maintenance/storage/trackBlobs.php

   1 <?php
   2 /**
   3  * Adds blobs from a given external storage cluster to the blob_tracking table.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Maintenance
  22  * @see wfWaitForSlaves()
  23  */
  24
  25 require __DIR__ . '/../commandLine.inc';
  26
  27 if ( count( $args ) < 1 ) {
  28         echo "Usage: php trackBlobs.php <cluster> [... <cluster>]\n";
  29         echo "Adds blobs from a given ES cluster to the blob_tracking table\n";
  30         echo "Automatically deletes the tracking table and starts from the start again when restarted.\n";
  31
  32         exit( 1 );
  33 }
  34 $tracker = new TrackBlobs( $args );
  35 $tracker->run();
  36 echo "All done.\n";
  37
  38 class TrackBlobs {
  39         public $clusters, $textClause;
  40         public $doBlobOrphans;
  41         public $trackedBlobs = [];
  42
  43         public $batchSize = 1000;
  44         public $reportingInterval = 10;
  45
  46         function __construct( $clusters ) {
  47                 $this->clusters = $clusters;
  48                 if ( extension_loaded( 'gmp' ) ) {
  49                         $this->doBlobOrphans = true;
  50                         foreach ( $clusters as $cluster ) {
  51                                 $this->trackedBlobs[$cluster] = gmp_init( 0 );
  52                         }
  53                 } else {
  54                         echo "Warning: the gmp extension is needed to find orphan blobs\n";
  55                 }
  56         }
  57
  58         function run() {
  59                 $this->checkIntegrity();
  60                 $this->initTrackingTable();
  61                 $this->trackRevisions();
  62                 $this->trackOrphanText();
  63                 if ( $this->doBlobOrphans ) {
  64                         $this->findOrphanBlobs();
  65                 }
  66         }
  67
  68         function checkIntegrity() {
  69                 echo "Doing integrity check...\n";
  70                 $dbr = wfGetDB( DB_REPLICA );
  71
  72                 // Scan for HistoryBlobStub objects in the text table (bug 20757)
  73
  74                 $exists = $dbr->selectField( 'text', 1,
  75                         'old_flags LIKE \'%object%\' AND old_flags NOT LIKE \'%external%\' ' .
  76                         'AND LOWER(CONVERT(LEFT(old_text,22) USING latin1)) = \'o:15:"historyblobstub"\'',
  77                         __METHOD__
  78                 );
  79
  80                 if ( $exists ) {
  81                         echo "Integrity check failed: found HistoryBlobStub objects in your text table.\n" .
  82                                 "This script could destroy these objects if it continued. Run resolveStubs.php\n" .
  83                                 "to fix this.\n";
  84                         exit( 1 );
  85                 }
  86
  87                 // Scan the archive table for HistoryBlobStub objects or external flags (bug 22624)
  88                 $flags = $dbr->selectField( 'archive', 'ar_flags',
  89                         'ar_flags LIKE \'%external%\' OR (' .
  90                         'ar_flags LIKE \'%object%\' ' .
  91                         'AND LOWER(CONVERT(LEFT(ar_text,22) USING latin1)) = \'o:15:"historyblobstub"\' )',
  92                         __METHOD__
  93                 );
  94
  95                 if ( strpos( $flags, 'external' ) !== false ) {
  96                         echo "Integrity check failed: found external storage pointers in your archive table.\n" .
  97                                 "Run normaliseArchiveTable.php to fix this.\n";
  98                         exit( 1 );
  99                 } elseif ( $flags ) {
 100                         echo "Integrity check failed: found HistoryBlobStub objects in your archive table.\n" .
 101                                 "These objects are probably already broken, continuing would make them\n" .
 102                                 "unrecoverable. Run \"normaliseArchiveTable.php --fix-cgz-bug\" to fix this.\n";
 103                         exit( 1 );
 104                 }
 105
 106                 echo "Integrity check OK\n";
 107         }
 108
 109         function initTrackingTable() {
 110                 $dbw = wfGetDB( DB_MASTER );
 111                 if ( $dbw->tableExists( 'blob_tracking' ) ) {
 112                         $dbw->query( 'DROP TABLE ' . $dbw->tableName( 'blob_tracking' ) );
 113                         $dbw->query( 'DROP TABLE ' . $dbw->tableName( 'blob_orphans' ) );
 114                 }
 115                 $dbw->sourceFile( __DIR__ . '/blob_tracking.sql' );
 116         }
 117
 118         function getTextClause() {
 119                 if ( !$this->textClause ) {
 120                         $dbr = wfGetDB( DB_REPLICA );
 121                         $this->textClause = '';
 122                         foreach ( $this->clusters as $cluster ) {
 123                                 if ( $this->textClause != '' ) {
 124                                         $this->textClause .= ' OR ';
 125                                 }
 126                                 $this->textClause .= 'old_text' . $dbr->buildLike( "DB://$cluster/", $dbr->anyString() );
 127                         }
 128                 }
 129
 130                 return $this->textClause;
 131         }
 132
 133         function interpretPointer( $text ) {
 134                 if ( !preg_match( '!^DB://(\w+)/(\d+)(?:/([0-9a-fA-F]+)|)$!', $text, $m ) ) {
 135                         return false;
 136                 }
 137
 138                 return [
 139                         'cluster' => $m[1],
 140                         'id' => intval( $m[2] ),
 141                         'hash' => isset( $m[3] ) ? $m[3] : null
 142                 ];
 143         }
 144
 145         /**
 146          *  Scan the revision table for rows stored in the specified clusters
 147          */
 148         function trackRevisions() {
 149                 $dbw = wfGetDB( DB_MASTER );
 150                 $dbr = wfGetDB( DB_REPLICA );
 151
 152                 $textClause = $this->getTextClause();
 153                 $startId = 0;
 154                 $endId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, __METHOD__ );
 155                 $batchesDone = 0;
 156                 $rowsInserted = 0;
 157
 158                 echo "Finding revisions...\n";
 159
 160                 while ( true ) {
 161                         $res = $dbr->select( [ 'revision', 'text' ],
 162                                 [ 'rev_id', 'rev_page', 'old_id', 'old_flags', 'old_text' ],
 163                                 [
 164                                         'rev_id > ' . $dbr->addQuotes( $startId ),
 165                                         'rev_text_id=old_id',
 166                                         $textClause,
 167                                         'old_flags ' . $dbr->buildLike( $dbr->anyString(), 'external', $dbr->anyString() ),
 168                                 ],
 169                                 __METHOD__,
 170                                 [
 171                                         'ORDER BY' => 'rev_id',
 172                                         'LIMIT' => $this->batchSize
 173                                 ]
 174                         );
 175                         if ( !$res->numRows() ) {
 176                                 break;
 177                         }
 178
 179                         $insertBatch = [];
 180                         foreach ( $res as $row ) {
 181                                 $startId = $row->rev_id;
 182                                 $info = $this->interpretPointer( $row->old_text );
 183                                 if ( !$info ) {
 184                                         echo "Invalid DB:// URL in rev_id {$row->rev_id}\n";
 185                                         continue;
 186                                 }
 187                                 if ( !in_array( $info['cluster'], $this->clusters ) ) {
 188                                         echo "Invalid cluster returned in SQL query: {$info['cluster']}\n";
 189                                         continue;
 190                                 }
 191                                 $insertBatch[] = [
 192                                         'bt_page' => $row->rev_page,
 193                                         'bt_rev_id' => $row->rev_id,
 194                                         'bt_text_id' => $row->old_id,
 195                                         'bt_cluster' => $info['cluster'],
 196                                         'bt_blob_id' => $info['id'],
 197                                         'bt_cgz_hash' => $info['hash']
 198                                 ];
 199                                 if ( $this->doBlobOrphans ) {
 200                                         gmp_setbit( $this->trackedBlobs[$info['cluster']], $info['id'] );
 201                                 }
 202                         }
 203                         $dbw->insert( 'blob_tracking', $insertBatch, __METHOD__ );
 204                         $rowsInserted += count( $insertBatch );
 205
 206                         ++$batchesDone;
 207                         if ( $batchesDone >= $this->reportingInterval ) {
 208                                 $batchesDone = 0;
 209                                 echo "$startId / $endId\n";
 210                                 wfWaitForSlaves();
 211                         }
 212                 }
 213                 echo "Found $rowsInserted revisions\n";
 214         }
 215
 216         /**
 217          * Scan the text table for orphan text
 218          * Orphan text here does not imply DB corruption -- deleted text tracked by the
 219          * archive table counts as orphan for our purposes.
 220          */
 221         function trackOrphanText() {
 222                 # Wait until the blob_tracking table is available in the replica DB
 223                 $dbw = wfGetDB( DB_MASTER );
 224                 $dbr = wfGetDB( DB_REPLICA );
 225                 $pos = $dbw->getMasterPos();
 226                 $dbr->masterPosWait( $pos, 100000 );
 227
 228                 $textClause = $this->getTextClause( $this->clusters );
 229                 $startId = 0;
 230                 $endId = $dbr->selectField( 'text', 'MAX(old_id)', false, __METHOD__ );
 231                 $rowsInserted = 0;
 232                 $batchesDone = 0;
 233
 234                 echo "Finding orphan text...\n";
 235
 236                 # Scan the text table for orphan text
 237                 while ( true ) {
 238                         $res = $dbr->select( [ 'text', 'blob_tracking' ],
 239                                 [ 'old_id', 'old_flags', 'old_text' ],
 240                                 [
 241                                         'old_id>' . $dbr->addQuotes( $startId ),
 242                                         $textClause,
 243                                         'old_flags ' . $dbr->buildLike( $dbr->anyString(), 'external', $dbr->anyString() ),
 244                                         'bt_text_id IS NULL'
 245                                 ],
 246                                 __METHOD__,
 247                                 [
 248                                         'ORDER BY' => 'old_id',
 249                                         'LIMIT' => $this->batchSize
 250                                 ],
 251                                 [ 'blob_tracking' => [ 'LEFT JOIN', 'bt_text_id=old_id' ] ]
 252                         );
 253                         $ids = [];
 254                         foreach ( $res as $row ) {
 255                                 $ids[] = $row->old_id;
 256                         }
 257
 258                         if ( !$res->numRows() ) {
 259                                 break;
 260                         }
 261
 262                         $insertBatch = [];
 263                         foreach ( $res as $row ) {
 264                                 $startId = $row->old_id;
 265                                 $info = $this->interpretPointer( $row->old_text );
 266                                 if ( !$info ) {
 267                                         echo "Invalid DB:// URL in old_id {$row->old_id}\n";
 268                                         continue;
 269                                 }
 270                                 if ( !in_array( $info['cluster'], $this->clusters ) ) {
 271                                         echo "Invalid cluster returned in SQL query\n";
 272                                         continue;
 273                                 }
 274
 275                                 $insertBatch[] = [
 276                                         'bt_page' => 0,
 277                                         'bt_rev_id' => 0,
 278                                         'bt_text_id' => $row->old_id,
 279                                         'bt_cluster' => $info['cluster'],
 280                                         'bt_blob_id' => $info['id'],
 281                                         'bt_cgz_hash' => $info['hash']
 282                                 ];
 283                                 if ( $this->doBlobOrphans ) {
 284                                         gmp_setbit( $this->trackedBlobs[$info['cluster']], $info['id'] );
 285                                 }
 286                         }
 287                         $dbw->insert( 'blob_tracking', $insertBatch, __METHOD__ );
 288
 289                         $rowsInserted += count( $insertBatch );
 290                         ++$batchesDone;
 291                         if ( $batchesDone >= $this->reportingInterval ) {
 292                                 $batchesDone = 0;
 293                                 echo "$startId / $endId\n";
 294                                 wfWaitForSlaves();
 295                         }
 296                 }
 297                 echo "Found $rowsInserted orphan text rows\n";
 298         }
 299
 300         /**
 301          * Scan the blobs table for rows not registered in blob_tracking (and thus not
 302          * registered in the text table).
 303          *
 304          * Orphan blobs are indicative of DB corruption. They are inaccessible and
 305          * should probably be deleted.
 306          */
 307         function findOrphanBlobs() {
 308                 if ( !extension_loaded( 'gmp' ) ) {
 309                         echo "Can't find orphan blobs, need bitfield support provided by GMP.\n";
 310
 311                         return;
 312                 }
 313
 314                 $dbw = wfGetDB( DB_MASTER );
 315
 316                 foreach ( $this->clusters as $cluster ) {
 317                         echo "Searching for orphan blobs in $cluster...\n";
 318                         $lb = wfGetLBFactory()->getExternalLB( $cluster );
 319                         try {
 320                                 $extDB = $lb->getConnection( DB_REPLICA );
 321                         } catch ( DBConnectionError $e ) {
 322                                 if ( strpos( $e->error, 'Unknown database' ) !== false ) {
 323                                         echo "No database on $cluster\n";
 324                                 } else {
 325                                         echo "Error on $cluster: " . $e->getMessage() . "\n";
 326                                 }
 327                                 continue;
 328                         }
 329                         $table = $extDB->getLBInfo( 'blobs table' );
 330                         if ( is_null( $table ) ) {
 331                                 $table = 'blobs';
 332                         }
 333                         if ( !$extDB->tableExists( $table ) ) {
 334                                 echo "No blobs table on cluster $cluster\n";
 335                                 continue;
 336                         }
 337                         $startId = 0;
 338                         $batchesDone = 0;
 339                         $actualBlobs = gmp_init( 0 );
 340                         $endId = $extDB->selectField( $table, 'MAX(blob_id)', false, __METHOD__ );
 341
 342                         // Build a bitmap of actual blob rows
 343                         while ( true ) {
 344                                 $res = $extDB->select( $table,
 345                                         [ 'blob_id' ],
 346                                         [ 'blob_id > ' . $extDB->addQuotes( $startId ) ],
 347                                         __METHOD__,
 348                                         [ 'LIMIT' => $this->batchSize, 'ORDER BY' => 'blob_id' ]
 349                                 );
 350
 351                                 if ( !$res->numRows() ) {
 352                                         break;
 353                                 }
 354
 355                                 foreach ( $res as $row ) {
 356                                         gmp_setbit( $actualBlobs, $row->blob_id );
 357                                 }
 358                                 $startId = $row->blob_id;
 359
 360                                 ++$batchesDone;
 361                                 if ( $batchesDone >= $this->reportingInterval ) {
 362                                         $batchesDone = 0;
 363                                         echo "$startId / $endId\n";
 364                                 }
 365                         }
 366
 367                         // Find actual blobs that weren't tracked by the previous passes
 368                         // This is a set-theoretic difference A \ B, or in bitwise terms, A & ~B
 369                         $orphans = gmp_and( $actualBlobs, gmp_com( $this->trackedBlobs[$cluster] ) );
 370
 371                         // Traverse the orphan list
 372                         $insertBatch = [];
 373                         $id = 0;
 374                         $numOrphans = 0;
 375                         while ( true ) {
 376                                 $id = gmp_scan1( $orphans, $id );
 377                                 if ( $id == -1 ) {
 378                                         break;
 379                                 }
 380                                 $insertBatch[] = [
 381                                         'bo_cluster' => $cluster,
 382                                         'bo_blob_id' => $id
 383                                 ];
 384                                 if ( count( $insertBatch ) > $this->batchSize ) {
 385                                         $dbw->insert( 'blob_orphans', $insertBatch, __METHOD__ );
 386                                         $insertBatch = [];
 387                                 }
 388
 389                                 ++$id;
 390                                 ++$numOrphans;
 391                         }
 392                         if ( $insertBatch ) {
 393                                 $dbw->insert( 'blob_orphans', $insertBatch, __METHOD__ );
 394                         }
 395                         echo "Found $numOrphans orphan(s) in $cluster\n";
 396                 }
 397         }
 398 }