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