MediaWiki  master
ExternalStoreDB.php
Go to the documentation of this file.
1 <?php
29 
47  public function fetchFromURL( $url ) {
48  list( $cluster, $id, $itemID ) = $this->parseURL( $url );
49  $ret = $this->fetchBlob( $cluster, $id, $itemID );
50 
51  if ( $itemID !== false && $ret !== false ) {
52  return $ret->getItem( $itemID );
53  }
54 
55  return $ret;
56  }
57 
67  public function batchFetchFromURLs( array $urls ) {
68  $batched = $inverseUrlMap = [];
69  foreach ( $urls as $url ) {
70  list( $cluster, $id, $itemID ) = $this->parseURL( $url );
71  $batched[$cluster][$id][] = $itemID;
72  // false $itemID gets cast to int, but should be ok
73  // since we do === from the $itemID in $batched
74  $inverseUrlMap[$cluster][$id][$itemID] = $url;
75  }
76  $ret = [];
77  foreach ( $batched as $cluster => $batchByCluster ) {
78  $res = $this->batchFetchBlobs( $cluster, $batchByCluster );
80  foreach ( $res as $id => $blob ) {
81  foreach ( $batchByCluster[$id] as $itemID ) {
82  $url = $inverseUrlMap[$cluster][$id][$itemID];
83  if ( $itemID === false ) {
84  $ret[$url] = $blob;
85  } else {
86  $ret[$url] = $blob->getItem( $itemID );
87  }
88  }
89  }
90  }
91 
92  return $ret;
93  }
94 
98  public function store( $location, $data ) {
99  $dbw = $this->getMaster( $location );
100  $dbw->insert( $this->getTable( $dbw ),
101  [ 'blob_text' => $data ],
102  __METHOD__ );
103  $id = $dbw->insertId();
104  if ( !$id ) {
105  throw new MWException( __METHOD__ . ': no insert ID' );
106  }
107 
108  return "DB://$location/$id";
109  }
110 
114  public function isReadOnly( $location ) {
115  $lb = $this->getLoadBalancer( $location );
116  $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
117  return ( $lb->getReadOnlyReason( $domainId ) !== false );
118  }
119 
126  private function getLoadBalancer( $cluster ) {
127  $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
128  return $lbFactory->getExternalLB( $cluster );
129  }
130 
137  public function getSlave( $cluster ) {
139 
140  $lb = $this->getLoadBalancer( $cluster );
141  $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
142 
143  if ( !in_array( "DB://" . $cluster, (array)$wgDefaultExternalStore ) ) {
144  wfDebug( "read only external store\n" );
145  $lb->allowLagged( true );
146  } else {
147  wfDebug( "writable external store\n" );
148  }
149 
150  $db = $lb->getConnectionRef( DB_REPLICA, [], $domainId );
151  $db->clearFlag( DBO_TRX ); // sanity
152 
153  return $db;
154  }
155 
162  public function getMaster( $cluster ) {
163  $lb = $this->getLoadBalancer( $cluster );
164  $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
165 
166  $db = $lb->getMaintenanceConnectionRef( DB_MASTER, [], $domainId );
167  $db->clearFlag( DBO_TRX ); // sanity
168 
169  return $db;
170  }
171 
176  private function getDomainId( array $server ) {
177  if ( isset( $this->params['wiki'] ) && $this->params['wiki'] !== false ) {
178  return $this->params['wiki']; // explicit domain
179  }
180 
181  if ( isset( $server['dbname'] ) ) {
182  // T200471: for b/c, treat any "dbname" field as forcing which database to use.
183  // MediaWiki/LoadBalancer previously did not enforce any concept of a local DB
184  // domain, but rather assumed that the LB server configuration matched $wgDBname.
185  // This check is useful when the external storage DB for this cluster does not use
186  // the same name as the corresponding "main" DB(s) for wikis.
187  $domain = new DatabaseDomain(
188  $server['dbname'],
189  $server['schema'] ?? null,
190  $server['tablePrefix'] ?? ''
191  );
192 
193  return $domain->getId();
194  }
195 
196  return false; // local LB domain
197  }
198 
205  public function getTable( $db ) {
206  $table = $db->getLBInfo( 'blobs table' );
207  if ( is_null( $table ) ) {
208  $table = 'blobs';
209  }
210 
211  return $table;
212  }
213 
223  private function fetchBlob( $cluster, $id, $itemID ) {
230  static $externalBlobCache = [];
231 
232  $cacheID = ( $itemID === false ) ? "$cluster/$id" : "$cluster/$id/";
233 
234  $wiki = $this->params['wiki'] ?? false;
235  $cacheID = ( $wiki === false ) ? $cacheID : "$cacheID@$wiki";
236 
237  if ( isset( $externalBlobCache[$cacheID] ) ) {
238  wfDebugLog( 'ExternalStoreDB-cache',
239  "ExternalStoreDB::fetchBlob cache hit on $cacheID" );
240 
241  return $externalBlobCache[$cacheID];
242  }
243 
244  wfDebugLog( 'ExternalStoreDB-cache',
245  "ExternalStoreDB::fetchBlob cache miss on $cacheID" );
246 
247  $dbr = $this->getSlave( $cluster );
248  $ret = $dbr->selectField( $this->getTable( $dbr ),
249  'blob_text', [ 'blob_id' => $id ], __METHOD__ );
250  if ( $ret === false ) {
251  wfDebugLog( 'ExternalStoreDB',
252  "ExternalStoreDB::fetchBlob master fallback on $cacheID" );
253  // Try the master
254  $dbw = $this->getMaster( $cluster );
255  $ret = $dbw->selectField( $this->getTable( $dbw ),
256  'blob_text', [ 'blob_id' => $id ], __METHOD__ );
257  if ( $ret === false ) {
258  wfDebugLog( 'ExternalStoreDB',
259  "ExternalStoreDB::fetchBlob master failed to find $cacheID" );
260  }
261  }
262  if ( $itemID !== false && $ret !== false ) {
263  // Unserialise object; caller extracts item
264  $ret = unserialize( $ret );
265  }
266 
267  $externalBlobCache = [ $cacheID => $ret ];
268 
269  return $ret;
270  }
271 
280  private function batchFetchBlobs( $cluster, array $ids ) {
281  $dbr = $this->getSlave( $cluster );
282  $res = $dbr->select( $this->getTable( $dbr ),
283  [ 'blob_id', 'blob_text' ], [ 'blob_id' => array_keys( $ids ) ], __METHOD__ );
284  $ret = [];
285  if ( $res !== false ) {
286  $this->mergeBatchResult( $ret, $ids, $res );
287  }
288  if ( $ids ) {
289  wfDebugLog( __CLASS__, __METHOD__ .
290  " master fallback on '$cluster' for: " .
291  implode( ',', array_keys( $ids ) ) );
292  // Try the master
293  $dbw = $this->getMaster( $cluster );
294  $res = $dbw->select( $this->getTable( $dbr ),
295  [ 'blob_id', 'blob_text' ],
296  [ 'blob_id' => array_keys( $ids ) ],
297  __METHOD__ );
298  if ( $res === false ) {
299  wfDebugLog( __CLASS__, __METHOD__ . " master failed on '$cluster'" );
300  } else {
301  $this->mergeBatchResult( $ret, $ids, $res );
302  }
303  }
304  if ( $ids ) {
305  wfDebugLog( __CLASS__, __METHOD__ .
306  " master on '$cluster' failed locating items: " .
307  implode( ',', array_keys( $ids ) ) );
308  }
309 
310  return $ret;
311  }
312 
319  private function mergeBatchResult( array &$ret, array &$ids, $res ) {
320  foreach ( $res as $row ) {
321  $id = $row->blob_id;
322  $itemIDs = $ids[$id];
323  unset( $ids[$id] ); // to track if everything is found
324  if ( count( $itemIDs ) === 1 && reset( $itemIDs ) === false ) {
325  // single result stored per blob
326  $ret[$id] = $row->blob_text;
327  } else {
328  // multi result stored per blob
329  $ret[$id] = unserialize( $row->blob_text );
330  }
331  }
332  }
333 
338  protected function parseURL( $url ) {
339  $path = explode( '/', $url );
340 
341  return [
342  $path[2], // cluster
343  $path[3], // id
344  $path[4] ?? false // itemID
345  ];
346  }
347 }
getDomainId(array $server)
deferred txt A few of the database updates required by various functions here can be deferred until after the result page is displayed to the user For updating the view updating the linked to tables after a etc PHP does not yet have any way to tell the server to actually return and disconnect while still running these but it might have such a feature in the future We handle these by creating a deferred update object and putting those objects on a global list
Definition: deferred.txt:11
array $wgDefaultExternalStore
The place to put new revisions, false to put them in the local text table.
Accessable external objects in a particular storage medium.
processing should stop and the error should be shown to the user * false
Definition: hooks.txt:187
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 noclasses & $ret
Definition: hooks.txt:1982
Apache License January AND DISTRIBUTION Definitions License shall mean the terms and conditions for use
fetchFromURL( $url)
The provided URL is in the form of DB://cluster/id or DB://cluster/id/itemid for concatened storage...
getSlave( $cluster)
Get a replica DB connection for the specified cluster.
batchFetchBlobs( $cluster, array $ids)
Fetch multiple blob items out of the database.
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 MediaWikiServices
Definition: injection.txt:23
const DB_MASTER
Definition: defines.php:26
batchFetchFromURLs(array $urls)
Fetch data from given external store URLs.
$res
Definition: database.txt:21
wfDebug( $text, $dest='all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
getLoadBalancer( $cluster)
Get a LoadBalancer for the specified cluster.
unserialize( $serialized)
this hook is for auditing only or null if authentication failed before getting that far or null if we can t even determine that When $user is not null
Definition: hooks.txt:780
mergeBatchResult(array &$ret, array &$ids, $res)
Helper function for self::batchFetchBlobs for merging master/replica DB results.
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
const DBO_TRX
Definition: defines.php:12
getTable( $db)
Get the &#39;blobs&#39; table name for this database.
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
Class to handle database/prefix specification for IDatabase domains.
wfDebugLog( $logGroup, $text, $dest='all', array $context=[])
Send a line to a supplementary debug log file, if configured, or main debug log if not...
store( $location, $data)
isReadOnly( $location)
const DB_REPLICA
Definition: defines.php:25
getMaster( $cluster)
Get a master database connection for the specified cluster.
DB accessible external objects.
fetchBlob( $cluster, $id, $itemID)
Fetch a blob item out of the database; a cache of the last-loaded blob will be kept so that multiple ...