includes/jobqueue/JobQueueDB.php

   1 <?php
   2 /**
   3  * Database-backed job queue code.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @author Aaron Schulz
  22  */
  23
  24 /**
  25  * Class to handle job queues stored in the DB
  26  *
  27  * @ingroup JobQueue
  28  * @since 1.21
  29  */
  30 class JobQueueDB extends JobQueue {
  31         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  32         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  33         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  34         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  35
  36         /** @var WANObjectCache */
  37         protected $cache;
  38
  39         /** @var bool|string Name of an external DB cluster. False if not set */
  40         protected $cluster = false;
  41
  42         /**
  43          * Additional parameters include:
  44          *   - cluster : The name of an external cluster registered via LBFactory.
  45          *               If not specified, the primary DB cluster for the wiki will be used.
  46          *               This can be overridden with a custom cluster so that DB handles will
  47          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  48          * @param array $params
  49          */
  50         protected function __construct( array $params ) {
  51                 parent::__construct( $params );
  52
  53                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  54                 $this->cache = ObjectCache::getMainWANInstance();
  55         }
  56
  57         protected function supportedOrders() {
  58                 return [ 'random', 'timestamp', 'fifo' ];
  59         }
  60
  61         protected function optimalOrder() {
  62                 return 'random';
  63         }
  64
  65         /**
  66          * @see JobQueue::doIsEmpty()
  67          * @return bool
  68          */
  69         protected function doIsEmpty() {
  70                 $dbr = $this->getSlaveDB();
  71                 try {
  72                         $found = $dbr->selectField( // unclaimed job
  73                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  74                         );
  75                 } catch ( DBError $e ) {
  76                         $this->throwDBException( $e );
  77                 }
  78
  79                 return !$found;
  80         }
  81
  82         /**
  83          * @see JobQueue::doGetSize()
  84          * @return int
  85          */
  86         protected function doGetSize() {
  87                 $key = $this->getCacheKey( 'size' );
  88
  89                 $size = $this->cache->get( $key );
  90                 if ( is_int( $size ) ) {
  91                         return $size;
  92                 }
  93
  94                 try {
  95                         $dbr = $this->getSlaveDB();
  96                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
  97                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
  98                                 __METHOD__
  99                         );
 100                 } catch ( DBError $e ) {
 101                         $this->throwDBException( $e );
 102                 }
 103                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 104
 105                 return $size;
 106         }
 107
 108         /**
 109          * @see JobQueue::doGetAcquiredCount()
 110          * @return int
 111          */
 112         protected function doGetAcquiredCount() {
 113                 if ( $this->claimTTL <= 0 ) {
 114                         return 0; // no acknowledgements
 115                 }
 116
 117                 $key = $this->getCacheKey( 'acquiredcount' );
 118
 119                 $count = $this->cache->get( $key );
 120                 if ( is_int( $count ) ) {
 121                         return $count;
 122                 }
 123
 124                 $dbr = $this->getSlaveDB();
 125                 try {
 126                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 127                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 128                                 __METHOD__
 129                         );
 130                 } catch ( DBError $e ) {
 131                         $this->throwDBException( $e );
 132                 }
 133                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 134
 135                 return $count;
 136         }
 137
 138         /**
 139          * @see JobQueue::doGetAbandonedCount()
 140          * @return int
 141          * @throws MWException
 142          */
 143         protected function doGetAbandonedCount() {
 144                 if ( $this->claimTTL <= 0 ) {
 145                         return 0; // no acknowledgements
 146                 }
 147
 148                 $key = $this->getCacheKey( 'abandonedcount' );
 149
 150                 $count = $this->cache->get( $key );
 151                 if ( is_int( $count ) ) {
 152                         return $count;
 153                 }
 154
 155                 $dbr = $this->getSlaveDB();
 156                 try {
 157                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 158                                 [
 159                                         'job_cmd' => $this->type,
 160                                         "job_token != {$dbr->addQuotes( '' )}",
 161                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 162                                 ],
 163                                 __METHOD__
 164                         );
 165                 } catch ( DBError $e ) {
 166                         $this->throwDBException( $e );
 167                 }
 168
 169                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 170
 171                 return $count;
 172         }
 173
 174         /**
 175          * @see JobQueue::doBatchPush()
 176          * @param IJobSpecification[] $jobs
 177          * @param int $flags
 178          * @throws DBError|Exception
 179          * @return void
 180          */
 181         protected function doBatchPush( array $jobs, $flags ) {
 182                 $dbw = $this->getMasterDB();
 183
 184                 $method = __METHOD__;
 185                 $dbw->onTransactionIdle(
 186                         function () use ( $dbw, $jobs, $flags, $method ) {
 187                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 188                         }
 189                 );
 190         }
 191
 192         /**
 193          * This function should *not* be called outside of JobQueueDB
 194          *
 195          * @param IDatabase $dbw
 196          * @param IJobSpecification[] $jobs
 197          * @param int $flags
 198          * @param string $method
 199          * @throws DBError
 200          * @return void
 201          */
 202         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 203                 if ( !count( $jobs ) ) {
 204                         return;
 205                 }
 206
 207                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 208                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 209                 foreach ( $jobs as $job ) {
 210                         $row = $this->insertFields( $job );
 211                         if ( $job->ignoreDuplicates() ) {
 212                                 $rowSet[$row['job_sha1']] = $row;
 213                         } else {
 214                                 $rowList[] = $row;
 215                         }
 216                 }
 217
 218                 if ( $flags & self::QOS_ATOMIC ) {
 219                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 220                 }
 221                 try {
 222                         // Strip out any duplicate jobs that are already in the queue...
 223                         if ( count( $rowSet ) ) {
 224                                 $res = $dbw->select( 'job', 'job_sha1',
 225                                         [
 226                                                 // No job_type condition since it's part of the job_sha1 hash
 227                                                 'job_sha1' => array_keys( $rowSet ),
 228                                                 'job_token' => '' // unclaimed
 229                                         ],
 230                                         $method
 231                                 );
 232                                 foreach ( $res as $row ) {
 233                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 234                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 235                                 }
 236                         }
 237                         // Build the full list of job rows to insert
 238                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 239                         // Insert the job rows in chunks to avoid slave lag...
 240                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 241                                 $dbw->insert( 'job', $rowBatch, $method );
 242                         }
 243                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 244                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 245                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 246                         );
 247                 } catch ( DBError $e ) {
 248                         $this->throwDBException( $e );
 249                 }
 250                 if ( $flags & self::QOS_ATOMIC ) {
 251                         $dbw->endAtomic( $method );
 252                 }
 253
 254                 return;
 255         }
 256
 257         /**
 258          * @see JobQueue::doPop()
 259          * @return Job|bool
 260          */
 261         protected function doPop() {
 262                 $dbw = $this->getMasterDB();
 263                 try {
 264                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 265                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 266                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 267                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 268                         } );
 269
 270                         $uuid = wfRandomString( 32 ); // pop attempt
 271                         $job = false; // job popped off
 272                         do { // retry when our row is invalid or deleted as a duplicate
 273                                 // Try to reserve a row in the DB...
 274                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 275                                         $row = $this->claimOldest( $uuid );
 276                                 } else { // random first
 277                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 278                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 279                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 280                                 }
 281                                 // Check if we found a row to reserve...
 282                                 if ( !$row ) {
 283                                         break; // nothing to do
 284                                 }
 285                                 JobQueue::incrStats( 'pops', $this->type );
 286                                 // Get the job object from the row...
 287                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 288                                 $job = Job::factory( $row->job_cmd, $title,
 289                                         self::extractBlob( $row->job_params ), $row->job_id );
 290                                 $job->metadata['id'] = $row->job_id;
 291                                 $job->metadata['timestamp'] = $row->job_timestamp;
 292                                 break; // done
 293                         } while ( true );
 294
 295                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 296                                 // Handled jobs that need to be recycled/deleted;
 297                                 // any recycled jobs will be picked up next attempt
 298                                 $this->recycleAndDeleteStaleJobs();
 299                         }
 300                 } catch ( DBError $e ) {
 301                         $this->throwDBException( $e );
 302                 }
 303
 304                 return $job;
 305         }
 306
 307         /**
 308          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 309          *
 310          * @param string $uuid 32 char hex string
 311          * @param int $rand Random unsigned integer (31 bits)
 312          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 313          * @return stdClass|bool Row|false
 314          */
 315         protected function claimRandom( $uuid, $rand, $gte ) {
 316                 $dbw = $this->getMasterDB();
 317                 // Check cache to see if the queue has <= OFFSET items
 318                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 319
 320                 $row = false; // the row acquired
 321                 $invertedDirection = false; // whether one job_random direction was already scanned
 322                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 323                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 324                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 325                 // be used here with MySQL.
 326                 do {
 327                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 328                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 329                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 330                                 $ineq = $gte ? '>=' : '<=';
 331                                 $dir = $gte ? 'ASC' : 'DESC';
 332                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 333                                         [
 334                                                 'job_cmd' => $this->type,
 335                                                 'job_token' => '', // unclaimed
 336                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 337                                         __METHOD__,
 338                                         [ 'ORDER BY' => "job_random {$dir}" ]
 339                                 );
 340                                 if ( !$row && !$invertedDirection ) {
 341                                         $gte = !$gte;
 342                                         $invertedDirection = true;
 343                                         continue; // try the other direction
 344                                 }
 345                         } else { // table *may* have >= MAX_OFFSET rows
 346                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 347                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 348                                 // instead of job_random for reducing excess claim retries.
 349                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 350                                         [
 351                                                 'job_cmd' => $this->type,
 352                                                 'job_token' => '', // unclaimed
 353                                         ],
 354                                         __METHOD__,
 355                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 356                                 );
 357                                 if ( !$row ) {
 358                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 359                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 360                                         continue; // use job_random
 361                                 }
 362                         }
 363
 364                         if ( $row ) { // claim the job
 365                                 $dbw->update( 'job', // update by PK
 366                                         [
 367                                                 'job_token' => $uuid,
 368                                                 'job_token_timestamp' => $dbw->timestamp(),
 369                                                 'job_attempts = job_attempts+1' ],
 370                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 371                                         __METHOD__
 372                                 );
 373                                 // This might get raced out by another runner when claiming the previously
 374                                 // selected row. The use of job_random should minimize this problem, however.
 375                                 if ( !$dbw->affectedRows() ) {
 376                                         $row = false; // raced out
 377                                 }
 378                         } else {
 379                                 break; // nothing to do
 380                         }
 381                 } while ( !$row );
 382
 383                 return $row;
 384         }
 385
 386         /**
 387          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 388          *
 389          * @param string $uuid 32 char hex string
 390          * @return stdClass|bool Row|false
 391          */
 392         protected function claimOldest( $uuid ) {
 393                 $dbw = $this->getMasterDB();
 394
 395                 $row = false; // the row acquired
 396                 do {
 397                         if ( $dbw->getType() === 'mysql' ) {
 398                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 399                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 400                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 401                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 402                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 403                                         "SET " .
 404                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 405                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 406                                                 "job_attempts = job_attempts+1 " .
 407                                         "WHERE ( " .
 408                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 409                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 410                                         ") ORDER BY job_id ASC LIMIT 1",
 411                                         __METHOD__
 412                                 );
 413                         } else {
 414                                 // Use a subquery to find the job, within an UPDATE to claim it.
 415                                 // This uses as much of the DB wrapper functions as possible.
 416                                 $dbw->update( 'job',
 417                                         [
 418                                                 'job_token' => $uuid,
 419                                                 'job_token_timestamp' => $dbw->timestamp(),
 420                                                 'job_attempts = job_attempts+1' ],
 421                                         [ 'job_id = (' .
 422                                                 $dbw->selectSQLText( 'job', 'job_id',
 423                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 424                                                         __METHOD__,
 425                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 426                                                 ')'
 427                                         ],
 428                                         __METHOD__
 429                                 );
 430                         }
 431                         // Fetch any row that we just reserved...
 432                         if ( $dbw->affectedRows() ) {
 433                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 434                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 435                                 );
 436                                 if ( !$row ) { // raced out by duplicate job removal
 437                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 438                                 }
 439                         } else {
 440                                 break; // nothing to do
 441                         }
 442                 } while ( !$row );
 443
 444                 return $row;
 445         }
 446
 447         /**
 448          * @see JobQueue::doAck()
 449          * @param Job $job
 450          * @throws MWException
 451          */
 452         protected function doAck( Job $job ) {
 453                 if ( !isset( $job->metadata['id'] ) ) {
 454                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 455                 }
 456
 457                 $dbw = $this->getMasterDB();
 458                 try {
 459                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 460                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 461                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 462                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 463                         } );
 464
 465                         // Delete a row with a single DELETE without holding row locks over RTTs...
 466                         $dbw->delete( 'job',
 467                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 468
 469                         JobQueue::incrStats( 'acks', $this->type );
 470                 } catch ( DBError $e ) {
 471                         $this->throwDBException( $e );
 472                 }
 473         }
 474
 475         /**
 476          * @see JobQueue::doDeduplicateRootJob()
 477          * @param IJobSpecification $job
 478          * @throws MWException
 479          * @return bool
 480          */
 481         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 482                 $params = $job->getParams();
 483                 if ( !isset( $params['rootJobSignature'] ) ) {
 484                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 485                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 486                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 487                 }
 488                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 489                 // Callers should call batchInsert() and then this function so that if the insert
 490                 // fails, the de-duplication registration will be aborted. Since the insert is
 491                 // deferred till "transaction idle", do the same here, so that the ordering is
 492                 // maintained. Having only the de-duplication registration succeed would cause
 493                 // jobs to become no-ops without any actual jobs that made them redundant.
 494                 $dbw = $this->getMasterDB();
 495                 $cache = $this->dupCache;
 496                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 497                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 498                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 499                                 return true; // a newer version of this root job was enqueued
 500                         }
 501
 502                         // Update the timestamp of the last root job started at the location...
 503                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 504                 } );
 505
 506                 return true;
 507         }
 508
 509         /**
 510          * @see JobQueue::doDelete()
 511          * @return bool
 512          */
 513         protected function doDelete() {
 514                 $dbw = $this->getMasterDB();
 515                 try {
 516                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 517                 } catch ( DBError $e ) {
 518                         $this->throwDBException( $e );
 519                 }
 520
 521                 return true;
 522         }
 523
 524         /**
 525          * @see JobQueue::doWaitForBackups()
 526          * @return void
 527          */
 528         protected function doWaitForBackups() {
 529                 wfWaitForSlaves( false, $this->wiki, $this->cluster ?: false );
 530         }
 531
 532         /**
 533          * @return void
 534          */
 535         protected function doFlushCaches() {
 536                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 537                         $this->cache->delete( $this->getCacheKey( $type ) );
 538                 }
 539         }
 540
 541         /**
 542          * @see JobQueue::getAllQueuedJobs()
 543          * @return Iterator
 544          */
 545         public function getAllQueuedJobs() {
 546                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 547         }
 548
 549         /**
 550          * @see JobQueue::getAllAcquiredJobs()
 551          * @return Iterator
 552          */
 553         public function getAllAcquiredJobs() {
 554                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 555         }
 556
 557         /**
 558          * @param array $conds Query conditions
 559          * @return Iterator
 560          */
 561         protected function getJobIterator( array $conds ) {
 562                 $dbr = $this->getSlaveDB();
 563                 try {
 564                         return new MappedIterator(
 565                                 $dbr->select( 'job', self::selectFields(), $conds ),
 566                                 function ( $row ) {
 567                                         $job = Job::factory(
 568                                                 $row->job_cmd,
 569                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 570                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 571                                         );
 572                                         $job->metadata['id'] = $row->job_id;
 573                                         $job->metadata['timestamp'] = $row->job_timestamp;
 574
 575                                         return $job;
 576                                 }
 577                         );
 578                 } catch ( DBError $e ) {
 579                         $this->throwDBException( $e );
 580                 }
 581         }
 582
 583         public function getCoalesceLocationInternal() {
 584                 return $this->cluster
 585                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 586                         : "LBFactory:{$this->wiki}";
 587         }
 588
 589         protected function doGetSiblingQueuesWithJobs( array $types ) {
 590                 $dbr = $this->getSlaveDB();
 591                 // @note: this does not check whether the jobs are claimed or not.
 592                 // This is useful so JobQueueGroup::pop() also sees queues that only
 593                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 594                 // failed jobs so that they can be popped again for that edge case.
 595                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 596                         [ 'job_cmd' => $types ], __METHOD__ );
 597
 598                 $types = [];
 599                 foreach ( $res as $row ) {
 600                         $types[] = $row->job_cmd;
 601                 }
 602
 603                 return $types;
 604         }
 605
 606         protected function doGetSiblingQueueSizes( array $types ) {
 607                 $dbr = $this->getSlaveDB();
 608                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 609                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 610
 611                 $sizes = [];
 612                 foreach ( $res as $row ) {
 613                         $sizes[$row->job_cmd] = (int)$row->count;
 614                 }
 615
 616                 return $sizes;
 617         }
 618
 619         /**
 620          * Recycle or destroy any jobs that have been claimed for too long
 621          *
 622          * @return int Number of jobs recycled/deleted
 623          */
 624         public function recycleAndDeleteStaleJobs() {
 625                 $now = time();
 626                 $count = 0; // affected rows
 627                 $dbw = $this->getMasterDB();
 628
 629                 try {
 630                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 631                                 return $count; // already in progress
 632                         }
 633
 634                         // Remove claims on jobs acquired for too long if enabled...
 635                         if ( $this->claimTTL > 0 ) {
 636                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 637                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 638                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 639                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 640                                 $res = $dbw->select( 'job', 'job_id',
 641                                         [
 642                                                 'job_cmd' => $this->type,
 643                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 644                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 645                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 646                                         __METHOD__
 647                                 );
 648                                 $ids = array_map(
 649                                         function ( $o ) {
 650                                                 return $o->job_id;
 651                                         }, iterator_to_array( $res )
 652                                 );
 653                                 if ( count( $ids ) ) {
 654                                         // Reset job_token for these jobs so that other runners will pick them up.
 655                                         // Set the timestamp to the current time, as it is useful to now that the job
 656                                         // was already tried before (the timestamp becomes the "released" time).
 657                                         $dbw->update( 'job',
 658                                                 [
 659                                                         'job_token' => '',
 660                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 661                                                 [
 662                                                         'job_id' => $ids ],
 663                                                 __METHOD__
 664                                         );
 665                                         $affected = $dbw->affectedRows();
 666                                         $count += $affected;
 667                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 668                                         $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 669                                 }
 670                         }
 671
 672                         // Just destroy any stale jobs...
 673                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 674                         $conds = [
 675                                 'job_cmd' => $this->type,
 676                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 677                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 678                         ];
 679                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 680                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 681                         }
 682                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 683                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 684                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 685                         $ids = array_map(
 686                                 function ( $o ) {
 687                                         return $o->job_id;
 688                                 }, iterator_to_array( $res )
 689                         );
 690                         if ( count( $ids ) ) {
 691                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 692                                 $affected = $dbw->affectedRows();
 693                                 $count += $affected;
 694                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 695                         }
 696
 697                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 698                 } catch ( DBError $e ) {
 699                         $this->throwDBException( $e );
 700                 }
 701
 702                 return $count;
 703         }
 704
 705         /**
 706          * @param IJobSpecification $job
 707          * @return array
 708          */
 709         protected function insertFields( IJobSpecification $job ) {
 710                 $dbw = $this->getMasterDB();
 711
 712                 return [
 713                         // Fields that describe the nature of the job
 714                         'job_cmd' => $job->getType(),
 715                         'job_namespace' => $job->getTitle()->getNamespace(),
 716                         'job_title' => $job->getTitle()->getDBkey(),
 717                         'job_params' => self::makeBlob( $job->getParams() ),
 718                         // Additional job metadata
 719                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 720                         'job_timestamp' => $dbw->timestamp(),
 721                         'job_sha1' => Wikimedia\base_convert(
 722                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 723                                 16, 36, 31
 724                         ),
 725                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 726                 ];
 727         }
 728
 729         /**
 730          * @throws JobQueueConnectionError
 731          * @return DBConnRef
 732          */
 733         protected function getSlaveDB() {
 734                 try {
 735                         return $this->getDB( DB_SLAVE );
 736                 } catch ( DBConnectionError $e ) {
 737                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 738                 }
 739         }
 740
 741         /**
 742          * @throws JobQueueConnectionError
 743          * @return DBConnRef
 744          */
 745         protected function getMasterDB() {
 746                 try {
 747                         return $this->getDB( DB_MASTER );
 748                 } catch ( DBConnectionError $e ) {
 749                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 750                 }
 751         }
 752
 753         /**
 754          * @param int $index (DB_SLAVE/DB_MASTER)
 755          * @return DBConnRef
 756          */
 757         protected function getDB( $index ) {
 758                 $lb = ( $this->cluster !== false )
 759                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 760                         : wfGetLB( $this->wiki );
 761
 762                 return $lb->getConnectionRef( $index, [], $this->wiki );
 763         }
 764
 765         /**
 766          * @param string $property
 767          * @return string
 768          */
 769         private function getCacheKey( $property ) {
 770                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 771                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 772
 773                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 774         }
 775
 776         /**
 777          * @param array|bool $params
 778          * @return string
 779          */
 780         protected static function makeBlob( $params ) {
 781                 if ( $params !== false ) {
 782                         return serialize( $params );
 783                 } else {
 784                         return '';
 785                 }
 786         }
 787
 788         /**
 789          * @param string $blob
 790          * @return bool|mixed
 791          */
 792         protected static function extractBlob( $blob ) {
 793                 if ( (string)$blob !== '' ) {
 794                         return unserialize( $blob );
 795                 } else {
 796                         return false;
 797                 }
 798         }
 799
 800         /**
 801          * @param DBError $e
 802          * @throws JobQueueError
 803          */
 804         protected function throwDBException( DBError $e ) {
 805                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 806         }
 807
 808         /**
 809          * Return the list of job fields that should be selected.
 810          * @since 1.23
 811          * @return array
 812          */
 813         public static function selectFields() {
 814                 return [
 815                         'job_id',
 816                         'job_cmd',
 817                         'job_namespace',
 818                         'job_title',
 819                         'job_timestamp',
 820                         'job_params',
 821                         'job_random',
 822                         'job_attempts',
 823                         'job_token',
 824                         'job_token_timestamp',
 825                         'job_sha1',
 826                 ];
 827         }
 828 }