3 * Database-backed job queue code.
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.
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.
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
21 * @author Aaron Schulz
25 * Class to handle job queues stored in the DB
30 class JobQueueDB
extends JobQueue
{
31 const CACHE_TTL_SHORT
= 30; // integer; seconds to cache info without re-validating
32 const CACHE_TTL_LONG
= 300; // integer; seconds to cache info that is kept up to date
33 const MAX_AGE_PRUNE
= 604800; // integer; seconds a job can live once claimed
34 const MAX_JOB_RANDOM
= 2147483647; // integer; 2^31 - 1, used for job_random
35 const MAX_OFFSET
= 255; // integer; maximum number of rows to skip
40 protected $cluster = false; // string; name of an external DB cluster
43 * Additional parameters include:
44 * - cluster : The name of an external cluster registered via LBFactory.
45 * If not specified, the primary DB cluster for the wiki will be used.
46 * This can be overridden with a custom cluster so that DB handles will
47 * be retrieved via LBFactory::getExternalLB() and getConnection().
48 * @param $params array
50 protected function __construct( array $params ) {
53 parent
::__construct( $params );
55 $this->cluster
= isset( $params['cluster'] ) ?
$params['cluster'] : false;
56 // Make sure that we don't use the SQL cache, which would be harmful
57 $this->cache
= ( $wgMemc instanceof SqlBagOStuff
) ?
new EmptyBagOStuff() : $wgMemc;
60 protected function supportedOrders() {
61 return array( 'random', 'timestamp', 'fifo' );
64 protected function optimalOrder() {
69 * @see JobQueue::doIsEmpty()
72 protected function doIsEmpty() {
73 $key = $this->getCacheKey( 'empty' );
75 $isEmpty = $this->cache
->get( $key );
76 if ( $isEmpty === 'true' ) {
78 } elseif ( $isEmpty === 'false' ) {
82 list( $dbr, $scope ) = $this->getSlaveDB();
83 $found = $dbr->selectField( // unclaimed job
84 'job', '1', array( 'job_cmd' => $this->type
, 'job_token' => '' ), __METHOD__
86 $this->cache
->add( $key, $found ?
'false' : 'true', self
::CACHE_TTL_LONG
);
92 * @see JobQueue::doGetSize()
95 protected function doGetSize() {
96 $key = $this->getCacheKey( 'size' );
98 $size = $this->cache
->get( $key );
99 if ( is_int( $size ) ) {
103 list( $dbr, $scope ) = $this->getSlaveDB();
104 $size = (int)$dbr->selectField( 'job', 'COUNT(*)',
105 array( 'job_cmd' => $this->type
, 'job_token' => '' ),
108 $this->cache
->set( $key, $size, self
::CACHE_TTL_SHORT
);
114 * @see JobQueue::doGetAcquiredCount()
117 protected function doGetAcquiredCount() {
118 if ( $this->claimTTL
<= 0 ) {
119 return 0; // no acknowledgements
122 $key = $this->getCacheKey( 'acquiredcount' );
124 $count = $this->cache
->get( $key );
125 if ( is_int( $count ) ) {
129 list( $dbr, $scope ) = $this->getSlaveDB();
130 $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
131 array( 'job_cmd' => $this->type
, "job_token != {$dbr->addQuotes( '' )}" ),
134 $this->cache
->set( $key, $count, self
::CACHE_TTL_SHORT
);
140 * @see JobQueue::doGetAbandonedCount()
142 * @throws MWException
144 protected function doGetAbandonedCount() {
147 if ( $this->claimTTL
<= 0 ) {
148 return 0; // no acknowledgements
151 $key = $this->getCacheKey( 'abandonedcount' );
153 $count = $wgMemc->get( $key );
154 if ( is_int( $count ) ) {
158 list( $dbr, $scope ) = $this->getSlaveDB();
159 $count = (int)$dbr->selectField( 'job', 'COUNT(*)',
161 'job_cmd' => $this->type
,
162 "job_token != {$dbr->addQuotes( '' )}",
163 "job_attempts >= " . $dbr->addQuotes( $this->maxTries
)
167 $wgMemc->set( $key, $count, self
::CACHE_TTL_SHORT
);
173 * @see JobQueue::doBatchPush()
176 * @throws DBError|Exception
179 protected function doBatchPush( array $jobs, $flags ) {
180 list( $dbw, $scope ) = $this->getMasterDB();
183 $method = __METHOD__
;
184 $dbw->onTransactionIdle(
185 function() use ( $dbw, $that, $jobs, $flags, $method, $scope ) {
186 $that->doBatchPushInternal( $dbw, $jobs, $flags, $method );
194 * This function should *not* be called outside of JobQueueDB
196 * @param DatabaseBase $dbw
199 * @param string $method
203 public function doBatchPushInternal( DatabaseBase
$dbw, array $jobs, $flags, $method ) {
204 if ( !count( $jobs ) ) {
208 $rowSet = array(); // (sha1 => job) map for jobs that are de-duplicated
209 $rowList = array(); // list of jobs for jobs that are are not de-duplicated
210 foreach ( $jobs as $job ) {
211 $row = $this->insertFields( $job );
212 if ( $job->ignoreDuplicates() ) {
213 $rowSet[$row['job_sha1']] = $row;
219 if ( $flags & self
::QOS_ATOMIC
) {
220 $dbw->begin( $method ); // wrap all the job additions in one transaction
223 // Strip out any duplicate jobs that are already in the queue...
224 if ( count( $rowSet ) ) {
225 $res = $dbw->select( 'job', 'job_sha1',
227 // No job_type condition since it's part of the job_sha1 hash
228 'job_sha1' => array_keys( $rowSet ),
229 'job_token' => '' // unclaimed
233 foreach ( $res as $row ) {
234 wfDebug( "Job with hash '{$row->job_sha1}' is a duplicate." );
235 unset( $rowSet[$row->job_sha1
] ); // already enqueued
238 // Build the full list of job rows to insert
239 $rows = array_merge( $rowList, array_values( $rowSet ) );
240 // Insert the job rows in chunks to avoid slave lag...
241 foreach ( array_chunk( $rows, 50 ) as $rowBatch ) {
242 $dbw->insert( 'job', $rowBatch, $method );
244 JobQueue
::incrStats( 'job-insert', $this->type
, count( $rows ) );
245 JobQueue
::incrStats( 'job-insert-duplicate', $this->type
,
246 count( $rowSet ) +
count( $rowList ) - count( $rows ) );
247 } catch ( DBError
$e ) {
248 if ( $flags & self
::QOS_ATOMIC
) {
249 $dbw->rollback( $method );
253 if ( $flags & self
::QOS_ATOMIC
) {
254 $dbw->commit( $method );
257 $this->cache
->set( $this->getCacheKey( 'empty' ), 'false', JobQueueDB
::CACHE_TTL_LONG
);
263 * @see JobQueue::doPop()
266 protected function doPop() {
267 if ( $this->cache
->get( $this->getCacheKey( 'empty' ) ) === 'true' ) {
268 return false; // queue is empty
271 list( $dbw, $scope ) = $this->getMasterDB();
272 $dbw->commit( __METHOD__
, 'flush' ); // flush existing transaction
274 $uuid = wfRandomString( 32 ); // pop attempt
275 $job = false; // job popped off
276 do { // retry when our row is invalid or deleted as a duplicate
277 // Try to reserve a row in the DB...
278 if ( in_array( $this->order
, array( 'fifo', 'timestamp' ) ) ) {
279 $row = $this->claimOldest( $uuid );
280 } else { // random first
281 $rand = mt_rand( 0, self
::MAX_JOB_RANDOM
); // encourage concurrent UPDATEs
282 $gte = (bool)mt_rand( 0, 1 ); // find rows with rand before/after $rand
283 $row = $this->claimRandom( $uuid, $rand, $gte );
285 // Check if we found a row to reserve...
287 $this->cache
->set( $this->getCacheKey( 'empty' ), 'true', self
::CACHE_TTL_LONG
);
288 break; // nothing to do
290 JobQueue
::incrStats( 'job-pop', $this->type
);
291 // Get the job object from the row...
292 $title = Title
::makeTitleSafe( $row->job_namespace
, $row->job_title
);
294 $dbw->delete( 'job', array( 'job_id' => $row->job_id
), __METHOD__
);
295 wfDebugLog( 'JobQueueDB', "Row has invalid title '{$row->job_title}'." );
296 continue; // try again
298 $job = Job
::factory( $row->job_cmd
, $title,
299 self
::extractBlob( $row->job_params
), $row->job_id
);
300 $job->metadata
['id'] = $row->job_id
;
301 $job->id
= $row->job_id
; // XXX: work around broken subclasses
309 * Reserve a row with a single UPDATE without holding row locks over RTTs...
311 * @param string $uuid 32 char hex string
312 * @param $rand integer Random unsigned integer (31 bits)
313 * @param bool $gte Search for job_random >= $random (otherwise job_random <= $random)
316 protected function claimRandom( $uuid, $rand, $gte ) {
317 list( $dbw, $scope ) = $this->getMasterDB();
318 // Check cache to see if the queue has <= OFFSET items
319 $tinyQueue = $this->cache
->get( $this->getCacheKey( 'small' ) );
321 $row = false; // the row acquired
322 $invertedDirection = false; // whether one job_random direction was already scanned
323 // This uses a replication safe method for acquiring jobs. One could use UPDATE+LIMIT
324 // instead, but that either uses ORDER BY (in which case it deadlocks in MySQL) or is
325 // not replication safe. Due to http://bugs.mysql.com/bug.php?id=6980, subqueries cannot
326 // be used here with MySQL.
328 if ( $tinyQueue ) { // queue has <= MAX_OFFSET rows
329 // For small queues, using OFFSET will overshoot and return no rows more often.
330 // Instead, this uses job_random to pick a row (possibly checking both directions).
331 $ineq = $gte ?
'>=' : '<=';
332 $dir = $gte ?
'ASC' : 'DESC';
333 $row = $dbw->selectRow( 'job', '*', // find a random job
335 'job_cmd' => $this->type
,
336 'job_token' => '', // unclaimed
337 "job_random {$ineq} {$dbw->addQuotes( $rand )}" ),
339 array( 'ORDER BY' => "job_random {$dir}" )
341 if ( !$row && !$invertedDirection ) {
343 $invertedDirection = true;
344 continue; // try the other direction
346 } else { // table *may* have >= MAX_OFFSET rows
347 // Bug 42614: "ORDER BY job_random" with a job_random inequality causes high CPU
348 // in MySQL if there are many rows for some reason. This uses a small OFFSET
349 // instead of job_random for reducing excess claim retries.
350 $row = $dbw->selectRow( 'job', '*', // find a random job
352 'job_cmd' => $this->type
,
353 'job_token' => '', // unclaimed
356 array( 'OFFSET' => mt_rand( 0, self
::MAX_OFFSET
) )
359 $tinyQueue = true; // we know the queue must have <= MAX_OFFSET rows
360 $this->cache
->set( $this->getCacheKey( 'small' ), 1, 30 );
361 continue; // use job_random
364 if ( $row ) { // claim the job
365 $dbw->update( 'job', // update by PK
367 'job_token' => $uuid,
368 'job_token_timestamp' => $dbw->timestamp(),
369 'job_attempts = job_attempts+1' ),
370 array( 'job_cmd' => $this->type
, 'job_id' => $row->job_id
, 'job_token' => '' ),
373 // This might get raced out by another runner when claiming the previously
374 // selected row. The use of job_random should minimize this problem, however.
375 if ( !$dbw->affectedRows() ) {
376 $row = false; // raced out
379 break; // nothing to do
387 * Reserve a row with a single UPDATE without holding row locks over RTTs...
389 * @param string $uuid 32 char hex string
392 protected function claimOldest( $uuid ) {
393 list( $dbw, $scope ) = $this->getMasterDB();
395 $row = false; // the row acquired
397 if ( $dbw->getType() === 'mysql' ) {
398 // Per http://bugs.mysql.com/bug.php?id=6980, we can't use subqueries on the
399 // same table being changed in an UPDATE query in MySQL (gives Error: 1093).
400 // Oracle and Postgre have no such limitation. However, MySQL offers an
401 // alternative here by supporting ORDER BY + LIMIT for UPDATE queries.
402 $dbw->query( "UPDATE {$dbw->tableName( 'job' )} " .
404 "job_token = {$dbw->addQuotes( $uuid ) }, " .
405 "job_token_timestamp = {$dbw->addQuotes( $dbw->timestamp() )}, " .
406 "job_attempts = job_attempts+1 " .
408 "job_cmd = {$dbw->addQuotes( $this->type )} " .
409 "AND job_token = {$dbw->addQuotes( '' )} " .
410 ") ORDER BY job_id ASC LIMIT 1",
414 // Use a subquery to find the job, within an UPDATE to claim it.
415 // This uses as much of the DB wrapper functions as possible.
418 'job_token' => $uuid,
419 'job_token_timestamp' => $dbw->timestamp(),
420 'job_attempts = job_attempts+1' ),
421 array( 'job_id = (' .
422 $dbw->selectSQLText( 'job', 'job_id',
423 array( 'job_cmd' => $this->type
, 'job_token' => '' ),
425 array( 'ORDER BY' => 'job_id ASC', 'LIMIT' => 1 ) ) .
431 // Fetch any row that we just reserved...
432 if ( $dbw->affectedRows() ) {
433 $row = $dbw->selectRow( 'job', '*',
434 array( 'job_cmd' => $this->type
, 'job_token' => $uuid ), __METHOD__
436 if ( !$row ) { // raced out by duplicate job removal
437 wfDebugLog( 'JobQueueDB', "Row deleted as duplicate by another process." );
440 break; // nothing to do
448 * @see JobQueue::doAck()
450 * @throws MWException
453 protected function doAck( Job
$job ) {
454 if ( !isset( $job->metadata
['id'] ) ) {
455 throw new MWException( "Job of type '{$job->getType()}' has no ID." );
458 list( $dbw, $scope ) = $this->getMasterDB();
459 $dbw->commit( __METHOD__
, 'flush' ); // flush existing transaction
461 // Delete a row with a single DELETE without holding row locks over RTTs...
463 array( 'job_cmd' => $this->type
, 'job_id' => $job->metadata
['id'] ), __METHOD__
);
469 * @see JobQueue::doDeduplicateRootJob()
471 * @throws MWException
474 protected function doDeduplicateRootJob( Job
$job ) {
475 $params = $job->getParams();
476 if ( !isset( $params['rootJobSignature'] ) ) {
477 throw new MWException( "Cannot register root job; missing 'rootJobSignature'." );
478 } elseif ( !isset( $params['rootJobTimestamp'] ) ) {
479 throw new MWException( "Cannot register root job; missing 'rootJobTimestamp'." );
481 $key = $this->getRootJobCacheKey( $params['rootJobSignature'] );
482 // Callers should call batchInsert() and then this function so that if the insert
483 // fails, the de-duplication registration will be aborted. Since the insert is
484 // deferred till "transaction idle", do the same here, so that the ordering is
485 // maintained. Having only the de-duplication registration succeed would cause
486 // jobs to become no-ops without any actual jobs that made them redundant.
487 list( $dbw, $scope ) = $this->getMasterDB();
488 $cache = $this->cache
;
489 $dbw->onTransactionIdle( function() use ( $cache, $params, $key, $scope ) {
490 $timestamp = $cache->get( $key ); // current last timestamp of this job
491 if ( $timestamp && $timestamp >= $params['rootJobTimestamp'] ) {
492 return true; // a newer version of this root job was enqueued
495 // Update the timestamp of the last root job started at the location...
496 return $cache->set( $key, $params['rootJobTimestamp'], JobQueueDB
::ROOTJOB_TTL
);
503 * @see JobQueue::doWaitForBackups()
506 protected function doWaitForBackups() {
513 protected function doGetPeriodicTasks() {
515 'recycleAndDeleteStaleJobs' => array(
516 'callback' => array( $this, 'recycleAndDeleteStaleJobs' ),
517 'period' => ceil( $this->claimTTL
/ 2 )
525 protected function doFlushCaches() {
526 foreach ( array( 'empty', 'size', 'acquiredcount' ) as $type ) {
527 $this->cache
->delete( $this->getCacheKey( $type ) );
532 * @see JobQueue::getAllQueuedJobs()
535 public function getAllQueuedJobs() {
536 list( $dbr, $scope ) = $this->getSlaveDB();
537 return new MappedIterator(
538 $dbr->select( 'job', '*', array( 'job_cmd' => $this->getType(), 'job_token' => '' ) ),
539 function( $row ) use ( $scope ) {
542 Title
::makeTitle( $row->job_namespace
, $row->job_title
),
543 strlen( $row->job_params
) ?
unserialize( $row->job_params
) : false,
546 $job->metadata
['id'] = $row->job_id
;
547 $job->id
= $row->job_id
; // XXX: work around broken subclasses
554 * Recycle or destroy any jobs that have been claimed for too long
556 * @return integer Number of jobs recycled/deleted
558 public function recycleAndDeleteStaleJobs() {
560 list( $dbw, $scope ) = $this->getMasterDB();
561 $count = 0; // affected rows
563 if ( !$dbw->lock( "jobqueue-recycle-{$this->type}", __METHOD__
, 1 ) ) {
564 return $count; // already in progress
567 // Remove claims on jobs acquired for too long if enabled...
568 if ( $this->claimTTL
> 0 ) {
569 $claimCutoff = $dbw->timestamp( $now - $this->claimTTL
);
570 // Get the IDs of jobs that have be claimed but not finished after too long.
571 // These jobs can be recycled into the queue by expiring the claim. Selecting
572 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
573 $res = $dbw->select( 'job', 'job_id',
575 'job_cmd' => $this->type
,
576 "job_token != {$dbw->addQuotes( '' )}", // was acquired
577 "job_token_timestamp < {$dbw->addQuotes( $claimCutoff )}", // stale
578 "job_attempts < {$dbw->addQuotes( $this->maxTries )}" ), // retries left
584 }, iterator_to_array( $res )
586 if ( count( $ids ) ) {
587 // Reset job_token for these jobs so that other runners will pick them up.
588 // Set the timestamp to the current time, as it is useful to now that the job
589 // was already tried before (the timestamp becomes the "released" time).
593 'job_token_timestamp' => $dbw->timestamp( $now ) ), // time of release
598 $count +
= $dbw->affectedRows();
599 JobQueue
::incrStats( 'job-recycle', $this->type
, $dbw->affectedRows() );
600 $this->cache
->set( $this->getCacheKey( 'empty' ), 'false', self
::CACHE_TTL_LONG
);
604 // Just destroy any stale jobs...
605 $pruneCutoff = $dbw->timestamp( $now - self
::MAX_AGE_PRUNE
);
607 'job_cmd' => $this->type
,
608 "job_token != {$dbw->addQuotes( '' )}", // was acquired
609 "job_token_timestamp < {$dbw->addQuotes( $pruneCutoff )}" // stale
611 if ( $this->claimTTL
> 0 ) { // only prune jobs attempted too many times...
612 $conds[] = "job_attempts >= {$dbw->addQuotes( $this->maxTries )}";
614 // Get the IDs of jobs that are considered stale and should be removed. Selecting
615 // the IDs first means that the UPDATE can be done by primary key (less deadlocks).
616 $res = $dbw->select( 'job', 'job_id', $conds, __METHOD__
);
620 }, iterator_to_array( $res )
622 if ( count( $ids ) ) {
623 $dbw->delete( 'job', array( 'job_id' => $ids ), __METHOD__
);
624 $count +
= $dbw->affectedRows();
625 JobQueue
::incrStats( 'job-abandon', $this->type
, $dbw->affectedRows() );
628 $dbw->unlock( "jobqueue-recycle-{$this->type}", __METHOD__
);
634 * @return Array (DatabaseBase, ScopedCallback)
636 protected function getSlaveDB() {
637 return $this->getDB( DB_SLAVE
);
641 * @return Array (DatabaseBase, ScopedCallback)
643 protected function getMasterDB() {
644 return $this->getDB( DB_MASTER
);
648 * @param $index integer (DB_SLAVE/DB_MASTER)
649 * @return Array (DatabaseBase, ScopedCallback)
651 protected function getDB( $index ) {
652 $lb = ( $this->cluster
!== false )
653 ?
wfGetLBFactory()->getExternalLB( $this->cluster
, $this->wiki
)
654 : wfGetLB( $this->wiki
);
655 $conn = $lb->getConnection( $index, array(), $this->wiki
);
658 new ScopedCallback( function() use ( $lb, $conn ) {
659 $lb->reuseConnection( $conn );
668 protected function insertFields( Job
$job ) {
669 list( $dbw, $scope ) = $this->getMasterDB();
671 // Fields that describe the nature of the job
672 'job_cmd' => $job->getType(),
673 'job_namespace' => $job->getTitle()->getNamespace(),
674 'job_title' => $job->getTitle()->getDBkey(),
675 'job_params' => self
::makeBlob( $job->getParams() ),
676 // Additional job metadata
677 'job_id' => $dbw->nextSequenceValue( 'job_job_id_seq' ),
678 'job_timestamp' => $dbw->timestamp(),
679 'job_sha1' => wfBaseConvert(
680 sha1( serialize( $job->getDeduplicationInfo() ) ),
683 'job_random' => mt_rand( 0, self
::MAX_JOB_RANDOM
)
690 private function getCacheKey( $property ) {
691 list( $db, $prefix ) = wfSplitWikiID( $this->wiki
);
692 return wfForeignMemcKey( $db, $prefix, 'jobqueue', $this->type
, $property );
699 protected static function makeBlob( $params ) {
700 if ( $params !== false ) {
701 return serialize( $params );
711 protected static function extractBlob( $blob ) {
712 if ( (string)$blob !== '' ) {
713 return unserialize( $blob );