includes/jobqueue/JobQueueDB.php

   1 <?php
   2 /**
   3  * Database-backed job queue code.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22 use Wikimedia\Rdbms\IDatabase;
  23 use Wikimedia\Rdbms\DBConnRef;
  24 use Wikimedia\Rdbms\DBConnectionError;
  25 use Wikimedia\Rdbms\DBError;
  26 use MediaWiki\MediaWikiServices;
  27 use Wikimedia\ScopedCallback;
  28
  29 /**
  30  * Class to handle job queues stored in the DB
  31  *
  32  * @ingroup JobQueue
  33  * @since 1.21
  34  */
  35 class JobQueueDB extends JobQueue {
  36         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  37         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  38         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  39         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  40
  41         /** @var WANObjectCache */
  42         protected $cache;
  43
  44         /** @var bool|string Name of an external DB cluster. False if not set */
  45         protected $cluster = false;
  46
  47         /**
  48          * Additional parameters include:
  49          *   - cluster : The name of an external cluster registered via LBFactory.
  50          *               If not specified, the primary DB cluster for the wiki will be used.
  51          *               This can be overridden with a custom cluster so that DB handles will
  52          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  53          * @param array $params
  54          */
  55         protected function __construct( array $params ) {
  56                 parent::__construct( $params );
  57
  58                 $this->cluster = $params['cluster'] ?? false;
  59                 $this->cache = ObjectCache::getMainWANInstance();
  60         }
  61
  62         protected function supportedOrders() {
  63                 return [ 'random', 'timestamp', 'fifo' ];
  64         }
  65
  66         protected function optimalOrder() {
  67                 return 'random';
  68         }
  69
  70         /**
  71          * @see JobQueue::doIsEmpty()
  72          * @return bool
  73          */
  74         protected function doIsEmpty() {
  75                 $dbr = $this->getReplicaDB();
  76                 try {
  77                         $found = $dbr->selectField( // unclaimed job
  78                                 'job', '1', [ 'job_cmd' => $this->type, 'job_token' => '' ], __METHOD__
  79                         );
  80                 } catch ( DBError $e ) {
  81                         $this->throwDBException( $e );
  82                 }
  83
  84                 return !$found;
  85         }
  86
  87         /**
  88          * @see JobQueue::doGetSize()
  89          * @return int
  90          */
  91         protected function doGetSize() {
  92                 $key = $this->getCacheKey( 'size' );
  93
  94                 $size = $this->cache->get( $key );
  95                 if ( is_int( $size ) ) {
  96                         return $size;
  97                 }
  98
  99                 try {
 100                         $dbr = $this->getReplicaDB();
 101                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 102                                 [ 'job_cmd' => $this->type, 'job_token' => '' ],
 103                                 __METHOD__
 104                         );
 105                 } catch ( DBError $e ) {
 106                         $this->throwDBException( $e );
 107                 }
 108                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 109
 110                 return $size;
 111         }
 112
 113         /**
 114          * @see JobQueue::doGetAcquiredCount()
 115          * @return int
 116          */
 117         protected function doGetAcquiredCount() {
 118                 if ( $this->claimTTL <= 0 ) {
 119                         return 0; // no acknowledgements
 120                 }
 121
 122                 $key = $this->getCacheKey( 'acquiredcount' );
 123
 124                 $count = $this->cache->get( $key );
 125                 if ( is_int( $count ) ) {
 126                         return $count;
 127                 }
 128
 129                 $dbr = $this->getReplicaDB();
 130                 try {
 131                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 132                                 [ 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ],
 133                                 __METHOD__
 134                         );
 135                 } catch ( DBError $e ) {
 136                         $this->throwDBException( $e );
 137                 }
 138                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 139
 140                 return $count;
 141         }
 142
 143         /**
 144          * @see JobQueue::doGetAbandonedCount()
 145          * @return int
 146          * @throws MWException
 147          */
 148         protected function doGetAbandonedCount() {
 149                 if ( $this->claimTTL <= 0 ) {
 150                         return 0; // no acknowledgements
 151                 }
 152
 153                 $key = $this->getCacheKey( 'abandonedcount' );
 154
 155                 $count = $this->cache->get( $key );
 156                 if ( is_int( $count ) ) {
 157                         return $count;
 158                 }
 159
 160                 $dbr = $this->getReplicaDB();
 161                 try {
 162                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 163                                 [
 164                                         'job_cmd' => $this->type,
 165                                         "job_token != {$dbr->addQuotes( '' )}",
 166                                         "job_attempts >= " . $dbr->addQuotes( $this->maxTries )
 167                                 ],
 168                                 __METHOD__
 169                         );
 170                 } catch ( DBError $e ) {
 171                         $this->throwDBException( $e );
 172                 }
 173
 174                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 175
 176                 return $count;
 177         }
 178
 179         /**
 180          * @see JobQueue::doBatchPush()
 181          * @param IJobSpecification[] $jobs
 182          * @param int $flags
 183          * @throws DBError|Exception
 184          * @return void
 185          */
 186         protected function doBatchPush( array $jobs, $flags ) {
 187                 $dbw = $this->getMasterDB();
 188                 // In general, there will be two cases here:
 189                 // a) sqlite; DB connection is probably a regular round-aware handle.
 190                 // If the connection is busy with a transaction, then defer the job writes
 191                 // until right before the main round commit step. Any errors that bubble
 192                 // up will rollback the main commit round.
 193                 // b) mysql/postgres; DB connection is generally a separate CONN_TRX_AUTOCOMMIT handle.
 194                 // No transaction is active nor will be started by writes, so enqueue the jobs
 195                 // now so that any errors will show up immediately as the interface expects. Any
 196                 // errors that bubble up will rollback the main commit round.
 197                 $fname = __METHOD__;
 198                 $dbw->onTransactionPreCommitOrIdle(
 199                         function ( IDatabase $dbw ) use ( $jobs, $flags, $fname ) {
 200                                 $this->doBatchPushInternal( $dbw, $jobs, $flags, $fname );
 201                         },
 202                         $fname
 203                 );
 204         }
 205
 206         /**
 207          * This function should *not* be called outside of JobQueueDB
 208          *
 209          * @param IDatabase $dbw
 210          * @param IJobSpecification[] $jobs
 211          * @param int $flags
 212          * @param string $method
 213          * @throws DBError
 214          * @return void
 215          */
 216         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 217                 if ( !count( $jobs ) ) {
 218                         return;
 219                 }
 220
 221                 $rowSet = []; // (sha1 => job) map for jobs that are de-duplicated
 222                 $rowList = []; // list of jobs for jobs that are not de-duplicated
 223                 foreach ( $jobs as $job ) {
 224                         $row = $this->insertFields( $job, $dbw );
 225                         if ( $job->ignoreDuplicates() ) {
 226                                 $rowSet[$row['job_sha1']] = $row;
 227                         } else {
 228                                 $rowList[] = $row;
 229                         }
 230                 }
 231
 232                 if ( $flags & self::QOS_ATOMIC ) {
 233                         $dbw->startAtomic( $method ); // wrap all the job additions in one transaction
 234                 }
 235                 try {
 236                         // Strip out any duplicate jobs that are already in the queue...
 237                         if ( count( $rowSet ) ) {
 238                                 $res = $dbw->select( 'job', 'job_sha1',
 239                                         [
 240                                                 // No job_type condition since it's part of the job_sha1 hash
 241                                                 'job_sha1' => array_keys( $rowSet ),
 242                                                 'job_token' => '' // unclaimed
 243                                         ],
 244                                         $method
 245                                 );
 246                                 foreach ( $res as $row ) {
 247                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 248                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 249                                 }
 250                         }
 251                         // Build the full list of job rows to insert
 252                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 253                         // Insert the job rows in chunks to avoid replica DB lag...
 254                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 255                                 $dbw->insert( 'job', $rowBatch, $method );
 256                         }
 257                         JobQueue::incrStats( 'inserts', $this->type, count( $rows ) );
 258                         JobQueue::incrStats( 'dupe_inserts', $this->type,
 259                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 260                         );
 261                 } catch ( DBError $e ) {
 262                         $this->throwDBException( $e );
 263                 }
 264                 if ( $flags & self::QOS_ATOMIC ) {
 265                         $dbw->endAtomic( $method );
 266                 }
 267         }
 268
 269         /**
 270          * @see JobQueue::doPop()
 271          * @return Job|bool
 272          */
 273         protected function doPop() {
 274                 $dbw = $this->getMasterDB();
 275                 try {
 276                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 277                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 278                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 279                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 280                         } );
 281
 282                         $uuid = wfRandomString( 32 ); // pop attempt
 283                         $job = false; // job popped off
 284                         do { // retry when our row is invalid or deleted as a duplicate
 285                                 // Try to reserve a row in the DB...
 286                                 if ( in_array( $this->order, [ 'fifo', 'timestamp' ] ) ) {
 287                                         $row = $this->claimOldest( $uuid );
 288                                 } else { // random first
 289                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 290                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 291                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 292                                 }
 293                                 // Check if we found a row to reserve...
 294                                 if ( !$row ) {
 295                                         break; // nothing to do
 296                                 }
 297                                 JobQueue::incrStats( 'pops', $this->type );
 298                                 // Get the job object from the row...
 299                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 300                                 $job = Job::factory( $row->job_cmd, $title,
 301                                         self::extractBlob( $row->job_params ), $row->job_id );
 302                                 $job->metadata['id'] = $row->job_id;
 303                                 $job->metadata['timestamp'] = $row->job_timestamp;
 304                                 break; // done
 305                         } while ( true );
 306
 307                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 308                                 // Handled jobs that need to be recycled/deleted;
 309                                 // any recycled jobs will be picked up next attempt
 310                                 $this->recycleAndDeleteStaleJobs();
 311                         }
 312                 } catch ( DBError $e ) {
 313                         $this->throwDBException( $e );
 314                 }
 315
 316                 return $job;
 317         }
 318
 319         /**
 320          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 321          *
 322          * @param string $uuid 32 char hex string
 323          * @param int $rand Random unsigned integer (31 bits)
 324          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 325          * @return stdClass|bool Row|false
 326          */
 327         protected function claimRandom( $uuid, $rand, $gte ) {
 328                 $dbw = $this->getMasterDB();
 329                 // Check cache to see if the queue has <= OFFSET items
 330                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 331
 332                 $row = false; // the row acquired
 333                 $invertedDirection = false; // whether one job_random direction was already scanned
 334                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 335                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 336                 // not replication safe. Due to https://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 337                 // be used here with MySQL.
 338                 do {
 339                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 340                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 341                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 342                                 $ineq = $gte ? '>=' : '<=';
 343                                 $dir = $gte ? 'ASC' : 'DESC';
 344                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 345                                         [
 346                                                 'job_cmd' => $this->type,
 347                                                 'job_token' => '', // unclaimed
 348                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ],
 349                                         __METHOD__,
 350                                         [ 'ORDER BY' => "job_random {$dir}" ]
 351                                 );
 352                                 if ( !$row && !$invertedDirection ) {
 353                                         $gte = !$gte;
 354                                         $invertedDirection = true;
 355                                         continue; // try the other direction
 356                                 }
 357                         } else { // table *may* have >= MAX_OFFSET rows
 358                                 // T44614: "ORDER BY job_random" with a job_random inequality causes high CPU
 359                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 360                                 // instead of job_random for reducing excess claim retries.
 361                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 362                                         [
 363                                                 'job_cmd' => $this->type,
 364                                                 'job_token' => '', // unclaimed
 365                                         ],
 366                                         __METHOD__,
 367                                         [ 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) ]
 368                                 );
 369                                 if ( !$row ) {
 370                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 371                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 372                                         continue; // use job_random
 373                                 }
 374                         }
 375
 376                         if ( $row ) { // claim the job
 377                                 $dbw->update( 'job', // update by PK
 378                                         [
 379                                                 'job_token' => $uuid,
 380                                                 'job_token_timestamp' => $dbw->timestamp(),
 381                                                 'job_attempts = job_attempts+1' ],
 382                                         [ 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ],
 383                                         __METHOD__
 384                                 );
 385                                 // This might get raced out by another runner when claiming the previously
 386                                 // selected row. The use of job_random should minimize this problem, however.
 387                                 if ( !$dbw->affectedRows() ) {
 388                                         $row = false; // raced out
 389                                 }
 390                         } else {
 391                                 break; // nothing to do
 392                         }
 393                 } while ( !$row );
 394
 395                 return $row;
 396         }
 397
 398         /**
 399          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 400          *
 401          * @param string $uuid 32 char hex string
 402          * @return stdClass|bool Row|false
 403          */
 404         protected function claimOldest( $uuid ) {
 405                 $dbw = $this->getMasterDB();
 406
 407                 $row = false; // the row acquired
 408                 do {
 409                         if ( $dbw->getType() === 'mysql' ) {
 410                                 // Per https://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 411                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 412                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 413                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 414                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 415                                         "SET " .
 416                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 417                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 418                                                 "job_attempts = job_attempts+1 " .
 419                                         "WHERE ( " .
 420                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 421                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 422                                         ") ORDER BY job_id ASC LIMIT 1",
 423                                         __METHOD__
 424                                 );
 425                         } else {
 426                                 // Use a subquery to find the job, within an UPDATE to claim it.
 427                                 // This uses as much of the DB wrapper functions as possible.
 428                                 $dbw->update( 'job',
 429                                         [
 430                                                 'job_token' => $uuid,
 431                                                 'job_token_timestamp' => $dbw->timestamp(),
 432                                                 'job_attempts = job_attempts+1' ],
 433                                         [ 'job_id = (' .
 434                                                 $dbw->selectSQLText( 'job', 'job_id',
 435                                                         [ 'job_cmd' => $this->type, 'job_token' => '' ],
 436                                                         __METHOD__,
 437                                                         [ 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ] ) .
 438                                                 ')'
 439                                         ],
 440                                         __METHOD__
 441                                 );
 442                         }
 443                         // Fetch any row that we just reserved...
 444                         if ( $dbw->affectedRows() ) {
 445                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 446                                         [ 'job_cmd' => $this->type, 'job_token' => $uuid ], __METHOD__
 447                                 );
 448                                 if ( !$row ) { // raced out by duplicate job removal
 449                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 450                                 }
 451                         } else {
 452                                 break; // nothing to do
 453                         }
 454                 } while ( !$row );
 455
 456                 return $row;
 457         }
 458
 459         /**
 460          * @see JobQueue::doAck()
 461          * @param Job $job
 462          * @throws MWException
 463          */
 464         protected function doAck( Job $job ) {
 465                 if ( !isset( $job->metadata['id'] ) ) {
 466                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 467                 }
 468
 469                 $dbw = $this->getMasterDB();
 470                 try {
 471                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 472                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 473                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 474                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 475                         } );
 476
 477                         // Delete a row with a single DELETE without holding row locks over RTTs...
 478                         $dbw->delete( 'job',
 479                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 480
 481                         JobQueue::incrStats( 'acks', $this->type );
 482                 } catch ( DBError $e ) {
 483                         $this->throwDBException( $e );
 484                 }
 485         }
 486
 487         /**
 488          * @see JobQueue::doDeduplicateRootJob()
 489          * @param IJobSpecification $job
 490          * @throws MWException
 491          * @return bool
 492          */
 493         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 494                 $params = $job->getParams();
 495                 if ( !isset( $params['rootJobSignature'] ) ) {
 496                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 497                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 498                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 499                 }
 500                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 501                 // Callers should call JobQueueGroup::push() before this method so that if the insert
 502                 // fails, the de-duplication registration will be aborted. Since the insert is
 503                 // deferred till "transaction idle", do the same here, so that the ordering is
 504                 // maintained. Having only the de-duplication registration succeed would cause
 505                 // jobs to become no-ops without any actual jobs that made them redundant.
 506                 $dbw = $this->getMasterDB();
 507                 $cache = $this->dupCache;
 508                 $dbw->onTransactionCommitOrIdle(
 509                         function () use ( $cache, $params, $key ) {
 510                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 511                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 512                                         return true; // a newer version of this root job was enqueued
 513                                 }
 514
 515                                 // Update the timestamp of the last root job started at the location...
 516                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 517                         },
 518                         __METHOD__
 519                 );
 520
 521                 return true;
 522         }
 523
 524         /**
 525          * @see JobQueue::doDelete()
 526          * @return bool
 527          */
 528         protected function doDelete() {
 529                 $dbw = $this->getMasterDB();
 530                 try {
 531                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 532                 } catch ( DBError $e ) {
 533                         $this->throwDBException( $e );
 534                 }
 535
 536                 return true;
 537         }
 538
 539         /**
 540          * @see JobQueue::doWaitForBackups()
 541          * @return void
 542          */
 543         protected function doWaitForBackups() {
 544                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 545                 $lbFactory->waitForReplication(
 546                         [ 'domain' => $this->domain, 'cluster' => $this->cluster ] );
 547         }
 548
 549         /**
 550          * @return void
 551          */
 552         protected function doFlushCaches() {
 553                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 554                         $this->cache->delete( $this->getCacheKey( $type ) );
 555                 }
 556         }
 557
 558         /**
 559          * @see JobQueue::getAllQueuedJobs()
 560          * @return Iterator
 561          */
 562         public function getAllQueuedJobs() {
 563                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 564         }
 565
 566         /**
 567          * @see JobQueue::getAllAcquiredJobs()
 568          * @return Iterator
 569          */
 570         public function getAllAcquiredJobs() {
 571                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 572         }
 573
 574         /**
 575          * @param array $conds Query conditions
 576          * @return Iterator
 577          */
 578         protected function getJobIterator( array $conds ) {
 579                 $dbr = $this->getReplicaDB();
 580                 try {
 581                         return new MappedIterator(
 582                                 $dbr->select( 'job', self::selectFields(), $conds ),
 583                                 function ( $row ) {
 584                                         $job = Job::factory(
 585                                                 $row->job_cmd,
 586                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 587                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 588                                         );
 589                                         $job->metadata['id'] = $row->job_id;
 590                                         $job->metadata['timestamp'] = $row->job_timestamp;
 591
 592                                         return $job;
 593                                 }
 594                         );
 595                 } catch ( DBError $e ) {
 596                         $this->throwDBException( $e );
 597                 }
 598         }
 599
 600         public function getCoalesceLocationInternal() {
 601                 return $this->cluster
 602                         ? "DBCluster:{$this->cluster}:{$this->domain}"
 603                         : "LBFactory:{$this->domain}";
 604         }
 605
 606         protected function doGetSiblingQueuesWithJobs( array $types ) {
 607                 $dbr = $this->getReplicaDB();
 608                 // @note: this does not check whether the jobs are claimed or not.
 609                 // This is useful so JobQueueGroup::pop() also sees queues that only
 610                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 611                 // failed jobs so that they can be popped again for that edge case.
 612                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 613                         [ 'job_cmd' => $types ], __METHOD__ );
 614
 615                 $types = [];
 616                 foreach ( $res as $row ) {
 617                         $types[] = $row->job_cmd;
 618                 }
 619
 620                 return $types;
 621         }
 622
 623         protected function doGetSiblingQueueSizes( array $types ) {
 624                 $dbr = $this->getReplicaDB();
 625                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 626                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 627
 628                 $sizes = [];
 629                 foreach ( $res as $row ) {
 630                         $sizes[$row->job_cmd] = (int)$row->count;
 631                 }
 632
 633                 return $sizes;
 634         }
 635
 636         /**
 637          * Recycle or destroy any jobs that have been claimed for too long
 638          *
 639          * @return int Number of jobs recycled/deleted
 640          */
 641         public function recycleAndDeleteStaleJobs() {
 642                 $now = time();
 643                 $count = 0; // affected rows
 644                 $dbw = $this->getMasterDB();
 645
 646                 try {
 647                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 648                                 return $count; // already in progress
 649                         }
 650
 651                         // Remove claims on jobs acquired for too long if enabled...
 652                         if ( $this->claimTTL > 0 ) {
 653                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 654                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 655                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 656                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 657                                 $res = $dbw->select( 'job', 'job_id',
 658                                         [
 659                                                 'job_cmd' => $this->type,
 660                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 661                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 662                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 663                                         __METHOD__
 664                                 );
 665                                 $ids = array_map(
 666                                         function ( $o ) {
 667                                                 return $o->job_id;
 668                                         }, iterator_to_array( $res )
 669                                 );
 670                                 if ( count( $ids ) ) {
 671                                         // Reset job_token for these jobs so that other runners will pick them up.
 672                                         // Set the timestamp to the current time, as it is useful to now that the job
 673                                         // was already tried before (the timestamp becomes the "released" time).
 674                                         $dbw->update( 'job',
 675                                                 [
 676                                                         'job_token' => '',
 677                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 678                                                 [
 679                                                         'job_id' => $ids ],
 680                                                 __METHOD__
 681                                         );
 682                                         $affected = $dbw->affectedRows();
 683                                         $count += $affected;
 684                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 685                                         $this->aggr->notifyQueueNonEmpty( $this->domain, $this->type );
 686                                 }
 687                         }
 688
 689                         // Just destroy any stale jobs...
 690                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 691                         $conds = [
 692                                 'job_cmd' => $this->type,
 693                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 694                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 695                         ];
 696                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 697                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 698                         }
 699                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 700                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 701                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 702                         $ids = array_map(
 703                                 function ( $o ) {
 704                                         return $o->job_id;
 705                                 }, iterator_to_array( $res )
 706                         );
 707                         if ( count( $ids ) ) {
 708                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 709                                 $affected = $dbw->affectedRows();
 710                                 $count += $affected;
 711                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 712                         }
 713
 714                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 715                 } catch ( DBError $e ) {
 716                         $this->throwDBException( $e );
 717                 }
 718
 719                 return $count;
 720         }
 721
 722         /**
 723          * @param IJobSpecification $job
 724          * @param IDatabase $db
 725          * @return array
 726          */
 727         protected function insertFields( IJobSpecification $job, IDatabase $db ) {
 728                 return [
 729                         // Fields that describe the nature of the job
 730                         'job_cmd' => $job->getType(),
 731                         'job_namespace' => $job->getTitle()->getNamespace(),
 732                         'job_title' => $job->getTitle()->getDBkey(),
 733                         'job_params' => self::makeBlob( $job->getParams() ),
 734                         // Additional job metadata
 735                         'job_timestamp' => $db->timestamp(),
 736                         'job_sha1' => Wikimedia\base_convert(
 737                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 738                                 16, 36, 31
 739                         ),
 740                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 741                 ];
 742         }
 743
 744         /**
 745          * @throws JobQueueConnectionError
 746          * @return DBConnRef
 747          */
 748         protected function getReplicaDB() {
 749                 try {
 750                         return $this->getDB( DB_REPLICA );
 751                 } catch ( DBConnectionError $e ) {
 752                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 753                 }
 754         }
 755
 756         /**
 757          * @throws JobQueueConnectionError
 758          * @return DBConnRef
 759          */
 760         protected function getMasterDB() {
 761                 try {
 762                         return $this->getDB( DB_MASTER );
 763                 } catch ( DBConnectionError $e ) {
 764                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 765                 }
 766         }
 767
 768         /**
 769          * @param int $index (DB_REPLICA/DB_MASTER)
 770          * @return DBConnRef
 771          */
 772         protected function getDB( $index ) {
 773                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 774                 $lb = ( $this->cluster !== false )
 775                         ? $lbFactory->getExternalLB( $this->cluster )
 776                         : $lbFactory->getMainLB( $this->domain );
 777
 778                 return ( $lb->getServerType( $lb->getWriterIndex() ) !== 'sqlite' )
 779                         // Keep a separate connection to avoid contention and deadlocks;
 780                         // However, SQLite has the opposite behavior due to DB-level locking.
 781                         ? $lb->getConnectionRef( $index, [], $this->domain, $lb::CONN_TRX_AUTOCOMMIT )
 782                         // Jobs insertion will be defered until the PRESEND stage to reduce contention.
 783                         : $lb->getConnectionRef( $index, [], $this->domain );
 784         }
 785
 786         /**
 787          * @param string $property
 788          * @return string
 789          */
 790         private function getCacheKey( $property ) {
 791                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 792
 793                 return $this->cache->makeGlobalKey(
 794                         'jobqueue',
 795                         $this->domain,
 796                         $cluster,
 797                         $this->type,
 798                         $property
 799                 );
 800         }
 801
 802         /**
 803          * @param array|bool $params
 804          * @return string
 805          */
 806         protected static function makeBlob( $params ) {
 807                 if ( $params !== false ) {
 808                         return serialize( $params );
 809                 } else {
 810                         return '';
 811                 }
 812         }
 813
 814         /**
 815          * @param string $blob
 816          * @return bool|mixed
 817          */
 818         protected static function extractBlob( $blob ) {
 819                 if ( (string)$blob !== '' ) {
 820                         return unserialize( $blob );
 821                 } else {
 822                         return false;
 823                 }
 824         }
 825
 826         /**
 827          * @param DBError $e
 828          * @throws JobQueueError
 829          */
 830         protected function throwDBException( DBError $e ) {
 831                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 832         }
 833
 834         /**
 835          * Return the list of job fields that should be selected.
 836          * @since 1.23
 837          * @return array
 838          */
 839         public static function selectFields() {
 840                 return [
 841                         'job_id',
 842                         'job_cmd',
 843                         'job_namespace',
 844                         'job_title',
 845                         'job_timestamp',
 846                         'job_params',
 847                         'job_random',
 848                         'job_attempts',
 849                         'job_token',
 850                         'job_token_timestamp',
 851                         'job_sha1',
 852                 ];
 853         }
 854 }