includes/libs/rdbms/loadmonitor/LoadMonitor.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Database
  20  */
  21
  22 use Psr\Log\LoggerInterface;
  23 use Wikimedia\ScopedCallback;
  24
  25 /**
  26  * Basic DB load monitor with no external dependencies
  27  * Uses memcached to cache the replication lag for a short time
  28  *
  29  * @ingroup Database
  30  */
  31 class LoadMonitor implements ILoadMonitor {
  32         /** @var ILoadBalancer */
  33         protected $parent;
  34         /** @var BagOStuff */
  35         protected $srvCache;
  36         /** @var BagOStuff */
  37         protected $mainCache;
  38         /** @var LoggerInterface */
  39         protected $replLogger;
  40
  41         /** @var float Moving average ratio (e.g. 0.1 for 10% weight to new weight) */
  42         private $movingAveRatio;
  43
  44         const VERSION = 1; // cache key version
  45
  46         public function __construct(
  47                 ILoadBalancer $lb, BagOStuff $srvCache, BagOStuff $cache, array $options = []
  48         ) {
  49                 $this->parent = $lb;
  50                 $this->srvCache = $srvCache;
  51                 $this->mainCache = $cache;
  52                 $this->replLogger = new \Psr\Log\NullLogger();
  53
  54                 $this->movingAveRatio = isset( $options['movingAveRatio'] )
  55                         ? $options['movingAveRatio']
  56                         : 0.1;
  57         }
  58
  59         public function setLogger( LoggerInterface $logger ) {
  60                 $this->replLogger = $logger;
  61         }
  62
  63         public function scaleLoads( array &$weightByServer, $domain ) {
  64                 $serverIndexes = array_keys( $weightByServer );
  65                 $states = $this->getServerStates( $serverIndexes, $domain );
  66                 $coefficientsByServer = $states['weightScales'];
  67                 foreach ( $weightByServer as $i => $weight ) {
  68                         if ( isset( $coefficientsByServer[$i] ) ) {
  69                                 $weightByServer[$i] = $weight * $coefficientsByServer[$i];
  70                         } else { // server recently added to config?
  71                                 $host = $this->parent->getServerName( $i );
  72                                 $this->replLogger->error( __METHOD__ . ": host $host not in cache" );
  73                         }
  74                 }
  75         }
  76
  77         public function getLagTimes( array $serverIndexes, $domain ) {
  78                 $states = $this->getServerStates( $serverIndexes, $domain );
  79
  80                 return $states['lagTimes'];
  81         }
  82
  83         protected function getServerStates( array $serverIndexes, $domain ) {
  84                 $writerIndex = $this->parent->getWriterIndex();
  85                 if ( count( $serverIndexes ) == 1 && reset( $serverIndexes ) == $writerIndex ) {
  86                         # Single server only, just return zero without caching
  87                         return [
  88                                 'lagTimes' => [ $writerIndex => 0 ],
  89                                 'weightScales' => [ $writerIndex => 1.0 ]
  90                         ];
  91                 }
  92
  93                 $key = $this->getCacheKey( $serverIndexes );
  94                 # Randomize TTLs to reduce stampedes (4.0 - 5.0 sec)
  95                 $ttl = mt_rand( 4e6, 5e6 ) / 1e6;
  96                 # Keep keys around longer as fallbacks
  97                 $staleTTL = 60;
  98
  99                 # (a) Check the local APC cache
 100                 $value = $this->srvCache->get( $key );
 101                 if ( $value && $value['timestamp'] > ( microtime( true ) - $ttl ) ) {
 102                         $this->replLogger->debug( __METHOD__ . ": got lag times ($key) from local cache" );
 103                         return $value; // cache hit
 104                 }
 105                 $staleValue = $value ?: false;
 106
 107                 # (b) Check the shared cache and backfill APC
 108                 $value = $this->mainCache->get( $key );
 109                 if ( $value && $value['timestamp'] > ( microtime( true ) - $ttl ) ) {
 110                         $this->srvCache->set( $key, $value, $staleTTL );
 111                         $this->replLogger->debug( __METHOD__ . ": got lag times ($key) from main cache" );
 112
 113                         return $value; // cache hit
 114                 }
 115                 $staleValue = $value ?: $staleValue;
 116
 117                 # (c) Cache key missing or expired; regenerate and backfill
 118                 if ( $this->mainCache->lock( $key, 0, 10 ) ) {
 119                         # Let this process alone update the cache value
 120                         $cache = $this->mainCache;
 121                         /** @noinspection PhpUnusedLocalVariableInspection */
 122                         $unlocker = new ScopedCallback( function () use ( $cache, $key ) {
 123                                 $cache->unlock( $key );
 124                         } );
 125                 } elseif ( $staleValue ) {
 126                         # Could not acquire lock but an old cache exists, so use it
 127                         return $staleValue;
 128                 }
 129
 130                 $lagTimes = [];
 131                 $weightScales = [];
 132                 $movAveRatio = $this->movingAveRatio;
 133                 foreach ( $serverIndexes as $i ) {
 134                         if ( $i == $this->parent->getWriterIndex() ) {
 135                                 $lagTimes[$i] = 0; // master always has no lag
 136                                 $weightScales[$i] = 1.0; // nominal weight
 137                                 continue;
 138                         }
 139
 140                         $conn = $this->parent->getAnyOpenConnection( $i );
 141                         if ( $conn ) {
 142                                 $close = false; // already open
 143                         } else {
 144                                 $conn = $this->parent->openConnection( $i, $domain );
 145                                 $close = true; // new connection
 146                         }
 147
 148                         $lastWeight = isset( $staleValue['weightScales'][$i] )
 149                                 ? $staleValue['weightScales'][$i]
 150                                 : 1.0;
 151                         $coefficient = $this->getWeightScale( $i, $conn ?: null );
 152                         $newWeight = $movAveRatio * $coefficient + ( 1 - $movAveRatio ) * $lastWeight;
 153
 154                         // Scale from 10% to 100% of nominal weight
 155                         $weightScales[$i] = max( $newWeight, .10 );
 156
 157                         if ( !$conn ) {
 158                                 $lagTimes[$i] = false;
 159                                 $host = $this->parent->getServerName( $i );
 160                                 $this->replLogger->error( __METHOD__ . ": host $host is unreachable" );
 161                                 continue;
 162                         }
 163
 164                         if ( $conn->getLBInfo( 'is static' ) ) {
 165                                 $lagTimes[$i] = 0;
 166                         } else {
 167                                 $lagTimes[$i] = $conn->getLag();
 168                                 if ( $lagTimes[$i] === false ) {
 169                                         $host = $this->parent->getServerName( $i );
 170                                         $this->replLogger->error( __METHOD__ . ": host $host is not replicating?" );
 171                                 }
 172                         }
 173
 174                         if ( $close ) {
 175                                 # Close the connection to avoid sleeper connections piling up.
 176                                 # Note that the caller will pick one of these DBs and reconnect,
 177                                 # which is slightly inefficient, but this only matters for the lag
 178                                 # time cache miss cache, which is far less common that cache hits.
 179                                 $this->parent->closeConnection( $conn );
 180                         }
 181                 }
 182
 183                 # Add a timestamp key so we know when it was cached
 184                 $value = [
 185                         'lagTimes' => $lagTimes,
 186                         'weightScales' => $weightScales,
 187                         'timestamp' => microtime( true )
 188                 ];
 189                 $this->mainCache->set( $key, $value, $staleTTL );
 190                 $this->srvCache->set( $key, $value, $staleTTL );
 191                 $this->replLogger->info( __METHOD__ . ": re-calculated lag times ($key)" );
 192
 193                 return $value;
 194         }
 195
 196         /**
 197          * @param integer $index Server index
 198          * @param IDatabase|null $conn Connection handle or null on connection failure
 199          * @return float
 200          */
 201         protected function getWeightScale( $index, IDatabase $conn = null ) {
 202                 return $conn ? 1.0 : 0.0;
 203         }
 204
 205         private function getCacheKey( array $serverIndexes ) {
 206                 sort( $serverIndexes );
 207                 // Lag is per-server, not per-DB, so key on the master DB name
 208                 return $this->srvCache->makeGlobalKey(
 209                         'lag-times',
 210                         self::VERSION,
 211                         $this->parent->getServerName( $this->parent->getWriterIndex() ),
 212                         implode( '-', $serverIndexes )
 213                 );
 214         }
 215 }