maintenance/storage/checkStorage.php

   1 <?php
   2
   3 /**
   4  * Fsck for MediaWiki
   5  */
   6
   7 define( 'CONCAT_HEADER', 'O:27:"concatenatedgziphistoryblob"' );
   8
   9 if ( !defined( 'MEDIAWIKI' ) ) {
  10         require_once( dirname(__FILE__) . '/../commandLine.inc' );
  11
  12         $cs = new CheckStorage;
  13         $fix = isset( $options['fix'] );
  14         if ( isset( $args[0] ) ) {
  15                 $xml = $args[0];
  16         } else {
  17                 $xml = false;
  18         }
  19         $cs->check( $fix, $xml );
  20 }
  21
  22
  23 //----------------------------------------------------------------------------------
  24
  25 class CheckStorage
  26 {
  27         var $oldIdMap, $errors;
  28         var $dbStore = null;
  29
  30         var $errorDescriptions = array(
  31                 'restore text' => 'Damaged text, need to be restored from a backup',
  32                 'restore revision' => 'Damaged revision row, need to be restored from a backup',
  33                 'unfixable' => 'Unexpected errors with no automated fixing method',
  34                 'fixed' => 'Errors already fixed',
  35                 'fixable' => 'Errors which would already be fixed if --fix was specified',
  36         );
  37
  38         function check( $fix = false, $xml = '' ) {
  39                 $fname = 'checkStorage';
  40                 $dbr = wfGetDB( DB_SLAVE );
  41                 if ( $fix ) {
  42                         $dbw = wfGetDB( DB_MASTER );
  43                         print "Checking, will fix errors if possible...\n";
  44                 } else {
  45                         print "Checking...\n";
  46                 }
  47                 $maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, $fname );
  48                 $chunkSize = 1000;
  49                 $flagStats = array();
  50                 $objectStats = array();
  51                 $knownFlags = array( 'external', 'gzip', 'object', 'utf-8' );
  52                 $this->errors = array(
  53                         'restore text' => array(),
  54                         'restore revision' => array(),
  55                         'unfixable' => array(),
  56                         'fixed' => array(),
  57                         'fixable' => array(),
  58                 );
  59
  60                 for ( $chunkStart = 1 ; $chunkStart < $maxRevId; $chunkStart += $chunkSize ) {
  61                         $chunkEnd = $chunkStart + $chunkSize - 1;
  62                         //print "$chunkStart of $maxRevId\n";
  63
  64                         // Fetch revision rows
  65                         $this->oldIdMap = array();
  66                         $dbr->ping();
  67                         $res = $dbr->select( 'revision', array( 'rev_id', 'rev_text_id' ),
  68                                 array( "rev_id BETWEEN $chunkStart AND $chunkEnd" ), $fname );
  69                         while ( $row = $dbr->fetchObject( $res ) ) {
  70                                 $this->oldIdMap[$row->rev_id] = $row->rev_text_id;
  71                         }
  72                         $dbr->freeResult( $res );
  73
  74                         if ( !count( $this->oldIdMap ) ) {
  75                                 continue;
  76                         }
  77
  78                         // Fetch old_flags
  79                         $missingTextRows = array_flip( $this->oldIdMap );
  80                         $externalRevs = array();
  81                         $objectRevs = array();
  82                         $res = $dbr->select( 'text', array( 'old_id', 'old_flags' ),
  83                                 'old_id IN (' . implode( ',', $this->oldIdMap ) . ')', $fname );
  84                         while ( $row = $dbr->fetchObject( $res ) ) {
  85                                 $flags = $row->old_flags;
  86                                 $id = $row->old_id;
  87
  88                                 // Create flagStats row if it doesn't exist
  89                                 $flagStats = $flagStats + array( $flags => 0 );
  90                                 // Increment counter
  91                                 $flagStats[$flags]++;
  92
  93                                 // Not missing
  94                                 unset( $missingTextRows[$row->old_id] );
  95
  96                                 // Check for external or object
  97                                 if ( $flags == '' ) {
  98                                         $flagArray = array();
  99                                 } else {
 100                                         $flagArray = explode( ',', $flags );
 101                                 }
 102                                 if ( in_array( 'external', $flagArray ) ) {
 103                                         $externalRevs[] = $id;
 104                                 } elseif ( in_array( 'object', $flagArray ) ) {
 105                                         $objectRevs[] = $id;
 106                                 }
 107
 108                                 // Check for unrecognised flags
 109                                 if ( $flags == '0' ) {
 110                                         // This is a known bug from 2004
 111                                         // It's safe to just erase the old_flags field
 112                                         if ( $fix ) {
 113                                                 $this->error( 'fixed', "Warning: old_flags set to 0", $id );
 114                                                 $dbw->ping();
 115                                                 $dbw->update( 'text', array( 'old_flags' => '' ),
 116                                                         array( 'old_id' => $id ), $fname );
 117                                                 echo "Fixed\n";
 118                                         } else {
 119                                                 $this->error( 'fixable', "Warning: old_flags set to 0", $id );
 120                                         }
 121                                 } elseif ( count( array_diff( $flagArray, $knownFlags ) ) ) {
 122                                         $this->error( 'unfixable', "Error: invalid flags field \"$flags\"", $id );
 123                                 }
 124                         }
 125                         $dbr->freeResult( $res );
 126
 127                         // Output errors for any missing text rows
 128                         foreach ( $missingTextRows as $oldId => $revId ) {
 129                                 $this->error( 'restore revision', "Error: missing text row", $oldId );
 130                         }
 131
 132                         // Verify external revisions
 133                         $externalConcatBlobs = array();
 134                         $externalNormalBlobs = array();
 135                         if ( count( $externalRevs ) ) {
 136                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', 'old_text' ),
 137                                         array( 'old_id IN (' . implode( ',', $externalRevs ) . ')' ), $fname );
 138                                 while ( $row = $dbr->fetchObject( $res ) ) {
 139                                         $urlParts = explode( '://', $row->old_text, 2 );
 140                                         if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {
 141                                                 $this->error( 'restore text', "Error: invalid URL \"{$row->old_text}\"", $row->old_id );
 142                                                 continue;
 143                                         }
 144                                         list( $proto, $path ) = $urlParts;
 145                                         if ( $proto != 'DB' ) {
 146                                                 $this->error( 'restore text', "Error: invalid external protocol \"$proto\"", $row->old_id );
 147                                                 continue;
 148                                         }
 149                                         $path = explode( '/', $row->old_text );
 150                                         $cluster = $path[2];
 151                                         $id = $path[3];
 152                                         if ( isset( $path[4] ) ) {
 153                                                 $externalConcatBlobs[$cluster][$id][] = $row->old_id;
 154                                         } else {
 155                                                 $externalNormalBlobs[$cluster][$id][] = $row->old_id;
 156                                         }
 157                                 }
 158                                 $dbr->freeResult( $res );
 159                         }
 160
 161                         // Check external concat blobs for the right header
 162                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 163
 164                         // Check external normal blobs for existence
 165                         if ( count( $externalNormalBlobs ) ) {
 166                                 if ( is_null( $this->dbStore ) ) {
 167                                         $this->dbStore = new ExternalStoreDB;
 168                                 }
 169                                 foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {
 170                                         $blobIds = array_keys( $xBlobIds );
 171                                         $extDb =& $this->dbStore->getSlave( $cluster );
 172                                         $blobsTable = $this->dbStore->getTable( $extDb );
 173                                         $res = $extDb->select( $blobsTable,
 174                                                 array( 'blob_id' ),
 175                                                 array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
 176                                         while ( $row = $extDb->fetchObject( $res ) ) {
 177                                                 unset( $xBlobIds[$row->blob_id] );
 178                                         }
 179                                         $extDb->freeResult( $res );
 180                                         // Print errors for missing blobs rows
 181                                         foreach ( $xBlobIds as $blobId => $oldId ) {
 182                                                 $this->error( 'restore text', "Error: missing target $blobId for one-part ES URL", $oldId );
 183                                         }
 184                                 }
 185                         }
 186
 187                         // Check local objects
 188                         $dbr->ping();
 189                         $concatBlobs = array();
 190                         $curIds = array();
 191                         if ( count( $objectRevs ) ) {
 192                                 $headerLength = 300;
 193                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ),
 194                                         array( 'old_id IN (' . implode( ',', $objectRevs ) . ')' ), $fname );
 195                                 while ( $row = $dbr->fetchObject( $res ) ) {
 196                                         $oldId = $row->old_id;
 197                                         $matches = array();
 198                                         if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {
 199                                                 $this->error( 'restore text', "Error: invalid object header", $oldId );
 200                                                 continue;
 201                                         }
 202
 203                                         $className = strtolower( $matches[2] );
 204                                         if ( strlen( $className ) != $matches[1] ) {
 205                                                 $this->error( 'restore text', "Error: invalid object header, wrong class name length", $oldId );
 206                                                 continue;
 207                                         }
 208
 209                                         $objectStats = $objectStats + array( $className => 0 );
 210                                         $objectStats[$className]++;
 211
 212                                         switch ( $className ) {
 213                                                 case 'concatenatedgziphistoryblob':
 214                                                         // Good
 215                                                         break;
 216                                                 case 'historyblobstub':
 217                                                 case 'historyblobcurstub':
 218                                                         if ( strlen( $row->header ) == $headerLength ) {
 219                                                                 $this->error( 'unfixable', "Error: overlong stub header", $oldId );
 220                                                                 continue;
 221                                                         }
 222                                                         $stubObj = unserialize( $row->header );
 223                                                         if ( !is_object( $stubObj ) ) {
 224                                                                 $this->error( 'restore text', "Error: unable to unserialize stub object", $oldId );
 225                                                                 continue;
 226                                                         }
 227                                                         if ( $className == 'historyblobstub' ) {
 228                                                                 $concatBlobs[$stubObj->mOldId][] = $oldId;
 229                                                         } else {
 230                                                                 $curIds[$stubObj->mCurId][] = $oldId;
 231                                                         }
 232                                                         break;
 233                                                 default:
 234                                                         $this->error( 'unfixable', "Error: unrecognised object class \"$className\"", $oldId );
 235                                         }
 236                                 }
 237                                 $dbr->freeResult( $res );
 238                         }
 239
 240                         // Check local concat blob validity
 241                         $externalConcatBlobs = array();
 242                         if ( count( $concatBlobs ) ) {
 243                                 $headerLength = 300;
 244                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ),
 245                                         array( 'old_id IN (' . implode( ',', array_keys( $concatBlobs ) ) . ')' ), $fname );
 246                                 while ( $row = $dbr->fetchObject( $res ) ) {
 247                                         $flags = explode( ',', $row->old_flags );
 248                                         if ( in_array( 'external', $flags ) ) {
 249                                                 // Concat blob is in external storage?
 250                                                 if ( in_array( 'object', $flags ) ) {
 251                                                         $urlParts = explode( '/', $row->header );
 252                                                         if ( $urlParts[0] != 'DB:' ) {
 253                                                                 $this->error( 'unfixable', "Error: unrecognised external storage type \"{$urlParts[0]}", $row->old_id );
 254                                                         } else {
 255                                                                 $cluster = $urlParts[2];
 256                                                                 $id = $urlParts[3];
 257                                                                 if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {
 258                                                                         $externalConcatBlobs[$cluster][$id] = array();
 259                                                                 }
 260                                                                 $externalConcatBlobs[$cluster][$id] = array_merge(
 261                                                                         $externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]
 262                                                                 );
 263                                                         }
 264                                                 } else {
 265                                                         $this->error( 'unfixable', "Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",
 266                                                                 $concatBlobs[$row->old_id] );
 267                                                 }
 268                                         } elseif ( strcasecmp( substr( $row->header, 0, strlen( CONCAT_HEADER ) ), CONCAT_HEADER ) ) {
 269                                                 $this->error( 'restore text', "Error: Incorrect object header for concat bulk row {$row->old_id}",
 270                                                         $concatBlobs[$row->old_id] );
 271                                         } # else good
 272
 273                                         unset( $concatBlobs[$row->old_id] );
 274                                 }
 275                                 $dbr->freeResult( $res );
 276                         }
 277
 278                         // Check targets of unresolved stubs
 279                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 280
 281                         // next chunk
 282                 }
 283
 284                 print "\n\nErrors:\n";
 285                 foreach( $this->errors as $name => $errors ) {
 286                         if ( count( $errors ) ) {
 287                                 $description = $this->errorDescriptions[$name];
 288                                 echo "$description: " . implode( ',', array_keys( $errors ) ) . "\n";
 289                         }
 290                 }
 291
 292                 if ( count( $this->errors['restore text'] ) && $fix ) {
 293                         if ( (string)$xml !== '' ) {
 294                                 $this->restoreText( array_keys( $this->errors['restore text'] ), $xml );
 295                         } else {
 296                                 echo "Can't fix text, no XML backup specified\n";
 297                         }
 298                 }
 299
 300                 print "\nFlag statistics:\n";
 301                 $total = array_sum( $flagStats );
 302                 foreach ( $flagStats as $flag => $count ) {
 303                         printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );
 304                 }
 305                 print "\nLocal object statistics:\n";
 306                 $total = array_sum( $objectStats );
 307                 foreach ( $objectStats as $className => $count ) {
 308                         printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );
 309                 }
 310         }
 311
 312
 313         function error( $type, $msg, $ids ) {
 314                 if ( is_array( $ids ) && count( $ids ) == 1 ) {
 315                         $ids = reset( $ids );
 316                 }
 317                 if ( is_array( $ids ) ) {
 318                         $revIds = array();
 319                         foreach ( $ids as $id ) {
 320                                 $revIds = array_merge( $revIds, array_keys( $this->oldIdMap, $id ) );
 321                         }
 322                         print "$msg in text rows " . implode( ', ', $ids ) .
 323                                 ", revisions " . implode( ', ', $revIds ) . "\n";
 324                 } else {
 325                         $id = $ids;
 326                         $revIds = array_keys( $this->oldIdMap, $id );
 327                         if ( count( $revIds ) == 1 ) {
 328                                 print "$msg in old_id $id, rev_id {$revIds[0]}\n";
 329                         } else {
 330                                 print "$msg in old_id $id, revisions " . implode( ', ', $revIds ) . "\n";
 331                         }
 332                 }
 333                 $this->errors[$type] = $this->errors[$type] + array_flip( $revIds );
 334         }
 335
 336         function checkExternalConcatBlobs( $externalConcatBlobs ) {
 337                 $fname = 'CheckStorage::checkExternalConcatBlobs';
 338                 if ( !count( $externalConcatBlobs ) ) {
 339                         return;
 340                 }
 341
 342                 if ( is_null( $this->dbStore ) ) {
 343                         $this->dbStore = new ExternalStoreDB;
 344                 }
 345
 346                 foreach ( $externalConcatBlobs as $cluster => $oldIds ) {
 347                         $blobIds = array_keys( $oldIds );
 348                         $extDb =& $this->dbStore->getSlave( $cluster );
 349                         $blobsTable = $this->dbStore->getTable( $extDb );
 350                         $headerLength = strlen( CONCAT_HEADER );
 351                         $res = $extDb->select( $blobsTable,
 352                                 array( 'blob_id', "LEFT(blob_text, $headerLength) AS header" ),
 353                                 array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
 354                         while ( $row = $extDb->fetchObject( $res ) ) {
 355                                 if ( strcasecmp( $row->header, CONCAT_HEADER ) ) {
 356                                         $this->error( 'restore text', "Error: invalid header on target $cluster/{$row->blob_id} of two-part ES URL",
 357                                                 $oldIds[$row->blob_id] );
 358                                 }
 359                                 unset( $oldIds[$row->blob_id] );
 360
 361                         }
 362                         $extDb->freeResult( $res );
 363
 364                         // Print errors for missing blobs rows
 365                         foreach ( $oldIds as $blobId => $oldIds ) {
 366                                 $this->error( 'restore text', "Error: missing target $cluster/$blobId for two-part ES URL", $oldIds );
 367                         }
 368                 }
 369         }
 370
 371         function restoreText( $revIds, $xml ) {
 372                 global $wgTmpDirectory, $wgDBname;
 373
 374                 if ( !count( $revIds ) ) {
 375                         return;
 376                 }
 377
 378                 print "Restoring text from XML backup...\n";
 379
 380                 $revFileName = "$wgTmpDirectory/broken-revlist-$wgDBname";
 381                 $filteredXmlFileName = "$wgTmpDirectory/filtered-$wgDBname.xml";
 382
 383                 // Write revision list
 384                 if ( !file_put_contents( $revFileName, implode( "\n", $revIds ) ) ) {
 385                         echo "Error writing revision list, can't restore text\n";
 386                         return;
 387                 }
 388
 389                 // Run mwdumper
 390                 echo "Filtering XML dump...\n";
 391                 $exitStatus = 0;
 392                 passthru( 'mwdumper ' .
 393                         wfEscapeShellArg(
 394                                 "--output=file:$filteredXmlFileName",
 395                                 "--filter=revlist:$revFileName",
 396                                 $xml
 397                         ), $exitStatus
 398                 );
 399
 400                 if ( $exitStatus ) {
 401                         echo "mwdumper died with exit status $exitStatus\n";
 402                         return;
 403                 }
 404
 405                 $file = fopen( $filteredXmlFileName, 'r' );
 406                 if ( !$file ) {
 407                         echo "Unable to open filtered XML file\n";
 408                         return;
 409                 }
 410
 411                 $dbr = wfGetDB( DB_SLAVE );
 412                 $dbw = wfGetDB( DB_MASTER );
 413                 $dbr->ping();
 414                 $dbw->ping();
 415
 416                 $source = new ImportStreamSource( $file );
 417                 $importer = new WikiImporter( $source );
 418                 $importer->setRevisionCallback( array( &$this, 'importRevision' ) );
 419                 $importer->doImport();
 420         }
 421
 422         function importRevision( &$revision, &$importer ) {
 423                 $fname = 'CheckStorage::importRevision';
 424
 425                 $id = $revision->getID();
 426                 $text = $revision->getText();
 427                 if ( $text === '' ) {
 428                         // This is what happens if the revision was broken at the time the
 429                         // dump was made. Unfortunately, it also happens if the revision was
 430                         // legitimately blank, so there's no way to tell the difference. To
 431                         // be safe, we'll skip it and leave it broken
 432                         $id = $id ? $id : '';
 433                         echo "Revision $id is blank in the dump, may have been broken before export\n";
 434                         return;
 435                 }
 436
 437                 if ( !$id )  {
 438                         // No ID, can't import
 439                         echo "No id tag in revision, can't import\n";
 440                         return;
 441                 }
 442
 443                 // Find text row again
 444                 $dbr = wfGetDB( DB_SLAVE );
 445                 $oldId = $dbr->selectField( 'revision', 'rev_text_id', array( 'rev_id' => $id ), $fname );
 446                 if ( !$oldId ) {
 447                         echo "Missing revision row for rev_id $id\n";
 448                         return;
 449                 }
 450
 451                 // Compress the text
 452                 $flags = Revision::compressRevisionText( $text );
 453
 454                 // Update the text row
 455                 $dbw = wfGetDB( DB_MASTER );
 456                 $dbw->update( 'text',
 457                         array( 'old_flags' => $flags, 'old_text' => $text ),
 458                         array( 'old_id' => $oldId ),
 459                         $fname, array( 'LIMIT' => 1 )
 460                 );
 461
 462                 // Remove it from the unfixed list and add it to the fixed list
 463                 unset( $this->errors['restore text'][$id] );
 464                 $this->errors['fixed'][$id] = true;
 465         }
 466 }
 467