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 namespace Wikimedia\Rdbms;
  23
  24 use Psr\Log\LoggerInterface;
  25 use Psr\Log\NullLogger;
  26 use Wikimedia\ScopedCallback;
  27 use BagOStuff;
  28
  29 /**
  30  * Basic DB load monitor with no external dependencies
  31  * Uses memcached to cache the replication lag for a short time
  32  *
  33  * @ingroup Database
  34  */
  35 class LoadMonitor implements ILoadMonitor {
  36         /** @var ILoadBalancer */
  37         protected $parent;
  38         /** @var BagOStuff */
  39         protected $srvCache;
  40         /** @var BagOStuff */
  41         protected $mainCache;
  42         /** @var LoggerInterface */
  43         protected $replLogger;
  44
  45         /** @var float Moving average ratio (e.g. 0.1 for 10% weight to new weight) */
  46         private $movingAveRatio;
  47
  48         const VERSION = 1; // cache key version
  49
  50         public function __construct(
  51                 ILoadBalancer $lb, BagOStuff $srvCache, BagOStuff $cache, array $options = []
  52         ) {
  53                 $this->parent = $lb;
  54                 $this->srvCache = $srvCache;
  55                 $this->mainCache = $cache;
  56                 $this->replLogger = new NullLogger();
  57
  58                 $this->movingAveRatio = isset( $options['movingAveRatio'] )
  59                         ? $options['movingAveRatio']
  60                         : 0.1;
  61         }
  62
  63         public function setLogger( LoggerInterface $logger ) {
  64                 $this->replLogger = $logger;
  65         }
  66
  67         public function scaleLoads( array &$weightByServer, $domain ) {
  68                 $serverIndexes = array_keys( $weightByServer );
  69                 $states = $this->getServerStates( $serverIndexes, $domain );
  70                 $coefficientsByServer = $states['weightScales'];
  71                 foreach ( $weightByServer as $i => $weight ) {
  72                         if ( isset( $coefficientsByServer[$i] ) ) {
  73                                 $weightByServer[$i] = $weight * $coefficientsByServer[$i];
  74                         } else { // server recently added to config?
  75                                 $host = $this->parent->getServerName( $i );
  76                                 $this->replLogger->error( __METHOD__ . ": host $host not in cache" );
  77                         }
  78                 }
  79         }
  80
  81         public function getLagTimes( array $serverIndexes, $domain ) {
  82                 $states = $this->getServerStates( $serverIndexes, $domain );
  83
  84                 return $states['lagTimes'];
  85         }
  86
  87         protected function getServerStates( array $serverIndexes, $domain ) {
  88                 $writerIndex = $this->parent->getWriterIndex();
  89                 if ( count( $serverIndexes ) == 1 && reset( $serverIndexes ) == $writerIndex ) {
  90                         # Single server only, just return zero without caching
  91                         return [
  92                                 'lagTimes' => [ $writerIndex => 0 ],
  93                                 'weightScales' => [ $writerIndex => 1.0 ]
  94                         ];
  95                 }
  96
  97                 $key = $this->getCacheKey( $serverIndexes );
  98                 # Randomize TTLs to reduce stampedes (4.0 - 5.0 sec)
  99                 $ttl = mt_rand( 4e6, 5e6 ) / 1e6;
 100                 # Keep keys around longer as fallbacks
 101                 $staleTTL = 60;
 102
 103                 # (a) Check the local APC cache
 104                 $value = $this->srvCache->get( $key );
 105                 if ( $value && $value['timestamp'] > ( microtime( true ) - $ttl ) ) {
 106                         $this->replLogger->debug( __METHOD__ . ": got lag times ($key) from local cache" );
 107                         return $value; // cache hit
 108                 }
 109                 $staleValue = $value ?: false;
 110
 111                 # (b) Check the shared cache and backfill APC
 112                 $value = $this->mainCache->get( $key );
 113                 if ( $value && $value['timestamp'] > ( microtime( true ) - $ttl ) ) {
 114                         $this->srvCache->set( $key, $value, $staleTTL );
 115                         $this->replLogger->debug( __METHOD__ . ": got lag times ($key) from main cache" );
 116
 117                         return $value; // cache hit
 118                 }
 119                 $staleValue = $value ?: $staleValue;
 120
 121                 # (c) Cache key missing or expired; regenerate and backfill
 122                 if ( $this->mainCache->lock( $key, 0, 10 ) ) {
 123                         # Let this process alone update the cache value
 124                         $cache = $this->mainCache;
 125                         /** @noinspection PhpUnusedLocalVariableInspection */
 126                         $unlocker = new ScopedCallback( function () use ( $cache, $key ) {
 127                                 $cache->unlock( $key );
 128                         } );
 129                 } elseif ( $staleValue ) {
 130                         # Could not acquire lock but an old cache exists, so use it
 131                         return $staleValue;
 132                 }
 133
 134                 $lagTimes = [];
 135                 $weightScales = [];
 136                 $movAveRatio = $this->movingAveRatio;
 137                 foreach ( $serverIndexes as $i ) {
 138                         if ( $i == $this->parent->getWriterIndex() ) {
 139                                 $lagTimes[$i] = 0; // master always has no lag
 140                                 $weightScales[$i] = 1.0; // nominal weight
 141                                 continue;
 142                         }
 143
 144                         $conn = $this->parent->getAnyOpenConnection( $i );
 145                         if ( $conn ) {
 146                                 $close = false; // already open
 147                         } else {
 148                                 $conn = $this->parent->openConnection( $i, $domain );
 149                                 $close = true; // new connection
 150                         }
 151
 152                         $lastWeight = isset( $staleValue['weightScales'][$i] )
 153                                 ? $staleValue['weightScales'][$i]
 154                                 : 1.0;
 155                         $coefficient = $this->getWeightScale( $i, $conn ?: null );
 156                         $newWeight = $movAveRatio * $coefficient + ( 1 - $movAveRatio ) * $lastWeight;
 157
 158                         // Scale from 10% to 100% of nominal weight
 159                         $weightScales[$i] = max( $newWeight, .10 );
 160
 161                         if ( !$conn ) {
 162                                 $lagTimes[$i] = false;
 163                                 $host = $this->parent->getServerName( $i );
 164                                 $this->replLogger->error(
 165                                         __METHOD__ . ": host {db_server} is unreachable",
 166                                         [ 'db_server' => $host ]
 167                                 );
 168                                 continue;
 169                         }
 170
 171                         if ( $conn->getLBInfo( 'is static' ) ) {
 172                                 $lagTimes[$i] = 0;
 173                         } else {
 174                                 $lagTimes[$i] = $conn->getLag();
 175                                 if ( $lagTimes[$i] === false ) {
 176                                         $host = $this->parent->getServerName( $i );
 177                                         $this->replLogger->error(
 178                                                 __METHOD__ . ": host {db_server} is not replicating?",
 179                                                 [ 'db_server' => $host ]
 180                                         );
 181                                 }
 182                         }
 183
 184                         if ( $close ) {
 185                                 # Close the connection to avoid sleeper connections piling up.
 186                                 # Note that the caller will pick one of these DBs and reconnect,
 187                                 # which is slightly inefficient, but this only matters for the lag
 188                                 # time cache miss cache, which is far less common that cache hits.
 189                                 $this->parent->closeConnection( $conn );
 190                         }
 191                 }
 192
 193                 # Add a timestamp key so we know when it was cached
 194                 $value = [
 195                         'lagTimes' => $lagTimes,
 196                         'weightScales' => $weightScales,
 197                         'timestamp' => microtime( true )
 198                 ];
 199                 $this->mainCache->set( $key, $value, $staleTTL );
 200                 $this->srvCache->set( $key, $value, $staleTTL );
 201                 $this->replLogger->info( __METHOD__ . ": re-calculated lag times ($key)" );
 202
 203                 return $value;
 204         }
 205
 206         /**
 207          * @param integer $index Server index
 208          * @param IDatabase|null $conn Connection handle or null on connection failure
 209          * @return float
 210          */
 211         protected function getWeightScale( $index, IDatabase $conn = null ) {
 212                 return $conn ? 1.0 : 0.0;
 213         }
 214
 215         private function getCacheKey( array $serverIndexes ) {
 216                 sort( $serverIndexes );
 217                 // Lag is per-server, not per-DB, so key on the master DB name
 218                 return $this->srvCache->makeGlobalKey(
 219                         'lag-times',
 220                         self::VERSION,
 221                         $this->parent->getServerName( $this->parent->getWriterIndex() ),
 222                         implode( '-', $serverIndexes )
 223                 );
 224         }
 225 }