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