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
  24 /**
  25  * Class to handle job queues stored in the DB
  26  *
  27  * @ingroup JobQueue
  28  * @since 1.21
  29  */
  30 class JobQueueDB extends JobQueue {
  31         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  32         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  33         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  34         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  35
  36         /** @var WANObjectCache */
  37         protected $cache;
  38
  39         /** @var bool|string Name of an external DB cluster. False if not set */
  40         protected $cluster = false;
  41
  42         /**
  43          * Additional parameters include:
  44          *   - cluster : The name of an external cluster registered via LBFactory.
  45          *               If not specified, the primary DB cluster for the wiki will be used.
  46          *               This can be overridden with a custom cluster so that DB handles will
  47          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  48          * @param array $params
  49          */
  50         protected function __construct( array $params ) {
  51                 parent::__construct( $params );
  52
  53                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  54                 $this->cache = ObjectCache::getMainWANInstance();
  55         }
  56
  57         protected function supportedOrders() {
  58                 return array( 'random', 'timestamp', 'fifo' );
  59         }
  60
  61         protected function optimalOrder() {
  62                 return 'random';
  63         }
  64
  65         /**
  66          * @see JobQueue::doIsEmpty()
  67          * @return bool
  68          */
  69         protected function doIsEmpty() {
  70                 $dbr = $this->getSlaveDB();
  71                 try {
  72                         $found = $dbr->selectField( // unclaimed job
  73                                 'job', '1', array( 'job_cmd' => $this->type, 'job_token' => '' ), __METHOD__
  74                         );
  75                 } catch ( DBError $e ) {
  76                         $this->throwDBException( $e );
  77                 }
  78
  79                 return !$found;
  80         }
  81
  82         /**
  83          * @see JobQueue::doGetSize()
  84          * @return int
  85          */
  86         protected function doGetSize() {
  87                 $key = $this->getCacheKey( 'size' );
  88
  89                 $size = $this->cache->get( $key );
  90                 if ( is_int( $size ) ) {
  91                         return $size;
  92                 }
  93
  94                 try {
  95                         $dbr = $this->getSlaveDB();
  96                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
  97                                 array( 'job_cmd' => $this->type, 'job_token' => '' ),
  98                                 __METHOD__
  99                         );
 100                 } catch ( DBError $e ) {
 101                         $this->throwDBException( $e );
 102                 }
 103                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 104
 105                 return $size;
 106         }
 107
 108         /**
 109          * @see JobQueue::doGetAcquiredCount()
 110          * @return int
 111          */
 112         protected function doGetAcquiredCount() {
 113                 if ( $this->claimTTL <= 0 ) {
 114                         return 0; // no acknowledgements
 115                 }
 116
 117                 $key = $this->getCacheKey( 'acquiredcount' );
 118
 119                 $count = $this->cache->get( $key );
 120                 if ( is_int( $count ) ) {
 121                         return $count;
 122                 }
 123
 124                 $dbr = $this->getSlaveDB();
 125                 try {
 126                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 127                                 array( 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ),
 128                                 __METHOD__
 129                         );
 130                 } catch ( DBError $e ) {
 131                         $this->throwDBException( $e );
 132                 }
 133                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 134
 135                 return $count;
 136         }
 137
 138         /**
 139          * @see JobQueue::doGetAbandonedCount()
 140          * @return int
 141          * @throws MWException
 142          */
 143         protected function doGetAbandonedCount() {
 144                 if ( $this->claimTTL <= 0 ) {
 145                         return 0; // no acknowledgements
 146                 }
 147
 148                 $key = $this->getCacheKey( 'abandonedcount' );
 149
 150                 $count = $this->cache->get( $key );
 151                 if ( is_int( $count ) ) {
 152                         return $count;
 153                 }
 154
 155                 $dbr = $this->getSlaveDB();
 156                 try {
 157                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 158                                 array(
 159                                         'job_cmd' => $this->type,
 160                                         "job_token != {$dbr->addQuotes( '' )}",
 161                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 162                                 ),
 163                                 __METHOD__
 164                         );
 165                 } catch ( DBError $e ) {
 166                         $this->throwDBException( $e );
 167                 }
 168
 169                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 170
 171                 return $count;
 172         }
 173
 174         /**
 175          * @see JobQueue::doBatchPush()
 176          * @param IJobSpecification[] $jobs
 177          * @param int $flags
 178          * @throws DBError|Exception
 179          * @return void
 180          */
 181         protected function doBatchPush( array $jobs, $flags ) {
 182                 $dbw = $this->getMasterDB();
 183
 184                 $that = $this;
 185                 $method = __METHOD__;
 186                 $dbw->onTransactionIdle(
 187                         function () use ( $dbw, $that, $jobs, $flags, $method ) {
 188                                 $that->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 189                         }
 190                 );
 191         }
 192
 193         /**
 194          * This function should *not* be called outside of JobQueueDB
 195          *
 196          * @param IDatabase $dbw
 197          * @param IJobSpecification[] $jobs
 198          * @param int $flags
 199          * @param string $method
 200          * @throws DBError
 201          * @return void
 202          */
 203         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 204                 if ( !count( $jobs ) ) {
 205                         return;
 206                 }
 207
 208                 $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
 209                 $rowList = array(); // list of jobs for jobs that are not de-duplicated
 210                 foreach ( $jobs as $job ) {
 211                         $row = $this->insertFields( $job );
 212                         if ( $job->ignoreDuplicates() ) {
 213                                 $rowSet[$row['job_sha1']] = $row;
 214                         } else {
 215                                 $rowList[] = $row;
 216                         }
 217                 }
 218
 219                 if ( $flags & self::QOS_ATOMIC ) {
 220                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 221                 }
 222                 try {
 223                         // Strip out any duplicate jobs that are already in the queue...
 224                         if ( count( $rowSet ) ) {
 225                                 $res = $dbw->select( 'job', 'job_sha1',
 226                                         array(
 227                                                 // No job_type condition since it's part of the job_sha1 hash
 228                                                 'job_sha1' => array_keys( $rowSet ),
 229                                                 'job_token' => '' // unclaimed
 230                                         ),
 231                                         $method
 232                                 );
 233                                 foreach ( $res as $row ) {
 234                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 235                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 236                                 }
 237                         }
 238                         // Build the full list of job rows to insert
 239                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 240                         // Insert the job rows in chunks to avoid slave lag...
 241                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 242                                 $dbw->insert( 'job', $rowBatch, $method );
 243                         }
 244                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 245                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 246                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 247                         );
 248                 } catch ( DBError $e ) {
 249                         if ( $flags & self::QOS_ATOMIC ) {
 250                                 $dbw->rollback( $method );
 251                         }
 252                         throw $e;
 253                 }
 254                 if ( $flags & self::QOS_ATOMIC ) {
 255                         $dbw->endAtomic( $method );
 256                 }
 257
 258                 return;
 259         }
 260
 261         /**
 262          * @see JobQueue::doPop()
 263          * @return Job|bool
 264          */
 265         protected function doPop() {
 266                 $dbw = $this->getMasterDB();
 267                 try {
 268                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 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, array( '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 http://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                                         array(
 339                                                 'job_cmd' => $this->type,
 340                                                 'job_token' => '', // unclaimed
 341                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 342                                         __METHOD__,
 343                                         array( 'ORDER BY' => "job_random {$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                                 // Bug 42614: "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                                         array(
 356                                                 'job_cmd' => $this->type,
 357                                                 'job_token' => '', // unclaimed
 358                                         ),
 359                                         __METHOD__,
 360                                         array( 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) )
 361                                 );
 362                                 if ( !$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                                         array(
 372                                                 'job_token' => $uuid,
 373                                                 'job_token_timestamp' => $dbw->timestamp(),
 374                                                 'job_attempts = job_attempts+1' ),
 375                                         array( '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 http://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                                         array(
 423                                                 'job_token' => $uuid,
 424                                                 'job_token_timestamp' => $dbw->timestamp(),
 425                                                 'job_attempts = job_attempts+1' ),
 426                                         array( 'job_id = (' .
 427                                                 $dbw->selectSQLText( 'job', 'job_id',
 428                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 429                                                         __METHOD__,
 430                                                         array( '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                                         array( '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          * @return Job|bool
 457          */
 458         protected function doAck( Job $job ) {
 459                 if ( !isset( $job->metadata['id'] ) ) {
 460                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 461                 }
 462
 463                 $dbw = $this->getMasterDB();
 464                 try {
 465                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 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                                 array( '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                 return true;
 482         }
 483
 484         /**
 485          * @see JobQueue::doDeduplicateRootJob()
 486          * @param IJobSpecification $job
 487          * @throws MWException
 488          * @return bool
 489          */
 490         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 491                 $params = $job->getParams();
 492                 if ( !isset( $params['rootJobSignature'] ) ) {
 493                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 494                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 495                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 496                 }
 497                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 498                 // Callers should call batchInsert() and then this function so that if the insert
 499                 // fails, the de-duplication registration will be aborted. Since the insert is
 500                 // deferred till "transaction idle", do the same here, so that the ordering is
 501                 // maintained. Having only the de-duplication registration succeed would cause
 502                 // jobs to become no-ops without any actual jobs that made them redundant.
 503                 $dbw = $this->getMasterDB();
 504                 $cache = $this->dupCache;
 505                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 506                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 507                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 508                                 return true; // a newer version of this root job was enqueued
 509                         }
 510
 511                         // Update the timestamp of the last root job started at the location...
 512                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 513                 } );
 514
 515                 return true;
 516         }
 517
 518         /**
 519          * @see JobQueue::doDelete()
 520          * @return bool
 521          */
 522         protected function doDelete() {
 523                 $dbw = $this->getMasterDB();
 524                 try {
 525                         $dbw->delete( 'job', array( 'job_cmd' => $this->type ) );
 526                 } catch ( DBError $e ) {
 527                         $this->throwDBException( $e );
 528                 }
 529
 530                 return true;
 531         }
 532
 533         /**
 534          * @see JobQueue::doWaitForBackups()
 535          * @return void
 536          */
 537         protected function doWaitForBackups() {
 538                 wfWaitForSlaves( false, $this->wiki, $this->cluster ?: false );
 539         }
 540
 541         /**
 542          * @return void
 543          */
 544         protected function doFlushCaches() {
 545                 foreach ( array( '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( array( '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( array( '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->getSlaveDB();
 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 ) : array()
 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->getSlaveDB();
 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                         array( 'job_cmd' => $types ), __METHOD__ );
 606
 607                 $types = array();
 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->getSlaveDB();
 617                 $res = $dbr->select( 'job', array( 'job_cmd', 'COUNT(*) AS count' ),
 618                         array( 'job_cmd' => $types ), __METHOD__, array( 'GROUP BY' => 'job_cmd' ) );
 619
 620                 $sizes = array();
 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                                         array(
 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                                                 array(
 668                                                         'job_token' => '',
 669                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 670                                                 array(
 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 = array(
 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', array( '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 array(
 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 getSlaveDB() {
 743                 try {
 744                         return $this->getDB( DB_SLAVE );
 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_SLAVE/DB_MASTER)
 764          * @return DBConnRef
 765          */
 766         protected function getDB( $index ) {
 767                 $lb = ( $this->cluster !== false )
 768                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 769                         : wfGetLB( $this->wiki );
 770
 771                 return $lb->getConnectionRef( $index, array(), $this->wiki );
 772         }
 773
 774         /**
 775          * @param string $property
 776          * @return string
 777          */
 778         private function getCacheKey( $property ) {
 779                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 780                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 781
 782                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 783         }
 784
 785         /**
 786          * @param array|bool $params
 787          * @return string
 788          */
 789         protected static function makeBlob( $params ) {
 790                 if ( $params !== false ) {
 791                         return serialize( $params );
 792                 } else {
 793                         return '';
 794                 }
 795         }
 796
 797         /**
 798          * @param string $blob
 799          * @return bool|mixed
 800          */
 801         protected static function extractBlob( $blob ) {
 802                 if ( (string)$blob !== '' ) {
 803                         return unserialize( $blob );
 804                 } else {
 805                         return false;
 806                 }
 807         }
 808
 809         /**
 810          * @param DBError $e
 811          * @throws JobQueueError
 812          */
 813         protected function throwDBException( DBError $e ) {
 814                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 815         }
 816
 817         /**
 818          * Return the list of job fields that should be selected.
 819          * @since 1.23
 820          * @return array
 821          */
 822         public static function selectFields() {
 823                 return array(
 824                         'job_id',
 825                         'job_cmd',
 826                         'job_namespace',
 827                         'job_title',
 828                         'job_timestamp',
 829                         'job_params',
 830                         'job_random',
 831                         'job_attempts',
 832                         'job_token',
 833                         'job_token_timestamp',
 834                         'job_sha1',
 835                 );
 836         }
 837 }