includes/jobqueue/JobQueueDB.php

   1 <?php
   2 /**
   3  * Database-backed job queue code.
   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  */
  22 use Wikimedia\Rdbms\IDatabase;
  23 use Wikimedia\Rdbms\DBConnRef;
  24 use Wikimedia\Rdbms\DBConnectionError;
  25 use Wikimedia\Rdbms\DBError;
  26 use MediaWiki\MediaWikiServices;
  27 use Wikimedia\ScopedCallback;
  28
  29 /**
  30  * Class to handle job queues stored in the DB
  31  *
  32  * @ingroup JobQueue
  33  * @since 1.21
  34  */
  35 class JobQueueDB extends JobQueue {
  36         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  37         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  38         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  39         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  40
  41         /** @var WANObjectCache */
  42         protected $cache;
  43
  44         /** @var bool|string Name of an external DB cluster. False if not set */
  45         protected $cluster = false;
  46
  47         /**
  48          * Additional parameters include:
  49          *   - cluster : The name of an external cluster registered via LBFactory.
  50          *               If not specified, the primary DB cluster for the wiki will be used.
  51          *               This can be overridden with a custom cluster so that DB handles will
  52          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  53          * @param array $params
  54          */
  55         protected function __construct( array $params ) {
  56                 parent::__construct( $params );
  57
  58                 $this->cluster = $params['cluster'] ?? false;
  59                 $this->cache = ObjectCache::getMainWANInstance();
  60         }
  61
  62         protected function supportedOrders() {
  63                 return [ 'random', 'timestamp', 'fifo' ];
  64         }
  65
  66         protected function optimalOrder() {
  67                 return 'random';
  68         }
  69
  70         /**
  71          * @see JobQueue::doIsEmpty()
  72          * @return bool
  73          */
  74         protected function doIsEmpty() {
  75                 $dbr = $this->getReplicaDB();
  76                 try {
  77                         $found = $dbr->selectField( // unclaimed job
  78                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  79                         );
  80                 } catch ( DBError $e ) {
  81                         $this->throwDBException( $e );
  82                 }
  83
  84                 return !$found;
  85         }
  86
  87         /**
  88          * @see JobQueue::doGetSize()
  89          * @return int
  90          */
  91         protected function doGetSize() {
  92                 $key = $this->getCacheKey( 'size' );
  93
  94                 $size = $this->cache->get( $key );
  95                 if ( is_int( $size ) ) {
  96                         return $size;
  97                 }
  98
  99                 try {
 100                         $dbr = $this->getReplicaDB();
 101                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 102                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
 103                                 __METHOD__
 104                         );
 105                 } catch ( DBError $e ) {
 106                         $this->throwDBException( $e );
 107                 }
 108                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 109
 110                 return $size;
 111         }
 112
 113         /**
 114          * @see JobQueue::doGetAcquiredCount()
 115          * @return int
 116          */
 117         protected function doGetAcquiredCount() {
 118                 if ( $this->claimTTL <= 0 ) {
 119                         return 0; // no acknowledgements
 120                 }
 121
 122                 $key = $this->getCacheKey( 'acquiredcount' );
 123
 124                 $count = $this->cache->get( $key );
 125                 if ( is_int( $count ) ) {
 126                         return $count;
 127                 }
 128
 129                 $dbr = $this->getReplicaDB();
 130                 try {
 131                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 132                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 133                                 __METHOD__
 134                         );
 135                 } catch ( DBError $e ) {
 136                         $this->throwDBException( $e );
 137                 }
 138                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 139
 140                 return $count;
 141         }
 142
 143         /**
 144          * @see JobQueue::doGetAbandonedCount()
 145          * @return int
 146          * @throws MWException
 147          */
 148         protected function doGetAbandonedCount() {
 149                 if ( $this->claimTTL <= 0 ) {
 150                         return 0; // no acknowledgements
 151                 }
 152
 153                 $key = $this->getCacheKey( 'abandonedcount' );
 154
 155                 $count = $this->cache->get( $key );
 156                 if ( is_int( $count ) ) {
 157                         return $count;
 158                 }
 159
 160                 $dbr = $this->getReplicaDB();
 161                 try {
 162                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 163                                 [
 164                                         'job_cmd' => $this->type,
 165                                         "job_token != {$dbr->addQuotes( '' )}",
 166                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 167                                 ],
 168                                 __METHOD__
 169                         );
 170                 } catch ( DBError $e ) {
 171                         $this->throwDBException( $e );
 172                 }
 173
 174                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 175
 176                 return $count;
 177         }
 178
 179         /**
 180          * @see JobQueue::doBatchPush()
 181          * @param IJobSpecification[] $jobs
 182          * @param int $flags
 183          * @throws DBError|Exception
 184          * @return void
 185          */
 186         protected function doBatchPush( array $jobs, $flags ) {
 187                 $dbw = $this->getMasterDB();
 188                 // In general, there will be two cases here:
 189                 // a) sqlite; DB connection is probably a regular round-aware handle.
 190                 // If the connection is busy with a transaction, then defer the job writes
 191                 // until right before the main round commit step. Any errors that bubble
 192                 // up will rollback the main commit round.
 193                 // b) mysql/postgres; DB connection is generally a separate CONN_TRX_AUTOCOMMIT handle.
 194                 // No transaction is active nor will be started by writes, so enqueue the jobs
 195                 // now so that any errors will show up immediately as the interface expects. Any
 196                 // errors that bubble up will rollback the main commit round.
 197                 $fname = __METHOD__;
 198                 $dbw->onTransactionPreCommitOrIdle(
 199                         function ( IDatabase $dbw ) use ( $jobs, $flags, $fname ) {
 200                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $fname );
 201                         },
 202                         $fname
 203                 );
 204         }
 205
 206         /**
 207          * This function should *not* be called outside of JobQueueDB
 208          *
 209          * @suppress SecurityCheck-SQLInjection Bug in phan-taint-check handling bulk inserts
 210          * @param IDatabase $dbw
 211          * @param IJobSpecification[] $jobs
 212          * @param int $flags
 213          * @param string $method
 214          * @throws DBError
 215          * @return void
 216          */
 217         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 218                 if ( $jobs === [] ) {
 219                         return;
 220                 }
 221
 222                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 223                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 224                 foreach ( $jobs as $job ) {
 225                         $row = $this->insertFields( $job, $dbw );
 226                         if ( $job->ignoreDuplicates() ) {
 227                                 $rowSet[$row['job_sha1']] = $row;
 228                         } else {
 229                                 $rowList[] = $row;
 230                         }
 231                 }
 232
 233                 if ( $flags & self::QOS_ATOMIC ) {
 234                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 235                 }
 236                 try {
 237                         // Strip out any duplicate jobs that are already in the queue...
 238                         if ( count( $rowSet ) ) {
 239                                 $res = $dbw->select( 'job', 'job_sha1',
 240                                         [
 241                                                 // No job_type condition since it's part of the job_sha1 hash
 242                                                 'job_sha1' => array_keys( $rowSet ),
 243                                                 'job_token' => '' // unclaimed
 244                                         ],
 245                                         $method
 246                                 );
 247                                 foreach ( $res as $row ) {
 248                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 249                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 250                                 }
 251                         }
 252                         // Build the full list of job rows to insert
 253                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 254                         // Insert the job rows in chunks to avoid replica DB lag...
 255                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 256                                 $dbw->insert( 'job', $rowBatch, $method );
 257                         }
 258                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 259                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 260                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 261                         );
 262                 } catch ( DBError $e ) {
 263                         $this->throwDBException( $e );
 264                 }
 265                 if ( $flags & self::QOS_ATOMIC ) {
 266                         $dbw->endAtomic( $method );
 267                 }
 268         }
 269
 270         /**
 271          * @see JobQueue::doPop()
 272          * @return Job|bool
 273          */
 274         protected function doPop() {
 275                 $dbw = $this->getMasterDB();
 276                 try {
 277                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 278                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 279                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 280                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 281                         } );
 282
 283                         $uuid = wfRandomString( 32 ); // pop attempt
 284                         $job = false; // job popped off
 285                         do { // retry when our row is invalid or deleted as a duplicate
 286                                 // Try to reserve a row in the DB...
 287                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 288                                         $row = $this->claimOldest( $uuid );
 289                                 } else { // random first
 290                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 291                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 292                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 293                                 }
 294                                 // Check if we found a row to reserve...
 295                                 if ( !$row ) {
 296                                         break; // nothing to do
 297                                 }
 298                                 JobQueue::incrStats( 'pops', $this->type );
 299                                 // Get the job object from the row...
 300                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 301                                 $job = Job::factory( $row->job_cmd, $title,
 302                                         self::extractBlob( $row->job_params ) );
 303                                 $job->metadata['id'] = $row->job_id;
 304                                 $job->metadata['timestamp'] = $row->job_timestamp;
 305                                 break; // done
 306                         } while ( true );
 307
 308                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 309                                 // Handled jobs that need to be recycled/deleted;
 310                                 // any recycled jobs will be picked up next attempt
 311                                 $this->recycleAndDeleteStaleJobs();
 312                         }
 313                 } catch ( DBError $e ) {
 314                         $this->throwDBException( $e );
 315                 }
 316
 317                 return $job;
 318         }
 319
 320         /**
 321          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 322          *
 323          * @param string $uuid 32 char hex string
 324          * @param int $rand Random unsigned integer (31 bits)
 325          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 326          * @return stdClass|bool Row|false
 327          */
 328         protected function claimRandom( $uuid, $rand, $gte ) {
 329                 $dbw = $this->getMasterDB();
 330                 // Check cache to see if the queue has <= OFFSET items
 331                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 332
 333                 $row = false; // the row acquired
 334                 $invertedDirection = false; // whether one job_random direction was already scanned
 335                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 336                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 337                 // not replication safe. Due to https://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 338                 // be used here with MySQL.
 339                 do {
 340                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 341                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 342                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 343                                 $ineq = $gte ? '>=' : '<=';
 344                                 $dir = $gte ? 'ASC' : 'DESC';
 345                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 346                                         [
 347                                                 'job_cmd' => $this->type,
 348                                                 'job_token' => '', // unclaimed
 349                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 350                                         __METHOD__,
 351                                         [ 'ORDER BY' => "job_random {$dir}" ]
 352                                 );
 353                                 if ( !$row && !$invertedDirection ) {
 354                                         $gte = !$gte;
 355                                         $invertedDirection = true;
 356                                         continue; // try the other direction
 357                                 }
 358                         } else { // table *may* have >= MAX_OFFSET rows
 359                                 // T44614: "ORDER BY job_random" with a job_random inequality causes high CPU
 360                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 361                                 // instead of job_random for reducing excess claim retries.
 362                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 363                                         [
 364                                                 'job_cmd' => $this->type,
 365                                                 'job_token' => '', // unclaimed
 366                                         ],
 367                                         __METHOD__,
 368                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 369                                 );
 370                                 if ( !$row ) {
 371                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 372                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 373                                         continue; // use job_random
 374                                 }
 375                         }
 376
 377                         if ( $row ) { // claim the job
 378                                 $dbw->update( 'job', // update by PK
 379                                         [
 380                                                 'job_token' => $uuid,
 381                                                 'job_token_timestamp' => $dbw->timestamp(),
 382                                                 'job_attempts = job_attempts+1' ],
 383                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 384                                         __METHOD__
 385                                 );
 386                                 // This might get raced out by another runner when claiming the previously
 387                                 // selected row. The use of job_random should minimize this problem, however.
 388                                 if ( !$dbw->affectedRows() ) {
 389                                         $row = false; // raced out
 390                                 }
 391                         } else {
 392                                 break; // nothing to do
 393                         }
 394                 } while ( !$row );
 395
 396                 return $row;
 397         }
 398
 399         /**
 400          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 401          *
 402          * @param string $uuid 32 char hex string
 403          * @return stdClass|bool Row|false
 404          */
 405         protected function claimOldest( $uuid ) {
 406                 $dbw = $this->getMasterDB();
 407
 408                 $row = false; // the row acquired
 409                 do {
 410                         if ( $dbw->getType() === 'mysql' ) {
 411                                 // Per https://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 412                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 413                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 414                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 415                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 416                                         "SET " .
 417                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 418                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 419                                                 "job_attempts = job_attempts+1 " .
 420                                         "WHERE ( " .
 421                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 422                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 423                                         ") ORDER BY job_id ASC LIMIT 1",
 424                                         __METHOD__
 425                                 );
 426                         } else {
 427                                 // Use a subquery to find the job, within an UPDATE to claim it.
 428                                 // This uses as much of the DB wrapper functions as possible.
 429                                 $dbw->update( 'job',
 430                                         [
 431                                                 'job_token' => $uuid,
 432                                                 'job_token_timestamp' => $dbw->timestamp(),
 433                                                 'job_attempts = job_attempts+1' ],
 434                                         [ 'job_id = (' .
 435                                                 $dbw->selectSQLText( 'job', 'job_id',
 436                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 437                                                         __METHOD__,
 438                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 439                                                 ')'
 440                                         ],
 441                                         __METHOD__
 442                                 );
 443                         }
 444                         // Fetch any row that we just reserved...
 445                         if ( $dbw->affectedRows() ) {
 446                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 447                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 448                                 );
 449                                 if ( !$row ) { // raced out by duplicate job removal
 450                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 451                                 }
 452                         } else {
 453                                 break; // nothing to do
 454                         }
 455                 } while ( !$row );
 456
 457                 return $row;
 458         }
 459
 460         /**
 461          * @see JobQueue::doAck()
 462          * @param Job $job
 463          * @throws MWException
 464          */
 465         protected function doAck( Job $job ) {
 466                 if ( !isset( $job->metadata['id'] ) ) {
 467                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 468                 }
 469
 470                 $dbw = $this->getMasterDB();
 471                 try {
 472                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 473                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 474                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 475                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 476                         } );
 477
 478                         // Delete a row with a single DELETE without holding row locks over RTTs...
 479                         $dbw->delete( 'job',
 480                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 481
 482                         JobQueue::incrStats( 'acks', $this->type );
 483                 } catch ( DBError $e ) {
 484                         $this->throwDBException( $e );
 485                 }
 486         }
 487
 488         /**
 489          * @see JobQueue::doDeduplicateRootJob()
 490          * @param IJobSpecification $job
 491          * @throws MWException
 492          * @return bool
 493          */
 494         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 495                 $params = $job->getParams();
 496                 if ( !isset( $params['rootJobSignature'] ) ) {
 497                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 498                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 499                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 500                 }
 501                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 502                 // Callers should call JobQueueGroup::push() before this method so that if the insert
 503                 // fails, the de-duplication registration will be aborted. Since the insert is
 504                 // deferred till "transaction idle", do the same here, so that the ordering is
 505                 // maintained. Having only the de-duplication registration succeed would cause
 506                 // jobs to become no-ops without any actual jobs that made them redundant.
 507                 $dbw = $this->getMasterDB();
 508                 $cache = $this->dupCache;
 509                 $dbw->onTransactionCommitOrIdle(
 510                         function () use ( $cache, $params, $key ) {
 511                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 512                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 513                                         return true; // a newer version of this root job was enqueued
 514                                 }
 515
 516                                 // Update the timestamp of the last root job started at the location...
 517                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 518                         },
 519                         __METHOD__
 520                 );
 521
 522                 return true;
 523         }
 524
 525         /**
 526          * @see JobQueue::doDelete()
 527          * @return bool
 528          */
 529         protected function doDelete() {
 530                 $dbw = $this->getMasterDB();
 531                 try {
 532                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 533                 } catch ( DBError $e ) {
 534                         $this->throwDBException( $e );
 535                 }
 536
 537                 return true;
 538         }
 539
 540         /**
 541          * @see JobQueue::doWaitForBackups()
 542          * @return void
 543          */
 544         protected function doWaitForBackups() {
 545                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 546                 $lbFactory->waitForReplication(
 547                         [ 'domain' => $this->domain, 'cluster' => $this->cluster ] );
 548         }
 549
 550         /**
 551          * @return void
 552          */
 553         protected function doFlushCaches() {
 554                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 555                         $this->cache->delete( $this->getCacheKey( $type ) );
 556                 }
 557         }
 558
 559         /**
 560          * @see JobQueue::getAllQueuedJobs()
 561          * @return Iterator
 562          */
 563         public function getAllQueuedJobs() {
 564                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 565         }
 566
 567         /**
 568          * @see JobQueue::getAllAcquiredJobs()
 569          * @return Iterator
 570          */
 571         public function getAllAcquiredJobs() {
 572                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 573         }
 574
 575         /**
 576          * @param array $conds Query conditions
 577          * @return Iterator
 578          */
 579         protected function getJobIterator( array $conds ) {
 580                 $dbr = $this->getReplicaDB();
 581                 try {
 582                         return new MappedIterator(
 583                                 $dbr->select( 'job', self::selectFields(), $conds ),
 584                                 function ( $row ) {
 585                                         $job = Job::factory(
 586                                                 $row->job_cmd,
 587                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 588                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 589                                         );
 590                                         $job->metadata['id'] = $row->job_id;
 591                                         $job->metadata['timestamp'] = $row->job_timestamp;
 592
 593                                         return $job;
 594                                 }
 595                         );
 596                 } catch ( DBError $e ) {
 597                         $this->throwDBException( $e );
 598                 }
 599         }
 600
 601         public function getCoalesceLocationInternal() {
 602                 return $this->cluster
 603                         ? "DBCluster:{$this->cluster}:{$this->domain}"
 604                         : "LBFactory:{$this->domain}";
 605         }
 606
 607         protected function doGetSiblingQueuesWithJobs( array $types ) {
 608                 $dbr = $this->getReplicaDB();
 609                 // @note: this does not check whether the jobs are claimed or not.
 610                 // This is useful so JobQueueGroup::pop() also sees queues that only
 611                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 612                 // failed jobs so that they can be popped again for that edge case.
 613                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 614                         [ 'job_cmd' => $types ], __METHOD__ );
 615
 616                 $types = [];
 617                 foreach ( $res as $row ) {
 618                         $types[] = $row->job_cmd;
 619                 }
 620
 621                 return $types;
 622         }
 623
 624         protected function doGetSiblingQueueSizes( array $types ) {
 625                 $dbr = $this->getReplicaDB();
 626                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 627                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 628
 629                 $sizes = [];
 630                 foreach ( $res as $row ) {
 631                         $sizes[$row->job_cmd] = (int)$row->count;
 632                 }
 633
 634                 return $sizes;
 635         }
 636
 637         /**
 638          * Recycle or destroy any jobs that have been claimed for too long
 639          *
 640          * @return int Number of jobs recycled/deleted
 641          */
 642         public function recycleAndDeleteStaleJobs() {
 643                 $now = time();
 644                 $count = 0; // affected rows
 645                 $dbw = $this->getMasterDB();
 646
 647                 try {
 648                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 649                                 return $count; // already in progress
 650                         }
 651
 652                         // Remove claims on jobs acquired for too long if enabled...
 653                         if ( $this->claimTTL > 0 ) {
 654                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 655                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 656                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 657                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 658                                 $res = $dbw->select( 'job', 'job_id',
 659                                         [
 660                                                 'job_cmd' => $this->type,
 661                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 662                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 663                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 664                                         __METHOD__
 665                                 );
 666                                 $ids = array_map(
 667                                         function ( $o ) {
 668                                                 return $o->job_id;
 669                                         }, iterator_to_array( $res )
 670                                 );
 671                                 if ( count( $ids ) ) {
 672                                         // Reset job_token for these jobs so that other runners will pick them up.
 673                                         // Set the timestamp to the current time, as it is useful to now that the job
 674                                         // was already tried before (the timestamp becomes the "released" time).
 675                                         $dbw->update( 'job',
 676                                                 [
 677                                                         'job_token' => '',
 678                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 679                                                 [
 680                                                         'job_id' => $ids ],
 681                                                 __METHOD__
 682                                         );
 683                                         $affected = $dbw->affectedRows();
 684                                         $count += $affected;
 685                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 686                                         $this->aggr->notifyQueueNonEmpty( $this->domain, $this->type );
 687                                 }
 688                         }
 689
 690                         // Just destroy any stale jobs...
 691                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 692                         $conds = [
 693                                 'job_cmd' => $this->type,
 694                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 695                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 696                         ];
 697                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 698                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 699                         }
 700                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 701                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 702                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 703                         $ids = array_map(
 704                                 function ( $o ) {
 705                                         return $o->job_id;
 706                                 }, iterator_to_array( $res )
 707                         );
 708                         if ( count( $ids ) ) {
 709                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 710                                 $affected = $dbw->affectedRows();
 711                                 $count += $affected;
 712                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 713                         }
 714
 715                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 716                 } catch ( DBError $e ) {
 717                         $this->throwDBException( $e );
 718                 }
 719
 720                 return $count;
 721         }
 722
 723         /**
 724          * @param IJobSpecification $job
 725          * @param IDatabase $db
 726          * @return array
 727          */
 728         protected function insertFields( IJobSpecification $job, IDatabase $db ) {
 729                 return [
 730                         // Fields that describe the nature of the job
 731                         'job_cmd' => $job->getType(),
 732                         'job_namespace' => $job->getTitle()->getNamespace(),
 733                         'job_title' => $job->getTitle()->getDBkey(),
 734                         'job_params' => self::makeBlob( $job->getParams() ),
 735                         // Additional job metadata
 736                         'job_timestamp' => $db->timestamp(),
 737                         'job_sha1' => Wikimedia\base_convert(
 738                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 739                                 16, 36, 31
 740                         ),
 741                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 742                 ];
 743         }
 744
 745         /**
 746          * @throws JobQueueConnectionError
 747          * @return DBConnRef
 748          */
 749         protected function getReplicaDB() {
 750                 try {
 751                         return $this->getDB( DB_REPLICA );
 752                 } catch ( DBConnectionError $e ) {
 753                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 754                 }
 755         }
 756
 757         /**
 758          * @throws JobQueueConnectionError
 759          * @return DBConnRef
 760          */
 761         protected function getMasterDB() {
 762                 try {
 763                         return $this->getDB( DB_MASTER );
 764                 } catch ( DBConnectionError $e ) {
 765                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 766                 }
 767         }
 768
 769         /**
 770          * @param int $index (DB_REPLICA/DB_MASTER)
 771          * @return DBConnRef
 772          */
 773         protected function getDB( $index ) {
 774                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 775                 $lb = ( $this->cluster !== false )
 776                         ? $lbFactory->getExternalLB( $this->cluster )
 777                         : $lbFactory->getMainLB( $this->domain );
 778
 779                 return ( $lb->getServerType( $lb->getWriterIndex() ) !== 'sqlite' )
 780                         // Keep a separate connection to avoid contention and deadlocks;
 781                         // However, SQLite has the opposite behavior due to DB-level locking.
 782                         ? $lb->getConnectionRef( $index, [], $this->domain, $lb::CONN_TRX_AUTOCOMMIT )
 783                         // Jobs insertion will be defered until the PRESEND stage to reduce contention.
 784                         : $lb->getConnectionRef( $index, [], $this->domain );
 785         }
 786
 787         /**
 788          * @param string $property
 789          * @return string
 790          */
 791         private function getCacheKey( $property ) {
 792                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 793
 794                 return $this->cache->makeGlobalKey(
 795                         'jobqueue',
 796                         $this->domain,
 797                         $cluster,
 798                         $this->type,
 799                         $property
 800                 );
 801         }
 802
 803         /**
 804          * @param array|bool $params
 805          * @return string
 806          */
 807         protected static function makeBlob( $params ) {
 808                 if ( $params !== false ) {
 809                         return serialize( $params );
 810                 } else {
 811                         return '';
 812                 }
 813         }
 814
 815         /**
 816          * @param string $blob
 817          * @return bool|mixed
 818          */
 819         protected static function extractBlob( $blob ) {
 820                 if ( (string)$blob !== '' ) {
 821                         return unserialize( $blob );
 822                 } else {
 823                         return false;
 824                 }
 825         }
 826
 827         /**
 828          * @param DBError $e
 829          * @throws JobQueueError
 830          */
 831         protected function throwDBException( DBError $e ) {
 832                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 833         }
 834
 835         /**
 836          * Return the list of job fields that should be selected.
 837          * @since 1.23
 838          * @return array
 839          */
 840         public static function selectFields() {
 841                 return [
 842                         'job_id',
 843                         'job_cmd',
 844                         'job_namespace',
 845                         'job_title',
 846                         'job_timestamp',
 847                         'job_params',
 848                         'job_random',
 849                         'job_attempts',
 850                         'job_token',
 851                         'job_token_timestamp',
 852                         'job_sha1',
 853                 ];
 854         }
 855 }