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          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 165          * Queue classes should use caching if they are any slower without memcached.
 166          *
 167          * If caching is used, this might return false when there are actually no jobs.
 168          * If pop() is called and returns false then it should correct the cache. Also,
 169          * calling flushCaches() first prevents this. However, this affect is typically
 170          * not distinguishable from the race condition between isEmpty() and pop().
 171          *
 172          * @return bool
 173          * @throws JobQueueError
 174          */
 175         final public function isEmpty() {
 176                 $res = $this->doIsEmpty();
 177
 178                 return $res;
 179         }
 180
 181         /**
 182          * @see JobQueue::isEmpty()
 183          * @return bool
 184          */
 185         abstract protected function doIsEmpty();
 186
 187         /**
 188          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 189          * Queue classes should use caching if they are any slower without memcached.
 190          *
 191          * If caching is used, this number might be out of date for a minute.
 192          *
 193          * @return int
 194          * @throws JobQueueError
 195          */
 196         final public function getSize() {
 197                 $res = $this->doGetSize();
 198
 199                 return $res;
 200         }
 201
 202         /**
 203          * @see JobQueue::getSize()
 204          * @return int
 205          */
 206         abstract protected function doGetSize();
 207
 208         /**
 209          * Get the number of acquired jobs (these are temporarily out of the queue).
 210          * Queue classes should use caching if they are any slower without memcached.
 211          *
 212          * If caching is used, this number might be out of date for a minute.
 213          *
 214          * @return int
 215          * @throws JobQueueError
 216          */
 217         final public function getAcquiredCount() {
 218                 $res = $this->doGetAcquiredCount();
 219
 220                 return $res;
 221         }
 222
 223         /**
 224          * @see JobQueue::getAcquiredCount()
 225          * @return int
 226          */
 227         abstract protected function doGetAcquiredCount();
 228
 229         /**
 230          * Get the number of delayed jobs (these are temporarily out of the queue).
 231          * Queue classes should use caching if they are any slower without memcached.
 232          *
 233          * If caching is used, this number might be out of date for a minute.
 234          *
 235          * @return int
 236          * @throws JobQueueError
 237          * @since 1.22
 238          */
 239         final public function getDelayedCount() {
 240                 $res = $this->doGetDelayedCount();
 241
 242                 return $res;
 243         }
 244
 245         /**
 246          * @see JobQueue::getDelayedCount()
 247          * @return int
 248          */
 249         protected function doGetDelayedCount() {
 250                 return 0; // not implemented
 251         }
 252
 253         /**
 254          * Get the number of acquired jobs that can no longer be attempted.
 255          * Queue classes should use caching if they are any slower without memcached.
 256          *
 257          * If caching is used, this number might be out of date for a minute.
 258          *
 259          * @return int
 260          * @throws JobQueueError
 261          */
 262         final public function getAbandonedCount() {
 263                 $res = $this->doGetAbandonedCount();
 264
 265                 return $res;
 266         }
 267
 268         /**
 269          * @see JobQueue::getAbandonedCount()
 270          * @return int
 271          */
 272         protected function doGetAbandonedCount() {
 273                 return 0; // not implemented
 274         }
 275
 276         /**
 277          * Push one or more jobs into the queue.
 278          * This does not require $wgJobClasses to be set for the given job type.
 279          * Outside callers should use JobQueueGroup::push() instead of this function.
 280          *
 281          * @param Job|array $jobs A single job or an array of Jobs
 282          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 283          * @return void
 284          * @throws JobQueueError
 285          */
 286         final public function push( $jobs, $flags = 0 ) {
 287                 $jobs = is_array( $jobs ) ? $jobs : array( $jobs );
 288                 $this->batchPush( $jobs, $flags );
 289         }
 290
 291         /**
 292          * Push a batch of jobs into the queue.
 293          * This does not require $wgJobClasses to be set for the given job type.
 294          * Outside callers should use JobQueueGroup::push() instead of this function.
 295          *
 296          * @param array $jobs List of Jobs
 297          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 298          * @return void
 299          * @throws MWException
 300          */
 301         final public function batchPush( array $jobs, $flags = 0 ) {
 302                 if ( !count( $jobs ) ) {
 303                         return; // nothing to do
 304                 }
 305
 306                 foreach ( $jobs as $job ) {
 307                         if ( $job->getType() !== $this->type ) {
 308                                 throw new MWException(
 309                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 310                         } elseif ( $job->getReleaseTimestamp() && !$this->supportsDelayedJobs() ) {
 311                                 throw new MWException(
 312                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 313                         }
 314                 }
 315
 316                 $this->doBatchPush( $jobs, $flags );
 317                 $this->aggr->notifyQueueNonEmpty( $this->wiki, $this->type );
 318         }
 319
 320         /**
 321          * @see JobQueue::batchPush()
 322          * @param array $jobs
 323          * @param int $flags
 324          */
 325         abstract protected function doBatchPush( array $jobs, $flags );
 326
 327         /**
 328          * Pop a job off of the queue.
 329          * This requires $wgJobClasses to be set for the given job type.
 330          * Outside callers should use JobQueueGroup::pop() instead of this function.
 331          *
 332          * @throws MWException
 333          * @return Job|bool Returns false if there are no jobs
 334          */
 335         final public function pop() {
 336                 global $wgJobClasses;
 337
 338                 if ( $this->wiki !== wfWikiID() ) {
 339                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 340                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 341                         // Do not pop jobs if there is no class for the queue type
 342                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 343                 }
 344
 345                 $job = $this->doPop();
 346
 347                 if ( !$job ) {
 348                         $this->aggr->notifyQueueEmpty( $this->wiki, $this->type );
 349                 }
 350
 351                 // Flag this job as an old duplicate based on its "root" job...
 352                 try {
 353                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 354                                 JobQueue::incrStats( 'job-pop-duplicate', $this->type, 1, $this->wiki );
 355                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 356                         }
 357                 } catch ( Exception $e ) {
 358                         // don't lose jobs over this
 359                 }
 360
 361                 return $job;
 362         }
 363
 364         /**
 365          * @see JobQueue::pop()
 366          * @return Job
 367          */
 368         abstract protected function doPop();
 369
 370         /**
 371          * Acknowledge that a job was completed.
 372          *
 373          * This does nothing for certain queue classes or if "claimTTL" is not set.
 374          * Outside callers should use JobQueueGroup::ack() instead of this function.
 375          *
 376          * @param Job $job
 377          * @return void
 378          * @throws MWException
 379          */
 380         final public function ack( Job $job ) {
 381                 if ( $job->getType() !== $this->type ) {
 382                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 383                 }
 384                 $this->doAck( $job );
 385         }
 386
 387         /**
 388          * @see JobQueue::ack()
 389          * @param Job $job
 390          */
 391         abstract protected function doAck( Job $job );
 392
 393         /**
 394          * Register the "root job" of a given job into the queue for de-duplication.
 395          * This should only be called right *after* all the new jobs have been inserted.
 396          * This is used to turn older, duplicate, job entries into no-ops. The root job
 397          * information will remain in the registry until it simply falls out of cache.
 398          *
 399          * This requires that $job has two special fields in the "params" array:
 400          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 401          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 402          *
 403          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 404          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 405          * spawned when a template is edited. One can think of the task as "update links
 406          * of pages that use template X" and an instance of that task as a "root job".
 407          * However, what actually goes into the queue are range and leaf job subtypes.
 408          * Since these jobs include things like page ID ranges and DB master positions,
 409          * and can morph into smaller jobs recursively, simple duplicate detection
 410          * for individual jobs being identical (like that of job_sha1) is not useful.
 411          *
 412          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 413          * is edited again, we want all of these old refreshLinks jobs for that template to become
 414          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 415          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 416          * previous "root job" for the same task of "update links of pages that use template X".
 417          *
 418          * This does nothing for certain queue classes.
 419          *
 420          * @param Job $job
 421          * @throws MWException
 422          * @return bool
 423          */
 424         final public function deduplicateRootJob( Job $job ) {
 425                 if ( $job->getType() !== $this->type ) {
 426                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 427                 }
 428                 $ok = $this->doDeduplicateRootJob( $job );
 429
 430                 return $ok;
 431         }
 432
 433         /**
 434          * @see JobQueue::deduplicateRootJob()
 435          * @param Job $job
 436          * @throws MWException
 437          * @return bool
 438          */
 439         protected function doDeduplicateRootJob( Job $job ) {
 440                 if ( !$job->hasRootJobParams() ) {
 441                         throw new MWException( "Cannot register root job; missing parameters." );
 442                 }
 443                 $params = $job->getRootJobParams();
 444
 445                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 446                 // Callers should call batchInsert() and then this function so that if the insert
 447                 // fails, the de-duplication registration will be aborted. Since the insert is
 448                 // deferred till "transaction idle", do the same here, so that the ordering is
 449                 // maintained. Having only the de-duplication registration succeed would cause
 450                 // jobs to become no-ops without any actual jobs that made them redundant.
 451                 $timestamp = $this->dupCache->get( $key ); // current last timestamp of this job
 452                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 453                         return true; // a newer version of this root job was enqueued
 454                 }
 455
 456                 // Update the timestamp of the last root job started at the location...
 457                 return $this->dupCache->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 458         }
 459
 460         /**
 461          * Check if the "root" job of a given job has been superseded by a newer one
 462          *
 463          * @param Job $job
 464          * @throws MWException
 465          * @return bool
 466          */
 467         final protected function isRootJobOldDuplicate( Job $job ) {
 468                 if ( $job->getType() !== $this->type ) {
 469                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 470                 }
 471                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 472
 473                 return $isDuplicate;
 474         }
 475
 476         /**
 477          * @see JobQueue::isRootJobOldDuplicate()
 478          * @param Job $job
 479          * @return bool
 480          */
 481         protected function doIsRootJobOldDuplicate( Job $job ) {
 482                 if ( !$job->hasRootJobParams() ) {
 483                         return false; // job has no de-deplication info
 484                 }
 485                 $params = $job->getRootJobParams();
 486
 487                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 488                 // Get the last time this root job was enqueued
 489                 $timestamp = $this->dupCache->get( $key );
 490
 491                 // Check if a new root job was started at the location after this one's...
 492                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 493         }
 494
 495         /**
 496          * @param string $signature Hash identifier of the root job
 497          * @return string
 498          */
 499         protected function getRootJobCacheKey( $signature ) {
 500                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 501
 502                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 503         }
 504
 505         /**
 506          * Deleted all unclaimed and delayed jobs from the queue
 507          *
 508          * @throws JobQueueError
 509          * @since 1.22
 510          * @return void
 511          */
 512         final public function delete() {
 513                 $this->doDelete();
 514         }
 515
 516         /**
 517          * @see JobQueue::delete()
 518          * @throws MWException
 519          */
 520         protected function doDelete() {
 521                 throw new MWException( "This method is not implemented." );
 522         }
 523
 524         /**
 525          * Wait for any slaves or backup servers to catch up.
 526          *
 527          * This does nothing for certain queue classes.
 528          *
 529          * @return void
 530          * @throws JobQueueError
 531          */
 532         final public function waitForBackups() {
 533                 $this->doWaitForBackups();
 534         }
 535
 536         /**
 537          * @see JobQueue::waitForBackups()
 538          * @return void
 539          */
 540         protected function doWaitForBackups() {
 541         }
 542
 543         /**
 544          * Return a map of task names to task definition maps.
 545          * A "task" is a fast periodic queue maintenance action.
 546          * Mutually exclusive tasks must implement their own locking in the callback.
 547          *
 548          * Each task value is an associative array with:
 549          *   - name     : the name of the task
 550          *   - callback : a PHP callable that performs the task
 551          *   - period   : the period in seconds corresponding to the task frequency
 552          *
 553          * @return array
 554          */
 555         final public function getPeriodicTasks() {
 556                 $tasks = $this->doGetPeriodicTasks();
 557                 foreach ( $tasks as $name => &$def ) {
 558                         $def['name'] = $name;
 559                 }
 560
 561                 return $tasks;
 562         }
 563
 564         /**
 565          * @see JobQueue::getPeriodicTasks()
 566          * @return array
 567          */
 568         protected function doGetPeriodicTasks() {
 569                 return array();
 570         }
 571
 572         /**
 573          * Clear any process and persistent caches
 574          *
 575          * @return void
 576          */
 577         final public function flushCaches() {
 578                 $this->doFlushCaches();
 579         }
 580
 581         /**
 582          * @see JobQueue::flushCaches()
 583          * @return void
 584          */
 585         protected function doFlushCaches() {
 586         }
 587
 588         /**
 589          * Get an iterator to traverse over all available jobs in this queue.
 590          * This does not include jobs that are currently acquired or delayed.
 591          * Note: results may be stale if the queue is concurrently modified.
 592          *
 593          * @return Iterator
 594          * @throws JobQueueError
 595          */
 596         abstract public function getAllQueuedJobs();
 597
 598         /**
 599          * Get an iterator to traverse over all delayed jobs in this queue.
 600          * Note: results may be stale if the queue is concurrently modified.
 601          *
 602          * @return Iterator
 603          * @throws JobQueueError
 604          * @since 1.22
 605          */
 606         public function getAllDelayedJobs() {
 607                 return new ArrayIterator( array() ); // not implemented
 608         }
 609
 610         /**
 611          * Do not use this function outside of JobQueue/JobQueueGroup
 612          *
 613          * @return string
 614          * @since 1.22
 615          */
 616         public function getCoalesceLocationInternal() {
 617                 return null;
 618         }
 619
 620         /**
 621          * Check whether each of the given queues are empty.
 622          * This is used for batching checks for queues stored at the same place.
 623          *
 624          * @param array $types List of queues types
 625          * @return array|null (list of non-empty queue types) or null if unsupported
 626          * @throws MWException
 627          * @since 1.22
 628          */
 629         final public function getSiblingQueuesWithJobs( array $types ) {
 630
 631                 return $this->doGetSiblingQueuesWithJobs( $types );
 632         }
 633
 634         /**
 635          * @see JobQueue::getSiblingQueuesWithJobs()
 636          * @param array $types List of queues types
 637          * @return array|null (list of queue types) or null if unsupported
 638          */
 639         protected function doGetSiblingQueuesWithJobs( array $types ) {
 640                 return null; // not supported
 641         }
 642
 643         /**
 644          * Check the size of each of the given queues.
 645          * For queues not served by the same store as this one, 0 is returned.
 646          * This is used for batching checks for queues stored at the same place.
 647          *
 648          * @param array $types List of queues types
 649          * @return array|null (job type => whether queue is empty) or null if unsupported
 650          * @throws MWException
 651          * @since 1.22
 652          */
 653         final public function getSiblingQueueSizes( array $types ) {
 654
 655                 return $this->doGetSiblingQueueSizes( $types );
 656         }
 657
 658         /**
 659          * @see JobQueue::getSiblingQueuesSize()
 660          * @param array $types List of queues types
 661          * @return array|null (list of queue types) or null if unsupported
 662          */
 663         protected function doGetSiblingQueueSizes( array $types ) {
 664                 return null; // not supported
 665         }
 666
 667         /**
 668          * Call wfIncrStats() for the queue overall and for the queue type
 669          *
 670          * @param string $key Event type
 671          * @param string $type Job type
 672          * @param int $delta
 673          * @param string $wiki Wiki ID (added in 1.23)
 674          * @since 1.22
 675          */
 676         public static function incrStats( $key, $type, $delta = 1, $wiki = null ) {
 677                 wfIncrStats( $key, $delta );
 678                 wfIncrStats( "{$key}-{$type}", $delta );
 679                 if ( $wiki !== null ) {
 680                         wfIncrStats( "{$key}-{$type}-{$wiki}", $delta );
 681                 }
 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 }