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