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          * @param array $jobs
  62          * @param $flags
  63          * @throws DBError|Exception
  64          * @return bool
  65          */
  66         protected function doBatchPush( array $jobs, $flags ) {
  67                 if ( count( $jobs ) ) {
  68                         $dbw = $this->getMasterDB();
  69
  70                         $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
  71                         $rowList = array(); // list of jobs for jobs that are are not de-duplicated
  72
  73                         foreach ( $jobs as $job ) {
  74                                 $row = $this->insertFields( $job );
  75                                 if ( $job->ignoreDuplicates() ) {
  76                                         $rowSet[$row['job_sha1']] = $row;
  77                                 } else {
  78                                         $rowList[] = $row;
  79                                 }
  80                         }
  81
  82                         $atomic = ( $flags & self::QoS_Atomic );
  83                         $key    = $this->getEmptinessCacheKey();
  84                         $ttl    = self::CACHE_TTL;
  85
  86                         $dbw->onTransactionIdle(
  87                                 function() use ( $dbw, $rowSet, $rowList, $atomic, $key, $ttl
  88                         ) {
  89                                 global $wgMemc;
  90
  91                                 if ( $atomic ) {
  92                                         $dbw->begin( __METHOD__ ); // wrap all the job additions in one 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                                         }
 120                                         throw $e;
 121                                 }
 122                                 if ( $atomic ) {
 123                                         $dbw->commit( __METHOD__ );
 124                                 }
 125
 126                                 $wgMemc->set( $key, 'false', $ttl ); // queue is not empty
 127                         } );
 128                 }
 129
 130                 return true;
 131         }
 132
 133         /**
 134          * @see JobQueue::doPop()
 135          * @return Job|bool
 136          */
 137         protected function doPop() {
 138                 global $wgMemc;
 139
 140                 if ( $wgMemc->get( $this->getEmptinessCacheKey() ) === 'true' ) {
 141                         return false; // queue is empty
 142                 }
 143
 144                 $dbw = $this->getMasterDB();
 145                 $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 146
 147                 $uuid = wfRandomString( 32 ); // pop attempt
 148                 $job = false; // job popped off
 149                 // Occasionally recycle jobs back into the queue that have been claimed too long
 150                 if ( mt_rand( 0, 99 ) == 0 ) {
 151                         $this->recycleStaleJobs();
 152                 }
 153                 do { // retry when our row is invalid or deleted as a duplicate
 154                         // Try to reserve a row in the DB...
 155                         if ( in_array( $this->order, array( 'fifo', 'timestamp' ) ) ) {
 156                                 $row = $this->claimOldest( $uuid );
 157                         } else { // random first
 158                                 $rand = mt_rand( 0, self::MAX_JOB_RANDOM ); // encourage concurrent UPDATEs
 159                                 $gte  = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
 160                                 $row  = $this->claimRandom( $uuid, $rand, $gte );
 161                                 if ( !$row ) { // need to try the other direction
 162                                         $row = $this->claimRandom( $uuid, $rand, !$gte );
 163                                 }
 164                         }
 165                         // Check if we found a row to reserve...
 166                         if ( !$row ) {
 167                                 $wgMemc->set( $this->getEmptinessCacheKey(), 'true', self::CACHE_TTL );
 168                                 break; // nothing to do
 169                         }
 170                         // Get the job object from the row...
 171                         $title = Title::makeTitleSafe( $row->job_namespace, $row->job_title );
 172                         if ( !$title ) {
 173                                 $dbw->delete( 'job', array( 'job_id' => $row->job_id ), __METHOD__ );
 174                                 wfIncrStats( 'job-pop' );
 175                                 wfDebugLog( 'JobQueueDB', "Row has invalid title '{$row->job_title}'." );
 176                                 continue; // try again
 177                         }
 178                         $job = Job::factory( $row->job_cmd, $title,
 179                                 self::extractBlob( $row->job_params ), $row->job_id );
 180                         $job->id = $row->job_id; // XXX: work around broken subclasses
 181                         // Flag this job as an old duplicate based on its "root" job...
 182                         if ( $this->isRootJobOldDuplicate( $job ) ) {
 183                                 $job = DuplicateJob::newFromJob( $job ); // convert to a no-op
 184                         }
 185                         break; // done
 186                 } while( true );
 187
 188                 return $job;
 189         }
 190
 191         /**
 192          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 193          *
 194          * @param $uuid string 32 char hex string
 195          * @param $rand integer Random unsigned integer (31 bits)
 196          * @param $gte bool Search for job_random >= $random (otherwise job_random <= $random)
 197          * @return Row|false
 198          */
 199         protected function claimRandom( $uuid, $rand, $gte ) {
 200                 $dbw  = $this->getMasterDB();
 201                 $dir  = $gte ? 'ASC' : 'DESC';
 202                 $ineq = $gte ? '>=' : '<=';
 203
 204                 $row = false; // the row acquired
 205                 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
 206                 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
 207                 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
 208                 // be used here with MySQL.
 209                 do {
 210                         $row = $dbw->selectRow( 'job', '*', // find a random job
 211                                 array(
 212                                         'job_cmd'   => $this->type,
 213                                         'job_token' => '',
 214                                         "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
 215                                 __METHOD__,
 216                                 array( 'ORDER BY' => "job_random {$dir}" )
 217                         );
 218                         if ( $row ) { // claim the job
 219                                 $dbw->update( 'job', // update by PK
 220                                         array(
 221                                                 'job_token'           => $uuid,
 222                                                 'job_token_timestamp' => $dbw->timestamp(),
 223                                                 'job_attempts = job_attempts+1' ),
 224                                         array( 'job_cmd' => $this->type, 'job_id' => $row->job_id, 'job_token' => '' ),
 225                                         __METHOD__
 226                                 );
 227                                 // This might get raced out by another runner when claiming the previously
 228                                 // selected row. The use of job_random should minimize this problem, however.
 229                                 if ( !$dbw->affectedRows() ) {
 230                                         $row = false; // raced out
 231                                 }
 232                         } else {
 233                                 break; // nothing to do
 234                         }
 235                 } while ( !$row );
 236
 237                 return $row;
 238         }
 239
 240         /**
 241          * Reserve a row with a single UPDATE without holding row locks over RTTs...
 242          *
 243          * @param $uuid string 32 char hex string
 244          * @return Row|false
 245          */
 246         protected function claimOldest( $uuid ) {
 247                 $dbw  = $this->getMasterDB();
 248
 249                 $row = false; // the row acquired
 250                 do {
 251                         if ( $dbw->getType() === 'mysql' ) {
 252                                 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
 253                                 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
 254                                 // Oracle and Postgre have no such limitation. However, MySQL offers an
 255                                 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
 256                                 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
 257                                         "SET " .
 258                                                 "job_token = {$dbw->addQuotes( $uuid ) }, " .
 259                                                 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
 260                                                 "job_attempts = job_attempts+1 " .
 261                                         "WHERE ( " .
 262                                                 "job_cmd = {$dbw->addQuotes( $this->type )} " .
 263                                                 "AND job_token = {$dbw->addQuotes( '' )} " .
 264                                         ") ORDER BY job_id ASC LIMIT 1",
 265                                         __METHOD__
 266                                 );
 267                         } else {
 268                                 // Use a subquery to find the job, within an UPDATE to claim it.
 269                                 // This uses as much of the DB wrapper functions as possible.
 270                                 $dbw->update( 'job',
 271                                         array(
 272                                                 'job_token'           => $uuid,
 273                                                 'job_token_timestamp' => $dbw->timestamp(),
 274                                                 'job_attempts = job_attempts+1' ),
 275                                         array( 'job_id = (' .
 276                                                 $dbw->selectSQLText( 'job', 'job_id',
 277                                                         array( 'job_cmd' => $this->type, 'job_token' => '' ),
 278                                                         __METHOD__,
 279                                                         array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
 280                                                 ')'
 281                                         ),
 282                                         __METHOD__
 283                                 );
 284                         }
 285                         // Fetch any row that we just reserved...
 286                         if ( $dbw->affectedRows() ) {
 287                                 $row = $dbw->selectRow( 'job', '*',
 288                                         array( 'job_cmd' => $this->type, 'job_token' => $uuid ), __METHOD__
 289                                 );
 290                                 if ( !$row ) { // raced out by duplicate job removal
 291                                         wfDebugLog( 'JobQueueDB', "Row deleted as duplicate by another process." );
 292                                 }
 293                         } else {
 294                                 break; // nothing to do
 295                         }
 296                 } while ( !$row );
 297
 298                 return $row;
 299         }
 300
 301         /**
 302          * Recycle or destroy any jobs that have been claimed for too long
 303          *
 304          * @return integer Number of jobs recycled/deleted
 305          */
 306         protected function recycleStaleJobs() {
 307                 $now   = time();
 308                 $dbw   = $this->getMasterDB();
 309                 $count = 0; // affected rows
 310
 311                 if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__, 1 ) ) {
 312                         return $count; // already in progress
 313                 }
 314
 315                 // Remove claims on jobs acquired for too long if enabled...
 316                 if ( $this->claimTTL > 0 ) {
 317                         $claimCutoff = $dbw->timestamp( $now - $this->claimTTL );
 318                         // Get the IDs of jobs that have be claimed but not finished after too long.
 319                         // These jobs can be recycled into the queue by expiring the claim. Selecting
 320                         // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 321                         $res = $dbw->select( 'job', 'job_id',
 322                                 array(
 323                                         'job_cmd' => $this->type,
 324                                         "job_token != {$dbw->addQuotes( '' )}", // was acquired
 325                                         "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
 326                                         "job_attempts < {$dbw->addQuotes( self::MAX_ATTEMPTS )}" ), // retries left
 327                                 __METHOD__
 328                         );
 329                         $ids = array_map( function( $o ) { return $o->job_id; }, iterator_to_array( $res ) );
 330                         if ( count( $ids ) ) {
 331                                 // Reset job_token for these jobs so that other runners will pick them up.
 332                                 // Set the timestamp to the current time, as it is useful to now that the job
 333                                 // was already tried before (the timestamp becomes the "released" time).
 334                                 $dbw->update( 'job',
 335                                         array(
 336                                                 'job_token' => '',
 337                                                 'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
 338                                         array(
 339                                                 'job_id' => $ids ),
 340                                         __METHOD__
 341                                 );
 342                                 $count += $dbw->affectedRows();
 343                         }
 344                 }
 345
 346                 // Just destroy any stale jobs...
 347                 $pruneCutoff = $dbw->timestamp( $now - self::MAX_AGE_PRUNE );
 348                 $conds = array(
 349                         'job_cmd' => $this->type,
 350                         "job_token != {$dbw->addQuotes( '' )}", // was acquired
 351                         "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
 352                 );
 353                 if ( $this->claimTTL > 0 ) { // only prune jobs attempted too many times...
 354                         $conds[] = "job_attempts >= {$dbw->addQuotes( self::MAX_ATTEMPTS )}";
 355                 }
 356                 // Get the IDs of jobs that are considered stale and should be removed. Selecting
 357                 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
 358                 $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__ );
 359                 $ids = array_map( function( $o ) { return $o->job_id; }, iterator_to_array( $res ) );
 360                 if ( count( $ids ) ) {
 361                         $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__ );
 362                         $count += $dbw->affectedRows();
 363                 }
 364
 365                 $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__ );
 366
 367                 return $count;
 368         }
 369
 370         /**
 371          * @see JobQueue::doAck()
 372          * @param Job $job
 373          * @throws MWException
 374          * @return Job|bool
 375          */
 376         protected function doAck( Job $job ) {
 377                 if ( !$job->getId() ) {
 378                         throw new MWException( "Job of type '{$job->getType()}' has no ID." );
 379                 }
 380
 381                 $dbw = $this->getMasterDB();
 382                 $dbw->commit( __METHOD__, 'flush' ); // flush existing transaction
 383
 384                 // Delete a row with a single DELETE without holding row locks over RTTs...
 385                 $dbw->delete( 'job',
 386                         array( 'job_cmd' => $this->type, 'job_id' => $job->getId() ), __METHOD__ );
 387
 388                 return true;
 389         }
 390
 391         /**
 392          * @see JobQueue::doDeduplicateRootJob()
 393          * @param Job $job
 394          * @throws MWException
 395          * @return bool
 396          */
 397         protected function doDeduplicateRootJob( Job $job ) {
 398                 $params = $job->getParams();
 399                 if ( !isset( $params['rootJobSignature'] ) ) {
 400                         throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
 401                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 402                         throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
 403                 }
 404                 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
 405                 // Callers should call batchInsert() and then this function so that if the insert
 406                 // fails, the de-duplication registration will be aborted. Since the insert is
 407                 // deferred till "transaction idle", do that same here, so that the ordering is
 408                 // maintained. Having only the de-duplication registration succeed would cause
 409                 // jobs to become no-ops without any actual jobs that made them redundant.
 410                 $this->getMasterDB()->onTransactionIdle( function() use ( $params, $key ) {
 411                         global $wgMemc;
 412
 413                         $timestamp = $wgMemc->get( $key ); // current last timestamp of this job
 414                         if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
 415                                 return true; // a newer version of this root job was enqueued
 416                         }
 417
 418                         // Update the timestamp of the last root job started at the location...
 419                         return $wgMemc->set( $key, $params['rootJobTimestamp'], 14*86400 ); // 2 weeks
 420                 } );
 421
 422                 return true;
 423         }
 424
 425         /**
 426          * Check if the "root" job of a given job has been superseded by a newer one
 427          *
 428          * @param $job Job
 429          * @return bool
 430          */
 431         protected function isRootJobOldDuplicate( Job $job ) {
 432                 global $wgMemc;
 433
 434                 $params = $job->getParams();
 435                 if ( !isset( $params['rootJobSignature'] ) ) {
 436                         return false; // job has no de-deplication info
 437                 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
 438                         trigger_error( "Cannot check root job; missing 'rootJobTimestamp'." );
 439                         return false;
 440                 }
 441
 442                 // Get the last time this root job was enqueued
 443                 $timestamp = $wgMemc->get( $this->getRootJobCacheKey( $params['rootJobSignature'] ) );
 444
 445                 // Check if a new root job was started at the location after this one's...
 446                 return ( $timestamp && $timestamp > $params['rootJobTimestamp'] );
 447         }
 448
 449         /**
 450          * @see JobQueue::doWaitForBackups()
 451          * @return void
 452          */
 453         protected function doWaitForBackups() {
 454                 wfWaitForSlaves();
 455         }
 456
 457         /**
 458          * @return DatabaseBase
 459          */
 460         protected function getSlaveDB() {
 461                 return wfGetDB( DB_SLAVE, array(), $this->wiki );
 462         }
 463
 464         /**
 465          * @return DatabaseBase
 466          */
 467         protected function getMasterDB() {
 468                 return wfGetDB( DB_MASTER, array(), $this->wiki );
 469         }
 470
 471         /**
 472          * @param $job Job
 473          * @return array
 474          */
 475         protected function insertFields( Job $job ) {
 476                 $dbw = $this->getMasterDB();
 477                 return array(
 478                         // Fields that describe the nature of the job
 479                         'job_cmd'       => $job->getType(),
 480                         'job_namespace' => $job->getTitle()->getNamespace(),
 481                         'job_title'     => $job->getTitle()->getDBkey(),
 482                         'job_params'    => self::makeBlob( $job->getParams() ),
 483                         // Additional job metadata
 484                         'job_id'        => $dbw->nextSequenceValue( 'job_job_id_seq' ),
 485                         'job_timestamp' => $dbw->timestamp(),
 486                         'job_sha1'      => wfBaseConvert(
 487                                 sha1( serialize( $job->getDeduplicationInfo() ) ),
 488                                 16, 36, 31
 489                         ),
 490                         'job_random'    => mt_rand( 0, self::MAX_JOB_RANDOM )
 491                 );
 492         }
 493
 494         /**
 495          * @return string
 496          */
 497         private function getEmptinessCacheKey() {
 498                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 499                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'isempty' );
 500         }
 501
 502         /**
 503          * @param string $signature Hash identifier of the root job
 504          * @return string
 505          */
 506         private function getRootJobCacheKey( $signature ) {
 507                 list( $db, $prefix ) = wfSplitWikiID( $this->wiki );
 508                 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type, 'rootjob', $signature );
 509         }
 510
 511         /**
 512          * @param $params
 513          * @return string
 514          */
 515         protected static function makeBlob( $params ) {
 516                 if ( $params !== false ) {
 517                         return serialize( $params );
 518                 } else {
 519                         return '';
 520                 }
 521         }
 522
 523         /**
 524          * @param $blob
 525          * @return bool|mixed
 526          */
 527         protected static function extractBlob( $blob ) {
 528                 if ( (string)$blob !== '' ) {
 529                         return unserialize( $blob );
 530                 } else {
 531                         return false;
 532                 }
 533         }
 534 }