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