includes/job/JobQueue.php

   1 <?php
   2 /**
   3  * Job queue base code
   4  *
   5  * @file
   6  * @defgroup JobQueue JobQueue
   7  */
   8
   9 if ( !defined( 'MEDIAWIKI' ) ) {
  10         die( "This file is part of MediaWiki, it is not a valid entry point\n" );
  11 }
  12
  13 /**
  14  * Class to both describe a background job and handle jobs.
  15  *
  16  * @ingroup JobQueue
  17  */
  18 abstract class Job {
  19
  20         /**
  21          * @var Title
  22          */
  23         var $title;
  24
  25         var $command,
  26                 $params,
  27                 $id,
  28                 $removeDuplicates,
  29                 $error;
  30
  31         /*-------------------------------------------------------------------------
  32          * Abstract functions
  33          *------------------------------------------------------------------------*/
  34
  35         /**
  36          * Run the job
  37          * @return boolean success
  38          */
  39         abstract function run();
  40
  41         /*-------------------------------------------------------------------------
  42          * Static functions
  43          *------------------------------------------------------------------------*/
  44
  45         /**
  46          * Pop a job of a certain type.  This tries less hard than pop() to
  47          * actually find a job; it may be adversely affected by concurrent job
  48          * runners.
  49          */
  50         static function pop_type( $type ) {
  51                 wfProfilein( __METHOD__ );
  52
  53                 $dbw = wfGetDB( DB_MASTER );
  54
  55                 $row = $dbw->selectRow(
  56                         'job',
  57                         '*',
  58                         array( 'job_cmd' => $type ),
  59                         __METHOD__,
  60                         array( 'LIMIT' => 1 )
  61                 );
  62
  63                 if ( $row === false ) {
  64                         wfProfileOut( __METHOD__ );
  65                         return false;
  66                 }
  67
  68                 /* Ensure we "own" this row */
  69                 $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
  70                 $affected = $dbw->affectedRows();
  71                 $dbw->commit();
  72
  73                 if ( $affected == 0 ) {
  74                         wfProfileOut( __METHOD__ );
  75                         return false;
  76                 }
  77
  78                 wfIncrStats( 'job-pop' );
  79                 $namespace = $row->job_namespace;
  80                 $dbkey = $row->job_title;
  81                 $title = Title::makeTitleSafe( $namespace, $dbkey );
  82                 $job = Job::factory( $row->job_cmd, $title, Job::extractBlob( $row->job_params ),
  83                         $row->job_id );
  84
  85                 $job->removeDuplicates();
  86
  87                 wfProfileOut( __METHOD__ );
  88                 return $job;
  89         }
  90
  91         /**
  92          * Pop a job off the front of the queue
  93          *
  94          * @param $offset Integer: Number of jobs to skip
  95          * @return Job or false if there's no jobs
  96          */
  97         static function pop( $offset = 0 ) {
  98                 wfProfileIn( __METHOD__ );
  99
 100                 $dbr = wfGetDB( DB_SLAVE );
 101
 102                 /* Get a job from the slave, start with an offset,
 103                         scan full set afterwards, avoid hitting purged rows
 104
 105                         NB: If random fetch previously was used, offset
 106                                 will always be ahead of few entries
 107                 */
 108
 109                 $row = $dbr->selectRow( 'job', '*', "job_id >= ${offset}", __METHOD__,
 110                         array( 'ORDER BY' => 'job_id', 'LIMIT' => 1 ) );
 111
 112                 // Refetching without offset is needed as some of job IDs could have had delayed commits
 113                 // and have lower IDs than jobs already executed, blame concurrency :)
 114                 //
 115                 if ( $row === false ) {
 116                         if ( $offset != 0 ) {
 117                                 $row = $dbr->selectRow( 'job', '*', '', __METHOD__,
 118                                         array( 'ORDER BY' => 'job_id', 'LIMIT' => 1 ) );
 119                         }
 120
 121                         if ( $row === false ) {
 122                                 wfProfileOut( __METHOD__ );
 123                                 return false;
 124                         }
 125                 }
 126
 127                 // Try to delete it from the master
 128                 $dbw = wfGetDB( DB_MASTER );
 129                 $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 130                 $affected = $dbw->affectedRows();
 131                 $dbw->commit();
 132
 133                 if ( !$affected ) {
 134                         // Failed, someone else beat us to it
 135                         // Try getting a random row
 136                         $row = $dbw->selectRow( 'job', array( 'MIN(job_id) as minjob',
 137                                 'MAX(job_id) as maxjob' ), '1=1', __METHOD__ );
 138                         if ( $row === false || is_null( $row->minjob ) || is_null( $row->maxjob ) ) {
 139                                 // No jobs to get
 140                                 wfProfileOut( __METHOD__ );
 141                                 return false;
 142                         }
 143                         // Get the random row
 144                         $row = $dbw->selectRow( 'job', '*',
 145                                 'job_id >= ' . mt_rand( $row->minjob, $row->maxjob ), __METHOD__ );
 146                         if ( $row === false ) {
 147                                 // Random job gone before we got the chance to select it
 148                                 // Give up
 149                                 wfProfileOut( __METHOD__ );
 150                                 return false;
 151                         }
 152                         // Delete the random row
 153                         $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 154                         $affected = $dbw->affectedRows();
 155                         $dbw->commit();
 156
 157                         if ( !$affected ) {
 158                                 // Random job gone before we exclusively deleted it
 159                                 // Give up
 160                                 wfProfileOut( __METHOD__ );
 161                                 return false;
 162                         }
 163                 }
 164
 165                 // If execution got to here, there's a row in $row that has been deleted from the database
 166                 // by this thread. Hence the concurrent pop was successful.
 167                 wfIncrStats( 'job-pop' );
 168                 $namespace = $row->job_namespace;
 169                 $dbkey = $row->job_title;
 170                 $title = Title::makeTitleSafe( $namespace, $dbkey );
 171                 $job = Job::factory( $row->job_cmd, $title, Job::extractBlob( $row->job_params ), $row->job_id );
 172
 173                 // Remove any duplicates it may have later in the queue
 174                 $job->removeDuplicates();
 175
 176                 wfProfileOut( __METHOD__ );
 177                 return $job;
 178         }
 179
 180         /**
 181          * Create the appropriate object to handle a specific job
 182          *
 183          * @param $command String: Job command
 184          * @param $title Title: Associated title
 185          * @param $params Array: Job parameters
 186          * @param $id Int: Job identifier
 187          * @return Job
 188          */
 189         static function factory( $command, $title, $params = false, $id = 0 ) {
 190                 global $wgJobClasses;
 191                 if( isset( $wgJobClasses[$command] ) ) {
 192                         $class = $wgJobClasses[$command];
 193                         return new $class( $title, $params, $id );
 194                 }
 195                 throw new MWException( "Invalid job command `{$command}`" );
 196         }
 197
 198         static function makeBlob( $params ) {
 199                 if ( $params !== false ) {
 200                         return serialize( $params );
 201                 } else {
 202                         return '';
 203                 }
 204         }
 205
 206         static function extractBlob( $blob ) {
 207                 if ( (string)$blob !== '' ) {
 208                         return unserialize( $blob );
 209                 } else {
 210                         return false;
 211                 }
 212         }
 213
 214         /**
 215          * Batch-insert a group of jobs into the queue.
 216          * This will be wrapped in a transaction with a forced commit.
 217          *
 218          * This may add duplicate at insert time, but they will be
 219          * removed later on, when the first one is popped.
 220          *
 221          * @param $jobs array of Job objects
 222          */
 223         static function batchInsert( $jobs ) {
 224                 if ( !count( $jobs ) ) {
 225                         return;
 226                 }
 227                 $dbw = wfGetDB( DB_MASTER );
 228                 $rows = array();
 229                 foreach ( $jobs as $job ) {
 230                         $rows[] = $job->insertFields();
 231                         if ( count( $rows ) >= 50 ) {
 232                                 # Do a small transaction to avoid slave lag
 233                                 $dbw->begin();
 234                                 $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 235                                 $dbw->commit();
 236                                 $rows = array();
 237                         }
 238                 }
 239                 if ( $rows ) { // last chunk
 240                         $dbw->begin();
 241                         $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 242                         $dbw->commit();
 243                 }
 244                 wfIncrStats( 'job-insert', count( $jobs ) );
 245         }
 246
 247         /**
 248          * Insert a group of jobs into the queue.
 249          *
 250          * Same as batchInsert() but does not commit and can thus
 251          * be rolled-back as part of a larger transaction. However,
 252          * large batches of jobs can cause slave lag.
 253          *
 254          * @param $jobs array of Job objects
 255          */
 256         static function safeBatchInsert( $jobs ) {
 257                 if ( !count( $jobs ) ) {
 258                         return;
 259                 }
 260                 $dbw = wfGetDB( DB_MASTER );
 261                 $rows = array();
 262                 foreach ( $jobs as $job ) {
 263                         $rows[] = $job->insertFields();
 264                         if ( count( $rows ) >= 500 ) {
 265                                 $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 266                                 $rows = array();
 267                         }
 268                 }
 269                 if ( $rows ) { // last chunk
 270                         $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 271                 }
 272                 wfIncrStats( 'job-insert', count( $jobs ) );
 273         }
 274
 275         /*-------------------------------------------------------------------------
 276          * Non-static functions
 277          *------------------------------------------------------------------------*/
 278
 279         /**
 280          * @param $command
 281          * @param $title
 282          * @param $params array
 283          * @param int $id
 284          */
 285         function __construct( $command, $title, $params = false, $id = 0 ) {
 286                 $this->command = $command;
 287                 $this->title = $title;
 288                 $this->params = $params;
 289                 $this->id = $id;
 290
 291                 // A bit of premature generalisation
 292                 // Oh well, the whole class is premature generalisation really
 293                 $this->removeDuplicates = true;
 294         }
 295
 296         /**
 297          * Insert a single job into the queue.
 298          * @return bool true on success
 299          */
 300         function insert() {
 301                 $fields = $this->insertFields();
 302
 303                 $dbw = wfGetDB( DB_MASTER );
 304
 305                 if ( $this->removeDuplicates ) {
 306                         $res = $dbw->select( 'job', array( '1' ), $fields, __METHOD__ );
 307                         if ( $dbw->numRows( $res ) ) {
 308                                 return;
 309                         }
 310                 }
 311                 wfIncrStats( 'job-insert' );
 312                 return $dbw->insert( 'job', $fields, __METHOD__ );
 313         }
 314
 315         protected function insertFields() {
 316                 $dbw = wfGetDB( DB_MASTER );
 317                 return array(
 318                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 319                         'job_cmd' => $this->command,
 320                         'job_namespace' => $this->title->getNamespace(),
 321                         'job_title' => $this->title->getDBkey(),
 322                         'job_params' => Job::makeBlob( $this->params )
 323                 );
 324         }
 325
 326         /**
 327          * Remove jobs in the job queue which are duplicates of this job.
 328          * This is deadlock-prone and so starts its own transaction.
 329          */
 330         function removeDuplicates() {
 331                 if ( !$this->removeDuplicates ) {
 332                         return;
 333                 }
 334
 335                 $fields = $this->insertFields();
 336                 unset( $fields['job_id'] );
 337                 $dbw = wfGetDB( DB_MASTER );
 338                 $dbw->begin();
 339                 $dbw->delete( 'job', $fields, __METHOD__ );
 340                 $affected = $dbw->affectedRows();
 341                 $dbw->commit();
 342                 if ( $affected ) {
 343                         wfIncrStats( 'job-dup-delete', $affected );
 344                 }
 345         }
 346
 347         function toString() {
 348                 $paramString = '';
 349                 if ( $this->params ) {
 350                         foreach ( $this->params as $key => $value ) {
 351                                 if ( $paramString != '' ) {
 352                                         $paramString .= ' ';
 353                                 }
 354                                 $paramString .= "$key=$value";
 355                         }
 356                 }
 357
 358                 if ( is_object( $this->title ) ) {
 359                         $s = "{$this->command} " . $this->title->getPrefixedDBkey();
 360                         if ( $paramString !== '' ) {
 361                                 $s .= ' ' . $paramString;
 362                         }
 363                         return $s;
 364                 } else {
 365                         return "{$this->command} $paramString";
 366                 }
 367         }
 368
 369         protected function setLastError( $error ) {
 370                 $this->error = $error;
 371         }
 372
 373         function getLastError() {
 374                 return $this->error;
 375         }
 376 }