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                                 $job->id = $row->job_id; // XXX: work around broken subclasses
 328                                 break; // done
 329                         } while ( true );
 330                 } catch ( DBError $e ) {
 331                         $this->throwDBException( $e );
 332                 }
 333
 334                 return $job;
 335         }
 336
 337         /**
 338          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 339          *
 340          * @param string $uuid 32 char hex string
 341          * @param $rand integer Random unsigned integer (31 bits)
 342          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 343          * @return stdClass|bool Row|false
 344          */
 345         protected function claimRandom( $uuid, $rand, $gte ) {
 346                 $dbw = $this->getMasterDB();
 347                 // Check cache to see if the queue has <= OFFSET items
 348                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 349
 350                 $row = false; // the row acquired
 351                 $invertedDirection = false; // whether one job_random direction was already scanned
 352                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 353                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 354                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 355                 // be used here with MySQL.
 356                 do {
 357                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 358                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 359                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 360                                 $ineq = $gte ? '>=' : '<=';
 361                                 $dir = $gte ? 'ASC' : 'DESC';
 362                                 $row = $dbw->selectRow( 'job', '*', // find a random job
 363                                         array(
 364                                                 'job_cmd' => $this->type,
 365                                                 'job_token' => '', // unclaimed
 366                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 367                                         __METHOD__,
 368                                         array( 'ORDER BY' => "job_random {$dir}" )
 369                                 );
 370                                 if ( !$row && !$invertedDirection ) {
 371                                         $gte = !$gte;
 372                                         $invertedDirection = true;
 373                                         continue; // try the other direction
 374                                 }
 375                         } else { // table *may* have >= MAX_OFFSET rows
 376                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 377                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 378                                 // instead of job_random for reducing excess claim retries.
 379                                 $row = $dbw->selectRow( 'job', '*', // find a random job
 380                                         array(
 381                                                 'job_cmd' => $this->type,
 382                                                 'job_token' => '', // unclaimed
 383                                         ),
 384                                         __METHOD__,
 385                                         array( 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) )
 386                                 );
 387                                 if ( !$row ) {
 388                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 389                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 390                                         continue; // use job_random
 391                                 }
 392                         }
 393
 394                         if ( $row ) { // claim the job
 395                                 $dbw->update( 'job', // update by PK
 396                                         array(
 397                                                 'job_token' => $uuid,
 398                                                 'job_token_timestamp' => $dbw->timestamp(),
 399                                                 'job_attempts = job_attempts+1' ),
 400                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 401                                         __METHOD__
 402                                 );
 403                                 // This might get raced out by another runner when claiming the previously
 404                                 // selected row. The use of job_random should minimize this problem, however.
 405                                 if ( !$dbw->affectedRows() ) {
 406                                         $row = false; // raced out
 407                                 }
 408                         } else {
 409                                 break; // nothing to do
 410                         }
 411                 } while ( !$row );
 412
 413                 return $row;
 414         }
 415
 416         /**
 417          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 418          *
 419          * @param string $uuid 32 char hex string
 420          * @return stdClass|bool Row|false
 421          */
 422         protected function claimOldest( $uuid ) {
 423                 $dbw = $this->getMasterDB();
 424
 425                 $row = false; // the row acquired
 426                 do {
 427                         if ( $dbw->getType() === 'mysql' ) {
 428                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 429                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 430                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 431                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 432                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 433                                         "SET " .
 434                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 435                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 436                                                 "job_attempts = job_attempts+1 " .
 437                                         "WHERE ( " .
 438                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 439                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 440                                         ") ORDER BY job_id ASC LIMIT 1",
 441                                         __METHOD__
 442                                 );
 443                         } else {
 444                                 // Use a subquery to find the job, within an UPDATE to claim it.
 445                                 // This uses as much of the DB wrapper functions as possible.
 446                                 $dbw->update( 'job',
 447                                         array(
 448                                                 'job_token' => $uuid,
 449                                                 'job_token_timestamp' => $dbw->timestamp(),
 450                                                 'job_attempts = job_attempts+1' ),
 451                                         array( 'job_id = (' .
 452                                                 $dbw->selectSQLText( 'job', 'job_id',
 453                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 454                                                         __METHOD__,
 455                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 456                                                 ')'
 457                                         ),
 458                                         __METHOD__
 459                                 );
 460                         }
 461                         // Fetch any row that we just reserved...
 462                         if ( $dbw->affectedRows() ) {
 463                                 $row = $dbw->selectRow( 'job', '*',
 464                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 465                                 );
 466                                 if ( !$row ) { // raced out by duplicate job removal
 467                                         wfDebug( "Row deleted as duplicate by another process." );
 468                                 }
 469                         } else {
 470                                 break; // nothing to do
 471                         }
 472                 } while ( !$row );
 473
 474                 return $row;
 475         }
 476
 477         /**
 478          * @see JobQueue::doAck()
 479          * @param Job $job
 480          * @throws MWException
 481          * @return Job|bool
 482          */
 483         protected function doAck( Job $job ) {
 484                 if ( !isset( $job->metadata['id'] ) ) {
 485                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 486                 }
 487
 488                 $dbw = $this->getMasterDB();
 489                 try {
 490                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 491                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 492                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 493                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 494                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 495                         } );
 496
 497                         // Delete a row with a single DELETE without holding row locks over RTTs...
 498                         $dbw->delete( 'job',
 499                                 array( 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ), __METHOD__ );
 500                 } catch ( DBError $e ) {
 501                         $this->throwDBException( $e );
 502                 }
 503
 504                 return true;
 505         }
 506
 507         /**
 508          * @see JobQueue::doDeduplicateRootJob()
 509          * @param Job $job
 510          * @throws MWException
 511          * @return bool
 512          */
 513         protected function doDeduplicateRootJob( Job $job ) {
 514                 $params = $job->getParams();
 515                 if ( !isset( $params['rootJobSignature'] ) ) {
 516                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 517                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 518                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 519                 }
 520                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 521                 // Callers should call batchInsert() and then this function so that if the insert
 522                 // fails, the de-duplication registration will be aborted. Since the insert is
 523                 // deferred till "transaction idle", do the same here, so that the ordering is
 524                 // maintained. Having only the de-duplication registration succeed would cause
 525                 // jobs to become no-ops without any actual jobs that made them redundant.
 526                 $dbw = $this->getMasterDB();
 527                 $cache = $this->dupCache;
 528                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 529                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 530                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 531                                 return true; // a newer version of this root job was enqueued
 532                         }
 533
 534                         // Update the timestamp of the last root job started at the location...
 535                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 536                 } );
 537
 538                 return true;
 539         }
 540
 541         /**
 542          * @see JobQueue::doDelete()
 543          * @return bool
 544          */
 545         protected function doDelete() {
 546                 $dbw = $this->getMasterDB();
 547                 try {
 548                         $dbw->delete( 'job', array( 'job_cmd' => $this->type ) );
 549                 } catch ( DBError $e ) {
 550                         $this->throwDBException( $e );
 551                 }
 552
 553                 return true;
 554         }
 555
 556         /**
 557          * @see JobQueue::doWaitForBackups()
 558          * @return void
 559          */
 560         protected function doWaitForBackups() {
 561                 wfWaitForSlaves();
 562         }
 563
 564         /**
 565          * @return array
 566          */
 567         protected function doGetPeriodicTasks() {
 568                 return array(
 569                         'recycleAndDeleteStaleJobs' => array(
 570                                 'callback' => array( $this, 'recycleAndDeleteStaleJobs' ),
 571                                 'period' => ceil( $this->claimTTL / 2 )
 572                         )
 573                 );
 574         }
 575
 576         /**
 577          * @return void
 578          */
 579         protected function doFlushCaches() {
 580                 foreach ( array( 'empty', 'size', 'acquiredcount' ) as $type ) {
 581                         $this->cache->delete( $this->getCacheKey( $type ) );
 582                 }
 583         }
 584
 585         /**
 586          * @see JobQueue::getAllQueuedJobs()
 587          * @return Iterator
 588          */
 589         public function getAllQueuedJobs() {
 590                 $dbr = $this->getSlaveDB();
 591                 try {
 592                         return new MappedIterator(
 593                                 $dbr->select( 'job', '*',
 594                                         array( 'job_cmd' => $this->getType(), 'job_token' => '' ) ),
 595                                 function ( $row ) use ( $dbr ) {
 596                                         $job = Job::factory(
 597                                                 $row->job_cmd,
 598                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 599                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : false,
 600                                                 $row->job_id
 601                                         );
 602                                         $job->metadata['id'] = $row->job_id;
 603                                         $job->id = $row->job_id; // XXX: work around broken subclasses
 604                                         return $job;
 605                                 }
 606                         );
 607                 } catch ( DBError $e ) {
 608                         $this->throwDBException( $e );
 609                 }
 610         }
 611
 612         public function getCoalesceLocationInternal() {
 613                 return $this->cluster
 614                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 615                         : "LBFactory:{$this->wiki}";
 616         }
 617
 618         protected function doGetSiblingQueuesWithJobs( array $types ) {
 619                 $dbr = $this->getSlaveDB();
 620                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 621                         array( 'job_cmd' => $types ), __METHOD__ );
 622
 623                 $types = array();
 624                 foreach ( $res as $row ) {
 625                         $types[] = $row->job_cmd;
 626                 }
 627
 628                 return $types;
 629         }
 630
 631         protected function doGetSiblingQueueSizes( array $types ) {
 632                 $dbr = $this->getSlaveDB();
 633                 $res = $dbr->select( 'job', array( 'job_cmd', 'COUNT(*) AS count' ),
 634                         array( 'job_cmd' => $types ), __METHOD__, array( 'GROUP BY' => 'job_cmd' ) );
 635
 636                 $sizes = array();
 637                 foreach ( $res as $row ) {
 638                         $sizes[$row->job_cmd] = (int)$row->count;
 639                 }
 640
 641                 return $sizes;
 642         }
 643
 644         /**
 645          * Recycle or destroy any jobs that have been claimed for too long
 646          *
 647          * @return int Number of jobs recycled/deleted
 648          */
 649         public function recycleAndDeleteStaleJobs() {
 650                 $now = time();
 651                 $count = 0; // affected rows
 652                 $dbw = $this->getMasterDB();
 653
 654                 try {
 655                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 656                                 return $count; // already in progress
 657                         }
 658
 659                         // Remove claims on jobs acquired for too long if enabled...
 660                         if ( $this->claimTTL > 0 ) {
 661                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 662                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 663                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 664                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 665                                 $res = $dbw->select( 'job', 'job_id',
 666                                         array(
 667                                                 'job_cmd' => $this->type,
 668                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 669                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 670                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ), // retries left
 671                                         __METHOD__
 672                                 );
 673                                 $ids = array_map(
 674                                         function ( $o ) {
 675                                                 return $o->job_id;
 676                                         }, iterator_to_array( $res )
 677                                 );
 678                                 if ( count( $ids ) ) {
 679                                         // Reset job_token for these jobs so that other runners will pick them up.
 680                                         // Set the timestamp to the current time, as it is useful to now that the job
 681                                         // was already tried before (the timestamp becomes the "released" time).
 682                                         $dbw->update( 'job',
 683                                                 array(
 684                                                         'job_token' => '',
 685                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 686                                                 array(
 687                                                         'job_id' => $ids ),
 688                                                 __METHOD__
 689                                         );
 690                                         $count += $dbw->affectedRows();
 691                                         JobQueue::incrStats( 'job-recycle', $this->type, $dbw->affectedRows() );
 692                                         $this->cache->set( $this->getCacheKey( 'empty' ), 'false', self::CACHE_TTL_LONG );
 693                                 }
 694                         }
 695
 696                         // Just destroy any stale jobs...
 697                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 698                         $conds = array(
 699                                 'job_cmd' => $this->type,
 700                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 701                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 702                         );
 703                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 704                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 705                         }
 706                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 707                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 708                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 709                         $ids = array_map(
 710                                 function ( $o ) {
 711                                         return $o->job_id;
 712                                 }, iterator_to_array( $res )
 713                         );
 714                         if ( count( $ids ) ) {
 715                                 $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 716                                 $count += $dbw->affectedRows();
 717                                 JobQueue::incrStats( 'job-abandon', $this->type, $dbw->affectedRows() );
 718                         }
 719
 720                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 721                 } catch ( DBError $e ) {
 722                         $this->throwDBException( $e );
 723                 }
 724
 725                 return $count;
 726         }
 727
 728         /**
 729          * @param Job $job
 730          * @return array
 731          */
 732         protected function insertFields( Job $job ) {
 733                 $dbw = $this->getMasterDB();
 734
 735                 return array(
 736                         // Fields that describe the nature of the job
 737                         'job_cmd' => $job->getType(),
 738                         'job_namespace' => $job->getTitle()->getNamespace(),
 739                         'job_title' => $job->getTitle()->getDBkey(),
 740                         'job_params' => self::makeBlob( $job->getParams() ),
 741                         // Additional job metadata
 742                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 743                         'job_timestamp' => $dbw->timestamp(),
 744                         'job_sha1' => wfBaseConvert(
 745                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 746                                 16, 36, 31
 747                         ),
 748                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 749                 );
 750         }
 751
 752         /**
 753          * @throws JobQueueConnectionError
 754          * @return DBConnRef
 755          */
 756         protected function getSlaveDB() {
 757                 try {
 758                         return $this->getDB( DB_SLAVE );
 759                 } catch ( DBConnectionError $e ) {
 760                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 761                 }
 762         }
 763
 764         /**
 765          * @throws JobQueueConnectionError
 766          * @return DBConnRef
 767          */
 768         protected function getMasterDB() {
 769                 try {
 770                         return $this->getDB( DB_MASTER );
 771                 } catch ( DBConnectionError $e ) {
 772                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 773                 }
 774         }
 775
 776         /**
 777          * @param $index integer (DB_SLAVE/DB_MASTER)
 778          * @return DBConnRef
 779          */
 780         protected function getDB( $index ) {
 781                 $lb = ( $this->cluster !== false )
 782                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 783                         : wfGetLB( $this->wiki );
 784
 785                 return $lb->getConnectionRef( $index, array(), $this->wiki );
 786         }
 787
 788         /**
 789          * @param $property
 790          * @return string
 791          */
 792         private function getCacheKey( $property ) {
 793                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 794                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 795
 796                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 797         }
 798
 799         /**
 800          * @param $params
 801          * @return string
 802          */
 803         protected static function makeBlob( $params ) {
 804                 if ( $params !== false ) {
 805                         return serialize( $params );
 806                 } else {
 807                         return '';
 808                 }
 809         }
 810
 811         /**
 812          * @param $blob
 813          * @return bool|mixed
 814          */
 815         protected static function extractBlob( $blob ) {
 816                 if ( (string)$blob !== '' ) {
 817                         return unserialize( $blob );
 818                 } else {
 819                         return false;
 820                 }
 821         }
 822
 823         /**
 824          * @param DBError $e
 825          * @throws JobQueueError
 826          */
 827         protected function throwDBException( DBError $e ) {
 828                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 829         }
 830 }