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