1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
|
<?php
/**
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
* @file
*/
/**
* Enqueue and run background jobs via a federated queue, for wiki farms.
*
* This class allows for queues to be partitioned into smaller queues.
* A partition is defined by the configuration for a JobQueue instance.
* For example, one can set $wgJobTypeConf['refreshLinks'] to point to a
* JobQueueFederated instance, which itself would consist of three JobQueueRedis
* instances, each using their own redis server. This would allow for the jobs
* to be split (evenly or based on weights) across multiple servers if a single
* server becomes impractical or expensive. Different JobQueue classes can be mixed.
*
* The basic queue configuration (e.g. "order", "claimTTL") of a federated queue
* is inherited by the partition queues. Additional configuration defines what
* section each wiki is in, what partition queues each section uses (and their weight),
* and the JobQueue configuration for each partition. Some sections might only need a
* single queue partition, like the sections for groups of small wikis.
*
* If used for performance, then $wgMainCacheType should be set to memcached/redis.
* Note that "fifo" cannot be used for the ordering, since the data is distributed.
* One can still use "timestamp" instead, as in "roughly timestamp ordered". Also,
* queue classes used by this should ignore down servers (with TTL) to avoid slowness.
*
* @since 1.22
* @ingroup JobQueue
*/
class JobQueueFederated extends JobQueue {
/** @var HashRing */
protected $partitionRing;
/** @var JobQueue[] (partition name => JobQueue) reverse sorted by weight */
protected $partitionQueues = [];
/** @var int Maximum number of partitions to try */
protected $maxPartitionsTry;
/**
* @param array $params Possible keys:
* - sectionsByWiki : A map of wiki IDs to section names.
* Wikis will default to using the section "default".
* - partitionsBySection : Map of section names to maps of (partition name => weight).
* A section called 'default' must be defined if not all wikis
* have explicitly defined sections.
* - configByPartition : Map of queue partition names to configuration arrays.
* These configuration arrays are passed to JobQueue::factory().
* The options set here are overridden by those passed to this
* the federated queue itself (e.g. 'order' and 'claimTTL').
* - maxPartitionsTry : Maximum number of times to attempt job insertion using
* different partition queues. This improves availability
* during failure, at the cost of added latency and somewhat
* less reliable job de-duplication mechanisms.
*/
protected function __construct( array $params ) {
parent::__construct( $params );
$section = $params['sectionsByWiki'][$this->domain] ?? 'default';
if ( !isset( $params['partitionsBySection'][$section] ) ) {
throw new InvalidArgumentException( "No configuration for section '$section'." );
}
$this->maxPartitionsTry = $params['maxPartitionsTry'] ?? 2;
// Get the full partition map
$partitionMap = $params['partitionsBySection'][$section];
arsort( $partitionMap, SORT_NUMERIC );
// Get the config to pass to merge into each partition queue config
$baseConfig = $params;
foreach ( [ 'class', 'sectionsByWiki', 'maxPartitionsTry',
'partitionsBySection', 'configByPartition', ] as $o
) {
unset( $baseConfig[$o] ); // partition queue doesn't care about this
}
// Get the partition queue objects
foreach ( $partitionMap as $partition => $w ) {
if ( !isset( $params['configByPartition'][$partition] ) ) {
throw new InvalidArgumentException( "No configuration for partition '$partition'." );
}
$this->partitionQueues[$partition] = JobQueue::factory(
$baseConfig + $params['configByPartition'][$partition] );
}
// Ring of all partitions
$this->partitionRing = new HashRing( $partitionMap );
}
protected function supportedOrders() {
// No FIFO due to partitioning, though "rough timestamp order" is supported
return [ 'undefined', 'random', 'timestamp' ];
}
protected function optimalOrder() {
return 'undefined'; // defer to the partitions
}
protected function supportsDelayedJobs() {
foreach ( $this->partitionQueues as $queue ) {
if ( !$queue->supportsDelayedJobs() ) {
return false;
}
}
return true;
}
protected function doIsEmpty() {
$empty = true;
$failed = 0;
foreach ( $this->partitionQueues as $queue ) {
try {
$empty = $empty && $queue->doIsEmpty();
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
return $empty;
}
protected function doGetSize() {
return $this->getCrossPartitionSum( 'size', 'doGetSize' );
}
protected function doGetAcquiredCount() {
return $this->getCrossPartitionSum( 'acquiredcount', 'doGetAcquiredCount' );
}
protected function doGetDelayedCount() {
return $this->getCrossPartitionSum( 'delayedcount', 'doGetDelayedCount' );
}
protected function doGetAbandonedCount() {
return $this->getCrossPartitionSum( 'abandonedcount', 'doGetAbandonedCount' );
}
/**
* @param string $type
* @param string $method
* @return int
*/
protected function getCrossPartitionSum( $type, $method ) {
$count = 0;
$failed = 0;
foreach ( $this->partitionQueues as $queue ) {
try {
$count += $queue->$method();
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
return $count;
}
protected function doBatchPush( array $jobs, $flags ) {
// Local ring variable that may be changed to point to a new ring on failure
$partitionRing = $this->partitionRing;
// Try to insert the jobs and update $partitionsTry on any failures.
// Retry to insert any remaining jobs again, ignoring the bad partitions.
$jobsLeft = $jobs;
for ( $i = $this->maxPartitionsTry; $i > 0 && count( $jobsLeft ); --$i ) {
try {
$partitionRing->getLiveLocationWeights();
} catch ( UnexpectedValueException $e ) {
break; // all servers down; nothing to insert to
}
$jobsLeft = $this->tryJobInsertions( $jobsLeft, $partitionRing, $flags );
}
if ( count( $jobsLeft ) ) {
throw new JobQueueError(
"Could not insert job(s), {$this->maxPartitionsTry} partitions tried." );
}
}
/**
* @param array $jobs
* @param HashRing &$partitionRing
* @param int $flags
* @throws JobQueueError
* @return IJobSpecification[] List of Job object that could not be inserted
*/
protected function tryJobInsertions( array $jobs, HashRing &$partitionRing, $flags ) {
$jobsLeft = [];
// Because jobs are spread across partitions, per-job de-duplication needs
// to use a consistent hash to avoid allowing duplicate jobs per partition.
// When inserting a batch of de-duplicated jobs, QOS_ATOMIC is disregarded.
$uJobsByPartition = []; // (partition name => job list)
/** @var Job $job */
foreach ( $jobs as $key => $job ) {
if ( $job->ignoreDuplicates() ) {
$sha1 = sha1( serialize( $job->getDeduplicationInfo() ) );
$uJobsByPartition[$partitionRing->getLiveLocation( $sha1 )][] = $job;
unset( $jobs[$key] );
}
}
// Get the batches of jobs that are not de-duplicated
if ( $flags & self::QOS_ATOMIC ) {
$nuJobBatches = [ $jobs ]; // all or nothing
} else {
// Split the jobs into batches and spread them out over servers if there
// are many jobs. This helps keep the partitions even. Otherwise, send all
// the jobs to a single partition queue to avoids the extra connections.
$nuJobBatches = array_chunk( $jobs, 300 );
}
// Insert the de-duplicated jobs into the queues...
foreach ( $uJobsByPartition as $partition => $jobBatch ) {
/** @var JobQueue $queue */
$queue = $this->partitionQueues[$partition];
try {
$ok = true;
$queue->doBatchPush( $jobBatch, $flags | self::QOS_ATOMIC );
} catch ( JobQueueError $e ) {
$ok = false;
$this->logException( $e );
}
if ( !$ok ) {
if ( !$partitionRing->ejectFromLiveRing( $partition, 5 ) ) {
throw new JobQueueError( "Could not insert job(s), no partitions available." );
}
$jobsLeft = array_merge( $jobsLeft, $jobBatch ); // not inserted
}
}
// Insert the jobs that are not de-duplicated into the queues...
foreach ( $nuJobBatches as $jobBatch ) {
$partition = ArrayUtils::pickRandom( $partitionRing->getLiveLocationWeights() );
$queue = $this->partitionQueues[$partition];
try {
$ok = true;
$queue->doBatchPush( $jobBatch, $flags | self::QOS_ATOMIC );
} catch ( JobQueueError $e ) {
$ok = false;
$this->logException( $e );
}
if ( !$ok ) {
if ( !$partitionRing->ejectFromLiveRing( $partition, 5 ) ) {
throw new JobQueueError( "Could not insert job(s), no partitions available." );
}
$jobsLeft = array_merge( $jobsLeft, $jobBatch ); // not inserted
}
}
return $jobsLeft;
}
protected function doPop() {
$partitionsTry = $this->partitionRing->getLiveLocationWeights(); // (partition => weight)
$failed = 0;
while ( count( $partitionsTry ) ) {
$partition = ArrayUtils::pickRandom( $partitionsTry );
if ( $partition === false ) {
break; // all partitions at 0 weight
}
/** @var JobQueue $queue */
$queue = $this->partitionQueues[$partition];
try {
$job = $queue->pop();
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
$job = false;
}
if ( $job ) {
$job->setMetadata( 'QueuePartition', $partition );
return $job;
} else {
unset( $partitionsTry[$partition] );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
return false;
}
protected function doAck( RunnableJob $job ) {
$partition = $job->getMetadata( 'QueuePartition' );
if ( $partition === null ) {
throw new UnexpectedValueException( "The given job has no defined partition name." );
}
$this->partitionQueues[$partition]->ack( $job );
}
protected function doIsRootJobOldDuplicate( IJobSpecification $job ) {
$signature = $job->getRootJobParams()['rootJobSignature'];
$partition = $this->partitionRing->getLiveLocation( $signature );
try {
return $this->partitionQueues[$partition]->doIsRootJobOldDuplicate( $job );
} catch ( JobQueueError $e ) {
if ( $this->partitionRing->ejectFromLiveRing( $partition, 5 ) ) {
$partition = $this->partitionRing->getLiveLocation( $signature );
return $this->partitionQueues[$partition]->doIsRootJobOldDuplicate( $job );
}
}
return false;
}
protected function doDeduplicateRootJob( IJobSpecification $job ) {
$signature = $job->getRootJobParams()['rootJobSignature'];
$partition = $this->partitionRing->getLiveLocation( $signature );
try {
return $this->partitionQueues[$partition]->doDeduplicateRootJob( $job );
} catch ( JobQueueError $e ) {
if ( $this->partitionRing->ejectFromLiveRing( $partition, 5 ) ) {
$partition = $this->partitionRing->getLiveLocation( $signature );
return $this->partitionQueues[$partition]->doDeduplicateRootJob( $job );
}
}
return false;
}
protected function doDelete() {
$failed = 0;
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
try {
$queue->doDelete();
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
return true;
}
protected function doWaitForBackups() {
$failed = 0;
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
try {
$queue->waitForBackups();
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
}
protected function doFlushCaches() {
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
$queue->doFlushCaches();
}
}
public function getAllQueuedJobs() {
$iterator = new AppendIterator();
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
$iterator->append( $queue->getAllQueuedJobs() );
}
return $iterator;
}
public function getAllDelayedJobs() {
$iterator = new AppendIterator();
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
$iterator->append( $queue->getAllDelayedJobs() );
}
return $iterator;
}
public function getAllAcquiredJobs() {
$iterator = new AppendIterator();
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
$iterator->append( $queue->getAllAcquiredJobs() );
}
return $iterator;
}
public function getAllAbandonedJobs() {
$iterator = new AppendIterator();
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
$iterator->append( $queue->getAllAbandonedJobs() );
}
return $iterator;
}
public function getCoalesceLocationInternal() {
return "JobQueueFederated:wiki:{$this->domain}" .
sha1( serialize( array_keys( $this->partitionQueues ) ) );
}
protected function doGetSiblingQueuesWithJobs( array $types ) {
$result = [];
$failed = 0;
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
try {
$nonEmpty = $queue->doGetSiblingQueuesWithJobs( $types );
if ( is_array( $nonEmpty ) ) {
$result = array_unique( array_merge( $result, $nonEmpty ) );
} else {
return null; // not supported on all partitions; bail
}
if ( count( $result ) == count( $types ) ) {
break; // short-circuit
}
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
return array_values( $result );
}
protected function doGetSiblingQueueSizes( array $types ) {
$result = [];
$failed = 0;
/** @var JobQueue $queue */
foreach ( $this->partitionQueues as $queue ) {
try {
$sizes = $queue->doGetSiblingQueueSizes( $types );
if ( is_array( $sizes ) ) {
foreach ( $sizes as $type => $size ) {
$result[$type] = ( $result[$type] ?? 0 ) + $size;
}
} else {
return null; // not supported on all partitions; bail
}
} catch ( JobQueueError $e ) {
++$failed;
$this->logException( $e );
}
}
$this->throwErrorIfAllPartitionsDown( $failed );
return $result;
}
protected function logException( Exception $e ) {
wfDebugLog( 'JobQueue', $e->getMessage() . "\n" . $e->getTraceAsString() );
}
/**
* Throw an error if no partitions available
*
* @param int $down The number of up partitions down
* @return void
* @throws JobQueueError
*/
protected function throwErrorIfAllPartitionsDown( $down ) {
if ( $down >= count( $this->partitionQueues ) ) {
throw new JobQueueError( 'No queue partitions available.' );
}
}
}
|