maintenance/storage/checkStorage.php

   1 <?php
   2 /**
   3  * Fsck for MediaWiki
   4  *
   5  * @file
   6  * @ingroup Maintenance ExternalStorage
   7  */
   8
   9 define( 'CONCAT_HEADER', 'O:27:"concatenatedgziphistoryblob"' );
  10
  11 if ( !defined( 'MEDIAWIKI' ) ) {
  12         require_once( dirname( __FILE__ ) . '/../commandLine.inc' );
  13
  14         $cs = new CheckStorage;
  15         $fix = isset( $options['fix'] );
  16         if ( isset( $args[0] ) ) {
  17                 $xml = $args[0];
  18         } else {
  19                 $xml = false;
  20         }
  21         $cs->check( $fix, $xml );
  22 }
  23
  24
  25 // ----------------------------------------------------------------------------------
  26
  27 /**
  28  * @ingroup Maintenance ExternalStorage
  29  */
  30 class CheckStorage {
  31         var $oldIdMap, $errors;
  32         var $dbStore = null;
  33
  34         var $errorDescriptions = array(
  35                 'restore text' => 'Damaged text, need to be restored from a backup',
  36                 'restore revision' => 'Damaged revision row, need to be restored from a backup',
  37                 'unfixable' => 'Unexpected errors with no automated fixing method',
  38                 'fixed' => 'Errors already fixed',
  39                 'fixable' => 'Errors which would already be fixed if --fix was specified',
  40         );
  41
  42         function check( $fix = false, $xml = '' ) {
  43                 $fname = 'checkStorage';
  44                 $dbr = wfGetDB( DB_SLAVE );
  45                 if ( $fix ) {
  46                         $dbw = wfGetDB( DB_MASTER );
  47                         print "Checking, will fix errors if possible...\n";
  48                 } else {
  49                         print "Checking...\n";
  50                 }
  51                 $maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, $fname );
  52                 $chunkSize = 1000;
  53                 $flagStats = array();
  54                 $objectStats = array();
  55                 $knownFlags = array( 'external', 'gzip', 'object', 'utf-8' );
  56                 $this->errors = array(
  57                         'restore text' => array(),
  58                         'restore revision' => array(),
  59                         'unfixable' => array(),
  60                         'fixed' => array(),
  61                         'fixable' => array(),
  62                 );
  63
  64                 for ( $chunkStart = 1 ; $chunkStart < $maxRevId; $chunkStart += $chunkSize ) {
  65                         $chunkEnd = $chunkStart + $chunkSize - 1;
  66                         // print "$chunkStart of $maxRevId\n";
  67
  68                         // Fetch revision rows
  69                         $this->oldIdMap = array();
  70                         $dbr->ping();
  71                         $res = $dbr->select( 'revision', array( 'rev_id', 'rev_text_id' ),
  72                                 array( "rev_id BETWEEN $chunkStart AND $chunkEnd" ), $fname );
  73                         foreach ( $res as $row ) {
  74                                 $this->oldIdMap[$row->rev_id] = $row->rev_text_id;
  75                         }
  76                         $dbr->freeResult( $res );
  77
  78                         if ( !count( $this->oldIdMap ) ) {
  79                                 continue;
  80                         }
  81
  82                         // Fetch old_flags
  83                         $missingTextRows = array_flip( $this->oldIdMap );
  84                         $externalRevs = array();
  85                         $objectRevs = array();
  86                         $res = $dbr->select( 'text', array( 'old_id', 'old_flags' ),
  87                                 'old_id IN (' . implode( ',', $this->oldIdMap ) . ')', $fname );
  88                         foreach ( $res as $row ) {
  89                                 $flags = $row->old_flags;
  90                                 $id = $row->old_id;
  91
  92                                 // Create flagStats row if it doesn't exist
  93                                 $flagStats = $flagStats + array( $flags => 0 );
  94                                 // Increment counter
  95                                 $flagStats[$flags]++;
  96
  97                                 // Not missing
  98                                 unset( $missingTextRows[$row->old_id] );
  99
 100                                 // Check for external or object
 101                                 if ( $flags == '' ) {
 102                                         $flagArray = array();
 103                                 } else {
 104                                         $flagArray = explode( ',', $flags );
 105                                 }
 106                                 if ( in_array( 'external', $flagArray ) ) {
 107                                         $externalRevs[] = $id;
 108                                 } elseif ( in_array( 'object', $flagArray ) ) {
 109                                         $objectRevs[] = $id;
 110                                 }
 111
 112                                 // Check for unrecognised flags
 113                                 if ( $flags == '0' ) {
 114                                         // This is a known bug from 2004
 115                                         // It's safe to just erase the old_flags field
 116                                         if ( $fix ) {
 117                                                 $this->error( 'fixed', "Warning: old_flags set to 0", $id );
 118                                                 $dbw->ping();
 119                                                 $dbw->update( 'text', array( 'old_flags' => '' ),
 120                                                         array( 'old_id' => $id ), $fname );
 121                                                 echo "Fixed\n";
 122                                         } else {
 123                                                 $this->error( 'fixable', "Warning: old_flags set to 0", $id );
 124                                         }
 125                                 } elseif ( count( array_diff( $flagArray, $knownFlags ) ) ) {
 126                                         $this->error( 'unfixable', "Error: invalid flags field \"$flags\"", $id );
 127                                 }
 128                         }
 129                         $dbr->freeResult( $res );
 130
 131                         // Output errors for any missing text rows
 132                         foreach ( $missingTextRows as $oldId => $revId ) {
 133                                 $this->error( 'restore revision', "Error: missing text row", $oldId );
 134                         }
 135
 136                         // Verify external revisions
 137                         $externalConcatBlobs = array();
 138                         $externalNormalBlobs = array();
 139                         if ( count( $externalRevs ) ) {
 140                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', 'old_text' ),
 141                                         array( 'old_id IN (' . implode( ',', $externalRevs ) . ')' ), $fname );
 142                                 foreach ( $res as $row ) {
 143                                         $urlParts = explode( '://', $row->old_text, 2 );
 144                                         if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {
 145                                                 $this->error( 'restore text', "Error: invalid URL \"{$row->old_text}\"", $row->old_id );
 146                                                 continue;
 147                                         }
 148                                         list( $proto, ) = $urlParts;
 149                                         if ( $proto != 'DB' ) {
 150                                                 $this->error( 'restore text', "Error: invalid external protocol \"$proto\"", $row->old_id );
 151                                                 continue;
 152                                         }
 153                                         $path = explode( '/', $row->old_text );
 154                                         $cluster = $path[2];
 155                                         $id = $path[3];
 156                                         if ( isset( $path[4] ) ) {
 157                                                 $externalConcatBlobs[$cluster][$id][] = $row->old_id;
 158                                         } else {
 159                                                 $externalNormalBlobs[$cluster][$id][] = $row->old_id;
 160                                         }
 161                                 }
 162                                 $dbr->freeResult( $res );
 163                         }
 164
 165                         // Check external concat blobs for the right header
 166                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 167
 168                         // Check external normal blobs for existence
 169                         if ( count( $externalNormalBlobs ) ) {
 170                                 if ( is_null( $this->dbStore ) ) {
 171                                         $this->dbStore = new ExternalStoreDB;
 172                                 }
 173                                 foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {
 174                                         $blobIds = array_keys( $xBlobIds );
 175                                         $extDb =& $this->dbStore->getSlave( $cluster );
 176                                         $blobsTable = $this->dbStore->getTable( $extDb );
 177                                         $res = $extDb->select( $blobsTable,
 178                                                 array( 'blob_id' ),
 179                                                 array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
 180                                         foreach ( $res as $row ) {
 181                                                 unset( $xBlobIds[$row->blob_id] );
 182                                         }
 183                                         $extDb->freeResult( $res );
 184                                         // Print errors for missing blobs rows
 185                                         foreach ( $xBlobIds as $blobId => $oldId ) {
 186                                                 $this->error( 'restore text', "Error: missing target $blobId for one-part ES URL", $oldId );
 187                                         }
 188                                 }
 189                         }
 190
 191                         // Check local objects
 192                         $dbr->ping();
 193                         $concatBlobs = array();
 194                         $curIds = array();
 195                         if ( count( $objectRevs ) ) {
 196                                 $headerLength = 300;
 197                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ),
 198                                         array( 'old_id IN (' . implode( ',', $objectRevs ) . ')' ), $fname );
 199                                 foreach ( $res as $row ) {
 200                                         $oldId = $row->old_id;
 201                                         $matches = array();
 202                                         if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {
 203                                                 $this->error( 'restore text', "Error: invalid object header", $oldId );
 204                                                 continue;
 205                                         }
 206
 207                                         $className = strtolower( $matches[2] );
 208                                         if ( strlen( $className ) != $matches[1] ) {
 209                                                 $this->error( 'restore text', "Error: invalid object header, wrong class name length", $oldId );
 210                                                 continue;
 211                                         }
 212
 213                                         $objectStats = $objectStats + array( $className => 0 );
 214                                         $objectStats[$className]++;
 215
 216                                         switch ( $className ) {
 217                                                 case 'concatenatedgziphistoryblob':
 218                                                         // Good
 219                                                         break;
 220                                                 case 'historyblobstub':
 221                                                 case 'historyblobcurstub':
 222                                                         if ( strlen( $row->header ) == $headerLength ) {
 223                                                                 $this->error( 'unfixable', "Error: overlong stub header", $oldId );
 224                                                                 continue;
 225                                                         }
 226                                                         $stubObj = unserialize( $row->header );
 227                                                         if ( !is_object( $stubObj ) ) {
 228                                                                 $this->error( 'restore text', "Error: unable to unserialize stub object", $oldId );
 229                                                                 continue;
 230                                                         }
 231                                                         if ( $className == 'historyblobstub' ) {
 232                                                                 $concatBlobs[$stubObj->mOldId][] = $oldId;
 233                                                         } else {
 234                                                                 $curIds[$stubObj->mCurId][] = $oldId;
 235                                                         }
 236                                                         break;
 237                                                 default:
 238                                                         $this->error( 'unfixable', "Error: unrecognised object class \"$className\"", $oldId );
 239                                         }
 240                                 }
 241                                 $dbr->freeResult( $res );
 242                         }
 243
 244                         // Check local concat blob validity
 245                         $externalConcatBlobs = array();
 246                         if ( count( $concatBlobs ) ) {
 247                                 $headerLength = 300;
 248                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ),
 249                                         array( 'old_id IN (' . implode( ',', array_keys( $concatBlobs ) ) . ')' ), $fname );
 250                                 foreach ( $res as $row ) {
 251                                         $flags = explode( ',', $row->old_flags );
 252                                         if ( in_array( 'external', $flags ) ) {
 253                                                 // Concat blob is in external storage?
 254                                                 if ( in_array( 'object', $flags ) ) {
 255                                                         $urlParts = explode( '/', $row->header );
 256                                                         if ( $urlParts[0] != 'DB:' ) {
 257                                                                 $this->error( 'unfixable', "Error: unrecognised external storage type \"{$urlParts[0]}", $row->old_id );
 258                                                         } else {
 259                                                                 $cluster = $urlParts[2];
 260                                                                 $id = $urlParts[3];
 261                                                                 if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {
 262                                                                         $externalConcatBlobs[$cluster][$id] = array();
 263                                                                 }
 264                                                                 $externalConcatBlobs[$cluster][$id] = array_merge(
 265                                                                         $externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]
 266                                                                 );
 267                                                         }
 268                                                 } else {
 269                                                         $this->error( 'unfixable', "Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",
 270                                                                 $concatBlobs[$row->old_id] );
 271                                                 }
 272                                         } elseif ( strcasecmp( substr( $row->header, 0, strlen( CONCAT_HEADER ) ), CONCAT_HEADER ) ) {
 273                                                 $this->error( 'restore text', "Error: Incorrect object header for concat bulk row {$row->old_id}",
 274                                                         $concatBlobs[$row->old_id] );
 275                                         } # else good
 276
 277                                         unset( $concatBlobs[$row->old_id] );
 278                                 }
 279                                 $dbr->freeResult( $res );
 280                         }
 281
 282                         // Check targets of unresolved stubs
 283                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 284
 285                         // next chunk
 286                 }
 287
 288                 print "\n\nErrors:\n";
 289                 foreach ( $this->errors as $name => $errors ) {
 290                         if ( count( $errors ) ) {
 291                                 $description = $this->errorDescriptions[$name];
 292                                 echo "$description: " . implode( ',', array_keys( $errors ) ) . "\n";
 293                         }
 294                 }
 295
 296                 if ( count( $this->errors['restore text'] ) && $fix ) {
 297                         if ( (string)$xml !== '' ) {
 298                                 $this->restoreText( array_keys( $this->errors['restore text'] ), $xml );
 299                         } else {
 300                                 echo "Can't fix text, no XML backup specified\n";
 301                         }
 302                 }
 303
 304                 print "\nFlag statistics:\n";
 305                 $total = array_sum( $flagStats );
 306                 foreach ( $flagStats as $flag => $count ) {
 307                         printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );
 308                 }
 309                 print "\nLocal object statistics:\n";
 310                 $total = array_sum( $objectStats );
 311                 foreach ( $objectStats as $className => $count ) {
 312                         printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );
 313                 }
 314         }
 315
 316
 317         function error( $type, $msg, $ids ) {
 318                 if ( is_array( $ids ) && count( $ids ) == 1 ) {
 319                         $ids = reset( $ids );
 320                 }
 321                 if ( is_array( $ids ) ) {
 322                         $revIds = array();
 323                         foreach ( $ids as $id ) {
 324                                 $revIds = array_merge( $revIds, array_keys( $this->oldIdMap, $id ) );
 325                         }
 326                         print "$msg in text rows " . implode( ', ', $ids ) .
 327                                 ", revisions " . implode( ', ', $revIds ) . "\n";
 328                 } else {
 329                         $id = $ids;
 330                         $revIds = array_keys( $this->oldIdMap, $id );
 331                         if ( count( $revIds ) == 1 ) {
 332                                 print "$msg in old_id $id, rev_id {$revIds[0]}\n";
 333                         } else {
 334                                 print "$msg in old_id $id, revisions " . implode( ', ', $revIds ) . "\n";
 335                         }
 336                 }
 337                 $this->errors[$type] = $this->errors[$type] + array_flip( $revIds );
 338         }
 339
 340         function checkExternalConcatBlobs( $externalConcatBlobs ) {
 341                 $fname = 'CheckStorage::checkExternalConcatBlobs';
 342                 if ( !count( $externalConcatBlobs ) ) {
 343                         return;
 344                 }
 345
 346                 if ( is_null( $this->dbStore ) ) {
 347                         $this->dbStore = new ExternalStoreDB;
 348                 }
 349
 350                 foreach ( $externalConcatBlobs as $cluster => $oldIds ) {
 351                         $blobIds = array_keys( $oldIds );
 352                         $extDb =& $this->dbStore->getSlave( $cluster );
 353                         $blobsTable = $this->dbStore->getTable( $extDb );
 354                         $headerLength = strlen( CONCAT_HEADER );
 355                         $res = $extDb->select( $blobsTable,
 356                                 array( 'blob_id', "LEFT(blob_text, $headerLength) AS header" ),
 357                                 array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
 358                         foreach ( $res as $row ) {
 359                                 if ( strcasecmp( $row->header, CONCAT_HEADER ) ) {
 360                                         $this->error( 'restore text', "Error: invalid header on target $cluster/{$row->blob_id} of two-part ES URL",
 361                                                 $oldIds[$row->blob_id] );
 362                                 }
 363                                 unset( $oldIds[$row->blob_id] );
 364
 365                         }
 366                         $extDb->freeResult( $res );
 367
 368                         // Print errors for missing blobs rows
 369                         foreach ( $oldIds as $blobId => $oldIds ) {
 370                                 $this->error( 'restore text', "Error: missing target $cluster/$blobId for two-part ES URL", $oldIds );
 371                         }
 372                 }
 373         }
 374
 375         function restoreText( $revIds, $xml ) {
 376                 global $wgTmpDirectory, $wgDBname;
 377
 378                 if ( !count( $revIds ) ) {
 379                         return;
 380                 }
 381
 382                 print "Restoring text from XML backup...\n";
 383
 384                 $revFileName = "$wgTmpDirectory/broken-revlist-$wgDBname";
 385                 $filteredXmlFileName = "$wgTmpDirectory/filtered-$wgDBname.xml";
 386
 387                 // Write revision list
 388                 if ( !file_put_contents( $revFileName, implode( "\n", $revIds ) ) ) {
 389                         echo "Error writing revision list, can't restore text\n";
 390                         return;
 391                 }
 392
 393                 // Run mwdumper
 394                 echo "Filtering XML dump...\n";
 395                 $exitStatus = 0;
 396                 passthru( 'mwdumper ' .
 397                         wfEscapeShellArg(
 398                                 "--output=file:$filteredXmlFileName",
 399                                 "--filter=revlist:$revFileName",
 400                                 $xml
 401                         ), $exitStatus
 402                 );
 403
 404                 if ( $exitStatus ) {
 405                         echo "mwdumper died with exit status $exitStatus\n";
 406                         return;
 407                 }
 408
 409                 $file = fopen( $filteredXmlFileName, 'r' );
 410                 if ( !$file ) {
 411                         echo "Unable to open filtered XML file\n";
 412                         return;
 413                 }
 414
 415                 $dbr = wfGetDB( DB_SLAVE );
 416                 $dbw = wfGetDB( DB_MASTER );
 417                 $dbr->ping();
 418                 $dbw->ping();
 419
 420                 $source = new ImportStreamSource( $file );
 421                 $importer = new WikiImporter( $source );
 422                 $importer->setRevisionCallback( array( &$this, 'importRevision' ) );
 423                 $importer->doImport();
 424         }
 425
 426         function importRevision( &$revision, &$importer ) {
 427                 $fname = 'CheckStorage::importRevision';
 428
 429                 $id = $revision->getID();
 430                 $text = $revision->getText();
 431                 if ( $text === '' ) {
 432                         // This is what happens if the revision was broken at the time the
 433                         // dump was made. Unfortunately, it also happens if the revision was
 434                         // legitimately blank, so there's no way to tell the difference. To
 435                         // be safe, we'll skip it and leave it broken
 436                         $id = $id ? $id : '';
 437                         echo "Revision $id is blank in the dump, may have been broken before export\n";
 438                         return;
 439                 }
 440
 441                 if ( !$id )  {
 442                         // No ID, can't import
 443                         echo "No id tag in revision, can't import\n";
 444                         return;
 445                 }
 446
 447                 // Find text row again
 448                 $dbr = wfGetDB( DB_SLAVE );
 449                 $oldId = $dbr->selectField( 'revision', 'rev_text_id', array( 'rev_id' => $id ), $fname );
 450                 if ( !$oldId ) {
 451                         echo "Missing revision row for rev_id $id\n";
 452                         return;
 453                 }
 454
 455                 // Compress the text
 456                 $flags = Revision::compressRevisionText( $text );
 457
 458                 // Update the text row
 459                 $dbw = wfGetDB( DB_MASTER );
 460                 $dbw->update( 'text',
 461                         array( 'old_flags' => $flags, 'old_text' => $text ),
 462                         array( 'old_id' => $oldId ),
 463                         $fname, array( 'LIMIT' => 1 )
 464                 );
 465
 466                 // Remove it from the unfixed list and add it to the fixed list
 467                 unset( $this->errors['restore text'][$id] );
 468                 $this->errors['fixed'][$id] = true;
 469         }
 470 }
 471