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
  41         const ROOTJOB_TTL = 2419200; // integer; seconds to remember root jobs (28 days)
  42
  43         /**
  44          * @param $params array
  45          */
  46         protected function __construct( array $params ) {
  47                 $this->wiki = $params['wiki'];
  48                 $this->type = $params['type'];
  49                 $this->claimTTL = isset( $params['claimTTL'] ) ? $params['claimTTL'] : 0;
  50                 $this->maxTries = isset( $params['maxTries'] ) ? $params['maxTries'] : 3;
  51                 if ( isset( $params['order'] ) && $params['order'] !== 'any' ) {
  52                         $this->order = $params['order'];
  53                 } else {
  54                         $this->order = $this->optimalOrder();
  55                 }
  56                 if ( !in_array( $this->order, $this->supportedOrders() ) ) {
  57                         throw new MWException( __CLASS__ . " does not support '{$this->order}' order." );
  58                 }
  59                 $this->checkDelay = !empty( $params['checkDelay'] );
  60                 if ( $this->checkDelay && !$this->supportsDelayedJobs() ) {
  61                         throw new MWException( __CLASS__ . " does not support delayed jobs." );
  62                 }
  63         }
  64
  65         /**
  66          * Get a job queue object of the specified type.
  67          * $params includes:
  68          *   - class      : What job class to use (determines job type)
  69          *   - wiki       : wiki ID of the wiki the jobs are for (defaults to current wiki)
  70          *   - type       : The name of the job types this queue handles
  71          *   - order      : Order that pop() selects jobs, one of "fifo", "timestamp" or "random".
  72          *                  If "fifo" is used, the queue will effectively be FIFO. Note that job
  73          *                  completion will not appear to be exactly FIFO if there are multiple
  74          *                  job runners since jobs can take different times to finish once popped.
  75          *                  If "timestamp" is used, the queue will at least be loosely ordered
  76          *                  by timestamp, allowing for some jobs to be popped off out of order.
  77          *                  If "random" is used, pop() will pick jobs in random order.
  78          *                  Note that it may only be weakly random (e.g. a lottery of the oldest X).
  79          *                  If "any" is choosen, the queue will use whatever order is the fastest.
  80          *                  This might be useful for improving concurrency for job acquisition.
  81          *   - claimTTL   : If supported, the queue will recycle jobs that have been popped
  82          *                  but not acknowledged as completed after this many seconds. Recycling
  83          *                  of jobs simple means re-inserting them into the queue. Jobs can be
  84          *                  attempted up to three times before being discarded.
  85          *   - checkDelay : If supported, respect Job::getReleaseTimestamp() in the push functions.
  86          *                  This lets delayed jobs wait in a staging area until a given timestamp is
  87          *                  reached, at which point they will enter the queue. If this is not enabled
  88          *                  or not supported, an exception will be thrown on delayed job insertion.
  89          *
  90          * Queue classes should throw an exception if they do not support the options given.
  91          *
  92          * @param $params array
  93          * @return JobQueue
  94          * @throws MWException
  95          */
  96         final public static function factory( array $params ) {
  97                 $class = $params['class'];
  98                 if ( !MWInit::classExists( $class ) ) {
  99                         throw new MWException( "Invalid job queue class '$class'." );
 100                 }
 101                 $obj = new $class( $params );
 102                 if ( !( $obj instanceof self ) ) {
 103                         throw new MWException( "Class '$class' is not a " . __CLASS__ . " class." );
 104                 }
 105                 return $obj;
 106         }
 107
 108         /**
 109          * @return string Wiki ID
 110          */
 111         final public function getWiki() {
 112                 return $this->wiki;
 113         }
 114
 115         /**
 116          * @return string Job type that this queue handles
 117          */
 118         final public function getType() {
 119                 return $this->type;
 120         }
 121
 122         /**
 123          * @return string One of (random, timestamp, fifo)
 124          */
 125         final public function getOrder() {
 126                 return $this->order;
 127         }
 128
 129         /**
 130          * @return Array Subset of (random, timestamp, fifo)
 131          */
 132         abstract protected function supportedOrders();
 133
 134         /**
 135          * @return string One of (random, timestamp, fifo)
 136          */
 137         abstract protected function optimalOrder();
 138
 139         /**
 140          * @return boolean Whether delayed jobs are supported
 141          */
 142         protected function supportsDelayedJobs() {
 143                 return false; // not implemented
 144         }
 145
 146         /**
 147          * Quickly check if the queue has no available (unacquired, non-delayed) jobs.
 148          * Queue classes should use caching if they are any slower without memcached.
 149          *
 150          * If caching is used, this might return false when there are actually no jobs.
 151          * If pop() is called and returns false then it should correct the cache. Also,
 152          * calling flushCaches() first prevents this. However, this affect is typically
 153          * not distinguishable from the race condition between isEmpty() and pop().
 154          *
 155          * @return bool
 156          * @throws MWException
 157          */
 158         final public function isEmpty() {
 159                 wfProfileIn( __METHOD__ );
 160                 $res = $this->doIsEmpty();
 161                 wfProfileOut( __METHOD__ );
 162                 return $res;
 163         }
 164
 165         /**
 166          * @see JobQueue::isEmpty()
 167          * @return bool
 168          */
 169         abstract protected function doIsEmpty();
 170
 171         /**
 172          * Get the number of available (unacquired, non-delayed) jobs in the queue.
 173          * Queue classes should use caching if they are any slower without memcached.
 174          *
 175          * If caching is used, this number might be out of date for a minute.
 176          *
 177          * @return integer
 178          * @throws MWException
 179          */
 180         final public function getSize() {
 181                 wfProfileIn( __METHOD__ );
 182                 $res = $this->doGetSize();
 183                 wfProfileOut( __METHOD__ );
 184                 return $res;
 185         }
 186
 187         /**
 188          * @see JobQueue::getSize()
 189          * @return integer
 190          */
 191         abstract protected function doGetSize();
 192
 193         /**
 194          * Get the number of acquired jobs (these are temporarily out of the queue).
 195          * Queue classes should use caching if they are any slower without memcached.
 196          *
 197          * If caching is used, this number might be out of date for a minute.
 198          *
 199          * @return integer
 200          * @throws MWException
 201          */
 202         final public function getAcquiredCount() {
 203                 wfProfileIn( __METHOD__ );
 204                 $res = $this->doGetAcquiredCount();
 205                 wfProfileOut( __METHOD__ );
 206                 return $res;
 207         }
 208
 209         /**
 210          * @see JobQueue::getAcquiredCount()
 211          * @return integer
 212          */
 213         abstract protected function doGetAcquiredCount();
 214
 215         /**
 216          * Get the number of delayed jobs (these are temporarily out of the queue).
 217          * Queue classes should use caching if they are any slower without memcached.
 218          *
 219          * If caching is used, this number might be out of date for a minute.
 220          *
 221          * @return integer
 222          * @throws MWException
 223          * @since 1.22
 224          */
 225         final public function getDelayedCount() {
 226                 wfProfileIn( __METHOD__ );
 227                 $res = $this->doGetDelayedCount();
 228                 wfProfileOut( __METHOD__ );
 229                 return $res;
 230         }
 231
 232         /**
 233          * @see JobQueue::getDelayedCount()
 234          * @return integer
 235          */
 236         protected function doGetDelayedCount() {
 237                 return 0; // not implemented
 238         }
 239
 240         /**
 241          * Push a single jobs into the queue.
 242          * This does not require $wgJobClasses to be set for the given job type.
 243          * Outside callers should use JobQueueGroup::push() instead of this function.
 244          *
 245          * @param $jobs Job|Array
 246          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 247          * @return bool Returns false on failure
 248          * @throws MWException
 249          */
 250         final public function push( $jobs, $flags = 0 ) {
 251                 return $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 252         }
 253
 254         /**
 255          * Push a batch of jobs into the queue.
 256          * This does not require $wgJobClasses to be set for the given job type.
 257          * Outside callers should use JobQueueGroup::push() instead of this function.
 258          *
 259          * @param array $jobs List of Jobs
 260          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 261          * @return bool Returns false on failure
 262          * @throws MWException
 263          */
 264         final public function batchPush( array $jobs, $flags = 0 ) {
 265                 if ( !count( $jobs ) ) {
 266                         return true; // nothing to do
 267                 }
 268
 269                 foreach ( $jobs as $job ) {
 270                         if ( $job->getType() !== $this->type ) {
 271                                 throw new MWException(
 272                                         "Got '{$job->getType()}' job; expected a '{$this->type}' job." );
 273                         } elseif ( $job->getReleaseTimestamp() && !$this->checkDelay ) {
 274                                 throw new MWException(
 275                                         "Got delayed '{$job->getType()}' job; delays are not supported." );
 276                         }
 277                 }
 278
 279                 wfProfileIn( __METHOD__ );
 280                 $ok = $this->doBatchPush( $jobs, $flags );
 281                 wfProfileOut( __METHOD__ );
 282                 return $ok;
 283         }
 284
 285         /**
 286          * @see JobQueue::batchPush()
 287          * @return bool
 288          */
 289         abstract protected function doBatchPush( array $jobs, $flags );
 290
 291         /**
 292          * Pop a job off of the queue.
 293          * This requires $wgJobClasses to be set for the given job type.
 294          * Outside callers should use JobQueueGroup::pop() instead of this function.
 295          *
 296          * @return Job|bool Returns false if there are no jobs
 297          * @throws MWException
 298          */
 299         final public function pop() {
 300                 global $wgJobClasses;
 301
 302                 if ( $this->wiki !== wfWikiID() ) {
 303                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 304                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 305                         // Do not pop jobs if there is no class for the queue type
 306                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 307                 }
 308
 309                 wfProfileIn( __METHOD__ );
 310                 $job = $this->doPop();
 311                 wfProfileOut( __METHOD__ );
 312
 313                 // Flag this job as an old duplicate based on its "root" job...
 314                 try {
 315                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 316                                 wfIncrStats( 'job-pop-duplicate' );
 317                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 318                         }
 319                 } catch ( MWException $e ) {} // don't lose jobs over this
 320
 321                 return $job;
 322         }
 323
 324         /**
 325          * @see JobQueue::pop()
 326          * @return Job
 327          */
 328         abstract protected function doPop();
 329
 330         /**
 331          * Acknowledge that a job was completed.
 332          *
 333          * This does nothing for certain queue classes or if "claimTTL" is not set.
 334          * Outside callers should use JobQueueGroup::ack() instead of this function.
 335          *
 336          * @param $job Job
 337          * @return bool
 338          * @throws MWException
 339          */
 340         final public function ack( Job $job ) {
 341                 if ( $job->getType() !== $this->type ) {
 342                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 343                 }
 344                 wfProfileIn( __METHOD__ );
 345                 $ok = $this->doAck( $job );
 346                 wfProfileOut( __METHOD__ );
 347                 return $ok;
 348         }
 349
 350         /**
 351          * @see JobQueue::ack()
 352          * @return bool
 353          */
 354         abstract protected function doAck( Job $job );
 355
 356         /**
 357          * Register the "root job" of a given job into the queue for de-duplication.
 358          * This should only be called right *after* all the new jobs have been inserted.
 359          * This is used to turn older, duplicate, job entries into no-ops. The root job
 360          * information will remain in the registry until it simply falls out of cache.
 361          *
 362          * This requires that $job has two special fields in the "params" array:
 363          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 364          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 365          *
 366          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 367          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 368          * spawned when a template is edited. One can think of the task as "update links
 369          * of pages that use template X" and an instance of that task as a "root job".
 370          * However, what actually goes into the queue are potentially many refreshLinks2 jobs.
 371          * Since these jobs include things like page ID ranges and DB master positions, and morph
 372          * into smaller refreshLinks2 jobs recursively, simple duplicate detection (like job_sha1)
 373          * for individual jobs being identical is not useful.
 374          *
 375          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 376          * is edited again, we want all of these old refreshLinks jobs for that template to become
 377          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 378          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 379          * previous "root job" for the same task of "update links of pages that use template X".
 380          *
 381          * This does nothing for certain queue classes.
 382          *
 383          * @param $job Job
 384          * @return bool
 385          * @throws MWException
 386          */
 387         final public function deduplicateRootJob( Job $job ) {
 388                 if ( $job->getType() !== $this->type ) {
 389                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 390                 }
 391                 wfProfileIn( __METHOD__ );
 392                 $ok = $this->doDeduplicateRootJob( $job );
 393                 wfProfileOut( __METHOD__ );
 394                 return $ok;
 395         }
 396
 397         /**
 398          * @see JobQueue::deduplicateRootJob()
 399          * @param $job Job
 400          * @return bool
 401          */
 402         protected function doDeduplicateRootJob( Job $job ) {
 403                 global $wgMemc;
 404
 405                 $params = $job->getParams();
 406                 if ( !isset( $params['rootJobSignature'] ) ) {
 407                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 408                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 409                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 410                 }
 411                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 412                 // Callers should call batchInsert() and then this function so that if the insert
 413                 // fails, the de-duplication registration will be aborted. Since the insert is
 414                 // deferred till "transaction idle", do the same here, so that the ordering is
 415                 // maintained. Having only the de-duplication registration succeed would cause
 416                 // jobs to become no-ops without any actual jobs that made them redundant.
 417                 $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 418                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 419                         return true; // a newer version of this root job was enqueued
 420                 }
 421
 422                 // Update the timestamp of the last root job started at the location...
 423                 return $wgMemc->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 424         }
 425
 426         /**
 427          * Check if the "root" job of a given job has been superseded by a newer one
 428          *
 429          * @param $job Job
 430          * @return bool
 431          * @throws MWException
 432          */
 433         final protected function isRootJobOldDuplicate( Job $job ) {
 434                 if ( $job->getType() !== $this->type ) {
 435                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 436                 }
 437                 wfProfileIn( __METHOD__ );
 438                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 439                 wfProfileOut( __METHOD__ );
 440                 return $isDuplicate;
 441         }
 442
 443         /**
 444          * @see JobQueue::isRootJobOldDuplicate()
 445          * @param Job $job
 446          * @return bool
 447          */
 448         protected function doIsRootJobOldDuplicate( Job $job ) {
 449                 global $wgMemc;
 450
 451                 $params = $job->getParams();
 452                 if ( !isset( $params['rootJobSignature'] ) ) {
 453                         return false; // job has no de-deplication info
 454                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 455                         trigger_error( "Cannot check root job; missing 'rootJobTimestamp'." );
 456                         return false;
 457                 }
 458
 459                 // Get the last time this root job was enqueued
 460                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 461
 462                 // Check if a new root job was started at the location after this one's...
 463                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 464         }
 465
 466         /**
 467          * @param string $signature Hash identifier of the root job
 468          * @return string
 469          */
 470         protected function getRootJobCacheKey( $signature ) {
 471                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 472                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 473         }
 474
 475         /**
 476          * Wait for any slaves or backup servers to catch up.
 477          *
 478          * This does nothing for certain queue classes.
 479          *
 480          * @return void
 481          * @throws MWException
 482          */
 483         final public function waitForBackups() {
 484                 wfProfileIn( __METHOD__ );
 485                 $this->doWaitForBackups();
 486                 wfProfileOut( __METHOD__ );
 487         }
 488
 489         /**
 490          * @see JobQueue::waitForBackups()
 491          * @return void
 492          */
 493         protected function doWaitForBackups() {}
 494
 495         /**
 496          * Return a map of task names to task definition maps.
 497          * A "task" is a fast periodic queue maintenance action.
 498          * Mutually exclusive tasks must implement their own locking in the callback.
 499          *
 500          * Each task value is an associative array with:
 501          *   - name     : the name of the task
 502          *   - callback : a PHP callable that performs the task
 503          *   - period   : the period in seconds corresponding to the task frequency
 504          *
 505          * @return Array
 506          */
 507         final public function getPeriodicTasks() {
 508                 $tasks = $this->doGetPeriodicTasks();
 509                 foreach ( $tasks as $name => &$def ) {
 510                         $def['name'] = $name;
 511                 }
 512                 return $tasks;
 513         }
 514
 515         /**
 516          * @see JobQueue::getPeriodicTasks()
 517          * @return Array
 518          */
 519         protected function doGetPeriodicTasks() {
 520                 return array();
 521         }
 522
 523         /**
 524          * Clear any process and persistent caches
 525          *
 526          * @return void
 527          */
 528         final public function flushCaches() {
 529                 wfProfileIn( __METHOD__ );
 530                 $this->doFlushCaches();
 531                 wfProfileOut( __METHOD__ );
 532         }
 533
 534         /**
 535          * @see JobQueue::flushCaches()
 536          * @return void
 537          */
 538         protected function doFlushCaches() {}
 539
 540         /**
 541          * Get an iterator to traverse over all available jobs in this queue.
 542          * This does not include jobs that are currently acquired or delayed.
 543          * This should only be called on a queue that is no longer being popped.
 544          *
 545          * @return Iterator|Traversable|Array
 546          * @throws MWException
 547          */
 548         abstract public function getAllQueuedJobs();
 549
 550         /**
 551          * Get an iterator to traverse over all delayed jobs in this queue.
 552          * This should only be called on a queue that is no longer being popped.
 553          *
 554          * @return Iterator|Traversable|Array
 555          * @throws MWException
 556          * @since 1.22
 557          */
 558         public function getAllDelayedJobs() {
 559                 return array(); // not implemented
 560         }
 561
 562         /**
 563          * Namespace the queue with a key to isolate it for testing
 564          *
 565          * @param $key string
 566          * @return void
 567          * @throws MWException
 568          */
 569         public function setTestingPrefix( $key ) {
 570                 throw new MWException( "Queue namespacing not supported for this queue type." );
 571         }
 572 }