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  * @author Aaron Schulz
  22  */
  23
  24 /**
  25  * Class to handle job queues stored in the DB
  26  *
  27  * @ingroup JobQueue
  28  * @since 1.21
  29  */
  30 class JobQueueDB extends JobQueue {
  31         const CACHE_TTL_SHORT = 30; // integer; seconds to cache info without re-validating
  32         const MAX_AGE_PRUNE = 604800; // integer; seconds a job can live once claimed
  33         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  34         const MAX_OFFSET = 255; // integer; maximum number of rows to skip
  35
  36         /** @var BagOStuff */
  37         protected $cache;
  38
  39         /** @var bool|string Name of an external DB cluster. False if not set */
  40         protected $cluster = false;
  41
  42         /**
  43          * Additional parameters include:
  44          *   - cluster : The name of an external cluster registered via LBFactory.
  45          *               If not specified, the primary DB cluster for the wiki will be used.
  46          *               This can be overridden with a custom cluster so that DB handles will
  47          *               be retrieved via LBFactory::getExternalLB() and getConnection().
  48          * @param array $params
  49          */
  50         protected function __construct( array $params ) {
  51                 global $wgMemc;
  52
  53                 parent::__construct( $params );
  54
  55                 $this->cluster = isset( $params['cluster'] ) ? $params['cluster'] : false;
  56                 // Make sure that we don't use the SQL cache, which would be harmful
  57                 $this->cache = ( $wgMemc instanceof SqlBagOStuff ) ? new EmptyBagOStuff() : $wgMemc;
  58         }
  59
  60         protected function supportedOrders() {
  61                 return array( 'random', 'timestamp', 'fifo' );
  62         }
  63
  64         protected function optimalOrder() {
  65                 return 'random';
  66         }
  67
  68         /**
  69          * @see JobQueue::doIsEmpty()
  70          * @return bool
  71          */
  72         protected function doIsEmpty() {
  73                 $dbr = $this->getSlaveDB();
  74                 try {
  75                         $found = $dbr->selectField( // unclaimed job
  76                                 'job', '1', array( 'job_cmd' => $this->type, 'job_token' => '' ), __METHOD__
  77                         );
  78                 } catch ( DBError $e ) {
  79                         $this->throwDBException( $e );
  80                 }
  81
  82                 return !$found;
  83         }
  84
  85         /**
  86          * @see JobQueue::doGetSize()
  87          * @return int
  88          */
  89         protected function doGetSize() {
  90                 $key = $this->getCacheKey( 'size' );
  91
  92                 $size = $this->cache->get( $key );
  93                 if ( is_int( $size ) ) {
  94                         return $size;
  95                 }
  96
  97                 try {
  98                         $dbr = $this->getSlaveDB();
  99                         $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
 100                                 array( 'job_cmd' => $this->type, 'job_token' => '' ),
 101                                 __METHOD__
 102                         );
 103                 } catch ( DBError $e ) {
 104                         $this->throwDBException( $e );
 105                 }
 106                 $this->cache->set( $key, $size, self::CACHE_TTL_SHORT );
 107
 108                 return $size;
 109         }
 110
 111         /**
 112          * @see JobQueue::doGetAcquiredCount()
 113          * @return int
 114          */
 115         protected function doGetAcquiredCount() {
 116                 if ( $this->claimTTL <= 0 ) {
 117                         return 0; // no acknowledgements
 118                 }
 119
 120                 $key = $this->getCacheKey( 'acquiredcount' );
 121
 122                 $count = $this->cache->get( $key );
 123                 if ( is_int( $count ) ) {
 124                         return $count;
 125                 }
 126
 127                 $dbr = $this->getSlaveDB();
 128                 try {
 129                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 130                                 array( 'job_cmd' => $this->type, "job_token != {$dbr->addQuotes( '' )}" ),
 131                                 __METHOD__
 132                         );
 133                 } catch ( DBError $e ) {
 134                         $this->throwDBException( $e );
 135                 }
 136                 $this->cache->set( $key, $count, self::CACHE_TTL_SHORT );
 137
 138                 return $count;
 139         }
 140
 141         /**
 142          * @see JobQueue::doGetAbandonedCount()
 143          * @return int
 144          * @throws MWException
 145          */
 146         protected function doGetAbandonedCount() {
 147                 global $wgMemc;
 148
 149                 if ( $this->claimTTL <= 0 ) {
 150                         return 0; // no acknowledgements
 151                 }
 152
 153                 $key = $this->getCacheKey( 'abandonedcount' );
 154
 155                 $count = $wgMemc->get( $key );
 156                 if ( is_int( $count ) ) {
 157                         return $count;
 158                 }
 159
 160                 $dbr = $this->getSlaveDB();
 161                 try {
 162                         $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
 163                                 array(
 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                 $wgMemc->set( $key, $count, self::CACHE_TTL_SHORT );
 174
 175                 return $count;
 176         }
 177
 178         /**
 179          * @see JobQueue::doBatchPush()
 180          * @param array $jobs
 181          * @param int $flags
 182          * @throws DBError|Exception
 183          * @return void
 184          */
 185         protected function doBatchPush( array $jobs, $flags ) {
 186                 $dbw = $this->getMasterDB();
 187
 188                 $that = $this;
 189                 $method = __METHOD__;
 190                 $dbw->onTransactionIdle(
 191                         function () use ( $dbw, $that, $jobs, $flags, $method ) {
 192                                 $that->doBatchPushInternal( $dbw, $jobs, $flags, $method );
 193                         }
 194                 );
 195         }
 196
 197         /**
 198          * This function should *not* be called outside of JobQueueDB
 199          *
 200          * @param IDatabase $dbw
 201          * @param array $jobs
 202          * @param int $flags
 203          * @param string $method
 204          * @throws DBError
 205          * @return void
 206          */
 207         public function doBatchPushInternal( IDatabase $dbw, array $jobs, $flags, $method ) {
 208                 if ( !count( $jobs ) ) {
 209                         return;
 210                 }
 211
 212                 $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
 213                 $rowList = array(); // list of jobs for jobs that are not de-duplicated
 214                 foreach ( $jobs as $job ) {
 215                         $row = $this->insertFields( $job );
 216                         if ( $job->ignoreDuplicates() ) {
 217                                 $rowSet[$row['job_sha1']] = $row;
 218                         } else {
 219                                 $rowList[] = $row;
 220                         }
 221                 }
 222
 223                 if ( $flags & self::QOS_ATOMIC ) {
 224                         $dbw->begin( $method ); // wrap all the job additions in one transaction
 225                 }
 226                 try {
 227                         // Strip out any duplicate jobs that are already in the queue...
 228                         if ( count( $rowSet ) ) {
 229                                 $res = $dbw->select( 'job', 'job_sha1',
 230                                         array(
 231                                                 // No job_type condition since it's part of the job_sha1 hash
 232                                                 'job_sha1' => array_keys( $rowSet ),
 233                                                 'job_token' => '' // unclaimed
 234                                         ),
 235                                         $method
 236                                 );
 237                                 foreach ( $res as $row ) {
 238                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate.\n" );
 239                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 240                                 }
 241                         }
 242                         // Build the full list of job rows to insert
 243                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 244                         // Insert the job rows in chunks to avoid slave lag...
 245                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 246                                 $dbw->insert( 'job', $rowBatch, $method );
 247                         }
 248                         JobQueue::incrStats( 'job-insert', $this->type, count( $rows ) );
 249                         JobQueue::incrStats(
 250                                 'job-insert-duplicate',
 251                                 $this->type,
 252                                 count( $rowSet ) + count( $rowList ) - count( $rows )
 253                         );
 254                 } catch ( DBError $e ) {
 255                         if ( $flags & self::QOS_ATOMIC ) {
 256                                 $dbw->rollback( $method );
 257                         }
 258                         throw $e;
 259                 }
 260                 if ( $flags & self::QOS_ATOMIC ) {
 261                         $dbw->commit( $method );
 262                 }
 263
 264                 return;
 265         }
 266
 267         /**
 268          * @see JobQueue::doPop()
 269          * @return Job|bool
 270          */
 271         protected function doPop() {
 272                 $dbw = $this->getMasterDB();
 273                 try {
 274                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 275                         $autoTrx = $dbw->getFlag( DBO_TRX ); // get current setting
 276                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 277                         $scopedReset = new ScopedCallback( function () use ( $dbw, $autoTrx ) {
 278                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore old setting
 279                         } );
 280
 281                         $uuid = wfRandomString( 32 ); // pop attempt
 282                         $job = false; // job popped off
 283                         do { // retry when our row is invalid or deleted as a duplicate
 284                                 // Try to reserve a row in the DB...
 285                                 if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 286                                         $row = $this->claimOldest( $uuid );
 287                                 } else { // random first
 288                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 289                                         $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 290                                         $row = $this->claimRandom( $uuid, $rand, $gte );
 291                                 }
 292                                 // Check if we found a row to reserve...
 293                                 if ( !$row ) {
 294                                         break; // nothing to do
 295                                 }
 296                                 JobQueue::incrStats( 'job-pop', $this->type );
 297                                 // Get the job object from the row...
 298                                 $title = Title::makeTitle( $row->job_namespace, $row->job_title );
 299                                 $job = Job::factory( $row->job_cmd, $title,
 300                                         self::extractBlob( $row->job_params ), $row->job_id );
 301                                 $job->metadata['id'] = $row->job_id;
 302                                 break; // done
 303                         } while ( true );
 304
 305                         if ( !$job || mt_rand( 0, 9 ) == 0 ) {
 306                                 // Handled jobs that need to be recycled/deleted;
 307                                 // any recycled jobs will be picked up next attempt
 308                                 $this->recycleAndDeleteStaleJobs();
 309                         }
 310                 } catch ( DBError $e ) {
 311                         $this->throwDBException( $e );
 312                 }
 313
 314                 return $job;
 315         }
 316
 317         /**
 318          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 319          *
 320          * @param string $uuid 32 char hex string
 321          * @param int $rand Random unsigned integer (31 bits)
 322          * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
 323          * @return stdClass|bool Row|false
 324          */
 325         protected function claimRandom( $uuid, $rand, $gte ) {
 326                 $dbw = $this->getMasterDB();
 327                 // Check cache to see if the queue has <= OFFSET items
 328                 $tinyQueue = $this->cache->get( $this->getCacheKey( 'small' ) );
 329
 330                 $row = false; // the row acquired
 331                 $invertedDirection = false; // whether one job_random direction was already scanned
 332                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 333                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 334                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 335                 // be used here with MySQL.
 336                 do {
 337                         if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
 338                                 // For small queues, using OFFSET will overshoot and return no rows more often.
 339                                 // Instead, this uses job_random to pick a row (possibly checking both directions).
 340                                 $ineq = $gte ? '>=' : '<=';
 341                                 $dir = $gte ? 'ASC' : 'DESC';
 342                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 343                                         array(
 344                                                 'job_cmd' => $this->type,
 345                                                 'job_token' => '', // unclaimed
 346                                                 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 347                                         __METHOD__,
 348                                         array( 'ORDER BY' => "job_random {$dir}" )
 349                                 );
 350                                 if ( !$row && !$invertedDirection ) {
 351                                         $gte = !$gte;
 352                                         $invertedDirection = true;
 353                                         continue; // try the other direction
 354                                 }
 355                         } else { // table *may* have >= MAX_OFFSET rows
 356                                 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
 357                                 // in MySQL if there are many rows for some reason. This uses a small OFFSET
 358                                 // instead of job_random for reducing excess claim retries.
 359                                 $row = $dbw->selectRow( 'job', self::selectFields(), // find a random job
 360                                         array(
 361                                                 'job_cmd' => $this->type,
 362                                                 'job_token' => '', // unclaimed
 363                                         ),
 364                                         __METHOD__,
 365                                         array( 'OFFSET' => mt_rand( 0, self::MAX_OFFSET ) )
 366                                 );
 367                                 if ( !$row ) {
 368                                         $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
 369                                         $this->cache->set( $this->getCacheKey( 'small' ), 1, 30 );
 370                                         continue; // use job_random
 371                                 }
 372                         }
 373
 374                         if ( $row ) { // claim the job
 375                                 $dbw->update( 'job', // update by PK
 376                                         array(
 377                                                 'job_token' => $uuid,
 378                                                 'job_token_timestamp' => $dbw->timestamp(),
 379                                                 'job_attempts = job_attempts+1' ),
 380                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 381                                         __METHOD__
 382                                 );
 383                                 // This might get raced out by another runner when claiming the previously
 384                                 // selected row. The use of job_random should minimize this problem, however.
 385                                 if ( !$dbw->affectedRows() ) {
 386                                         $row = false; // raced out
 387                                 }
 388                         } else {
 389                                 break; // nothing to do
 390                         }
 391                 } while ( !$row );
 392
 393                 return $row;
 394         }
 395
 396         /**
 397          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 398          *
 399          * @param string $uuid 32 char hex string
 400          * @return stdClass|bool Row|false
 401          */
 402         protected function claimOldest( $uuid ) {
 403                 $dbw = $this->getMasterDB();
 404
 405                 $row = false; // the row acquired
 406                 do {
 407                         if ( $dbw->getType() === 'mysql' ) {
 408                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 409                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 410                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 411                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 412                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 413                                         "SET " .
 414                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 415                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 416                                                 "job_attempts = job_attempts+1 " .
 417                                         "WHERE ( " .
 418                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 419                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 420                                         ") ORDER BY job_id ASC LIMIT 1",
 421                                         __METHOD__
 422                                 );
 423                         } else {
 424                                 // Use a subquery to find the job, within an UPDATE to claim it.
 425                                 // This uses as much of the DB wrapper functions as possible.
 426                                 $dbw->update( 'job',
 427                                         array(
 428                                                 'job_token' => $uuid,
 429                                                 'job_token_timestamp' => $dbw->timestamp(),
 430                                                 'job_attempts = job_attempts+1' ),
 431                                         array( 'job_id = (' .
 432                                                 $dbw->selectSQLText( 'job', 'job_id',
 433                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 434                                                         __METHOD__,
 435                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 436                                                 ')'
 437                                         ),
 438                                         __METHOD__
 439                                 );
 440                         }
 441                         // Fetch any row that we just reserved...
 442                         if ( $dbw->affectedRows() ) {
 443                                 $row = $dbw->selectRow( 'job', self::selectFields(),
 444                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 445                                 );
 446                                 if ( !$row ) { // raced out by duplicate job removal
 447                                         wfDebug( "Row deleted as duplicate by another process.\n" );
 448                                 }
 449                         } else {
 450                                 break; // nothing to do
 451                         }
 452                 } while ( !$row );
 453
 454                 return $row;
 455         }
 456
 457         /**
 458          * @see JobQueue::doAck()
 459          * @param Job $job
 460          * @throws MWException
 461          * @return Job|bool
 462          */
 463         protected function doAck( Job $job ) {
 464                 if ( !isset( $job->metadata['id'] ) ) {
 465                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 466                 }
 467
 468                 $dbw = $this->getMasterDB();
 469                 try {
 470                         $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 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                                 array( 'job_cmd' => $this->type, 'job_id' => $job->metadata['id'] ), __METHOD__ );
 480
 481                         JobQueue::incrStats( 'job-ack', $this->type );
 482                 } catch ( DBError $e ) {
 483                         $this->throwDBException( $e );
 484                 }
 485
 486                 return true;
 487         }
 488
 489         /**
 490          * @see JobQueue::doDeduplicateRootJob()
 491          * @param Job $job
 492          * @throws MWException
 493          * @return bool
 494          */
 495         protected function doDeduplicateRootJob( Job $job ) {
 496                 $params = $job->getParams();
 497                 if ( !isset( $params['rootJobSignature'] ) ) {
 498                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 499                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 500                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 501                 }
 502                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 503                 // Callers should call batchInsert() and then this function so that if the insert
 504                 // fails, the de-duplication registration will be aborted. Since the insert is
 505                 // deferred till "transaction idle", do the same here, so that the ordering is
 506                 // maintained. Having only the de-duplication registration succeed would cause
 507                 // jobs to become no-ops without any actual jobs that made them redundant.
 508                 $dbw = $this->getMasterDB();
 509                 $cache = $this->dupCache;
 510                 $dbw->onTransactionIdle( function () use ( $cache, $params, $key, $dbw ) {
 511                         $timestamp = $cache->get( $key ); // current last timestamp of this job
 512                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 513                                 return true; // a newer version of this root job was enqueued
 514                         }
 515
 516                         // Update the timestamp of the last root job started at the location...
 517                         return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 518                 } );
 519
 520                 return true;
 521         }
 522
 523         /**
 524          * @see JobQueue::doDelete()
 525          * @return bool
 526          */
 527         protected function doDelete() {
 528                 $dbw = $this->getMasterDB();
 529                 try {
 530                         $dbw->delete( 'job', array( 'job_cmd' => $this->type ) );
 531                 } catch ( DBError $e ) {
 532                         $this->throwDBException( $e );
 533                 }
 534
 535                 return true;
 536         }
 537
 538         /**
 539          * @see JobQueue::doWaitForBackups()
 540          * @return void
 541          */
 542         protected function doWaitForBackups() {
 543                 wfWaitForSlaves( false, $this->wiki, $this->cluster ?: false );
 544         }
 545
 546         /**
 547          * @return void
 548          */
 549         protected function doFlushCaches() {
 550                 foreach ( array( 'size', 'acquiredcount' ) as $type ) {
 551                         $this->cache->delete( $this->getCacheKey( $type ) );
 552                 }
 553         }
 554
 555         /**
 556          * @see JobQueue::getAllQueuedJobs()
 557          * @return Iterator
 558          */
 559         public function getAllQueuedJobs() {
 560                 $dbr = $this->getSlaveDB();
 561                 try {
 562                         return new MappedIterator(
 563                                 $dbr->select( 'job', self::selectFields(),
 564                                         array( 'job_cmd' => $this->getType(), 'job_token' => '' ) ),
 565                                 function ( $row ) use ( $dbr ) {
 566                                         $job = Job::factory(
 567                                                 $row->job_cmd,
 568                                                 Title::makeTitle( $row->job_namespace, $row->job_title ),
 569                                                 strlen( $row->job_params ) ? unserialize( $row->job_params ) : false
 570                                         );
 571                                         $job->metadata['id'] = $row->job_id;
 572                                         return $job;
 573                                 }
 574                         );
 575                 } catch ( DBError $e ) {
 576                         $this->throwDBException( $e );
 577                 }
 578         }
 579
 580         public function getCoalesceLocationInternal() {
 581                 return $this->cluster
 582                         ? "DBCluster:{$this->cluster}:{$this->wiki}"
 583                         : "LBFactory:{$this->wiki}";
 584         }
 585
 586         protected function doGetSiblingQueuesWithJobs( array $types ) {
 587                 $dbr = $this->getSlaveDB();
 588                 // @note: this does not check whether the jobs are claimed or not.
 589                 // This is useful so JobQueueGroup::pop() also sees queues that only
 590                 // have stale jobs. This lets recycleAndDeleteStaleJobs() re-enqueue
 591                 // failed jobs so that they can be popped again for that edge case.
 592                 $res = $dbr->select( 'job', 'DISTINCT job_cmd',
 593                         array( 'job_cmd' => $types ), __METHOD__ );
 594
 595                 $types = array();
 596                 foreach ( $res as $row ) {
 597                         $types[] = $row->job_cmd;
 598                 }
 599
 600                 return $types;
 601         }
 602
 603         protected function doGetSiblingQueueSizes( array $types ) {
 604                 $dbr = $this->getSlaveDB();
 605                 $res = $dbr->select( 'job', array( 'job_cmd', 'COUNT(*) AS count' ),
 606                         array( 'job_cmd' => $types ), __METHOD__, array( 'GROUP BY' => 'job_cmd' ) );
 607
 608                 $sizes = array();
 609                 foreach ( $res as $row ) {
 610                         $sizes[$row->job_cmd] = (int)$row->count;
 611                 }
 612
 613                 return $sizes;
 614         }
 615
 616         /**
 617          * Recycle or destroy any jobs that have been claimed for too long
 618          *
 619          * @return int Number of jobs recycled/deleted
 620          */
 621         public function recycleAndDeleteStaleJobs() {
 622                 $now = time();
 623                 $count = 0; // affected rows
 624                 $dbw = $this->getMasterDB();
 625
 626                 try {
 627                         if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 628                                 return $count; // already in progress
 629                         }
 630
 631                         // Remove claims on jobs acquired for too long if enabled...
 632                         if ( $this->claimTTL > 0 ) {
 633                                 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 634                                 // Get the IDs of jobs that have be claimed but not finished after too long.
 635                                 // These jobs can be recycled into the queue by expiring the claim. Selecting
 636                                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 637                                 $res = $dbw->select( 'job', 'job_id',
 638                                         array(
 639                                                 'job_cmd' => $this->type,
 640                                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 641                                                 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 642                                                 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ), // retries left
 643                                         __METHOD__
 644                                 );
 645                                 $ids = array_map(
 646                                         function ( $o ) {
 647                                                 return $o->job_id;
 648                                         }, iterator_to_array( $res )
 649                                 );
 650                                 if ( count( $ids ) ) {
 651                                         // Reset job_token for these jobs so that other runners will pick them up.
 652                                         // Set the timestamp to the current time, as it is useful to now that the job
 653                                         // was already tried before (the timestamp becomes the "released" time).
 654                                         $dbw->update( 'job',
 655                                                 array(
 656                                                         'job_token' => '',
 657                                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 658                                                 array(
 659                                                         'job_id' => $ids ),
 660                                                 __METHOD__
 661                                         );
 662                                         $affected = $dbw->affectedRows();
 663                                         $count += $affected;
 664                                         JobQueue::incrStats( 'job-recycle', $this->type, $affected );
 665                                         $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 666                                 }
 667                         }
 668
 669                         // Just destroy any stale jobs...
 670                         $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 671                         $conds = array(
 672                                 'job_cmd' => $this->type,
 673                                 "job_token != {$dbw->addQuotes( '' )}", // was acquired
 674                                 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 675                         );
 676                         if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 677                                 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
 678                         }
 679                         // Get the IDs of jobs that are considered stale and should be removed. Selecting
 680                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 681                         $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 682                         $ids = array_map(
 683                                 function ( $o ) {
 684                                         return $o->job_id;
 685                                 }, iterator_to_array( $res )
 686                         );
 687                         if ( count( $ids ) ) {
 688                                 $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 689                                 $affected = $dbw->affectedRows();
 690                                 $count += $affected;
 691                                 JobQueue::incrStats( 'job-abandon', $this->type, $affected );
 692                         }
 693
 694                         $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 695                 } catch ( DBError $e ) {
 696                         $this->throwDBException( $e );
 697                 }
 698
 699                 return $count;
 700         }
 701
 702         /**
 703          * @param IJobSpecification $job
 704          * @return array
 705          */
 706         protected function insertFields( IJobSpecification $job ) {
 707                 $dbw = $this->getMasterDB();
 708
 709                 return array(
 710                         // Fields that describe the nature of the job
 711                         'job_cmd' => $job->getType(),
 712                         'job_namespace' => $job->getTitle()->getNamespace(),
 713                         'job_title' => $job->getTitle()->getDBkey(),
 714                         'job_params' => self::makeBlob( $job->getParams() ),
 715                         // Additional job metadata
 716                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 717                         'job_timestamp' => $dbw->timestamp(),
 718                         'job_sha1' => wfBaseConvert(
 719                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 720                                 16, 36, 31
 721                         ),
 722                         'job_random' => mt_rand( 0, self::MAX_JOB_RANDOM )
 723                 );
 724         }
 725
 726         /**
 727          * @throws JobQueueConnectionError
 728          * @return DBConnRef
 729          */
 730         protected function getSlaveDB() {
 731                 try {
 732                         return $this->getDB( DB_SLAVE );
 733                 } catch ( DBConnectionError $e ) {
 734                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 735                 }
 736         }
 737
 738         /**
 739          * @throws JobQueueConnectionError
 740          * @return DBConnRef
 741          */
 742         protected function getMasterDB() {
 743                 try {
 744                         return $this->getDB( DB_MASTER );
 745                 } catch ( DBConnectionError $e ) {
 746                         throw new JobQueueConnectionError( "DBConnectionError:" . $e->getMessage() );
 747                 }
 748         }
 749
 750         /**
 751          * @param int $index (DB_SLAVE/DB_MASTER)
 752          * @return DBConnRef
 753          */
 754         protected function getDB( $index ) {
 755                 $lb = ( $this->cluster !== false )
 756                         ? wfGetLBFactory()->getExternalLB( $this->cluster, $this->wiki )
 757                         : wfGetLB( $this->wiki );
 758
 759                 return $lb->getConnectionRef( $index, array(), $this->wiki );
 760         }
 761
 762         /**
 763          * @param string $property
 764          * @return string
 765          */
 766         private function getCacheKey( $property ) {
 767                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 768                 $cluster = is_string( $this->cluster ) ? $this->cluster : 'main';
 769
 770                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $cluster, $this->type, $property );
 771         }
 772
 773         /**
 774          * @param array|bool $params
 775          * @return string
 776          */
 777         protected static function makeBlob( $params ) {
 778                 if ( $params !== false ) {
 779                         return serialize( $params );
 780                 } else {
 781                         return '';
 782                 }
 783         }
 784
 785         /**
 786          * @param string $blob
 787          * @return bool|mixed
 788          */
 789         protected static function extractBlob( $blob ) {
 790                 if ( (string)$blob !== '' ) {
 791                         return unserialize( $blob );
 792                 } else {
 793                         return false;
 794                 }
 795         }
 796
 797         /**
 798          * @param DBError $e
 799          * @throws JobQueueError
 800          */
 801         protected function throwDBException( DBError $e ) {
 802                 throw new JobQueueError( get_class( $e ) . ": " . $e->getMessage() );
 803         }
 804
 805         /**
 806          * Return the list of job fields that should be selected.
 807          * @since 1.23
 808          * @return array
 809          */
 810         public static function selectFields() {
 811                 return array(
 812                         'job_id',
 813                         'job_cmd',
 814                         'job_namespace',
 815                         'job_title',
 816                         'job_timestamp',
 817                         'job_params',
 818                         'job_random',
 819                         'job_attempts',
 820                         'job_token',
 821                         'job_token_timestamp',
 822                         'job_sha1',
 823                 );
 824         }
 825 }