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  */
  23 use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface;
  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 DB domain ID */
  33         protected $domain;
  34         /** @var string Job type */
  35         protected $type;
  36         /** @var string Job priority for pop() */
  37         protected $order;
  38         /** @var int Time to live in seconds */
  39         protected $claimTTL;
  40         /** @var int Maximum number of times to try a job */
  41         protected $maxTries;
  42         /** @var string|bool Read only rationale (or false if r/w) */
  43         protected $readOnlyReason;
  44         /** @var StatsdDataFactoryInterface */
  45         protected $stats;
  46
  47         /** @var BagOStuff */
  48         protected $dupCache;
  49
  50         const QOS_ATOMIC = 1; // integer; "all-or-nothing" job insertions
  51
  52         const ROOTJOB_TTL = 2419200; // integer; seconds to remember root jobs (28 days)
  53
  54         /**
  55          * @param array $params
  56          * @throws JobQueueError
  57          */
  58         protected function __construct( array $params ) {
  59                 $this->domain = $params['domain'] ?? $params['wiki']; // b/c
  60                 $this->type = $params['type'];
  61                 $this->claimTTL = $params['claimTTL'] ?? 0;
  62                 $this->maxTries = $params['maxTries'] ?? 3;
  63                 if ( isset( $params['order'] ) && $params['order'] !== 'any' ) {
  64                         $this->order = $params['order'];
  65                 } else {
  66                         $this->order = $this->optimalOrder();
  67                 }
  68                 if ( !in_array( $this->order, $this->supportedOrders() ) ) {
  69                         throw new JobQueueError( __CLASS__ . " does not support '{$this->order}' order." );
  70                 }
  71                 $this->readOnlyReason = $params['readOnlyReason'] ?? false;
  72                 $this->stats = $params['stats'] ?? new NullStatsdDataFactory();
  73                 $this->dupCache = $params['stash'] ?? new EmptyBagOStuff();
  74         }
  75
  76         /**
  77          * Get a job queue object of the specified type.
  78          * $params includes:
  79          *   - class      : What job class to use (determines job type)
  80          *   - domain     : Database domain ID of the wiki the jobs are for (defaults to current wiki)
  81          *   - type       : The name of the job types this queue handles
  82          *   - order      : Order that pop() selects jobs, one of "fifo", "timestamp" or "random".
  83          *                  If "fifo" is used, the queue will effectively be FIFO. Note that job
  84          *                  completion will not appear to be exactly FIFO if there are multiple
  85          *                  job runners since jobs can take different times to finish once popped.
  86          *                  If "timestamp" is used, the queue will at least be loosely ordered
  87          *                  by timestamp, allowing for some jobs to be popped off out of order.
  88          *                  If "random" is used, pop() will pick jobs in random order.
  89          *                  Note that it may only be weakly random (e.g. a lottery of the oldest X).
  90          *                  If "any" is choosen, the queue will use whatever order is the fastest.
  91          *                  This might be useful for improving concurrency for job acquisition.
  92          *   - claimTTL   : If supported, the queue will recycle jobs that have been popped
  93          *                  but not acknowledged as completed after this many seconds. Recycling
  94          *                  of jobs simply means re-inserting them into the queue. Jobs can be
  95          *                  attempted up to three times before being discarded.
  96          *   - readOnlyReason : Set this to a string to make the queue read-only.
  97          *   - stash      : A BagOStuff instance that can be used for root job deduplication
  98          *   - stats      : A StatsdDataFactoryInterface [optional]
  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 JobQueueError
 105          */
 106         final public static function factory( array $params ) {
 107                 $class = $params['class'];
 108                 if ( !class_exists( $class ) ) {
 109                         throw new JobQueueError( "Invalid job queue class '$class'." );
 110                 }
 111                 $obj = new $class( $params );
 112                 if ( !( $obj instanceof self ) ) {
 113                         throw new JobQueueError( "Class '$class' is not a " . __CLASS__ . " class." );
 114                 }
 115
 116                 return $obj;
 117         }
 118
 119         /**
 120          * @return string Database domain ID
 121          */
 122         final public function getDomain() {
 123                 return $this->domain;
 124         }
 125
 126         /**
 127          * @return string Wiki ID
 128          * @deprecated 1.33
 129          */
 130         final public function getWiki() {
 131                 return WikiMap::getWikiIdFromDbDomain( $this->domain );
 132         }
 133
 134         /**
 135          * @return string Job type that this queue handles
 136          */
 137         final public function getType() {
 138                 return $this->type;
 139         }
 140
 141         /**
 142          * @return string One of (random, timestamp, fifo, undefined)
 143          */
 144         final public function getOrder() {
 145                 return $this->order;
 146         }
 147
 148         /**
 149          * Get the allowed queue orders for configuration validation
 150          *
 151          * @return array Subset of (random, timestamp, fifo, undefined)
 152          */
 153         abstract protected function supportedOrders();
 154
 155         /**
 156          * Get the default queue order to use if configuration does not specify one
 157          *
 158          * @return string One of (random, timestamp, fifo, undefined)
 159          */
 160         abstract protected function optimalOrder();
 161
 162         /**
 163          * Find out if delayed jobs are supported for configuration validation
 164          *
 165          * @return bool Whether delayed jobs are supported
 166          */
 167         protected function supportsDelayedJobs() {
 168                 return false; // not implemented
 169         }
 170
 171         /**
 172          * @return bool Whether delayed jobs are enabled
 173          * @since 1.22
 174          */
 175         final public function delayedJobsEnabled() {
 176                 return $this->supportsDelayedJobs();
 177         }
 178
 179         /**
 180          * @return string|bool Read-only rational or false if r/w
 181          * @since 1.27
 182          */
 183         public function getReadOnlyReason() {
 184                 return $this->readOnlyReason;
 185         }
 186
 187         /**
 188          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 189          * Queue classes should use caching if they are any slower without memcached.
 190          *
 191          * If caching is used, this might return false when there are actually no jobs.
 192          * If pop() is called and returns false then it should correct the cache. Also,
 193          * calling flushCaches() first prevents this. However, this affect is typically
 194          * not distinguishable from the race condition between isEmpty() and pop().
 195          *
 196          * @return bool
 197          * @throws JobQueueError
 198          */
 199         final public function isEmpty() {
 200                 $res = $this->doIsEmpty();
 201
 202                 return $res;
 203         }
 204
 205         /**
 206          * @see JobQueue::isEmpty()
 207          * @return bool
 208          */
 209         abstract protected function doIsEmpty();
 210
 211         /**
 212          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 213          * Queue classes should use caching if they are any slower without memcached.
 214          *
 215          * If caching is used, this number might be out of date for a minute.
 216          *
 217          * @return int
 218          * @throws JobQueueError
 219          */
 220         final public function getSize() {
 221                 $res = $this->doGetSize();
 222
 223                 return $res;
 224         }
 225
 226         /**
 227          * @see JobQueue::getSize()
 228          * @return int
 229          */
 230         abstract protected function doGetSize();
 231
 232         /**
 233          * Get the number of acquired jobs (these are temporarily out of the queue).
 234          * Queue classes should use caching if they are any slower without memcached.
 235          *
 236          * If caching is used, this number might be out of date for a minute.
 237          *
 238          * @return int
 239          * @throws JobQueueError
 240          */
 241         final public function getAcquiredCount() {
 242                 $res = $this->doGetAcquiredCount();
 243
 244                 return $res;
 245         }
 246
 247         /**
 248          * @see JobQueue::getAcquiredCount()
 249          * @return int
 250          */
 251         abstract protected function doGetAcquiredCount();
 252
 253         /**
 254          * Get the number of delayed jobs (these are temporarily out of the queue).
 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          * @since 1.22
 262          */
 263         final public function getDelayedCount() {
 264                 $res = $this->doGetDelayedCount();
 265
 266                 return $res;
 267         }
 268
 269         /**
 270          * @see JobQueue::getDelayedCount()
 271          * @return int
 272          */
 273         protected function doGetDelayedCount() {
 274                 return 0; // not implemented
 275         }
 276
 277         /**
 278          * Get the number of acquired jobs that can no longer be attempted.
 279          * Queue classes should use caching if they are any slower without memcached.
 280          *
 281          * If caching is used, this number might be out of date for a minute.
 282          *
 283          * @return int
 284          * @throws JobQueueError
 285          */
 286         final public function getAbandonedCount() {
 287                 $res = $this->doGetAbandonedCount();
 288
 289                 return $res;
 290         }
 291
 292         /**
 293          * @see JobQueue::getAbandonedCount()
 294          * @return int
 295          */
 296         protected function doGetAbandonedCount() {
 297                 return 0; // not implemented
 298         }
 299
 300         /**
 301          * Push one or more jobs into the queue.
 302          * This does not require $wgJobClasses to be set for the given job type.
 303          * Outside callers should use JobQueueGroup::push() instead of this function.
 304          *
 305          * @param IJobSpecification|IJobSpecification[] $jobs
 306          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 307          * @return void
 308          * @throws JobQueueError
 309          */
 310         final public function push( $jobs, $flags = 0 ) {
 311                 $jobs = is_array( $jobs ) ? $jobs : [ $jobs ];
 312                 $this->batchPush( $jobs, $flags );
 313         }
 314
 315         /**
 316          * Push a batch of jobs into the queue.
 317          * This does not require $wgJobClasses to be set for the given job type.
 318          * Outside callers should use JobQueueGroup::push() instead of this function.
 319          *
 320          * @param IJobSpecification[] $jobs
 321          * @param int $flags Bitfield (supports JobQueue::QOS_ATOMIC)
 322          * @return void
 323          * @throws JobQueueError
 324          */
 325         final public function batchPush( array $jobs, $flags = 0 ) {
 326                 $this->assertNotReadOnly();
 327
 328                 if ( $jobs === [] ) {
 329                         return; // nothing to do
 330                 }
 331
 332                 foreach ( $jobs as $job ) {
 333                         if ( $job->getType() !== $this->type ) {
 334                                 throw new JobQueueError(
 335                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 336                         } elseif ( $job->getReleaseTimestamp() && !$this->supportsDelayedJobs() ) {
 337                                 throw new JobQueueError(
 338                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 339                         }
 340                 }
 341
 342                 $this->doBatchPush( $jobs, $flags );
 343
 344                 foreach ( $jobs as $job ) {
 345                         if ( $job->isRootJob() ) {
 346                                 $this->deduplicateRootJob( $job );
 347                         }
 348                 }
 349         }
 350
 351         /**
 352          * @see JobQueue::batchPush()
 353          * @param IJobSpecification[] $jobs
 354          * @param int $flags
 355          */
 356         abstract protected function doBatchPush( array $jobs, $flags );
 357
 358         /**
 359          * Pop a job off of the queue.
 360          * This requires $wgJobClasses to be set for the given job type.
 361          * Outside callers should use JobQueueGroup::pop() instead of this function.
 362          *
 363          * @throws JobQueueError
 364          * @return Job|bool Returns false if there are no jobs
 365          */
 366         final public function pop() {
 367                 $this->assertNotReadOnly();
 368
 369                 $job = $this->doPop();
 370
 371                 // Flag this job as an old duplicate based on its "root" job...
 372                 try {
 373                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 374                                 $this->incrStats( 'dupe_pops', $this->type );
 375                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 376                         }
 377                 } catch ( Exception $e ) {
 378                         // don't lose jobs over this
 379                 }
 380
 381                 return $job;
 382         }
 383
 384         /**
 385          * @see JobQueue::pop()
 386          * @return Job|bool
 387          */
 388         abstract protected function doPop();
 389
 390         /**
 391          * Acknowledge that a job was completed.
 392          *
 393          * This does nothing for certain queue classes or if "claimTTL" is not set.
 394          * Outside callers should use JobQueueGroup::ack() instead of this function.
 395          *
 396          * @param Job $job
 397          * @return void
 398          * @throws JobQueueError
 399          */
 400         final public function ack( Job $job ) {
 401                 $this->assertNotReadOnly();
 402                 if ( $job->getType() !== $this->type ) {
 403                         throw new JobQueueError( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 404                 }
 405
 406                 $this->doAck( $job );
 407         }
 408
 409         /**
 410          * @see JobQueue::ack()
 411          * @param Job $job
 412          */
 413         abstract protected function doAck( Job $job );
 414
 415         /**
 416          * Register the "root job" of a given job into the queue for de-duplication.
 417          * This should only be called right *after* all the new jobs have been inserted.
 418          * This is used to turn older, duplicate, job entries into no-ops. The root job
 419          * information will remain in the registry until it simply falls out of cache.
 420          *
 421          * This requires that $job has two special fields in the "params" array:
 422          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 423          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 424          *
 425          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 426          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 427          * spawned when a template is edited. One can think of the task as "update links
 428          * of pages that use template X" and an instance of that task as a "root job".
 429          * However, what actually goes into the queue are range and leaf job subtypes.
 430          * Since these jobs include things like page ID ranges and DB master positions,
 431          * and can morph into smaller jobs recursively, simple duplicate detection
 432          * for individual jobs being identical (like that of job_sha1) is not useful.
 433          *
 434          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 435          * is edited again, we want all of these old refreshLinks jobs for that template to become
 436          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 437          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 438          * previous "root job" for the same task of "update links of pages that use template X".
 439          *
 440          * This does nothing for certain queue classes.
 441          *
 442          * @param IJobSpecification $job
 443          * @throws JobQueueError
 444          * @return bool
 445          */
 446         final public function deduplicateRootJob( IJobSpecification $job ) {
 447                 $this->assertNotReadOnly();
 448                 if ( $job->getType() !== $this->type ) {
 449                         throw new JobQueueError( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 450                 }
 451
 452                 return $this->doDeduplicateRootJob( $job );
 453         }
 454
 455         /**
 456          * @see JobQueue::deduplicateRootJob()
 457          * @param IJobSpecification $job
 458          * @throws JobQueueError
 459          * @return bool
 460          */
 461         protected function doDeduplicateRootJob( IJobSpecification $job ) {
 462                 if ( !$job->hasRootJobParams() ) {
 463                         throw new JobQueueError( "Cannot register root job; missing parameters." );
 464                 }
 465                 $params = $job->getRootJobParams();
 466
 467                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 468                 // Callers should call JobQueueGroup::push() before this method so that if the insert
 469                 // fails, the de-duplication registration will be aborted. Since the insert is
 470                 // deferred till "transaction idle", do the same here, so that the ordering is
 471                 // maintained. Having only the de-duplication registration succeed would cause
 472                 // jobs to become no-ops without any actual jobs that made them redundant.
 473                 $timestamp = $this->dupCache->get( $key ); // current last timestamp of this job
 474                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 475                         return true; // a newer version of this root job was enqueued
 476                 }
 477
 478                 // Update the timestamp of the last root job started at the location...
 479                 return $this->dupCache->set( $key, $params['rootJobTimestamp'], self::ROOTJOB_TTL );
 480         }
 481
 482         /**
 483          * Check if the "root" job of a given job has been superseded by a newer one
 484          *
 485          * @param Job $job
 486          * @throws JobQueueError
 487          * @return bool
 488          */
 489         final protected function isRootJobOldDuplicate( Job $job ) {
 490                 if ( $job->getType() !== $this->type ) {
 491                         throw new JobQueueError( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 492                 }
 493                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 494
 495                 return $isDuplicate;
 496         }
 497
 498         /**
 499          * @see JobQueue::isRootJobOldDuplicate()
 500          * @param Job $job
 501          * @return bool
 502          */
 503         protected function doIsRootJobOldDuplicate( Job $job ) {
 504                 if ( !$job->hasRootJobParams() ) {
 505                         return false; // job has no de-deplication info
 506                 }
 507                 $params = $job->getRootJobParams();
 508
 509                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 510                 // Get the last time this root job was enqueued
 511                 $timestamp = $this->dupCache->get( $key );
 512
 513                 // Check if a new root job was started at the location after this one's...
 514                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 515         }
 516
 517         /**
 518          * @param string $signature Hash identifier of the root job
 519          * @return string
 520          */
 521         protected function getRootJobCacheKey( $signature ) {
 522                 return $this->dupCache->makeGlobalKey(
 523                         'jobqueue',
 524                         $this->domain,
 525                         $this->type,
 526                         'rootjob',
 527                         $signature
 528                 );
 529         }
 530
 531         /**
 532          * Deleted all unclaimed and delayed jobs from the queue
 533          *
 534          * @throws JobQueueError
 535          * @since 1.22
 536          * @return void
 537          */
 538         final public function delete() {
 539                 $this->assertNotReadOnly();
 540
 541                 $this->doDelete();
 542         }
 543
 544         /**
 545          * @see JobQueue::delete()
 546          * @throws JobQueueError
 547          */
 548         protected function doDelete() {
 549                 throw new JobQueueError( "This method is not implemented." );
 550         }
 551
 552         /**
 553          * Wait for any replica DBs or backup servers to catch up.
 554          *
 555          * This does nothing for certain queue classes.
 556          *
 557          * @return void
 558          * @throws JobQueueError
 559          */
 560         final public function waitForBackups() {
 561                 $this->doWaitForBackups();
 562         }
 563
 564         /**
 565          * @see JobQueue::waitForBackups()
 566          * @return void
 567          */
 568         protected function doWaitForBackups() {
 569         }
 570
 571         /**
 572          * Clear any process and persistent caches
 573          *
 574          * @return void
 575          */
 576         final public function flushCaches() {
 577                 $this->doFlushCaches();
 578         }
 579
 580         /**
 581          * @see JobQueue::flushCaches()
 582          * @return void
 583          */
 584         protected function doFlushCaches() {
 585         }
 586
 587         /**
 588          * Get an iterator to traverse over all available jobs in this queue.
 589          * This does not include jobs that are currently acquired or delayed.
 590          * Note: results may be stale if the queue is concurrently modified.
 591          *
 592          * @return Iterator
 593          * @throws JobQueueError
 594          */
 595         abstract public function getAllQueuedJobs();
 596
 597         /**
 598          * Get an iterator to traverse over all delayed jobs in this queue.
 599          * Note: results may be stale if the queue is concurrently modified.
 600          *
 601          * @return Iterator
 602          * @throws JobQueueError
 603          * @since 1.22
 604          */
 605         public function getAllDelayedJobs() {
 606                 return new ArrayIterator( [] ); // not implemented
 607         }
 608
 609         /**
 610          * Get an iterator to traverse over all claimed jobs in this queue
 611          *
 612          * Callers should be quick to iterator over it or few results
 613          * will be returned due to jobs being acknowledged and deleted
 614          *
 615          * @return Iterator
 616          * @throws JobQueueError
 617          * @since 1.26
 618          */
 619         public function getAllAcquiredJobs() {
 620                 return new ArrayIterator( [] ); // not implemented
 621         }
 622
 623         /**
 624          * Get an iterator to traverse over all abandoned jobs in this queue
 625          *
 626          * @return Iterator
 627          * @throws JobQueueError
 628          * @since 1.25
 629          */
 630         public function getAllAbandonedJobs() {
 631                 return new ArrayIterator( [] ); // not implemented
 632         }
 633
 634         /**
 635          * Do not use this function outside of JobQueue/JobQueueGroup
 636          *
 637          * @return string
 638          * @since 1.22
 639          */
 640         public function getCoalesceLocationInternal() {
 641                 return null;
 642         }
 643
 644         /**
 645          * Check whether each of the given queues are empty.
 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 (list of non-empty queue types) or null if unsupported
 650          * @throws JobQueueError
 651          * @since 1.22
 652          */
 653         final public function getSiblingQueuesWithJobs( array $types ) {
 654                 return $this->doGetSiblingQueuesWithJobs( $types );
 655         }
 656
 657         /**
 658          * @see JobQueue::getSiblingQueuesWithJobs()
 659          * @param array $types List of queues types
 660          * @return array|null (list of queue types) or null if unsupported
 661          */
 662         protected function doGetSiblingQueuesWithJobs( array $types ) {
 663                 return null; // not supported
 664         }
 665
 666         /**
 667          * Check the size of each of the given queues.
 668          * For queues not served by the same store as this one, 0 is returned.
 669          * This is used for batching checks for queues stored at the same place.
 670          *
 671          * @param array $types List of queues types
 672          * @return array|null (job type => whether queue is empty) or null if unsupported
 673          * @throws JobQueueError
 674          * @since 1.22
 675          */
 676         final public function getSiblingQueueSizes( array $types ) {
 677                 return $this->doGetSiblingQueueSizes( $types );
 678         }
 679
 680         /**
 681          * @see JobQueue::getSiblingQueuesSize()
 682          * @param array $types List of queues types
 683          * @return array|null (list of queue types) or null if unsupported
 684          */
 685         protected function doGetSiblingQueueSizes( array $types ) {
 686                 return null; // not supported
 687         }
 688
 689         /**
 690          * @throws JobQueueReadOnlyError
 691          */
 692         protected function assertNotReadOnly() {
 693                 if ( $this->readOnlyReason !== false ) {
 694                         throw new JobQueueReadOnlyError( "Job queue is read-only: {$this->readOnlyReason}" );
 695                 }
 696         }
 697
 698         /**
 699          * Call wfIncrStats() for the queue overall and for the queue type
 700          *
 701          * @param string $key Event type
 702          * @param string $type Job type
 703          * @param int $delta
 704          * @since 1.22
 705          */
 706         protected function incrStats( $key, $type, $delta = 1 ) {
 707                 $this->stats->updateCount( "jobqueue.{$key}.all", $delta );
 708                 $this->stats->updateCount( "jobqueue.{$key}.{$type}", $delta );
 709         }
 710 }