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                 /** @noinspection PhpUnusedLocalVariableInspection */
  87                 $scope = $this->getScopedNoTrxFlag( $dbr );
  88                 try {
  89                         $found = $dbr->selectField( // unclaimed job
  90                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  91                         );
  92                 } catch ( DBError $e ) {
  93                         $this->throwDBException( $e );
  94                 }
  95
  96                 return !$found;
  97         }
  98
  99         /**
 100          * @see JobQueue::doGetSize()
 101          * @return int
 102          */
 103         protected function doGetSize() {
 104                 $key = $this->getCacheKey( 'size' );
 105
 106                 $size = $this->cache->get( $key );
 107                 if ( is_int( $size ) ) {
 108                         return $size;
 109                 }
 110
 111                 $dbr = $this->getReplicaDB();
 112                 /** @noinspection PhpUnusedLocalVariableInspection */
 113                 $scope = $this->getScopedNoTrxFlag( $dbr );
 114                 try {
 115                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 116                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
 117                                 __METHOD__
 118                         );
 119                 } catch ( DBError $e ) {
 120                         $this->throwDBException( $e );
 121                 }
 122                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 123
 124                 return $size;
 125         }
 126
 127         /**
 128          * @see JobQueue::doGetAcquiredCount()
 129          * @return int
 130          */
 131         protected function doGetAcquiredCount() {
 132                 if ( $this->claimTTL <= 0 ) {
 133                         return 0; // no acknowledgements
 134                 }
 135
 136                 $key = $this->getCacheKey( 'acquiredcount' );
 137
 138                 $count = $this->cache->get( $key );
 139                 if ( is_int( $count ) ) {
 140                         return $count;
 141                 }
 142
 143                 $dbr = $this->getReplicaDB();
 144                 /** @noinspection PhpUnusedLocalVariableInspection */
 145                 $scope = $this->getScopedNoTrxFlag( $dbr );
 146                 try {
 147                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 148                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 149                                 __METHOD__
 150                         );
 151                 } catch ( DBError $e ) {
 152                         $this->throwDBException( $e );
 153                 }
 154                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 155
 156                 return $count;
 157         }
 158
 159         /**
 160          * @see JobQueue::doGetAbandonedCount()
 161          * @return int
 162          * @throws MWException
 163          */
 164         protected function doGetAbandonedCount() {
 165                 if ( $this->claimTTL <= 0 ) {
 166                         return 0; // no acknowledgements
 167                 }
 168
 169                 $key = $this->getCacheKey( 'abandonedcount' );
 170
 171                 $count = $this->cache->get( $key );
 172                 if ( is_int( $count ) ) {
 173                         return $count;
 174                 }
 175
 176                 $dbr = $this->getReplicaDB();
 177                 /** @noinspection PhpUnusedLocalVariableInspection */
 178                 $scope = $this->getScopedNoTrxFlag( $dbr );
 179                 try {
 180                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 181                                 [
 182                                         'job_cmd' => $this->type,
 183                                         "job_token != {$dbr->addQuotes( '' )}",
 184                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 185                                 ],
 186                                 __METHOD__
 187                         );
 188                 } catch ( DBError $e ) {
 189                         $this->throwDBException( $e );
 190                 }
 191
 192                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 193
 194                 return $count;
 195         }
 196
 197         /**
 198          * @see JobQueue::doBatchPush()
 199          * @param IJobSpecification[] $jobs
 200          * @param int $flags
 201          * @throws DBError|Exception
 202          * @return void
 203          */
 204         protected function doBatchPush( array $jobs, $flags ) {
 205                 $dbw = $this->getMasterDB();
 206                 /** @noinspection PhpUnusedLocalVariableInspection */
 207                 $scope = $this->getScopedNoTrxFlag( $dbw );
 208                 // In general, there will be two cases here:
 209                 // a) sqlite; DB connection is probably a regular round-aware handle.
 210                 // If the connection is busy with a transaction, then defer the job writes
 211                 // until right before the main round commit step. Any errors that bubble
 212                 // up will rollback the main commit round.
 213                 // b) mysql/postgres; DB connection is generally a separate CONN_TRX_AUTOCOMMIT handle.
 214                 // No transaction is active nor will be started by writes, so enqueue the jobs
 215                 // now so that any errors will show up immediately as the interface expects. Any
 216                 // errors that bubble up will rollback the main commit round.
 217                 $fname = __METHOD__;
 218                 $dbw->onTransactionPreCommitOrIdle(
 219                         function ( IDatabase $dbw ) use ( $jobs, $flags, $fname ) {
 220                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $fname );
 221                         },
 222                         $fname
 223                 );
 224         }
 225
 226         /**
 227          * This function should *not* be called outside of JobQueueDB
 228          *
 229          * @suppress SecurityCheck-SQLInjection Bug in phan-taint-check handling bulk inserts
 230          * @param IDatabase $dbw
 231          * @param IJobSpecification[] $jobs
 232          * @param int $flags
 233          * @param string $method
 234          * @throws DBError
 235          * @return void
 236          */
 237         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 238                 if ( $jobs === [] ) {
 239                         return;
 240                 }
 241
 242                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 243                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 244                 foreach ( $jobs as $job ) {
 245                         $row = $this->insertFields( $job, $dbw );
 246                         if ( $job->ignoreDuplicates() ) {
 247                                 $rowSet[$row['job_sha1']] = $row;
 248                         } else {
 249                                 $rowList[] = $row;
 250                         }
 251                 }
 252
 253                 if ( $flags & self::QOS_ATOMIC ) {
 254                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 255                 }
 256                 try {
 257                         // Strip out any duplicate jobs that are already in the queue...
 258                         if ( count( $rowSet ) ) {
 259                                 $res = $dbw->select( 'job', 'job_sha1',
 260                                         [
 261                                                 // No job_type condition since it's part of the job_sha1 hash
 262                                                 'job_sha1' => array_keys( $rowSet ),
 263                                                 'job_token' => '' // unclaimed
 264                                         ],
 265                                         $method
 266                                 );
 267                                 foreach ( $res as $row ) {
 268                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 269                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 270                                 }
 271                         }
 272                         // Build the full list of job rows to insert
 273                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 274                         // Insert the job rows in chunks to avoid replica DB lag...
 275                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 276                                 $dbw->insert( 'job', $rowBatch, $method );
 277                         }
 278                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 279                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 280                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 281                         );
 282                 } catch ( DBError $e ) {
 283                         $this->throwDBException( $e );
 284                 }
 285                 if ( $flags & self::QOS_ATOMIC ) {
 286                         $dbw->endAtomic( $method );
 287                 }
 288         }
 289
 290         /**
 291          * @see JobQueue::doPop()
 292          * @return Job|bool
 293          */
 294         protected function doPop() {
 295                 $dbw = $this->getMasterDB();
 296                 /** @noinspection PhpUnusedLocalVariableInspection */
 297                 $scope = $this->getScopedNoTrxFlag( $dbw );
 298
 299                 $job = false; // job popped off
 300                 try {
 301                         $uuid = wfRandomString( 32 ); // pop attempt
 302                         do { // retry when our row is invalid or deleted as a duplicate
 303                                 // Try to reserve a row in the DB...
 304                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 305                                         $row = $this->claimOldest( $uuid );
 306                                 } else { // random first
 307                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 308                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 309                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 310                                 }
 311                                 // Check if we found a row to reserve...
 312                                 if ( !$row ) {
 313                                         break; // nothing to do
 314                                 }
 315                                 JobQueue::incrStats( 'pops', $this->type );
 316                                 // Get the job object from the row...
 317                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 318                                 $job = Job::factory( $row->job_cmd, $title,
 319                                         self::extractBlob( $row->job_params ) );
 320                                 $job->setMetadata( 'id', $row->job_id );
 321                                 $job->setMetadata( 'timestamp', $row->job_timestamp );
 322                                 break; // done
 323                         } while ( true );
 324
 325                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 326                                 // Handled jobs that need to be recycled/deleted;
 327                                 // any recycled jobs will be picked up next attempt
 328                                 $this->recycleAndDeleteStaleJobs();
 329                         }
 330                 } catch ( DBError $e ) {
 331                         $this->throwDBException( $e );
 332                 }
 333
 334                 return $job;
 335         }
 336
 337         /**
 338          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 339          *
 340          * @param string $uuid 32 char hex string
 341          * @param int $rand Random unsigned integer (31 bits)
 342          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 343          * @return stdClass|bool Row|false
 344          */
 345         protected function claimRandom( $uuid, $rand, $gte ) {
 346                 $dbw = $this->getMasterDB();
 347                 /** @noinspection PhpUnusedLocalVariableInspection */
 348                 $scope = $this->getScopedNoTrxFlag( $dbw );
 349                 // Check cache to see if the queue has <= OFFSET items
 350                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 351
 352                 $row = false; // the row acquired
 353                 $invertedDirection = false; // whether one job_random direction was already scanned
 354                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 355                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 356                 // not replication safe. Due to https://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 357                 // be used here with MySQL.
 358                 do {
 359                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 360                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 361                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 362                                 $ineq = $gte ? '>=' : '<=';
 363                                 $dir = $gte ? 'ASC' : 'DESC';
 364                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 365                                         [
 366                                                 'job_cmd' => $this->type,
 367                                                 'job_token' => '', // unclaimed
 368                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 369                                         __METHOD__,
 370                                         [ 'ORDER BY' => "job_random {$dir}" ]
 371                                 );
 372                                 if ( !$row && !$invertedDirection ) {
 373                                         $gte = !$gte;
 374                                         $invertedDirection = true;
 375                                         continue; // try the other direction
 376                                 }
 377                         } else { // table *may* have >= MAX_OFFSET rows
 378                                 // T44614: "ORDER BY job_random" with a job_random inequality causes high CPU
 379                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 380                                 // instead of job_random for reducing excess claim retries.
 381                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 382                                         [
 383                                                 'job_cmd' => $this->type,
 384                                                 'job_token' => '', // unclaimed
 385                                         ],
 386                                         __METHOD__,
 387                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 388                                 );
 389                                 if ( !$row ) {
 390                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 391                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 392                                         continue; // use job_random
 393                                 }
 394                         }
 395
 396                         if ( $row ) { // claim the job
 397                                 $dbw->update( 'job', // update by PK
 398                                         [
 399                                                 'job_token' => $uuid,
 400                                                 'job_token_timestamp' => $dbw->timestamp(),
 401                                                 'job_attempts = job_attempts+1' ],
 402                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 403                                         __METHOD__
 404                                 );
 405                                 // This might get raced out by another runner when claiming the previously
 406                                 // selected row. The use of job_random should minimize this problem, however.
 407                                 if ( !$dbw->affectedRows() ) {
 408                                         $row = false; // raced out
 409                                 }
 410                         } else {
 411                                 break; // nothing to do
 412                         }
 413                 } while ( !$row );
 414
 415                 return $row;
 416         }
 417
 418         /**
 419          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 420          *
 421          * @param string $uuid 32 char hex string
 422          * @return stdClass|bool Row|false
 423          */
 424         protected function claimOldest( $uuid ) {
 425                 $dbw = $this->getMasterDB();
 426                 /** @noinspection PhpUnusedLocalVariableInspection */
 427                 $scope = $this->getScopedNoTrxFlag( $dbw );
 428
 429                 $row = false; // the row acquired
 430                 do {
 431                         if ( $dbw->getType() === 'mysql' ) {
 432                                 // Per https://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 433                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 434                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 435                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 436                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 437                                         "SET " .
 438                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 439                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 440                                                 "job_attempts = job_attempts+1 " .
 441                                         "WHERE ( " .
 442                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 443                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 444                                         ") ORDER BY job_id ASC LIMIT 1",
 445                                         __METHOD__
 446                                 );
 447                         } else {
 448                                 // Use a subquery to find the job, within an UPDATE to claim it.
 449                                 // This uses as much of the DB wrapper functions as possible.
 450                                 $dbw->update( 'job',
 451                                         [
 452                                                 'job_token' => $uuid,
 453                                                 'job_token_timestamp' => $dbw->timestamp(),
 454                                                 'job_attempts = job_attempts+1' ],
 455                                         [ 'job_id = (' .
 456                                                 $dbw->selectSQLText( 'job', 'job_id',
 457                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 458                                                         __METHOD__,
 459                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 460                                                 ')'
 461                                         ],
 462                                         __METHOD__
 463                                 );
 464                         }
 465                         // Fetch any row that we just reserved...
 466                         if ( $dbw->affectedRows() ) {
 467                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 468                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 469                                 );
 470                                 if ( !$row ) { // raced out by duplicate job removal
 471                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 472                                 }
 473                         } else {
 474                                 break; // nothing to do
 475                         }
 476                 } while ( !$row );
 477
 478                 return $row;
 479         }
 480
 481         /**
 482          * @see JobQueue::doAck()
 483          * @param Job $job
 484          * @throws MWException
 485          */
 486         protected function doAck( Job $job ) {
 487                 $id = $job->getMetadata( 'id' );
 488                 if ( $id === null ) {
 489                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 490                 }
 491
 492                 $dbw = $this->getMasterDB();
 493                 /** @noinspection PhpUnusedLocalVariableInspection */
 494                 $scope = $this->getScopedNoTrxFlag( $dbw );
 495                 try {
 496                         // Delete a row with a single DELETE without holding row locks over RTTs...
 497                         $dbw->delete(
 498                                 'job',
 499                                 [ 'job_cmd' => $this->type, 'job_id' => $id ],
 500                                 __METHOD__
 501                         );
 502
 503                         JobQueue::incrStats( 'acks', $this->type );
 504                 } catch ( DBError $e ) {
 505                         $this->throwDBException( $e );
 506                 }
 507         }
 508
 509         /**
 510          * @see JobQueue::doDeduplicateRootJob()
 511          * @param IJobSpecification $job
 512          * @throws MWException
 513          * @return bool
 514          */
 515         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 516                 $params = $job->getParams();
 517                 if ( !isset( $params['rootJobSignature'] ) ) {
 518                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 519                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 520                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 521                 }
 522                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 523                 // Callers should call JobQueueGroup::push() before this method so that if the insert
 524                 // fails, the de-duplication registration will be aborted. Since the insert is
 525                 // deferred till "transaction idle", do the same here, so that the ordering is
 526                 // maintained. Having only the de-duplication registration succeed would cause
 527                 // jobs to become no-ops without any actual jobs that made them redundant.
 528                 $dbw = $this->getMasterDB();
 529                 /** @noinspection PhpUnusedLocalVariableInspection */
 530                 $scope = $this->getScopedNoTrxFlag( $dbw );
 531
 532                 $cache = $this->dupCache;
 533                 $dbw->onTransactionCommitOrIdle(
 534                         function () use ( $cache, $params, $key ) {
 535                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 536                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 537                                         return true; // a newer version of this root job was enqueued
 538                                 }
 539
 540                                 // Update the timestamp of the last root job started at the location...
 541                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 542                         },
 543                         __METHOD__
 544                 );
 545
 546                 return true;
 547         }
 548
 549         /**
 550          * @see JobQueue::doDelete()
 551          * @return bool
 552          */
 553         protected function doDelete() {
 554                 $dbw = $this->getMasterDB();
 555                 /** @noinspection PhpUnusedLocalVariableInspection */
 556                 $scope = $this->getScopedNoTrxFlag( $dbw );
 557                 try {
 558                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 559                 } catch ( DBError $e ) {
 560                         $this->throwDBException( $e );
 561                 }
 562
 563                 return true;
 564         }
 565
 566         /**
 567          * @see JobQueue::doWaitForBackups()
 568          * @return void
 569          */
 570         protected function doWaitForBackups() {
 571                 if ( $this->server ) {
 572                         return; // not using LBFactory instance
 573                 }
 574
 575                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 576                 $lbFactory->waitForReplication( [
 577                         'domain' => $this->domain,
 578                         'cluster' => is_string( $this->cluster ) ? $this->cluster : false
 579                 ] );
 580         }
 581
 582         /**
 583          * @return void
 584          */
 585         protected function doFlushCaches() {
 586                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 587                         $this->cache->delete( $this->getCacheKey( $type ) );
 588                 }
 589         }
 590
 591         /**
 592          * @see JobQueue::getAllQueuedJobs()
 593          * @return Iterator
 594          */
 595         public function getAllQueuedJobs() {
 596                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 597         }
 598
 599         /**
 600          * @see JobQueue::getAllAcquiredJobs()
 601          * @return Iterator
 602          */
 603         public function getAllAcquiredJobs() {
 604                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 605         }
 606
 607         /**
 608          * @param array $conds Query conditions
 609          * @return Iterator
 610          */
 611         protected function getJobIterator( array $conds ) {
 612                 $dbr = $this->getReplicaDB();
 613                 /** @noinspection PhpUnusedLocalVariableInspection */
 614                 $scope = $this->getScopedNoTrxFlag( $dbr );
 615                 try {
 616                         return new MappedIterator(
 617                                 $dbr->select( 'job', self::selectFields(), $conds ),
 618                                 function ( $row ) {
 619                                         $job = Job::factory(
 620                                                 $row->job_cmd,
 621                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 622                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 623                                         );
 624                                         $job->setMetadata( 'id', $row->job_id );
 625                                         $job->setMetadata( 'timestamp', $row->job_timestamp );
 626
 627                                         return $job;
 628                                 }
 629                         );
 630                 } catch ( DBError $e ) {
 631                         $this->throwDBException( $e );
 632                 }
 633         }
 634
 635         public function getCoalesceLocationInternal() {
 636                 if ( $this->server ) {
 637                         return null; // not using the LBFactory instance
 638                 }
 639
 640                 return is_string( $this->cluster )
 641                         ? "DBCluster:{$this->cluster}:{$this->domain}"
 642                         : "LBFactory:{$this->domain}";
 643         }
 644
 645         protected function doGetSiblingQueuesWithJobs( array $types ) {
 646                 $dbr = $this->getReplicaDB();
 647                 /** @noinspection PhpUnusedLocalVariableInspection */
 648                 $scope = $this->getScopedNoTrxFlag( $dbr );
 649                 // @note: this does not check whether the jobs are claimed or not.
 650                 // This is useful so JobQueueGroup::pop() also sees queues that only
 651                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 652                 // failed jobs so that they can be popped again for that edge case.
 653                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 654                         [ 'job_cmd' => $types ], __METHOD__ );
 655
 656                 $types = [];
 657                 foreach ( $res as $row ) {
 658                         $types[] = $row->job_cmd;
 659                 }
 660
 661                 return $types;
 662         }
 663
 664         protected function doGetSiblingQueueSizes( array $types ) {
 665                 $dbr = $this->getReplicaDB();
 666                 /** @noinspection PhpUnusedLocalVariableInspection */
 667                 $scope = $this->getScopedNoTrxFlag( $dbr );
 668
 669                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 670                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 671
 672                 $sizes = [];
 673                 foreach ( $res as $row ) {
 674                         $sizes[$row->job_cmd] = (int)$row->count;
 675                 }
 676
 677                 return $sizes;
 678         }
 679
 680         /**
 681          * Recycle or destroy any jobs that have been claimed for too long
 682          *
 683          * @return int Number of jobs recycled/deleted
 684          */
 685         public function recycleAndDeleteStaleJobs() {
 686                 $now = time();
 687                 $count = 0; // affected rows
 688                 $dbw = $this->getMasterDB();
 689                 /** @noinspection PhpUnusedLocalVariableInspection */
 690                 $scope = $this->getScopedNoTrxFlag( $dbw );
 691
 692                 try {
 693                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 694                                 return $count; // already in progress
 695                         }
 696
 697                         // Remove claims on jobs acquired for too long if enabled...
 698                         if ( $this->claimTTL > 0 ) {
 699                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 700                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 701                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 702                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 703                                 $res = $dbw->select( 'job', 'job_id',
 704                                         [
 705                                                 'job_cmd' => $this->type,
 706                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 707                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 708                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 709                                         __METHOD__
 710                                 );
 711                                 $ids = array_map(
 712                                         function ( $o ) {
 713                                                 return $o->job_id;
 714                                         }, iterator_to_array( $res )
 715                                 );
 716                                 if ( count( $ids ) ) {
 717                                         // Reset job_token for these jobs so that other runners will pick them up.
 718                                         // Set the timestamp to the current time, as it is useful to now that the job
 719                                         // was already tried before (the timestamp becomes the "released" time).
 720                                         $dbw->update( 'job',
 721                                                 [
 722                                                         'job_token' => '',
 723                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 724                                                 [
 725                                                         'job_id' => $ids ],
 726                                                 __METHOD__
 727                                         );
 728                                         $affected = $dbw->affectedRows();
 729                                         $count += $affected;
 730                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 731                                 }
 732                         }
 733
 734                         // Just destroy any stale jobs...
 735                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 736                         $conds = [
 737                                 'job_cmd' => $this->type,
 738                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 739                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 740                         ];
 741                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 742                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 743                         }
 744                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 745                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 746                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 747                         $ids = array_map(
 748                                 function ( $o ) {
 749                                         return $o->job_id;
 750                                 }, iterator_to_array( $res )
 751                         );
 752                         if ( count( $ids ) ) {
 753                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 754                                 $affected = $dbw->affectedRows();
 755                                 $count += $affected;
 756                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 757                         }
 758
 759                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 760                 } catch ( DBError $e ) {
 761                         $this->throwDBException( $e );
 762                 }
 763
 764                 return $count;
 765         }
 766
 767         /**
 768          * @param IJobSpecification $job
 769          * @param IDatabase $db
 770          * @return array
 771          */
 772         protected function insertFields( IJobSpecification $job, IDatabase $db ) {
 773                 return [
 774                         // Fields that describe the nature of the job
 775                         'job_cmd' => $job->getType(),
 776                         'job_namespace' => $job->getTitle()->getNamespace(),
 777                         'job_title' => $job->getTitle()->getDBkey(),
 778                         'job_params' => self::makeBlob( $job->getParams() ),
 779                         // Additional job metadata
 780                         'job_timestamp' => $db->timestamp(),
 781                         'job_sha1' => Wikimedia\base_convert(
 782                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 783                                 16, 36, 31
 784                         ),
 785                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 786                 ];
 787         }
 788
 789         /**
 790          * @throws JobQueueConnectionError
 791          * @return IDatabase
 792          */
 793         protected function getReplicaDB() {
 794                 try {
 795                         return $this->getDB( DB_REPLICA );
 796                 } catch ( DBConnectionError $e ) {
 797                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 798                 }
 799         }
 800
 801         /**
 802          * @throws JobQueueConnectionError
 803          * @return IDatabase
 804          */
 805         protected function getMasterDB() {
 806                 try {
 807                         return $this->getDB( DB_MASTER );
 808                 } catch ( DBConnectionError $e ) {
 809                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 810                 }
 811         }
 812
 813         /**
 814          * @param int $index (DB_REPLICA/DB_MASTER)
 815          * @return IDatabase
 816          */
 817         protected function getDB( $index ) {
 818                 if ( $this->server ) {
 819                         if ( $this->conn instanceof IDatabase ) {
 820                                 return $this->conn;
 821                         } elseif ( $this->conn instanceof DBError ) {
 822                                 throw $this->conn;
 823                         }
 824
 825                         try {
 826                                 $this->conn = Database::factory( $this->server['type'], $this->server );
 827                         } catch ( DBError $e ) {
 828                                 $this->conn = $e;
 829                                 throw $e;
 830                         }
 831
 832                         return $this->conn;
 833                 } else {
 834                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 835                         $lb = is_string( $this->cluster )
 836                                 ? $lbFactory->getExternalLB( $this->cluster )
 837                                 : $lbFactory->getMainLB( $this->domain );
 838
 839                         return ( $lb->getServerType( $lb->getWriterIndex() ) !== 'sqlite' )
 840                                 // Keep a separate connection to avoid contention and deadlocks;
 841                                 // However, SQLite has the opposite behavior due to DB-level locking.
 842                                 ? $lb->getConnectionRef( $index, [], $this->domain, $lb::CONN_TRX_AUTOCOMMIT )
 843                                 // Jobs insertion will be defered until the PRESEND stage to reduce contention.
 844                                 : $lb->getConnectionRef( $index, [], $this->domain );
 845                 }
 846         }
 847
 848         /**
 849          * @param IDatabase $db
 850          * @return ScopedCallback
 851          */
 852         private function getScopedNoTrxFlag( IDatabase $db ) {
 853                 $autoTrx = $db->getFlag( DBO_TRX ); // get current setting
 854                 $db->clearFlag( DBO_TRX ); // make each query its own transaction
 855
 856                 return new ScopedCallback( function () use ( $db, $autoTrx ) {
 857                         if ( $autoTrx ) {
 858                                 $db->setFlag( DBO_TRX ); // restore old setting
 859                         }
 860                 } );
 861         }
 862
 863         /**
 864          * @param string $property
 865          * @return string
 866          */
 867         private function getCacheKey( $property ) {
 868                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 869
 870                 return $this->cache->makeGlobalKey(
 871                         'jobqueue',
 872                         $this->domain,
 873                         $cluster,
 874                         $this->type,
 875                         $property
 876                 );
 877         }
 878
 879         /**
 880          * @param array|bool $params
 881          * @return string
 882          */
 883         protected static function makeBlob( $params ) {
 884                 if ( $params !== false ) {
 885                         return serialize( $params );
 886                 } else {
 887                         return '';
 888                 }
 889         }
 890
 891         /**
 892          * @param string $blob
 893          * @return bool|mixed
 894          */
 895         protected static function extractBlob( $blob ) {
 896                 if ( (string)$blob !== '' ) {
 897                         return unserialize( $blob );
 898                 } else {
 899                         return false;
 900                 }
 901         }
 902
 903         /**
 904          * @param DBError $e
 905          * @throws JobQueueError
 906          */
 907         protected function throwDBException( DBError $e ) {
 908                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 909         }
 910
 911         /**
 912          * Return the list of job fields that should be selected.
 913          * @since 1.23
 914          * @return array
 915          */
 916         public static function selectFields() {
 917                 return [
 918                         'job_id',
 919                         'job_cmd',
 920                         'job_namespace',
 921                         'job_title',
 922                         'job_timestamp',
 923                         'job_params',
 924                         'job_random',
 925                         'job_attempts',
 926                         'job_token',
 927                         'job_token_timestamp',
 928                         'job_sha1',
 929                 ];
 930         }
 931 }