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
  38         const QoS_Atomic = 1; // integer; "all-or-nothing" job insertions
  39
  40         const ROOTJOB_TTL = 2419200; // integer; seconds to remember root jobs (28 days)
  41
  42         /**
  43          * @param $params array
  44          */
  45         protected function __construct( array $params ) {
  46                 $this->wiki = $params['wiki'];
  47                 $this->type = $params['type'];
  48                 $this->claimTTL = isset( $params['claimTTL'] ) ? $params['claimTTL'] : 0;
  49                 $this->maxTries = isset( $params['maxTries'] ) ? $params['maxTries'] : 3;
  50                 if ( isset( $params['order'] ) && $params['order'] !== 'any' ) {
  51                         $this->order = $params['order'];
  52                 } else {
  53                         $this->order = $this->optimalOrder();
  54                 }
  55                 if ( !in_array( $this->order, $this->supportedOrders() ) ) {
  56                         throw new MWException( __CLASS__ . " does not support '{$this->order}' order." );
  57                 }
  58         }
  59
  60         /**
  61          * Get a job queue object of the specified type.
  62          * $params includes:
  63          *   - class    : What job class to use (determines job type)
  64          *   - wiki     : wiki ID of the wiki the jobs are for (defaults to current wiki)
  65          *   - type     : The name of the job types this queue handles
  66          *   - order    : Order that pop() selects jobs, one of "fifo", "timestamp" or "random".
  67          *                If "fifo" is used, the queue will effectively be FIFO. Note that
  68          *                job completion will not appear to be exactly FIFO if there are multiple
  69          *                job runners since jobs can take different times to finish once popped.
  70          *                If "timestamp" is used, the queue will at least be loosely ordered
  71          *                by timestamp, allowing for some jobs to be popped off out of order.
  72          *                If "random" is used, pop() will pick jobs in random order.
  73          *                Note that it may only be weakly random (e.g. a lottery of the oldest X).
  74          *                If "any" is choosen, the queue will use whatever order is the fastest.
  75          *                This might be useful for improving concurrency for job acquisition.
  76          *   - claimTTL : If supported, the queue will recycle jobs that have been popped
  77          *                but not acknowledged as completed after this many seconds. Recycling
  78          *                of jobs simple means re-inserting them into the queue. Jobs can be
  79          *                attempted up to three times before being discarded.
  80          *
  81          * Queue classes should throw an exception if they do not support the options given.
  82          *
  83          * @param $params array
  84          * @return JobQueue
  85          * @throws MWException
  86          */
  87         final public static function factory( array $params ) {
  88                 $class = $params['class'];
  89                 if ( !MWInit::classExists( $class ) ) {
  90                         throw new MWException( "Invalid job queue class '$class'." );
  91                 }
  92                 $obj = new $class( $params );
  93                 if ( !( $obj instanceof self ) ) {
  94                         throw new MWException( "Class '$class' is not a " . __CLASS__ . " class." );
  95                 }
  96                 return $obj;
  97         }
  98
  99         /**
 100          * @return string Wiki ID
 101          */
 102         final public function getWiki() {
 103                 return $this->wiki;
 104         }
 105
 106         /**
 107          * @return string Job type that this queue handles
 108          */
 109         final public function getType() {
 110                 return $this->type;
 111         }
 112
 113         /**
 114          * @return string One of (random, timestamp, fifo)
 115          */
 116         final public function getOrder() {
 117                 return $this->order;
 118         }
 119
 120         /**
 121          * @return Array Subset of (random, timestamp, fifo)
 122          */
 123         abstract protected function supportedOrders();
 124
 125         /**
 126          * @return string One of (random, timestamp, fifo)
 127          */
 128         abstract protected function optimalOrder();
 129
 130         /**
 131          * Quickly check if the queue is empty (has no available jobs).
 132          * Queue classes should use caching if they are any slower without memcached.
 133          *
 134          * If caching is used, this might return false when there are actually no jobs.
 135          * If pop() is called and returns false then it should correct the cache. Also,
 136          * calling flushCaches() first prevents this. However, this affect is typically
 137          * not distinguishable from the race condition between isEmpty() and pop().
 138          *
 139          * @return bool
 140          * @throws MWException
 141          */
 142         final public function isEmpty() {
 143                 wfProfileIn( __METHOD__ );
 144                 $res = $this->doIsEmpty();
 145                 wfProfileOut( __METHOD__ );
 146                 return $res;
 147         }
 148
 149         /**
 150          * @see JobQueue::isEmpty()
 151          * @return bool
 152          */
 153         abstract protected function doIsEmpty();
 154
 155         /**
 156          * Get the number of available (unacquired) jobs in the queue.
 157          * Queue classes should use caching if they are any slower without memcached.
 158          *
 159          * If caching is used, this number might be out of date for a minute.
 160          *
 161          * @return integer
 162          * @throws MWException
 163          */
 164         final public function getSize() {
 165                 wfProfileIn( __METHOD__ );
 166                 $res = $this->doGetSize();
 167                 wfProfileOut( __METHOD__ );
 168                 return $res;
 169         }
 170
 171         /**
 172          * @see JobQueue::getSize()
 173          * @return integer
 174          */
 175         abstract protected function doGetSize();
 176
 177         /**
 178          * Get the number of acquired jobs (these are temporarily out of the queue).
 179          * Queue classes should use caching if they are any slower without memcached.
 180          *
 181          * If caching is used, this number might be out of date for a minute.
 182          *
 183          * @return integer
 184          * @throws MWException
 185          */
 186         final public function getAcquiredCount() {
 187                 wfProfileIn( __METHOD__ );
 188                 $res = $this->doGetAcquiredCount();
 189                 wfProfileOut( __METHOD__ );
 190                 return $res;
 191         }
 192
 193         /**
 194          * @see JobQueue::getAcquiredCount()
 195          * @return integer
 196          */
 197         abstract protected function doGetAcquiredCount();
 198
 199         /**
 200          * Push a single jobs into the queue.
 201          * This does not require $wgJobClasses to be set for the given job type.
 202          * Outside callers should use JobQueueGroup::push() instead of this function.
 203          *
 204          * @param $jobs Job|Array
 205          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 206          * @return bool Returns false on failure
 207          * @throws MWException
 208          */
 209         final public function push( $jobs, $flags = 0 ) {
 210                 return $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 211         }
 212
 213         /**
 214          * Push a batch of jobs into the queue.
 215          * This does not require $wgJobClasses to be set for the given job type.
 216          * Outside callers should use JobQueueGroup::push() instead of this function.
 217          *
 218          * @param array $jobs List of Jobs
 219          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 220          * @return bool Returns false on failure
 221          * @throws MWException
 222          */
 223         final public function batchPush( array $jobs, $flags = 0 ) {
 224                 if ( !count( $jobs ) ) {
 225                         return true; // nothing to do
 226                 }
 227
 228                 foreach ( $jobs as $job ) {
 229                         if ( $job->getType() !== $this->type ) {
 230                                 throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 231                         }
 232                 }
 233
 234                 wfProfileIn( __METHOD__ );
 235                 $ok = $this->doBatchPush( $jobs, $flags );
 236                 wfProfileOut( __METHOD__ );
 237                 return $ok;
 238         }
 239
 240         /**
 241          * @see JobQueue::batchPush()
 242          * @return bool
 243          */
 244         abstract protected function doBatchPush( array $jobs, $flags );
 245
 246         /**
 247          * Pop a job off of the queue.
 248          * This requires $wgJobClasses to be set for the given job type.
 249          * Outside callers should use JobQueueGroup::pop() instead of this function.
 250          *
 251          * @return Job|bool Returns false if there are no jobs
 252          * @throws MWException
 253          */
 254         final public function pop() {
 255                 global $wgJobClasses;
 256
 257                 if ( $this->wiki !== wfWikiID() ) {
 258                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 259                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 260                         // Do not pop jobs if there is no class for the queue type
 261                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 262                 }
 263
 264                 wfProfileIn( __METHOD__ );
 265                 $job = $this->doPop();
 266                 wfProfileOut( __METHOD__ );
 267
 268                 // Flag this job as an old duplicate based on its "root" job...
 269                 try {
 270                         if ( $job && $this->isRootJobOldDuplicate( $job ) ) {
 271                                 wfIncrStats( 'job-pop-duplicate' );
 272                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 273                         }
 274                 } catch ( MWException $e ) {} // don't lose jobs over this
 275
 276                 return $job;
 277         }
 278
 279         /**
 280          * @see JobQueue::pop()
 281          * @return Job
 282          */
 283         abstract protected function doPop();
 284
 285         /**
 286          * Acknowledge that a job was completed.
 287          *
 288          * This does nothing for certain queue classes or if "claimTTL" is not set.
 289          * Outside callers should use JobQueueGroup::ack() instead of this function.
 290          *
 291          * @param $job Job
 292          * @return bool
 293          * @throws MWException
 294          */
 295         final public function ack( Job $job ) {
 296                 if ( $job->getType() !== $this->type ) {
 297                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 298                 }
 299                 wfProfileIn( __METHOD__ );
 300                 $ok = $this->doAck( $job );
 301                 wfProfileOut( __METHOD__ );
 302                 return $ok;
 303         }
 304
 305         /**
 306          * @see JobQueue::ack()
 307          * @return bool
 308          */
 309         abstract protected function doAck( Job $job );
 310
 311         /**
 312          * Register the "root job" of a given job into the queue for de-duplication.
 313          * This should only be called right *after* all the new jobs have been inserted.
 314          * This is used to turn older, duplicate, job entries into no-ops. The root job
 315          * information will remain in the registry until it simply falls out of cache.
 316          *
 317          * This requires that $job has two special fields in the "params" array:
 318          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 319          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 320          *
 321          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 322          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 323          * spawned when a template is edited. One can think of the task as "update links
 324          * of pages that use template X" and an instance of that task as a "root job".
 325          * However, what actually goes into the queue are potentially many refreshLinks2 jobs.
 326          * Since these jobs include things like page ID ranges and DB master positions, and morph
 327          * into smaller refreshLinks2 jobs recursively, simple duplicate detection (like job_sha1)
 328          * for individual jobs being identical is not useful.
 329          *
 330          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 331          * is edited again, we want all of these old refreshLinks jobs for that template to become
 332          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 333          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 334          * previous "root job" for the same task of "update links of pages that use template X".
 335          *
 336          * This does nothing for certain queue classes.
 337          *
 338          * @param $job Job
 339          * @return bool
 340          * @throws MWException
 341          */
 342         final public function deduplicateRootJob( Job $job ) {
 343                 if ( $job->getType() !== $this->type ) {
 344                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 345                 }
 346                 wfProfileIn( __METHOD__ );
 347                 $ok = $this->doDeduplicateRootJob( $job );
 348                 wfProfileOut( __METHOD__ );
 349                 return $ok;
 350         }
 351
 352         /**
 353          * @see JobQueue::deduplicateRootJob()
 354          * @param $job Job
 355          * @return bool
 356          */
 357         protected function doDeduplicateRootJob( Job $job ) {
 358                 global $wgMemc;
 359
 360                 $params = $job->getParams();
 361                 if ( !isset( $params['rootJobSignature'] ) ) {
 362                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 363                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 364                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 365                 }
 366                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 367                 // Callers should call batchInsert() and then this function so that if the insert
 368                 // fails, the de-duplication registration will be aborted. Since the insert is
 369                 // deferred till "transaction idle", do the same here, so that the ordering is
 370                 // maintained. Having only the de-duplication registration succeed would cause
 371                 // jobs to become no-ops without any actual jobs that made them redundant.
 372                 $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 373                 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 374                         return true; // a newer version of this root job was enqueued
 375                 }
 376
 377                 // Update the timestamp of the last root job started at the location...
 378                 return $wgMemc->set( $key, $params['rootJobTimestamp'], JobQueueDB::ROOTJOB_TTL );
 379         }
 380
 381         /**
 382          * Check if the "root" job of a given job has been superseded by a newer one
 383          *
 384          * @param $job Job
 385          * @return bool
 386          * @throws MWException
 387          */
 388         final protected function isRootJobOldDuplicate( Job $job ) {
 389                 if ( $job->getType() !== $this->type ) {
 390                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 391                 }
 392                 wfProfileIn( __METHOD__ );
 393                 $isDuplicate = $this->doIsRootJobOldDuplicate( $job );
 394                 wfProfileOut( __METHOD__ );
 395                 return $isDuplicate;
 396         }
 397
 398         /**
 399          * @see JobQueue::isRootJobOldDuplicate()
 400          * @param Job $job
 401          * @return bool
 402          */
 403         protected function doIsRootJobOldDuplicate( Job $job ) {
 404                 global $wgMemc;
 405
 406                 $params = $job->getParams();
 407                 if ( !isset( $params['rootJobSignature'] ) ) {
 408                         return false; // job has no de-deplication info
 409                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 410                         trigger_error( "Cannot check root job; missing 'rootJobTimestamp'." );
 411                         return false;
 412                 }
 413
 414                 // Get the last time this root job was enqueued
 415                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 416
 417                 // Check if a new root job was started at the location after this one's...
 418                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 419         }
 420
 421         /**
 422          * @param string $signature Hash identifier of the root job
 423          * @return string
 424          */
 425         protected function getRootJobCacheKey( $signature ) {
 426                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 427                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 428         }
 429
 430         /**
 431          * Wait for any slaves or backup servers to catch up.
 432          *
 433          * This does nothing for certain queue classes.
 434          *
 435          * @return void
 436          * @throws MWException
 437          */
 438         final public function waitForBackups() {
 439                 wfProfileIn( __METHOD__ );
 440                 $this->doWaitForBackups();
 441                 wfProfileOut( __METHOD__ );
 442         }
 443
 444         /**
 445          * @see JobQueue::waitForBackups()
 446          * @return void
 447          */
 448         protected function doWaitForBackups() {}
 449
 450         /**
 451          * Return a map of task names to task definition maps.
 452          * A "task" is a fast periodic queue maintenance action.
 453          * Mutually exclusive tasks must implement their own locking in the callback.
 454          *
 455          * Each task value is an associative array with:
 456          *   - name     : the name of the task
 457          *   - callback : a PHP callable that performs the task
 458          *   - period   : the period in seconds corresponding to the task frequency
 459          *
 460          * @return Array
 461          */
 462         final public function getPeriodicTasks() {
 463                 $tasks = $this->doGetPeriodicTasks();
 464                 foreach ( $tasks as $name => &$def ) {
 465                         $def['name'] = $name;
 466                 }
 467                 return $tasks;
 468         }
 469
 470         /**
 471          * @see JobQueue::getPeriodicTasks()
 472          * @return Array
 473          */
 474         protected function doGetPeriodicTasks() {
 475                 return array();
 476         }
 477
 478         /**
 479          * Clear any process and persistent caches
 480          *
 481          * @return void
 482          */
 483         final public function flushCaches() {
 484                 wfProfileIn( __METHOD__ );
 485                 $this->doFlushCaches();
 486                 wfProfileOut( __METHOD__ );
 487         }
 488
 489         /**
 490          * @see JobQueue::flushCaches()
 491          * @return void
 492          */
 493         protected function doFlushCaches() {}
 494
 495         /**
 496          * Get an iterator to traverse over all of the jobs in this queue.
 497          * This does not include jobs that are current acquired. In general,
 498          * this should only be called on a queue that is no longer being popped.
 499          *
 500          * @return Iterator|Traversable|Array
 501          * @throws MWException
 502          */
 503         abstract public function getAllQueuedJobs();
 504
 505         /**
 506          * Namespace the queue with a key to isolate it for testing
 507          *
 508          * @param $key string
 509          * @return void
 510          * @throws MWException
 511          */
 512         public function setTestingPrefix( $key ) {
 513                 throw new MWException( "Queue namespacing not supported for this queue type." );
 514         }
 515 }