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 MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * Class to handle job queues stored in the DB
  27  *
  28  * @ingroup JobQueue
  29  * @since 1.21
  30  */
  31 class JobQueueDB extends JobQueue {
  32         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  33         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  34         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  35         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  36
  37         /** @var WANObjectCache */
  38         protected $cache;
  39
  40         /** @var bool|string Name of an external DB cluster. False if not set */
  41         protected $cluster = false;
  42
  43         /**
  44          * Additional parameters include:
  45          *   - cluster : The name of an external cluster registered via LBFactory.
  46          *               If not specified, the primary DB cluster for the wiki will be used.
  47          *               This can be overridden with a custom cluster so that DB handles will
  48          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  49          * @param array $params
  50          */
  51         protected function __construct( array $params ) {
  52                 parent::__construct( $params );
  53
  54                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  55                 $this->cache = ObjectCache::getMainWANInstance();
  56         }
  57
  58         protected function supportedOrders() {
  59                 return [ 'random', 'timestamp', 'fifo' ];
  60         }
  61
  62         protected function optimalOrder() {
  63                 return 'random';
  64         }
  65
  66         /**
  67          * @see JobQueue::doIsEmpty()
  68          * @return bool
  69          */
  70         protected function doIsEmpty() {
  71                 $dbr = $this->getSlaveDB();
  72                 try {
  73                         $found = $dbr->selectField( // unclaimed job
  74                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  75                         );
  76                 } catch ( DBError $e ) {
  77                         $this->throwDBException( $e );
  78                 }
  79
  80                 return !$found;
  81         }
  82
  83         /**
  84          * @see JobQueue::doGetSize()
  85          * @return int
  86          */
  87         protected function doGetSize() {
  88                 $key = $this->getCacheKey( 'size' );
  89
  90                 $size = $this->cache->get( $key );
  91                 if ( is_int( $size ) ) {
  92                         return $size;
  93                 }
  94
  95                 try {
  96                         $dbr = $this->getSlaveDB();
  97                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
  98                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
  99                                 __METHOD__
 100                         );
 101                 } catch ( DBError $e ) {
 102                         $this->throwDBException( $e );
 103                 }
 104                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 105
 106                 return $size;
 107         }
 108
 109         /**
 110          * @see JobQueue::doGetAcquiredCount()
 111          * @return int
 112          */
 113         protected function doGetAcquiredCount() {
 114                 if ( $this->claimTTL <= 0 ) {
 115                         return 0; // no acknowledgements
 116                 }
 117
 118                 $key = $this->getCacheKey( 'acquiredcount' );
 119
 120                 $count = $this->cache->get( $key );
 121                 if ( is_int( $count ) ) {
 122                         return $count;
 123                 }
 124
 125                 $dbr = $this->getSlaveDB();
 126                 try {
 127                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 128                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 129                                 __METHOD__
 130                         );
 131                 } catch ( DBError $e ) {
 132                         $this->throwDBException( $e );
 133                 }
 134                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 135
 136                 return $count;
 137         }
 138
 139         /**
 140          * @see JobQueue::doGetAbandonedCount()
 141          * @return int
 142          * @throws MWException
 143          */
 144         protected function doGetAbandonedCount() {
 145                 if ( $this->claimTTL <= 0 ) {
 146                         return 0; // no acknowledgements
 147                 }
 148
 149                 $key = $this->getCacheKey( 'abandonedcount' );
 150
 151                 $count = $this->cache->get( $key );
 152                 if ( is_int( $count ) ) {
 153                         return $count;
 154                 }
 155
 156                 $dbr = $this->getSlaveDB();
 157                 try {
 158                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 159                                 [
 160                                         'job_cmd' => $this->type,
 161                                         "job_token != {$dbr->addQuotes( '' )}",
 162                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 163                                 ],
 164                                 __METHOD__
 165                         );
 166                 } catch ( DBError $e ) {
 167                         $this->throwDBException( $e );
 168                 }
 169
 170                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 171
 172                 return $count;
 173         }
 174
 175         /**
 176          * @see JobQueue::doBatchPush()
 177          * @param IJobSpecification[] $jobs
 178          * @param int $flags
 179          * @throws DBError|Exception
 180          * @return void
 181          */
 182         protected function doBatchPush( array $jobs, $flags ) {
 183                 $dbw = $this->getMasterDB();
 184
 185                 $method = __METHOD__;
 186                 $dbw->onTransactionIdle(
 187                         function () use ( $dbw, $jobs, $flags, $method ) {
 188                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 189                         },
 190                         __METHOD__
 191                 );
 192         }
 193
 194         /**
 195          * This function should *not* be called outside of JobQueueDB
 196          *
 197          * @param IDatabase $dbw
 198          * @param IJobSpecification[] $jobs
 199          * @param int $flags
 200          * @param string $method
 201          * @throws DBError
 202          * @return void
 203          */
 204         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 205                 if ( !count( $jobs ) ) {
 206                         return;
 207                 }
 208
 209                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 210                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 211                 foreach ( $jobs as $job ) {
 212                         $row = $this->insertFields( $job );
 213                         if ( $job->ignoreDuplicates() ) {
 214                                 $rowSet[$row['job_sha1']] = $row;
 215                         } else {
 216                                 $rowList[] = $row;
 217                         }
 218                 }
 219
 220                 if ( $flags & self::QOS_ATOMIC ) {
 221                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 222                 }
 223                 try {
 224                         // Strip out any duplicate jobs that are already in the queue...
 225                         if ( count( $rowSet ) ) {
 226                                 $res = $dbw->select( 'job', 'job_sha1',
 227                                         [
 228                                                 // No job_type condition since it's part of the job_sha1 hash
 229                                                 'job_sha1' => array_keys( $rowSet ),
 230                                                 'job_token' => '' // unclaimed
 231                                         ],
 232                                         $method
 233                                 );
 234                                 foreach ( $res as $row ) {
 235                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 236                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 237                                 }
 238                         }
 239                         // Build the full list of job rows to insert
 240                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 241                         // Insert the job rows in chunks to avoid replica DB lag...
 242                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 243                                 $dbw->insert( 'job', $rowBatch, $method );
 244                         }
 245                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 246                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 247                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 248                         );
 249                 } catch ( DBError $e ) {
 250                         $this->throwDBException( $e );
 251                 }
 252                 if ( $flags & self::QOS_ATOMIC ) {
 253                         $dbw->endAtomic( $method );
 254                 }
 255
 256                 return;
 257         }
 258
 259         /**
 260          * @see JobQueue::doPop()
 261          * @return Job|bool
 262          */
 263         protected function doPop() {
 264                 $dbw = $this->getMasterDB();
 265                 try {
 266                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 267                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 268                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 269                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 270                         } );
 271
 272                         $uuid = wfRandomString( 32 ); // pop attempt
 273                         $job = false; // job popped off
 274                         do { // retry when our row is invalid or deleted as a duplicate
 275                                 // Try to reserve a row in the DB...
 276                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 277                                         $row = $this->claimOldest( $uuid );
 278                                 } else { // random first
 279                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 280                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 281                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 282                                 }
 283                                 // Check if we found a row to reserve...
 284                                 if ( !$row ) {
 285                                         break; // nothing to do
 286                                 }
 287                                 JobQueue::incrStats( 'pops', $this->type );
 288                                 // Get the job object from the row...
 289                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 290                                 $job = Job::factory( $row->job_cmd, $title,
 291                                         self::extractBlob( $row->job_params ), $row->job_id );
 292                                 $job->metadata['id'] = $row->job_id;
 293                                 $job->metadata['timestamp'] = $row->job_timestamp;
 294                                 break; // done
 295                         } while ( true );
 296
 297                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 298                                 // Handled jobs that need to be recycled/deleted;
 299                                 // any recycled jobs will be picked up next attempt
 300                                 $this->recycleAndDeleteStaleJobs();
 301                         }
 302                 } catch ( DBError $e ) {
 303                         $this->throwDBException( $e );
 304                 }
 305
 306                 return $job;
 307         }
 308
 309         /**
 310          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 311          *
 312          * @param string $uuid 32 char hex string
 313          * @param int $rand Random unsigned integer (31 bits)
 314          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 315          * @return stdClass|bool Row|false
 316          */
 317         protected function claimRandom( $uuid, $rand, $gte ) {
 318                 $dbw = $this->getMasterDB();
 319                 // Check cache to see if the queue has <= OFFSET items
 320                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 321
 322                 $row = false; // the row acquired
 323                 $invertedDirection = false; // whether one job_random direction was already scanned
 324                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 325                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 326                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 327                 // be used here with MySQL.
 328                 do {
 329                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 330                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 331                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 332                                 $ineq = $gte ? '>=' : '<=';
 333                                 $dir = $gte ? 'ASC' : 'DESC';
 334                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 335                                         [
 336                                                 'job_cmd' => $this->type,
 337                                                 'job_token' => '', // unclaimed
 338                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 339                                         __METHOD__,
 340                                         [ 'ORDER BY' => "job_random {$dir}" ]
 341                                 );
 342                                 if ( !$row && !$invertedDirection ) {
 343                                         $gte = !$gte;
 344                                         $invertedDirection = true;
 345                                         continue; // try the other direction
 346                                 }
 347                         } else { // table *may* have >= MAX_OFFSET rows
 348                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 349                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 350                                 // instead of job_random for reducing excess claim retries.
 351                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 352                                         [
 353                                                 'job_cmd' => $this->type,
 354                                                 'job_token' => '', // unclaimed
 355                                         ],
 356                                         __METHOD__,
 357                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 358                                 );
 359                                 if ( !$row ) {
 360                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 361                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 362                                         continue; // use job_random
 363                                 }
 364                         }
 365
 366                         if ( $row ) { // claim the job
 367                                 $dbw->update( 'job', // update by PK
 368                                         [
 369                                                 'job_token' => $uuid,
 370                                                 'job_token_timestamp' => $dbw->timestamp(),
 371                                                 'job_attempts = job_attempts+1' ],
 372                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 373                                         __METHOD__
 374                                 );
 375                                 // This might get raced out by another runner when claiming the previously
 376                                 // selected row. The use of job_random should minimize this problem, however.
 377                                 if ( !$dbw->affectedRows() ) {
 378                                         $row = false; // raced out
 379                                 }
 380                         } else {
 381                                 break; // nothing to do
 382                         }
 383                 } while ( !$row );
 384
 385                 return $row;
 386         }
 387
 388         /**
 389          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 390          *
 391          * @param string $uuid 32 char hex string
 392          * @return stdClass|bool Row|false
 393          */
 394         protected function claimOldest( $uuid ) {
 395                 $dbw = $this->getMasterDB();
 396
 397                 $row = false; // the row acquired
 398                 do {
 399                         if ( $dbw->getType() === 'mysql' ) {
 400                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 401                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 402                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 403                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 404                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 405                                         "SET " .
 406                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 407                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 408                                                 "job_attempts = job_attempts+1 " .
 409                                         "WHERE ( " .
 410                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 411                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 412                                         ") ORDER BY job_id ASC LIMIT 1",
 413                                         __METHOD__
 414                                 );
 415                         } else {
 416                                 // Use a subquery to find the job, within an UPDATE to claim it.
 417                                 // This uses as much of the DB wrapper functions as possible.
 418                                 $dbw->update( 'job',
 419                                         [
 420                                                 'job_token' => $uuid,
 421                                                 'job_token_timestamp' => $dbw->timestamp(),
 422                                                 'job_attempts = job_attempts+1' ],
 423                                         [ 'job_id = (' .
 424                                                 $dbw->selectSQLText( 'job', 'job_id',
 425                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 426                                                         __METHOD__,
 427                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 428                                                 ')'
 429                                         ],
 430                                         __METHOD__
 431                                 );
 432                         }
 433                         // Fetch any row that we just reserved...
 434                         if ( $dbw->affectedRows() ) {
 435                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 436                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 437                                 );
 438                                 if ( !$row ) { // raced out by duplicate job removal
 439                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 440                                 }
 441                         } else {
 442                                 break; // nothing to do
 443                         }
 444                 } while ( !$row );
 445
 446                 return $row;
 447         }
 448
 449         /**
 450          * @see JobQueue::doAck()
 451          * @param Job $job
 452          * @throws MWException
 453          */
 454         protected function doAck( Job $job ) {
 455                 if ( !isset( $job->metadata['id'] ) ) {
 456                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 457                 }
 458
 459                 $dbw = $this->getMasterDB();
 460                 try {
 461                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 462                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 463                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 464                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 465                         } );
 466
 467                         // Delete a row with a single DELETE without holding row locks over RTTs...
 468                         $dbw->delete( 'job',
 469                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 470
 471                         JobQueue::incrStats( 'acks', $this->type );
 472                 } catch ( DBError $e ) {
 473                         $this->throwDBException( $e );
 474                 }
 475         }
 476
 477         /**
 478          * @see JobQueue::doDeduplicateRootJob()
 479          * @param IJobSpecification $job
 480          * @throws MWException
 481          * @return bool
 482          */
 483         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 484                 $params = $job->getParams();
 485                 if ( !isset( $params['rootJobSignature'] ) ) {
 486                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 487                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 488                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 489                 }
 490                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 491                 // Callers should call batchInsert() and then this function so that if the insert
 492                 // fails, the de-duplication registration will be aborted. Since the insert is
 493                 // deferred till "transaction idle", do the same here, so that the ordering is
 494                 // maintained. Having only the de-duplication registration succeed would cause
 495                 // jobs to become no-ops without any actual jobs that made them redundant.
 496                 $dbw = $this->getMasterDB();
 497                 $cache = $this->dupCache;
 498                 $dbw->onTransactionIdle(
 499                         function () use ( $cache, $params, $key, $dbw ) {
 500                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 501                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 502                                         return true; // a newer version of this root job was enqueued
 503                                 }
 504
 505                                 // Update the timestamp of the last root job started at the location...
 506                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 507                         },
 508                         __METHOD__
 509                 );
 510
 511                 return true;
 512         }
 513
 514         /**
 515          * @see JobQueue::doDelete()
 516          * @return bool
 517          */
 518         protected function doDelete() {
 519                 $dbw = $this->getMasterDB();
 520                 try {
 521                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 522                 } catch ( DBError $e ) {
 523                         $this->throwDBException( $e );
 524                 }
 525
 526                 return true;
 527         }
 528
 529         /**
 530          * @see JobQueue::doWaitForBackups()
 531          * @return void
 532          */
 533         protected function doWaitForBackups() {
 534                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 535                 $lbFactory->waitForReplication( [ 'wiki' => $this->wiki, 'cluster' => $this->cluster ] );
 536         }
 537
 538         /**
 539          * @return void
 540          */
 541         protected function doFlushCaches() {
 542                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 543                         $this->cache->delete( $this->getCacheKey( $type ) );
 544                 }
 545         }
 546
 547         /**
 548          * @see JobQueue::getAllQueuedJobs()
 549          * @return Iterator
 550          */
 551         public function getAllQueuedJobs() {
 552                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 553         }
 554
 555         /**
 556          * @see JobQueue::getAllAcquiredJobs()
 557          * @return Iterator
 558          */
 559         public function getAllAcquiredJobs() {
 560                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 561         }
 562
 563         /**
 564          * @param array $conds Query conditions
 565          * @return Iterator
 566          */
 567         protected function getJobIterator( array $conds ) {
 568                 $dbr = $this->getSlaveDB();
 569                 try {
 570                         return new MappedIterator(
 571                                 $dbr->select( 'job', self::selectFields(), $conds ),
 572                                 function ( $row ) {
 573                                         $job = Job::factory(
 574                                                 $row->job_cmd,
 575                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 576                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 577                                         );
 578                                         $job->metadata['id'] = $row->job_id;
 579                                         $job->metadata['timestamp'] = $row->job_timestamp;
 580
 581                                         return $job;
 582                                 }
 583                         );
 584                 } catch ( DBError $e ) {
 585                         $this->throwDBException( $e );
 586                 }
 587         }
 588
 589         public function getCoalesceLocationInternal() {
 590                 return $this->cluster
 591                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 592                         : "LBFactory:{$this->wiki}";
 593         }
 594
 595         protected function doGetSiblingQueuesWithJobs( array $types ) {
 596                 $dbr = $this->getSlaveDB();
 597                 // @note: this does not check whether the jobs are claimed or not.
 598                 // This is useful so JobQueueGroup::pop() also sees queues that only
 599                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 600                 // failed jobs so that they can be popped again for that edge case.
 601                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 602                         [ 'job_cmd' => $types ], __METHOD__ );
 603
 604                 $types = [];
 605                 foreach ( $res as $row ) {
 606                         $types[] = $row->job_cmd;
 607                 }
 608
 609                 return $types;
 610         }
 611
 612         protected function doGetSiblingQueueSizes( array $types ) {
 613                 $dbr = $this->getSlaveDB();
 614                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 615                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 616
 617                 $sizes = [];
 618                 foreach ( $res as $row ) {
 619                         $sizes[$row->job_cmd] = (int)$row->count;
 620                 }
 621
 622                 return $sizes;
 623         }
 624
 625         /**
 626          * Recycle or destroy any jobs that have been claimed for too long
 627          *
 628          * @return int Number of jobs recycled/deleted
 629          */
 630         public function recycleAndDeleteStaleJobs() {
 631                 $now = time();
 632                 $count = 0; // affected rows
 633                 $dbw = $this->getMasterDB();
 634
 635                 try {
 636                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 637                                 return $count; // already in progress
 638                         }
 639
 640                         // Remove claims on jobs acquired for too long if enabled...
 641                         if ( $this->claimTTL > 0 ) {
 642                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 643                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 644                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 645                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 646                                 $res = $dbw->select( 'job', 'job_id',
 647                                         [
 648                                                 'job_cmd' => $this->type,
 649                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 650                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 651                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 652                                         __METHOD__
 653                                 );
 654                                 $ids = array_map(
 655                                         function ( $o ) {
 656                                                 return $o->job_id;
 657                                         }, iterator_to_array( $res )
 658                                 );
 659                                 if ( count( $ids ) ) {
 660                                         // Reset job_token for these jobs so that other runners will pick them up.
 661                                         // Set the timestamp to the current time, as it is useful to now that the job
 662                                         // was already tried before (the timestamp becomes the "released" time).
 663                                         $dbw->update( 'job',
 664                                                 [
 665                                                         'job_token' => '',
 666                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 667                                                 [
 668                                                         'job_id' => $ids ],
 669                                                 __METHOD__
 670                                         );
 671                                         $affected = $dbw->affectedRows();
 672                                         $count += $affected;
 673                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 674                                         $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 675                                 }
 676                         }
 677
 678                         // Just destroy any stale jobs...
 679                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 680                         $conds = [
 681                                 'job_cmd' => $this->type,
 682                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 683                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 684                         ];
 685                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 686                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 687                         }
 688                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 689                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 690                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 691                         $ids = array_map(
 692                                 function ( $o ) {
 693                                         return $o->job_id;
 694                                 }, iterator_to_array( $res )
 695                         );
 696                         if ( count( $ids ) ) {
 697                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 698                                 $affected = $dbw->affectedRows();
 699                                 $count += $affected;
 700                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 701                         }
 702
 703                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 704                 } catch ( DBError $e ) {
 705                         $this->throwDBException( $e );
 706                 }
 707
 708                 return $count;
 709         }
 710
 711         /**
 712          * @param IJobSpecification $job
 713          * @return array
 714          */
 715         protected function insertFields( IJobSpecification $job ) {
 716                 $dbw = $this->getMasterDB();
 717
 718                 return [
 719                         // Fields that describe the nature of the job
 720                         'job_cmd' => $job->getType(),
 721                         'job_namespace' => $job->getTitle()->getNamespace(),
 722                         'job_title' => $job->getTitle()->getDBkey(),
 723                         'job_params' => self::makeBlob( $job->getParams() ),
 724                         // Additional job metadata
 725                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 726                         'job_timestamp' => $dbw->timestamp(),
 727                         'job_sha1' => Wikimedia\base_convert(
 728                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 729                                 16, 36, 31
 730                         ),
 731                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 732                 ];
 733         }
 734
 735         /**
 736          * @throws JobQueueConnectionError
 737          * @return DBConnRef
 738          */
 739         protected function getSlaveDB() {
 740                 try {
 741                         return $this->getDB( DB_REPLICA );
 742                 } catch ( DBConnectionError $e ) {
 743                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 744                 }
 745         }
 746
 747         /**
 748          * @throws JobQueueConnectionError
 749          * @return DBConnRef
 750          */
 751         protected function getMasterDB() {
 752                 try {
 753                         return $this->getDB( DB_MASTER );
 754                 } catch ( DBConnectionError $e ) {
 755                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 756                 }
 757         }
 758
 759         /**
 760          * @param int $index (DB_REPLICA/DB_MASTER)
 761          * @return DBConnRef
 762          */
 763         protected function getDB( $index ) {
 764                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 765                 $lb = ( $this->cluster !== false )
 766                         ? $lbFactory->getExternalLB( $this->cluster, $this->wiki )
 767                         : $lbFactory->getMainLB( $this->wiki );
 768
 769                 return $lb->getConnectionRef( $index, [], $this->wiki );
 770         }
 771
 772         /**
 773          * @param string $property
 774          * @return string
 775          */
 776         private function getCacheKey( $property ) {
 777                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 778                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 779
 780                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 781         }
 782
 783         /**
 784          * @param array|bool $params
 785          * @return string
 786          */
 787         protected static function makeBlob( $params ) {
 788                 if ( $params !== false ) {
 789                         return serialize( $params );
 790                 } else {
 791                         return '';
 792                 }
 793         }
 794
 795         /**
 796          * @param string $blob
 797          * @return bool|mixed
 798          */
 799         protected static function extractBlob( $blob ) {
 800                 if ( (string)$blob !== '' ) {
 801                         return unserialize( $blob );
 802                 } else {
 803                         return false;
 804                 }
 805         }
 806
 807         /**
 808          * @param DBError $e
 809          * @throws JobQueueError
 810          */
 811         protected function throwDBException( DBError $e ) {
 812                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 813         }
 814
 815         /**
 816          * Return the list of job fields that should be selected.
 817          * @since 1.23
 818          * @return array
 819          */
 820         public static function selectFields() {
 821                 return [
 822                         'job_id',
 823                         'job_cmd',
 824                         'job_namespace',
 825                         'job_title',
 826                         'job_timestamp',
 827                         'job_params',
 828                         'job_random',
 829                         'job_attempts',
 830                         'job_token',
 831                         'job_token_timestamp',
 832                         'job_sha1',
 833                 ];
 834         }
 835 }