbr->ping();
  87
  88                         // Fetch revision rows
  89                         if ( $wgMultiContentRevisionSchemaMigrationStage & SCHEMA_COMPAT_READ_OLD ) {
  90                                 $res = $dbr->select( 'revision', [ 'rev_id', 'rev_text_id' ],
  91                                         [ "rev_id BETWEEN $chunkStart AND $chunkEnd" ], __METHOD__ );
  92                                 foreach ( $res as $row ) {
  93                                         if ( !isset( $this->oldIdMap[ $row->rev_text_id ] ) ) {
  94                                                 $this->oldIdMap[ $row->rev_text_id ] = [ $row->rev_id ];
  95                                         } elseif ( !in_array( $row->rev_id, $this->oldIdMap[ $row->rev_text_id ] ) ) {
  96                                                 $this->oldIdMap[ $row->rev_text_id ][] = $row->rev_id;
  97                                         }
  98                                 }
  99                         } else {
 100                                 $res = $dbr->select(
 101                                         [ 'slots', 'content' ],
 102                                         [ 'slot_revision_id', 'content_address' ],
 103                                         [ "slot_revision_id BETWEEN $chunkStart AND $chunkEnd" ],
 104                                         __METHOD__,
 105                                         [],
 106                                         [ 'content' => [ 'INNER JOIN', [ 'content_id = slot_content_id' ] ] ]
 107                                 );
 108                                 $blobStore = MediaWikiServices::getInstance()->getBlobStore();
 109                                 foreach ( $res as $row ) {
 110                                         $textId = $blobStore->getTextIdFromAddress( $row->content_address );
 111                                         if ( $textId ) {
 112                                                 if ( !isset( $this->oldIdMap[$textId] ) ) {
 113                                                         $this->oldIdMap[ $textId ] = [ $row->slot_revision_id ];
 114                                                 } elseif ( !in_array( $row->slot_revision_id, $this->oldIdMap[$textId] ) ) {
 115                                                         $this->oldIdMap[ $textId ][] = $row->slot_revision_id;
 116                                                 }
 117                                         }
 118                                 }
 119                         }
 120
 121                         if ( !count( $this->oldIdMap ) ) {
 122                                 continue;
 123                         }
 124
 125                         // Fetch old_flags
 126                         $missingTextRows = $this->oldIdMap;
 127                         $externalRevs = [];
 128                         $objectRevs = [];
 129                         $res = $dbr->select(
 130                                 'text',
 131                                 [ 'old_id', 'old_flags' ],
 132                                 [ 'old_id' => array_keys( $this->oldIdMap ) ],
 133                                 __METHOD__
 134                         );
 135                         foreach ( $res as $row ) {
 136                                 /**
 137                                  * @var int $flags
 138                                  */
 139                                 $flags = $row->old_flags;
 140                                 $id = $row->old_id;
 141
 142                                 // Create flagStats row if it doesn't exist
 143                                 $flagStats = $flagStats + [ $flags => 0 ];
 144                                 // Increment counter
 145                                 $flagStats[$flags]++;
 146
 147                                 // Not missing
 148                                 unset( $missingTextRows[$row->old_id] );
 149
 150                                 // Check for external or object
 151                                 if ( $flags == '' ) {
 152                                         $flagArray = [];
 153                                 } else {
 154                                         $flagArray = explode( ',', $flags );
 155                                 }
 156                                 if ( in_array( 'external', $flagArray ) ) {
 157                                         $externalRevs[] = $id;
 158                                 } elseif ( in_array( 'object', $flagArray ) ) {
 159                                         $objectRevs[] = $id;
 160                                 }
 161
 162                                 // Check for unrecognised flags
 163                                 if ( $flags == '0' ) {
 164                                         // This is a known bug from 2004
 165                                         // It's safe to just erase the old_flags field
 166                                         if ( $fix ) {
 167                                                 $this->addError( 'fixed', "Warning: old_flags set to 0", $id );
 168                                                 $dbw = wfGetDB( DB_MASTER );
 169                                                 $dbw->ping();
 170                                                 $dbw->update( 'text', [ 'old_flags' => '' ],
 171                                                         [ 'old_id' => $id ], __METHOD__ );
 172                                                 echo "Fixed\n";
 173                                         } else {
 174                                                 $this->addError( 'fixable', "Warning: old_flags set to 0", $id );
 175                                         }
 176                                 } elseif ( count( array_diff( $flagArray, $knownFlags ) ) ) {
 177                                         $this->addError( 'unfixable', "Error: invalid flags field \"$flags\"", $id );
 178                                 }
 179                         }
 180
 181                         // Output errors for any missing text rows
 182                         foreach ( $missingTextRows as $oldId => $revIds ) {
 183                                 $this->addError( 'restore revision', "Error: missing text row", $oldId );
 184                         }
 185
 186                         // Verify external revisions
 187                         $externalConcatBlobs = [];
 188                         $externalNormalBlobs = [];
 189                         if ( count( $externalRevs ) ) {
 190                                 $res = $dbr->select(
 191                                         'text',
 192                                         [ 'old_id', 'old_flags', 'old_text' ],
 193                                         [ 'old_id' => $externalRevs ],
 194                                         __METHOD__
 195                                 );
 196                                 foreach ( $res as $row ) {
 197                                         $urlParts = explode( '://', $row->old_text, 2 );
 198                                         if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {
 199                                                 $this->addError( 'restore text', "Error: invalid URL \"{$row->old_text}\"", $row->old_id );
 200                                                 continue;
 201                                         }
 202                                         list( $proto, ) = $urlParts;
 203                                         if ( $proto != 'DB' ) {
 204                                                 $this->addError(
 205                                                         'restore text',
 206                                                         "Error: invalid external protocol \"$proto\"",
 207                                                         $row->old_id );
 208                                                 continue;
 209                                         }
 210                                         $path = explode( '/', $row->old_text );
 211                                         $cluster = $path[2];
 212                                         $id = $path[3];
 213                                         if ( isset( $path[4] ) ) {
 214                                                 $externalConcatBlobs[$cluster][$id][] = $row->old_id;
 215                                         } else {
 216                                                 $externalNormalBlobs[$cluster][$id][] = $row->old_id;
 217                                         }
 218                                 }
 219                         }
 220
 221                         // Check external concat blobs for the right header
 222                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 223
 224                         // Check external normal blobs for existence
 225                         if ( count( $externalNormalBlobs ) ) {
 226                                 if ( is_null( $this->dbStore ) ) {
 227                                         $esFactory = MediaWikiServices::getInstance()->getExternalStoreFactory();
 228                                         $this->dbStore = $esFactory->getStore( 'DB' );
 229                                 }
 230                                 foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {
 231                                         $blobIds = array_keys( $xBlobIds );
 232                                         $extDb =& $this->dbStore->getSlave( $cluster );
 233                                         $blobsTable = $this->dbStore->getTable( $extDb );
 234                                         $res = $extDb->select( $blobsTable,
 235                                                 [ 'blob_id' ],
 236                                                 [ 'blob_id' => $blobIds ],
 237                                                 __METHOD__
 238                                         );
 239                                         foreach ( $res as $row ) {
 240                                                 unset( $xBlobIds[$row->blob_id] );
 241                                         }
 242                                         // Print errors for missing blobs rows
 243                                         foreach ( $xBlobIds as $blobId => $oldId ) {
 244                                                 $this->addError(
 245                                                         'restore text',
 246                                                         "Error: missing target $blobId for one-part ES URL",
 247                                                         $oldId );
 248                                         }
 249                                 }
 250                         }
 251
 252                         // Check local objects
 253                         $dbr->ping();
 254                         $concatBlobs = [];
 255                         $curIds = [];
 256                         if ( count( $objectRevs ) ) {
 257                                 $headerLength = 300;
 258                                 $res = $dbr->select(
 259                                         'text',
 260                                         [ 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ],
 261                                         [ 'old_id' => $objectRevs ],
 262                                         __METHOD__
 263                                 );
 264                                 foreach ( $res as $row ) {
 265                                         $oldId = $row->old_id;
 266                                         $matches = [];
 267                                         if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {
 268                                                 $this->addError( 'restore text', "Error: invalid object header", $oldId );
 269                                                 continue;
 270                                         }
 271
 272                                         $className = strtolower( $matches[2] );
 273                                         if ( strlen( $className ) != $matches[1] ) {
 274                                                 $this->addError(
 275                                                         'restore text',
 276                                                         "Error: invalid object header, wrong class name length",
 277                                                         $oldId
 278                                                 );
 279                                                 continue;
 280                                         }
 281
 282                                         $objectStats = $objectStats + [ $className => 0 ];
 283                                         $objectStats[$className]++;
 284
 285                                         switch ( $className ) {
 286                                                 case 'concatenatedgziphistoryblob':
 287                                                         // Good
 288                                                         break;
 289                                                 case 'historyblobstub':
 290                                                 case 'historyblobcurstub':
 291                                                         if ( strlen( $row->header ) == $headerLength ) {
 292                                                                 $this->addError( 'unfixable', "Error: overlong stub header", $oldId );
 293                                                                 break;
 294                                                         }
 295                                                         $stubObj = unserialize( $row->header );
 296                                                         if ( !is_object( $stubObj ) ) {
 297                                                                 $this->addError( 'restore text', "Error: unable to unserialize stub object", $oldId );
 298                                                                 break;
 299                                                         }
 300                                                         if ( $className == 'historyblobstub' ) {
 301                                                                 $concatBlobs[$stubObj->mOldId][] = $oldId;
 302                                                         } else {
 303                                                                 $curIds[$stubObj->mCurId][] = $oldId;
 304                                                         }
 305                                                         break;
 306                                                 default:
 307                                                         $this->addError( 'unfixable', "Error: unrecognised object class \"$className\"", $oldId );
 308                                         }
 309                                 }
 310                         }
 311
 312                         // Check local concat blob validity
 313                         $externalConcatBlobs = [];
 314                         if ( count( $concatBlobs ) ) {
 315                                 $headerLength = 300;
 316                                 $res = $dbr->select(
 317                                         'text',
 318                                         [ 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ],
 319                                         [ 'old_id' => array_keys( $concatBlobs ) ],
 320                                         __METHOD__
 321                                 );
 322                                 foreach ( $res as $row ) {
 323                                         $flags = explode( ',', $row->old_flags );
 324                                         if ( in_array( 'external', $flags ) ) {
 325                                                 // Concat blob is in external storage?
 326                                                 if ( in_array( 'object', $flags ) ) {
 327                                                         $urlParts = explode( '/', $row->header );
 328                                                         if ( $urlParts[0] != 'DB:' ) {
 329                                                                 $this->addError(
 330                                                                         'unfixable',
 331                                                                         "Error: unrecognised external storage type \"{$urlParts[0]}",
 332                                                                         $row->old_id
 333                                                                 );
 334                                                         } else {
 335                                                                 $cluster = $urlParts[2];
 336                                                                 $id = $urlParts[3];
 337                                                                 if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {
 338                                                                         $externalConcatBlobs[$cluster][$id] = [];
 339                                                                 }
 340                                                                 $externalConcatBlobs[$cluster][$id] = array_merge(
 341                                                                         $externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]
 342                                                                 );
 343                                                         }
 344                                                 } else {
 345                                                         $this->addError(
 346                                                                 'unfixable',
 347                                                                 "Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",
 348                                                                 $concatBlobs[$row->old_id] );
 349                                                 }
 350                                         } elseif ( strcasecmp(
 351                                                 substr( $row->header, 0, strlen( self::CONCAT_HEADER ) ),
 352                                                 self::CONCAT_HEADER
 353                                         ) ) {
 354                                                 $this->addError(
 355                                                         'restore text',
 356                                                         "Error: Incorrect object header for concat bulk row {$row->old_id}",
 357                                                         $concatBlobs[$row->old_id]
 358                                                 );
 359                                         } # else good
 360
 361                                         unset( $concatBlobs[$row->old_id] );
 362                                 }
 363                         }
 364
 365                         // Check targets of unresolved stubs
 366                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 367                         // next chunk
 368                 }
 369
 370                 print "\n\nErrors:\n";
 371                 foreach ( $this->errors as $name => $errors ) {
 372                         if ( count( $errors ) ) {
 373                                 $description = $this->errorDescriptions[$name];
 374                                 echo "$description: " . implode( ',', array_keys( $errors ) ) . "\n";
 375                         }
 376                 }
 377
 378                 if ( count( $this->errors['restore text'] ) && $fix ) {
 379                         if ( (string)$xml !== '' ) {
 380                                 $this->restoreText( array_keys( $this->errors['restore text'] ), $xml );
 381                         } else {
 382                                 echo "Can't fix text, no XML backup specified\n";
 383                         }
 384                 }
 385
 386                 print "\nFlag statistics:\n";
 387                 $total = array_sum( $flagStats );
 388                 foreach ( $flagStats as $flag => $count ) {
 389                         printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );
 390                 }
 391                 print "\nLocal object statistics:\n";
 392                 $total = array_sum( $objectStats );
 393                 foreach ( $objectStats as $className => $count ) {
 394                         printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );
 395                 }
 396         }
 397
 398         function addError( $type, $msg, $ids ) {
 399                 if ( is_array( $ids ) && count( $ids ) == 1 ) {
 400                         $ids = reset( $ids );
 401                 }
 402                 if ( is_array( $ids ) ) {
 403                         $revIds = [];
 404                         foreach ( $ids as $id ) {
 405                                 $revIds = array_unique( array_merge( $revIds, $this->oldIdMap[$id] ) );
 406                         }
 407                         print "$msg in text rows " . implode( ', ', $ids ) .
 408                                 ", revisions " . implode( ', ', $revIds ) . "\n";
 409                 } else {
 410                         $id = $ids;
 411                         $revIds = $this->oldIdMap[$id];
 412                         if ( count( $revIds ) == 1 ) {
 413                                 print "$msg in old_id $id, rev_id {$revIds[0]}\n";
 414                         } else {
 415                                 print "$msg in old_id $id, revisions " . implode( ', ', $revIds ) . "\n";
 416                         }
 417                 }
 418                 $this->errors[$type] = $this->errors[$type] + array_flip( $revIds );
 419         }
 420
 421         function checkExternalConcatBlobs( $externalConcatBlobs ) {
 422                 if ( !count( $externalConcatBlobs ) ) {
 423                         return;
 424                 }
 425
 426                 if ( is_null( $this->dbStore ) ) {
 427                         $esFactory = MediaWikiServices::getInstance()->getExternalStoreFactory();
 428                         $this->dbStore = $esFactory->getStore( 'DB' );
 429                 }
 430
 431                 foreach ( $externalConcatBlobs as $cluster => $oldIds ) {
 432                         $blobIds = array_keys( $oldIds );
 433                         $extDb =& $this->dbStore->getSlave( $cluster );
 434                         $blobsTable = $this->dbStore->getTable( $extDb );
 435                         $headerLength = strlen( self::CONCAT_HEADER );
 436                         $res = $extDb->select( $blobsTable,
 437                                 [ 'blob_id', "LEFT(blob_text, $headerLength) AS header" ],
 438                                 [ 'blob_id' => $blobIds ],
 439                                 __METHOD__
 440                         );
 441                         foreach ( $res as $row ) {
 442                                 if ( strcasecmp( $row->header, self::CONCAT_HEADER ) ) {
 443                                         $this->addError(
 444                                                 'restore text',
 445                                                 "Error: invalid header on target $cluster/{$row->blob_id} of two-part ES URL",
 446                                                 $oldIds[$row->blob_id]
 447                                         );
 448                                 }
 449                                 unset( $oldIds[$row->blob_id] );
 450                         }
 451
 452                         // Print errors for missing blobs rows
 453                         foreach ( $oldIds as $blobId => $oldIds2 ) {
 454                                 $this->addError(
 455                                         'restore text',
 456                                         "Error: missing target $cluster/$blobId for two-part ES URL",
 457                                         $oldIds2
 458                                 );
 459                         }
 460                 }
 461         }
 462
 463         function restoreText( $revIds, $xml ) {
 464                 global $wgDBname;
 465                 $tmpDir = wfTempDir();
 466
 467                 if ( !count( $revIds ) ) {
 468                         return;
 469                 }
 470
 471                 print "Restoring text from XML backup...\n";
 472
 473                 $revFileName = "$tmpDir/broken-revlist-$wgDBname";
 474                 $filteredXmlFileName = "$tmpDir/filtered-$wgDBname.xml";
 475
 476                 // Write revision list
 477                 if ( !file_put_contents( $revFileName, implode( "\n", $revIds ) ) ) {
 478                         echo "Error writing revision list, can't restore text\n";
 479
 480                         return;
 481                 }
 482
 483                 // Run mwdumper
 484                 echo "Filtering XML dump...\n";
 485                 $exitStatus = 0;
 486                 passthru( 'mwdumper ' .
 487                         Shell::escape(
 488                                 "--output=file:$filteredXmlFileName",
 489                                 "--filter=revlist:$revFileName",
 490                                 $xml
 491                         ), $exitStatus
 492                 );
 493
 494                 if ( $exitStatus ) {
 495                         echo "mwdumper died with exit status $exitStatus\n";
 496
 497                         return;
 498                 }
 499
 500                 $file = fopen( $filteredXmlFileName, 'r' );
 501                 if ( !$file ) {
 502                         echo "Unable to open filtered XML file\n";
 503
 504                         return;
 505                 }
 506
 507                 $dbr = wfGetDB( DB_REPLICA );
 508                 $dbw = wfGetDB( DB_MASTER );
 509                 $dbr->ping();
 510                 $dbw->ping();
 511
 512                 $source = new ImportStreamSource( $file );
 513                 $importer = new WikiImporter(
 514                         $source,
 515                         MediaWikiServices::getInstance()->getMainConfig()
 516                 );
 517                 $importer->setRevisionCallback( [ $this, 'importRevision' ] );
 518                 $importer->setNoticeCallback( function ( $msg, $params ) {
 519                         echo wfMessage( $msg, $params )->text() . "\n";
 520                 } );
 521                 $importer->doImport();
 522         }
 523
 524         function importRevision( &$revision, &$importer ) {
 525                 $id = $revision->getID();
 526                 $content = $revision->getContent( Revision::RAW );
 527                 $id = $id ?: '';
 528
 529                 if ( $content === null ) {
 530                         echo "Revision $id is broken, we have no content available\n";
 531
 532                         return;
 533                 }
 534
 535                 $text = $content->serialize();
 536                 if ( $text === '' ) {
 537                         // This is what happens if the revision was broken at the time the
 538                         // dump was made. Unfortunately, it also happens if the revision was
 539                         // legitimately blank, so there's no way to tell the difference. To
 540                         // be safe, we'll skip it and leave it broken
 541
 542                         echo "Revision $id is blank in the dump, may have been broken before export\n";
 543
 544                         return;
 545                 }
 546
 547                 if ( !$id ) {
 548                         // No ID, can't import
 549                         echo "No id tag in revision, can't import\n";
 550
 551                         return;
 552                 }
 553
 554                 // Find text row again
 555                 $dbr = wfGetDB( DB_REPLICA );
 556                 $oldId = $dbr->selectField( 'revision', 'rev_text_id', [ 'rev_id' => $id ], __METHOD__ );
 557                 if ( !$oldId ) {
 558                         echo "Missing revision row for rev_id $id\n";
 559
 560                         return;
 561                 }
 562
 563                 // Compress the text
 564                 $flags = Revision::compressRevisionText( $text );
 565
 566                 // Update the text row
 567                 $dbw = wfGetDB( DB_MASTER );
 568                 $dbw->update( 'text',
 569                         [ 'old_flags' => $flags, 'old_text' => $text ],
 570                         [ 'old_id' => $oldId ],
 571                         __METHOD__, [ 'LIMIT' => 1 ]
 572                 );
 573
 574                 // Remove it from the unfixed list and add it to the fixed list
 575                 unset( $this->errors['restore text'][$id] );
 576                 $this->errors['fixed'][$id] = true;
 577         }
 578 }