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 bool Allow delayed jobs */
  48         protected $checkDelay;
  49
  50         /** @var BagOStuff */
  51         protected $dupCache;
  52
  53         const QOS_ATOMIC = 1; // integer; "all-or-nothing" job insertions
  54
  55         const ROOTJOB_TTL = 2419200; // integer; seconds to remember root jobs (28 days)
  56
  57         /**
  58          * @param array $params
  59          * @throws MWException
  60          */
  61         protected function __construct( array $params ) {
  62                 $this->wiki = $params['wiki'];
  63                 $this->type = $params['type'];
  64                 $this->claimTTL = isset( $params['claimTTL'] ) ? $params['claimTTL'] : 0;
  65                 $this->maxTries = isset( $params['maxTries'] ) ? $params['maxTries'] : 3;
  66                 if ( isset( $params['order'] ) && $params['order'] !== 'any' ) {
  67                         $this->order = $params['order'];
  68                 } else {
  69                         $this->order = $this->optimalOrder();
  70                 }
  71                 if ( !in_array( $this->order, $this->supportedOrders() ) ) {
  72                         throw new MWException( __CLASS__ . " does not support '{$this->order}' order." );
  73                 }
  74                 $this->checkDelay = !empty( $params['checkDelay'] );
  75                 if ( $this->checkDelay && !$this->supportsDelayedJobs() ) {
  76                         throw new MWException( __CLASS__ . " does not support delayed jobs." );
  77                 }
  78                 $this->dupCache = wfGetCache( CACHE_ANYTHING );
  79         }
  80
  81         /**
  82          * Get a job queue object of the specified type.
  83          * $params includes:
  84          *   - class      : What job class to use (determines job type)
  85          *   - wiki       : wiki ID of the wiki the jobs are for (defaults to current wiki)
  86          *   - type       : The name of the job types this queue handles
  87          *   - order      : Order that pop() selects jobs, one of "fifo", "timestamp" or "random".
  88          *                  If "fifo" is used, the queue will effectively be FIFO. Note that job
  89          *                  completion will not appear to be exactly FIFO if there are multiple
  90          *                  job runners since jobs can take different times to finish once popped.
  91          *                  If "timestamp" is used, the queue will at least be loosely ordered
  92          *                  by timestamp, allowing for some jobs to be popped off out of order.
  93          *                  If "random" is used, pop() will pick jobs in random order.
  94          *                  Note that it may only be weakly random (e.g. a lottery of the oldest X).
  95          *                  If "any" is choosen, the queue will use whatever order is the fastest.
  96          *                  This might be useful for improving concurrency for job acquisition.
  97          *   - claimTTL   : If supported, the queue will recycle jobs that have been popped
  98          *                  but not acknowledged as completed after this many seconds. Recycling
  99          *                  of jobs simple means re-inserting them into the queue. Jobs can be
 100          *                  attempted up to three times before being discarded.
 101          *   - checkDelay : If supported, respect Job::getReleaseTimestamp() in the push functions.
 102          *                  This lets delayed jobs wait in a staging area until a given timestamp is
 103          *                  reached, at which point they will enter the queue. If this is not enabled
 104          *                  or not supported, an exception will be thrown on delayed job insertion.
 105          *
 106          * Queue classes should throw an exception if they do not support the options given.
 107          *
 108          * @param array $params
 109          * @return JobQueue
 110          * @throws MWException
 111          */
 112         final public static function factory( array $params ) {
 113                 $class = $params['class'];
 114                 if ( !class_exists( $class ) ) {
 115                         throw new MWException( "Invalid job queue class '$class'." );
 116                 }
 117                 $obj = new $class( $params );
 118                 if ( !( $obj instanceof self ) ) {
 119                         throw new MWException( "Class '$class' is not a " . __CLASS__ . " class." );
 120                 }
 121
 122                 return $obj;
 123         }
 124
 125         /**
 126          * @return string Wiki ID
 127          */
 128         final public function getWiki() {
 129                 return $this->wiki;
 130         }
 131
 132         /**
 133          * @return string Job type that this queue handles
 134          */
 135         final public function getType() {
 136                 return $this->type;
 137         }
 138
 139         /**
 140          * @return string One of (random, timestamp, fifo, undefined)
 141          */
 142         final public function getOrder() {
 143                 return $this->order;
 144         }
 145
 146         /**
 147          * @return bool Whether delayed jobs are enabled
 148          * @since 1.22
 149          */
 150         final public function delayedJobsEnabled() {
 151                 return $this->checkDelay;
 152         }
 153
 154         /**
 155          * Get the allowed queue orders for configuration validation
 156          *
 157          * @return array Subset of (random, timestamp, fifo, undefined)
 158          */
 159         abstract protected function supportedOrders();
 160
 161         /**
 162          * Get the default queue order to use if configuration does not specify one
 163          *
 164          * @return string One of (random, timestamp, fifo, undefined)
 165          */
 166         abstract protected function optimalOrder();
 167
 168         /**
 169          * Find out if delayed jobs are supported for configuration validation
 170          *
 171          * @return bool Whether delayed jobs are supported
 172          */
 173         protected function supportsDelayedJobs() {
 174                 return false; // not implemented
 175         }
 176
 177         /**
 178          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 179          * Queue classes should use caching if they are any slower without memcached.
 180          *
 181          * If caching is used, this might return false when there are actually no jobs.
 182          * If pop() is called and returns false then it should correct the cache. Also,
 183          * calling flushCaches() first prevents this. However, this affect is typically
 184          * not distinguishable from the race condition between isEmpty() and pop().
 185          *
 186          * @return bool
 187          * @throws JobQueueError
 188          */
 189         final public function isEmpty() {
 190                 $res = $this->doIsEmpty();
 191
 192                 return $res;
 193         }
 194
 195         /**
 196          * @see JobQueue::isEmpty()
 197          * @return bool
 198          */
 199         abstract protected function doIsEmpty();
 200
 201         /**
 202          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 203          * Queue classes should use caching if they are any slower without memcached.
 204          *
 205          * If caching is used, this number might be out of date for a minute.
 206          *
 207          * @return int
 208          * @throws JobQueueError
 209          */
 210         final public function getSize() {
 211                 $res = $this->doGetSize();
 212
 213                 return $res;
 214         }
 215
 216         /**
 217          * @see JobQueue::getSize()
 218          * @return int
 219          */
 220         abstract protected function doGetSize();
 221
 222         /**
 223          * Get the number of acquired jobs (these are temporarily out of the queue).
 224          * Queue classes should use caching if they are any slower without memcached.
 225          *
 226          * If caching is used, this number might be out of date for a minute.
 227          *
 228          * @return int
 229          * @throws JobQueueError
 230          */
 231         final public function getAcquiredCount() {
 232                 $res = $this->doGetAcquiredCount();
 233
 234                 return $res;
 235         }
 236
 237         /**
 238          * @see JobQueue::getAcquiredCount()
 239          * @return int
 240          */
 241         abstract protected function doGetAcquiredCount();
 242
 243         /**
 244          * Get the number of delayed jobs (these are temporarily out of the queue).
 245          * Queue classes should use caching if they are any slower without memcached.
 246          *
 247          * If caching is used, this number might be out of date for a minute.
 248          *
 249          * @return int
 250          * @throws JobQueueError
 251          * @since 1.22
 252          */
 253         final public function getDelayedCount() {
 254                 $res = $this->doGetDelayedCount();
 255
 256                 return $res;
 257         }
 258
 259         /**
 260          * @see JobQueue::getDelayedCount()
 261          * @return int
 262          */
 263         protected function doGetDelayedCount() {
 264                 return 0; // not implemented
 265         }
 266
 267         /**
 268          * Get the number of acquired jobs that can no longer be attempted.
 269          * Queue classes should use caching if they are any slower without memcached.
 270          *
 271          * If caching is used, this number might be out of date for a minute.
 272          *
 273          * @return int
 274          * @throws JobQueueError
 275          */
 276         final public function getAbandonedCount() {
 277                 $res = $this->doGetAbandonedCount();
 278
 279                 return $res;
 280         }
 281
 282         /**
 283          * @see JobQueue::getAbandonedCount()
 284          * @return int
 285          */
 286         protected function doGetAbandonedCount() {
 287                 return 0; // not implemented
 288         }
 289
 290         /**
 291          * Push one or more jobs into the queue.
 292          * This does not require $wgJobClasses to be set for the given job type.
 293          * Outside callers should use JobQueueGroup::push() instead of this function.
 294          *
 295          * @param Job|array $jobs A single job or an array of Jobs
 296          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 297          * @return void
 298          * @throws JobQueueError
 299          */
 300         final public function push( $jobs, $flags = 0 ) {
 301                 $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 302         }
 303
 304         /**
 305          * Push a batch of jobs into the queue.
 306          * This does not require $wgJobClasses to be set for the given job type.
 307          * Outside callers should use JobQueueGroup::push() instead of this function.
 308          *
 309          * @param array $jobs List of Jobs
 310          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 311          * @return void
 312          * @throws MWException
 313          */
 314         final public function batchPush( array $jobs, $flags = 0 ) {
 315                 if ( !count( $jobs ) ) {
 316                         return; // nothing to do
 317                 }
 318
 319                 foreach ( $jobs as $job ) {
 320                         if ( $job->getType() !== $this->type ) {
 321                                 throw new MWException(
 322                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 323                         } elseif ( $job->getReleaseTimestamp() && !$this->checkDelay ) {
 324                                 throw new MWException(
 325                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 326                         }
 327                 }
 328
 329                 $this->doBatchPush( $jobs, $flags );
 330         }
 331
 332         /**
 333          * @see JobQueue::batchPush()
 334          * @param array $jobs
 335          * @param int $flags
 336          */
 337         abstract protected function doBatchPush( array $jobs, $flags );
 338
 339         /**
 340          * Pop a job off of the queue.
 341          * This requires $wgJobClasses to be set for the given job type.
 342          * Outside callers should use JobQueueGroup::pop() instead of this function.
 343          *
 344          * @throws MWException
 345          * @return Job|bool Returns false if there are no jobs
 346          */
 347         final public function pop() {
 348                 global $wgJobClasses;
 349
 350                 if ( $this->wiki !== wfWikiID() ) {
 351                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 352                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 353                         // Do not pop jobs if there is no class for the queue type
 354                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 355                 }
 356
 357                 $job = $this->doPop();
 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, 1, $this->wiki );
 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 Job $job
 429          * @throws MWException
 430          * @return bool
 431          */
 432         final public function deduplicateRootJob( Job $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 Job $job
 444          * @throws MWException
 445          * @return bool
 446          */
 447         protected function doDeduplicateRootJob( Job $job ) {
 448                 if ( !$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          * Return a map of task names to task definition maps.
 553          * A "task" is a fast periodic queue maintenance action.
 554          * Mutually exclusive tasks must implement their own locking in the callback.
 555          *
 556          * Each task value is an associative array with:
 557          *   - name     : the name of the task
 558          *   - callback : a PHP callable that performs the task
 559          *   - period   : the period in seconds corresponding to the task frequency
 560          *
 561          * @return array
 562          */
 563         final public function getPeriodicTasks() {
 564                 $tasks = $this->doGetPeriodicTasks();
 565                 foreach ( $tasks as $name => &$def ) {
 566                         $def['name'] = $name;
 567                 }
 568
 569                 return $tasks;
 570         }
 571
 572         /**
 573          * @see JobQueue::getPeriodicTasks()
 574          * @return array
 575          */
 576         protected function doGetPeriodicTasks() {
 577                 return array();
 578         }
 579
 580         /**
 581          * Clear any process and persistent caches
 582          *
 583          * @return void
 584          */
 585         final public function flushCaches() {
 586                 $this->doFlushCaches();
 587         }
 588
 589         /**
 590          * @see JobQueue::flushCaches()
 591          * @return void
 592          */
 593         protected function doFlushCaches() {
 594         }
 595
 596         /**
 597          * Get an iterator to traverse over all available jobs in this queue.
 598          * This does not include jobs that are currently acquired or delayed.
 599          * Note: results may be stale if the queue is concurrently modified.
 600          *
 601          * @return Iterator
 602          * @throws JobQueueError
 603          */
 604         abstract public function getAllQueuedJobs();
 605
 606         /**
 607          * Get an iterator to traverse over all delayed jobs in this queue.
 608          * Note: results may be stale if the queue is concurrently modified.
 609          *
 610          * @return Iterator
 611          * @throws JobQueueError
 612          * @since 1.22
 613          */
 614         public function getAllDelayedJobs() {
 615                 return new ArrayIterator( array() ); // not implemented
 616         }
 617
 618         /**
 619          * Do not use this function outside of JobQueue/JobQueueGroup
 620          *
 621          * @return string
 622          * @since 1.22
 623          */
 624         public function getCoalesceLocationInternal() {
 625                 return null;
 626         }
 627
 628         /**
 629          * Check whether each of the given queues are empty.
 630          * This is used for batching checks for queues stored at the same place.
 631          *
 632          * @param array $types List of queues types
 633          * @return array|null (list of non-empty queue types) or null if unsupported
 634          * @throws MWException
 635          * @since 1.22
 636          */
 637         final public function getSiblingQueuesWithJobs( array $types ) {
 638
 639                 return $this->doGetSiblingQueuesWithJobs( $types );
 640         }
 641
 642         /**
 643          * @see JobQueue::getSiblingQueuesWithJobs()
 644          * @param array $types List of queues types
 645          * @return array|null (list of queue types) or null if unsupported
 646          */
 647         protected function doGetSiblingQueuesWithJobs( array $types ) {
 648                 return null; // not supported
 649         }
 650
 651         /**
 652          * Check the size of each of the given queues.
 653          * For queues not served by the same store as this one, 0 is returned.
 654          * This is used for batching checks for queues stored at the same place.
 655          *
 656          * @param array $types List of queues types
 657          * @return array|null (job type => whether queue is empty) or null if unsupported
 658          * @throws MWException
 659          * @since 1.22
 660          */
 661         final public function getSiblingQueueSizes( array $types ) {
 662
 663                 return $this->doGetSiblingQueueSizes( $types );
 664         }
 665
 666         /**
 667          * @see JobQueue::getSiblingQueuesSize()
 668          * @param array $types List of queues types
 669          * @return array|null (list of queue types) or null if unsupported
 670          */
 671         protected function doGetSiblingQueueSizes( array $types ) {
 672                 return null; // not supported
 673         }
 674
 675         /**
 676          * Call wfIncrStats() for the queue overall and for the queue type
 677          *
 678          * @param string $key Event type
 679          * @param string $type Job type
 680          * @param int $delta
 681          * @param string $wiki Wiki ID (added in 1.23)
 682          * @since 1.22
 683          */
 684         public static function incrStats( $key, $type, $delta = 1, $wiki = null ) {
 685                 wfIncrStats( $key, $delta );
 686                 wfIncrStats( "{$key}-{$type}", $delta );
 687                 if ( $wiki !== null ) {
 688                         wfIncrStats( "{$key}-{$type}-{$wiki}", $delta );
 689                 }
 690         }
 691
 692         /**
 693          * Namespace the queue with a key to isolate it for testing
 694          *
 695          * @param string $key
 696          * @return void
 697          * @throws MWException
 698          */
 699         public function setTestingPrefix( $key ) {
 700                 throw new MWException( "Queue namespacing not supported for this queue type." );
 701         }
 702 }
 703
 704 /**
 705  * @ingroup JobQueue
 706  * @since 1.22
 707  */
 708 class JobQueueError extends MWException {
 709 }
 710
 711 class JobQueueConnectionError extends JobQueueError {
 712 }