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