includes/job/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 CACHE_TTL_LONG = 300; // integer; seconds to cache info that is kept up to date
  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 BagOStuff */
  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                 global $wgMemc;
  53
  54                 parent::__construct( $params );
  55
  56                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  57                 // Make sure that we don't use the SQL cache, which would be harmful
  58                 $this->cache = ( $wgMemc instanceof SqlBagOStuff ) ? new EmptyBagOStuff() : $wgMemc;
  59         }
  60
  61         protected function supportedOrders() {
  62                 return array( '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                 $key = $this->getCacheKey( 'empty' );
  75
  76                 $isEmpty = $this->cache->get( $key );
  77                 if ( $isEmpty === 'true' ) {
  78                         return true;
  79                 } elseif ( $isEmpty === 'false' ) {
  80                         return false;
  81                 }
  82
  83                 $dbr = $this->getSlaveDB();
  84                 try {
  85                         $found = $dbr->selectField( // unclaimed job
  86                                 'job', '1', array( 'job_cmd' => $this->type, 'job_token' => '' ), __METHOD__
  87                         );
  88                 } catch ( DBError $e ) {
  89                         $this->throwDBException( $e );
  90                 }
  91                 $this->cache->add( $key, $found ? 'false' : 'true', self::CACHE_TTL_LONG );
  92
  93                 return !$found;
  94         }
  95
  96         /**
  97          * @see JobQueue::doGetSize()
  98          * @return int
  99          */
 100         protected function doGetSize() {
 101                 $key = $this->getCacheKey( 'size' );
 102
 103                 $size = $this->cache->get( $key );
 104                 if ( is_int( $size ) ) {
 105                         return $size;
 106                 }
 107
 108                 try {
 109                         $dbr = $this->getSlaveDB();
 110                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 111                                 array( 'job_cmd' => $this->type, 'job_token' => '' ),
 112                                 __METHOD__
 113                         );
 114                 } catch ( DBError $e ) {
 115                         $this->throwDBException( $e );
 116                 }
 117                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 118
 119                 return $size;
 120         }
 121
 122         /**
 123          * @see JobQueue::doGetAcquiredCount()
 124          * @return int
 125          */
 126         protected function doGetAcquiredCount() {
 127                 if ( $this->claimTTL <= 0 ) {
 128                         return 0; // no acknowledgements
 129                 }
 130
 131                 $key = $this->getCacheKey( 'acquiredcount' );
 132
 133                 $count = $this->cache->get( $key );
 134                 if ( is_int( $count ) ) {
 135                         return $count;
 136                 }
 137
 138                 $dbr = $this->getSlaveDB();
 139                 try {
 140                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 141                                 array( 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ),
 142                                 __METHOD__
 143                         );
 144                 } catch ( DBError $e ) {
 145                         $this->throwDBException( $e );
 146                 }
 147                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 148
 149                 return $count;
 150         }
 151
 152         /**
 153          * @see JobQueue::doGetAbandonedCount()
 154          * @return int
 155          * @throws MWException
 156          */
 157         protected function doGetAbandonedCount() {
 158                 global $wgMemc;
 159
 160                 if ( $this->claimTTL <= 0 ) {
 161                         return 0; // no acknowledgements
 162                 }
 163
 164                 $key = $this->getCacheKey( 'abandonedcount' );
 165
 166                 $count = $wgMemc->get( $key );
 167                 if ( is_int( $count ) ) {
 168                         return $count;
 169                 }
 170
 171                 $dbr = $this->getSlaveDB();
 172                 try {
 173                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 174                                 array(
 175                                         'job_cmd' => $this->type,
 176                                         "job_token != {$dbr->addQuotes( '' )}",
 177                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 178                                 ),
 179                                 __METHOD__
 180                         );
 181                 } catch ( DBError $e ) {
 182                         $this->throwDBException( $e );
 183                 }
 184                 $wgMemc->set( $key, $count, self::CACHE_TTL_SHORT );
 185
 186                 return $count;
 187         }
 188
 189         /**
 190          * @see JobQueue::doBatchPush()
 191          * @param array $jobs
 192          * @param $flags
 193          * @throws DBError|Exception
 194          * @return bool
 195          */
 196         protected function doBatchPush( array $jobs, $flags ) {
 197                 $dbw = $this->getMasterDB();
 198
 199                 $that = $this;
 200                 $method = __METHOD__;
 201                 $dbw->onTransactionIdle(
 202                         function () use ( $dbw, $that, $jobs, $flags, $method ) {
 203                                 $that->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 204                         }
 205                 );
 206
 207                 return true;
 208         }
 209
 210         /**
 211          * This function should *not* be called outside of JobQueueDB
 212          *
 213          * @param IDatabase $dbw
 214          * @param array $jobs
 215          * @param int $flags
 216          * @param string $method
 217          * @throws DBError
 218          * @return bool
 219          */
 220         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 221                 if ( !count( $jobs ) ) {
 222                         return true;
 223                 }
 224
 225                 $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
 226                 $rowList = array(); // list of jobs for jobs that are are not de-duplicated
 227                 foreach ( $jobs as $job ) {
 228                         $row = $this->insertFields( $job );
 229                         if ( $job->ignoreDuplicates() ) {
 230                                 $rowSet[$row['job_sha1']] = $row;
 231                         } else {
 232                                 $rowList[] = $row;
 233                         }
 234                 }
 235
 236                 if ( $flags & self::QOS_ATOMIC ) {
 237                         $dbw->begin( $method ); // wrap all the job additions in one transaction
 238                 }
 239                 try {
 240                         // Strip out any duplicate jobs that are already in the queue...
 241                         if ( count( $rowSet ) ) {
 242                                 $res = $dbw->select( 'job', 'job_sha1',
 243                                         array(
 244                                                 // No job_type condition since it's part of the job_sha1 hash
 245                                                 'job_sha1' => array_keys( $rowSet ),
 246                                                 'job_token' => '' // unclaimed
 247                                         ),
 248                                         $method
 249                                 );
 250                                 foreach ( $res as $row ) {
 251                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 252                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 253                                 }
 254                         }
 255                         // Build the full list of job rows to insert
 256                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 257                         // Insert the job rows in chunks to avoid slave lag...
 258                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 259                                 $dbw->insert( 'job', $rowBatch, $method );
 260                         }
 261                         JobQueue::incrStats( 'job-insert', $this->type, count( $rows ) );
 262                         JobQueue::incrStats(
 263                                 'job-insert-duplicate',
 264                                 $this->type,
 265                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 266                         );
 267                 } catch ( DBError $e ) {
 268                         if ( $flags & self::QOS_ATOMIC ) {
 269                                 $dbw->rollback( $method );
 270                         }
 271                         throw $e;
 272                 }
 273                 if ( $flags & self::QOS_ATOMIC ) {
 274                         $dbw->commit( $method );
 275                 }
 276
 277                 $this->cache->set( $this->getCacheKey( 'empty' ), 'false', JobQueueDB::CACHE_TTL_LONG );
 278
 279                 return true;
 280         }
 281
 282         /**
 283          * @see JobQueue::doPop()
 284          * @return Job|bool
 285          */
 286         protected function doPop() {
 287                 if ( $this->cache->get( $this->getCacheKey( 'empty' ) ) === 'true' ) {
 288                         return false; // queue is empty
 289                 }
 290
 291                 $dbw = $this->getMasterDB();
 292                 try {
 293                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 294                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 295                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 296                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 297                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 298                         } );
 299
 300                         $uuid = wfRandomString( 32 ); // pop attempt
 301                         $job = false; // job popped off
 302                         do { // retry when our row is invalid or deleted as a duplicate
 303                                 // Try to reserve a row in the DB...
 304                                 if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 305                                         $row = $this->claimOldest( $uuid );
 306                                 } else { // random first
 307                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 308                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 309                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 310                                 }
 311                                 // Check if we found a row to reserve...
 312                                 if ( !$row ) {
 313                                         $this->cache->set( $this->getCacheKey( 'empty' ), 'true', self::CACHE_TTL_LONG );
 314                                         break; // nothing to do
 315                                 }
 316                                 JobQueue::incrStats( 'job-pop', $this->type );
 317                                 // Get the job object from the row...
 318                                 $title = Title::makeTitleSafe( $row->job_namespace, $row->job_title );
 319                                 if ( !$title ) {
 320                                         $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 321                                         wfDebug( "Row has invalid title '{$row->job_title}'." );
 322                                         continue; // try again
 323                                 }
 324                                 $job = Job::factory( $row->job_cmd, $title,
 325                                         self::extractBlob( $row->job_params ), $row->job_id );
 326                                 $job->metadata['id'] = $row->job_id;
 327                                 break; // done
 328                         } while ( true );
 329                 } catch ( DBError $e ) {
 330                         $this->throwDBException( $e );
 331                 }
 332
 333                 return $job;
 334         }
 335
 336         /**
 337          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 338          *
 339          * @param string $uuid 32 char hex string
 340          * @param $rand integer Random unsigned integer (31 bits)
 341          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 342          * @return stdClass|bool Row|false
 343          */
 344         protected function claimRandom( $uuid, $rand, $gte ) {
 345                 $dbw = $this->getMasterDB();
 346                 // Check cache to see if the queue has <= OFFSET items
 347                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 348
 349                 $row = false; // the row acquired
 350                 $invertedDirection = false; // whether one job_random direction was already scanned
 351                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 352                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 353                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 354                 // be used here with MySQL.
 355                 do {
 356                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 357                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 358                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 359                                 $ineq = $gte ? '>=' : '<=';
 360                                 $dir = $gte ? 'ASC' : 'DESC';
 361                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 362                                         array(
 363                                                 'job_cmd' => $this->type,
 364                                                 'job_token' => '', // unclaimed
 365                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 366                                         __METHOD__,
 367                                         array( 'ORDER BY' => "job_random {$dir}" )
 368                                 );
 369                                 if ( !$row && !$invertedDirection ) {
 370                                         $gte = !$gte;
 371                                         $invertedDirection = true;
 372                                         continue; // try the other direction
 373                                 }
 374                         } else { // table *may* have >= MAX_OFFSET rows
 375                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 376                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 377                                 // instead of job_random for reducing excess claim retries.
 378                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 379                                         array(
 380                                                 'job_cmd' => $this->type,
 381                                                 'job_token' => '', // unclaimed
 382                                         ),
 383                                         __METHOD__,
 384                                         array( 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) )
 385                                 );
 386                                 if ( !$row ) {
 387                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 388                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 389                                         continue; // use job_random
 390                                 }
 391                         }
 392
 393                         if ( $row ) { // claim the job
 394                                 $dbw->update( 'job', // update by PK
 395                                         array(
 396                                                 'job_token' => $uuid,
 397                                                 'job_token_timestamp' => $dbw->timestamp(),
 398                                                 'job_attempts = job_attempts+1' ),
 399                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 400                                         __METHOD__
 401                                 );
 402                                 // This might get raced out by another runner when claiming the previously
 403                                 // selected row. The use of job_random should minimize this problem, however.
 404                                 if ( !$dbw->affectedRows() ) {
 405                                         $row = false; // raced out
 406                                 }
 407                         } else {
 408                                 break; // nothing to do
 409                         }
 410                 } while ( !$row );
 411
 412                 return $row;
 413         }
 414
 415         /**
 416          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 417          *
 418          * @param string $uuid 32 char hex string
 419          * @return stdClass|bool Row|false
 420          */
 421         protected function claimOldest( $uuid ) {
 422                 $dbw = $this->getMasterDB();
 423
 424                 $row = false; // the row acquired
 425                 do {
 426                         if ( $dbw->getType() === 'mysql' ) {
 427                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 428                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 429                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 430                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 431                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 432                                         "SET " .
 433                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 434                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 435                                                 "job_attempts = job_attempts+1 " .
 436                                         "WHERE ( " .
 437                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 438                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 439                                         ") ORDER BY job_id ASC LIMIT 1",
 440                                         __METHOD__
 441                                 );
 442                         } else {
 443                                 // Use a subquery to find the job, within an UPDATE to claim it.
 444                                 // This uses as much of the DB wrapper functions as possible.
 445                                 $dbw->update( 'job',
 446                                         array(
 447                                                 'job_token' => $uuid,
 448                                                 'job_token_timestamp' => $dbw->timestamp(),
 449                                                 'job_attempts = job_attempts+1' ),
 450                                         array( 'job_id = (' .
 451                                                 $dbw->selectSQLText( 'job', 'job_id',
 452                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 453                                                         __METHOD__,
 454                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 455                                                 ')'
 456                                         ),
 457                                         __METHOD__
 458                                 );
 459                         }
 460                         // Fetch any row that we just reserved...
 461                         if ( $dbw->affectedRows() ) {
 462                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 463                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 464                                 );
 465                                 if ( !$row ) { // raced out by duplicate job removal
 466                                         wfDebug( "Row deleted as duplicate by another process." );
 467                                 }
 468                         } else {
 469                                 break; // nothing to do
 470                         }
 471                 } while ( !$row );
 472
 473                 return $row;
 474         }
 475
 476         /**
 477          * @see JobQueue::doAck()
 478          * @param Job $job
 479          * @throws MWException
 480          * @return Job|bool
 481          */
 482         protected function doAck( Job $job ) {
 483                 if ( !isset( $job->metadata['id'] ) ) {
 484                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 485                 }
 486
 487                 $dbw = $this->getMasterDB();
 488                 try {
 489                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 490                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 491                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 492                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 493                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 494                         } );
 495
 496                         // Delete a row with a single DELETE without holding row locks over RTTs...
 497                         $dbw->delete( 'job',
 498                                 array( 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ), __METHOD__ );
 499                 } catch ( DBError $e ) {
 500                         $this->throwDBException( $e );
 501                 }
 502
 503                 return true;
 504         }
 505
 506         /**
 507          * @see JobQueue::doDeduplicateRootJob()
 508          * @param Job $job
 509          * @throws MWException
 510          * @return bool
 511          */
 512         protected function doDeduplicateRootJob( Job $job ) {
 513                 $params = $job->getParams();
 514                 if ( !isset( $params['rootJobSignature'] ) ) {
 515                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 516                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 517                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 518                 }
 519                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 520                 // Callers should call batchInsert() and then this function so that if the insert
 521                 // fails, the de-duplication registration will be aborted. Since the insert is
 522                 // deferred till "transaction idle", do the same here, so that the ordering is
 523                 // maintained. Having only the de-duplication registration succeed would cause
 524                 // jobs to become no-ops without any actual jobs that made them redundant.
 525                 $dbw = $this->getMasterDB();
 526                 $cache = $this->dupCache;
 527                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 528                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 529                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 530                                 return true; // a newer version of this root job was enqueued
 531                         }
 532
 533                         // Update the timestamp of the last root job started at the location...
 534                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 535                 } );
 536
 537                 return true;
 538         }
 539
 540         /**
 541          * @see JobQueue::doDelete()
 542          * @return bool
 543          */
 544         protected function doDelete() {
 545                 $dbw = $this->getMasterDB();
 546                 try {
 547                         $dbw->delete( 'job', array( 'job_cmd' => $this->type ) );
 548                 } catch ( DBError $e ) {
 549                         $this->throwDBException( $e );
 550                 }
 551
 552                 return true;
 553         }
 554
 555         /**
 556          * @see JobQueue::doWaitForBackups()
 557          * @return void
 558          */
 559         protected function doWaitForBackups() {
 560                 wfWaitForSlaves();
 561         }
 562
 563         /**
 564          * @return array
 565          */
 566         protected function doGetPeriodicTasks() {
 567                 return array(
 568                         'recycleAndDeleteStaleJobs' => array(
 569                                 'callback' => array( $this, 'recycleAndDeleteStaleJobs' ),
 570                                 'period' => ceil( $this->claimTTL / 2 )
 571                         )
 572                 );
 573         }
 574
 575         /**
 576          * @return void
 577          */
 578         protected function doFlushCaches() {
 579                 foreach ( array( 'empty', 'size', 'acquiredcount' ) as $type ) {
 580                         $this->cache->delete( $this->getCacheKey( $type ) );
 581                 }
 582         }
 583
 584         /**
 585          * @see JobQueue::getAllQueuedJobs()
 586          * @return Iterator
 587          */
 588         public function getAllQueuedJobs() {
 589                 $dbr = $this->getSlaveDB();
 590                 try {
 591                         return new MappedIterator(
 592                                 $dbr->select( 'job', self::selectFields(),
 593                                         array( 'job_cmd' => $this->getType(), 'job_token' => '' ) ),
 594                                 function ( $row ) use ( $dbr ) {
 595                                         $job = Job::factory(
 596                                                 $row->job_cmd,
 597                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 598                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : false
 599                                         );
 600                                         $job->metadata['id'] = $row->job_id;
 601                                         return $job;
 602                                 }
 603                         );
 604                 } catch ( DBError $e ) {
 605                         $this->throwDBException( $e );
 606                 }
 607         }
 608
 609         public function getCoalesceLocationInternal() {
 610                 return $this->cluster
 611                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 612                         : "LBFactory:{$this->wiki}";
 613         }
 614
 615         protected function doGetSiblingQueuesWithJobs( array $types ) {
 616                 $dbr = $this->getSlaveDB();
 617                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 618                         array( 'job_cmd' => $types ), __METHOD__ );
 619
 620                 $types = array();
 621                 foreach ( $res as $row ) {
 622                         $types[] = $row->job_cmd;
 623                 }
 624
 625                 return $types;
 626         }
 627
 628         protected function doGetSiblingQueueSizes( array $types ) {
 629                 $dbr = $this->getSlaveDB();
 630                 $res = $dbr->select( 'job', array( 'job_cmd', 'COUNT(*) AS count' ),
 631                         array( 'job_cmd' => $types ), __METHOD__, array( 'GROUP BY' => 'job_cmd' ) );
 632
 633                 $sizes = array();
 634                 foreach ( $res as $row ) {
 635                         $sizes[$row->job_cmd] = (int)$row->count;
 636                 }
 637
 638                 return $sizes;
 639         }
 640
 641         /**
 642          * Recycle or destroy any jobs that have been claimed for too long
 643          *
 644          * @return int Number of jobs recycled/deleted
 645          */
 646         public function recycleAndDeleteStaleJobs() {
 647                 $now = time();
 648                 $count = 0; // affected rows
 649                 $dbw = $this->getMasterDB();
 650
 651                 try {
 652                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 653                                 return $count; // already in progress
 654                         }
 655
 656                         // Remove claims on jobs acquired for too long if enabled...
 657                         if ( $this->claimTTL > 0 ) {
 658                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 659                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 660                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 661                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 662                                 $res = $dbw->select( 'job', 'job_id',
 663                                         array(
 664                                                 'job_cmd' => $this->type,
 665                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 666                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 667                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ), // retries left
 668                                         __METHOD__
 669                                 );
 670                                 $ids = array_map(
 671                                         function ( $o ) {
 672                                                 return $o->job_id;
 673                                         }, iterator_to_array( $res )
 674                                 );
 675                                 if ( count( $ids ) ) {
 676                                         // Reset job_token for these jobs so that other runners will pick them up.
 677                                         // Set the timestamp to the current time, as it is useful to now that the job
 678                                         // was already tried before (the timestamp becomes the "released" time).
 679                                         $dbw->update( 'job',
 680                                                 array(
 681                                                         'job_token' => '',
 682                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 683                                                 array(
 684                                                         'job_id' => $ids ),
 685                                                 __METHOD__
 686                                         );
 687                                         $count += $dbw->affectedRows();
 688                                         JobQueue::incrStats( 'job-recycle', $this->type, $dbw->affectedRows() );
 689                                         $this->cache->set( $this->getCacheKey( 'empty' ), 'false', self::CACHE_TTL_LONG );
 690                                 }
 691                         }
 692
 693                         // Just destroy any stale jobs...
 694                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 695                         $conds = array(
 696                                 'job_cmd' => $this->type,
 697                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 698                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 699                         );
 700                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 701                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 702                         }
 703                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 704                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 705                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 706                         $ids = array_map(
 707                                 function ( $o ) {
 708                                         return $o->job_id;
 709                                 }, iterator_to_array( $res )
 710                         );
 711                         if ( count( $ids ) ) {
 712                                 $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 713                                 $count += $dbw->affectedRows();
 714                                 JobQueue::incrStats( 'job-abandon', $this->type, $dbw->affectedRows() );
 715                         }
 716
 717                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 718                 } catch ( DBError $e ) {
 719                         $this->throwDBException( $e );
 720                 }
 721
 722                 return $count;
 723         }
 724
 725         /**
 726          * @param Job $job
 727          * @return array
 728          */
 729         protected function insertFields( Job $job ) {
 730                 $dbw = $this->getMasterDB();
 731
 732                 return array(
 733                         // Fields that describe the nature of the job
 734                         'job_cmd' => $job->getType(),
 735                         'job_namespace' => $job->getTitle()->getNamespace(),
 736                         'job_title' => $job->getTitle()->getDBkey(),
 737                         'job_params' => self::makeBlob( $job->getParams() ),
 738                         // Additional job metadata
 739                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 740                         'job_timestamp' => $dbw->timestamp(),
 741                         'job_sha1' => wfBaseConvert(
 742                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 743                                 16, 36, 31
 744                         ),
 745                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 746                 );
 747         }
 748
 749         /**
 750          * @throws JobQueueConnectionError
 751          * @return DBConnRef
 752          */
 753         protected function getSlaveDB() {
 754                 try {
 755                         return $this->getDB( DB_SLAVE );
 756                 } catch ( DBConnectionError $e ) {
 757                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 758                 }
 759         }
 760
 761         /**
 762          * @throws JobQueueConnectionError
 763          * @return DBConnRef
 764          */
 765         protected function getMasterDB() {
 766                 try {
 767                         return $this->getDB( DB_MASTER );
 768                 } catch ( DBConnectionError $e ) {
 769                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 770                 }
 771         }
 772
 773         /**
 774          * @param $index integer (DB_SLAVE/DB_MASTER)
 775          * @return DBConnRef
 776          */
 777         protected function getDB( $index ) {
 778                 $lb = ( $this->cluster !== false )
 779                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 780                         : wfGetLB( $this->wiki );
 781
 782                 return $lb->getConnectionRef( $index, array(), $this->wiki );
 783         }
 784
 785         /**
 786          * @param $property
 787          * @return string
 788          */
 789         private function getCacheKey( $property ) {
 790                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 791                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 792
 793                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 794         }
 795
 796         /**
 797          * @param $params
 798          * @return string
 799          */
 800         protected static function makeBlob( $params ) {
 801                 if ( $params !== false ) {
 802                         return serialize( $params );
 803                 } else {
 804                         return '';
 805                 }
 806         }
 807
 808         /**
 809          * @param $blob
 810          * @return bool|mixed
 811          */
 812         protected static function extractBlob( $blob ) {
 813                 if ( (string)$blob !== '' ) {
 814                         return unserialize( $blob );
 815                 } else {
 816                         return false;
 817                 }
 818         }
 819
 820         /**
 821          * @param DBError $e
 822          * @throws JobQueueError
 823          */
 824         protected function throwDBException( DBError $e ) {
 825                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 826         }
 827
 828         /**
 829          * Return the list of job fields that should be selected.
 830          * @since 1.23
 831          * @return array
 832          */
 833         public static function selectFields() {
 834                 return array(
 835                         'job_id',
 836                         'job_cmd',
 837                         'job_namespace',
 838                         'job_title',
 839                         'job_timestamp',
 840                         'job_params',
 841                         'job_random',
 842                         'job_attempts',
 843                         'job_token',
 844                         'job_token_timestamp',
 845                         'job_sha1',
 846                 );
 847         }
 848 }