MediaWiki  1.33.0
LoadMonitor.php
Go to the documentation of this file.
1 <?php
22 namespace Wikimedia\Rdbms;
23 
24 use Psr\Log\LoggerInterface;
25 use Psr\Log\NullLogger;
26 use Wikimedia\ScopedCallback;
29 
36 class LoadMonitor implements ILoadMonitor {
38  protected $parent;
40  protected $srvCache;
42  protected $wanCache;
44  protected $replLogger;
45 
47  private $movingAveRatio;
50 
52  const VERSION = 1;
54  const LAG_WARN_THRESHOLD = 10;
55 
64  public function __construct(
66  ) {
67  $this->parent = $lb;
68  $this->srvCache = $srvCache;
69  $this->wanCache = $wCache;
70  $this->replLogger = new NullLogger();
71 
72  $this->movingAveRatio = $options['movingAveRatio'] ?? 0.1;
73  $this->lagWarnThreshold = $options['lagWarnThreshold'] ?? self::LAG_WARN_THRESHOLD;
74  }
75 
76  public function setLogger( LoggerInterface $logger ) {
77  $this->replLogger = $logger;
78  }
79 
80  final public function scaleLoads( array &$weightByServer, $domain ) {
81  $serverIndexes = array_keys( $weightByServer );
82  $states = $this->getServerStates( $serverIndexes, $domain );
83  $newScalesByServer = $states['weightScales'];
84  foreach ( $weightByServer as $i => $weight ) {
85  if ( isset( $newScalesByServer[$i] ) ) {
86  $weightByServer[$i] = $weight * $newScalesByServer[$i];
87  } else { // server recently added to config?
88  $host = $this->parent->getServerName( $i );
89  $this->replLogger->error( __METHOD__ . ": host $host not in cache" );
90  }
91  }
92  }
93 
94  final public function getLagTimes( array $serverIndexes, $domain ) {
95  return $this->getServerStates( $serverIndexes, $domain )['lagTimes'];
96  }
97 
98  protected function getServerStates( array $serverIndexes, $domain ) {
99  $writerIndex = $this->parent->getWriterIndex();
100  if ( count( $serverIndexes ) == 1 && reset( $serverIndexes ) == $writerIndex ) {
101  # Single server only, just return zero without caching
102  return [
103  'lagTimes' => [ $writerIndex => 0 ],
104  'weightScales' => [ $writerIndex => 1.0 ]
105  ];
106  }
107 
108  $key = $this->getCacheKey( $serverIndexes );
109  # Randomize TTLs to reduce stampedes (4.0 - 5.0 sec)
110  $ttl = mt_rand( 4e6, 5e6 ) / 1e6;
111  # Keep keys around longer as fallbacks
112  $staleTTL = 60;
113 
114  # (a) Check the local APC cache
115  $value = $this->srvCache->get( $key );
116  if ( $value && $value['timestamp'] > ( microtime( true ) - $ttl ) ) {
117  $this->replLogger->debug( __METHOD__ . ": got lag times ($key) from local cache" );
118  return $value; // cache hit
119  }
120  $staleValue = $value ?: false;
121 
122  # (b) Check the shared cache and backfill APC
123  $value = $this->wanCache->get( $key );
124  if ( $value && $value['timestamp'] > ( microtime( true ) - $ttl ) ) {
125  $this->srvCache->set( $key, $value, $staleTTL );
126  $this->replLogger->debug( __METHOD__ . ": got lag times ($key) from main cache" );
127 
128  return $value; // cache hit
129  }
130  $staleValue = $value ?: $staleValue;
131 
132  # (c) Cache key missing or expired; regenerate and backfill
133  if ( $this->srvCache->lock( $key, 0, 10 ) ) {
134  # Let only this process update the cache value on this server
135  $sCache = $this->srvCache;
137  $unlocker = new ScopedCallback( function () use ( $sCache, $key ) {
138  $sCache->unlock( $key );
139  } );
140  } elseif ( $staleValue ) {
141  # Could not acquire lock but an old cache exists, so use it
142  return $staleValue;
143  }
144 
145  $lagTimes = [];
146  $weightScales = [];
147  $movAveRatio = $this->movingAveRatio;
148  foreach ( $serverIndexes as $i ) {
149  if ( $i == $this->parent->getWriterIndex() ) {
150  $lagTimes[$i] = 0; // master always has no lag
151  $weightScales[$i] = 1.0; // nominal weight
152  continue;
153  }
154 
155  # Handles with open transactions are avoided since they might be subject
156  # to REPEATABLE-READ snapshots, which could affect the lag estimate query.
157  $flags = ILoadBalancer::CONN_TRX_AUTOCOMMIT;
158  $conn = $this->parent->getAnyOpenConnection( $i, $flags );
159  if ( $conn ) {
160  $close = false; // already open
161  } else {
162  $conn = $this->parent->openConnection( $i, ILoadBalancer::DOMAIN_ANY, $flags );
163  $close = true; // new connection
164  }
165 
166  $lastWeight = $staleValue['weightScales'][$i] ?? 1.0;
167  $coefficient = $this->getWeightScale( $i, $conn ?: null );
168  $newWeight = $movAveRatio * $coefficient + ( 1 - $movAveRatio ) * $lastWeight;
169 
170  // Scale from 10% to 100% of nominal weight
171  $weightScales[$i] = max( $newWeight, 0.10 );
172 
173  $host = $this->parent->getServerName( $i );
174 
175  if ( !$conn ) {
176  $lagTimes[$i] = false;
177  $this->replLogger->error(
178  __METHOD__ . ": host {db_server} is unreachable",
179  [ 'db_server' => $host ]
180  );
181  continue;
182  }
183 
184  if ( $conn->getLBInfo( 'is static' ) ) {
185  $lagTimes[$i] = 0;
186  } else {
187  $lagTimes[$i] = $conn->getLag();
188  if ( $lagTimes[$i] === false ) {
189  $this->replLogger->error(
190  __METHOD__ . ": host {db_server} is not replicating?",
191  [ 'db_server' => $host ]
192  );
193  } elseif ( $lagTimes[$i] > $this->lagWarnThreshold ) {
194  $this->replLogger->warning(
195  "Server {host} has {lag} seconds of lag (>= {maxlag})",
196  [
197  'host' => $host,
198  'lag' => $lagTimes[$i],
199  'maxlag' => $this->lagWarnThreshold
200  ]
201  );
202  }
203  }
204 
205  if ( $close ) {
206  # Close the connection to avoid sleeper connections piling up.
207  # Note that the caller will pick one of these DBs and reconnect,
208  # which is slightly inefficient, but this only matters for the lag
209  # time cache miss cache, which is far less common that cache hits.
210  $this->parent->closeConnection( $conn );
211  }
212  }
213 
214  # Add a timestamp key so we know when it was cached
215  $value = [
216  'lagTimes' => $lagTimes,
217  'weightScales' => $weightScales,
218  'timestamp' => microtime( true )
219  ];
220  $this->wanCache->set( $key, $value, $staleTTL );
221  $this->srvCache->set( $key, $value, $staleTTL );
222  $this->replLogger->info( __METHOD__ . ": re-calculated lag times ($key)" );
223 
224  return $value;
225  }
226 
232  protected function getWeightScale( $index, IDatabase $conn = null ) {
233  return $conn ? 1.0 : 0.0;
234  }
235 
236  private function getCacheKey( array $serverIndexes ) {
237  sort( $serverIndexes );
238  // Lag is per-server, not per-DB, so key on the master DB name
239  return $this->srvCache->makeGlobalKey(
240  'lag-times',
241  self::VERSION,
242  $this->parent->getServerName( $this->parent->getWriterIndex() ),
243  implode( '-', $serverIndexes )
244  );
245  }
246 }
Wikimedia\Rdbms\LoadMonitor\$replLogger
LoggerInterface $replLogger
Definition: LoadMonitor.php:44
Wikimedia\Rdbms\LoadMonitor
Basic DB load monitor with no external dependencies Uses memcached to cache the replication lag for a...
Definition: LoadMonitor.php:36
Wikimedia\Rdbms\LoadMonitor\$srvCache
BagOStuff $srvCache
Definition: LoadMonitor.php:40
Wikimedia\Rdbms\ILoadMonitor
An interface for database load monitoring.
Definition: ILoadMonitor.php:35
captcha-old.count
count
Definition: captcha-old.py:249
Wikimedia\Rdbms\LoadMonitor\$parent
ILoadBalancer $parent
Definition: LoadMonitor.php:38
Wikimedia\Rdbms
Definition: ChronologyProtector.php:24
Wikimedia\Rdbms\LoadMonitor\getCacheKey
getCacheKey(array $serverIndexes)
Definition: LoadMonitor.php:236
BagOStuff
Class representing a cache/ephemeral data store.
Definition: BagOStuff.php:58
Wikimedia\Rdbms\LoadMonitor\getWeightScale
getWeightScale( $index, IDatabase $conn=null)
Definition: LoadMonitor.php:232
php
injection txt This is an overview of how MediaWiki makes use of dependency injection The design described here grew from the discussion of RFC T384 The term dependency this means that anything an object needs to operate should be injected from the the object itself should only know narrow no concrete implementation of the logic it relies on The requirement to inject everything typically results in an architecture that based on two main types of and essentially stateless service objects that use other service objects to operate on the value objects As of the beginning MediaWiki is only starting to use the DI approach Much of the code still relies on global state or direct resulting in a highly cyclical dependency which acts as the top level factory for services in MediaWiki which can be used to gain access to default instances of various services MediaWikiServices however also allows new services to be defined and default services to be redefined Services are defined or redefined by providing a callback the instantiator that will return a new instance of the service When it will create an instance of MediaWikiServices and populate it with the services defined in the files listed by thereby bootstrapping the DI framework Per $wgServiceWiringFiles lists includes ServiceWiring php
Definition: injection.txt:35
Wikimedia\Rdbms\IDatabase
Basic database interface for live and lazy-loaded relation database handles.
Definition: IDatabase.php:38
Wikimedia\Rdbms\LoadMonitor\$wanCache
WANObjectCache $wanCache
Definition: LoadMonitor.php:42
Wikimedia\Rdbms\LoadMonitor\$lagWarnThreshold
int $lagWarnThreshold
Amount of replication lag in seconds before warnings are logged.
Definition: LoadMonitor.php:49
Wikimedia\Rdbms\LoadMonitor\$movingAveRatio
float $movingAveRatio
Moving average ratio (e.g.
Definition: LoadMonitor.php:47
use
as see the revision history and available at free of to any person obtaining a copy of this software and associated documentation to deal in the Software without including without limitation the rights to use
Definition: MIT-LICENSE.txt:10
Wikimedia\Rdbms\LoadMonitor\scaleLoads
scaleLoads(array &$weightByServer, $domain)
Perform load ratio adjustment before deciding which server to use.
Definition: LoadMonitor.php:80
array
The wiki should then use memcached to cache various data To use multiple just add more items to the array To increase the weight of a make its entry a array("192.168.0.1:11211", 2))
Wikimedia\Rdbms\LoadMonitor\getServerStates
getServerStates(array $serverIndexes, $domain)
Definition: LoadMonitor.php:98
$value
$value
Definition: styleTest.css.php:49
Wikimedia\Rdbms\LoadMonitor\setLogger
setLogger(LoggerInterface $logger)
Definition: LoadMonitor.php:76
WANObjectCache
Multi-datacenter aware caching interface.
Definition: WANObjectCache.php:116
Wikimedia\Rdbms\LoadMonitor\getLagTimes
getLagTimes(array $serverIndexes, $domain)
Get an estimate of replication lag (in seconds) for each server.
Definition: LoadMonitor.php:94
$options
null means default in associative array with keys and values unescaped Should be merged with default with a value of false meaning to suppress the attribute in associative array with keys and values unescaped & $options
Definition: hooks.txt:1985
as
This document is intended to provide useful advice for parties seeking to redistribute MediaWiki to end users It s targeted particularly at maintainers for Linux since it s been observed that distribution packages of MediaWiki often break We ve consistently had to recommend that users seeking support use official tarballs instead of their distribution s and this often solves whatever problem the user is having It would be nice if this could such as
Definition: distributors.txt:9
Wikimedia\Rdbms\LoadMonitor\__construct
__construct(ILoadBalancer $lb, BagOStuff $srvCache, WANObjectCache $wCache, array $options=[])
Definition: LoadMonitor.php:64
Wikimedia\Rdbms\ILoadBalancer
Database cluster connection, tracking, load balancing, and transaction manager interface.
Definition: ILoadBalancer.php:78