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