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 BagOStuff */
  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                 global $wgMemc;
  52
  53                 parent::__construct( $params );
  54
  55                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  56                 // Make sure that we don't use the SQL cache, which would be harmful
  57                 $this->cache = ( $wgMemc instanceof SqlBagOStuff ) ? new EmptyBagOStuff() : $wgMemc;
  58         }
  59
  60         protected function supportedOrders() {
  61                 return array( 'random', 'timestamp', 'fifo' );
  62         }
  63
  64         protected function optimalOrder() {
  65                 return 'random';
  66         }
  67
  68         /**
  69          * @see JobQueue::doIsEmpty()
  70          * @return bool
  71          */
  72         protected function doIsEmpty() {
  73                 $dbr = $this->getSlaveDB();
  74                 try {
  75                         $found = $dbr->selectField( // unclaimed job
  76                                 'job', '1', array( 'job_cmd' => $this->type, 'job_token' => '' ), __METHOD__
  77                         );
  78                 } catch ( DBError $e ) {
  79                         $this->throwDBException( $e );
  80                 }
  81
  82                 return !$found;
  83         }
  84
  85         /**
  86          * @see JobQueue::doGetSize()
  87          * @return int
  88          */
  89         protected function doGetSize() {
  90                 $key = $this->getCacheKey( 'size' );
  91
  92                 $size = $this->cache->get( $key );
  93                 if ( is_int( $size ) ) {
  94                         return $size;
  95                 }
  96
  97                 try {
  98                         $dbr = $this->getSlaveDB();
  99                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 100                                 array( 'job_cmd' => $this->type, 'job_token' => '' ),
 101                                 __METHOD__
 102                         );
 103                 } catch ( DBError $e ) {
 104                         $this->throwDBException( $e );
 105                 }
 106                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 107
 108                 return $size;
 109         }
 110
 111         /**
 112          * @see JobQueue::doGetAcquiredCount()
 113          * @return int
 114          */
 115         protected function doGetAcquiredCount() {
 116                 if ( $this->claimTTL <= 0 ) {
 117                         return 0; // no acknowledgements
 118                 }
 119
 120                 $key = $this->getCacheKey( 'acquiredcount' );
 121
 122                 $count = $this->cache->get( $key );
 123                 if ( is_int( $count ) ) {
 124                         return $count;
 125                 }
 126
 127                 $dbr = $this->getSlaveDB();
 128                 try {
 129                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 130                                 array( 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ),
 131                                 __METHOD__
 132                         );
 133                 } catch ( DBError $e ) {
 134                         $this->throwDBException( $e );
 135                 }
 136                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 137
 138                 return $count;
 139         }
 140
 141         /**
 142          * @see JobQueue::doGetAbandonedCount()
 143          * @return int
 144          * @throws MWException
 145          */
 146         protected function doGetAbandonedCount() {
 147                 if ( $this->claimTTL <= 0 ) {
 148                         return 0; // no acknowledgements
 149                 }
 150
 151                 $key = $this->getCacheKey( 'abandonedcount' );
 152
 153                 $count = $this->cache->get( $key );
 154                 if ( is_int( $count ) ) {
 155                         return $count;
 156                 }
 157
 158                 $dbr = $this->getSlaveDB();
 159                 try {
 160                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 161                                 array(
 162                                         'job_cmd' => $this->type,
 163                                         "job_token != {$dbr->addQuotes( '' )}",
 164                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 165                                 ),
 166                                 __METHOD__
 167                         );
 168                 } catch ( DBError $e ) {
 169                         $this->throwDBException( $e );
 170                 }
 171
 172                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 173
 174                 return $count;
 175         }
 176
 177         /**
 178          * @see JobQueue::doBatchPush()
 179          * @param IJobSpecification[] $jobs
 180          * @param int $flags
 181          * @throws DBError|Exception
 182          * @return void
 183          */
 184         protected function doBatchPush( array $jobs, $flags ) {
 185                 $dbw = $this->getMasterDB();
 186
 187                 $that = $this;
 188                 $method = __METHOD__;
 189                 $dbw->onTransactionIdle(
 190                         function () use ( $dbw, $that, $jobs, $flags, $method ) {
 191                                 $that->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 192                         }
 193                 );
 194         }
 195
 196         /**
 197          * This function should *not* be called outside of JobQueueDB
 198          *
 199          * @param IDatabase $dbw
 200          * @param IJobSpecification[] $jobs
 201          * @param int $flags
 202          * @param string $method
 203          * @throws DBError
 204          * @return void
 205          */
 206         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 207                 if ( !count( $jobs ) ) {
 208                         return;
 209                 }
 210
 211                 $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
 212                 $rowList = array(); // list of jobs for jobs that are not de-duplicated
 213                 foreach ( $jobs as $job ) {
 214                         $row = $this->insertFields( $job );
 215                         if ( $job->ignoreDuplicates() ) {
 216                                 $rowSet[$row['job_sha1']] = $row;
 217                         } else {
 218                                 $rowList[] = $row;
 219                         }
 220                 }
 221
 222                 if ( $flags & self::QOS_ATOMIC ) {
 223                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 224                 }
 225                 try {
 226                         // Strip out any duplicate jobs that are already in the queue...
 227                         if ( count( $rowSet ) ) {
 228                                 $res = $dbw->select( 'job', 'job_sha1',
 229                                         array(
 230                                                 // No job_type condition since it's part of the job_sha1 hash
 231                                                 'job_sha1' => array_keys( $rowSet ),
 232                                                 'job_token' => '' // unclaimed
 233                                         ),
 234                                         $method
 235                                 );
 236                                 foreach ( $res as $row ) {
 237                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 238                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 239                                 }
 240                         }
 241                         // Build the full list of job rows to insert
 242                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 243                         // Insert the job rows in chunks to avoid slave lag...
 244                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 245                                 $dbw->insert( 'job', $rowBatch, $method );
 246                         }
 247                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 248                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 249                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 250                         );
 251                 } catch ( DBError $e ) {
 252                         if ( $flags & self::QOS_ATOMIC ) {
 253                                 $dbw->rollback( $method );
 254                         }
 255                         throw $e;
 256                 }
 257                 if ( $flags & self::QOS_ATOMIC ) {
 258                         $dbw->endAtomic( $method );
 259                 }
 260
 261                 return;
 262         }
 263
 264         /**
 265          * @see JobQueue::doPop()
 266          * @return Job|bool
 267          */
 268         protected function doPop() {
 269                 $dbw = $this->getMasterDB();
 270                 try {
 271                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 272                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 273                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 274                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 275                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 276                         } );
 277
 278                         $uuid = wfRandomString( 32 ); // pop attempt
 279                         $job = false; // job popped off
 280                         do { // retry when our row is invalid or deleted as a duplicate
 281                                 // Try to reserve a row in the DB...
 282                                 if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 283                                         $row = $this->claimOldest( $uuid );
 284                                 } else { // random first
 285                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 286                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 287                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 288                                 }
 289                                 // Check if we found a row to reserve...
 290                                 if ( !$row ) {
 291                                         break; // nothing to do
 292                                 }
 293                                 JobQueue::incrStats( 'pops', $this->type );
 294                                 // Get the job object from the row...
 295                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 296                                 $job = Job::factory( $row->job_cmd, $title,
 297                                         self::extractBlob( $row->job_params ), $row->job_id );
 298                                 $job->metadata['id'] = $row->job_id;
 299                                 $job->metadata['timestamp'] = $row->job_timestamp;
 300                                 break; // done
 301                         } while ( true );
 302
 303                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 304                                 // Handled jobs that need to be recycled/deleted;
 305                                 // any recycled jobs will be picked up next attempt
 306                                 $this->recycleAndDeleteStaleJobs();
 307                         }
 308                 } catch ( DBError $e ) {
 309                         $this->throwDBException( $e );
 310                 }
 311
 312                 return $job;
 313         }
 314
 315         /**
 316          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 317          *
 318          * @param string $uuid 32 char hex string
 319          * @param int $rand Random unsigned integer (31 bits)
 320          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 321          * @return stdClass|bool Row|false
 322          */
 323         protected function claimRandom( $uuid, $rand, $gte ) {
 324                 $dbw = $this->getMasterDB();
 325                 // Check cache to see if the queue has <= OFFSET items
 326                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 327
 328                 $row = false; // the row acquired
 329                 $invertedDirection = false; // whether one job_random direction was already scanned
 330                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 331                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 332                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 333                 // be used here with MySQL.
 334                 do {
 335                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 336                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 337                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 338                                 $ineq = $gte ? '>=' : '<=';
 339                                 $dir = $gte ? 'ASC' : 'DESC';
 340                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 341                                         array(
 342                                                 'job_cmd' => $this->type,
 343                                                 'job_token' => '', // unclaimed
 344                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 345                                         __METHOD__,
 346                                         array( 'ORDER BY' => "job_random {$dir}" )
 347                                 );
 348                                 if ( !$row && !$invertedDirection ) {
 349                                         $gte = !$gte;
 350                                         $invertedDirection = true;
 351                                         continue; // try the other direction
 352                                 }
 353                         } else { // table *may* have >= MAX_OFFSET rows
 354                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 355                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 356                                 // instead of job_random for reducing excess claim retries.
 357                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 358                                         array(
 359                                                 'job_cmd' => $this->type,
 360                                                 'job_token' => '', // unclaimed
 361                                         ),
 362                                         __METHOD__,
 363                                         array( 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) )
 364                                 );
 365                                 if ( !$row ) {
 366                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 367                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 368                                         continue; // use job_random
 369                                 }
 370                         }
 371
 372                         if ( $row ) { // claim the job
 373                                 $dbw->update( 'job', // update by PK
 374                                         array(
 375                                                 'job_token' => $uuid,
 376                                                 'job_token_timestamp' => $dbw->timestamp(),
 377                                                 'job_attempts = job_attempts+1' ),
 378                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 379                                         __METHOD__
 380                                 );
 381                                 // This might get raced out by another runner when claiming the previously
 382                                 // selected row. The use of job_random should minimize this problem, however.
 383                                 if ( !$dbw->affectedRows() ) {
 384                                         $row = false; // raced out
 385                                 }
 386                         } else {
 387                                 break; // nothing to do
 388                         }
 389                 } while ( !$row );
 390
 391                 return $row;
 392         }
 393
 394         /**
 395          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 396          *
 397          * @param string $uuid 32 char hex string
 398          * @return stdClass|bool Row|false
 399          */
 400         protected function claimOldest( $uuid ) {
 401                 $dbw = $this->getMasterDB();
 402
 403                 $row = false; // the row acquired
 404                 do {
 405                         if ( $dbw->getType() === 'mysql' ) {
 406                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 407                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 408                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 409                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 410                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 411                                         "SET " .
 412                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 413                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 414                                                 "job_attempts = job_attempts+1 " .
 415                                         "WHERE ( " .
 416                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 417                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 418                                         ") ORDER BY job_id ASC LIMIT 1",
 419                                         __METHOD__
 420                                 );
 421                         } else {
 422                                 // Use a subquery to find the job, within an UPDATE to claim it.
 423                                 // This uses as much of the DB wrapper functions as possible.
 424                                 $dbw->update( 'job',
 425                                         array(
 426                                                 'job_token' => $uuid,
 427                                                 'job_token_timestamp' => $dbw->timestamp(),
 428                                                 'job_attempts = job_attempts+1' ),
 429                                         array( 'job_id = (' .
 430                                                 $dbw->selectSQLText( 'job', 'job_id',
 431                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 432                                                         __METHOD__,
 433                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 434                                                 ')'
 435                                         ),
 436                                         __METHOD__
 437                                 );
 438                         }
 439                         // Fetch any row that we just reserved...
 440                         if ( $dbw->affectedRows() ) {
 441                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 442                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 443                                 );
 444                                 if ( !$row ) { // raced out by duplicate job removal
 445                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 446                                 }
 447                         } else {
 448                                 break; // nothing to do
 449                         }
 450                 } while ( !$row );
 451
 452                 return $row;
 453         }
 454
 455         /**
 456          * @see JobQueue::doAck()
 457          * @param Job $job
 458          * @throws MWException
 459          * @return Job|bool
 460          */
 461         protected function doAck( Job $job ) {
 462                 if ( !isset( $job->metadata['id'] ) ) {
 463                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 464                 }
 465
 466                 $dbw = $this->getMasterDB();
 467                 try {
 468                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 469                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 470                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 471                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 472                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 473                         } );
 474
 475                         // Delete a row with a single DELETE without holding row locks over RTTs...
 476                         $dbw->delete( 'job',
 477                                 array( 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ), __METHOD__ );
 478
 479                         JobQueue::incrStats( 'acks', $this->type );
 480                 } catch ( DBError $e ) {
 481                         $this->throwDBException( $e );
 482                 }
 483
 484                 return true;
 485         }
 486
 487         /**
 488          * @see JobQueue::doDeduplicateRootJob()
 489          * @param IJobSpecification $job
 490          * @throws MWException
 491          * @return bool
 492          */
 493         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 494                 $params = $job->getParams();
 495                 if ( !isset( $params['rootJobSignature'] ) ) {
 496                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 497                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 498                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 499                 }
 500                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 501                 // Callers should call batchInsert() and then this function so that if the insert
 502                 // fails, the de-duplication registration will be aborted. Since the insert is
 503                 // deferred till "transaction idle", do the same here, so that the ordering is
 504                 // maintained. Having only the de-duplication registration succeed would cause
 505                 // jobs to become no-ops without any actual jobs that made them redundant.
 506                 $dbw = $this->getMasterDB();
 507                 $cache = $this->dupCache;
 508                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 509                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 510                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 511                                 return true; // a newer version of this root job was enqueued
 512                         }
 513
 514                         // Update the timestamp of the last root job started at the location...
 515                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 516                 } );
 517
 518                 return true;
 519         }
 520
 521         /**
 522          * @see JobQueue::doDelete()
 523          * @return bool
 524          */
 525         protected function doDelete() {
 526                 $dbw = $this->getMasterDB();
 527                 try {
 528                         $dbw->delete( 'job', array( 'job_cmd' => $this->type ) );
 529                 } catch ( DBError $e ) {
 530                         $this->throwDBException( $e );
 531                 }
 532
 533                 return true;
 534         }
 535
 536         /**
 537          * @see JobQueue::doWaitForBackups()
 538          * @return void
 539          */
 540         protected function doWaitForBackups() {
 541                 wfWaitForSlaves( false, $this->wiki, $this->cluster ?: false );
 542         }
 543
 544         /**
 545          * @return void
 546          */
 547         protected function doFlushCaches() {
 548                 foreach ( array( 'size', 'acquiredcount' ) as $type ) {
 549                         $this->cache->delete( $this->getCacheKey( $type ) );
 550                 }
 551         }
 552
 553         /**
 554          * @see JobQueue::getAllQueuedJobs()
 555          * @return Iterator
 556          */
 557         public function getAllQueuedJobs() {
 558                 return $this->getJobIterator( array( 'job_cmd' => $this->getType(), 'job_token' => '' ) );
 559         }
 560
 561         /**
 562          * @see JobQueue::getAllAcquiredJobs()
 563          * @return Iterator
 564          */
 565         public function getAllAcquiredJobs() {
 566                 return $this->getJobIterator( array( 'job_cmd' => $this->getType(), "job_token > ''" ) );
 567         }
 568
 569         /**
 570          * @param array $conds Query conditions
 571          * @return Iterator
 572          */
 573         protected function getJobIterator( array $conds ) {
 574                 $dbr = $this->getSlaveDB();
 575                 try {
 576                         return new MappedIterator(
 577                                 $dbr->select( 'job', self::selectFields(), $conds ),
 578                                 function ( $row ) {
 579                                         $job = Job::factory(
 580                                                 $row->job_cmd,
 581                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 582                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : array()
 583                                         );
 584                                         $job->metadata['id'] = $row->job_id;
 585                                         $job->metadata['timestamp'] = $row->job_timestamp;
 586
 587                                         return $job;
 588                                 }
 589                         );
 590                 } catch ( DBError $e ) {
 591                         $this->throwDBException( $e );
 592                 }
 593         }
 594
 595         public function getCoalesceLocationInternal() {
 596                 return $this->cluster
 597                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 598                         : "LBFactory:{$this->wiki}";
 599         }
 600
 601         protected function doGetSiblingQueuesWithJobs( array $types ) {
 602                 $dbr = $this->getSlaveDB();
 603                 // @note: this does not check whether the jobs are claimed or not.
 604                 // This is useful so JobQueueGroup::pop() also sees queues that only
 605                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 606                 // failed jobs so that they can be popped again for that edge case.
 607                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 608                         array( 'job_cmd' => $types ), __METHOD__ );
 609
 610                 $types = array();
 611                 foreach ( $res as $row ) {
 612                         $types[] = $row->job_cmd;
 613                 }
 614
 615                 return $types;
 616         }
 617
 618         protected function doGetSiblingQueueSizes( array $types ) {
 619                 $dbr = $this->getSlaveDB();
 620                 $res = $dbr->select( 'job', array( 'job_cmd', 'COUNT(*) AS count' ),
 621                         array( 'job_cmd' => $types ), __METHOD__, array( 'GROUP BY' => 'job_cmd' ) );
 622
 623                 $sizes = array();
 624                 foreach ( $res as $row ) {
 625                         $sizes[$row->job_cmd] = (int)$row->count;
 626                 }
 627
 628                 return $sizes;
 629         }
 630
 631         /**
 632          * Recycle or destroy any jobs that have been claimed for too long
 633          *
 634          * @return int Number of jobs recycled/deleted
 635          */
 636         public function recycleAndDeleteStaleJobs() {
 637                 $now = time();
 638                 $count = 0; // affected rows
 639                 $dbw = $this->getMasterDB();
 640
 641                 try {
 642                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 643                                 return $count; // already in progress
 644                         }
 645
 646                         // Remove claims on jobs acquired for too long if enabled...
 647                         if ( $this->claimTTL > 0 ) {
 648                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 649                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 650                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 651                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 652                                 $res = $dbw->select( 'job', 'job_id',
 653                                         array(
 654                                                 'job_cmd' => $this->type,
 655                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 656                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 657                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ), // retries left
 658                                         __METHOD__
 659                                 );
 660                                 $ids = array_map(
 661                                         function ( $o ) {
 662                                                 return $o->job_id;
 663                                         }, iterator_to_array( $res )
 664                                 );
 665                                 if ( count( $ids ) ) {
 666                                         // Reset job_token for these jobs so that other runners will pick them up.
 667                                         // Set the timestamp to the current time, as it is useful to now that the job
 668                                         // was already tried before (the timestamp becomes the "released" time).
 669                                         $dbw->update( 'job',
 670                                                 array(
 671                                                         'job_token' => '',
 672                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 673                                                 array(
 674                                                         'job_id' => $ids ),
 675                                                 __METHOD__
 676                                         );
 677                                         $affected = $dbw->affectedRows();
 678                                         $count += $affected;
 679                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 680                                         $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 681                                 }
 682                         }
 683
 684                         // Just destroy any stale jobs...
 685                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 686                         $conds = array(
 687                                 'job_cmd' => $this->type,
 688                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 689                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 690                         );
 691                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 692                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 693                         }
 694                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 695                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 696                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 697                         $ids = array_map(
 698                                 function ( $o ) {
 699                                         return $o->job_id;
 700                                 }, iterator_to_array( $res )
 701                         );
 702                         if ( count( $ids ) ) {
 703                                 $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 704                                 $affected = $dbw->affectedRows();
 705                                 $count += $affected;
 706                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 707                         }
 708
 709                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 710                 } catch ( DBError $e ) {
 711                         $this->throwDBException( $e );
 712                 }
 713
 714                 return $count;
 715         }
 716
 717         /**
 718          * @param IJobSpecification $job
 719          * @return array
 720          */
 721         protected function insertFields( IJobSpecification $job ) {
 722                 $dbw = $this->getMasterDB();
 723
 724                 return array(
 725                         // Fields that describe the nature of the job
 726                         'job_cmd' => $job->getType(),
 727                         'job_namespace' => $job->getTitle()->getNamespace(),
 728                         'job_title' => $job->getTitle()->getDBkey(),
 729                         'job_params' => self::makeBlob( $job->getParams() ),
 730                         // Additional job metadata
 731                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 732                         'job_timestamp' => $dbw->timestamp(),
 733                         'job_sha1' => wfBaseConvert(
 734                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 735                                 16, 36, 31
 736                         ),
 737                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 738                 );
 739         }
 740
 741         /**
 742          * @throws JobQueueConnectionError
 743          * @return DBConnRef
 744          */
 745         protected function getSlaveDB() {
 746                 try {
 747                         return $this->getDB( DB_SLAVE );
 748                 } catch ( DBConnectionError $e ) {
 749                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 750                 }
 751         }
 752
 753         /**
 754          * @throws JobQueueConnectionError
 755          * @return DBConnRef
 756          */
 757         protected function getMasterDB() {
 758                 try {
 759                         return $this->getDB( DB_MASTER );
 760                 } catch ( DBConnectionError $e ) {
 761                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 762                 }
 763         }
 764
 765         /**
 766          * @param int $index (DB_SLAVE/DB_MASTER)
 767          * @return DBConnRef
 768          */
 769         protected function getDB( $index ) {
 770                 $lb = ( $this->cluster !== false )
 771                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 772                         : wfGetLB( $this->wiki );
 773
 774                 return $lb->getConnectionRef( $index, array(), $this->wiki );
 775         }
 776
 777         /**
 778          * @param string $property
 779          * @return string
 780          */
 781         private function getCacheKey( $property ) {
 782                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 783                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 784
 785                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 786         }
 787
 788         /**
 789          * @param array|bool $params
 790          * @return string
 791          */
 792         protected static function makeBlob( $params ) {
 793                 if ( $params !== false ) {
 794                         return serialize( $params );
 795                 } else {
 796                         return '';
 797                 }
 798         }
 799
 800         /**
 801          * @param string $blob
 802          * @return bool|mixed
 803          */
 804         protected static function extractBlob( $blob ) {
 805                 if ( (string)$blob !== '' ) {
 806                         return unserialize( $blob );
 807                 } else {
 808                         return false;
 809                 }
 810         }
 811
 812         /**
 813          * @param DBError $e
 814          * @throws JobQueueError
 815          */
 816         protected function throwDBException( DBError $e ) {
 817                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 818         }
 819
 820         /**
 821          * Return the list of job fields that should be selected.
 822          * @since 1.23
 823          * @return array
 824          */
 825         public static function selectFields() {
 826                 return array(
 827                         'job_id',
 828                         'job_cmd',
 829                         'job_namespace',
 830                         'job_title',
 831                         'job_timestamp',
 832                         'job_params',
 833                         'job_random',
 834                         'job_attempts',
 835                         'job_token',
 836                         'job_token_timestamp',
 837                         'job_sha1',
 838                 );
 839         }
 840 }