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, $found ? 'false' : 'true', self::CACHE_TTL );
  57                 return (bool)$found;
  58         }
  59
  60         /**
  61          * @see JobQueue::doBatchPush()
  62          * @return bool
  63          */
  64         protected function doBatchPush( array $jobs, $flags ) {
  65                 if ( count( $jobs ) ) {
  66                         $dbw = $this->getMasterDB();
  67
  68                         $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
  69                         $rowList = array(); // list of jobs for jobs that are are not de-duplicated
  70
  71                         foreach ( $jobs as $job ) {
  72                                 $row = $this->insertFields( $job );
  73                                 if ( $job->ignoreDuplicates() ) {
  74                                         $rowSet[$row['job_sha1']] = $row;
  75                                 } else {
  76                                         $rowList[] = $row;
  77                                 }
  78                         }
  79
  80                         $atomic = ( $flags & self::QoS_Atomic );
  81                         $key    = $this->getEmptinessCacheKey();
  82                         $ttl    = self::CACHE_TTL;
  83
  84                         $dbw->onTransactionIdle(
  85                                 function() use ( $dbw, $rowSet, $rowList, $atomic, $key, $ttl
  86                         ) {
  87                                 global $wgMemc;
  88
  89                                 if ( $atomic ) {
  90                                         $dbw->begin( __METHOD__ ); // wrap all the job additions in one transaction
  91                                 }
  92                                 try {
  93                                         // Strip out any duplicate jobs that are already in the queue...
  94                                         if ( count( $rowSet ) ) {
  95                                                 $res = $dbw->select( 'job', 'job_sha1',
  96                                                         array(
  97                                                                 // No job_type condition since it's part of the job_sha1 hash
  98                                                                 'job_sha1'  => array_keys( $rowSet ),
  99                                                                 'job_token' => '' // unclaimed
 100                                                         ),
 101                                                         __METHOD__
 102                                                 );
 103                                                 foreach ( $res as $row ) {
 104                                                         wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate." );
 105                                                         unset( $rowSet[$row->job_sha1] ); // already enqueued
 106                                                 }
 107                                         }
 108                                         // Build the full list of job rows to insert
 109                                         $rows = array_merge( $rowList, array_values( $rowSet ) );
 110                                         // Insert the job rows in chunks to avoid slave lag...
 111                                         foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
 112                                                 $dbw->insert( 'job', $rowBatch, __METHOD__ );
 113                                         }
 114                                 } catch ( DBError $e ) {
 115                                         if ( $atomic ) {
 116                                                 $dbw->rollback( __METHOD__ );
 117                                         }
 118                                         throw $e;
 119                                 }
 120                                 if ( $atomic ) {
 121                                         $dbw->commit( __METHOD__ );
 122                                 }
 123
 124                                 $wgMemc->set( $key, 'false', $ttl ); // queue is not empty
 125                         } );
 126                 }
 127
 128                 return true;
 129         }
 130
 131         /**
 132          * @see JobQueue::doPop()
 133          * @return Job|bool
 134          */
 135         protected function doPop() {
 136                 global $wgMemc;
 137
 138                 if ( $wgMemc->get( $this->getEmptinessCacheKey() ) === 'true' ) {
 139                         return false; // queue is empty
 140                 }
 141
 142                 $dbw = $this->getMasterDB();
 143                 $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 144
 145                 $uuid = wfRandomString( 32 ); // pop attempt
 146                 $job = false; // job popped off
 147                 // Occasionally recycle jobs back into the queue that have been claimed too long
 148                 if ( mt_rand( 0, 99 ) == 0 ) {
 149                         $this->recycleStaleJobs();
 150                 }
 151                 do { // retry when our row is invalid or deleted as a duplicate
 152                         // Try to reserve a row in the DB...
 153                         if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 154                                 $row = $this->claimOldest( $uuid );
 155                         } else { // random first
 156                                 $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 157                                 $gte  = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 158                                 $row  = $this->claimRandom( $uuid, $rand, $gte );
 159                                 if ( !$row ) { // need to try the other direction
 160                                         $row = $this->claimRandom( $uuid, $rand, !$gte );
 161                                 }
 162                         }
 163                         // Check if we found a row to reserve...
 164                         if ( !$row ) {
 165                                 $wgMemc->set( $this->getEmptinessCacheKey(), 'true', self::CACHE_TTL );
 166                                 break; // nothing to do
 167                         }
 168                         // Get the job object from the row...
 169                         $title = Title::makeTitleSafe( $row->job_namespace, $row->job_title );
 170                         if ( !$title ) {
 171                                 $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 172                                 wfIncrStats( 'job-pop' );
 173                                 wfDebugLog( 'JobQueueDB', "Row has invalid title '{$row->job_title}'." );
 174                                 continue; // try again
 175                         }
 176                         $job = Job::factory( $row->job_cmd, $title,
 177                                 self::extractBlob( $row->job_params ), $row->job_id );
 178                         $job->id = $row->job_id; // XXX: work around broken subclasses
 179                         // Flag this job as an old duplicate based on its "root" job...
 180                         if ( $this->isRootJobOldDuplicate( $job ) ) {
 181                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 182                         }
 183                         break; // done
 184                 } while( true );
 185
 186                 return $job;
 187         }
 188
 189         /**
 190          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 191          *
 192          * @param $uuid string 32 char hex string
 193          * @param $rand integer Random unsigned integer (31 bits)
 194          * @param $gte bool Search for job_random >= $random (otherwise job_random <= $random)
 195          * @return Row|false
 196          */
 197         protected function claimRandom( $uuid, $rand, $gte ) {
 198                 $dbw  = $this->getMasterDB();
 199                 $dir  = $gte ? 'ASC' : 'DESC';
 200                 $ineq = $gte ? '>=' : '<=';
 201
 202                 $row = false; // the row acquired
 203                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 204                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 205                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 206                 // be used here with MySQL.
 207                 do {
 208                         $row = $dbw->selectRow( 'job', '*', // find a random job
 209                                 array(
 210                                         'job_cmd'   => $this->type,
 211                                         'job_token' => '',
 212                                         "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 213                                 __METHOD__,
 214                                 array( 'ORDER BY' => "job_random {$dir}" )
 215                         );
 216                         if ( $row ) { // claim the job
 217                                 $dbw->update( 'job', // update by PK
 218                                         array(
 219                                                 'job_token'           => $uuid,
 220                                                 'job_token_timestamp' => $dbw->timestamp(),
 221                                                 'job_attempts = job_attempts+1' ),
 222                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 223                                         __METHOD__
 224                                 );
 225                                 // This might get raced out by another runner when claiming the previously
 226                                 // selected row. The use of job_random should minimize this problem, however.
 227                                 if ( !$dbw->affectedRows() ) {
 228                                         $row = false; // raced out
 229                                 }
 230                         } else {
 231                                 break; // nothing to do
 232                         }
 233                 } while ( !$row );
 234
 235                 return $row;
 236         }
 237
 238         /**
 239          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 240          *
 241          * @param $uuid string 32 char hex string
 242          * @return Row|false
 243          */
 244         protected function claimOldest( $uuid ) {
 245                 $dbw  = $this->getMasterDB();
 246
 247                 $row = false; // the row acquired
 248                 do {
 249                         if ( $dbw->getType() === 'mysql' ) {
 250                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 251                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 252                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 253                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 254                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 255                                         "SET " .
 256                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 257                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 258                                                 "job_attempts = job_attempts+1 " .
 259                                         "WHERE ( " .
 260                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 261                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 262                                         ") ORDER BY job_id ASC LIMIT 1",
 263                                         __METHOD__
 264                                 );
 265                         } else {
 266                                 // Use a subquery to find the job, within an UPDATE to claim it.
 267                                 // This uses as much of the DB wrapper functions as possible.
 268                                 $dbw->update( 'job',
 269                                         array(
 270                                                 'job_token'           => $uuid,
 271                                                 'job_token_timestamp' => $dbw->timestamp(),
 272                                                 'job_attempts = job_attempts+1' ),
 273                                         array( 'job_id = (' .
 274                                                 $dbw->selectSQLText( 'job', 'job_id',
 275                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 276                                                         __METHOD__,
 277                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 278                                                 ')'
 279                                         ),
 280                                         __METHOD__
 281                                 );
 282                         }
 283                         // Fetch any row that we just reserved...
 284                         if ( $dbw->affectedRows() ) {
 285                                 $row = $dbw->selectRow( 'job', '*',
 286                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 287                                 );
 288                                 if ( !$row ) { // raced out by duplicate job removal
 289                                         wfDebugLog( 'JobQueueDB', "Row deleted as duplicate by another process." );
 290                                 }
 291                         } else {
 292                                 break; // nothing to do
 293                         }
 294                 } while ( !$row );
 295
 296                 return $row;
 297         }
 298
 299         /**
 300          * Recycle or destroy any jobs that have been claimed for too long
 301          *
 302          * @return integer Number of jobs recycled/deleted
 303          */
 304         protected function recycleStaleJobs() {
 305                 $now   = time();
 306                 $dbw   = $this->getMasterDB();
 307                 $count = 0; // affected rows
 308
 309                 if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 310                         return $count; // already in progress
 311                 }
 312
 313                 // Remove claims on jobs acquired for too long if enabled...
 314                 if ( $this->claimTTL > 0 ) {
 315                         $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 316                         // Get the IDs of jobs that have be claimed but not finished after too long.
 317                         // These jobs can be recycled into the queue by expiring the claim. Selecting
 318                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 319                         $res = $dbw->select( 'job', 'job_id',
 320                                 array(
 321                                         'job_cmd' => $this->type,
 322                                         "job_token != {$dbw->addQuotes( '' )}", // was acquired
 323                                         "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 324                                         "job_attempts < {$dbw->addQuotes( self::MAX_ATTEMPTS )}" ), // retries left
 325                                 __METHOD__
 326                         );
 327                         $ids = array_map( function( $o ) { return $o->job_id; }, iterator_to_array( $res ) );
 328                         if ( count( $ids ) ) {
 329                                 // Reset job_token for these jobs so that other runners will pick them up.
 330                                 // Set the timestamp to the current time, as it is useful to now that the job
 331                                 // was already tried before (the timestamp becomes the "released" time).
 332                                 $dbw->update( 'job',
 333                                         array(
 334                                                 'job_token' => '',
 335                                                 'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 336                                         array(
 337                                                 'job_id' => $ids ),
 338                                         __METHOD__
 339                                 );
 340                                 $count += $dbw->affectedRows();
 341                         }
 342                 }
 343
 344                 // Just destroy any stale jobs...
 345                 $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 346                 $conds = array(
 347                         'job_cmd' => $this->type,
 348                         "job_token != {$dbw->addQuotes( '' )}", // was acquired
 349                         "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 350                 );
 351                 if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 352                         $conds[] = "job_attempts >= {$dbw->addQuotes( self::MAX_ATTEMPTS )}";
 353                 }
 354                 // Get the IDs of jobs that are considered stale and should be removed. Selecting
 355                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 356                 $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 357                 $ids = array_map( function( $o ) { return $o->job_id; }, iterator_to_array( $res ) );
 358                 if ( count( $ids ) ) {
 359                         $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 360                         $count += $dbw->affectedRows();
 361                 }
 362
 363                 $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 364
 365                 return $count;
 366         }
 367
 368         /**
 369          * @see JobQueue::doAck()
 370          * @return Job|bool
 371          */
 372         protected function doAck( Job $job ) {
 373                 if ( !$job->getId() ) {
 374                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 375                 }
 376
 377                 $dbw = $this->getMasterDB();
 378                 $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 379
 380                 // Delete a row with a single DELETE without holding row locks over RTTs...
 381                 $dbw->delete( 'job',
 382                         array( 'job_cmd' => $this->type, 'job_id' => $job->getId() ), __METHOD__ );
 383
 384                 return true;
 385         }
 386
 387         /**
 388          * @see JobQueue::doDeduplicateRootJob()
 389          * @return bool
 390          */
 391         protected function doDeduplicateRootJob( Job $job ) {
 392                 $params = $job->getParams();
 393                 if ( !isset( $params['rootJobSignature'] ) ) {
 394                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 395                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 396                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 397                 }
 398                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 399                 // Callers should call batchInsert() and then this function so that if the insert
 400                 // fails, the de-duplication registration will be aborted. Since the insert is
 401                 // deferred till "transaction idle", do that same here, so that the ordering is
 402                 // maintained. Having only the de-duplication registration succeed would cause
 403                 // jobs to become no-ops without any actual jobs that made them redundant.
 404                 $this->getMasterDB()->onTransactionIdle( function() use ( $params, $key ) {
 405                         global $wgMemc;
 406
 407                         $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 408                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 409                                 return true; // a newer version of this root job was enqueued
 410                         }
 411
 412                         // Update the timestamp of the last root job started at the location...
 413                         return $wgMemc->set( $key, $params['rootJobTimestamp'], 14*86400 ); // 2 weeks
 414                 } );
 415
 416                 return true;
 417         }
 418
 419         /**
 420          * Check if the "root" job of a given job has been superseded by a newer one
 421          *
 422          * @param $job Job
 423          * @return bool
 424          */
 425         protected function isRootJobOldDuplicate( Job $job ) {
 426                 global $wgMemc;
 427
 428                 $params = $job->getParams();
 429                 if ( !isset( $params['rootJobSignature'] ) ) {
 430                         return false; // job has no de-deplication info
 431                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 432                         trigger_error( "Cannot check root job; missing 'rootJobTimestamp'." );
 433                         return false;
 434                 }
 435
 436                 // Get the last time this root job was enqueued
 437                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 438
 439                 // Check if a new root job was started at the location after this one's...
 440                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 441         }
 442
 443         /**
 444          * @see JobQueue::doWaitForBackups()
 445          * @return void
 446          */
 447         protected function doWaitForBackups() {
 448                 wfWaitForSlaves();
 449         }
 450
 451         /**
 452          * @return DatabaseBase
 453          */
 454         protected function getSlaveDB() {
 455                 return wfGetDB( DB_SLAVE, array(), $this->wiki );
 456         }
 457
 458         /**
 459          * @return DatabaseBase
 460          */
 461         protected function getMasterDB() {
 462                 return wfGetDB( DB_MASTER, array(), $this->wiki );
 463         }
 464
 465         /**
 466          * @param $job Job
 467          * @return array
 468          */
 469         protected function insertFields( Job $job ) {
 470                 $dbw = $this->getMasterDB();
 471                 return array(
 472                         // Fields that describe the nature of the job
 473                         'job_cmd'       => $job->getType(),
 474                         'job_namespace' => $job->getTitle()->getNamespace(),
 475                         'job_title'     => $job->getTitle()->getDBkey(),
 476                         'job_params'    => self::makeBlob( $job->getParams() ),
 477                         // Additional job metadata
 478                         'job_id'        => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 479                         'job_timestamp' => $dbw->timestamp(),
 480                         'job_sha1'      => wfBaseConvert(
 481                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 482                                 16, 36, 31
 483                         ),
 484                         'job_random'    => mt_rand( 0, self::MAX_JOB_RANDOM )
 485                 );
 486         }
 487
 488         /**
 489          * @return string
 490          */
 491         private function getEmptinessCacheKey() {
 492                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 493                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'isempty' );
 494         }
 495
 496         /**
 497          * @param string $signature Hash identifier of the root job
 498          * @return string
 499          */
 500         private function getRootJobCacheKey( $signature ) {
 501                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 502                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 503         }
 504
 505         /**
 506          * @param $params
 507          * @return string
 508          */
 509         protected static function makeBlob( $params ) {
 510                 if ( $params !== false ) {
 511                         return serialize( $params );
 512                 } else {
 513                         return '';
 514                 }
 515         }
 516
 517         /**
 518          * @param $blob
 519          * @return bool|mixed
 520          */
 521         protected static function extractBlob( $blob ) {
 522                 if ( (string)$blob !== '' ) {
 523                         return unserialize( $blob );
 524                 } else {
 525                         return false;
 526                 }
 527         }
 528 }