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