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->metadata['id'] = $row->job_id;
 321                                 $job->metadata['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                 if ( !isset( $job->metadata['id'] ) ) {
 488                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 489                 }
 490
 491                 $dbw = $this->getMasterDB();
 492                 /** @noinspection PhpUnusedLocalVariableInspection */
 493                 $scope = $this->getScopedNoTrxFlag( $dbw );
 494                 try {
 495                         // Delete a row with a single DELETE without holding row locks over RTTs...
 496                         $dbw->delete( 'job',
 497                                 [ 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ], __METHOD__ );
 498
 499                         JobQueue::incrStats( 'acks', $this->type );
 500                 } catch ( DBError $e ) {
 501                         $this->throwDBException( $e );
 502                 }
 503         }
 504
 505         /**
 506          * @see JobQueue::doDeduplicateRootJob()
 507          * @param IJobSpecification $job
 508          * @throws MWException
 509          * @return bool
 510          */
 511         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 512                 $params = $job->getParams();
 513                 if ( !isset( $params['rootJobSignature'] ) ) {
 514                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 515                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 516                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 517                 }
 518                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 519                 // Callers should call JobQueueGroup::push() before this method so that if the insert
 520                 // fails, the de-duplication registration will be aborted. Since the insert is
 521                 // deferred till "transaction idle", do the same here, so that the ordering is
 522                 // maintained. Having only the de-duplication registration succeed would cause
 523                 // jobs to become no-ops without any actual jobs that made them redundant.
 524                 $dbw = $this->getMasterDB();
 525                 /** @noinspection PhpUnusedLocalVariableInspection */
 526                 $scope = $this->getScopedNoTrxFlag( $dbw );
 527
 528                 $cache = $this->dupCache;
 529                 $dbw->onTransactionCommitOrIdle(
 530                         function () use ( $cache, $params, $key ) {
 531                                 $timestamp = $cache->get( $key ); // current last timestamp of this job
 532                                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 533                                         return true; // a newer version of this root job was enqueued
 534                                 }
 535
 536                                 // Update the timestamp of the last root job started at the location...
 537                                 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 538                         },
 539                         __METHOD__
 540                 );
 541
 542                 return true;
 543         }
 544
 545         /**
 546          * @see JobQueue::doDelete()
 547          * @return bool
 548          */
 549         protected function doDelete() {
 550                 $dbw = $this->getMasterDB();
 551                 /** @noinspection PhpUnusedLocalVariableInspection */
 552                 $scope = $this->getScopedNoTrxFlag( $dbw );
 553                 try {
 554                         $dbw->delete( 'job', [ 'job_cmd' => $this->type ] );
 555                 } catch ( DBError $e ) {
 556                         $this->throwDBException( $e );
 557                 }
 558
 559                 return true;
 560         }
 561
 562         /**
 563          * @see JobQueue::doWaitForBackups()
 564          * @return void
 565          */
 566         protected function doWaitForBackups() {
 567                 if ( $this->server ) {
 568                         return; // not using LBFactory instance
 569                 }
 570
 571                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 572                 $lbFactory->waitForReplication( [
 573                         'domain' => $this->domain,
 574                         'cluster' => is_string( $this->cluster ) ? $this->cluster : false
 575                 ] );
 576         }
 577
 578         /**
 579          * @return void
 580          */
 581         protected function doFlushCaches() {
 582                 foreach ( [ 'size', 'acquiredcount' ] as $type ) {
 583                         $this->cache->delete( $this->getCacheKey( $type ) );
 584                 }
 585         }
 586
 587         /**
 588          * @see JobQueue::getAllQueuedJobs()
 589          * @return Iterator
 590          */
 591         public function getAllQueuedJobs() {
 592                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), 'job_token' => '' ] );
 593         }
 594
 595         /**
 596          * @see JobQueue::getAllAcquiredJobs()
 597          * @return Iterator
 598          */
 599         public function getAllAcquiredJobs() {
 600                 return $this->getJobIterator( [ 'job_cmd' => $this->getType(), "job_token > ''" ] );
 601         }
 602
 603         /**
 604          * @param array $conds Query conditions
 605          * @return Iterator
 606          */
 607         protected function getJobIterator( array $conds ) {
 608                 $dbr = $this->getReplicaDB();
 609                 /** @noinspection PhpUnusedLocalVariableInspection */
 610                 $scope = $this->getScopedNoTrxFlag( $dbr );
 611                 try {
 612                         return new MappedIterator(
 613                                 $dbr->select( 'job', self::selectFields(), $conds ),
 614                                 function ( $row ) {
 615                                         $job = Job::factory(
 616                                                 $row->job_cmd,
 617                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 618                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : []
 619                                         );
 620                                         $job->metadata['id'] = $row->job_id;
 621                                         $job->metadata['timestamp'] = $row->job_timestamp;
 622
 623                                         return $job;
 624                                 }
 625                         );
 626                 } catch ( DBError $e ) {
 627                         $this->throwDBException( $e );
 628                 }
 629         }
 630
 631         public function getCoalesceLocationInternal() {
 632                 if ( $this->server ) {
 633                         return null; // not using the LBFactory instance
 634                 }
 635
 636                 return is_string( $this->cluster )
 637                         ? "DBCluster:{$this->cluster}:{$this->domain}"
 638                         : "LBFactory:{$this->domain}";
 639         }
 640
 641         protected function doGetSiblingQueuesWithJobs( array $types ) {
 642                 $dbr = $this->getReplicaDB();
 643                 /** @noinspection PhpUnusedLocalVariableInspection */
 644                 $scope = $this->getScopedNoTrxFlag( $dbr );
 645                 // @note: this does not check whether the jobs are claimed or not.
 646                 // This is useful so JobQueueGroup::pop() also sees queues that only
 647                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 648                 // failed jobs so that they can be popped again for that edge case.
 649                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 650                         [ 'job_cmd' => $types ], __METHOD__ );
 651
 652                 $types = [];
 653                 foreach ( $res as $row ) {
 654                         $types[] = $row->job_cmd;
 655                 }
 656
 657                 return $types;
 658         }
 659
 660         protected function doGetSiblingQueueSizes( array $types ) {
 661                 $dbr = $this->getReplicaDB();
 662                 /** @noinspection PhpUnusedLocalVariableInspection */
 663                 $scope = $this->getScopedNoTrxFlag( $dbr );
 664
 665                 $res = $dbr->select( 'job', [ 'job_cmd', 'COUNT(*) AS count' ],
 666                         [ 'job_cmd' => $types ], __METHOD__, [ 'GROUP BY' => 'job_cmd' ] );
 667
 668                 $sizes = [];
 669                 foreach ( $res as $row ) {
 670                         $sizes[$row->job_cmd] = (int)$row->count;
 671                 }
 672
 673                 return $sizes;
 674         }
 675
 676         /**
 677          * Recycle or destroy any jobs that have been claimed for too long
 678          *
 679          * @return int Number of jobs recycled/deleted
 680          */
 681         public function recycleAndDeleteStaleJobs() {
 682                 $now = time();
 683                 $count = 0; // affected rows
 684                 $dbw = $this->getMasterDB();
 685                 /** @noinspection PhpUnusedLocalVariableInspection */
 686                 $scope = $this->getScopedNoTrxFlag( $dbw );
 687
 688                 try {
 689                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 690                                 return $count; // already in progress
 691                         }
 692
 693                         // Remove claims on jobs acquired for too long if enabled...
 694                         if ( $this->claimTTL > 0 ) {
 695                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 696                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 697                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 698                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 699                                 $res = $dbw->select( 'job', 'job_id',
 700                                         [
 701                                                 'job_cmd' => $this->type,
 702                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 703                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 704                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ], // retries left
 705                                         __METHOD__
 706                                 );
 707                                 $ids = array_map(
 708                                         function ( $o ) {
 709                                                 return $o->job_id;
 710                                         }, iterator_to_array( $res )
 711                                 );
 712                                 if ( count( $ids ) ) {
 713                                         // Reset job_token for these jobs so that other runners will pick them up.
 714                                         // Set the timestamp to the current time, as it is useful to now that the job
 715                                         // was already tried before (the timestamp becomes the "released" time).
 716                                         $dbw->update( 'job',
 717                                                 [
 718                                                         'job_token' => '',
 719                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ], // time of release
 720                                                 [
 721                                                         'job_id' => $ids ],
 722                                                 __METHOD__
 723                                         );
 724                                         $affected = $dbw->affectedRows();
 725                                         $count += $affected;
 726                                         JobQueue::incrStats( 'recycles', $this->type, $affected );
 727                                         $this->aggr->notifyQueueNonEmpty( $this->domain, $this->type );
 728                                 }
 729                         }
 730
 731                         // Just destroy any stale jobs...
 732                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 733                         $conds = [
 734                                 'job_cmd' => $this->type,
 735                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 736                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 737                         ];
 738                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 739                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 740                         }
 741                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 742                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 743                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 744                         $ids = array_map(
 745                                 function ( $o ) {
 746                                         return $o->job_id;
 747                                 }, iterator_to_array( $res )
 748                         );
 749                         if ( count( $ids ) ) {
 750                                 $dbw->delete( 'job', [ 'job_id' => $ids ], __METHOD__ );
 751                                 $affected = $dbw->affectedRows();
 752                                 $count += $affected;
 753                                 JobQueue::incrStats( 'abandons', $this->type, $affected );
 754                         }
 755
 756                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 757                 } catch ( DBError $e ) {
 758                         $this->throwDBException( $e );
 759                 }
 760
 761                 return $count;
 762         }
 763
 764         /**
 765          * @param IJobSpecification $job
 766          * @param IDatabase $db
 767          * @return array
 768          */
 769         protected function insertFields( IJobSpecification $job, IDatabase $db ) {
 770                 return [
 771                         // Fields that describe the nature of the job
 772                         'job_cmd' => $job->getType(),
 773                         'job_namespace' => $job->getTitle()->getNamespace(),
 774                         'job_title' => $job->getTitle()->getDBkey(),
 775                         'job_params' => self::makeBlob( $job->getParams() ),
 776                         // Additional job metadata
 777                         'job_timestamp' => $db->timestamp(),
 778                         'job_sha1' => Wikimedia\base_convert(
 779                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 780                                 16, 36, 31
 781                         ),
 782                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 783                 ];
 784         }
 785
 786         /**
 787          * @throws JobQueueConnectionError
 788          * @return IDatabase
 789          */
 790         protected function getReplicaDB() {
 791                 try {
 792                         return $this->getDB( DB_REPLICA );
 793                 } catch ( DBConnectionError $e ) {
 794                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 795                 }
 796         }
 797
 798         /**
 799          * @throws JobQueueConnectionError
 800          * @return IDatabase
 801          */
 802         protected function getMasterDB() {
 803                 try {
 804                         return $this->getDB( DB_MASTER );
 805                 } catch ( DBConnectionError $e ) {
 806                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 807                 }
 808         }
 809
 810         /**
 811          * @param int $index (DB_REPLICA/DB_MASTER)
 812          * @return IDatabase
 813          */
 814         protected function getDB( $index ) {
 815                 if ( $this->server ) {
 816                         if ( $this->conn instanceof IDatabase ) {
 817                                 return $this->conn;
 818                         } elseif ( $this->conn instanceof DBError ) {
 819                                 throw $this->conn;
 820                         }
 821
 822                         try {
 823                                 $this->conn = Database::factory( $this->server['type'], $this->server );
 824                         } catch ( DBError $e ) {
 825                                 $this->conn = $e;
 826                                 throw $e;
 827                         }
 828
 829                         return $this->conn;
 830                 } else {
 831                         $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 832                         $lb = is_string( $this->cluster )
 833                                 ? $lbFactory->getExternalLB( $this->cluster )
 834                                 : $lbFactory->getMainLB( $this->domain );
 835
 836                         return ( $lb->getServerType( $lb->getWriterIndex() ) !== 'sqlite' )
 837                                 // Keep a separate connection to avoid contention and deadlocks;
 838                                 // However, SQLite has the opposite behavior due to DB-level locking.
 839                                 ? $lb->getConnectionRef( $index, [], $this->domain, $lb::CONN_TRX_AUTOCOMMIT )
 840                                 // Jobs insertion will be defered until the PRESEND stage to reduce contention.
 841                                 : $lb->getConnectionRef( $index, [], $this->domain );
 842                 }
 843         }
 844
 845         /**
 846          * @param IDatabase $db
 847          * @return ScopedCallback
 848          */
 849         private function getScopedNoTrxFlag( IDatabase $db ) {
 850                 $autoTrx = $db->getFlag( DBO_TRX ); // get current setting
 851                 $db->clearFlag( DBO_TRX ); // make each query its own transaction
 852
 853                 return new ScopedCallback( function () use ( $db, $autoTrx ) {
 854                         if ( $autoTrx ) {
 855                                 $db->setFlag( DBO_TRX ); // restore old setting
 856                         }
 857                 } );
 858         }
 859
 860         /**
 861          * @param string $property
 862          * @return string
 863          */
 864         private function getCacheKey( $property ) {
 865                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 866
 867                 return $this->cache->makeGlobalKey(
 868                         'jobqueue',
 869                         $this->domain,
 870                         $cluster,
 871                         $this->type,
 872                         $property
 873                 );
 874         }
 875
 876         /**
 877          * @param array|bool $params
 878          * @return string
 879          */
 880         protected static function makeBlob( $params ) {
 881                 if ( $params !== false ) {
 882                         return serialize( $params );
 883                 } else {
 884                         return '';
 885                 }
 886         }
 887
 888         /**
 889          * @param string $blob
 890          * @return bool|mixed
 891          */
 892         protected static function extractBlob( $blob ) {
 893                 if ( (string)$blob !== '' ) {
 894                         return unserialize( $blob );
 895                 } else {
 896                         return false;
 897                 }
 898         }
 899
 900         /**
 901          * @param DBError $e
 902          * @throws JobQueueError
 903          */
 904         protected function throwDBException( DBError $e ) {
 905                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 906         }
 907
 908         /**
 909          * Return the list of job fields that should be selected.
 910          * @since 1.23
 911          * @return array
 912          */
 913         public static function selectFields() {
 914                 return [
 915                         'job_id',
 916                         'job_cmd',
 917                         'job_namespace',
 918                         'job_title',
 919                         'job_timestamp',
 920                         'job_params',
 921                         'job_random',
 922                         'job_attempts',
 923                         'job_token',
 924                         'job_token_timestamp',
 925                         'job_sha1',
 926                 ];
 927         }
 928 }