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 
95  public function store( $location, $data ) {
96  $dbw = $this->getMaster( $location );
97  $dbw->insert( $this->getTable( $dbw ),
98  [ 'blob_text' => $data ],
99  __METHOD__ );
100  $id = $dbw->insertId();
101  if ( !$id ) {
102  throw new MWException( __METHOD__ . ': no insert ID' );
103  }
104 
105  return "DB://$location/$id";
106  }
107 
108  public function isReadOnly( $location ) {
109  return ( $this->getLoadBalancer( $location )->getReadOnlyReason() !== false );
110  }
111 
118  private function getLoadBalancer( $cluster ) {
119  $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
120  return $lbFactory->getExternalLB( $cluster );
121  }
122 
129  public function getSlave( $cluster ) {
131 
132  $lb = $this->getLoadBalancer( $cluster );
133  $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
134 
135  if ( !in_array( "DB://" . $cluster, (array)$wgDefaultExternalStore ) ) {
136  wfDebug( "read only external store\n" );
137  $lb->allowLagged( true );
138  } else {
139  wfDebug( "writable external store\n" );
140  }
141 
142  $db = $lb->getConnectionRef( DB_REPLICA, [], $domainId );
143  $db->clearFlag( DBO_TRX ); // sanity
144 
145  return $db;
146  }
147 
154  public function getMaster( $cluster ) {
155  $lb = $this->getLoadBalancer( $cluster );
156  $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
157 
158  $db = $lb->getMaintenanceConnectionRef( DB_MASTER, [], $domainId );
159  $db->clearFlag( DBO_TRX ); // sanity
160 
161  return $db;
162  }
163 
168  private function getDomainId( array $server ) {
169  if ( isset( $this->params['wiki'] ) ) {
170  return $this->params['wiki']; // explicit domain
171  }
172 
173  if ( isset( $server['dbname'] ) ) {
174  // T200471: for b/c, treat any "dbname" field as forcing which database to use.
175  // MediaWiki/LoadBalancer previously did not enforce any concept of a local DB
176  // domain, but rather assumed that the LB server configuration matched $wgDBname.
177  // This check is useful when the external storage DB for this cluster does not use
178  // the same name as the corresponding "main" DB(s) for wikis.
179  $domain = new DatabaseDomain(
180  $server['dbname'],
181  $server['schema'] ?? null,
182  $server['tablePrefix'] ?? ''
183  );
184 
185  return $domain->getId();
186  }
187 
188  return false; // local LB domain
189  }
190 
197  public function getTable( $db ) {
198  $table = $db->getLBInfo( 'blobs table' );
199  if ( is_null( $table ) ) {
200  $table = 'blobs';
201  }
202 
203  return $table;
204  }
205 
215  private function fetchBlob( $cluster, $id, $itemID ) {
222  static $externalBlobCache = [];
223 
224  $cacheID = ( $itemID === false ) ? "$cluster/$id" : "$cluster/$id/";
225 
226  $wiki = $this->params['wiki'] ?? false;
227  $cacheID = ( $wiki === false ) ? $cacheID : "$cacheID@$wiki";
228 
229  if ( isset( $externalBlobCache[$cacheID] ) ) {
230  wfDebugLog( 'ExternalStoreDB-cache',
231  "ExternalStoreDB::fetchBlob cache hit on $cacheID" );
232 
233  return $externalBlobCache[$cacheID];
234  }
235 
236  wfDebugLog( 'ExternalStoreDB-cache',
237  "ExternalStoreDB::fetchBlob cache miss on $cacheID" );
238 
239  $dbr = $this->getSlave( $cluster );
240  $ret = $dbr->selectField( $this->getTable( $dbr ),
241  'blob_text', [ 'blob_id' => $id ], __METHOD__ );
242  if ( $ret === false ) {
243  wfDebugLog( 'ExternalStoreDB',
244  "ExternalStoreDB::fetchBlob master fallback on $cacheID" );
245  // Try the master
246  $dbw = $this->getMaster( $cluster );
247  $ret = $dbw->selectField( $this->getTable( $dbw ),
248  'blob_text', [ 'blob_id' => $id ], __METHOD__ );
249  if ( $ret === false ) {
250  wfDebugLog( 'ExternalStoreDB',
251  "ExternalStoreDB::fetchBlob master failed to find $cacheID" );
252  }
253  }
254  if ( $itemID !== false && $ret !== false ) {
255  // Unserialise object; caller extracts item
256  $ret = unserialize( $ret );
257  }
258 
259  $externalBlobCache = [ $cacheID => $ret ];
260 
261  return $ret;
262  }
263 
272  private function batchFetchBlobs( $cluster, array $ids ) {
273  $dbr = $this->getSlave( $cluster );
274  $res = $dbr->select( $this->getTable( $dbr ),
275  [ 'blob_id', 'blob_text' ], [ 'blob_id' => array_keys( $ids ) ], __METHOD__ );
276  $ret = [];
277  if ( $res !== false ) {
278  $this->mergeBatchResult( $ret, $ids, $res );
279  }
280  if ( $ids ) {
281  wfDebugLog( __CLASS__, __METHOD__ .
282  " master fallback on '$cluster' for: " .
283  implode( ',', array_keys( $ids ) ) );
284  // Try the master
285  $dbw = $this->getMaster( $cluster );
286  $res = $dbw->select( $this->getTable( $dbr ),
287  [ 'blob_id', 'blob_text' ],
288  [ 'blob_id' => array_keys( $ids ) ],
289  __METHOD__ );
290  if ( $res === false ) {
291  wfDebugLog( __CLASS__, __METHOD__ . " master failed on '$cluster'" );
292  } else {
293  $this->mergeBatchResult( $ret, $ids, $res );
294  }
295  }
296  if ( $ids ) {
297  wfDebugLog( __CLASS__, __METHOD__ .
298  " master on '$cluster' failed locating items: " .
299  implode( ',', array_keys( $ids ) ) );
300  }
301 
302  return $ret;
303  }
304 
311  private function mergeBatchResult( array &$ret, array &$ids, $res ) {
312  foreach ( $res as $row ) {
313  $id = $row->blob_id;
314  $itemIDs = $ids[$id];
315  unset( $ids[$id] ); // to track if everything is found
316  if ( count( $itemIDs ) === 1 && reset( $itemIDs ) === false ) {
317  // single result stored per blob
318  $ret[$id] = $row->blob_text;
319  } else {
320  // multi result stored per blob
321  $ret[$id] = unserialize( $row->blob_text );
322  }
323  }
324  }
325 
330  protected function parseURL( $url ) {
331  $path = explode( '/', $url );
332 
333  return [
334  $path[2], // cluster
335  $path[3], // id
336  $path[4] ?? false // itemID
337  ];
338  }
339 }
getDomainId(array $server)
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))
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
$data
Utility to generate mapping file used in mw.Title (phpCharToUpper.json)
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 ...