includes/libs/rdbms/lbfactory/LBFactory.php

   1 <?php
   2 /**
   3  * Generator and manager of database load balancing objects
   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  * @ingroup Database
  22  */
  23
  24 namespace Wikimedia\Rdbms;
  25
  26 use Psr\Log\LoggerInterface;
  27 use Wikimedia\ScopedCallback;
  28 use BagOStuff;
  29 use EmptyBagOStuff;
  30 use WANObjectCache;
  31 use Exception;
  32 use RuntimeException;
  33
  34 /**
  35  * An interface for generating database load balancers
  36  * @ingroup Database
  37  */
  38 abstract class LBFactory implements ILBFactory {
  39         /** @var ChronologyProtector */
  40         protected $chronProt;
  41         /** @var object|string Class name or object With profileIn/profileOut methods */
  42         protected $profiler;
  43         /** @var TransactionProfiler */
  44         protected $trxProfiler;
  45         /** @var LoggerInterface */
  46         protected $replLogger;
  47         /** @var LoggerInterface */
  48         protected $connLogger;
  49         /** @var LoggerInterface */
  50         protected $queryLogger;
  51         /** @var LoggerInterface */
  52         protected $perfLogger;
  53         /** @var callable Error logger */
  54         protected $errorLogger;
  55         /** @var BagOStuff */
  56         protected $srvCache;
  57         /** @var BagOStuff */
  58         protected $memStash;
  59         /** @var WANObjectCache */
  60         protected $wanCache;
  61
  62         /** @var DatabaseDomain Local domain */
  63         protected $localDomain;
  64         /** @var string Local hostname of the app server */
  65         protected $hostname;
  66         /** @var array Web request information about the client */
  67         protected $requestInfo;
  68
  69         /** @var mixed */
  70         protected $ticket;
  71         /** @var string|bool String if a requested DBO_TRX transaction round is active */
  72         protected $trxRoundId = false;
  73         /** @var string|bool Reason all LBs are read-only or false if not */
  74         protected $readOnlyReason = false;
  75         /** @var callable[] */
  76         protected $replicationWaitCallbacks = [];
  77
  78         /** @var bool Whether this PHP instance is for a CLI script */
  79         protected $cliMode;
  80         /** @var string Agent name for query profiling */
  81         protected $agent;
  82
  83         private static $loggerFields =
  84                 [ 'replLogger', 'connLogger', 'queryLogger', 'perfLogger' ];
  85
  86         public function __construct( array $conf ) {
  87                 $this->localDomain = isset( $conf['localDomain'] )
  88                         ? DatabaseDomain::newFromId( $conf['localDomain'] )
  89                         : DatabaseDomain::newUnspecified();
  90
  91                 if ( isset( $conf['readOnlyReason'] ) && is_string( $conf['readOnlyReason'] ) ) {
  92                         $this->readOnlyReason = $conf['readOnlyReason'];
  93                 }
  94
  95                 $this->srvCache = isset( $conf['srvCache'] ) ? $conf['srvCache'] : new EmptyBagOStuff();
  96                 $this->memStash = isset( $conf['memStash'] ) ? $conf['memStash'] : new EmptyBagOStuff();
  97                 $this->wanCache = isset( $conf['wanCache'] )
  98                         ? $conf['wanCache']
  99                         : WANObjectCache::newEmpty();
 100
 101                 foreach ( self::$loggerFields as $key ) {
 102                         $this->$key = isset( $conf[$key] ) ? $conf[$key] : new \Psr\Log\NullLogger();
 103                 }
 104                 $this->errorLogger = isset( $conf['errorLogger'] )
 105                         ? $conf['errorLogger']
 106                         : function ( Exception $e ) {
 107                                 trigger_error( E_USER_WARNING, get_class( $e ) . ': ' . $e->getMessage() );
 108                         };
 109
 110                 $this->profiler = isset( $conf['profiler'] ) ? $conf['profiler'] : null;
 111                 $this->trxProfiler = isset( $conf['trxProfiler'] )
 112                         ? $conf['trxProfiler']
 113                         : new TransactionProfiler();
 114
 115                 $this->requestInfo = [
 116                         'IPAddress' => isset( $_SERVER[ 'REMOTE_ADDR' ] ) ? $_SERVER[ 'REMOTE_ADDR' ] : '',
 117                         'UserAgent' => isset( $_SERVER['HTTP_USER_AGENT'] ) ? $_SERVER['HTTP_USER_AGENT'] : '',
 118                         'ChronologyProtection' => 'true'
 119                 ];
 120
 121                 $this->cliMode = isset( $conf['cliMode'] ) ? $conf['cliMode'] : PHP_SAPI === 'cli';
 122                 $this->hostname = isset( $conf['hostname'] ) ? $conf['hostname'] : gethostname();
 123                 $this->agent = isset( $conf['agent'] ) ? $conf['agent'] : '';
 124
 125                 $this->ticket = mt_rand();
 126         }
 127
 128         public function destroy() {
 129                 $this->shutdown( self::SHUTDOWN_NO_CHRONPROT );
 130                 $this->forEachLBCallMethod( 'disable' );
 131         }
 132
 133         public function shutdown(
 134                 $mode = self::SHUTDOWN_CHRONPROT_SYNC, callable $workCallback = null
 135         ) {
 136                 $chronProt = $this->getChronologyProtector();
 137                 if ( $mode === self::SHUTDOWN_CHRONPROT_SYNC ) {
 138                         $this->shutdownChronologyProtector( $chronProt, $workCallback, 'sync' );
 139                 } elseif ( $mode === self::SHUTDOWN_CHRONPROT_ASYNC ) {
 140                         $this->shutdownChronologyProtector( $chronProt, null, 'async' );
 141                 }
 142
 143                 $this->commitMasterChanges( __METHOD__ ); // sanity
 144         }
 145
 146         /**
 147          * @see ILBFactory::newMainLB()
 148          * @param bool $domain
 149          * @return LoadBalancer
 150          */
 151         abstract public function newMainLB( $domain = false );
 152
 153         /**
 154          * @see ILBFactory::getMainLB()
 155          * @param bool $domain
 156          * @return LoadBalancer
 157          */
 158         abstract public function getMainLB( $domain = false );
 159
 160         /**
 161          * @see ILBFactory::newExternalLB()
 162          * @param string $cluster
 163          * @return LoadBalancer
 164          */
 165         abstract public function newExternalLB( $cluster );
 166
 167         /**
 168          * @see ILBFactory::getExternalLB()
 169          * @param string $cluster
 170          * @return LoadBalancer
 171          */
 172         abstract public function getExternalLB( $cluster );
 173
 174         /**
 175          * Call a method of each tracked load balancer
 176          *
 177          * @param string $methodName
 178          * @param array $args
 179          */
 180         protected function forEachLBCallMethod( $methodName, array $args = [] ) {
 181                 $this->forEachLB(
 182                         function ( ILoadBalancer $loadBalancer, $methodName, array $args ) {
 183                                 call_user_func_array( [ $loadBalancer, $methodName ], $args );
 184                         },
 185                         [ $methodName, $args ]
 186                 );
 187         }
 188
 189         public function flushReplicaSnapshots( $fname = __METHOD__ ) {
 190                 $this->forEachLBCallMethod( 'flushReplicaSnapshots', [ $fname ] );
 191         }
 192
 193         public function commitAll( $fname = __METHOD__, array $options = [] ) {
 194                 $this->commitMasterChanges( $fname, $options );
 195                 $this->forEachLBCallMethod( 'commitAll', [ $fname ] );
 196         }
 197
 198         public function beginMasterChanges( $fname = __METHOD__ ) {
 199                 if ( $this->trxRoundId !== false ) {
 200                         throw new DBTransactionError(
 201                                 null,
 202                                 "$fname: transaction round '{$this->trxRoundId}' already started."
 203                         );
 204                 }
 205                 $this->trxRoundId = $fname;
 206                 // Set DBO_TRX flags on all appropriate DBs
 207                 $this->forEachLBCallMethod( 'beginMasterChanges', [ $fname ] );
 208         }
 209
 210         public function commitMasterChanges( $fname = __METHOD__, array $options = [] ) {
 211                 if ( $this->trxRoundId !== false && $this->trxRoundId !== $fname ) {
 212                         throw new DBTransactionError(
 213                                 null,
 214                                 "$fname: transaction round '{$this->trxRoundId}' still running."
 215                         );
 216                 }
 217                 /** @noinspection PhpUnusedLocalVariableInspection */
 218                 $scope = $this->getScopedPHPBehaviorForCommit(); // try to ignore client aborts
 219                 // Run pre-commit callbacks and suppress post-commit callbacks, aborting on failure
 220                 $this->forEachLBCallMethod( 'finalizeMasterChanges' );
 221                 $this->trxRoundId = false;
 222                 // Perform pre-commit checks, aborting on failure
 223                 $this->forEachLBCallMethod( 'approveMasterChanges', [ $options ] );
 224                 // Log the DBs and methods involved in multi-DB transactions
 225                 $this->logIfMultiDbTransaction();
 226                 // Actually perform the commit on all master DB connections and revert DBO_TRX
 227                 $this->forEachLBCallMethod( 'commitMasterChanges', [ $fname ] );
 228                 // Run all post-commit callbacks
 229                 /** @var Exception $e */
 230                 $e = null; // first callback exception
 231                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( &$e ) {
 232                         $ex = $lb->runMasterPostTrxCallbacks( IDatabase::TRIGGER_COMMIT );
 233                         $e = $e ?: $ex;
 234                 } );
 235                 // Commit any dangling DBO_TRX transactions from callbacks on one DB to another DB
 236                 $this->forEachLBCallMethod( 'commitMasterChanges', [ $fname ] );
 237                 // Throw any last post-commit callback error
 238                 if ( $e instanceof Exception ) {
 239                         throw $e;
 240                 }
 241         }
 242
 243         public function rollbackMasterChanges( $fname = __METHOD__ ) {
 244                 $this->trxRoundId = false;
 245                 $this->forEachLBCallMethod( 'suppressTransactionEndCallbacks' );
 246                 $this->forEachLBCallMethod( 'rollbackMasterChanges', [ $fname ] );
 247                 // Run all post-rollback callbacks
 248                 $this->forEachLB( function ( ILoadBalancer $lb ) {
 249                         $lb->runMasterPostTrxCallbacks( IDatabase::TRIGGER_ROLLBACK );
 250                 } );
 251         }
 252
 253         public function hasTransactionRound() {
 254                 return ( $this->trxRoundId !== false );
 255         }
 256
 257         /**
 258          * Log query info if multi DB transactions are going to be committed now
 259          */
 260         private function logIfMultiDbTransaction() {
 261                 $callersByDB = [];
 262                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( &$callersByDB ) {
 263                         $masterName = $lb->getServerName( $lb->getWriterIndex() );
 264                         $callers = $lb->pendingMasterChangeCallers();
 265                         if ( $callers ) {
 266                                 $callersByDB[$masterName] = $callers;
 267                         }
 268                 } );
 269
 270                 if ( count( $callersByDB ) >= 2 ) {
 271                         $dbs = implode( ', ', array_keys( $callersByDB ) );
 272                         $msg = "Multi-DB transaction [{$dbs}]:\n";
 273                         foreach ( $callersByDB as $db => $callers ) {
 274                                 $msg .= "$db: " . implode( '; ', $callers ) . "\n";
 275                         }
 276                         $this->queryLogger->info( $msg );
 277                 }
 278         }
 279
 280         public function hasMasterChanges() {
 281                 $ret = false;
 282                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( &$ret ) {
 283                         $ret = $ret || $lb->hasMasterChanges();
 284                 } );
 285
 286                 return $ret;
 287         }
 288
 289         public function laggedReplicaUsed() {
 290                 $ret = false;
 291                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( &$ret ) {
 292                         $ret = $ret || $lb->laggedReplicaUsed();
 293                 } );
 294
 295                 return $ret;
 296         }
 297
 298         public function hasOrMadeRecentMasterChanges( $age = null ) {
 299                 $ret = false;
 300                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( $age, &$ret ) {
 301                         $ret = $ret || $lb->hasOrMadeRecentMasterChanges( $age );
 302                 } );
 303                 return $ret;
 304         }
 305
 306         public function waitForReplication( array $opts = [] ) {
 307                 $opts += [
 308                         'domain' => false,
 309                         'cluster' => false,
 310                         'timeout' => 60,
 311                         'ifWritesSince' => null
 312                 ];
 313
 314                 if ( $opts['domain'] === false && isset( $opts['wiki'] ) ) {
 315                         $opts['domain'] = $opts['wiki']; // b/c
 316                 }
 317
 318                 // Figure out which clusters need to be checked
 319                 /** @var ILoadBalancer[] $lbs */
 320                 $lbs = [];
 321                 if ( $opts['cluster'] !== false ) {
 322                         $lbs[] = $this->getExternalLB( $opts['cluster'] );
 323                 } elseif ( $opts['domain'] !== false ) {
 324                         $lbs[] = $this->getMainLB( $opts['domain'] );
 325                 } else {
 326                         $this->forEachLB( function ( ILoadBalancer $lb ) use ( &$lbs ) {
 327                                 $lbs[] = $lb;
 328                         } );
 329                         if ( !$lbs ) {
 330                                 return; // nothing actually used
 331                         }
 332                 }
 333
 334                 // Get all the master positions of applicable DBs right now.
 335                 // This can be faster since waiting on one cluster reduces the
 336                 // time needed to wait on the next clusters.
 337                 $masterPositions = array_fill( 0, count( $lbs ), false );
 338                 foreach ( $lbs as $i => $lb ) {
 339                         if ( $lb->getServerCount() <= 1 ) {
 340                                 // T29975 - Don't try to wait for replica DBs if there are none
 341                                 // Prevents permission error when getting master position
 342                                 continue;
 343                         } elseif ( $opts['ifWritesSince']
 344                                 && $lb->lastMasterChangeTimestamp() < $opts['ifWritesSince']
 345                         ) {
 346                                 continue; // no writes since the last wait
 347                         }
 348                         $masterPositions[$i] = $lb->getMasterPos();
 349                 }
 350
 351                 // Run any listener callbacks *after* getting the DB positions. The more
 352                 // time spent in the callbacks, the less time is spent in waitForAll().
 353                 foreach ( $this->replicationWaitCallbacks as $callback ) {
 354                         $callback();
 355                 }
 356
 357                 $failed = [];
 358                 foreach ( $lbs as $i => $lb ) {
 359                         if ( $masterPositions[$i] ) {
 360                                 // The DBMS may not support getMasterPos()
 361                                 if ( !$lb->waitForAll( $masterPositions[$i], $opts['timeout'] ) ) {
 362                                         $failed[] = $lb->getServerName( $lb->getWriterIndex() );
 363                                 }
 364                         }
 365                 }
 366
 367                 if ( $failed ) {
 368                         throw new DBReplicationWaitError(
 369                                 null,
 370                                 "Could not wait for replica DBs to catch up to " .
 371                                 implode( ', ', $failed )
 372                         );
 373                 }
 374         }
 375
 376         public function setWaitForReplicationListener( $name, callable $callback = null ) {
 377                 if ( $callback ) {
 378                         $this->replicationWaitCallbacks[$name] = $callback;
 379                 } else {
 380                         unset( $this->replicationWaitCallbacks[$name] );
 381                 }
 382         }
 383
 384         public function getEmptyTransactionTicket( $fname ) {
 385                 if ( $this->hasMasterChanges() ) {
 386                         $this->queryLogger->error( __METHOD__ . ": $fname does not have outer scope.\n" .
 387                                 ( new RuntimeException() )->getTraceAsString() );
 388
 389                         return null;
 390                 }
 391
 392                 return $this->ticket;
 393         }
 394
 395         public function commitAndWaitForReplication( $fname, $ticket, array $opts = [] ) {
 396                 if ( $ticket !== $this->ticket ) {
 397                         $this->perfLogger->error( __METHOD__ . ": $fname does not have outer scope.\n" .
 398                                 ( new RuntimeException() )->getTraceAsString() );
 399
 400                         return;
 401                 }
 402
 403                 // The transaction owner and any caller with the empty transaction ticket can commit
 404                 // so that getEmptyTransactionTicket() callers don't risk seeing DBTransactionError.
 405                 if ( $this->trxRoundId !== false && $fname !== $this->trxRoundId ) {
 406                         $this->queryLogger->info( "$fname: committing on behalf of {$this->trxRoundId}." );
 407                         $fnameEffective = $this->trxRoundId;
 408                 } else {
 409                         $fnameEffective = $fname;
 410                 }
 411
 412                 $this->commitMasterChanges( $fnameEffective );
 413                 $this->waitForReplication( $opts );
 414                 // If a nested caller committed on behalf of $fname, start another empty $fname
 415                 // transaction, leaving the caller with the same empty transaction state as before.
 416                 if ( $fnameEffective !== $fname ) {
 417                         $this->beginMasterChanges( $fnameEffective );
 418                 }
 419         }
 420
 421         public function getChronologyProtectorTouched( $dbName ) {
 422                 return $this->getChronologyProtector()->getTouched( $dbName );
 423         }
 424
 425         public function disableChronologyProtection() {
 426                 $this->getChronologyProtector()->setEnabled( false );
 427         }
 428
 429         /**
 430          * @return ChronologyProtector
 431          */
 432         protected function getChronologyProtector() {
 433                 if ( $this->chronProt ) {
 434                         return $this->chronProt;
 435                 }
 436
 437                 $this->chronProt = new ChronologyProtector(
 438                         $this->memStash,
 439                         [
 440                                 'ip' => $this->requestInfo['IPAddress'],
 441                                 'agent' => $this->requestInfo['UserAgent'],
 442                         ],
 443                         isset( $_GET['cpPosTime'] ) ? $_GET['cpPosTime'] : null
 444                 );
 445                 $this->chronProt->setLogger( $this->replLogger );
 446
 447                 if ( $this->cliMode ) {
 448                         $this->chronProt->setEnabled( false );
 449                 } elseif ( $this->requestInfo['ChronologyProtection'] === 'false' ) {
 450                         // Request opted out of using position wait logic. This is useful for requests
 451                         // done by the job queue or background ETL that do not have a meaningful session.
 452                         $this->chronProt->setWaitEnabled( false );
 453                 }
 454
 455                 $this->replLogger->debug( __METHOD__ . ': using request info ' .
 456                         json_encode( $this->requestInfo, JSON_PRETTY_PRINT ) );
 457
 458                 return $this->chronProt;
 459         }
 460
 461         /**
 462          * Get and record all of the staged DB positions into persistent memory storage
 463          *
 464          * @param ChronologyProtector $cp
 465          * @param callable|null $workCallback Work to do instead of waiting on syncing positions
 466          * @param string $mode One of (sync, async); whether to wait on remote datacenters
 467          */
 468         protected function shutdownChronologyProtector(
 469                 ChronologyProtector $cp, $workCallback, $mode
 470         ) {
 471                 // Record all the master positions needed
 472                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( $cp ) {
 473                         $cp->shutdownLB( $lb );
 474                 } );
 475                 // Write them to the persistent stash. Try to do something useful by running $work
 476                 // while ChronologyProtector waits for the stash write to replicate to all DCs.
 477                 $unsavedPositions = $cp->shutdown( $workCallback, $mode );
 478                 if ( $unsavedPositions && $workCallback ) {
 479                         // Invoke callback in case it did not cache the result yet
 480                         $workCallback(); // work now to block for less time in waitForAll()
 481                 }
 482                 // If the positions failed to write to the stash, at least wait on local datacenter
 483                 // replica DBs to catch up before responding. Even if there are several DCs, this increases
 484                 // the chance that the user will see their own changes immediately afterwards. As long
 485                 // as the sticky DC cookie applies (same domain), this is not even an issue.
 486                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( $unsavedPositions ) {
 487                         $masterName = $lb->getServerName( $lb->getWriterIndex() );
 488                         if ( isset( $unsavedPositions[$masterName] ) ) {
 489                                 $lb->waitForAll( $unsavedPositions[$masterName] );
 490                         }
 491                 } );
 492         }
 493
 494         /**
 495          * Base parameters to LoadBalancer::__construct()
 496          * @return array
 497          */
 498         final protected function baseLoadBalancerParams() {
 499                 return [
 500                         'localDomain' => $this->localDomain,
 501                         'readOnlyReason' => $this->readOnlyReason,
 502                         'srvCache' => $this->srvCache,
 503                         'wanCache' => $this->wanCache,
 504                         'profiler' => $this->profiler,
 505                         'trxProfiler' => $this->trxProfiler,
 506                         'queryLogger' => $this->queryLogger,
 507                         'connLogger' => $this->connLogger,
 508                         'replLogger' => $this->replLogger,
 509                         'errorLogger' => $this->errorLogger,
 510                         'hostname' => $this->hostname,
 511                         'cliMode' => $this->cliMode,
 512                         'agent' => $this->agent,
 513                         'chronologyProtector' => $this->getChronologyProtector()
 514                 ];
 515         }
 516
 517         /**
 518          * @param ILoadBalancer $lb
 519          */
 520         protected function initLoadBalancer( ILoadBalancer $lb ) {
 521                 if ( $this->trxRoundId !== false ) {
 522                         $lb->beginMasterChanges( $this->trxRoundId ); // set DBO_TRX
 523                 }
 524         }
 525
 526         public function setDomainPrefix( $prefix ) {
 527                 $this->localDomain = new DatabaseDomain(
 528                         $this->localDomain->getDatabase(),
 529                         null,
 530                         $prefix
 531                 );
 532
 533                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( $prefix ) {
 534                         $lb->setDomainPrefix( $prefix );
 535                 } );
 536         }
 537
 538         public function closeAll() {
 539                 $this->forEachLBCallMethod( 'closeAll', [] );
 540         }
 541
 542         public function setAgentName( $agent ) {
 543                 $this->agent = $agent;
 544         }
 545
 546         public function appendPreShutdownTimeAsQuery( $url, $time ) {
 547                 $usedCluster = 0;
 548                 $this->forEachLB( function ( ILoadBalancer $lb ) use ( &$usedCluster ) {
 549                         $usedCluster |= ( $lb->getServerCount() > 1 );
 550                 } );
 551
 552                 if ( !$usedCluster ) {
 553                         return $url; // no master/replica clusters touched
 554                 }
 555
 556                 return strpos( $url, '?' ) === false ? "$url?cpPosTime=$time" : "$url&cpPosTime=$time";
 557         }
 558
 559         public function setRequestInfo( array $info ) {
 560                 $this->requestInfo = $info + $this->requestInfo;
 561         }
 562
 563         /**
 564          * Make PHP ignore user aborts/disconnects until the returned
 565          * value leaves scope. This returns null and does nothing in CLI mode.
 566          *
 567          * @return ScopedCallback|null
 568          */
 569         final protected function getScopedPHPBehaviorForCommit() {
 570                 if ( PHP_SAPI != 'cli' ) { // https://bugs.php.net/bug.php?id=47540
 571                         $old = ignore_user_abort( true ); // avoid half-finished operations
 572                         return new ScopedCallback( function () use ( $old ) {
 573                                 ignore_user_abort( $old );
 574                         } );
 575                 }
 576
 577                 return null;
 578         }
 579
 580         function __destruct() {
 581                 $this->destroy();
 582         }
 583 }
 584
 585 class_alias( LBFactory::class, 'LBFactory' );