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