includes/job/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         protected $wiki; // string; wiki ID
  33         protected $type; // string; job type
  34         protected $order; // string; job priority for pop()
  35         protected $claimTTL; // integer; seconds
  36         protected $maxTries; // integer; maximum number of times to try a job
  37         protected $checkDelay; // boolean; allow delayed jobs
  38
  39         const QOS_ATOMIC = 1; // integer; "all-or-nothing" job insertions
  40         const QoS_Atomic = 1; // integer; "all-or-nothing" job insertions (b/c)
  41
  42         const ROOTJOB_TTL = 2419200; // integer; seconds to remember root jobs (28 days)
  43
  44         /**
  45          * @param $params array
  46          */
  47         protected function __construct( array $params ) {
  48                 $this->wiki = $params['wiki'];
  49                 $this->type = $params['type'];
  50                 $this->claimTTL = isset( $params['claimTTL'] ) ? $params['claimTTL'] : 0;
  51                 $this->maxTries = isset( $params['maxTries'] ) ? $params['maxTries'] : 3;
  52                 if ( isset( $params['order'] ) && $params['order'] !== 'any' ) {
  53                         $this->order = $params['order'];
  54                 } else {
  55                         $this->order = $this->optimalOrder();
  56                 }
  57                 if ( !in_array( $this->order, $this->supportedOrders() ) ) {
  58                         throw new MWException( __CLASS__ . " does not support '{$this->order}' order." );
  59                 }
  60                 $this->checkDelay = !empty( $params['checkDelay'] );
  61                 if ( $this->checkDelay && !$this->supportsDelayedJobs() ) {
  62                         throw new MWException( __CLASS__ . " does not support delayed jobs." );
  63                 }
  64         }
  65
  66         /**
  67          * Get a job queue object of the specified type.
  68          * $params includes:
  69          *   - class      : What job class to use (determines job type)
  70          *   - wiki       : wiki ID of the wiki the jobs are for (defaults to current wiki)
  71          *   - type       : The name of the job types this queue handles
  72          *   - order      : Order that pop() selects jobs, one of "fifo", "timestamp" or "random".
  73          *                  If "fifo" is used, the queue will effectively be FIFO. Note that job
  74          *                  completion will not appear to be exactly FIFO if there are multiple
  75          *                  job runners since jobs can take different times to finish once popped.
  76          *                  If "timestamp" is used, the queue will at least be loosely ordered
  77          *                  by timestamp, allowing for some jobs to be popped off out of order.
  78          *                  If "random" is used, pop() will pick jobs in random order.
  79          *                  Note that it may only be weakly random (e.g. a lottery of the oldest X).
  80          *                  If "any" is choosen, the queue will use whatever order is the fastest.
  81          *                  This might be useful for improving concurrency for job acquisition.
  82          *   - claimTTL   : If supported, the queue will recycle jobs that have been popped
  83          *                  but not acknowledged as completed after this many seconds. Recycling
  84          *                  of jobs simple means re-inserting them into the queue. Jobs can be
  85          *                  attempted up to three times before being discarded.
  86          *   - checkDelay : If supported, respect Job::getReleaseTimestamp() in the push functions.
  87          *                  This lets delayed jobs wait in a staging area until a given timestamp is
  88          *                  reached, at which point they will enter the queue. If this is not enabled
  89          *                  or not supported, an exception will be thrown on delayed job insertion.
  90          *
  91          * Queue classes should throw an exception if they do not support the options given.
  92          *
  93          * @param $params array
  94          * @return JobQueue
  95          * @throws MWException
  96          */
  97         final public static function factory( array $params ) {
  98                 $class = $params['class'];
  99                 if ( !class_exists( $class ) ) {
 100                         throw new MWException( "Invalid job queue class '$class'." );
 101                 }
 102                 $obj = new $class( $params );
 103                 if ( !( $obj instanceof self ) ) {
 104                         throw new MWException( "Class '$class' is not a " . __CLASS__ . " class." );
 105                 }
 106                 return $obj;
 107         }
 108
 109         /**
 110          * @return string Wiki ID
 111          */
 112         final public function getWiki() {
 113                 return $this->wiki;
 114         }
 115
 116         /**
 117          * @return string Job type that this queue handles
 118          */
 119         final public function getType() {
 120                 return $this->type;
 121         }
 122
 123         /**
 124          * @return string One of (random, timestamp, fifo, undefined)
 125          */
 126         final public function getOrder() {
 127                 return $this->order;
 128         }
 129
 130         /**
 131          * @return bool Whether delayed jobs are enabled
 132          * @since 1.22
 133          */
 134         final public function delayedJobsEnabled() {
 135                 return $this->checkDelay;
 136         }
 137
 138         /**
 139          * Get the allowed queue orders for configuration validation
 140          *
 141          * @return Array Subset of (random, timestamp, fifo, undefined)
 142          */
 143         abstract protected function supportedOrders();
 144
 145         /**
 146          * Get the default queue order to use if configuration does not specify one
 147          *
 148          * @return string One of (random, timestamp, fifo, undefined)
 149          */
 150         abstract protected function optimalOrder();
 151
 152         /**
 153          * Find out if delayed jobs are supported for configuration validation
 154          *
 155          * @return boolean Whether delayed jobs are supported
 156          */
 157         protected function supportsDelayedJobs() {
 158                 return false; // not implemented
 159         }
 160
 161         /**
 162          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 163          * Queue classes should use caching if they are any slower without memcached.
 164          *
 165          * If caching is used, this might return false when there are actually no jobs.
 166          * If pop() is called and returns false then it should correct the cache. Also,
 167          * calling flushCaches() first prevents this. However, this affect is typically
 168          * not distinguishable from the race condition between isEmpty() and pop().
 169          *
 170          * @return bool
 171          * @throws JobQueueError
 172          */
 173         final public function isEmpty() {
 174                 wfProfileIn( __METHOD__ );
 175                 $res = $this->doIsEmpty();
 176                 wfProfileOut( __METHOD__ );
 177                 return $res;
 178         }
 179
 180         /**
 181          * @see JobQueue::isEmpty()
 182          * @return bool
 183          */
 184         abstract protected function doIsEmpty();
 185
 186         /**
 187          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 188          * Queue classes should use caching if they are any slower without memcached.
 189          *
 190          * If caching is used, this number might be out of date for a minute.
 191          *
 192          * @return integer
 193          * @throws JobQueueError
 194          */
 195         final public function getSize() {
 196                 wfProfileIn( __METHOD__ );
 197                 $res = $this->doGetSize();
 198                 wfProfileOut( __METHOD__ );
 199                 return $res;
 200         }
 201
 202         /**
 203          * @see JobQueue::getSize()
 204          * @return integer
 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 integer
 215          * @throws JobQueueError
 216          */
 217         final public function getAcquiredCount() {
 218                 wfProfileIn( __METHOD__ );
 219                 $res = $this->doGetAcquiredCount();
 220                 wfProfileOut( __METHOD__ );
 221                 return $res;
 222         }
 223
 224         /**
 225          * @see JobQueue::getAcquiredCount()
 226          * @return integer
 227          */
 228         abstract protected function doGetAcquiredCount();
 229
 230         /**
 231          * Get the number of delayed jobs (these are temporarily out of the queue).
 232          * Queue classes should use caching if they are any slower without memcached.
 233          *
 234          * If caching is used, this number might be out of date for a minute.
 235          *
 236          * @return integer
 237          * @throws JobQueueError
 238          * @since 1.22
 239          */
 240         final public function getDelayedCount() {
 241                 wfProfileIn( __METHOD__ );
 242                 $res = $this->doGetDelayedCount();
 243                 wfProfileOut( __METHOD__ );
 244                 return $res;
 245         }
 246
 247         /**
 248          * @see JobQueue::getDelayedCount()
 249          * @return integer
 250          */
 251         protected function doGetDelayedCount() {
 252                 return 0; // not implemented
 253         }
 254
 255         /**
 256          * Get the number of acquired jobs that can no longer be attempted.
 257          * Queue classes should use caching if they are any slower without memcached.
 258          *
 259          * If caching is used, this number might be out of date for a minute.
 260          *
 261          * @return integer
 262          * @throws JobQueueError
 263          */
 264         final public function getAbandonedCount() {
 265                 wfProfileIn( __METHOD__ );
 266                 $res = $this->doGetAbandonedCount();
 267                 wfProfileOut( __METHOD__ );
 268                 return $res;
 269         }
 270
 271         /**
 272          * @see JobQueue::getAbandonedCount()
 273          * @return integer
 274          */
 275         protected function doGetAbandonedCount() {
 276                 return 0; // not implemented
 277         }
 278
 279         /**
 280          * Push a single jobs into the queue.
 281          * This does not require $wgJobClasses to be set for the given job type.
 282          * Outside callers should use JobQueueGroup::push() instead of this function.
 283          *
 284          * @param $jobs Job|Array
 285          * @param $flags integer Bitfield (supports JobQueue::QOS_ATOMIC)
 286          * @return bool Returns false on failure
 287          * @throws JobQueueError
 288          */
 289         final public function push( $jobs, $flags = 0 ) {
 290                 return $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 291         }
 292
 293         /**
 294          * Push a batch of jobs into the queue.
 295          * This does not require $wgJobClasses to be set for the given job type.
 296          * Outside callers should use JobQueueGroup::push() instead of this function.
 297          *
 298          * @param array $jobs List of Jobs
 299          * @param $flags integer Bitfield (supports JobQueue::QOS_ATOMIC)
 300          * @return bool Returns false on failure
 301          * @throws JobQueueError
 302          */
 303         final public function batchPush( array $jobs, $flags = 0 ) {
 304                 if ( !count( $jobs ) ) {
 305                         return true; // nothing to do
 306                 }
 307
 308                 foreach ( $jobs as $job ) {
 309                         if ( $job->getType() !== $this->type ) {
 310                                 throw new MWException(
 311                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 312                         } elseif ( $job->getReleaseTimestamp() && !$this->checkDelay ) {
 313                                 throw new MWException(
 314                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 315                         }
 316                 }
 317
 318                 wfProfileIn( __METHOD__ );
 319                 $ok = $this->doBatchPush( $jobs, $flags );
 320                 wfProfileOut( __METHOD__ );
 321                 return $ok;
 322         }
 323
 324         /**
 325          * @see JobQueue::batchPush()
 326          * @return bool
 327          */
 328         abstract protected function doBatchPush( array $jobs, $flags );
 329
 330         /**
 331          * Pop a job off of the queue.
 332          * This requires $wgJobClasses to be set for the given job type.
 333          * Outside callers should use JobQueueGroup::pop() instead of this function.
 334          *
 335          * @return Job|bool Returns false if there are no jobs
 336          * @throws JobQueueError
 337          */
 338         final public function pop() {
 339                 global $wgJobClasses;
 340
 341                 if ( $this->wiki !== wfWikiID() ) {
 342                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 343                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 344                         // Do not pop jobs if there is no class for the queue type
 345                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 346                 }
 347
 348                 wfProfileIn( __METHOD__ );
 349                 $job = $this->doPop();
 350                 wfProfileOut( __METHOD__ );
 351
 352                 // Flag this job as an old duplicate based on its "root" job...
 353                 try {
 354                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 355                                 JobQueue::incrStats( 'job-pop-duplicate', $this->type );
 356                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 357                         }
 358                 } catch ( MWException $e ) {} // don't lose jobs over this
 359
 360                 return $job;
 361         }
 362
 363         /**
 364          * @see JobQueue::pop()
 365          * @return Job
 366          */
 367         abstract protected function doPop();
 368
 369         /**
 370          * Acknowledge that a job was completed.
 371          *
 372          * This does nothing for certain queue classes or if "claimTTL" is not set.
 373          * Outside callers should use JobQueueGroup::ack() instead of this function.
 374          *
 375          * @param $job Job
 376          * @return bool
 377          * @throws JobQueueError
 378          */
 379         final public function ack( Job $job ) {
 380                 if ( $job->getType() !== $this->type ) {
 381                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 382                 }
 383                 wfProfileIn( __METHOD__ );
 384                 $ok = $this->doAck( $job );
 385                 wfProfileOut( __METHOD__ );
 386                 return $ok;
 387         }
 388
 389         /**
 390          * @see JobQueue::ack()
 391          * @return bool
 392          */
 393         abstract protected function doAck( Job $job );
 394
 395         /**
 396          * Register the "root job" of a given job into the queue for de-duplication.
 397          * This should only be called right *after* all the new jobs have been inserted.
 398          * This is used to turn older, duplicate, job entries into no-ops. The root job
 399          * information will remain in the registry until it simply falls out of cache.
 400          *
 401          * This requires that $job has two special fields in the "params" array:
 402          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 403          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 404          *
 405          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 406          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 407          * spawned when a template is edited. One can think of the task as "update links
 408          * of pages that use template X" and an instance of that task as a "root job".
 409          * However, what actually goes into the queue are potentially many refreshLinks2 jobs.
 410          * Since these jobs include things like page ID ranges and DB master positions, and morph
 411          * into smaller refreshLinks2 jobs recursively, simple duplicate detection (like job_sha1)
 412          * for individual jobs being identical is not useful.
 413          *
 414          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 415          * is edited again, we want all of these old refreshLinks jobs for that template to become
 416          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 417          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 418          * previous "root job" for the same task of "update links of pages that use template X".
 419          *
 420          * This does nothing for certain queue classes.
 421          *
 422          * @param $job Job
 423          * @return bool
 424          * @throws JobQueueError
 425          */
 426         final public function deduplicateRootJob( Job $job ) {
 427                 if ( $job->getType() !== $this->type ) {
 428                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 429                 }
 430                 wfProfileIn( __METHOD__ );
 431                 $ok = $this->doDeduplicateRootJob( $job );
 432                 wfProfileOut( __METHOD__ );
 433                 return $ok;
 434         }
 435
 436         /**
 437          * @see JobQueue::deduplicateRootJob()
 438          * @param $job Job
 439          * @return bool
 440          */
 441         protected function doDeduplicateRootJob( Job $job ) {
 442                 global $wgMemc;
 443
 444                 if ( !$job->hasRootJobParams() ) {
 445                         throw new MWException( "Cannot register root job; missing parameters." );
 446                 }
 447                 $params = $job->getRootJobParams();
 448
 449                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 450                 // Callers should call batchInsert() and then this function so that if the insert
 451                 // fails, the de-duplication registration will be aborted. Since the insert is
 452                 // deferred till "transaction idle", do the same here, so that the ordering is
 453                 // maintained. Having only the de-duplication registration succeed would cause
 454                 // jobs to become no-ops without any actual jobs that made them redundant.
 455                 $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 456                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 457                         return true; // a newer version of this root job was enqueued
 458                 }
 459
 460                 // Update the timestamp of the last root job started at the location...
 461                 return $wgMemc->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 462         }
 463
 464         /**
 465          * Check if the "root" job of a given job has been superseded by a newer one
 466          *
 467          * @param $job Job
 468          * @return bool
 469          * @throws JobQueueError
 470          */
 471         final protected function isRootJobOldDuplicate( Job $job ) {
 472                 if ( $job->getType() !== $this->type ) {
 473                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 474                 }
 475                 wfProfileIn( __METHOD__ );
 476                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 477                 wfProfileOut( __METHOD__ );
 478                 return $isDuplicate;
 479         }
 480
 481         /**
 482          * @see JobQueue::isRootJobOldDuplicate()
 483          * @param Job $job
 484          * @return bool
 485          */
 486         protected function doIsRootJobOldDuplicate( Job $job ) {
 487                 global $wgMemc;
 488
 489                 if ( !$job->hasRootJobParams() ) {
 490                         return false; // job has no de-deplication info
 491                 }
 492                 $params = $job->getRootJobParams();
 493
 494                 // Get the last time this root job was enqueued
 495                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 496
 497                 // Check if a new root job was started at the location after this one's...
 498                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 499         }
 500
 501         /**
 502          * @param string $signature Hash identifier of the root job
 503          * @return string
 504          */
 505         protected function getRootJobCacheKey( $signature ) {
 506                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 507                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 508         }
 509
 510         /**
 511          * Deleted all unclaimed and delayed jobs from the queue
 512          *
 513          * @return bool Success
 514          * @throws JobQueueError
 515          * @since 1.22
 516          */
 517         final public function delete() {
 518                 wfProfileIn( __METHOD__ );
 519                 $res = $this->doDelete();
 520                 wfProfileOut( __METHOD__ );
 521                 return $res;
 522         }
 523
 524         /**
 525          * @see JobQueue::delete()
 526          * @return bool Success
 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                 wfProfileIn( __METHOD__ );
 542                 $this->doWaitForBackups();
 543                 wfProfileOut( __METHOD__ );
 544         }
 545
 546         /**
 547          * @see JobQueue::waitForBackups()
 548          * @return void
 549          */
 550         protected function doWaitForBackups() {}
 551
 552         /**
 553          * Return a map of task names to task definition maps.
 554          * A "task" is a fast periodic queue maintenance action.
 555          * Mutually exclusive tasks must implement their own locking in the callback.
 556          *
 557          * Each task value is an associative array with:
 558          *   - name     : the name of the task
 559          *   - callback : a PHP callable that performs the task
 560          *   - period   : the period in seconds corresponding to the task frequency
 561          *
 562          * @return Array
 563          */
 564         final public function getPeriodicTasks() {
 565                 $tasks = $this->doGetPeriodicTasks();
 566                 foreach ( $tasks as $name => &$def ) {
 567                         $def['name'] = $name;
 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                 wfProfileIn( __METHOD__ );
 587                 $this->doFlushCaches();
 588                 wfProfileOut( __METHOD__ );
 589         }
 590
 591         /**
 592          * @see JobQueue::flushCaches()
 593          * @return void
 594          */
 595         protected function doFlushCaches() {}
 596
 597         /**
 598          * Get an iterator to traverse over all available jobs in this queue.
 599          * This does not include jobs that are currently acquired or delayed.
 600          * Note: results may be stale if the queue is concurrently modified.
 601          *
 602          * @return Iterator
 603          * @throws JobQueueError
 604          */
 605         abstract public function getAllQueuedJobs();
 606
 607         /**
 608          * Get an iterator to traverse over all delayed jobs in this queue.
 609          * Note: results may be stale if the queue is concurrently modified.
 610          *
 611          * @return Iterator
 612          * @throws JobQueueError
 613          * @since 1.22
 614          */
 615         public function getAllDelayedJobs() {
 616                 return new ArrayIterator( array() ); // not implemented
 617         }
 618
 619         /**
 620          * Call wfIncrStats() for the queue overall and for the queue type
 621          *
 622          * @param string $key Event type
 623          * @param string $type Job type
 624          * @param integer $delta
 625          * @since 1.22
 626          */
 627         public static function incrStats( $key, $type, $delta = 1 ) {
 628                 wfIncrStats( $key, $delta );
 629                 wfIncrStats( "{$key}-{$type}", $delta );
 630         }
 631
 632         /**
 633          * Namespace the queue with a key to isolate it for testing
 634          *
 635          * @param $key string
 636          * @return void
 637          * @throws MWException
 638          */
 639         public function setTestingPrefix( $key ) {
 640                 throw new MWException( "Queue namespacing not supported for this queue type." );
 641         }
 642 }
 643
 644 /**
 645  * @ingroup JobQueue
 646  * @since 1.22
 647  */
 648 class JobQueueError extends MWException {}
 649 class JobQueueConnectionError extends JobQueueError {}