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 use MediaWiki\MediaWikiServices;
  26 use Wikimedia\Rdbms\DBConnectionError;
  27
  28 require __DIR__ . '/../commandLine.inc';
  29
  30 if ( count( $args ) < 1 ) {
  31         echo "Usage: php trackBlobs.php <cluster> [... <cluster>]\n";
  32         echo "Adds blobs from a given ES cluster to the blob_tracking table\n";
  33         echo "Automatically deletes the tracking table and starts from the start again when restarted.\n";
  34
  35         exit( 1 );
  36 }
  37 $tracker = new TrackBlobs( $args );
  38 $tracker->run();
  39 echo "All done.\n";
  40
  41 class TrackBlobs {
  42         public $clusters, $textClause;
  43         public $doBlobOrphans;
  44         public $trackedBlobs = [];
  45
  46         public $batchSize = 1000;
  47         public $reportingInterval = 10;
  48
  49         function __construct( $clusters ) {
  50                 $this->clusters = $clusters;
  51                 if ( extension_loaded( 'gmp' ) ) {
  52                         $this->doBlobOrphans = true;
  53                         foreach ( $clusters as $cluster ) {
  54                                 $this->trackedBlobs[$cluster] = gmp_init( 0 );
  55                         }
  56                 } else {
  57                         echo "Warning: the gmp extension is needed to find orphan blobs\n";
  58                 }
  59         }
  60
  61         function run() {
  62                 $this->checkIntegrity();
  63                 $this->initTrackingTable();
  64                 $this->trackRevisions();
  65                 $this->trackOrphanText();
  66                 if ( $this->doBlobOrphans ) {
  67                         $this->findOrphanBlobs();
  68                 }
  69         }
  70
  71         function checkIntegrity() {
  72                 echo "Doing integrity check...\n";
  73                 $dbr = wfGetDB( DB_REPLICA );
  74
  75                 // Scan for HistoryBlobStub objects in the text table (T22757)
  76
  77                 $exists = $dbr->selectField( 'text', 1,
  78                         'old_flags LIKE \'%object%\' AND old_flags NOT LIKE \'%external%\' ' .
  79                         'AND LOWER(CONVERT(LEFT(old_text,22) USING latin1)) = \'o:15:"historyblobstub"\'',
  80                         __METHOD__
  81                 );
  82
  83                 if ( $exists ) {
  84                         echo "Integrity check failed: found HistoryBlobStub objects in your text table.\n" .
  85                                 "This script could destroy these objects if it continued. Run resolveStubs.php\n" .
  86                                 "to fix this.\n";
  87                         exit( 1 );
  88                 }
  89
  90                 // Scan the archive table for HistoryBlobStub objects or external flags (T24624)
  91                 $flags = $dbr->selectField( 'archive', 'ar_flags',
  92                         'ar_flags LIKE \'%external%\' OR (' .
  93                         'ar_flags LIKE \'%object%\' ' .
  94                         'AND LOWER(CONVERT(LEFT(ar_text,22) USING latin1)) = \'o:15:"historyblobstub"\' )',
  95                         __METHOD__
  96                 );
  97
  98                 if ( strpos( $flags, 'external' ) !== false ) {
  99                         echo "Integrity check failed: found external storage pointers in your archive table.\n" .
 100                                 "Run normaliseArchiveTable.php to fix this.\n";
 101                         exit( 1 );
 102                 } elseif ( $flags ) {
 103                         echo "Integrity check failed: found HistoryBlobStub objects in your archive table.\n" .
 104                                 "These objects are probably already broken, continuing would make them\n" .
 105                                 "unrecoverable. Run \"normaliseArchiveTable.php --fix-cgz-bug\" to fix this.\n";
 106                         exit( 1 );
 107                 }
 108
 109                 echo "Integrity check OK\n";
 110         }
 111
 112         function initTrackingTable() {
 113                 $dbw = wfGetDB( DB_MASTER );
 114                 if ( $dbw->tableExists( 'blob_tracking' ) ) {
 115                         $dbw->query( 'DROP TABLE ' . $dbw->tableName( 'blob_tracking' ) );
 116                         $dbw->query( 'DROP TABLE ' . $dbw->tableName( 'blob_orphans' ) );
 117                 }
 118                 $dbw->sourceFile( __DIR__ . '/blob_tracking.sql' );
 119         }
 120
 121         function getTextClause() {
 122                 if ( !$this->textClause ) {
 123                         $dbr = wfGetDB( DB_REPLICA );
 124                         $this->textClause = '';
 125                         foreach ( $this->clusters as $cluster ) {
 126                                 if ( $this->textClause != '' ) {
 127                                         $this->textClause .= ' OR ';
 128                                 }
 129                                 $this->textClause .= 'old_text' . $dbr->buildLike( "DB://$cluster/", $dbr->anyString() );
 130                         }
 131                 }
 132
 133                 return $this->textClause;
 134         }
 135
 136         function interpretPointer( $text ) {
 137                 if ( !preg_match( '!^DB://(\w+)/(\d+)(?:/([0-9a-fA-F]+)|)$!', $text, $m ) ) {
 138                         return false;
 139                 }
 140
 141                 return [
 142                         'cluster' => $m[1],
 143                         'id' => intval( $m[2] ),
 144                         'hash' => isset( $m[3] ) ? $m[3] : null
 145                 ];
 146         }
 147
 148         /**
 149          *  Scan the revision table for rows stored in the specified clusters
 150          */
 151         function trackRevisions() {
 152                 $dbw = wfGetDB( DB_MASTER );
 153                 $dbr = wfGetDB( DB_REPLICA );
 154
 155                 $textClause = $this->getTextClause();
 156                 $startId = 0;
 157                 $endId = $dbr->selectField( 'revision', 'MAX(rev_id)', '', __METHOD__ );
 158                 $batchesDone = 0;
 159                 $rowsInserted = 0;
 160
 161                 echo "Finding revisions...\n";
 162
 163                 while ( true ) {
 164                         $res = $dbr->select( [ 'revision', 'text' ],
 165                                 [ 'rev_id', 'rev_page', 'old_id', 'old_flags', 'old_text' ],
 166                                 [
 167                                         'rev_id > ' . $dbr->addQuotes( $startId ),
 168                                         'rev_text_id=old_id',
 169                                         $textClause,
 170                                         'old_flags ' . $dbr->buildLike( $dbr->anyString(), 'external', $dbr->anyString() ),
 171                                 ],
 172                                 __METHOD__,
 173                                 [
 174                                         'ORDER BY' => 'rev_id',
 175                                         'LIMIT' => $this->batchSize
 176                                 ]
 177                         );
 178                         if ( !$res->numRows() ) {
 179                                 break;
 180                         }
 181
 182                         $insertBatch = [];
 183                         foreach ( $res as $row ) {
 184                                 $startId = $row->rev_id;
 185                                 $info = $this->interpretPointer( $row->old_text );
 186                                 if ( !$info ) {
 187                                         echo "Invalid DB:// URL in rev_id {$row->rev_id}\n";
 188                                         continue;
 189                                 }
 190                                 if ( !in_array( $info['cluster'], $this->clusters ) ) {
 191                                         echo "Invalid cluster returned in SQL query: {$info['cluster']}\n";
 192                                         continue;
 193                                 }
 194                                 $insertBatch[] = [
 195                                         'bt_page' => $row->rev_page,
 196                                         'bt_rev_id' => $row->rev_id,
 197                                         'bt_text_id' => $row->old_id,
 198                                         'bt_cluster' => $info['cluster'],
 199                                         'bt_blob_id' => $info['id'],
 200                                         'bt_cgz_hash' => $info['hash']
 201                                 ];
 202                                 if ( $this->doBlobOrphans ) {
 203                                         gmp_setbit( $this->trackedBlobs[$info['cluster']], $info['id'] );
 204                                 }
 205                         }
 206                         $dbw->insert( 'blob_tracking', $insertBatch, __METHOD__ );
 207                         $rowsInserted += count( $insertBatch );
 208
 209                         ++$batchesDone;
 210                         if ( $batchesDone >= $this->reportingInterval ) {
 211                                 $batchesDone = 0;
 212                                 echo "$startId / $endId\n";
 213                                 wfWaitForSlaves();
 214                         }
 215                 }
 216                 echo "Found $rowsInserted revisions\n";
 217         }
 218
 219         /**
 220          * Scan the text table for orphan text
 221          * Orphan text here does not imply DB corruption -- deleted text tracked by the
 222          * archive table counts as orphan for our purposes.
 223          */
 224         function trackOrphanText() {
 225                 # Wait until the blob_tracking table is available in the replica DB
 226                 $dbw = wfGetDB( DB_MASTER );
 227                 $dbr = wfGetDB( DB_REPLICA );
 228                 $pos = $dbw->getMasterPos();
 229                 $dbr->masterPosWait( $pos, 100000 );
 230
 231                 $textClause = $this->getTextClause( $this->clusters );
 232                 $startId = 0;
 233                 $endId = $dbr->selectField( 'text', 'MAX(old_id)', '', __METHOD__ );
 234                 $rowsInserted = 0;
 235                 $batchesDone = 0;
 236
 237                 echo "Finding orphan text...\n";
 238
 239                 # Scan the text table for orphan text
 240                 while ( true ) {
 241                         $res = $dbr->select( [ 'text', 'blob_tracking' ],
 242                                 [ 'old_id', 'old_flags', 'old_text' ],
 243                                 [
 244                                         'old_id>' . $dbr->addQuotes( $startId ),
 245                                         $textClause,
 246                                         'old_flags ' . $dbr->buildLike( $dbr->anyString(), 'external', $dbr->anyString() ),
 247                                         'bt_text_id IS NULL'
 248                                 ],
 249                                 __METHOD__,
 250                                 [
 251                                         'ORDER BY' => 'old_id',
 252                                         'LIMIT' => $this->batchSize
 253                                 ],
 254                                 [ 'blob_tracking' => [ 'LEFT JOIN', 'bt_text_id=old_id' ] ]
 255                         );
 256                         $ids = [];
 257                         foreach ( $res as $row ) {
 258                                 $ids[] = $row->old_id;
 259                         }
 260
 261                         if ( !$res->numRows() ) {
 262                                 break;
 263                         }
 264
 265                         $insertBatch = [];
 266                         foreach ( $res as $row ) {
 267                                 $startId = $row->old_id;
 268                                 $info = $this->interpretPointer( $row->old_text );
 269                                 if ( !$info ) {
 270                                         echo "Invalid DB:// URL in old_id {$row->old_id}\n";
 271                                         continue;
 272                                 }
 273                                 if ( !in_array( $info['cluster'], $this->clusters ) ) {
 274                                         echo "Invalid cluster returned in SQL query\n";
 275                                         continue;
 276                                 }
 277
 278                                 $insertBatch[] = [
 279                                         'bt_page' => 0,
 280                                         'bt_rev_id' => 0,
 281                                         'bt_text_id' => $row->old_id,
 282                                         'bt_cluster' => $info['cluster'],
 283                                         'bt_blob_id' => $info['id'],
 284                                         'bt_cgz_hash' => $info['hash']
 285                                 ];
 286                                 if ( $this->doBlobOrphans ) {
 287                                         gmp_setbit( $this->trackedBlobs[$info['cluster']], $info['id'] );
 288                                 }
 289                         }
 290                         $dbw->insert( 'blob_tracking', $insertBatch, __METHOD__ );
 291
 292                         $rowsInserted += count( $insertBatch );
 293                         ++$batchesDone;
 294                         if ( $batchesDone >= $this->reportingInterval ) {
 295                                 $batchesDone = 0;
 296                                 echo "$startId / $endId\n";
 297                                 wfWaitForSlaves();
 298                         }
 299                 }
 300                 echo "Found $rowsInserted orphan text rows\n";
 301         }
 302
 303         /**
 304          * Scan the blobs table for rows not registered in blob_tracking (and thus not
 305          * registered in the text table).
 306          *
 307          * Orphan blobs are indicative of DB corruption. They are inaccessible and
 308          * should probably be deleted.
 309          */
 310         function findOrphanBlobs() {
 311                 if ( !extension_loaded( 'gmp' ) ) {
 312                         echo "Can't find orphan blobs, need bitfield support provided by GMP.\n";
 313
 314                         return;
 315                 }
 316
 317                 $dbw = wfGetDB( DB_MASTER );
 318
 319                 foreach ( $this->clusters as $cluster ) {
 320                         echo "Searching for orphan blobs in $cluster...\n";
 321                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 322                         $lb = $lbFactory->getExternalLB( $cluster );
 323                         try {
 324                                 $extDB = $lb->getConnection( DB_REPLICA );
 325                         } catch ( DBConnectionError $e ) {
 326                                 if ( strpos( $e->error, 'Unknown database' ) !== false ) {
 327                                         echo "No database on $cluster\n";
 328                                 } else {
 329                                         echo "Error on $cluster: " . $e->getMessage() . "\n";
 330                                 }
 331                                 continue;
 332                         }
 333                         $table = $extDB->getLBInfo( 'blobs table' );
 334                         if ( is_null( $table ) ) {
 335                                 $table = 'blobs';
 336                         }
 337                         if ( !$extDB->tableExists( $table ) ) {
 338                                 echo "No blobs table on cluster $cluster\n";
 339                                 continue;
 340                         }
 341                         $startId = 0;
 342                         $batchesDone = 0;
 343                         $actualBlobs = gmp_init( 0 );
 344                         $endId = $extDB->selectField( $table, 'MAX(blob_id)', '', __METHOD__ );
 345
 346                         // Build a bitmap of actual blob rows
 347                         while ( true ) {
 348                                 $res = $extDB->select( $table,
 349                                         [ 'blob_id' ],
 350                                         [ 'blob_id > ' . $extDB->addQuotes( $startId ) ],
 351                                         __METHOD__,
 352                                         [ 'LIMIT' => $this->batchSize, 'ORDER BY' => 'blob_id' ]
 353                                 );
 354
 355                                 if ( !$res->numRows() ) {
 356                                         break;
 357                                 }
 358
 359                                 foreach ( $res as $row ) {
 360                                         gmp_setbit( $actualBlobs, $row->blob_id );
 361                                 }
 362                                 $startId = $row->blob_id;
 363
 364                                 ++$batchesDone;
 365                                 if ( $batchesDone >= $this->reportingInterval ) {
 366                                         $batchesDone = 0;
 367                                         echo "$startId / $endId\n";
 368                                 }
 369                         }
 370
 371                         // Find actual blobs that weren't tracked by the previous passes
 372                         // This is a set-theoretic difference A \ B, or in bitwise terms, A & ~B
 373                         $orphans = gmp_and( $actualBlobs, gmp_com( $this->trackedBlobs[$cluster] ) );
 374
 375                         // Traverse the orphan list
 376                         $insertBatch = [];
 377                         $id = 0;
 378                         $numOrphans = 0;
 379                         while ( true ) {
 380                                 $id = gmp_scan1( $orphans, $id );
 381                                 if ( $id == -1 ) {
 382                                         break;
 383                                 }
 384                                 $insertBatch[] = [
 385                                         'bo_cluster' => $cluster,
 386                                         'bo_blob_id' => $id
 387                                 ];
 388                                 if ( count( $insertBatch ) > $this->batchSize ) {
 389                                         $dbw->insert( 'blob_orphans', $insertBatch, __METHOD__ );
 390                                         $insertBatch = [];
 391                                 }
 392
 393                                 ++$id;
 394                                 ++$numOrphans;
 395                         }
 396                         if ( $insertBatch ) {
 397                                 $dbw->insert( 'blob_orphans', $insertBatch, __METHOD__ );
 398                         }
 399                         echo "Found $numOrphans orphan(s) in $cluster\n";
 400                 }
 401         }
 402 }