includes/deferred/DeferredUpdates.php

   1 <?php
   2 /**
   3  * Interface and manager for deferred updates.
   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  */
  22
  23 use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface;
  24 use MediaWiki\Logger\LoggerFactory;
  25 use Psr\Log\LoggerInterface;
  26 use Wikimedia\Rdbms\IDatabase;
  27 use MediaWiki\MediaWikiServices;
  28 use Wikimedia\Rdbms\LBFactory;
  29 use Wikimedia\Rdbms\ILBFactory;
  30 use Wikimedia\Rdbms\LoadBalancer;
  31 use Wikimedia\Rdbms\DBTransactionError;
  32
  33 /**
  34  * Class for managing the deferred updates
  35  *
  36  * In web request mode, deferred updates can be run at the end of the request, either before or
  37  * after the HTTP response has been sent. In either case, they run after the DB commit step. If
  38  * an update runs after the response is sent, it will not block clients. If sent before, it will
  39  * run synchronously. These two modes are defined via PRESEND and POSTSEND constants, the latter
  40  * being the default for addUpdate() and addCallableUpdate().
  41  *
  42  * Updates that work through this system will be more likely to complete by the time the client
  43  * makes their next request after this one than with the JobQueue system.
  44  *
  45  * In CLI mode, deferred updates will run:
  46  *   - a) During DeferredUpdates::addUpdate if no LBFactory DB handles have writes pending
  47  *   - b) On commit of an LBFactory DB handle if no other such handles have writes pending
  48  *   - c) During an LBFactory::waitForReplication call if no LBFactory DBs have writes pending
  49  *   - d) When the queue is large and an LBFactory DB handle commits (EnqueueableDataUpdate only)
  50  *   - e) At the completion of Maintenance::execute()
  51  *
  52  * @see Maintenance::setLBFactoryTriggers
  53  *
  54  * When updates are deferred, they go into one two FIFO "top-queues" (one for pre-send and one
  55  * for post-send). Updates enqueued *during* doUpdate() of a "top" update go into the "sub-queue"
  56  * for that update. After that method finishes, the sub-queue is run until drained. This continues
  57  * for each top-queue job until the entire top queue is drained. This happens for the pre-send
  58  * top-queue, and later on, the post-send top-queue, in execute().
  59  *
  60  * @since 1.19
  61  */
  62 class DeferredUpdates {
  63         /** @var DeferrableUpdate[] Updates to be deferred until before request end */
  64         private static $preSendUpdates = [];
  65         /** @var DeferrableUpdate[] Updates to be deferred until after request end */
  66         private static $postSendUpdates = [];
  67
  68         const ALL = 0; // all updates; in web requests, use only after flushing the output buffer
  69         const PRESEND = 1; // for updates that should run before flushing output buffer
  70         const POSTSEND = 2; // for updates that should run after flushing output buffer
  71
  72         const BIG_QUEUE_SIZE = 100;
  73
  74         /** @var array|null Information about the current execute() call or null if not running */
  75         private static $executeContext;
  76
  77         /**
  78          * Add an update to the deferred list to be run later by execute()
  79          *
  80          * In CLI mode, callback magic will also be used to run updates when safe
  81          *
  82          * @param DeferrableUpdate $update Some object that implements doUpdate()
  83          * @param int $stage DeferredUpdates constant (PRESEND or POSTSEND) (since 1.27)
  84          */
  85         public static function addUpdate( DeferrableUpdate $update, $stage = self::POSTSEND ) {
  86                 global $wgCommandLineMode;
  87
  88                 if (
  89                         self::$executeContext &&
  90                         self::$executeContext['stage'] >= $stage &&
  91                         !( $update instanceof MergeableUpdate )
  92                 ) {
  93                         // This is a sub-DeferredUpdate; run it right after its parent update.
  94                         // Also, while post-send updates are running, push any "pre-send" jobs to the
  95                         // active post-send queue to make sure they get run this round (or at all).
  96                         self::$executeContext['subqueue'][] = $update;
  97
  98                         return;
  99                 }
 100
 101                 if ( $stage === self::PRESEND ) {
 102                         self::push( self::$preSendUpdates, $update );
 103                 } else {
 104                         self::push( self::$postSendUpdates, $update );
 105                 }
 106
 107                 // Try to run the updates now if in CLI mode and no transaction is active.
 108                 // This covers scripts that don't/barely use the DB but make updates to other stores.
 109                 if ( $wgCommandLineMode ) {
 110                         self::tryOpportunisticExecute( 'run' );
 111                 }
 112         }
 113
 114         /**
 115          * Add a callable update. In a lot of cases, we just need a callback/closure,
 116          * defining a new DeferrableUpdate object is not necessary
 117          *
 118          * @see MWCallableUpdate::__construct()
 119          *
 120          * @param callable $callable
 121          * @param int $stage DeferredUpdates constant (PRESEND or POSTSEND) (since 1.27)
 122          * @param IDatabase|IDatabase[]|null $dbw Abort if this DB is rolled back [optional] (since 1.28)
 123          */
 124         public static function addCallableUpdate(
 125                 $callable, $stage = self::POSTSEND, $dbw = null
 126         ) {
 127                 self::addUpdate( new MWCallableUpdate( $callable, wfGetCaller(), $dbw ), $stage );
 128         }
 129
 130         /**
 131          * Do any deferred updates and clear the list
 132          *
 133          * If $stage is self::ALL then the queue of PRESEND updates will be resolved,
 134          * followed by the queue of POSTSEND updates
 135          *
 136          * @param string $mode Use "enqueue" to use the job queue when possible [Default: "run"]
 137          * @param int $stage DeferredUpdates constant (PRESEND, POSTSEND, or ALL) (since 1.27)
 138          */
 139         public static function doUpdates( $mode = 'run', $stage = self::ALL ) {
 140                 $stageEffective = ( $stage === self::ALL ) ? self::POSTSEND : $stage;
 141                 // For ALL mode, make sure that any PRESEND updates added along the way get run.
 142                 // Normally, these use the subqueue, but that isn't true for MergeableUpdate items.
 143                 do {
 144                         if ( $stage === self::ALL || $stage === self::PRESEND ) {
 145                                 self::handleUpdateQueue( self::$preSendUpdates, $mode, $stageEffective );
 146                         }
 147
 148                         if ( $stage === self::ALL || $stage == self::POSTSEND ) {
 149                                 self::handleUpdateQueue( self::$postSendUpdates, $mode, $stageEffective );
 150                         }
 151                 } while ( $stage === self::ALL && self::$preSendUpdates );
 152         }
 153
 154         /**
 155          * @param DeferrableUpdate[] $queue
 156          * @param DeferrableUpdate $update
 157          */
 158         private static function push( array &$queue, DeferrableUpdate $update ) {
 159                 if ( $update instanceof MergeableUpdate ) {
 160                         $class = get_class( $update ); // fully-qualified class
 161                         if ( isset( $queue[$class] ) ) {
 162                                 /** @var MergeableUpdate $existingUpdate */
 163                                 $existingUpdate = $queue[$class];
 164                                 $existingUpdate->merge( $update );
 165                                 // Move the update to the end to handle things like mergeable purge
 166                                 // updates that might depend on the prior updates in the queue running
 167                                 unset( $queue[$class] );
 168                                 $queue[$class] = $existingUpdate;
 169                         } else {
 170                                 $queue[$class] = $update;
 171                         }
 172                 } else {
 173                         $queue[] = $update;
 174                 }
 175         }
 176
 177         /**
 178          * Immediately run or enqueue a list of updates
 179          *
 180          * @param DeferrableUpdate[] &$queue List of DeferrableUpdate objects
 181          * @param string $mode Either "run" or "enqueue" (to use the job queue when possible)
 182          * @param int $stage Class constant (PRESEND, POSTSEND) (since 1.28)
 183          * @throws ErrorPageError Happens on top-level calls
 184          * @throws Exception Happens on second-level calls
 185          */
 186         protected static function handleUpdateQueue( array &$queue, $mode, $stage ) {
 187                 $services = MediaWikiServices::getInstance();
 188                 $stats = $services->getStatsdDataFactory();
 189                 $lbf = $services->getDBLoadBalancerFactory();
 190                 $logger = LoggerFactory::getInstance( 'DeferredUpdates' );
 191                 $httpMethod = $services->getMainConfig()->get( 'CommandLineMode' )
 192                         ? 'cli'
 193                         : strtolower( RequestContext::getMain()->getRequest()->getMethod() );
 194
 195                 /** @var ErrorPageError $guiEx */
 196                 $guiEx = null;
 197                 /** @var DeferrableUpdate[] $updates Snapshot of queue */
 198                 $updates = $queue;
 199
 200                 // Keep doing rounds of updates until none get enqueued...
 201                 while ( $updates ) {
 202                         $queue = []; // clear the queue
 203
 204                         // Segregate the queue into one for DataUpdate and one for everything else
 205                         $dataUpdateQueue = [];
 206                         $genericUpdateQueue = [];
 207                         foreach ( $updates as $update ) {
 208                                 if ( $update instanceof DataUpdate ) {
 209                                         $dataUpdateQueue[] = $update;
 210                                 } else {
 211                                         $genericUpdateQueue[] = $update;
 212                                 }
 213                         }
 214                         // Execute all DataUpdate queue followed by the DeferrableUpdate queue...
 215                         foreach ( [ $dataUpdateQueue, $genericUpdateQueue ] as $updateQueue ) {
 216                                 foreach ( $updateQueue as $du ) {
 217                                         // Enqueue the task into the job queue system instead if applicable
 218                                         if ( $mode === 'enqueue' && $du instanceof EnqueueableDataUpdate ) {
 219                                                 self::jobify( $du, $lbf, $logger, $stats, $httpMethod );
 220                                                 continue;
 221                                         }
 222                                         // Otherwise, execute the task and any subtasks that it spawns
 223                                         self::$executeContext = [ 'stage' => $stage, 'subqueue' => [] ];
 224                                         try {
 225                                                 $e = self::run( $du, $lbf, $logger, $stats, $httpMethod );
 226                                                 $guiEx = $guiEx ?: ( $e instanceof ErrorPageError ? $e : null );
 227                                                 // Do the subqueue updates for $update until there are none
 228                                                 while ( self::$executeContext['subqueue'] ) {
 229                                                         $duChild = reset( self::$executeContext['subqueue'] );
 230                                                         $firstKey = key( self::$executeContext['subqueue'] );
 231                                                         unset( self::$executeContext['subqueue'][$firstKey] );
 232
 233                                                         $e = self::run( $duChild, $lbf, $logger, $stats, $httpMethod );
 234                                                         $guiEx = $guiEx ?: ( $e instanceof ErrorPageError ? $e : null );
 235                                                 }
 236                                         } finally {
 237                                                 // Make sure we always clean up the context.
 238                                                 // Losing updates while rewinding the stack is acceptable,
 239                                                 // losing updates that are added later is not.
 240                                                 self::$executeContext = null;
 241                                         }
 242                                 }
 243                         }
 244
 245                         $updates = $queue; // new snapshot of queue (check for new entries)
 246                 }
 247
 248                 // Throw the first of any GUI errors as long as the context is HTTP pre-send. However,
 249                 // callers should check permissions *before* enqueueing updates. If the main transaction
 250                 // round actions succeed but some deferred updates fail due to permissions errors then
 251                 // there is a risk that some secondary data was not properly updated.
 252                 if ( $guiEx && $stage === self::PRESEND && !headers_sent() ) {
 253                         throw $guiEx;
 254                 }
 255         }
 256
 257         /**
 258          * Run a task and catch/log any exceptions
 259          *
 260          * @param DeferrableUpdate $update
 261          * @param LBFactory $lbFactory
 262          * @param LoggerInterface $logger
 263          * @param StatsdDataFactoryInterface $stats
 264          * @param string $httpMethod
 265          * @return Exception|Throwable|null
 266          */
 267         private static function run(
 268                 DeferrableUpdate $update,
 269                 LBFactory $lbFactory,
 270                 LoggerInterface $logger,
 271                 StatsdDataFactoryInterface $stats,
 272                 $httpMethod
 273         ) {
 274                 $name = get_class( $update );
 275                 $suffix = ( $update instanceof DeferrableCallback ) ? "_{$update->getOrigin()}" : '';
 276                 $stats->increment( "deferred_updates.$httpMethod.{$name}{$suffix}" );
 277
 278                 $e = null;
 279                 try {
 280                         self::attemptUpdate( $update, $lbFactory );
 281                 } catch ( Exception $e ) {
 282                 } catch ( Throwable $e ) {
 283                 }
 284
 285                 if ( $e ) {
 286                         $logger->error(
 287                                 "Deferred update {type} failed: {message}",
 288                                 [
 289                                         'type' => $name . $suffix,
 290                                         'message' => $e->getMessage(),
 291                                         'trace' => $e->getTraceAsString()
 292                                 ]
 293                         );
 294                         $lbFactory->rollbackMasterChanges( __METHOD__ );
 295                         // VW-style hack to work around T190178, so we can make sure
 296                         // PageMetaDataUpdater doesn't throw exceptions.
 297                         if ( defined( 'MW_PHPUNIT_TEST' ) ) {
 298                                 throw $e;
 299                         }
 300                 }
 301
 302                 return $e;
 303         }
 304
 305         /**
 306          * Push a task into the job queue system and catch/log any exceptions
 307          *
 308          * @param EnqueueableDataUpdate $update
 309          * @param LBFactory $lbFactory
 310          * @param LoggerInterface $logger
 311          * @param StatsdDataFactoryInterface $stats
 312          * @param string $httpMethod
 313          */
 314         private static function jobify(
 315                 EnqueueableDataUpdate $update,
 316                 LBFactory $lbFactory,
 317                 LoggerInterface $logger,
 318                 StatsdDataFactoryInterface $stats,
 319                 $httpMethod
 320         ) {
 321                 $stats->increment( "deferred_updates.$httpMethod." . get_class( $update ) );
 322
 323                 $e = null;
 324                 try {
 325                         $spec = $update->getAsJobSpecification();
 326                         JobQueueGroup::singleton( $spec['domain'] ?? $spec['wiki'] )->push( $spec['job'] );
 327                 } catch ( Exception $e ) {
 328                 } catch ( Throwable $e ) {
 329                 }
 330
 331                 if ( $e ) {
 332                         $logger->error(
 333                                 "Job insertion of deferred update {type} failed: {message}",
 334                                 [
 335                                         'type' => get_class( $update ),
 336                                         'message' => $e->getMessage(),
 337                                         'trace' => $e->getTraceAsString()
 338                                 ]
 339                         );
 340                         $lbFactory->rollbackMasterChanges( __METHOD__ );
 341                 }
 342         }
 343
 344         /**
 345          * Attempt to run an update with the appropriate transaction round state it expects
 346          *
 347          * DeferredUpdate classes that wrap the execution of bundles of other DeferredUpdate
 348          * instances can use this method to run the updates. Any such wrapper class should
 349          * always use TRX_ROUND_ABSENT itself.
 350          *
 351          * @param DeferrableUpdate $update
 352          * @param ILBFactory $lbFactory
 353          * @since 1.34
 354          */
 355         public static function attemptUpdate( DeferrableUpdate $update, ILBFactory $lbFactory ) {
 356                 $ticket = $lbFactory->getEmptyTransactionTicket( __METHOD__ );
 357                 if ( !$ticket || $lbFactory->hasTransactionRound() ) {
 358                         throw new DBTransactionError( null, "A database transaction round is pending." );
 359                 }
 360
 361                 if ( $update instanceof DataUpdate ) {
 362                         $update->setTransactionTicket( $ticket );
 363                 }
 364
 365                 $fnameTrxOwner = get_class( $update ) . '::doUpdate';
 366                 $useExplicitTrxRound = !(
 367                         $update instanceof TransactionRoundAwareUpdate &&
 368                         $update->getTransactionRoundRequirement() == $update::TRX_ROUND_ABSENT
 369                 );
 370                 // Flush any pending changes left over from an implicit transaction round
 371                 if ( $useExplicitTrxRound ) {
 372                         $lbFactory->beginMasterChanges( $fnameTrxOwner ); // new explicit round
 373                 } else {
 374                         $lbFactory->commitMasterChanges( $fnameTrxOwner ); // new implicit round
 375                 }
 376                 // Run the update after any stale master view snapshots have been flushed
 377                 $update->doUpdate();
 378                 // Commit any pending changes from the explicit or implicit transaction round
 379                 $lbFactory->commitMasterChanges( $fnameTrxOwner );
 380         }
 381
 382         /**
 383          * Run all deferred updates immediately if there are no DB writes active
 384          *
 385          * If there are many deferred updates pending, $mode is 'run', and there
 386          * are still busy LBFactory database handles, then any EnqueueableDataUpdate
 387          * tasks might be enqueued as jobs to be executed later.
 388          *
 389          * @param string $mode Use "enqueue" to use the job queue when possible
 390          * @return bool Whether updates were allowed to run
 391          * @since 1.28
 392          */
 393         public static function tryOpportunisticExecute( $mode = 'run' ) {
 394                 // execute() loop is already running
 395                 if ( self::$executeContext ) {
 396                         return false;
 397                 }
 398
 399                 // Avoiding running updates without them having outer scope
 400                 if ( !self::areDatabaseTransactionsActive() ) {
 401                         self::doUpdates( $mode );
 402                         return true;
 403                 }
 404
 405                 if ( self::pendingUpdatesCount() >= self::BIG_QUEUE_SIZE ) {
 406                         // If we cannot run the updates with outer transaction context, try to
 407                         // at least enqueue all the updates that support queueing to job queue
 408                         self::$preSendUpdates = self::enqueueUpdates( self::$preSendUpdates );
 409                         self::$postSendUpdates = self::enqueueUpdates( self::$postSendUpdates );
 410                 }
 411
 412                 return !self::pendingUpdatesCount();
 413         }
 414
 415         /**
 416          * Enqueue a job for each EnqueueableDataUpdate item and return the other items
 417          *
 418          * @param DeferrableUpdate[] $updates A list of deferred update instances
 419          * @return DeferrableUpdate[] Remaining updates that do not support being queued
 420          */
 421         private static function enqueueUpdates( array $updates ) {
 422                 $remaining = [];
 423
 424                 foreach ( $updates as $update ) {
 425                         if ( $update instanceof EnqueueableDataUpdate ) {
 426                                 $spec = $update->getAsJobSpecification();
 427                                 $domain = $spec['domain'] ?? $spec['wiki'];
 428                                 JobQueueGroup::singleton( $domain )->push( $spec['job'] );
 429                         } else {
 430                                 $remaining[] = $update;
 431                         }
 432                 }
 433
 434                 return $remaining;
 435         }
 436
 437         /**
 438          * @return int Number of enqueued updates
 439          * @since 1.28
 440          */
 441         public static function pendingUpdatesCount() {
 442                 return count( self::$preSendUpdates ) + count( self::$postSendUpdates );
 443         }
 444
 445         /**
 446          * @param int $stage DeferredUpdates constant (PRESEND, POSTSEND, or ALL)
 447          * @return DeferrableUpdate[]
 448          * @since 1.29
 449          */
 450         public static function getPendingUpdates( $stage = self::ALL ) {
 451                 $updates = [];
 452                 if ( $stage === self::ALL || $stage === self::PRESEND ) {
 453                         $updates = array_merge( $updates, self::$preSendUpdates );
 454                 }
 455                 if ( $stage === self::ALL || $stage === self::POSTSEND ) {
 456                         $updates = array_merge( $updates, self::$postSendUpdates );
 457                 }
 458                 return $updates;
 459         }
 460
 461         /**
 462          * Clear all pending updates without performing them. Generally, you don't
 463          * want or need to call this. Unit tests need it though.
 464          */
 465         public static function clearPendingUpdates() {
 466                 self::$preSendUpdates = [];
 467                 self::$postSendUpdates = [];
 468         }
 469
 470         /**
 471          * @return bool If a transaction round is active or connection is not ready for commit()
 472          */
 473         private static function areDatabaseTransactionsActive() {
 474                 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 475                 if ( $lbFactory->hasTransactionRound() || !$lbFactory->isReadyForRoundOperations() ) {
 476                         return true;
 477                 }
 478
 479                 $connsBusy = false;
 480                 $lbFactory->forEachLB( function ( LoadBalancer $lb ) use ( &$connsBusy ) {
 481                         $lb->forEachOpenMasterConnection( function ( IDatabase $conn ) use ( &$connsBusy ) {
 482                                 if ( $conn->writesOrCallbacksPending() || $conn->explicitTrxActive() ) {
 483                                         $connsBusy = true;
 484                                 }
 485                         } );
 486                 } );
 487
 488                 return $connsBusy;
 489         }
 490 }