includes/jobqueue/JobQueue.php

   1 <?php
   2 /**
   3  * Job queue base 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  * @defgroup JobQueue JobQueue
  22  * @author Aaron Schulz
  23  */
  24
  25 /**
  26  * Class to handle enqueueing and running of background jobs
  27  *
  28  * @ingroup JobQueue
  29  * @since 1.21
  30  */
  31 abstract class JobQueue {
  32         /** @var string Wiki ID */
  33         protected $wiki;
  34
  35         /** @var string Job type */
  36         protected $type;
  37
  38         /** @var string Job priority for pop() */
  39         protected $order;
  40
  41         /** @var int Time to live in seconds */
  42         protected $claimTTL;
  43
  44         /** @var int Maximum number of times to try a job */
  45         protected $maxTries;
  46
  47         /** @var BagOStuff */
  48         protected $dupCache;
  49         /** @var JobQueueAggregator */
  50         protected $aggr;
  51
  52         const QOS_ATOMIC = 1; // integer; "all-or-nothing" job insertions
  53
  54         const ROOTJOB_TTL = 2419200; // integer; seconds to remember root jobs (28 days)
  55
  56         /**
  57          * @param array $params
  58          * @throws MWException
  59          */
  60         protected function __construct( array $params ) {
  61                 $this->wiki = $params['wiki'];
  62                 $this->type = $params['type'];
  63                 $this->claimTTL = isset( $params['claimTTL'] ) ? $params['claimTTL'] : 0;
  64                 $this->maxTries = isset( $params['maxTries'] ) ? $params['maxTries'] : 3;
  65                 if ( isset( $params['order'] ) && $params['order'] !== 'any' ) {
  66                         $this->order = $params['order'];
  67                 } else {
  68                         $this->order = $this->optimalOrder();
  69                 }
  70                 if ( !in_array( $this->order, $this->supportedOrders() ) ) {
  71                         throw new MWException( __CLASS__ . " does not support '{$this->order}' order." );
  72                 }
  73                 $this->dupCache = wfGetCache( CACHE_ANYTHING );
  74                 $this->aggr = isset( $params['aggregator'] )
  75                         ? $params['aggregator']
  76                         : new JobQueueAggregatorNull( array() );
  77         }
  78
  79         /**
  80          * Get a job queue object of the specified type.
  81          * $params includes:
  82          *   - class      : What job class to use (determines job type)
  83          *   - wiki       : wiki ID of the wiki the jobs are for (defaults to current wiki)
  84          *   - type       : The name of the job types this queue handles
  85          *   - order      : Order that pop() selects jobs, one of "fifo", "timestamp" or "random".
  86          *                  If "fifo" is used, the queue will effectively be FIFO. Note that job
  87          *                  completion will not appear to be exactly FIFO if there are multiple
  88          *                  job runners since jobs can take different times to finish once popped.
  89          *                  If "timestamp" is used, the queue will at least be loosely ordered
  90          *                  by timestamp, allowing for some jobs to be popped off out of order.
  91          *                  If "random" is used, pop() will pick jobs in random order.
  92          *                  Note that it may only be weakly random (e.g. a lottery of the oldest X).
  93          *                  If "any" is choosen, the queue will use whatever order is the fastest.
  94          *                  This might be useful for improving concurrency for job acquisition.
  95          *   - claimTTL   : If supported, the queue will recycle jobs that have been popped
  96          *                  but not acknowledged as completed after this many seconds. Recycling
  97          *                  of jobs simple means re-inserting them into the queue. Jobs can be
  98          *                  attempted up to three times before being discarded.
  99          *
 100          * Queue classes should throw an exception if they do not support the options given.
 101          *
 102          * @param array $params
 103          * @return JobQueue
 104          * @throws MWException
 105          */
 106         final public static function factory( array $params ) {
 107                 $class = $params['class'];
 108                 if ( !class_exists( $class ) ) {
 109                         throw new MWException( "Invalid job queue class '$class'." );
 110                 }
 111                 $obj = new $class( $params );
 112                 if ( !( $obj instanceof self ) ) {
 113                         throw new MWException( "Class '$class' is not a " . __CLASS__ . " class." );
 114                 }
 115
 116                 return $obj;
 117         }
 118
 119         /**
 120          * @return string Wiki ID
 121          */
 122         final public function getWiki() {
 123                 return $this->wiki;
 124         }
 125
 126         /**
 127          * @return string Job type that this queue handles
 128          */
 129         final public function getType() {
 130                 return $this->type;
 131         }
 132
 133         /**
 134          * @return string One of (random, timestamp, fifo, undefined)
 135          */
 136         final public function getOrder() {
 137                 return $this->order;
 138         }
 139
 140         /**
 141          * Get the allowed queue orders for configuration validation
 142          *
 143          * @return array Subset of (random, timestamp, fifo, undefined)
 144          */
 145         abstract protected function supportedOrders();
 146
 147         /**
 148          * Get the default queue order to use if configuration does not specify one
 149          *
 150          * @return string One of (random, timestamp, fifo, undefined)
 151          */
 152         abstract protected function optimalOrder();
 153
 154         /**
 155          * Find out if delayed jobs are supported for configuration validation
 156          *
 157          * @return bool Whether delayed jobs are supported
 158          */
 159         protected function supportsDelayedJobs() {
 160                 return false; // not implemented
 161         }
 162
 163         /**
 164          * @return bool Whether delayed jobs are enabled
 165          * @since 1.22
 166          */
 167         final public function delayedJobsEnabled() {
 168                 return $this->supportsDelayedJobs();
 169         }
 170
 171         /**
 172          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 173          * Queue classes should use caching if they are any slower without memcached.
 174          *
 175          * If caching is used, this might return false when there are actually no jobs.
 176          * If pop() is called and returns false then it should correct the cache. Also,
 177          * calling flushCaches() first prevents this. However, this affect is typically
 178          * not distinguishable from the race condition between isEmpty() and pop().
 179          *
 180          * @return bool
 181          * @throws JobQueueError
 182          */
 183         final public function isEmpty() {
 184                 $res = $this->doIsEmpty();
 185
 186                 return $res;
 187         }
 188
 189         /**
 190          * @see JobQueue::isEmpty()
 191          * @return bool
 192          */
 193         abstract protected function doIsEmpty();
 194
 195         /**
 196          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 197          * Queue classes should use caching if they are any slower without memcached.
 198          *
 199          * If caching is used, this number might be out of date for a minute.
 200          *
 201          * @return int
 202          * @throws JobQueueError
 203          */
 204         final public function getSize() {
 205                 $res = $this->doGetSize();
 206
 207                 return $res;
 208         }
 209
 210         /**
 211          * @see JobQueue::getSize()
 212          * @return int
 213          */
 214         abstract protected function doGetSize();
 215
 216         /**
 217          * Get the number of acquired jobs (these are temporarily out of the queue).
 218          * Queue classes should use caching if they are any slower without memcached.
 219          *
 220          * If caching is used, this number might be out of date for a minute.
 221          *
 222          * @return int
 223          * @throws JobQueueError
 224          */
 225         final public function getAcquiredCount() {
 226                 $res = $this->doGetAcquiredCount();
 227
 228                 return $res;
 229         }
 230
 231         /**
 232          * @see JobQueue::getAcquiredCount()
 233          * @return int
 234          */
 235         abstract protected function doGetAcquiredCount();
 236
 237         /**
 238          * Get the number of delayed jobs (these are temporarily out of the queue).
 239          * Queue classes should use caching if they are any slower without memcached.
 240          *
 241          * If caching is used, this number might be out of date for a minute.
 242          *
 243          * @return int
 244          * @throws JobQueueError
 245          * @since 1.22
 246          */
 247         final public function getDelayedCount() {
 248                 $res = $this->doGetDelayedCount();
 249
 250                 return $res;
 251         }
 252
 253         /**
 254          * @see JobQueue::getDelayedCount()
 255          * @return int
 256          */
 257         protected function doGetDelayedCount() {
 258                 return 0; // not implemented
 259         }
 260
 261         /**
 262          * Get the number of acquired jobs that can no longer be attempted.
 263          * Queue classes should use caching if they are any slower without memcached.
 264          *
 265          * If caching is used, this number might be out of date for a minute.
 266          *
 267          * @return int
 268          * @throws JobQueueError
 269          */
 270         final public function getAbandonedCount() {
 271                 $res = $this->doGetAbandonedCount();
 272
 273                 return $res;
 274         }
 275
 276         /**
 277          * @see JobQueue::getAbandonedCount()
 278          * @return int
 279          */
 280         protected function doGetAbandonedCount() {
 281                 return 0; // not implemented
 282         }
 283
 284         /**
 285          * Push one or more jobs into the queue.
 286          * This does not require $wgJobClasses to be set for the given job type.
 287          * Outside callers should use JobQueueGroup::push() instead of this function.
 288          *
 289          * @param JobSpecification|JobSpecification[] $jobs
 290          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 291          * @return void
 292          * @throws JobQueueError
 293          */
 294         final public function push( $jobs, $flags = 0 ) {
 295                 $jobs = is_array( $jobs ) ? $jobs : array( $jobs );
 296                 $this->batchPush( $jobs, $flags );
 297         }
 298
 299         /**
 300          * Push a batch of jobs into the queue.
 301          * This does not require $wgJobClasses to be set for the given job type.
 302          * Outside callers should use JobQueueGroup::push() instead of this function.
 303          *
 304          * @param JobSpecification[] $jobs
 305          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 306          * @return void
 307          * @throws MWException
 308          */
 309         final public function batchPush( array $jobs, $flags = 0 ) {
 310                 if ( !count( $jobs ) ) {
 311                         return; // nothing to do
 312                 }
 313
 314                 foreach ( $jobs as $job ) {
 315                         if ( $job->getType() !== $this->type ) {
 316                                 throw new MWException(
 317                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 318                         } elseif ( $job->getReleaseTimestamp() && !$this->supportsDelayedJobs() ) {
 319                                 throw new MWException(
 320                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 321                         }
 322                 }
 323
 324                 $this->doBatchPush( $jobs, $flags );
 325                 $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 326         }
 327
 328         /**
 329          * @see JobQueue::batchPush()
 330          * @param JobSpecification[] $jobs
 331          * @param int $flags
 332          */
 333         abstract protected function doBatchPush( array $jobs, $flags );
 334
 335         /**
 336          * Pop a job off of the queue.
 337          * This requires $wgJobClasses to be set for the given job type.
 338          * Outside callers should use JobQueueGroup::pop() instead of this function.
 339          *
 340          * @throws MWException
 341          * @return Job|bool Returns false if there are no jobs
 342          */
 343         final public function pop() {
 344                 global $wgJobClasses;
 345
 346                 if ( $this->wiki !== wfWikiID() ) {
 347                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 348                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 349                         // Do not pop jobs if there is no class for the queue type
 350                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 351                 }
 352
 353                 $job = $this->doPop();
 354
 355                 if ( !$job ) {
 356                         $this->aggr->notifyQueueEmpty( $this->wiki, $this->type );
 357                 }
 358
 359                 // Flag this job as an old duplicate based on its "root" job...
 360                 try {
 361                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 362                                 JobQueue::incrStats( 'job-pop-duplicate', $this->type );
 363                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 364                         }
 365                 } catch ( Exception $e ) {
 366                         // don't lose jobs over this
 367                 }
 368
 369                 return $job;
 370         }
 371
 372         /**
 373          * @see JobQueue::pop()
 374          * @return Job
 375          */
 376         abstract protected function doPop();
 377
 378         /**
 379          * Acknowledge that a job was completed.
 380          *
 381          * This does nothing for certain queue classes or if "claimTTL" is not set.
 382          * Outside callers should use JobQueueGroup::ack() instead of this function.
 383          *
 384          * @param Job $job
 385          * @return void
 386          * @throws MWException
 387          */
 388         final public function ack( Job $job ) {
 389                 if ( $job->getType() !== $this->type ) {
 390                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 391                 }
 392                 $this->doAck( $job );
 393         }
 394
 395         /**
 396          * @see JobQueue::ack()
 397          * @param Job $job
 398          */
 399         abstract protected function doAck( Job $job );
 400
 401         /**
 402          * Register the "root job" of a given job into the queue for de-duplication.
 403          * This should only be called right *after* all the new jobs have been inserted.
 404          * This is used to turn older, duplicate, job entries into no-ops. The root job
 405          * information will remain in the registry until it simply falls out of cache.
 406          *
 407          * This requires that $job has two special fields in the "params" array:
 408          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 409          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 410          *
 411          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 412          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 413          * spawned when a template is edited. One can think of the task as "update links
 414          * of pages that use template X" and an instance of that task as a "root job".
 415          * However, what actually goes into the queue are range and leaf job subtypes.
 416          * Since these jobs include things like page ID ranges and DB master positions,
 417          * and can morph into smaller jobs recursively, simple duplicate detection
 418          * for individual jobs being identical (like that of job_sha1) is not useful.
 419          *
 420          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 421          * is edited again, we want all of these old refreshLinks jobs for that template to become
 422          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 423          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 424          * previous "root job" for the same task of "update links of pages that use template X".
 425          *
 426          * This does nothing for certain queue classes.
 427          *
 428          * @param IJobSpecification $job
 429          * @throws MWException
 430          * @return bool
 431          */
 432         final public function deduplicateRootJob( IJobSpecification $job ) {
 433                 if ( $job->getType() !== $this->type ) {
 434                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 435                 }
 436                 $ok = $this->doDeduplicateRootJob( $job );
 437
 438                 return $ok;
 439         }
 440
 441         /**
 442          * @see JobQueue::deduplicateRootJob()
 443          * @param IJobSpecification $job
 444          * @throws MWException
 445          * @return bool
 446          */
 447         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 448                 if ( !$job->hasRootJobParams() ) {
 449                         throw new MWException( "Cannot register root job; missing parameters." );
 450                 }
 451                 $params = $job->getRootJobParams();
 452
 453                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 454                 // Callers should call batchInsert() and then this function so that if the insert
 455                 // fails, the de-duplication registration will be aborted. Since the insert is
 456                 // deferred till "transaction idle", do the same here, so that the ordering is
 457                 // maintained. Having only the de-duplication registration succeed would cause
 458                 // jobs to become no-ops without any actual jobs that made them redundant.
 459                 $timestamp = $this->dupCache->get( $key ); // current last timestamp of this job
 460                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 461                         return true; // a newer version of this root job was enqueued
 462                 }
 463
 464                 // Update the timestamp of the last root job started at the location...
 465                 return $this->dupCache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 466         }
 467
 468         /**
 469          * Check if the "root" job of a given job has been superseded by a newer one
 470          *
 471          * @param Job $job
 472          * @throws MWException
 473          * @return bool
 474          */
 475         final protected function isRootJobOldDuplicate( Job $job ) {
 476                 if ( $job->getType() !== $this->type ) {
 477                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 478                 }
 479                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 480
 481                 return $isDuplicate;
 482         }
 483
 484         /**
 485          * @see JobQueue::isRootJobOldDuplicate()
 486          * @param Job $job
 487          * @return bool
 488          */
 489         protected function doIsRootJobOldDuplicate( Job $job ) {
 490                 if ( !$job->hasRootJobParams() ) {
 491                         return false; // job has no de-deplication info
 492                 }
 493                 $params = $job->getRootJobParams();
 494
 495                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 496                 // Get the last time this root job was enqueued
 497                 $timestamp = $this->dupCache->get( $key );
 498
 499                 // Check if a new root job was started at the location after this one's...
 500                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 501         }
 502
 503         /**
 504          * @param string $signature Hash identifier of the root job
 505          * @return string
 506          */
 507         protected function getRootJobCacheKey( $signature ) {
 508                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 509
 510                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 511         }
 512
 513         /**
 514          * Deleted all unclaimed and delayed jobs from the queue
 515          *
 516          * @throws JobQueueError
 517          * @since 1.22
 518          * @return void
 519          */
 520         final public function delete() {
 521                 $this->doDelete();
 522         }
 523
 524         /**
 525          * @see JobQueue::delete()
 526          * @throws MWException
 527          */
 528         protected function doDelete() {
 529                 throw new MWException( "This method is not implemented." );
 530         }
 531
 532         /**
 533          * Wait for any slaves or backup servers to catch up.
 534          *
 535          * This does nothing for certain queue classes.
 536          *
 537          * @return void
 538          * @throws JobQueueError
 539          */
 540         final public function waitForBackups() {
 541                 $this->doWaitForBackups();
 542         }
 543
 544         /**
 545          * @see JobQueue::waitForBackups()
 546          * @return void
 547          */
 548         protected function doWaitForBackups() {
 549         }
 550
 551         /**
 552          * Clear any process and persistent caches
 553          *
 554          * @return void
 555          */
 556         final public function flushCaches() {
 557                 $this->doFlushCaches();
 558         }
 559
 560         /**
 561          * @see JobQueue::flushCaches()
 562          * @return void
 563          */
 564         protected function doFlushCaches() {
 565         }
 566
 567         /**
 568          * Get an iterator to traverse over all available jobs in this queue.
 569          * This does not include jobs that are currently acquired or delayed.
 570          * Note: results may be stale if the queue is concurrently modified.
 571          *
 572          * @return Iterator
 573          * @throws JobQueueError
 574          */
 575         abstract public function getAllQueuedJobs();
 576
 577         /**
 578          * Get an iterator to traverse over all delayed jobs in this queue.
 579          * Note: results may be stale if the queue is concurrently modified.
 580          *
 581          * @return Iterator
 582          * @throws JobQueueError
 583          * @since 1.22
 584          */
 585         public function getAllDelayedJobs() {
 586                 return new ArrayIterator( array() ); // not implemented
 587         }
 588
 589         /**
 590          * Get an iterator to traverse over all claimed jobs in this queue
 591          *
 592          * Callers should be quick to iterator over it or few results
 593          * will be returned due to jobs being acknowledged and deleted
 594          *
 595          * @return Iterator
 596          * @throws JobQueueError
 597          * @since 1.26
 598          */
 599         public function getAllAcquiredJobs() {
 600                 return new ArrayIterator( array() ); // not implemented
 601         }
 602
 603         /**
 604          * Get an iterator to traverse over all abandoned jobs in this queue
 605          *
 606          * @return Iterator
 607          * @throws JobQueueError
 608          * @since 1.25
 609          */
 610         public function getAllAbandonedJobs() {
 611                 return new ArrayIterator( array() ); // not implemented
 612         }
 613
 614         /**
 615          * Do not use this function outside of JobQueue/JobQueueGroup
 616          *
 617          * @return string
 618          * @since 1.22
 619          */
 620         public function getCoalesceLocationInternal() {
 621                 return null;
 622         }
 623
 624         /**
 625          * Check whether each of the given queues are empty.
 626          * This is used for batching checks for queues stored at the same place.
 627          *
 628          * @param array $types List of queues types
 629          * @return array|null (list of non-empty queue types) or null if unsupported
 630          * @throws MWException
 631          * @since 1.22
 632          */
 633         final public function getSiblingQueuesWithJobs( array $types ) {
 634
 635                 return $this->doGetSiblingQueuesWithJobs( $types );
 636         }
 637
 638         /**
 639          * @see JobQueue::getSiblingQueuesWithJobs()
 640          * @param array $types List of queues types
 641          * @return array|null (list of queue types) or null if unsupported
 642          */
 643         protected function doGetSiblingQueuesWithJobs( array $types ) {
 644                 return null; // not supported
 645         }
 646
 647         /**
 648          * Check the size of each of the given queues.
 649          * For queues not served by the same store as this one, 0 is returned.
 650          * This is used for batching checks for queues stored at the same place.
 651          *
 652          * @param array $types List of queues types
 653          * @return array|null (job type => whether queue is empty) or null if unsupported
 654          * @throws MWException
 655          * @since 1.22
 656          */
 657         final public function getSiblingQueueSizes( array $types ) {
 658
 659                 return $this->doGetSiblingQueueSizes( $types );
 660         }
 661
 662         /**
 663          * @see JobQueue::getSiblingQueuesSize()
 664          * @param array $types List of queues types
 665          * @return array|null (list of queue types) or null if unsupported
 666          */
 667         protected function doGetSiblingQueueSizes( array $types ) {
 668                 return null; // not supported
 669         }
 670
 671         /**
 672          * Call wfIncrStats() for the queue overall and for the queue type
 673          *
 674          * @param string $key Event type
 675          * @param string $type Job type
 676          * @param int $delta
 677          * @since 1.22
 678          */
 679         public static function incrStats( $key, $type, $delta = 1 ) {
 680                 wfIncrStats( $key, $delta );
 681                 wfIncrStats( "{$key}-{$type}", $delta );
 682         }
 683
 684         /**
 685          * Namespace the queue with a key to isolate it for testing
 686          *
 687          * @param string $key
 688          * @return void
 689          * @throws MWException
 690          */
 691         public function setTestingPrefix( $key ) {
 692                 throw new MWException( "Queue namespacing not supported for this queue type." );
 693         }
 694 }
 695
 696 /**
 697  * @ingroup JobQueue
 698  * @since 1.22
 699  */
 700 class JobQueueError extends MWException {
 701 }
 702
 703 class JobQueueConnectionError extends JobQueueError {
 704 }