includes/job/JobQueueDB.php

   1 <?php
   2 /**
   3  * Database-backed job queue 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  * @author Aaron Schulz
  22  */
  23
  24 /**
  25  * Class to handle job queues stored in the DB
  26  *
  27  * @ingroup JobQueue
  28  * @since 1.21
  29  */
  30 class JobQueueDB extends JobQueue {
  31         const CACHE_TTL      = 300; // integer; seconds to cache queue information
  32         const MAX_AGE_PRUNE  = 604800; // integer; seconds a job can live once claimed
  33         const MAX_ATTEMPTS   = 3; // integer; number of times to try a job
  34         const MAX_JOB_RANDOM = 2147483647; // integer; 2^31 - 1, used for job_random
  35
  36         /**
  37          * @see JobQueue::doIsEmpty()
  38          * @return bool
  39          */
  40         protected function doIsEmpty() {
  41                 global $wgMemc;
  42
  43                 $key = $this->getEmptinessCacheKey();
  44
  45                 $isEmpty = $wgMemc->get( $key );
  46                 if ( $isEmpty === 'true' ) {
  47                         return true;
  48                 } elseif ( $isEmpty === 'false' ) {
  49                         return false;
  50                 }
  51
  52                 $found = $this->getSlaveDB()->selectField(
  53                         'job', '1', array( 'job_cmd' => $this->type ), __METHOD__
  54                 );
  55
  56                 $wgMemc->add( $key, $found ? 'false' : 'true', self::CACHE_TTL );
  57         }
  58
  59         /**
  60          * @see JobQueue::doBatchPush()
  61          * @return bool
  62          */
  63         protected function doBatchPush( array $jobs, $flags ) {
  64                 if ( count( $jobs ) ) {
  65                         $dbw = $this->getMasterDB();
  66
  67                         $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
  68                         $rowList = array(); // list of jobs for jobs that are are not de-duplicated
  69
  70                         foreach ( $jobs as $job ) {
  71                                 $row = $this->insertFields( $job );
  72                                 if ( $job->ignoreDuplicates() ) {
  73                                         $rowSet[$row['job_sha1']] = $row;
  74                                 } else {
  75                                         $rowList[] = $row;
  76                                 }
  77                         }
  78
  79                         $atomic = ( $flags & self::QoS_Atomic );
  80                         $key    = $this->getEmptinessCacheKey();
  81                         $ttl    = self::CACHE_TTL;
  82
  83                         $dbw->onTransactionIdle(
  84                                 function() use ( $dbw, $rowSet, $rowList, $atomic, $key, $ttl
  85                         ) {
  86                                 global $wgMemc;
  87
  88                                 $autoTrx = $dbw->getFlag( DBO_TRX ); // automatic begin() enabled?
  89                                 if ( $atomic ) {
  90                                         $dbw->begin( __METHOD__ ); // wrap all the job additions in one transaction
  91                                 } else {
  92                                         $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
  93                                 }
  94                                 try {
  95                                         // Strip out any duplicate jobs that are already in the queue...
  96                                         if ( count( $rowSet ) ) {
  97                                                 $res = $dbw->select( 'job', 'job_sha1',
  98                                                         array(
  99                                                                 // No job_type condition since it's part of the job_sha1 hash
 100                                                                 'job_sha1'  => array_keys( $rowSet ),
 101                                                                 'job_token' => '' // unclaimed
 102                                                         ),
 103                                                         __METHOD__
 104                                                 );
 105                                                 foreach ( $res as $row ) {
 106                                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate." );
 107                                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 108                                                 }
 109                                         }
 110                                         // Build the full list of job rows to insert
 111                                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 112                                         // Insert the job rows in chunks to avoid slave lag...
 113                                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 114                                                 $dbw->insert( 'job', $rowBatch, __METHOD__ );
 115                                         }
 116                                 } catch ( DBError $e ) {
 117                                         if ( $atomic ) {
 118                                                 $dbw->rollback( __METHOD__ );
 119                                         } else {
 120                                                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore automatic begin()
 121                                         }
 122                                         throw $e;
 123                                 }
 124                                 if ( $atomic ) {
 125                                         $dbw->commit( __METHOD__ );
 126                                 } else {
 127                                         $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore automatic begin()
 128                                 }
 129
 130                                 $wgMemc->set( $key, 'false', $ttl ); // queue is not empty
 131                         } );
 132                 }
 133
 134                 return true;
 135         }
 136
 137         /**
 138          * @see JobQueue::doPop()
 139          * @return Job|bool
 140          */
 141         protected function doPop() {
 142                 global $wgMemc;
 143
 144                 if ( $wgMemc->get( $this->getEmptinessCacheKey() ) === 'true' ) {
 145                         return false; // queue is empty
 146                 }
 147
 148                 $dbw = $this->getMasterDB();
 149                 $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 150
 151                 $uuid = wfRandomString( 32 ); // pop attempt
 152                 $job = false; // job popped off
 153                 $autoTrx = $dbw->getFlag( DBO_TRX ); // automatic begin() enabled?
 154                 $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 155                 try {
 156                         // Occasionally recycle jobs back into the queue that have been claimed too long
 157                         if ( mt_rand( 0, 99 ) == 0 ) {
 158                                 $this->recycleStaleJobs();
 159                         }
 160                         do { // retry when our row is invalid or deleted as a duplicate
 161                                 // Try to reserve a row in the DB...
 162                                 if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 163                                         $row = $this->claimOldest( $uuid );
 164                                 } else { // random first
 165                                         $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 166                                         $gte  = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 167                                         $row  = $this->claimRandom( $uuid, $rand, $gte );
 168                                         if ( !$row ) { // need to try the other direction
 169                                                 $row = $this->claimRandom( $uuid, $rand, !$gte );
 170                                         }
 171                                 }
 172                                 // Check if we found a row to reserve...
 173                                 if ( !$row ) {
 174                                         $wgMemc->set( $this->getEmptinessCacheKey(), 'true', self::CACHE_TTL );
 175                                         break; // nothing to do
 176                                 }
 177                                 // Get the job object from the row...
 178                                 $title = Title::makeTitleSafe( $row->job_namespace, $row->job_title );
 179                                 if ( !$title ) {
 180                                         $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 181                                         wfIncrStats( 'job-pop' );
 182                                         wfDebugLog( 'JobQueueDB', "Row has invalid title '{$row->job_title}'." );
 183                                         continue; // try again
 184                                 }
 185                                 $job = Job::factory( $row->job_cmd, $title,
 186                                         self::extractBlob( $row->job_params ), $row->job_id );
 187                                 // Flag this job as an old duplicate based on its "root" job...
 188                                 if ( $this->isRootJobOldDuplicate( $job ) ) {
 189                                         $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 190                                 }
 191                                 break; // done
 192                         } while( true );
 193                 } catch ( DBError $e ) {
 194                         $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore automatic begin()
 195                         throw $e;
 196                 }
 197                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore automatic begin()
 198
 199                 return $job;
 200         }
 201
 202         /**
 203          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 204          *
 205          * @param $uuid string 32 char hex string
 206          * @param $rand integer Random unsigned integer (31 bits)
 207          * @param $gte bool Search for job_random >= $random (otherwise job_random <= $random)
 208          * @return Row|false
 209          */
 210         protected function claimRandom( $uuid, $rand, $gte ) {
 211                 $dbw  = $this->getMasterDB();
 212                 $dir  = $gte ? 'ASC' : 'DESC';
 213                 $ineq = $gte ? '>=' : '<=';
 214
 215                 $row = false; // the row acquired
 216                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 217                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 218                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 219                 // be used here with MySQL.
 220                 do {
 221                         $row = $dbw->selectRow( 'job', '*', // find a random job
 222                                 array(
 223                                         'job_cmd'   => $this->type,
 224                                         'job_token' => '',
 225                                         "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 226                                 __METHOD__,
 227                                 array( 'ORDER BY' => "job_random {$dir}" )
 228                         );
 229                         if ( $row ) { // claim the job
 230                                 $dbw->update( 'job', // update by PK
 231                                         array(
 232                                                 'job_token'           => $uuid,
 233                                                 'job_token_timestamp' => $dbw->timestamp(),
 234                                                 'job_attempts = job_attempts+1' ),
 235                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 236                                         __METHOD__
 237                                 );
 238                                 // This might get raced out by another runner when claiming the previously
 239                                 // selected row. The use of job_random should minimize this problem, however.
 240                                 if ( !$dbw->affectedRows() ) {
 241                                         $row = false; // raced out
 242                                 }
 243                         } else {
 244                                 break; // nothing to do
 245                         }
 246                 } while ( !$row );
 247
 248                 return $row;
 249         }
 250
 251         /**
 252          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 253          *
 254          * @param $uuid string 32 char hex string
 255          * @return Row|false
 256          */
 257         protected function claimOldest( $uuid ) {
 258                 $dbw  = $this->getMasterDB();
 259
 260                 $row = false; // the row acquired
 261                 do {
 262                         if ( $dbw->getType() === 'mysql' ) {
 263                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 264                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 265                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 266                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 267                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 268                                         "SET " .
 269                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 270                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 271                                                 "job_attempts = job_attempts+1 " .
 272                                         "WHERE ( " .
 273                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 274                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 275                                         ") ORDER BY job_id ASC LIMIT 1",
 276                                         __METHOD__
 277                                 );
 278                         } else {
 279                                 // Use a subquery to find the job, within an UPDATE to claim it.
 280                                 // This uses as much of the DB wrapper functions as possible.
 281                                 $dbw->update( 'job',
 282                                         array(
 283                                                 'job_token'           => $uuid,
 284                                                 'job_token_timestamp' => $dbw->timestamp(),
 285                                                 'job_attempts = job_attempts+1' ),
 286                                         array( 'job_id = (' .
 287                                                 $dbw->selectSQLText( 'job', 'job_id',
 288                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 289                                                         __METHOD__,
 290                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 291                                                 ')'
 292                                         ),
 293                                         __METHOD__
 294                                 );
 295                         }
 296                         // Fetch any row that we just reserved...
 297                         if ( $dbw->affectedRows() ) {
 298                                 $row = $dbw->selectRow( 'job', '*',
 299                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 300                                 );
 301                                 if ( !$row ) { // raced out by duplicate job removal
 302                                         wfDebugLog( 'JobQueueDB', "Row deleted as duplicate by another process." );
 303                                 }
 304                         } else {
 305                                 break; // nothing to do
 306                         }
 307                 } while ( !$row );
 308
 309                 return $row;
 310         }
 311
 312         /**
 313          * Recycle or destroy any jobs that have been claimed for too long
 314          *
 315          * @return integer Number of jobs recycled/deleted
 316          */
 317         protected function recycleStaleJobs() {
 318                 $now   = time();
 319                 $dbw   = $this->getMasterDB();
 320                 $count = 0; // affected rows
 321
 322                 if ( $this->claimTTL > 0 ) { // re-try stale jobs...
 323                         $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 324                         // Reset job_token for these jobs so that other runners will pick them up.
 325                         // Set the timestamp to the current time, as it is useful to now that the job
 326                         // was already tried before.
 327                         $dbw->update( 'job',
 328                                 array(
 329                                         'job_token' => '',
 330                                         'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 331                                 array(
 332                                         'job_cmd' => $this->type,
 333                                         "job_token != {$dbw->addQuotes( '' )}", // was acquired
 334                                         "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 335                                         "job_attempts < {$dbw->addQuotes( self::MAX_ATTEMPTS )}" ),
 336                                 __METHOD__
 337                         );
 338                         $count += $dbw->affectedRows();
 339                 }
 340
 341                 // Just destroy stale jobs...
 342                 $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 343                 $conds = array(
 344                         'job_cmd' => $this->type,
 345                         "job_token != {$dbw->addQuotes( '' )}", // was acquired
 346                         "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 347                 );
 348                 if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 349                         $conds[] = "job_attempts >= {$dbw->addQuotes( self::MAX_ATTEMPTS )}";
 350                 }
 351                 $dbw->delete( 'job', $conds, __METHOD__ );
 352                 $count += $dbw->affectedRows();
 353
 354                 return $count;
 355         }
 356
 357         /**
 358          * @see JobQueue::doAck()
 359          * @return Job|bool
 360          */
 361         protected function doAck( Job $job ) {
 362                 $dbw = $this->getMasterDB();
 363                 $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 364
 365                 $autoTrx = $dbw->getFlag( DBO_TRX ); // automatic begin() enabled?
 366                 $dbw->clearFlag( DBO_TRX ); // make each query its own transaction
 367                 try {
 368                         // Delete a row with a single DELETE without holding row locks over RTTs...
 369                         $dbw->delete( 'job', array( 'job_cmd' => $this->type, 'job_id' => $job->getId() ) );
 370                 } catch ( Exception $e ) {
 371                         $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore automatic begin()
 372                         throw $e;
 373                 }
 374                 $dbw->setFlag( $autoTrx ? DBO_TRX : 0 ); // restore automatic begin()
 375
 376                 return true;
 377         }
 378
 379         /**
 380          * @see JobQueue::doDeduplicateRootJob()
 381          * @return bool
 382          */
 383         protected function doDeduplicateRootJob( Job $job ) {
 384                 $params = $job->getParams();
 385                 if ( !isset( $params['rootJobSignature'] ) ) {
 386                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 387                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 388                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 389                 }
 390                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 391                 // Callers should call batchInsert() and then this function so that if the insert
 392                 // fails, the de-duplication registration will be aborted. Since the insert is
 393                 // deferred till "transaction idle", do that same here, so that the ordering is
 394                 // maintained. Having only the de-duplication registration succeed would cause
 395                 // jobs to become no-ops without any actual jobs that made them redundant.
 396                 $this->getMasterDB()->onTransactionIdle( function() use ( $params, $key ) {
 397                         global $wgMemc;
 398
 399                         $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 400                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 401                                 return true; // a newer version of this root job was enqueued
 402                         }
 403
 404                         // Update the timestamp of the last root job started at the location...
 405                         return $wgMemc->set( $key, $params['rootJobTimestamp'], 14*86400 ); // 2 weeks
 406                 } );
 407
 408                 return true;
 409         }
 410
 411         /**
 412          * Check if the "root" job of a given job has been superseded by a newer one
 413          *
 414          * @param $job Job
 415          * @return bool
 416          */
 417         protected function isRootJobOldDuplicate( Job $job ) {
 418                 global $wgMemc;
 419
 420                 $params = $job->getParams();
 421                 if ( !isset( $params['rootJobSignature'] ) ) {
 422                         return false; // job has no de-deplication info
 423                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 424                         trigger_error( "Cannot check root job; missing 'rootJobTimestamp'." );
 425                         return false;
 426                 }
 427
 428                 // Get the last time this root job was enqueued
 429                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 430
 431                 // Check if a new root job was started at the location after this one's...
 432                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 433         }
 434
 435         /**
 436          * @see JobQueue::doWaitForBackups()
 437          * @return void
 438          */
 439         protected function doWaitForBackups() {
 440                 wfWaitForSlaves();
 441         }
 442
 443         /**
 444          * @return DatabaseBase
 445          */
 446         protected function getSlaveDB() {
 447                 return wfGetDB( DB_SLAVE, array(), $this->wiki );
 448         }
 449
 450         /**
 451          * @return DatabaseBase
 452          */
 453         protected function getMasterDB() {
 454                 return wfGetDB( DB_MASTER, array(), $this->wiki );
 455         }
 456
 457         /**
 458          * @param $job Job
 459          * @return array
 460          */
 461         protected function insertFields( Job $job ) {
 462                 $dbw = $this->getMasterDB();
 463                 return array(
 464                         // Fields that describe the nature of the job
 465                         'job_cmd'       => $job->getType(),
 466                         'job_namespace' => $job->getTitle()->getNamespace(),
 467                         'job_title'     => $job->getTitle()->getDBkey(),
 468                         'job_params'    => self::makeBlob( $job->getParams() ),
 469                         // Additional job metadata
 470                         'job_id'        => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 471                         'job_timestamp' => $dbw->timestamp(),
 472                         'job_sha1'      => wfBaseConvert(
 473                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 474                                 16, 36, 31
 475                         ),
 476                         'job_random'    => mt_rand( 0, self::MAX_JOB_RANDOM )
 477                 );
 478         }
 479
 480         /**
 481          * @return string
 482          */
 483         private function getEmptinessCacheKey() {
 484                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 485                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'isempty' );
 486         }
 487
 488         /**
 489          * @param string $signature Hash identifier of the root job
 490          * @return string
 491          */
 492         private function getRootJobCacheKey( $signature ) {
 493                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 494                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 495         }
 496
 497         /**
 498          * @param $params
 499          * @return string
 500          */
 501         protected static function makeBlob( $params ) {
 502                 if ( $params !== false ) {
 503                         return serialize( $params );
 504                 } else {
 505                         return '';
 506                 }
 507         }
 508
 509         /**
 510          * @param $blob
 511          * @return bool|mixed
 512          */
 513         protected static function extractBlob( $blob ) {
 514                 if ( (string)$blob !== '' ) {
 515                         return unserialize( $blob );
 516                 } else {
 517                         return false;
 518                 }
 519         }
 520 }