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