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