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\Database;
  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         /** @var IDatabase|DBError|null */
  44         protected $conn;
  45
  46         /** @var array|null Server configuration array */
  47         protected $server;
  48         /** @var string|null Name of an external DB cluster or null for the local DB cluster */
  49         protected $cluster;
  50
  51         /**
  52          * Additional parameters include:
  53          *   - server  : Server configuration array for Database::factory. Overrides "cluster".
  54          *   - cluster : The name of an external cluster registered via LBFactory.
  55          *               If not specified, the primary DB cluster for the wiki will be used.
  56          *               This can be overridden with a custom cluster so that DB handles will
  57          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  58          * @param array $params
  59          */
  60         protected function __construct( array $params ) {
  61                 parent::__construct( $params );
  62
  63                 if ( isset( $params['server'] ) ) {
  64                         $this->server = $params['server'];
  65                 } elseif ( isset( $params['cluster'] ) && is_string( $params['cluster'] ) ) {
  66                         $this->cluster = $params['cluster'];
  67                 }
  68
  69                 $this->cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
  70         }
  71
  72         protected function supportedOrders() {
  73                 return [ 'random', 'timestamp', 'fifo' ];
  74         }
  75
  76         protected function optimalOrder() {
  77                 return 'random';
  78         }
  79
  80         /**
  81          * @see JobQueue::doIsEmpty()
  82          * @return bool
  83          */
  84         protected function doIsEmpty() {
  85                 $dbr = $this->getReplicaDB();
  86                 try {
  87                         $found = $dbr->selectField( // unclaimed job
  88                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  89                         );
  90                 } catch ( DBError $e ) {
  91                         $this->throwDBException( $e );
  92                 }
  93
  94                 return !$found;
  95         }
  96
  97         /**
  98          * @see JobQueue::doGetSize()
  99          * @return int
 100          */
 101         protected function doGetSize() {
 102                 $key = $this->getCacheKey( 'size' );
 103
 104                 $size = $this->cache->get( $key );
 105                 if ( is_int( $size ) ) {
 106                         return $size;
 107                 }
 108
 109                 try {
 110                         $dbr = $this->getReplicaDB();
 111                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 112                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
 113                                 __METHOD__
 114                         );
 115                 } catch ( DBError $e ) {
 116                         $this->throwDBException( $e );
 117                 }
 118                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 119
 120                 return $size;
 121         }
 122
 123         /**
 124          * @see JobQueue::doGetAcquiredCount()
 125          * @return int
 126          */
 127         protected function doGetAcquiredCount() {
 128                 if ( $this->claimTTL <= 0 ) {
 129                         return 0; // no acknowledgements
 130                 }
 131
 132                 $key = $this->getCacheKey( 'acquiredcount' );
 133
 134                 $count = $this->cache->get( $key );
 135                 if ( is_int( $count ) ) {
 136                         return $count;
 137                 }
 138
 139                 $dbr = $this->getReplicaDB();
 140                 try {
 141                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 142                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 143                                 __METHOD__
 144                         );
 145                 } catch ( DBError $e ) {
 146                         $this->throwDBException( $e );
 147                 }
 148                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 149
 150                 return $count;
 151         }
 152
 153         /**
 154          * @see JobQueue::doGetAbandonedCount()
 155          * @return int
 156          * @throws MWException
 157          */
 158         protected function doGetAbandonedCount() {
 159                 if ( $this->claimTTL <= 0 ) {
 160                         return 0; // no acknowledgements
 161                 }
 162
 163                 $key = $this->getCacheKey( 'abandonedcount' );
 164
 165                 $count = $this->cache->get( $key );
 166                 if ( is_int( $count ) ) {
 167                         return $count;
 168                 }
 169
 170                 $dbr = $this->getReplicaDB();
 171                 try {
 172                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 173                                 [
 174                                         'job_cmd' => $this->type,
 175                                         "job_token != {$dbr->addQuotes( '' )}",
 176                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 177                                 ],
 178                                 __METHOD__
 179                         );
 180                 } catch ( DBError $e ) {
 181                         $this->throwDBException( $e );
 182                 }
 183
 184                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 185
 186                 return $count;
 187         }
 188
 189         /**
 190          * @see JobQueue::doBatchPush()
 191          * @param IJobSpecification[] $jobs
 192          * @param int $flags
 193          * @throws DBError|Exception
 194          * @return void
 195          */
 196         protected function doBatchPush( array $jobs, $flags ) {
 197                 $dbw = $this->getMasterDB();
 198                 // In general, there will be two cases here:
 199                 // a) sqlite; DB connection is probably a regular round-aware handle.
 200                 // If the connection is busy with a transaction, then defer the job writes
 201                 // until right before the main round commit step. Any errors that bubble
 202                 // up will rollback the main commit round.
 203                 // b) mysql/postgres; DB connection is generally a separate CONN_TRX_AUTOCOMMIT handle.
 204                 // No transaction is active nor will be started by writes, so enqueue the jobs
 205                 // now so that any errors will show up immediately as the interface expects. Any
 206                 // errors that bubble up will rollback the main commit round.
 207                 $fname = __METHOD__;
 208                 $dbw->onTransactionPreCommitOrIdle(
 209                         function ( IDatabase $dbw ) use ( $jobs, $flags, $fname ) {
 210                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $fname );
 211                         },
 212                         $fname
 213                 );
 214         }
 215
 216         /**
 217          * This function should *not* be called outside of JobQueueDB
 218          *
 219          * @suppress SecurityCheck-SQLInjection Bug in phan-taint-check handling bulk inserts
 220          * @param IDatabase $dbw
 221          * @param IJobSpecification[] $jobs
 222          * @param int $flags
 223          * @param string $method
 224          * @throws DBError
 225          * @return void
 226          */
 227         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 228                 if ( $jobs === [] ) {
 229                         return;
 230                 }
 231
 232                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 233                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 234                 foreach ( $jobs as $job ) {
 235                         $row = $this->insertFields( $job, $dbw );
 236                         if ( $job->ignoreDuplicates() ) {
 237                                 $rowSet[$row['job_sha1']] = $row;
 238                         } else {
 239                                 $rowList[] = $row;
 240                         }
 241                 }
 242
 243                 if ( $flags & self::QOS_ATOMIC ) {
 244                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 245                 }
 246                 try {
 247                         // Strip out any duplicate jobs that are already in the queue...
 248                         if ( count( $rowSet ) ) {
 249                                 $res = $dbw->select( 'job', 'job_sha1',
 250                                         [
 251                                                 // No job_type condition since it's part of the job_sha1 hash
 252                                                 'job_sha1' => array_keys( $rowSet ),
 253                                                 'job_token' => '' // unclaimed
 254                                         ],
 255                                         $method
 256                                 );
 257                                 foreach ( $res as $row ) {
 258                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 259                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 260                                 }
 261                         }
 262                         // Build the full list of job rows to insert
 263                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 264                         // Insert the job rows in chunks to avoid replica DB lag...
 265                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 266                                 $dbw->insert( 'job', $rowBatch, $method );
 267                         }
 268                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 269                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 270                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 271                         );
 272                 } catch ( DBError $e ) {
 273                         $this->throwDBException( $e );
 274                 }
 275                 if ( $flags & self::QOS_ATOMIC ) {
 276                         $dbw->endAtomic( $method );
 277                 }
 278         }
 279
 280         /**
 281          * @see JobQueue::doPop()
 282          * @return Job|bool
 283          */
 284         protected function doPop() {
 285                 $dbw = $this->getMasterDB();
 286                 try {
 287                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 288                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 289                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 290                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 291                         } );
 292
 293                         $uuid = wfRandomString( 32 ); // pop attempt
 294                         $job = false; // job popped off
 295                         do { // retry when our row is invalid or deleted as a duplicate
 296                                 // Try to reserve a row in the DB...
 297                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 298                                         $row = $this->claimOldest( $uuid );
 299                                 } else { // random first
 300                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 301                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 302                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 303                                 }
 304                                 // Check if we found a row to reserve...
 305                                 if ( !$row ) {
 306                                         break; // nothing to do
 307                                 }
 308                                 JobQueue::incrStats( 'pops', $this->type );
 309                                 // Get the job object from the row...
 310                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 311                                 $job = Job::factory( $row->job_cmd, $title,
 312                                         self::extractBlob( $row->job_params ) );
 313                                 $job->metadata['id'] = $row->job_id;
 314                                 $job->metadata['timestamp'] = $row->job_timestamp;
 315                                 break; // done
 316                         } while ( true );
 317
 318                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 319                                 // Handled jobs that need to be recycled/deleted;
 320                                 // any recycled jobs will be picked up next attempt
 321                                 $this->recycleAndDeleteStaleJobs();
 322                         }
 323                 } catch ( DBError $e ) {
 324                         $this->throwDBException( $e );
 325                 }
 326
 327                 return $job;
 328         }
 329
 330         /**
 331          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 332          *
 333          * @param string $uuid 32 char hex string
 334          * @param int $rand Random unsigned integer (31 bits)
 335          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 336          * @return stdClass|bool Row|false
 337          */
 338         protected function claimRandom( $uuid, $rand, $gte ) {
 339                 $dbw = $this->getMasterDB();
 340                 // Check cache to see if the queue has <= OFFSET items
 341                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 342
 343                 $row = false; // the row acquired
 344                 $invertedDirection = false; // whether one job_random direction was already scanned
 345                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 346                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 347                 // not replication safe. Due to https://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 348                 // be used here with MySQL.
 349                 do {
 350                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 351                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 352                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 353                                 $ineq = $gte ? '>=' : '<=';
 354                                 $dir = $gte ? 'ASC' : 'DESC';
 355                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 356                                         [
 357                                                 'job_cmd' => $this->type,
 358                                                 'job_token' => '', // unclaimed
 359                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 360                                         __METHOD__,
 361                                         [ 'ORDER BY' => "job_random {$dir}" ]
 362                                 );
 363                                 if ( !$row && !$invertedDirection ) {
 364                                         $gte = !$gte;
 365                                         $invertedDirection = true;
 366                                         continue; // try the other direction
 367                                 }
 368                         } else { // table *may* have >= MAX_OFFSET rows
 369                                 // T44614: "ORDER BY job_random" with a job_random inequality causes high CPU
 370                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 371                                 // instead of job_random for reducing excess claim retries.
 372                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 373                                         [
 374                                                 'job_cmd' => $this->type,
 375                                                 'job_token' => '', // unclaimed
 376                                         ],
 377                                         __METHOD__,
 378                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 379                                 );
 380                                 if ( !$row ) {
 381                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 382                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 383                                         continue; // use job_random
 384                                 }
 385                         }
 386
 387                         if ( $row ) { // claim the job
 388                                 $dbw->update( 'job', // update by PK
 389                                         [
 390                                                 'job_token' => $uuid,
 391                                                 'job_token_timestamp' => $dbw->timestamp(),
 392                                                 'job_attempts = job_attempts+1' ],
 393                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 394                                         __METHOD__
 395                                 );
 396                                 // This might get raced out by another runner when claiming the previously
 397                                 // selected row. The use of job_random should minimize this problem, however.
 398                                 if ( !$dbw->affectedRows() ) {
 399                                         $row = false; // raced out
 400                                 }
 401                         } else {
 402                                 break; // nothing to do
 403                         }
 404                 } while ( !$row );
 405
 406                 return $row;
 407         }
 408
 409         /**
 410          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 411          *
 412          * @param string $uuid 32 char hex string
 413          * @return stdClass|bool Row|false
 414          */
 415         protected function claimOldest( $uuid ) {
 416                 $dbw = $this->getMasterDB();
 417
 418                 $row = false; // the row acquired
 419                 do {
 420                         if ( $dbw->getType() === 'mysql' ) {
 421                                 // Per https://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 422                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 423                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 424                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 425                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 426                                         "SET " .
 427                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 428                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 429                                                 "job_attempts = job_attempts+1 " .
 430                                         "WHERE ( " .
 431                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 432                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 433                                         ") ORDER BY job_id ASC LIMIT 1",
 434                                         __METHOD__
 435                                 );
 436                         } else {
 437                                 // Use a subquery to find the job, within an UPDATE to claim it.
 438                                 // This uses as much of the DB wrapper functions as possible.
 439                                 $dbw->update( 'job',
 440                                         [
 441                                                 'job_token' => $uuid,
 442                                                 'job_token_timestamp' => $dbw->timestamp(),
 443                                                 'job_attempts = job_attempts+1' ],
 444                                         [ 'job_id = (' .
 445                                                 $dbw->selectSQLText( 'job', 'job_id',
 446                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 447                                                         __METHOD__,
 448                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 449                                                 ')'
 450                                         ],
 451                                         __METHOD__
 452                                 );
 453                         }
 454                         // Fetch any row that we just reserved...
 455                         if ( $dbw->affectedRows() ) {
 456                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 457                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 458                                 );
 459                                 if ( !$row ) { // raced out by duplicate job removal
 460                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 461                                 }
 462                         } else {
 463                                 break; // nothing to do
 464                         }
 465                 } while ( !$row );
 466
 467                 return $row;
 468         }
 469
 470         /**
 471          * @see JobQueue::doAck()
 472          * @param Job $job
 473          * @throws MWException
 474          */
 475         protected function doAck( Job $job ) {
 476                 if ( !isset( $job->metadata['id'] ) ) {
 477                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 478                 }
 479
 480                 $dbw = $this->getMasterDB();
 481                 try {
 482                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 483                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 484                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 485                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 486                         } );
 487
 488                         // Delete a row with a single DELETE without holding row locks over RTTs...
 489                         $dbw->delete( 'job',
 490                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 491
 492                         JobQueue::incrStats( 'acks', $this->type );
 493                 } catch ( DBError $e ) {
 494                         $this->throwDBException( $e );
 495                 }
 496         }
 497
 498         /**
 499          * @see JobQueue::doDeduplicateRootJob()
 500          * @param IJobSpecification $job
 501          * @throws MWException
 502          * @return bool
 503          */
 504         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 505                 $params = $job->getParams();
 506                 if ( !isset( $params['rootJobSignature'] ) ) {
 507                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 508                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 509                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 510                 }
 511                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 512                 // Callers should call JobQueueGroup::push() before this method so that if the insert
 513                 // fails, the de-duplication registration will be aborted. Since the insert is
 514                 // deferred till "transaction idle", do the same here, so that the ordering is
 515                 // maintained. Having only the de-duplication registration succeed would cause
 516                 // jobs to become no-ops without any actual jobs that made them redundant.
 517                 $dbw = $this->getMasterDB();
 518                 $cache = $this->dupCache;
 519                 $dbw->onTransactionCommitOrIdle(
 520                         function () use ( $cache, $params, $key ) {
 521                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 522                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 523                                         return true; // a newer version of this root job was enqueued
 524                                 }
 525
 526                                 // Update the timestamp of the last root job started at the location...
 527                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 528                         },
 529                         __METHOD__
 530                 );
 531
 532                 return true;
 533         }
 534
 535         /**
 536          * @see JobQueue::doDelete()
 537          * @return bool
 538          */
 539         protected function doDelete() {
 540                 $dbw = $this->getMasterDB();
 541                 try {
 542                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 543                 } catch ( DBError $e ) {
 544                         $this->throwDBException( $e );
 545                 }
 546
 547                 return true;
 548         }
 549
 550         /**
 551          * @see JobQueue::doWaitForBackups()
 552          * @return void
 553          */
 554         protected function doWaitForBackups() {
 555                 if ( $this->server ) {
 556                         return; // not using LBFactory instance
 557                 }
 558
 559                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 560                 $lbFactory->waitForReplication( [
 561                         'domain' => $this->domain,
 562                         'cluster' => is_string( $this->cluster ) ? $this->cluster : false
 563                 ] );
 564         }
 565
 566         /**
 567          * @return void
 568          */
 569         protected function doFlushCaches() {
 570                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 571                         $this->cache->delete( $this->getCacheKey( $type ) );
 572                 }
 573         }
 574
 575         /**
 576          * @see JobQueue::getAllQueuedJobs()
 577          * @return Iterator
 578          */
 579         public function getAllQueuedJobs() {
 580                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 581         }
 582
 583         /**
 584          * @see JobQueue::getAllAcquiredJobs()
 585          * @return Iterator
 586          */
 587         public function getAllAcquiredJobs() {
 588                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 589         }
 590
 591         /**
 592          * @param array $conds Query conditions
 593          * @return Iterator
 594          */
 595         protected function getJobIterator( array $conds ) {
 596                 $dbr = $this->getReplicaDB();
 597                 try {
 598                         return new MappedIterator(
 599                                 $dbr->select( 'job', self::selectFields(), $conds ),
 600                                 function ( $row ) {
 601                                         $job = Job::factory(
 602                                                 $row->job_cmd,
 603                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 604                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 605                                         );
 606                                         $job->metadata['id'] = $row->job_id;
 607                                         $job->metadata['timestamp'] = $row->job_timestamp;
 608
 609                                         return $job;
 610                                 }
 611                         );
 612                 } catch ( DBError $e ) {
 613                         $this->throwDBException( $e );
 614                 }
 615         }
 616
 617         public function getCoalesceLocationInternal() {
 618                 if ( $this->server ) {
 619                         return null; // not using the LBFactory instance
 620                 }
 621
 622                 return is_string( $this->cluster )
 623                         ? "DBCluster:{$this->cluster}:{$this->domain}"
 624                         : "LBFactory:{$this->domain}";
 625         }
 626
 627         protected function doGetSiblingQueuesWithJobs( array $types ) {
 628                 $dbr = $this->getReplicaDB();
 629                 // @note: this does not check whether the jobs are claimed or not.
 630                 // This is useful so JobQueueGroup::pop() also sees queues that only
 631                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 632                 // failed jobs so that they can be popped again for that edge case.
 633                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 634                         [ 'job_cmd' => $types ], __METHOD__ );
 635
 636                 $types = [];
 637                 foreach ( $res as $row ) {
 638                         $types[] = $row->job_cmd;
 639                 }
 640
 641                 return $types;
 642         }
 643
 644         protected function doGetSiblingQueueSizes( array $types ) {
 645                 $dbr = $this->getReplicaDB();
 646                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 647                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 648
 649                 $sizes = [];
 650                 foreach ( $res as $row ) {
 651                         $sizes[$row->job_cmd] = (int)$row->count;
 652                 }
 653
 654                 return $sizes;
 655         }
 656
 657         /**
 658          * Recycle or destroy any jobs that have been claimed for too long
 659          *
 660          * @return int Number of jobs recycled/deleted
 661          */
 662         public function recycleAndDeleteStaleJobs() {
 663                 $now = time();
 664                 $count = 0; // affected rows
 665                 $dbw = $this->getMasterDB();
 666
 667                 try {
 668                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 669                                 return $count; // already in progress
 670                         }
 671
 672                         // Remove claims on jobs acquired for too long if enabled...
 673                         if ( $this->claimTTL > 0 ) {
 674                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 675                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 676                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 677                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 678                                 $res = $dbw->select( 'job', 'job_id',
 679                                         [
 680                                                 'job_cmd' => $this->type,
 681                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 682                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 683                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 684                                         __METHOD__
 685                                 );
 686                                 $ids = array_map(
 687                                         function ( $o ) {
 688                                                 return $o->job_id;
 689                                         }, iterator_to_array( $res )
 690                                 );
 691                                 if ( count( $ids ) ) {
 692                                         // Reset job_token for these jobs so that other runners will pick them up.
 693                                         // Set the timestamp to the current time, as it is useful to now that the job
 694                                         // was already tried before (the timestamp becomes the "released" time).
 695                                         $dbw->update( 'job',
 696                                                 [
 697                                                         'job_token' => '',
 698                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 699                                                 [
 700                                                         'job_id' => $ids ],
 701                                                 __METHOD__
 702                                         );
 703                                         $affected = $dbw->affectedRows();
 704                                         $count += $affected;
 705                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 706                                         $this->aggr->notifyQueueNonEmpty( $this->domain, $this->type );
 707                                 }
 708                         }
 709
 710                         // Just destroy any stale jobs...
 711                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 712                         $conds = [
 713                                 'job_cmd' => $this->type,
 714                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 715                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 716                         ];
 717                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 718                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 719                         }
 720                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 721                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 722                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 723                         $ids = array_map(
 724                                 function ( $o ) {
 725                                         return $o->job_id;
 726                                 }, iterator_to_array( $res )
 727                         );
 728                         if ( count( $ids ) ) {
 729                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 730                                 $affected = $dbw->affectedRows();
 731                                 $count += $affected;
 732                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 733                         }
 734
 735                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 736                 } catch ( DBError $e ) {
 737                         $this->throwDBException( $e );
 738                 }
 739
 740                 return $count;
 741         }
 742
 743         /**
 744          * @param IJobSpecification $job
 745          * @param IDatabase $db
 746          * @return array
 747          */
 748         protected function insertFields( IJobSpecification $job, IDatabase $db ) {
 749                 return [
 750                         // Fields that describe the nature of the job
 751                         'job_cmd' => $job->getType(),
 752                         'job_namespace' => $job->getTitle()->getNamespace(),
 753                         'job_title' => $job->getTitle()->getDBkey(),
 754                         'job_params' => self::makeBlob( $job->getParams() ),
 755                         // Additional job metadata
 756                         'job_timestamp' => $db->timestamp(),
 757                         'job_sha1' => Wikimedia\base_convert(
 758                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 759                                 16, 36, 31
 760                         ),
 761                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 762                 ];
 763         }
 764
 765         /**
 766          * @throws JobQueueConnectionError
 767          * @return IDatabase
 768          */
 769         protected function getReplicaDB() {
 770                 try {
 771                         return $this->getDB( DB_REPLICA );
 772                 } catch ( DBConnectionError $e ) {
 773                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 774                 }
 775         }
 776
 777         /**
 778          * @throws JobQueueConnectionError
 779          * @return IDatabase
 780          */
 781         protected function getMasterDB() {
 782                 try {
 783                         return $this->getDB( DB_MASTER );
 784                 } catch ( DBConnectionError $e ) {
 785                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 786                 }
 787         }
 788
 789         /**
 790          * @param int $index (DB_REPLICA/DB_MASTER)
 791          * @return IDatabase
 792          */
 793         protected function getDB( $index ) {
 794                 if ( $this->server ) {
 795                         if ( $this->conn instanceof IDatabase ) {
 796                                 return $this->conn;
 797                         } elseif ( $this->conn instanceof DBError ) {
 798                                 throw $this->conn;
 799                         }
 800
 801                         try {
 802                                 $this->conn = Database::factory( $this->server['type'], $this->server );
 803                         } catch ( DBError $e ) {
 804                                 $this->conn = $e;
 805                                 throw $e;
 806                         }
 807
 808                         return $this->conn;
 809                 } else {
 810                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 811                         $lb = is_string( $this->cluster )
 812                                 ? $lbFactory->getExternalLB( $this->cluster )
 813                                 : $lbFactory->getMainLB( $this->domain );
 814
 815                         return ( $lb->getServerType( $lb->getWriterIndex() ) !== 'sqlite' )
 816                                 // Keep a separate connection to avoid contention and deadlocks;
 817                                 // However, SQLite has the opposite behavior due to DB-level locking.
 818                                 ? $lb->getConnectionRef( $index, [], $this->domain, $lb::CONN_TRX_AUTOCOMMIT )
 819                                 // Jobs insertion will be defered until the PRESEND stage to reduce contention.
 820                                 : $lb->getConnectionRef( $index, [], $this->domain );
 821                 }
 822         }
 823
 824         /**
 825          * @param string $property
 826          * @return string
 827          */
 828         private function getCacheKey( $property ) {
 829                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 830
 831                 return $this->cache->makeGlobalKey(
 832                         'jobqueue',
 833                         $this->domain,
 834                         $cluster,
 835                         $this->type,
 836                         $property
 837                 );
 838         }
 839
 840         /**
 841          * @param array|bool $params
 842          * @return string
 843          */
 844         protected static function makeBlob( $params ) {
 845                 if ( $params !== false ) {
 846                         return serialize( $params );
 847                 } else {
 848                         return '';
 849                 }
 850         }
 851
 852         /**
 853          * @param string $blob
 854          * @return bool|mixed
 855          */
 856         protected static function extractBlob( $blob ) {
 857                 if ( (string)$blob !== '' ) {
 858                         return unserialize( $blob );
 859                 } else {
 860                         return false;
 861                 }
 862         }
 863
 864         /**
 865          * @param DBError $e
 866          * @throws JobQueueError
 867          */
 868         protected function throwDBException( DBError $e ) {
 869                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 870         }
 871
 872         /**
 873          * Return the list of job fields that should be selected.
 874          * @since 1.23
 875          * @return array
 876          */
 877         public static function selectFields() {
 878                 return [
 879                         'job_id',
 880                         'job_cmd',
 881                         'job_namespace',
 882                         'job_title',
 883                         'job_timestamp',
 884                         'job_params',
 885                         'job_random',
 886                         'job_attempts',
 887                         'job_token',
 888                         'job_token_timestamp',
 889                         'job_sha1',
 890                 ];
 891         }
 892 }