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         protected $cluster = false; // string; name of an external DB cluster
  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 $params array
  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                 $key = $this->getCacheKey( 'empty' );
  74
  75                 $isEmpty = $this->cache->get( $key );
  76                 if ( $isEmpty === 'true' ) {
  77                         return true;
  78                 } elseif ( $isEmpty === 'false' ) {
  79                         return false;
  80                 }
  81
  82                 $dbr = $this->getSlaveDB();
  83                 try {
  84                         $found = $dbr->selectField( // unclaimed job
  85                                 'job', '1', array( 'job_cmd' => $this->type, 'job_token' => '' ), __METHOD__
  86                         );
  87                 } catch ( DBError $e ) {
  88                         $this->throwDBException( $e );
  89                 }
  90                 $this->cache->add( $key, $found ? 'false' : 'true', self::CACHE_TTL_LONG );
  91
  92                 return !$found;
  93         }
  94
  95         /**
  96          * @see JobQueue::doGetSize()
  97          * @return integer
  98          */
  99         protected function doGetSize() {
 100                 $key = $this->getCacheKey( 'size' );
 101
 102                 $size = $this->cache->get( $key );
 103                 if ( is_int( $size ) ) {
 104                         return $size;
 105                 }
 106
 107                 try {
 108                         $dbr = $this->getSlaveDB();
 109                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 110                                 array( 'job_cmd' => $this->type, 'job_token' => '' ),
 111                                 __METHOD__
 112                         );
 113                 } catch ( DBError $e ) {
 114                         $this->throwDBException( $e );
 115                 }
 116                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 117
 118                 return $size;
 119         }
 120
 121         /**
 122          * @see JobQueue::doGetAcquiredCount()
 123          * @return integer
 124          */
 125         protected function doGetAcquiredCount() {
 126                 if ( $this->claimTTL <= 0 ) {
 127                         return 0; // no acknowledgements
 128                 }
 129
 130                 $key = $this->getCacheKey( 'acquiredcount' );
 131
 132                 $count = $this->cache->get( $key );
 133                 if ( is_int( $count ) ) {
 134                         return $count;
 135                 }
 136
 137                 $dbr = $this->getSlaveDB();
 138                 try {
 139                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 140                                 array( 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ),
 141                                 __METHOD__
 142                         );
 143                 } catch ( DBError $e ) {
 144                         $this->throwDBException( $e );
 145                 }
 146                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 147
 148                 return $count;
 149         }
 150
 151         /**
 152          * @see JobQueue::doGetAbandonedCount()
 153          * @return integer
 154          * @throws MWException
 155          */
 156         protected function doGetAbandonedCount() {
 157                 global $wgMemc;
 158
 159                 if ( $this->claimTTL <= 0 ) {
 160                         return 0; // no acknowledgements
 161                 }
 162
 163                 $key = $this->getCacheKey( 'abandonedcount' );
 164
 165                 $count = $wgMemc->get( $key );
 166                 if ( is_int( $count ) ) {
 167                         return $count;
 168                 }
 169
 170                 $dbr = $this->getSlaveDB();
 171                 try {
 172                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 173                                 array(
 174                                         'job_cmd' => $this->type,
 175                                         "job_token != {$dbr->addQuotes( '' )}",
 176                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 177                                 ),
 178                                 __METHOD__
 179                         );
 180                 } catch ( DBError $e ) {
 181                         $this->throwDBException( $e );
 182                 }
 183                 $wgMemc->set( $key, $count, self::CACHE_TTL_SHORT );
 184
 185                 return $count;
 186         }
 187
 188         /**
 189          * @see JobQueue::doBatchPush()
 190          * @param array $jobs
 191          * @param $flags
 192          * @throws DBError|Exception
 193          * @return bool
 194          */
 195         protected function doBatchPush( array $jobs, $flags ) {
 196                 $dbw = $this->getMasterDB();
 197
 198                 $that = $this;
 199                 $method = __METHOD__;
 200                 $dbw->onTransactionIdle(
 201                         function () use ( $dbw, $that, $jobs, $flags, $method ) {
 202                                 $that->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 203                         }
 204                 );
 205
 206                 return true;
 207         }
 208
 209         /**
 210          * This function should *not* be called outside of JobQueueDB
 211          *
 212          * @param DatabaseBase $dbw
 213          * @param array $jobs
 214          * @param int $flags
 215          * @param string $method
 216          * @return boolean
 217          * @throws type
 218          */
 219         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 220                 if ( !count( $jobs ) ) {
 221                         return true;
 222                 }
 223
 224                 $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
 225                 $rowList = array(); // list of jobs for jobs that are are not de-duplicated
 226                 foreach ( $jobs as $job ) {
 227                         $row = $this->insertFields( $job );
 228                         if ( $job->ignoreDuplicates() ) {
 229                                 $rowSet[$row['job_sha1']] = $row;
 230                         } else {
 231                                 $rowList[] = $row;
 232                         }
 233                 }
 234
 235                 if ( $flags & self::QOS_ATOMIC ) {
 236                         $dbw->begin( $method ); // wrap all the job additions in one transaction
 237                 }
 238                 try {
 239                         // Strip out any duplicate jobs that are already in the queue...
 240                         if ( count( $rowSet ) ) {
 241                                 $res = $dbw->select( 'job', 'job_sha1',
 242                                         array(
 243                                                 // No job_type condition since it's part of the job_sha1 hash
 244                                                 'job_sha1' => array_keys( $rowSet ),
 245                                                 'job_token' => '' // unclaimed
 246                                         ),
 247                                         $method
 248                                 );
 249                                 foreach ( $res as $row ) {
 250                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 251                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 252                                 }
 253                         }
 254                         // Build the full list of job rows to insert
 255                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 256                         // Insert the job rows in chunks to avoid slave lag...
 257                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 258                                 $dbw->insert( 'job', $rowBatch, $method );
 259                         }
 260                         JobQueue::incrStats( 'job-insert', $this->type, count( $rows ) );
 261                         JobQueue::incrStats( 'job-insert-duplicate', $this->type,
 262                                 count( $rowSet ) + count( $rowList ) - count( $rows ) );
 263                 } catch ( DBError $e ) {
 264                         if ( $flags & self::QOS_ATOMIC ) {
 265                                 $dbw->rollback( $method );
 266                         }
 267                         throw $e;
 268                 }
 269                 if ( $flags & self::QOS_ATOMIC ) {
 270                         $dbw->commit( $method );
 271                 }
 272
 273                 $this->cache->set( $this->getCacheKey( 'empty' ), 'false', JobQueueDB::CACHE_TTL_LONG );
 274
 275                 return true;
 276         }
 277
 278         /**
 279          * @see JobQueue::doPop()
 280          * @return Job|bool
 281          */
 282         protected function doPop() {
 283                 if ( $this->cache->get( $this->getCacheKey( 'empty' ) ) === 'true' ) {
 284                         return false; // queue is empty
 285                 }
 286
 287                 $dbw = $this->getMasterDB();
 288                 try {
 289                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 290                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 291                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 292                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 293                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 294                         } );
 295
 296                         $uuid = wfRandomString( 32 ); // pop attempt
 297                         $job = false; // job popped off
 298                         do { // retry when our row is invalid or deleted as a duplicate
 299                                 // Try to reserve a row in the DB...
 300                                 if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 301                                         $row = $this->claimOldest( $uuid );
 302                                 } else { // random first
 303                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 304                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 305                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 306                                 }
 307                                 // Check if we found a row to reserve...
 308                                 if ( !$row ) {
 309                                         $this->cache->set( $this->getCacheKey( 'empty' ), 'true', self::CACHE_TTL_LONG );
 310                                         break; // nothing to do
 311                                 }
 312                                 JobQueue::incrStats( 'job-pop', $this->type );
 313                                 // Get the job object from the row...
 314                                 $title = Title::makeTitleSafe( $row->job_namespace, $row->job_title );
 315                                 if ( !$title ) {
 316                                         $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 317                                         wfDebug( "Row has invalid title '{$row->job_title}'." );
 318                                         continue; // try again
 319                                 }
 320                                 $job = Job::factory( $row->job_cmd, $title,
 321                                         self::extractBlob( $row->job_params ), $row->job_id );
 322                                 $job->metadata['id'] = $row->job_id;
 323                                 $job->id = $row->job_id; // XXX: work around broken subclasses
 324                                 break; // done
 325                         } while ( true );
 326                 } catch ( DBError $e ) {
 327                         $this->throwDBException( $e );
 328                 }
 329
 330                 return $job;
 331         }
 332
 333         /**
 334          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 335          *
 336          * @param string $uuid 32 char hex string
 337          * @param $rand integer Random unsigned integer (31 bits)
 338          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 339          * @return Row|false
 340          */
 341         protected function claimRandom( $uuid, $rand, $gte ) {
 342                 $dbw = $this->getMasterDB();
 343                 // Check cache to see if the queue has <= OFFSET items
 344                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 345
 346                 $row = false; // the row acquired
 347                 $invertedDirection = false; // whether one job_random direction was already scanned
 348                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 349                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 350                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 351                 // be used here with MySQL.
 352                 do {
 353                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 354                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 355                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 356                                 $ineq = $gte ? '>=' : '<=';
 357                                 $dir = $gte ? 'ASC' : 'DESC';
 358                                 $row = $dbw->selectRow( 'job', '*', // find a random job
 359                                         array(
 360                                                 'job_cmd' => $this->type,
 361                                                 'job_token' => '', // unclaimed
 362                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 363                                         __METHOD__,
 364                                         array( 'ORDER BY' => "job_random {$dir}" )
 365                                 );
 366                                 if ( !$row && !$invertedDirection ) {
 367                                         $gte = !$gte;
 368                                         $invertedDirection = true;
 369                                         continue; // try the other direction
 370                                 }
 371                         } else { // table *may* have >= MAX_OFFSET rows
 372                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 373                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 374                                 // instead of job_random for reducing excess claim retries.
 375                                 $row = $dbw->selectRow( 'job', '*', // find a random job
 376                                         array(
 377                                                 'job_cmd' => $this->type,
 378                                                 'job_token' => '', // unclaimed
 379                                         ),
 380                                         __METHOD__,
 381                                         array( 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) )
 382                                 );
 383                                 if ( !$row ) {
 384                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 385                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 386                                         continue; // use job_random
 387                                 }
 388                         }
 389                         if ( $row ) { // claim the job
 390                                 $dbw->update( 'job', // update by PK
 391                                         array(
 392                                                 'job_token' => $uuid,
 393                                                 'job_token_timestamp' => $dbw->timestamp(),
 394                                                 'job_attempts = job_attempts+1' ),
 395                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 396                                         __METHOD__
 397                                 );
 398                                 // This might get raced out by another runner when claiming the previously
 399                                 // selected row. The use of job_random should minimize this problem, however.
 400                                 if ( !$dbw->affectedRows() ) {
 401                                         $row = false; // raced out
 402                                 }
 403                         } else {
 404                                 break; // nothing to do
 405                         }
 406                 } while ( !$row );
 407
 408                 return $row;
 409         }
 410
 411         /**
 412          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 413          *
 414          * @param string $uuid 32 char hex string
 415          * @return Row|false
 416          */
 417         protected function claimOldest( $uuid ) {
 418                 $dbw = $this->getMasterDB();
 419
 420                 $row = false; // the row acquired
 421                 do {
 422                         if ( $dbw->getType() === 'mysql' ) {
 423                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 424                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 425                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 426                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 427                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 428                                         "SET " .
 429                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 430                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 431                                                 "job_attempts = job_attempts+1 " .
 432                                         "WHERE ( " .
 433                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 434                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 435                                         ") ORDER BY job_id ASC LIMIT 1",
 436                                         __METHOD__
 437                                 );
 438                         } else {
 439                                 // Use a subquery to find the job, within an UPDATE to claim it.
 440                                 // This uses as much of the DB wrapper functions as possible.
 441                                 $dbw->update( 'job',
 442                                         array(
 443                                                 'job_token' => $uuid,
 444                                                 'job_token_timestamp' => $dbw->timestamp(),
 445                                                 'job_attempts = job_attempts+1' ),
 446                                         array( 'job_id = (' .
 447                                                 $dbw->selectSQLText( 'job', 'job_id',
 448                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 449                                                         __METHOD__,
 450                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 451                                                 ')'
 452                                         ),
 453                                         __METHOD__
 454                                 );
 455                         }
 456                         // Fetch any row that we just reserved...
 457                         if ( $dbw->affectedRows() ) {
 458                                 $row = $dbw->selectRow( 'job', '*',
 459                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 460                                 );
 461                                 if ( !$row ) { // raced out by duplicate job removal
 462                                         wfDebug( "Row deleted as duplicate by another process." );
 463                                 }
 464                         } else {
 465                                 break; // nothing to do
 466                         }
 467                 } while ( !$row );
 468
 469                 return $row;
 470         }
 471
 472         /**
 473          * @see JobQueue::doAck()
 474          * @param Job $job
 475          * @throws MWException
 476          * @return Job|bool
 477          */
 478         protected function doAck( Job $job ) {
 479                 if ( !isset( $job->metadata['id'] ) ) {
 480                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 481                 }
 482
 483                 $dbw = $this->getMasterDB();
 484                 try {
 485                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 486                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 487                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 488                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 489                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 490                         } );
 491
 492                         // Delete a row with a single DELETE without holding row locks over RTTs...
 493                         $dbw->delete( 'job',
 494                                 array( 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ), __METHOD__ );
 495                 } catch ( DBError $e ) {
 496                         $this->throwDBException( $e );
 497                 }
 498
 499                 return true;
 500         }
 501
 502         /**
 503          * @see JobQueue::doDeduplicateRootJob()
 504          * @param Job $job
 505          * @throws MWException
 506          * @return bool
 507          */
 508         protected function doDeduplicateRootJob( Job $job ) {
 509                 $params = $job->getParams();
 510                 if ( !isset( $params['rootJobSignature'] ) ) {
 511                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 512                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 513                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 514                 }
 515                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 516                 // Callers should call batchInsert() and then this function so that if the insert
 517                 // fails, the de-duplication registration will be aborted. Since the insert is
 518                 // deferred till "transaction idle", do the same here, so that the ordering is
 519                 // maintained. Having only the de-duplication registration succeed would cause
 520                 // jobs to become no-ops without any actual jobs that made them redundant.
 521                 $dbw = $this->getMasterDB();
 522                 $cache = $this->dupCache;
 523                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 524                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 525                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 526                                 return true; // a newer version of this root job was enqueued
 527                         }
 528
 529                         // Update the timestamp of the last root job started at the location...
 530                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 531                 } );
 532
 533                 return true;
 534         }
 535
 536         /**
 537          * @see JobQueue::doDelete()
 538          * @return bool
 539          */
 540         protected function doDelete() {
 541                 $dbw = $this->getMasterDB();
 542                 try {
 543                         $dbw->delete( 'job', array( 'job_cmd' => $this->type ) );
 544                 } catch ( DBError $e ) {
 545                         $this->throwDBException( $e );
 546                 }
 547
 548                 return true;
 549         }
 550
 551         /**
 552          * @see JobQueue::doWaitForBackups()
 553          * @return void
 554          */
 555         protected function doWaitForBackups() {
 556                 wfWaitForSlaves();
 557         }
 558
 559         /**
 560          * @return Array
 561          */
 562         protected function doGetPeriodicTasks() {
 563                 return array(
 564                         'recycleAndDeleteStaleJobs' => array(
 565                                 'callback' => array( $this, 'recycleAndDeleteStaleJobs' ),
 566                                 'period' => ceil( $this->claimTTL / 2 )
 567                         )
 568                 );
 569         }
 570
 571         /**
 572          * @return void
 573          */
 574         protected function doFlushCaches() {
 575                 foreach ( array( 'empty', 'size', 'acquiredcount' ) as $type ) {
 576                         $this->cache->delete( $this->getCacheKey( $type ) );
 577                 }
 578         }
 579
 580         /**
 581          * @see JobQueue::getAllQueuedJobs()
 582          * @return Iterator
 583          */
 584         public function getAllQueuedJobs() {
 585                 $dbr = $this->getSlaveDB();
 586                 try {
 587                         return new MappedIterator(
 588                                 $dbr->select( 'job', '*',
 589                                         array( 'job_cmd' => $this->getType(), 'job_token' => '' ) ),
 590                                 function ( $row ) use ( $dbr ) {
 591                                         $job = Job::factory(
 592                                                 $row->job_cmd,
 593                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 594                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : false,
 595                                                 $row->job_id
 596                                         );
 597                                         $job->metadata['id'] = $row->job_id;
 598                                         $job->id = $row->job_id; // XXX: work around broken subclasses
 599                                         return $job;
 600                                 }
 601                         );
 602                 } catch ( DBError $e ) {
 603                         $this->throwDBException( $e );
 604                 }
 605         }
 606
 607         public function getCoalesceLocationInternal() {
 608                 return $this->cluster
 609                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 610                         : "LBFactory:{$this->wiki}";
 611         }
 612
 613         protected function doGetSiblingQueuesWithJobs( array $types ) {
 614                 $dbr = $this->getSlaveDB();
 615                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 616                         array( 'job_cmd' => $types ), __METHOD__ );
 617
 618                 $types = array();
 619                 foreach ( $res as $row ) {
 620                         $types[] = $row->job_cmd;
 621                 }
 622
 623                 return $types;
 624         }
 625
 626         protected function doGetSiblingQueueSizes( array $types ) {
 627                 $dbr = $this->getSlaveDB();
 628                 $res = $dbr->select( 'job', array( 'job_cmd', 'COUNT(*) AS count' ),
 629                         array( 'job_cmd' => $types ), __METHOD__, array( 'GROUP BY' => 'job_cmd' ) );
 630
 631                 $sizes = array();
 632                 foreach ( $res as $row ) {
 633                         $sizes[$row->job_cmd] = (int)$row->count;
 634                 }
 635
 636                 return $sizes;
 637         }
 638
 639         /**
 640          * Recycle or destroy any jobs that have been claimed for too long
 641          *
 642          * @return integer Number of jobs recycled/deleted
 643          */
 644         public function recycleAndDeleteStaleJobs() {
 645                 $now = time();
 646                 $count = 0; // affected rows
 647                 $dbw = $this->getMasterDB();
 648
 649                 try {
 650                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 651                                 return $count; // already in progress
 652                         }
 653
 654                         // Remove claims on jobs acquired for too long if enabled...
 655                         if ( $this->claimTTL > 0 ) {
 656                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 657                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 658                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 659                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 660                                 $res = $dbw->select( 'job', 'job_id',
 661                                         array(
 662                                                 'job_cmd' => $this->type,
 663                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 664                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 665                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ), // retries left
 666                                         __METHOD__
 667                                 );
 668                                 $ids = array_map(
 669                                         function ( $o ) {
 670                                                 return $o->job_id;
 671                                         }, iterator_to_array( $res )
 672                                 );
 673                                 if ( count( $ids ) ) {
 674                                         // Reset job_token for these jobs so that other runners will pick them up.
 675                                         // Set the timestamp to the current time, as it is useful to now that the job
 676                                         // was already tried before (the timestamp becomes the "released" time).
 677                                         $dbw->update( 'job',
 678                                                 array(
 679                                                         'job_token' => '',
 680                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 681                                                 array(
 682                                                         'job_id' => $ids ),
 683                                                 __METHOD__
 684                                         );
 685                                         $count += $dbw->affectedRows();
 686                                         JobQueue::incrStats( 'job-recycle', $this->type, $dbw->affectedRows() );
 687                                         $this->cache->set( $this->getCacheKey( 'empty' ), 'false', self::CACHE_TTL_LONG );
 688                                 }
 689                         }
 690
 691                         // Just destroy any stale jobs...
 692                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 693                         $conds = array(
 694                                 'job_cmd' => $this->type,
 695                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 696                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 697                         );
 698                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 699                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 700                         }
 701                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 702                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 703                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 704                         $ids = array_map(
 705                                 function ( $o ) {
 706                                         return $o->job_id;
 707                                 }, iterator_to_array( $res )
 708                         );
 709                         if ( count( $ids ) ) {
 710                                 $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 711                                 $count += $dbw->affectedRows();
 712                                 JobQueue::incrStats( 'job-abandon', $this->type, $dbw->affectedRows() );
 713                         }
 714
 715                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 716                 } catch ( DBError $e ) {
 717                         $this->throwDBException( $e );
 718                 }
 719
 720                 return $count;
 721         }
 722
 723         /**
 724          * @param $job Job
 725          * @return array
 726          */
 727         protected function insertFields( Job $job ) {
 728                 $dbw = $this->getMasterDB();
 729
 730                 return array(
 731                         // Fields that describe the nature of the job
 732                         'job_cmd' => $job->getType(),
 733                         'job_namespace' => $job->getTitle()->getNamespace(),
 734                         'job_title' => $job->getTitle()->getDBkey(),
 735                         'job_params' => self::makeBlob( $job->getParams() ),
 736                         // Additional job metadata
 737                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 738                         'job_timestamp' => $dbw->timestamp(),
 739                         'job_sha1' => wfBaseConvert(
 740                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 741                                 16, 36, 31
 742                         ),
 743                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 744                 );
 745         }
 746
 747         /**
 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          * @return DBConnRef
 760          */
 761         protected function getMasterDB() {
 762                 try {
 763                         return $this->getDB( DB_MASTER );
 764                 } catch ( DBConnectionError $e ) {
 765                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 766                 }
 767         }
 768
 769         /**
 770          * @param $index integer (DB_SLAVE/DB_MASTER)
 771          * @return DBConnRef
 772          */
 773         protected function getDB( $index ) {
 774                 $lb = ( $this->cluster !== false )
 775                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 776                         : wfGetLB( $this->wiki );
 777
 778                 return $lb->getConnectionRef( $index, array(), $this->wiki );
 779         }
 780
 781         /**
 782          * @return string
 783          */
 784         private function getCacheKey( $property ) {
 785                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 786                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 787
 788                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 789         }
 790
 791         /**
 792          * @param $params
 793          * @return string
 794          */
 795         protected static function makeBlob( $params ) {
 796                 if ( $params !== false ) {
 797                         return serialize( $params );
 798                 } else {
 799                         return '';
 800                 }
 801         }
 802
 803         /**
 804          * @param $blob
 805          * @return bool|mixed
 806          */
 807         protected static function extractBlob( $blob ) {
 808                 if ( (string)$blob !== '' ) {
 809                         return unserialize( $blob );
 810                 } else {
 811                         return false;
 812                 }
 813         }
 814
 815         /**
 816          * @param DBError $e
 817          * @throws JobQueueError
 818          */
 819         protected function throwDBException( DBError $e ) {
 820                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 821         }
 822 }