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 use MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * Class to handle job queues stored in the DB
  27  *
  28  * @ingroup JobQueue
  29  * @since 1.21
  30  */
  31 class JobQueueDB extends JobQueue {
  32         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  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 WANObjectCache */
  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                 parent::__construct( $params );
  53
  54                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  55                 $this->cache = ObjectCache::getMainWANInstance();
  56         }
  57
  58         protected function supportedOrders() {
  59                 return [ 'random', 'timestamp', 'fifo' ];
  60         }
  61
  62         protected function optimalOrder() {
  63                 return 'random';
  64         }
  65
  66         /**
  67          * @see JobQueue::doIsEmpty()
  68          * @return bool
  69          */
  70         protected function doIsEmpty() {
  71                 $dbr = $this->getSlaveDB();
  72                 try {
  73                         $found = $dbr->selectField( // unclaimed job
  74                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  75                         );
  76                 } catch ( DBError $e ) {
  77                         $this->throwDBException( $e );
  78                 }
  79
  80                 return !$found;
  81         }
  82
  83         /**
  84          * @see JobQueue::doGetSize()
  85          * @return int
  86          */
  87         protected function doGetSize() {
  88                 $key = $this->getCacheKey( 'size' );
  89
  90                 $size = $this->cache->get( $key );
  91                 if ( is_int( $size ) ) {
  92                         return $size;
  93                 }
  94
  95                 try {
  96                         $dbr = $this->getSlaveDB();
  97                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
  98                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
  99                                 __METHOD__
 100                         );
 101                 } catch ( DBError $e ) {
 102                         $this->throwDBException( $e );
 103                 }
 104                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 105
 106                 return $size;
 107         }
 108
 109         /**
 110          * @see JobQueue::doGetAcquiredCount()
 111          * @return int
 112          */
 113         protected function doGetAcquiredCount() {
 114                 if ( $this->claimTTL <= 0 ) {
 115                         return 0; // no acknowledgements
 116                 }
 117
 118                 $key = $this->getCacheKey( 'acquiredcount' );
 119
 120                 $count = $this->cache->get( $key );
 121                 if ( is_int( $count ) ) {
 122                         return $count;
 123                 }
 124
 125                 $dbr = $this->getSlaveDB();
 126                 try {
 127                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 128                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 129                                 __METHOD__
 130                         );
 131                 } catch ( DBError $e ) {
 132                         $this->throwDBException( $e );
 133                 }
 134                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 135
 136                 return $count;
 137         }
 138
 139         /**
 140          * @see JobQueue::doGetAbandonedCount()
 141          * @return int
 142          * @throws MWException
 143          */
 144         protected function doGetAbandonedCount() {
 145                 if ( $this->claimTTL <= 0 ) {
 146                         return 0; // no acknowledgements
 147                 }
 148
 149                 $key = $this->getCacheKey( 'abandonedcount' );
 150
 151                 $count = $this->cache->get( $key );
 152                 if ( is_int( $count ) ) {
 153                         return $count;
 154                 }
 155
 156                 $dbr = $this->getSlaveDB();
 157                 try {
 158                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 159                                 [
 160                                         'job_cmd' => $this->type,
 161                                         "job_token != {$dbr->addQuotes( '' )}",
 162                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 163                                 ],
 164                                 __METHOD__
 165                         );
 166                 } catch ( DBError $e ) {
 167                         $this->throwDBException( $e );
 168                 }
 169
 170                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 171
 172                 return $count;
 173         }
 174
 175         /**
 176          * @see JobQueue::doBatchPush()
 177          * @param IJobSpecification[] $jobs
 178          * @param int $flags
 179          * @throws DBError|Exception
 180          * @return void
 181          */
 182         protected function doBatchPush( array $jobs, $flags ) {
 183                 $dbw = $this->getMasterDB();
 184
 185                 $method = __METHOD__;
 186                 $dbw->onTransactionIdle(
 187                         function () use ( $dbw, $jobs, $flags, $method ) {
 188                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 189                         }
 190                 );
 191         }
 192
 193         /**
 194          * This function should *not* be called outside of JobQueueDB
 195          *
 196          * @param IDatabase $dbw
 197          * @param IJobSpecification[] $jobs
 198          * @param int $flags
 199          * @param string $method
 200          * @throws DBError
 201          * @return void
 202          */
 203         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 204                 if ( !count( $jobs ) ) {
 205                         return;
 206                 }
 207
 208                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 209                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 210                 foreach ( $jobs as $job ) {
 211                         $row = $this->insertFields( $job );
 212                         if ( $job->ignoreDuplicates() ) {
 213                                 $rowSet[$row['job_sha1']] = $row;
 214                         } else {
 215                                 $rowList[] = $row;
 216                         }
 217                 }
 218
 219                 if ( $flags & self::QOS_ATOMIC ) {
 220                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 221                 }
 222                 try {
 223                         // Strip out any duplicate jobs that are already in the queue...
 224                         if ( count( $rowSet ) ) {
 225                                 $res = $dbw->select( 'job', 'job_sha1',
 226                                         [
 227                                                 // No job_type condition since it's part of the job_sha1 hash
 228                                                 'job_sha1' => array_keys( $rowSet ),
 229                                                 'job_token' => '' // unclaimed
 230                                         ],
 231                                         $method
 232                                 );
 233                                 foreach ( $res as $row ) {
 234                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 235                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 236                                 }
 237                         }
 238                         // Build the full list of job rows to insert
 239                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 240                         // Insert the job rows in chunks to avoid replica DB lag...
 241                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 242                                 $dbw->insert( 'job', $rowBatch, $method );
 243                         }
 244                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 245                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 246                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 247                         );
 248                 } catch ( DBError $e ) {
 249                         $this->throwDBException( $e );
 250                 }
 251                 if ( $flags & self::QOS_ATOMIC ) {
 252                         $dbw->endAtomic( $method );
 253                 }
 254
 255                 return;
 256         }
 257
 258         /**
 259          * @see JobQueue::doPop()
 260          * @return Job|bool
 261          */
 262         protected function doPop() {
 263                 $dbw = $this->getMasterDB();
 264                 try {
 265                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 266                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 267                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 268                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 269                         } );
 270
 271                         $uuid = wfRandomString( 32 ); // pop attempt
 272                         $job = false; // job popped off
 273                         do { // retry when our row is invalid or deleted as a duplicate
 274                                 // Try to reserve a row in the DB...
 275                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 276                                         $row = $this->claimOldest( $uuid );
 277                                 } else { // random first
 278                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 279                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 280                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 281                                 }
 282                                 // Check if we found a row to reserve...
 283                                 if ( !$row ) {
 284                                         break; // nothing to do
 285                                 }
 286                                 JobQueue::incrStats( 'pops', $this->type );
 287                                 // Get the job object from the row...
 288                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 289                                 $job = Job::factory( $row->job_cmd, $title,
 290                                         self::extractBlob( $row->job_params ), $row->job_id );
 291                                 $job->metadata['id'] = $row->job_id;
 292                                 $job->metadata['timestamp'] = $row->job_timestamp;
 293                                 break; // done
 294                         } while ( true );
 295
 296                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 297                                 // Handled jobs that need to be recycled/deleted;
 298                                 // any recycled jobs will be picked up next attempt
 299                                 $this->recycleAndDeleteStaleJobs();
 300                         }
 301                 } catch ( DBError $e ) {
 302                         $this->throwDBException( $e );
 303                 }
 304
 305                 return $job;
 306         }
 307
 308         /**
 309          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 310          *
 311          * @param string $uuid 32 char hex string
 312          * @param int $rand Random unsigned integer (31 bits)
 313          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 314          * @return stdClass|bool Row|false
 315          */
 316         protected function claimRandom( $uuid, $rand, $gte ) {
 317                 $dbw = $this->getMasterDB();
 318                 // Check cache to see if the queue has <= OFFSET items
 319                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 320
 321                 $row = false; // the row acquired
 322                 $invertedDirection = false; // whether one job_random direction was already scanned
 323                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 324                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 325                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 326                 // be used here with MySQL.
 327                 do {
 328                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 329                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 330                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 331                                 $ineq = $gte ? '>=' : '<=';
 332                                 $dir = $gte ? 'ASC' : 'DESC';
 333                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 334                                         [
 335                                                 'job_cmd' => $this->type,
 336                                                 'job_token' => '', // unclaimed
 337                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 338                                         __METHOD__,
 339                                         [ 'ORDER BY' => "job_random {$dir}" ]
 340                                 );
 341                                 if ( !$row && !$invertedDirection ) {
 342                                         $gte = !$gte;
 343                                         $invertedDirection = true;
 344                                         continue; // try the other direction
 345                                 }
 346                         } else { // table *may* have >= MAX_OFFSET rows
 347                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 348                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 349                                 // instead of job_random for reducing excess claim retries.
 350                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 351                                         [
 352                                                 'job_cmd' => $this->type,
 353                                                 'job_token' => '', // unclaimed
 354                                         ],
 355                                         __METHOD__,
 356                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 357                                 );
 358                                 if ( !$row ) {
 359                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 360                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 361                                         continue; // use job_random
 362                                 }
 363                         }
 364
 365                         if ( $row ) { // claim the job
 366                                 $dbw->update( 'job', // update by PK
 367                                         [
 368                                                 'job_token' => $uuid,
 369                                                 'job_token_timestamp' => $dbw->timestamp(),
 370                                                 'job_attempts = job_attempts+1' ],
 371                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 372                                         __METHOD__
 373                                 );
 374                                 // This might get raced out by another runner when claiming the previously
 375                                 // selected row. The use of job_random should minimize this problem, however.
 376                                 if ( !$dbw->affectedRows() ) {
 377                                         $row = false; // raced out
 378                                 }
 379                         } else {
 380                                 break; // nothing to do
 381                         }
 382                 } while ( !$row );
 383
 384                 return $row;
 385         }
 386
 387         /**
 388          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 389          *
 390          * @param string $uuid 32 char hex string
 391          * @return stdClass|bool Row|false
 392          */
 393         protected function claimOldest( $uuid ) {
 394                 $dbw = $this->getMasterDB();
 395
 396                 $row = false; // the row acquired
 397                 do {
 398                         if ( $dbw->getType() === 'mysql' ) {
 399                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 400                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 401                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 402                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 403                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 404                                         "SET " .
 405                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 406                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 407                                                 "job_attempts = job_attempts+1 " .
 408                                         "WHERE ( " .
 409                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 410                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 411                                         ") ORDER BY job_id ASC LIMIT 1",
 412                                         __METHOD__
 413                                 );
 414                         } else {
 415                                 // Use a subquery to find the job, within an UPDATE to claim it.
 416                                 // This uses as much of the DB wrapper functions as possible.
 417                                 $dbw->update( 'job',
 418                                         [
 419                                                 'job_token' => $uuid,
 420                                                 'job_token_timestamp' => $dbw->timestamp(),
 421                                                 'job_attempts = job_attempts+1' ],
 422                                         [ 'job_id = (' .
 423                                                 $dbw->selectSQLText( 'job', 'job_id',
 424                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 425                                                         __METHOD__,
 426                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 427                                                 ')'
 428                                         ],
 429                                         __METHOD__
 430                                 );
 431                         }
 432                         // Fetch any row that we just reserved...
 433                         if ( $dbw->affectedRows() ) {
 434                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 435                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 436                                 );
 437                                 if ( !$row ) { // raced out by duplicate job removal
 438                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 439                                 }
 440                         } else {
 441                                 break; // nothing to do
 442                         }
 443                 } while ( !$row );
 444
 445                 return $row;
 446         }
 447
 448         /**
 449          * @see JobQueue::doAck()
 450          * @param Job $job
 451          * @throws MWException
 452          */
 453         protected function doAck( Job $job ) {
 454                 if ( !isset( $job->metadata['id'] ) ) {
 455                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 456                 }
 457
 458                 $dbw = $this->getMasterDB();
 459                 try {
 460                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 461                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 462                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 463                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 464                         } );
 465
 466                         // Delete a row with a single DELETE without holding row locks over RTTs...
 467                         $dbw->delete( 'job',
 468                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 469
 470                         JobQueue::incrStats( 'acks', $this->type );
 471                 } catch ( DBError $e ) {
 472                         $this->throwDBException( $e );
 473                 }
 474         }
 475
 476         /**
 477          * @see JobQueue::doDeduplicateRootJob()
 478          * @param IJobSpecification $job
 479          * @throws MWException
 480          * @return bool
 481          */
 482         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 483                 $params = $job->getParams();
 484                 if ( !isset( $params['rootJobSignature'] ) ) {
 485                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 486                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 487                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 488                 }
 489                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 490                 // Callers should call batchInsert() and then this function so that if the insert
 491                 // fails, the de-duplication registration will be aborted. Since the insert is
 492                 // deferred till "transaction idle", do the same here, so that the ordering is
 493                 // maintained. Having only the de-duplication registration succeed would cause
 494                 // jobs to become no-ops without any actual jobs that made them redundant.
 495                 $dbw = $this->getMasterDB();
 496                 $cache = $this->dupCache;
 497                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 498                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 499                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 500                                 return true; // a newer version of this root job was enqueued
 501                         }
 502
 503                         // Update the timestamp of the last root job started at the location...
 504                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 505                 } );
 506
 507                 return true;
 508         }
 509
 510         /**
 511          * @see JobQueue::doDelete()
 512          * @return bool
 513          */
 514         protected function doDelete() {
 515                 $dbw = $this->getMasterDB();
 516                 try {
 517                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 518                 } catch ( DBError $e ) {
 519                         $this->throwDBException( $e );
 520                 }
 521
 522                 return true;
 523         }
 524
 525         /**
 526          * @see JobQueue::doWaitForBackups()
 527          * @return void
 528          */
 529         protected function doWaitForBackups() {
 530                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 531                 $lbFactory->waitForReplication( [ 'wiki' => $this->wiki, 'cluster' => $this->cluster ] );
 532         }
 533
 534         /**
 535          * @return void
 536          */
 537         protected function doFlushCaches() {
 538                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 539                         $this->cache->delete( $this->getCacheKey( $type ) );
 540                 }
 541         }
 542
 543         /**
 544          * @see JobQueue::getAllQueuedJobs()
 545          * @return Iterator
 546          */
 547         public function getAllQueuedJobs() {
 548                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 549         }
 550
 551         /**
 552          * @see JobQueue::getAllAcquiredJobs()
 553          * @return Iterator
 554          */
 555         public function getAllAcquiredJobs() {
 556                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 557         }
 558
 559         /**
 560          * @param array $conds Query conditions
 561          * @return Iterator
 562          */
 563         protected function getJobIterator( array $conds ) {
 564                 $dbr = $this->getSlaveDB();
 565                 try {
 566                         return new MappedIterator(
 567                                 $dbr->select( 'job', self::selectFields(), $conds ),
 568                                 function ( $row ) {
 569                                         $job = Job::factory(
 570                                                 $row->job_cmd,
 571                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 572                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 573                                         );
 574                                         $job->metadata['id'] = $row->job_id;
 575                                         $job->metadata['timestamp'] = $row->job_timestamp;
 576
 577                                         return $job;
 578                                 }
 579                         );
 580                 } catch ( DBError $e ) {
 581                         $this->throwDBException( $e );
 582                 }
 583         }
 584
 585         public function getCoalesceLocationInternal() {
 586                 return $this->cluster
 587                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 588                         : "LBFactory:{$this->wiki}";
 589         }
 590
 591         protected function doGetSiblingQueuesWithJobs( array $types ) {
 592                 $dbr = $this->getSlaveDB();
 593                 // @note: this does not check whether the jobs are claimed or not.
 594                 // This is useful so JobQueueGroup::pop() also sees queues that only
 595                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 596                 // failed jobs so that they can be popped again for that edge case.
 597                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 598                         [ 'job_cmd' => $types ], __METHOD__ );
 599
 600                 $types = [];
 601                 foreach ( $res as $row ) {
 602                         $types[] = $row->job_cmd;
 603                 }
 604
 605                 return $types;
 606         }
 607
 608         protected function doGetSiblingQueueSizes( array $types ) {
 609                 $dbr = $this->getSlaveDB();
 610                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 611                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 612
 613                 $sizes = [];
 614                 foreach ( $res as $row ) {
 615                         $sizes[$row->job_cmd] = (int)$row->count;
 616                 }
 617
 618                 return $sizes;
 619         }
 620
 621         /**
 622          * Recycle or destroy any jobs that have been claimed for too long
 623          *
 624          * @return int Number of jobs recycled/deleted
 625          */
 626         public function recycleAndDeleteStaleJobs() {
 627                 $now = time();
 628                 $count = 0; // affected rows
 629                 $dbw = $this->getMasterDB();
 630
 631                 try {
 632                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 633                                 return $count; // already in progress
 634                         }
 635
 636                         // Remove claims on jobs acquired for too long if enabled...
 637                         if ( $this->claimTTL > 0 ) {
 638                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 639                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 640                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 641                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 642                                 $res = $dbw->select( 'job', 'job_id',
 643                                         [
 644                                                 'job_cmd' => $this->type,
 645                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 646                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 647                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 648                                         __METHOD__
 649                                 );
 650                                 $ids = array_map(
 651                                         function ( $o ) {
 652                                                 return $o->job_id;
 653                                         }, iterator_to_array( $res )
 654                                 );
 655                                 if ( count( $ids ) ) {
 656                                         // Reset job_token for these jobs so that other runners will pick them up.
 657                                         // Set the timestamp to the current time, as it is useful to now that the job
 658                                         // was already tried before (the timestamp becomes the "released" time).
 659                                         $dbw->update( 'job',
 660                                                 [
 661                                                         'job_token' => '',
 662                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 663                                                 [
 664                                                         'job_id' => $ids ],
 665                                                 __METHOD__
 666                                         );
 667                                         $affected = $dbw->affectedRows();
 668                                         $count += $affected;
 669                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 670                                         $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 671                                 }
 672                         }
 673
 674                         // Just destroy any stale jobs...
 675                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 676                         $conds = [
 677                                 'job_cmd' => $this->type,
 678                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 679                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 680                         ];
 681                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 682                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 683                         }
 684                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 685                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 686                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 687                         $ids = array_map(
 688                                 function ( $o ) {
 689                                         return $o->job_id;
 690                                 }, iterator_to_array( $res )
 691                         );
 692                         if ( count( $ids ) ) {
 693                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 694                                 $affected = $dbw->affectedRows();
 695                                 $count += $affected;
 696                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 697                         }
 698
 699                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 700                 } catch ( DBError $e ) {
 701                         $this->throwDBException( $e );
 702                 }
 703
 704                 return $count;
 705         }
 706
 707         /**
 708          * @param IJobSpecification $job
 709          * @return array
 710          */
 711         protected function insertFields( IJobSpecification $job ) {
 712                 $dbw = $this->getMasterDB();
 713
 714                 return [
 715                         // Fields that describe the nature of the job
 716                         'job_cmd' => $job->getType(),
 717                         'job_namespace' => $job->getTitle()->getNamespace(),
 718                         'job_title' => $job->getTitle()->getDBkey(),
 719                         'job_params' => self::makeBlob( $job->getParams() ),
 720                         // Additional job metadata
 721                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 722                         'job_timestamp' => $dbw->timestamp(),
 723                         'job_sha1' => Wikimedia\base_convert(
 724                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 725                                 16, 36, 31
 726                         ),
 727                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 728                 ];
 729         }
 730
 731         /**
 732          * @throws JobQueueConnectionError
 733          * @return DBConnRef
 734          */
 735         protected function getSlaveDB() {
 736                 try {
 737                         return $this->getDB( DB_REPLICA );
 738                 } catch ( DBConnectionError $e ) {
 739                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 740                 }
 741         }
 742
 743         /**
 744          * @throws JobQueueConnectionError
 745          * @return DBConnRef
 746          */
 747         protected function getMasterDB() {
 748                 try {
 749                         return $this->getDB( DB_MASTER );
 750                 } catch ( DBConnectionError $e ) {
 751                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 752                 }
 753         }
 754
 755         /**
 756          * @param int $index (DB_REPLICA/DB_MASTER)
 757          * @return DBConnRef
 758          */
 759         protected function getDB( $index ) {
 760                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 761                 $lb = ( $this->cluster !== false )
 762                         ? $lbFactory->getExternalLB( $this->cluster, $this->wiki )
 763                         : $lbFactory->getMainLB( $this->wiki );
 764
 765                 return $lb->getConnectionRef( $index, [], $this->wiki );
 766         }
 767
 768         /**
 769          * @param string $property
 770          * @return string
 771          */
 772         private function getCacheKey( $property ) {
 773                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 774                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 775
 776                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 777         }
 778
 779         /**
 780          * @param array|bool $params
 781          * @return string
 782          */
 783         protected static function makeBlob( $params ) {
 784                 if ( $params !== false ) {
 785                         return serialize( $params );
 786                 } else {
 787                         return '';
 788                 }
 789         }
 790
 791         /**
 792          * @param string $blob
 793          * @return bool|mixed
 794          */
 795         protected static function extractBlob( $blob ) {
 796                 if ( (string)$blob !== '' ) {
 797                         return unserialize( $blob );
 798                 } else {
 799                         return false;
 800                 }
 801         }
 802
 803         /**
 804          * @param DBError $e
 805          * @throws JobQueueError
 806          */
 807         protected function throwDBException( DBError $e ) {
 808                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 809         }
 810
 811         /**
 812          * Return the list of job fields that should be selected.
 813          * @since 1.23
 814          * @return array
 815          */
 816         public static function selectFields() {
 817                 return [
 818                         'job_id',
 819                         'job_cmd',
 820                         'job_namespace',
 821                         'job_title',
 822                         'job_timestamp',
 823                         'job_params',
 824                         'job_random',
 825                         'job_attempts',
 826                         'job_token',
 827                         'job_token_timestamp',
 828                         'job_sha1',
 829                 ];
 830         }
 831 }