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