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          * If caching is used, this might return false when there are actually no jobs.
 133          * If pop() is called and returns false then it should correct the cache. Also,
 134          * calling flushCaches() first prevents this. However, this affect is typically
 135          * not distinguishable from the race condition between isEmpty() and pop().
 136          *
 137          * @return bool
 138          * @throws MWException
 139          */
 140         final public function isEmpty() {
 141                 wfProfileIn( __METHOD__ );
 142                 $res = $this->doIsEmpty();
 143                 wfProfileOut( __METHOD__ );
 144                 return $res;
 145         }
 146
 147         /**
 148          * @see JobQueue::isEmpty()
 149          * @return bool
 150          */
 151         abstract protected function doIsEmpty();
 152
 153         /**
 154          * Get the number of available (unacquired) jobs in the queue.
 155          * Queue classes should use caching if they are any slower without memcached.
 156          *
 157          * If caching is used, this number might be out of date for a minute.
 158          *
 159          * @return integer
 160          * @throws MWException
 161          */
 162         final public function getSize() {
 163                 wfProfileIn( __METHOD__ );
 164                 $res = $this->doGetSize();
 165                 wfProfileOut( __METHOD__ );
 166                 return $res;
 167         }
 168
 169         /**
 170          * @see JobQueue::getSize()
 171          * @return integer
 172          */
 173         abstract protected function doGetSize();
 174
 175         /**
 176          * Get the number of acquired jobs (these are temporarily out of the queue).
 177          * Queue classes should use caching if they are any slower without memcached.
 178          *
 179          * If caching is used, this number might be out of date for a minute.
 180          *
 181          * @return integer
 182          * @throws MWException
 183          */
 184         final public function getAcquiredCount() {
 185                 wfProfileIn( __METHOD__ );
 186                 $res = $this->doGetAcquiredCount();
 187                 wfProfileOut( __METHOD__ );
 188                 return $res;
 189         }
 190
 191         /**
 192          * @see JobQueue::getAcquiredCount()
 193          * @return integer
 194          */
 195         abstract protected function doGetAcquiredCount();
 196
 197         /**
 198          * Push a single jobs into the queue.
 199          * This does not require $wgJobClasses to be set for the given job type.
 200          * Outside callers should use JobQueueGroup::push() instead of this function.
 201          *
 202          * @param $jobs Job|Array
 203          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 204          * @return bool Returns false on failure
 205          * @throws MWException
 206          */
 207         final public function push( $jobs, $flags = 0 ) {
 208                 return $this->batchPush( is_array( $jobs ) ? $jobs : array( $jobs ), $flags );
 209         }
 210
 211         /**
 212          * Push a batch of jobs into the queue.
 213          * This does not require $wgJobClasses to be set for the given job type.
 214          * Outside callers should use JobQueueGroup::push() instead of this function.
 215          *
 216          * @param array $jobs List of Jobs
 217          * @param $flags integer Bitfield (supports JobQueue::QoS_Atomic)
 218          * @return bool Returns false on failure
 219          * @throws MWException
 220          */
 221         final public function batchPush( array $jobs, $flags = 0 ) {
 222                 if ( !count( $jobs ) ) {
 223                         return true; // nothing to do
 224                 }
 225
 226                 foreach ( $jobs as $job ) {
 227                         if ( $job->getType() !== $this->type ) {
 228                                 throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 229                         }
 230                 }
 231
 232                 wfProfileIn( __METHOD__ );
 233                 $ok = $this->doBatchPush( $jobs, $flags );
 234                 wfProfileOut( __METHOD__ );
 235                 return $ok;
 236         }
 237
 238         /**
 239          * @see JobQueue::batchPush()
 240          * @return bool
 241          */
 242         abstract protected function doBatchPush( array $jobs, $flags );
 243
 244         /**
 245          * Pop a job off of the queue.
 246          * This requires $wgJobClasses to be set for the given job type.
 247          * Outside callers should use JobQueueGroup::pop() instead of this function.
 248          *
 249          * @return Job|bool Returns false if there are no jobs
 250          * @throws MWException
 251          */
 252         final public function pop() {
 253                 global $wgJobClasses;
 254
 255                 if ( $this->wiki !== wfWikiID() ) {
 256                         throw new MWException( "Cannot pop '{$this->type}' job off foreign wiki queue." );
 257                 } elseif ( !isset( $wgJobClasses[$this->type] ) ) {
 258                         // Do not pop jobs if there is no class for the queue type
 259                         throw new MWException( "Unrecognized job type '{$this->type}'." );
 260                 }
 261
 262                 wfProfileIn( __METHOD__ );
 263                 $job = $this->doPop();
 264                 wfProfileOut( __METHOD__ );
 265                 return $job;
 266         }
 267
 268         /**
 269          * @see JobQueue::pop()
 270          * @return Job
 271          */
 272         abstract protected function doPop();
 273
 274         /**
 275          * Acknowledge that a job was completed.
 276          *
 277          * This does nothing for certain queue classes or if "claimTTL" is not set.
 278          * Outside callers should use JobQueueGroup::ack() instead of this function.
 279          *
 280          * @param $job Job
 281          * @return bool
 282          * @throws MWException
 283          */
 284         final public function ack( Job $job ) {
 285                 if ( $job->getType() !== $this->type ) {
 286                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 287                 }
 288                 wfProfileIn( __METHOD__ );
 289                 $ok = $this->doAck( $job );
 290                 wfProfileOut( __METHOD__ );
 291                 return $ok;
 292         }
 293
 294         /**
 295          * @see JobQueue::ack()
 296          * @return bool
 297          */
 298         abstract protected function doAck( Job $job );
 299
 300         /**
 301          * Register the "root job" of a given job into the queue for de-duplication.
 302          * This should only be called right *after* all the new jobs have been inserted.
 303          * This is used to turn older, duplicate, job entries into no-ops. The root job
 304          * information will remain in the registry until it simply falls out of cache.
 305          *
 306          * This requires that $job has two special fields in the "params" array:
 307          *   - rootJobSignature : hash (e.g. SHA1) that identifies the task
 308          *   - rootJobTimestamp : TS_MW timestamp of this instance of the task
 309          *
 310          * A "root job" is a conceptual job that consist of potentially many smaller jobs
 311          * that are actually inserted into the queue. For example, "refreshLinks" jobs are
 312          * spawned when a template is edited. One can think of the task as "update links
 313          * of pages that use template X" and an instance of that task as a "root job".
 314          * However, what actually goes into the queue are potentially many refreshLinks2 jobs.
 315          * Since these jobs include things like page ID ranges and DB master positions, and morph
 316          * into smaller refreshLinks2 jobs recursively, simple duplicate detection (like job_sha1)
 317          * for individual jobs being identical is not useful.
 318          *
 319          * In the case of "refreshLinks", if these jobs are still in the queue when the template
 320          * is edited again, we want all of these old refreshLinks jobs for that template to become
 321          * no-ops. This can greatly reduce server load, since refreshLinks jobs involves parsing.
 322          * Essentially, the new batch of jobs belong to a new "root job" and the older ones to a
 323          * previous "root job" for the same task of "update links of pages that use template X".
 324          *
 325          * This does nothing for certain queue classes.
 326          *
 327          * @param $job Job
 328          * @return bool
 329          * @throws MWException
 330          */
 331         final public function deduplicateRootJob( Job $job ) {
 332                 if ( $job->getType() !== $this->type ) {
 333                         throw new MWException( "Got '{$job->getType()}' job; expected '{$this->type}'." );
 334                 }
 335                 wfProfileIn( __METHOD__ );
 336                 $ok = $this->doDeduplicateRootJob( $job );
 337                 wfProfileOut( __METHOD__ );
 338                 return $ok;
 339         }
 340
 341         /**
 342          * @see JobQueue::deduplicateRootJob()
 343          * @param $job Job
 344          * @return bool
 345          */
 346         protected function doDeduplicateRootJob( Job $job ) {
 347                 return true;
 348         }
 349
 350         /**
 351          * Wait for any slaves or backup servers to catch up.
 352          *
 353          * This does nothing for certain queue classes.
 354          *
 355          * @return void
 356          * @throws MWException
 357          */
 358         final public function waitForBackups() {
 359                 wfProfileIn( __METHOD__ );
 360                 $this->doWaitForBackups();
 361                 wfProfileOut( __METHOD__ );
 362         }
 363
 364         /**
 365          * @see JobQueue::waitForBackups()
 366          * @return void
 367          */
 368         protected function doWaitForBackups() {}
 369
 370         /**
 371          * Return a map of task names to task definition maps.
 372          * A "task" is a fast periodic queue maintenance action.
 373          * Mutually exclusive tasks must implement their own locking in the callback.
 374          *
 375          * Each task value is an associative array with:
 376          *   - name     : the name of the task
 377          *   - callback : a PHP callable that performs the task
 378          *   - period   : the period in seconds corresponding to the task frequency
 379          *
 380          * @return Array
 381          */
 382         final public function getPeriodicTasks() {
 383                 $tasks = $this->doGetPeriodicTasks();
 384                 foreach ( $tasks as $name => &$def ) {
 385                         $def['name'] = $name;
 386                 }
 387                 return $tasks;
 388         }
 389
 390         /**
 391          * @see JobQueue::getPeriodicTasks()
 392          * @return Array
 393          */
 394         protected function doGetPeriodicTasks() {
 395                 return array();
 396         }
 397
 398         /**
 399          * Clear any process and persistent caches
 400          *
 401          * @return void
 402          */
 403         final public function flushCaches() {
 404                 wfProfileIn( __METHOD__ );
 405                 $this->doFlushCaches();
 406                 wfProfileOut( __METHOD__ );
 407         }
 408
 409         /**
 410          * @see JobQueue::flushCaches()
 411          * @return void
 412          */
 413         protected function doFlushCaches() {}
 414
 415         /**
 416          * Get an iterator to traverse over all of the jobs in this queue.
 417          * This does not include jobs that are current acquired. In general,
 418          * this should only be called on a queue that is no longer being popped.
 419          *
 420          * @return Iterator|Traversable|Array
 421          * @throws MWException
 422          */
 423         abstract public function getAllQueuedJobs();
 424
 425         /**
 426          * Namespace the queue with a key to isolate it for testing
 427          *
 428          * @param $key string
 429          * @return void
 430          * @throws MWException
 431          */
 432         public function setTestingPrefix( $key ) {
 433                 throw new MWException( "Queue namespacing not supported for this queue type." );
 434         }
 435 }