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          * @param $type string
  51          *
  52          * @return Job
  53          */
  54         static function pop_type( $type ) {
  55                 wfProfilein( __METHOD__ );
  56
  57                 $dbw = wfGetDB( DB_MASTER );
  58
  59                 $dbw->begin();
  60
  61                 $row = $dbw->selectRow(
  62                         'job',
  63                         '*',
  64                         array( 'job_cmd' => $type ),
  65                         __METHOD__,
  66                         array( 'LIMIT' => 1, 'FOR UPDATE' )
  67                 );
  68
  69                 if ( $row === false ) {
  70                         $dbw->commit();
  71                         wfProfileOut( __METHOD__ );
  72                         return false;
  73                 }
  74
  75                 /* Ensure we "own" this row */
  76                 $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
  77                 $affected = $dbw->affectedRows();
  78                 $dbw->commit();
  79
  80                 if ( $affected == 0 ) {
  81                         wfProfileOut( __METHOD__ );
  82                         return false;
  83                 }
  84
  85                 wfIncrStats( 'job-pop' );
  86                 $namespace = $row->job_namespace;
  87                 $dbkey = $row->job_title;
  88                 $title = Title::makeTitleSafe( $namespace, $dbkey );
  89                 $job = Job::factory( $row->job_cmd, $title, Job::extractBlob( $row->job_params ),
  90                         $row->job_id );
  91
  92                 $job->removeDuplicates();
  93
  94                 wfProfileOut( __METHOD__ );
  95                 return $job;
  96         }
  97
  98         /**
  99          * Pop a job off the front of the queue
 100          *
 101          * @param $offset Integer: Number of jobs to skip
 102          * @return Job or false if there's no jobs
 103          */
 104         static function pop( $offset = 0 ) {
 105                 global $wgJobTypesExcludedFromDefaultQueue;
 106                 wfProfileIn( __METHOD__ );
 107
 108                 $dbr = wfGetDB( DB_SLAVE );
 109
 110                 /* Get a job from the slave, start with an offset,
 111                         scan full set afterwards, avoid hitting purged rows
 112
 113                         NB: If random fetch previously was used, offset
 114                                 will always be ahead of few entries
 115                 */
 116                 $conditions = array();
 117                 if ( count( $wgJobTypesExcludedFromDefaultQueue ) != 0 ) {
 118                         foreach ( $wgJobTypesExcludedFromDefaultQueue as $cmdType ) {
 119                                 $conditions[] = "job_cmd != " . $dbr->addQuotes( $cmdType );
 120                         }
 121                 }
 122                 $offset = intval( $offset );
 123                 $options = array( 'ORDER BY' => 'job_id', 'USE INDEX' => 'PRIMARY' );
 124
 125                 $row = $dbr->selectRow( 'job', '*',
 126                         array_merge( $conditions, array( "job_id >= $offset" ) ),
 127                         __METHOD__,
 128                         $options
 129                 );
 130
 131                 // Refetching without offset is needed as some of job IDs could have had delayed commits
 132                 // and have lower IDs than jobs already executed, blame concurrency :)
 133                 //
 134                 if ( $row === false ) {
 135                         if ( $offset != 0 ) {
 136                                 $row = $dbr->selectRow( 'job', '*', $conditions, __METHOD__, $options );
 137                         }
 138
 139                         if ( $row === false ) {
 140                                 wfProfileOut( __METHOD__ );
 141                                 return false;
 142                         }
 143                 }
 144
 145                 // Try to delete it from the master
 146                 $dbw = wfGetDB( DB_MASTER );
 147                 $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 148                 $affected = $dbw->affectedRows();
 149                 $dbw->commit();
 150
 151                 if ( !$affected ) {
 152                         // Failed, someone else beat us to it
 153                         // Try getting a random row
 154                         $row = $dbw->selectRow( 'job', array( 'MIN(job_id) as minjob',
 155                                 'MAX(job_id) as maxjob' ), '1=1', __METHOD__ );
 156                         if ( $row === false || is_null( $row->minjob ) || is_null( $row->maxjob ) ) {
 157                                 // No jobs to get
 158                                 wfProfileOut( __METHOD__ );
 159                                 return false;
 160                         }
 161                         // Get the random row
 162                         $row = $dbw->selectRow( 'job', '*',
 163                                 'job_id >= ' . mt_rand( $row->minjob, $row->maxjob ), __METHOD__ );
 164                         if ( $row === false ) {
 165                                 // Random job gone before we got the chance to select it
 166                                 // Give up
 167                                 wfProfileOut( __METHOD__ );
 168                                 return false;
 169                         }
 170                         // Delete the random row
 171                         $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 172                         $affected = $dbw->affectedRows();
 173                         $dbw->commit();
 174
 175                         if ( !$affected ) {
 176                                 // Random job gone before we exclusively deleted it
 177                                 // Give up
 178                                 wfProfileOut( __METHOD__ );
 179                                 return false;
 180                         }
 181                 }
 182
 183                 // If execution got to here, there's a row in $row that has been deleted from the database
 184                 // by this thread. Hence the concurrent pop was successful.
 185                 wfIncrStats( 'job-pop' );
 186                 $namespace = $row->job_namespace;
 187                 $dbkey = $row->job_title;
 188                 $title = Title::makeTitleSafe( $namespace, $dbkey );
 189                 $job = Job::factory( $row->job_cmd, $title, Job::extractBlob( $row->job_params ), $row->job_id );
 190
 191                 // Remove any duplicates it may have later in the queue
 192                 $job->removeDuplicates();
 193
 194                 wfProfileOut( __METHOD__ );
 195                 return $job;
 196         }
 197
 198         /**
 199          * Create the appropriate object to handle a specific job
 200          *
 201          * @param $command String: Job command
 202          * @param $title Title: Associated title
 203          * @param $params Array: Job parameters
 204          * @param $id Int: Job identifier
 205          * @return Job
 206          */
 207         static function factory( $command, $title, $params = false, $id = 0 ) {
 208                 global $wgJobClasses;
 209                 if( isset( $wgJobClasses[$command] ) ) {
 210                         $class = $wgJobClasses[$command];
 211                         return new $class( $title, $params, $id );
 212                 }
 213                 throw new MWException( "Invalid job command `{$command}`" );
 214         }
 215
 216         static function makeBlob( $params ) {
 217                 if ( $params !== false ) {
 218                         return serialize( $params );
 219                 } else {
 220                         return '';
 221                 }
 222         }
 223
 224         static function extractBlob( $blob ) {
 225                 if ( (string)$blob !== '' ) {
 226                         return unserialize( $blob );
 227                 } else {
 228                         return false;
 229                 }
 230         }
 231
 232         /**
 233          * Batch-insert a group of jobs into the queue.
 234          * This will be wrapped in a transaction with a forced commit.
 235          *
 236          * This may add duplicate at insert time, but they will be
 237          * removed later on, when the first one is popped.
 238          *
 239          * @param $jobs array of Job objects
 240          */
 241         static function batchInsert( $jobs ) {
 242                 if ( !count( $jobs ) ) {
 243                         return;
 244                 }
 245                 $dbw = wfGetDB( DB_MASTER );
 246                 $rows = array();
 247                 foreach ( $jobs as $job ) {
 248                         $rows[] = $job->insertFields();
 249                         if ( count( $rows ) >= 50 ) {
 250                                 # Do a small transaction to avoid slave lag
 251                                 $dbw->begin();
 252                                 $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 253                                 $dbw->commit();
 254                                 $rows = array();
 255                         }
 256                 }
 257                 if ( $rows ) { // last chunk
 258                         $dbw->begin();
 259                         $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 260                         $dbw->commit();
 261                 }
 262                 wfIncrStats( 'job-insert', count( $jobs ) );
 263         }
 264
 265         /**
 266          * Insert a group of jobs into the queue.
 267          *
 268          * Same as batchInsert() but does not commit and can thus
 269          * be rolled-back as part of a larger transaction. However,
 270          * large batches of jobs can cause slave lag.
 271          *
 272          * @param $jobs array of Job objects
 273          */
 274         static function safeBatchInsert( $jobs ) {
 275                 if ( !count( $jobs ) ) {
 276                         return;
 277                 }
 278                 $dbw = wfGetDB( DB_MASTER );
 279                 $rows = array();
 280                 foreach ( $jobs as $job ) {
 281                         $rows[] = $job->insertFields();
 282                         if ( count( $rows ) >= 500 ) {
 283                                 $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 284                                 $rows = array();
 285                         }
 286                 }
 287                 if ( $rows ) { // last chunk
 288                         $dbw->insert( 'job', $rows, __METHOD__, 'IGNORE' );
 289                 }
 290                 wfIncrStats( 'job-insert', count( $jobs ) );
 291         }
 292
 293         /*-------------------------------------------------------------------------
 294          * Non-static functions
 295          *------------------------------------------------------------------------*/
 296
 297         /**
 298          * @param $command
 299          * @param $title
 300          * @param $params array
 301          * @param int $id
 302          */
 303         function __construct( $command, $title, $params = false, $id = 0 ) {
 304                 $this->command = $command;
 305                 $this->title = $title;
 306                 $this->params = $params;
 307                 $this->id = $id;
 308
 309                 // A bit of premature generalisation
 310                 // Oh well, the whole class is premature generalisation really
 311                 $this->removeDuplicates = true;
 312         }
 313
 314         /**
 315          * Insert a single job into the queue.
 316          * @return bool true on success
 317          */
 318         function insert() {
 319                 $fields = $this->insertFields();
 320
 321                 $dbw = wfGetDB( DB_MASTER );
 322
 323                 if ( $this->removeDuplicates ) {
 324                         $res = $dbw->select( 'job', array( '1' ), $fields, __METHOD__ );
 325                         if ( $dbw->numRows( $res ) ) {
 326                                 return;
 327                         }
 328                 }
 329                 wfIncrStats( 'job-insert' );
 330                 return $dbw->insert( 'job', $fields, __METHOD__ );
 331         }
 332
 333         protected function insertFields() {
 334                 $dbw = wfGetDB( DB_MASTER );
 335                 return array(
 336                         'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 337                         'job_cmd' => $this->command,
 338                         'job_namespace' => $this->title->getNamespace(),
 339                         'job_title' => $this->title->getDBkey(),
 340                         'job_params' => Job::makeBlob( $this->params )
 341                 );
 342         }
 343
 344         /**
 345          * Remove jobs in the job queue which are duplicates of this job.
 346          * This is deadlock-prone and so starts its own transaction.
 347          */
 348         function removeDuplicates() {
 349                 if ( !$this->removeDuplicates ) {
 350                         return;
 351                 }
 352
 353                 $fields = $this->insertFields();
 354                 unset( $fields['job_id'] );
 355                 $dbw = wfGetDB( DB_MASTER );
 356                 $dbw->begin();
 357                 $dbw->delete( 'job', $fields, __METHOD__ );
 358                 $affected = $dbw->affectedRows();
 359                 $dbw->commit();
 360                 if ( $affected ) {
 361                         wfIncrStats( 'job-dup-delete', $affected );
 362                 }
 363         }
 364
 365         function toString() {
 366                 $paramString = '';
 367                 if ( $this->params ) {
 368                         foreach ( $this->params as $key => $value ) {
 369                                 if ( $paramString != '' ) {
 370                                         $paramString .= ' ';
 371                                 }
 372                                 $paramString .= "$key=$value";
 373                         }
 374                 }
 375
 376                 if ( is_object( $this->title ) ) {
 377                         $s = "{$this->command} " . $this->title->getPrefixedDBkey();
 378                         if ( $paramString !== '' ) {
 379                                 $s .= ' ' . $paramString;
 380                         }
 381                         return $s;
 382                 } else {
 383                         return "{$this->command} $paramString";
 384                 }
 385         }
 386
 387         protected function setLastError( $error ) {
 388                 $this->error = $error;
 389         }
 390
 391         function getLastError() {
 392                 return $this->error;
 393         }
 394 }