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  */
  22 use Wikimedia\Rdbms\IDatabase;
  23 use Wikimedia\Rdbms\DBConnRef;
  24 use Wikimedia\Rdbms\DBConnectionError;
  25 use Wikimedia\Rdbms\DBError;
  26 use MediaWiki\MediaWikiServices;
  27 use Wikimedia\ScopedCallback;
  28
  29 /**
  30  * Class to handle job queues stored in the DB
  31  *
  32  * @ingroup JobQueue
  33  * @since 1.21
  34  */
  35 class JobQueueDB extends JobQueue {
  36         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  37         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  38         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  39         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  40
  41         /** @var WANObjectCache */
  42         protected $cache;
  43
  44         /** @var bool|string Name of an external DB cluster. False if not set */
  45         protected $cluster = false;
  46
  47         /**
  48          * Additional parameters include:
  49          *   - cluster : The name of an external cluster registered via LBFactory.
  50          *               If not specified, the primary DB cluster for the wiki will be used.
  51          *               This can be overridden with a custom cluster so that DB handles will
  52          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  53          * @param array $params
  54          */
  55         protected function __construct( array $params ) {
  56                 parent::__construct( $params );
  57
  58                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  59                 $this->cache = ObjectCache::getMainWANInstance();
  60         }
  61
  62         protected function supportedOrders() {
  63                 return [ 'random', 'timestamp', 'fifo' ];
  64         }
  65
  66         protected function optimalOrder() {
  67                 return 'random';
  68         }
  69
  70         /**
  71          * @see JobQueue::doIsEmpty()
  72          * @return bool
  73          */
  74         protected function doIsEmpty() {
  75                 $dbr = $this->getReplicaDB();
  76                 try {
  77                         $found = $dbr->selectField( // unclaimed job
  78                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  79                         );
  80                 } catch ( DBError $e ) {
  81                         $this->throwDBException( $e );
  82                 }
  83
  84                 return !$found;
  85         }
  86
  87         /**
  88          * @see JobQueue::doGetSize()
  89          * @return int
  90          */
  91         protected function doGetSize() {
  92                 $key = $this->getCacheKey( 'size' );
  93
  94                 $size = $this->cache->get( $key );
  95                 if ( is_int( $size ) ) {
  96                         return $size;
  97                 }
  98
  99                 try {
 100                         $dbr = $this->getReplicaDB();
 101                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 102                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
 103                                 __METHOD__
 104                         );
 105                 } catch ( DBError $e ) {
 106                         $this->throwDBException( $e );
 107                 }
 108                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 109
 110                 return $size;
 111         }
 112
 113         /**
 114          * @see JobQueue::doGetAcquiredCount()
 115          * @return int
 116          */
 117         protected function doGetAcquiredCount() {
 118                 if ( $this->claimTTL <= 0 ) {
 119                         return 0; // no acknowledgements
 120                 }
 121
 122                 $key = $this->getCacheKey( 'acquiredcount' );
 123
 124                 $count = $this->cache->get( $key );
 125                 if ( is_int( $count ) ) {
 126                         return $count;
 127                 }
 128
 129                 $dbr = $this->getReplicaDB();
 130                 try {
 131                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 132                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 133                                 __METHOD__
 134                         );
 135                 } catch ( DBError $e ) {
 136                         $this->throwDBException( $e );
 137                 }
 138                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 139
 140                 return $count;
 141         }
 142
 143         /**
 144          * @see JobQueue::doGetAbandonedCount()
 145          * @return int
 146          * @throws MWException
 147          */
 148         protected function doGetAbandonedCount() {
 149                 if ( $this->claimTTL <= 0 ) {
 150                         return 0; // no acknowledgements
 151                 }
 152
 153                 $key = $this->getCacheKey( 'abandonedcount' );
 154
 155                 $count = $this->cache->get( $key );
 156                 if ( is_int( $count ) ) {
 157                         return $count;
 158                 }
 159
 160                 $dbr = $this->getReplicaDB();
 161                 try {
 162                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 163                                 [
 164                                         'job_cmd' => $this->type,
 165                                         "job_token != {$dbr->addQuotes( '' )}",
 166                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 167                                 ],
 168                                 __METHOD__
 169                         );
 170                 } catch ( DBError $e ) {
 171                         $this->throwDBException( $e );
 172                 }
 173
 174                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 175
 176                 return $count;
 177         }
 178
 179         /**
 180          * @see JobQueue::doBatchPush()
 181          * @param IJobSpecification[] $jobs
 182          * @param int $flags
 183          * @throws DBError|Exception
 184          * @return void
 185          */
 186         protected function doBatchPush( array $jobs, $flags ) {
 187                 $dbw = $this->getMasterDB();
 188                 // In general, there will be two cases here:
 189                 // a) sqlite; DB connection is probably a regular round-aware handle.
 190                 // If the connection is busy with a transaction, then defer the job writes
 191                 // until right before the main round commit step. Any errors that bubble
 192                 // up will rollback the main commit round.
 193                 // b) mysql/postgres; DB connection is generally a separate CONN_TRX_AUTO handle.
 194                 // No transaction is active nor will be started by writes, so enqueue the jobs
 195                 // now so that any errors will show up immediately as the interface expects. Any
 196                 // errors that bubble up will rollback the main commit round.
 197                 $fname = __METHOD__;
 198                 $dbw->onTransactionPreCommitOrIdle(
 199                         function () use ( $dbw, $jobs, $flags, $fname ) {
 200                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $fname );
 201                         },
 202                         $fname
 203                 );
 204         }
 205
 206         /**
 207          * This function should *not* be called outside of JobQueueDB
 208          *
 209          * @param IDatabase $dbw
 210          * @param IJobSpecification[] $jobs
 211          * @param int $flags
 212          * @param string $method
 213          * @throws DBError
 214          * @return void
 215          */
 216         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 217                 if ( !count( $jobs ) ) {
 218                         return;
 219                 }
 220
 221                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 222                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 223                 foreach ( $jobs as $job ) {
 224                         $row = $this->insertFields( $job );
 225                         if ( $job->ignoreDuplicates() ) {
 226                                 $rowSet[$row['job_sha1']] = $row;
 227                         } else {
 228                                 $rowList[] = $row;
 229                         }
 230                 }
 231
 232                 if ( $flags & self::QOS_ATOMIC ) {
 233                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 234                 }
 235                 try {
 236                         // Strip out any duplicate jobs that are already in the queue...
 237                         if ( count( $rowSet ) ) {
 238                                 $res = $dbw->select( 'job', 'job_sha1',
 239                                         [
 240                                                 // No job_type condition since it's part of the job_sha1 hash
 241                                                 'job_sha1' => array_keys( $rowSet ),
 242                                                 'job_token' => '' // unclaimed
 243                                         ],
 244                                         $method
 245                                 );
 246                                 foreach ( $res as $row ) {
 247                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 248                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 249                                 }
 250                         }
 251                         // Build the full list of job rows to insert
 252                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 253                         // Insert the job rows in chunks to avoid replica DB lag...
 254                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 255                                 $dbw->insert( 'job', $rowBatch, $method );
 256                         }
 257                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 258                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 259                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 260                         );
 261                 } catch ( DBError $e ) {
 262                         $this->throwDBException( $e );
 263                 }
 264                 if ( $flags & self::QOS_ATOMIC ) {
 265                         $dbw->endAtomic( $method );
 266                 }
 267
 268                 return;
 269         }
 270
 271         /**
 272          * @see JobQueue::doPop()
 273          * @return Job|bool
 274          */
 275         protected function doPop() {
 276                 $dbw = $this->getMasterDB();
 277                 try {
 278                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 279                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 280                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 281                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 282                         } );
 283
 284                         $uuid = wfRandomString( 32 ); // pop attempt
 285                         $job = false; // job popped off
 286                         do { // retry when our row is invalid or deleted as a duplicate
 287                                 // Try to reserve a row in the DB...
 288                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 289                                         $row = $this->claimOldest( $uuid );
 290                                 } else { // random first
 291                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 292                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 293                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 294                                 }
 295                                 // Check if we found a row to reserve...
 296                                 if ( !$row ) {
 297                                         break; // nothing to do
 298                                 }
 299                                 JobQueue::incrStats( 'pops', $this->type );
 300                                 // Get the job object from the row...
 301                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 302                                 $job = Job::factory( $row->job_cmd, $title,
 303                                         self::extractBlob( $row->job_params ), $row->job_id );
 304                                 $job->metadata['id'] = $row->job_id;
 305                                 $job->metadata['timestamp'] = $row->job_timestamp;
 306                                 break; // done
 307                         } while ( true );
 308
 309                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 310                                 // Handled jobs that need to be recycled/deleted;
 311                                 // any recycled jobs will be picked up next attempt
 312                                 $this->recycleAndDeleteStaleJobs();
 313                         }
 314                 } catch ( DBError $e ) {
 315                         $this->throwDBException( $e );
 316                 }
 317
 318                 return $job;
 319         }
 320
 321         /**
 322          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 323          *
 324          * @param string $uuid 32 char hex string
 325          * @param int $rand Random unsigned integer (31 bits)
 326          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 327          * @return stdClass|bool Row|false
 328          */
 329         protected function claimRandom( $uuid, $rand, $gte ) {
 330                 $dbw = $this->getMasterDB();
 331                 // Check cache to see if the queue has <= OFFSET items
 332                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 333
 334                 $row = false; // the row acquired
 335                 $invertedDirection = false; // whether one job_random direction was already scanned
 336                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 337                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 338                 // not replication safe. Due to https://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 339                 // be used here with MySQL.
 340                 do {
 341                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 342                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 343                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 344                                 $ineq = $gte ? '>=' : '<=';
 345                                 $dir = $gte ? 'ASC' : 'DESC';
 346                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 347                                         [
 348                                                 'job_cmd' => $this->type,
 349                                                 'job_token' => '', // unclaimed
 350                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 351                                         __METHOD__,
 352                                         [ 'ORDER BY' => "job_random {$dir}" ]
 353                                 );
 354                                 if ( !$row && !$invertedDirection ) {
 355                                         $gte = !$gte;
 356                                         $invertedDirection = true;
 357                                         continue; // try the other direction
 358                                 }
 359                         } else { // table *may* have >= MAX_OFFSET rows
 360                                 // T44614: "ORDER BY job_random" with a job_random inequality causes high CPU
 361                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 362                                 // instead of job_random for reducing excess claim retries.
 363                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 364                                         [
 365                                                 'job_cmd' => $this->type,
 366                                                 'job_token' => '', // unclaimed
 367                                         ],
 368                                         __METHOD__,
 369                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 370                                 );
 371                                 if ( !$row ) {
 372                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 373                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 374                                         continue; // use job_random
 375                                 }
 376                         }
 377
 378                         if ( $row ) { // claim the job
 379                                 $dbw->update( 'job', // update by PK
 380                                         [
 381                                                 'job_token' => $uuid,
 382                                                 'job_token_timestamp' => $dbw->timestamp(),
 383                                                 'job_attempts = job_attempts+1' ],
 384                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 385                                         __METHOD__
 386                                 );
 387                                 // This might get raced out by another runner when claiming the previously
 388                                 // selected row. The use of job_random should minimize this problem, however.
 389                                 if ( !$dbw->affectedRows() ) {
 390                                         $row = false; // raced out
 391                                 }
 392                         } else {
 393                                 break; // nothing to do
 394                         }
 395                 } while ( !$row );
 396
 397                 return $row;
 398         }
 399
 400         /**
 401          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 402          *
 403          * @param string $uuid 32 char hex string
 404          * @return stdClass|bool Row|false
 405          */
 406         protected function claimOldest( $uuid ) {
 407                 $dbw = $this->getMasterDB();
 408
 409                 $row = false; // the row acquired
 410                 do {
 411                         if ( $dbw->getType() === 'mysql' ) {
 412                                 // Per https://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 413                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 414                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 415                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 416                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 417                                         "SET " .
 418                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 419                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 420                                                 "job_attempts = job_attempts+1 " .
 421                                         "WHERE ( " .
 422                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 423                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 424                                         ") ORDER BY job_id ASC LIMIT 1",
 425                                         __METHOD__
 426                                 );
 427                         } else {
 428                                 // Use a subquery to find the job, within an UPDATE to claim it.
 429                                 // This uses as much of the DB wrapper functions as possible.
 430                                 $dbw->update( 'job',
 431                                         [
 432                                                 'job_token' => $uuid,
 433                                                 'job_token_timestamp' => $dbw->timestamp(),
 434                                                 'job_attempts = job_attempts+1' ],
 435                                         [ 'job_id = (' .
 436                                                 $dbw->selectSQLText( 'job', 'job_id',
 437                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 438                                                         __METHOD__,
 439                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 440                                                 ')'
 441                                         ],
 442                                         __METHOD__
 443                                 );
 444                         }
 445                         // Fetch any row that we just reserved...
 446                         if ( $dbw->affectedRows() ) {
 447                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 448                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 449                                 );
 450                                 if ( !$row ) { // raced out by duplicate job removal
 451                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 452                                 }
 453                         } else {
 454                                 break; // nothing to do
 455                         }
 456                 } while ( !$row );
 457
 458                 return $row;
 459         }
 460
 461         /**
 462          * @see JobQueue::doAck()
 463          * @param Job $job
 464          * @throws MWException
 465          */
 466         protected function doAck( Job $job ) {
 467                 if ( !isset( $job->metadata['id'] ) ) {
 468                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 469                 }
 470
 471                 $dbw = $this->getMasterDB();
 472                 try {
 473                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 474                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 475                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 476                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 477                         } );
 478
 479                         // Delete a row with a single DELETE without holding row locks over RTTs...
 480                         $dbw->delete( 'job',
 481                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 482
 483                         JobQueue::incrStats( 'acks', $this->type );
 484                 } catch ( DBError $e ) {
 485                         $this->throwDBException( $e );
 486                 }
 487         }
 488
 489         /**
 490          * @see JobQueue::doDeduplicateRootJob()
 491          * @param IJobSpecification $job
 492          * @throws MWException
 493          * @return bool
 494          */
 495         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 496                 $params = $job->getParams();
 497                 if ( !isset( $params['rootJobSignature'] ) ) {
 498                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 499                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 500                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 501                 }
 502                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 503                 // Callers should call batchInsert() and then this function so that if the insert
 504                 // fails, the de-duplication registration will be aborted. Since the insert is
 505                 // deferred till "transaction idle", do the same here, so that the ordering is
 506                 // maintained. Having only the de-duplication registration succeed would cause
 507                 // jobs to become no-ops without any actual jobs that made them redundant.
 508                 $dbw = $this->getMasterDB();
 509                 $cache = $this->dupCache;
 510                 $dbw->onTransactionIdle(
 511                         function () use ( $cache, $params, $key, $dbw ) {
 512                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 513                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 514                                         return true; // a newer version of this root job was enqueued
 515                                 }
 516
 517                                 // Update the timestamp of the last root job started at the location...
 518                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 519                         },
 520                         __METHOD__
 521                 );
 522
 523                 return true;
 524         }
 525
 526         /**
 527          * @see JobQueue::doDelete()
 528          * @return bool
 529          */
 530         protected function doDelete() {
 531                 $dbw = $this->getMasterDB();
 532                 try {
 533                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 534                 } catch ( DBError $e ) {
 535                         $this->throwDBException( $e );
 536                 }
 537
 538                 return true;
 539         }
 540
 541         /**
 542          * @see JobQueue::doWaitForBackups()
 543          * @return void
 544          */
 545         protected function doWaitForBackups() {
 546                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 547                 $lbFactory->waitForReplication( [ 'wiki' => $this->wiki, 'cluster' => $this->cluster ] );
 548         }
 549
 550         /**
 551          * @return void
 552          */
 553         protected function doFlushCaches() {
 554                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 555                         $this->cache->delete( $this->getCacheKey( $type ) );
 556                 }
 557         }
 558
 559         /**
 560          * @see JobQueue::getAllQueuedJobs()
 561          * @return Iterator
 562          */
 563         public function getAllQueuedJobs() {
 564                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 565         }
 566
 567         /**
 568          * @see JobQueue::getAllAcquiredJobs()
 569          * @return Iterator
 570          */
 571         public function getAllAcquiredJobs() {
 572                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 573         }
 574
 575         /**
 576          * @param array $conds Query conditions
 577          * @return Iterator
 578          */
 579         protected function getJobIterator( array $conds ) {
 580                 $dbr = $this->getReplicaDB();
 581                 try {
 582                         return new MappedIterator(
 583                                 $dbr->select( 'job', self::selectFields(), $conds ),
 584                                 function ( $row ) {
 585                                         $job = Job::factory(
 586                                                 $row->job_cmd,
 587                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 588                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 589                                         );
 590                                         $job->metadata['id'] = $row->job_id;
 591                                         $job->metadata['timestamp'] = $row->job_timestamp;
 592
 593                                         return $job;
 594                                 }
 595                         );
 596                 } catch ( DBError $e ) {
 597                         $this->throwDBException( $e );
 598                 }
 599         }
 600
 601         public function getCoalesceLocationInternal() {
 602                 return $this->cluster
 603                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 604                         : "LBFactory:{$this->wiki}";
 605         }
 606
 607         protected function doGetSiblingQueuesWithJobs( array $types ) {
 608                 $dbr = $this->getReplicaDB();
 609                 // @note: this does not check whether the jobs are claimed or not.
 610                 // This is useful so JobQueueGroup::pop() also sees queues that only
 611                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 612                 // failed jobs so that they can be popped again for that edge case.
 613                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 614                         [ 'job_cmd' => $types ], __METHOD__ );
 615
 616                 $types = [];
 617                 foreach ( $res as $row ) {
 618                         $types[] = $row->job_cmd;
 619                 }
 620
 621                 return $types;
 622         }
 623
 624         protected function doGetSiblingQueueSizes( array $types ) {
 625                 $dbr = $this->getReplicaDB();
 626                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 627                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 628
 629                 $sizes = [];
 630                 foreach ( $res as $row ) {
 631                         $sizes[$row->job_cmd] = (int)$row->count;
 632                 }
 633
 634                 return $sizes;
 635         }
 636
 637         /**
 638          * Recycle or destroy any jobs that have been claimed for too long
 639          *
 640          * @return int Number of jobs recycled/deleted
 641          */
 642         public function recycleAndDeleteStaleJobs() {
 643                 $now = time();
 644                 $count = 0; // affected rows
 645                 $dbw = $this->getMasterDB();
 646
 647                 try {
 648                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 649                                 return $count; // already in progress
 650                         }
 651
 652                         // Remove claims on jobs acquired for too long if enabled...
 653                         if ( $this->claimTTL > 0 ) {
 654                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 655                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 656                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 657                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 658                                 $res = $dbw->select( 'job', 'job_id',
 659                                         [
 660                                                 'job_cmd' => $this->type,
 661                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 662                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 663                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 664                                         __METHOD__
 665                                 );
 666                                 $ids = array_map(
 667                                         function ( $o ) {
 668                                                 return $o->job_id;
 669                                         }, iterator_to_array( $res )
 670                                 );
 671                                 if ( count( $ids ) ) {
 672                                         // Reset job_token for these jobs so that other runners will pick them up.
 673                                         // Set the timestamp to the current time, as it is useful to now that the job
 674                                         // was already tried before (the timestamp becomes the "released" time).
 675                                         $dbw->update( 'job',
 676                                                 [
 677                                                         'job_token' => '',
 678                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 679                                                 [
 680                                                         'job_id' => $ids ],
 681                                                 __METHOD__
 682                                         );
 683                                         $affected = $dbw->affectedRows();
 684                                         $count += $affected;
 685                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 686                                         $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 687                                 }
 688                         }
 689
 690                         // Just destroy any stale jobs...
 691                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 692                         $conds = [
 693                                 'job_cmd' => $this->type,
 694                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 695                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 696                         ];
 697                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 698                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 699                         }
 700                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 701                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 702                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 703                         $ids = array_map(
 704                                 function ( $o ) {
 705                                         return $o->job_id;
 706                                 }, iterator_to_array( $res )
 707                         );
 708                         if ( count( $ids ) ) {
 709                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 710                                 $affected = $dbw->affectedRows();
 711                                 $count += $affected;
 712                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 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 IJobSpecification $job
 725          * @return array
 726          */
 727         protected function insertFields( IJobSpecification $job ) {
 728                 $dbw = $this->getMasterDB();
 729
 730                 return [
 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' => Wikimedia\base_convert(
 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          * @throws JobQueueConnectionError
 749          * @return DBConnRef
 750          */
 751         protected function getReplicaDB() {
 752                 try {
 753                         return $this->getDB( DB_REPLICA );
 754                 } catch ( DBConnectionError $e ) {
 755                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 756                 }
 757         }
 758
 759         /**
 760          * @throws JobQueueConnectionError
 761          * @return DBConnRef
 762          */
 763         protected function getMasterDB() {
 764                 try {
 765                         return $this->getDB( DB_MASTER );
 766                 } catch ( DBConnectionError $e ) {
 767                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 768                 }
 769         }
 770
 771         /**
 772          * @param int $index (DB_REPLICA/DB_MASTER)
 773          * @return DBConnRef
 774          */
 775         protected function getDB( $index ) {
 776                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 777                 $lb = ( $this->cluster !== false )
 778                         ? $lbFactory->getExternalLB( $this->cluster )
 779                         : $lbFactory->getMainLB( $this->wiki );
 780
 781                 return ( $lb->getServerType( $lb->getWriterIndex() ) !== 'sqlite' )
 782                         // Keep a separate connection to avoid contention and deadlocks;
 783                         // However, SQLite has the opposite behavior due to DB-level locking.
 784                         ? $lb->getConnectionRef( $index, [], $this->wiki, $lb::CONN_TRX_AUTO )
 785                         // Jobs insertion will be defered until the PRESEND stage to reduce contention.
 786                         : $lb->getConnectionRef( $index, [], $this->wiki );
 787         }
 788
 789         /**
 790          * @param string $property
 791          * @return string
 792          */
 793         private function getCacheKey( $property ) {
 794                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 795                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 796
 797                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 798         }
 799
 800         /**
 801          * @param array|bool $params
 802          * @return string
 803          */
 804         protected static function makeBlob( $params ) {
 805                 if ( $params !== false ) {
 806                         return serialize( $params );
 807                 } else {
 808                         return '';
 809                 }
 810         }
 811
 812         /**
 813          * @param string $blob
 814          * @return bool|mixed
 815          */
 816         protected static function extractBlob( $blob ) {
 817                 if ( (string)$blob !== '' ) {
 818                         return unserialize( $blob );
 819                 } else {
 820                         return false;
 821                 }
 822         }
 823
 824         /**
 825          * @param DBError $e
 826          * @throws JobQueueError
 827          */
 828         protected function throwDBException( DBError $e ) {
 829                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 830         }
 831
 832         /**
 833          * Return the list of job fields that should be selected.
 834          * @since 1.23
 835          * @return array
 836          */
 837         public static function selectFields() {
 838                 return [
 839                         'job_id',
 840                         'job_cmd',
 841                         'job_namespace',
 842                         'job_title',
 843                         'job_timestamp',
 844                         'job_params',
 845                         'job_random',
 846                         'job_attempts',
 847                         'job_token',
 848                         'job_token_timestamp',
 849                         'job_sha1',
 850                 ];
 851         }
 852 }