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          *
 168          * @param $jobs Job|Array
 169          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 170          * @return bool Returns false on failure
 171          * @throws MWException
 172          */
 173         final public function push( $jobs, $flags = 0 ) {
 174                 return $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 175         }
 176
 177         /**
 178          * Push a batch of jobs into the queue.
 179          * This does not require $wgJobClasses to be set for the given job type.
 180          *
 181          * @param $jobs array List of Jobs
 182          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 183          * @return bool Returns false on failure
 184          * @throws MWException
 185          */
 186         final public function batchPush( array $jobs, $flags = 0 ) {
 187                 if ( !count( $jobs ) ) {
 188                         return true; // nothing to do
 189                 }
 190                 foreach ( $jobs as $job ) {
 191                         if ( $job->getType() !== $this->type ) {
 192                                 throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 193                         }
 194                 }
 195
 196                 wfProfileIn( __METHOD__ );
 197                 $ok = $this->doBatchPush( $jobs, $flags );
 198                 wfProfileOut( __METHOD__ );
 199                 return $ok;
 200         }
 201
 202         /**
 203          * @see JobQueue::batchPush()
 204          * @return bool
 205          */
 206         abstract protected function doBatchPush( array $jobs, $flags );
 207
 208         /**
 209          * Pop a job off of the queue.
 210          * This requires $wgJobClasses to be set for the given job type.
 211          *
 212          * @return Job|bool Returns false if there are no jobs
 213          * @throws MWException
 214          */
 215         final public function pop() {
 216                 global $wgJobClasses;
 217
 218                 if ( $this->wiki !== wfWikiID() ) {
 219                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 220                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 221                         // Do not pop jobs if there is no class for the queue type
 222                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 223                 }
 224
 225                 wfProfileIn( __METHOD__ );
 226                 $job = $this->doPop();
 227                 wfProfileOut( __METHOD__ );
 228                 return $job;
 229         }
 230
 231         /**
 232          * @see JobQueue::pop()
 233          * @return Job
 234          */
 235         abstract protected function doPop();
 236
 237         /**
 238          * Acknowledge that a job was completed.
 239          *
 240          * This does nothing for certain queue classes or if "claimTTL" is not set.
 241          *
 242          * @param $job Job
 243          * @return bool
 244          * @throws MWException
 245          */
 246         final public function ack( Job $job ) {
 247                 if ( $job->getType() !== $this->type ) {
 248                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 249                 }
 250                 wfProfileIn( __METHOD__ );
 251                 $ok = $this->doAck( $job );
 252                 wfProfileOut( __METHOD__ );
 253                 return $ok;
 254         }
 255
 256         /**
 257          * @see JobQueue::ack()
 258          * @return bool
 259          */
 260         abstract protected function doAck( Job $job );
 261
 262         /**
 263          * Register the "root job" of a given job into the queue for de-duplication.
 264          * This should only be called right *after* all the new jobs have been inserted.
 265          * This is used to turn older, duplicate, job entries into no-ops. The root job
 266          * information will remain in the registry until it simply falls out of cache.
 267          *
 268          * This requires that $job has two special fields in the "params" array:
 269          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 270          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 271          *
 272          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 273          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 274          * spawned when a template is edited. One can think of the task as "update links
 275          * of pages that use template X" and an instance of that task as a "root job".
 276          * However, what actually goes into the queue are potentially many refreshLinks2 jobs.
 277          * Since these jobs include things like page ID ranges and DB master positions, and morph
 278          * into smaller refreshLinks2 jobs recursively, simple duplicate detection (like job_sha1)
 279          * for individual jobs being identical is not useful.
 280          *
 281          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 282          * is edited again, we want all of these old refreshLinks jobs for that template to become
 283          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 284          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 285          * previous "root job" for the same task of "update links of pages that use template X".
 286          *
 287          * This does nothing for certain queue classes.
 288          *
 289          * @param $job Job
 290          * @return bool
 291          * @throws MWException
 292          */
 293         final public function deduplicateRootJob( Job $job ) {
 294                 if ( $job->getType() !== $this->type ) {
 295                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 296                 }
 297                 wfProfileIn( __METHOD__ );
 298                 $ok = $this->doDeduplicateRootJob( $job );
 299                 wfProfileOut( __METHOD__ );
 300                 return $ok;
 301         }
 302
 303         /**
 304          * @see JobQueue::deduplicateRootJob()
 305          * @param $job Job
 306          * @return bool
 307          */
 308         protected function doDeduplicateRootJob( Job $job ) {
 309                 return true;
 310         }
 311
 312         /**
 313          * Wait for any slaves or backup servers to catch up.
 314          *
 315          * This does nothing for certain queue classes.
 316          *
 317          * @return void
 318          * @throws MWException
 319          */
 320         final public function waitForBackups() {
 321                 wfProfileIn( __METHOD__ );
 322                 $this->doWaitForBackups();
 323                 wfProfileOut( __METHOD__ );
 324         }
 325
 326         /**
 327          * @see JobQueue::waitForBackups()
 328          * @return void
 329          */
 330         protected function doWaitForBackups() {}
 331
 332         /**
 333          * Return a map of task names to task definition maps.
 334          * A "task" is a fast periodic queue maintenance action.
 335          * Mutually exclusive tasks must implement their own locking in the callback.
 336          *
 337          * Each task value is an associative array with:
 338          *   - name     : the name of the task
 339          *   - callback : a PHP callable that performs the task
 340          *   - period   : the period in seconds corresponding to the task frequency
 341          *
 342          * @return Array
 343          */
 344         final public function getPeriodicTasks() {
 345                 $tasks = $this->doGetPeriodicTasks();
 346                 foreach ( $tasks as $name => &$def ) {
 347                         $def['name'] = $name;
 348                 }
 349                 return $tasks;
 350         }
 351
 352         /**
 353          * @see JobQueue::getPeriodicTasks()
 354          * @return Array
 355          */
 356         protected function doGetPeriodicTasks() {
 357                 return array();
 358         }
 359
 360         /**
 361          * Clear any process and persistent caches
 362          *
 363          * @return void
 364          */
 365         final public function flushCaches() {
 366                 wfProfileIn( __METHOD__ );
 367                 $this->doFlushCaches();
 368                 wfProfileOut( __METHOD__ );
 369         }
 370
 371         /**
 372          * @see JobQueue::flushCaches()
 373          * @return void
 374          */
 375         protected function doFlushCaches() {}
 376
 377         /**
 378          * Namespace the queue with a key to isolate it for testing
 379          *
 380          * @param $key string
 381          * @return void
 382          * @throws MWException
 383          */
 384         public function setTestingPrefix( $key ) {
 385                 throw new MWException( "Queue namespacing not support for this queue type." );
 386         }
 387 }