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 ( !MWInit::classExists( $class ) ) {
 100                         throw new MWException( "Invalid job queue class '$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)
 125          */
 126         final public function getOrder() {
 127                 return $this->order;
 128         }
 129
 130         /**
 131          * @return Array Subset of (random, timestamp, fifo)
 132          */
 133         abstract protected function supportedOrders();
 134
 135         /**
 136          * @return string One of (random, timestamp, fifo)
 137          */
 138         abstract protected function optimalOrder();
 139
 140         /**
 141          * @return boolean Whether delayed jobs are supported
 142          */
 143         protected function supportsDelayedJobs() {
 144                 return false; // not implemented
 145         }
 146
 147         /**
 148          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 149          * Queue classes should use caching if they are any slower without memcached.
 150          *
 151          * If caching is used, this might return false when there are actually no jobs.
 152          * If pop() is called and returns false then it should correct the cache. Also,
 153          * calling flushCaches() first prevents this. However, this affect is typically
 154          * not distinguishable from the race condition between isEmpty() and pop().
 155          *
 156          * @return bool
 157          * @throws MWException
 158          */
 159         final public function isEmpty() {
 160                 wfProfileIn( __METHOD__ );
 161                 $res = $this->doIsEmpty();
 162                 wfProfileOut( __METHOD__ );
 163                 return $res;
 164         }
 165
 166         /**
 167          * @see JobQueue::isEmpty()
 168          * @return bool
 169          */
 170         abstract protected function doIsEmpty();
 171
 172         /**
 173          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 174          * Queue classes should use caching if they are any slower without memcached.
 175          *
 176          * If caching is used, this number might be out of date for a minute.
 177          *
 178          * @return integer
 179          * @throws MWException
 180          */
 181         final public function getSize() {
 182                 wfProfileIn( __METHOD__ );
 183                 $res = $this->doGetSize();
 184                 wfProfileOut( __METHOD__ );
 185                 return $res;
 186         }
 187
 188         /**
 189          * @see JobQueue::getSize()
 190          * @return integer
 191          */
 192         abstract protected function doGetSize();
 193
 194         /**
 195          * Get the number of acquired jobs (these are temporarily out of the queue).
 196          * Queue classes should use caching if they are any slower without memcached.
 197          *
 198          * If caching is used, this number might be out of date for a minute.
 199          *
 200          * @return integer
 201          * @throws MWException
 202          */
 203         final public function getAcquiredCount() {
 204                 wfProfileIn( __METHOD__ );
 205                 $res = $this->doGetAcquiredCount();
 206                 wfProfileOut( __METHOD__ );
 207                 return $res;
 208         }
 209
 210         /**
 211          * @see JobQueue::getAcquiredCount()
 212          * @return integer
 213          */
 214         abstract protected function doGetAcquiredCount();
 215
 216         /**
 217          * Get the number of delayed jobs (these are temporarily out of the queue).
 218          * Queue classes should use caching if they are any slower without memcached.
 219          *
 220          * If caching is used, this number might be out of date for a minute.
 221          *
 222          * @return integer
 223          * @throws MWException
 224          * @since 1.22
 225          */
 226         final public function getDelayedCount() {
 227                 wfProfileIn( __METHOD__ );
 228                 $res = $this->doGetDelayedCount();
 229                 wfProfileOut( __METHOD__ );
 230                 return $res;
 231         }
 232
 233         /**
 234          * @see JobQueue::getDelayedCount()
 235          * @return integer
 236          */
 237         protected function doGetDelayedCount() {
 238                 return 0; // not implemented
 239         }
 240
 241         /**
 242          * Get the number of acquired jobs that can no longer be attempted.
 243          * Queue classes should use caching if they are any slower without memcached.
 244          *
 245          * If caching is used, this number might be out of date for a minute.
 246          *
 247          * @return integer
 248          * @throws MWException
 249          */
 250         final public function getAbandonedCount() {
 251                 wfProfileIn( __METHOD__ );
 252                 $res = $this->doGetAbandonedCount();
 253                 wfProfileOut( __METHOD__ );
 254                 return $res;
 255         }
 256
 257         /**
 258          * @see JobQueue::getAbandonedCount()
 259          * @return integer
 260          */
 261         protected function doGetAbandonedCount() {
 262                 return 0; // not implemented
 263         }
 264
 265         /**
 266          * Push a single jobs into the queue.
 267          * This does not require $wgJobClasses to be set for the given job type.
 268          * Outside callers should use JobQueueGroup::push() instead of this function.
 269          *
 270          * @param $jobs Job|Array
 271          * @param $flags integer Bitfield (supports JobQueue::QOS_ATOMIC)
 272          * @return bool Returns false on failure
 273          * @throws MWException
 274          */
 275         final public function push( $jobs, $flags = 0 ) {
 276                 return $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 277         }
 278
 279         /**
 280          * Push a batch of 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 array $jobs List of Jobs
 285          * @param $flags integer Bitfield (supports JobQueue::QOS_ATOMIC)
 286          * @return bool Returns false on failure
 287          * @throws MWException
 288          */
 289         final public function batchPush( array $jobs, $flags = 0 ) {
 290                 if ( !count( $jobs ) ) {
 291                         return true; // nothing to do
 292                 }
 293
 294                 foreach ( $jobs as $job ) {
 295                         if ( $job->getType() !== $this->type ) {
 296                                 throw new MWException(
 297                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 298                         } elseif ( $job->getReleaseTimestamp() && !$this->checkDelay ) {
 299                                 throw new MWException(
 300                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 301                         }
 302                 }
 303
 304                 wfProfileIn( __METHOD__ );
 305                 $ok = $this->doBatchPush( $jobs, $flags );
 306                 wfProfileOut( __METHOD__ );
 307                 return $ok;
 308         }
 309
 310         /**
 311          * @see JobQueue::batchPush()
 312          * @return bool
 313          */
 314         abstract protected function doBatchPush( array $jobs, $flags );
 315
 316         /**
 317          * Pop a job off of the queue.
 318          * This requires $wgJobClasses to be set for the given job type.
 319          * Outside callers should use JobQueueGroup::pop() instead of this function.
 320          *
 321          * @return Job|bool Returns false if there are no jobs
 322          * @throws MWException
 323          */
 324         final public function pop() {
 325                 global $wgJobClasses;
 326
 327                 if ( $this->wiki !== wfWikiID() ) {
 328                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 329                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 330                         // Do not pop jobs if there is no class for the queue type
 331                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 332                 }
 333
 334                 wfProfileIn( __METHOD__ );
 335                 $job = $this->doPop();
 336                 wfProfileOut( __METHOD__ );
 337
 338                 // Flag this job as an old duplicate based on its "root" job...
 339                 try {
 340                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 341                                 wfIncrStats( 'job-pop-duplicate' );
 342                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 343                         }
 344                 } catch ( MWException $e ) {} // don't lose jobs over this
 345
 346                 return $job;
 347         }
 348
 349         /**
 350          * @see JobQueue::pop()
 351          * @return Job
 352          */
 353         abstract protected function doPop();
 354
 355         /**
 356          * Acknowledge that a job was completed.
 357          *
 358          * This does nothing for certain queue classes or if "claimTTL" is not set.
 359          * Outside callers should use JobQueueGroup::ack() instead of this function.
 360          *
 361          * @param $job Job
 362          * @return bool
 363          * @throws MWException
 364          */
 365         final public function ack( Job $job ) {
 366                 if ( $job->getType() !== $this->type ) {
 367                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 368                 }
 369                 wfProfileIn( __METHOD__ );
 370                 $ok = $this->doAck( $job );
 371                 wfProfileOut( __METHOD__ );
 372                 return $ok;
 373         }
 374
 375         /**
 376          * @see JobQueue::ack()
 377          * @return bool
 378          */
 379         abstract protected function doAck( Job $job );
 380
 381         /**
 382          * Register the "root job" of a given job into the queue for de-duplication.
 383          * This should only be called right *after* all the new jobs have been inserted.
 384          * This is used to turn older, duplicate, job entries into no-ops. The root job
 385          * information will remain in the registry until it simply falls out of cache.
 386          *
 387          * This requires that $job has two special fields in the "params" array:
 388          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 389          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 390          *
 391          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 392          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 393          * spawned when a template is edited. One can think of the task as "update links
 394          * of pages that use template X" and an instance of that task as a "root job".
 395          * However, what actually goes into the queue are potentially many refreshLinks2 jobs.
 396          * Since these jobs include things like page ID ranges and DB master positions, and morph
 397          * into smaller refreshLinks2 jobs recursively, simple duplicate detection (like job_sha1)
 398          * for individual jobs being identical is not useful.
 399          *
 400          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 401          * is edited again, we want all of these old refreshLinks jobs for that template to become
 402          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 403          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 404          * previous "root job" for the same task of "update links of pages that use template X".
 405          *
 406          * This does nothing for certain queue classes.
 407          *
 408          * @param $job Job
 409          * @return bool
 410          * @throws MWException
 411          */
 412         final public function deduplicateRootJob( Job $job ) {
 413                 if ( $job->getType() !== $this->type ) {
 414                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 415                 }
 416                 wfProfileIn( __METHOD__ );
 417                 $ok = $this->doDeduplicateRootJob( $job );
 418                 wfProfileOut( __METHOD__ );
 419                 return $ok;
 420         }
 421
 422         /**
 423          * @see JobQueue::deduplicateRootJob()
 424          * @param $job Job
 425          * @return bool
 426          */
 427         protected function doDeduplicateRootJob( Job $job ) {
 428                 global $wgMemc;
 429
 430                 $params = $job->getParams();
 431                 if ( !isset( $params['rootJobSignature'] ) ) {
 432                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 433                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 434                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 435                 }
 436                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 437                 // Callers should call batchInsert() and then this function so that if the insert
 438                 // fails, the de-duplication registration will be aborted. Since the insert is
 439                 // deferred till "transaction idle", do the same here, so that the ordering is
 440                 // maintained. Having only the de-duplication registration succeed would cause
 441                 // jobs to become no-ops without any actual jobs that made them redundant.
 442                 $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 443                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 444                         return true; // a newer version of this root job was enqueued
 445                 }
 446
 447                 // Update the timestamp of the last root job started at the location...
 448                 return $wgMemc->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 449         }
 450
 451         /**
 452          * Check if the "root" job of a given job has been superseded by a newer one
 453          *
 454          * @param $job Job
 455          * @return bool
 456          * @throws MWException
 457          */
 458         final protected function isRootJobOldDuplicate( Job $job ) {
 459                 if ( $job->getType() !== $this->type ) {
 460                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 461                 }
 462                 wfProfileIn( __METHOD__ );
 463                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 464                 wfProfileOut( __METHOD__ );
 465                 return $isDuplicate;
 466         }
 467
 468         /**
 469          * @see JobQueue::isRootJobOldDuplicate()
 470          * @param Job $job
 471          * @return bool
 472          */
 473         protected function doIsRootJobOldDuplicate( Job $job ) {
 474                 global $wgMemc;
 475
 476                 $params = $job->getParams();
 477                 if ( !isset( $params['rootJobSignature'] ) ) {
 478                         return false; // job has no de-deplication info
 479                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 480                         trigger_error( "Cannot check root job; missing 'rootJobTimestamp'." );
 481                         return false;
 482                 }
 483
 484                 // Get the last time this root job was enqueued
 485                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 486
 487                 // Check if a new root job was started at the location after this one's...
 488                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 489         }
 490
 491         /**
 492          * @param string $signature Hash identifier of the root job
 493          * @return string
 494          */
 495         protected function getRootJobCacheKey( $signature ) {
 496                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 497                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 498         }
 499
 500         /**
 501          * Wait for any slaves or backup servers to catch up.
 502          *
 503          * This does nothing for certain queue classes.
 504          *
 505          * @return void
 506          * @throws MWException
 507          */
 508         final public function waitForBackups() {
 509                 wfProfileIn( __METHOD__ );
 510                 $this->doWaitForBackups();
 511                 wfProfileOut( __METHOD__ );
 512         }
 513
 514         /**
 515          * @see JobQueue::waitForBackups()
 516          * @return void
 517          */
 518         protected function doWaitForBackups() {}
 519
 520         /**
 521          * Return a map of task names to task definition maps.
 522          * A "task" is a fast periodic queue maintenance action.
 523          * Mutually exclusive tasks must implement their own locking in the callback.
 524          *
 525          * Each task value is an associative array with:
 526          *   - name     : the name of the task
 527          *   - callback : a PHP callable that performs the task
 528          *   - period   : the period in seconds corresponding to the task frequency
 529          *
 530          * @return Array
 531          */
 532         final public function getPeriodicTasks() {
 533                 $tasks = $this->doGetPeriodicTasks();
 534                 foreach ( $tasks as $name => &$def ) {
 535                         $def['name'] = $name;
 536                 }
 537                 return $tasks;
 538         }
 539
 540         /**
 541          * @see JobQueue::getPeriodicTasks()
 542          * @return Array
 543          */
 544         protected function doGetPeriodicTasks() {
 545                 return array();
 546         }
 547
 548         /**
 549          * Clear any process and persistent caches
 550          *
 551          * @return void
 552          */
 553         final public function flushCaches() {
 554                 wfProfileIn( __METHOD__ );
 555                 $this->doFlushCaches();
 556                 wfProfileOut( __METHOD__ );
 557         }
 558
 559         /**
 560          * @see JobQueue::flushCaches()
 561          * @return void
 562          */
 563         protected function doFlushCaches() {}
 564
 565         /**
 566          * Get an iterator to traverse over all available jobs in this queue.
 567          * This does not include jobs that are currently acquired or delayed.
 568          * This should only be called on a queue that is no longer being popped.
 569          *
 570          * @return Iterator|Traversable|Array
 571          * @throws MWException
 572          */
 573         abstract public function getAllQueuedJobs();
 574
 575         /**
 576          * Get an iterator to traverse over all delayed jobs in this queue.
 577          * This should only be called on a queue that is no longer being popped.
 578          *
 579          * @return Iterator|Traversable|Array
 580          * @throws MWException
 581          * @since 1.22
 582          */
 583         public function getAllDelayedJobs() {
 584                 return array(); // not implemented
 585         }
 586
 587         /**
 588          * Namespace the queue with a key to isolate it for testing
 589          *
 590          * @param $key string
 591          * @return void
 592          * @throws MWException
 593          */
 594         public function setTestingPrefix( $key ) {
 595                 throw new MWException( "Queue namespacing not supported for this queue type." );
 596         }
 597 }