Merge "ResourceLoader: Deprecate ResourceLoader::makeConfigSetScript"
[mediawiki.git] / includes / libs / rdbms / lbfactory / LBFactoryMulti.php
blob0ccce1d2da4dbe8567c4219c0da2b57457ab54f7
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.
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.
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
18 * @file
20 namespace Wikimedia\Rdbms;
22 use InvalidArgumentException;
23 use LogicException;
24 use UnexpectedValueException;
26 /**
27 * LoadBalancer manager for sites with several "main" database clusters
29 * Each database cluster consists of a "primary" server and any number of replica servers,
30 * all of which converge, as soon as possible, to contain the same schemas and records. If
31 * a replication topology has multiple primaries, then the "primary" is merely the preferred
32 * co-primary for the current context (e.g. datacenter).
34 * For single-primary topologies, the schemas and records of the primary define the "dataset".
35 * For multiple-primary topologies, the "dataset" is the convergent result of applying/merging
36 * all committed events (regardless of the co-primary they originated on); it possible that no
37 * co-primary has yet converged upon this state at any given time (especially when there are
38 * frequent writes and co-primaries are geographically distant).
40 * A "main" cluster contain a "main" dataset, which consists of data that is compact, highly
41 * relational (e.g. read by JOIN queries), and essential to one or more sites. The "external"
42 * clusters each store an "external" dataset, which consists of data that is non-relational
43 * (e.g. key/value pairs), self-contained (e.g. JOIN queries and transactions thereof never
44 * involve a main dataset), or too bulky to reside in a main dataset (e.g. text blobs).
46 * The class allows for large site farms to split up their data in the following ways:
47 * - Vertically shard compact site-specific data by site (e.g. page/comment metadata)
48 * - Vertically shard compact global data by module (e.g. account/notification data)
49 * - Horizontally shard any bulk data by blob key (e.g. page/comment content blobs)
51 * @ingroup Database
53 class LBFactoryMulti extends LBFactory {
54 /** @var array<string,LoadBalancer> Map of (main section => tracked LoadBalancer) */
55 private $mainLBs = [];
56 /** @var array<string,LoadBalancer> Map of (external cluster => tracked LoadBalancer) */
57 private $externalLBs = [];
59 /** @var string[] Map of (server name => IP address) */
60 private $hostsByServerName;
61 /** @var string[] Map of (database name => main section) */
62 private $sectionsByDB;
63 /** @var int[][][] Map of (main section => group => server name => load ratio) */
64 private $groupLoadsBySection;
65 /** @var int[][] Map of (external cluster => server name => load ratio) */
66 private $externalLoadsByCluster;
67 /** @var array Server config map ("host", "serverName", "load", and "groupLoads" ignored) */
68 private $serverTemplate;
69 /** @var array Server config map overriding "serverTemplate" for all external servers */
70 private $externalTemplateOverrides;
71 /** @var array[] Map of (main section => server config map overrides) */
72 private $templateOverridesBySection;
73 /** @var array[] Map of (external cluster => server config map overrides) */
74 private $templateOverridesByCluster;
75 /** @var array Server config override map for all main/external primary DB servers */
76 private $masterTemplateOverrides;
77 /** @var array[] Map of (server name => server config map overrides) for all servers */
78 private $templateOverridesByServer;
79 /** @var string[]|bool[] A map of (main section => read-only message) */
80 private $readOnlyBySection;
81 /** @var array Configuration for the LoadMonitor to use within LoadBalancer instances */
82 private $loadMonitorConfig;
83 /** @var DatabaseDomain[] Map of (domain ID => domain instance) */
84 private $nonLocalDomainCache = [];
86 /**
87 * Template override precedence (highest => lowest):
88 * - templateOverridesByServer
89 * - masterTemplateOverrides
90 * - templateOverridesBySection/templateOverridesByCluster
91 * - externalTemplateOverrides
92 * - serverTemplate
93 * Overrides only work on top level keys (so nested values will not be merged).
95 * Server config maps should be of the format Database::factory() requires.
96 * Additionally, a 'max lag' key should also be set on server maps, indicating how stale the
97 * data can be before the load balancer tries to avoid using it. The map can have 'is static'
98 * set to disable blocking replication sync checks (intended for archive servers with
99 * unchanging data).
101 * @see LBFactory::__construct()
102 * @param array $conf Additional parameters include:
103 * - hostsByName: map of (server name => IP address). [optional]
104 * - sectionsByDB: map of (database => main section). The database name "DEFAULT" is
105 * interpreted as a catch-all for all databases not otherwise mentioned. If no section
106 * name is specified for "DEFAULT", then the catch-all section is assumed to be named
107 * "DEFAULT". [optional]
108 * - sectionLoads: map of (main section => server name => load ratio); the first host
109 * listed in each section is the primary DB server for that section. [optional]
110 * - groupLoadsBySection: map of (main section => group => server name => group load ratio).
111 * Any ILoadBalancer::GROUP_GENERIC group will be ignored. [optional]
112 * - externalLoads: map of (cluster => server name => load ratio) map. [optional]
113 * - serverTemplate: server config map for Database::factory().
114 * Note that "host", "serverName" and "load" entries will be overridden by
115 * "groupLoadsBySection" and "hostsByName". [optional]
116 * - externalTemplateOverrides: server config map overrides for external stores;
117 * respects the override precedence described above. [optional]
118 * - templateOverridesBySection: map of (main section => server config map overrides);
119 * respects the override precedence described above. [optional]
120 * - templateOverridesByCluster: map of (external cluster => server config map overrides);
121 * respects the override precedence described above. [optional]
122 * - masterTemplateOverrides: server config map overrides for masters;
123 * respects the override precedence described above. [optional]
124 * - templateOverridesByServer: map of (server name => server config map overrides);
125 * respects the override precedence described above and applies to both core
126 * and external storage. [optional]
127 * - loadMonitor: LoadMonitor::__construct() parameters with "class" field. [optional]
128 * - readOnlyBySection: map of (main section => message text or false).
129 * String values make sections read only, whereas anything else does not
130 * restrict read/write mode. [optional]
131 * - configCallback: A callback that returns a conf array that can be passed to
132 * the reconfigure() method. This will be used to autoReconfigure() to load
133 * any updated configuration.
135 public function __construct( array $conf ) {
136 parent::__construct( $conf );
138 $this->hostsByServerName = $conf['hostsByName'] ?? [];
139 $this->sectionsByDB = $conf['sectionsByDB'];
140 $this->sectionsByDB += [ self::CLUSTER_MAIN_DEFAULT => self::CLUSTER_MAIN_DEFAULT ];
141 $this->groupLoadsBySection = $conf['groupLoadsBySection'] ?? [];
142 foreach ( ( $conf['sectionLoads'] ?? [] ) as $section => $loadsByServerName ) {
143 $this->groupLoadsBySection[$section][ILoadBalancer::GROUP_GENERIC] = $loadsByServerName;
145 $this->externalLoadsByCluster = $conf['externalLoads'] ?? [];
146 $this->serverTemplate = $conf['serverTemplate'] ?? [];
147 $this->externalTemplateOverrides = $conf['externalTemplateOverrides'] ?? [];
148 $this->templateOverridesBySection = $conf['templateOverridesBySection'] ?? [];
149 $this->templateOverridesByCluster = $conf['templateOverridesByCluster'] ?? [];
150 $this->masterTemplateOverrides = $conf['masterTemplateOverrides'] ?? [];
151 $this->templateOverridesByServer = $conf['templateOverridesByServer'] ?? [];
152 $this->readOnlyBySection = $conf['readOnlyBySection'] ?? [];
154 if ( isset( $conf['loadMonitor'] ) ) {
155 $this->loadMonitorConfig = $conf['loadMonitor'];
156 } elseif ( isset( $conf['loadMonitorClass'] ) ) { // b/c
157 $this->loadMonitorConfig = [ 'class' => $conf['loadMonitorClass'] ];
158 } else {
159 $this->loadMonitorConfig = [ 'class' => LoadMonitor::class ];
162 foreach ( $this->externalLoadsByCluster as $cluster => $_ ) {
163 if ( isset( $this->groupLoadsBySection[$cluster] ) ) {
164 throw new LogicException(
165 "External cluster '$cluster' has the same name as a main section/cluster"
171 public function newMainLB( $domain = false ): ILoadBalancerForOwner {
172 $domainInstance = $this->resolveDomainInstance( $domain );
173 $database = $domainInstance->getDatabase();
174 $section = $this->getSectionFromDatabase( $database );
176 if ( !isset( $this->groupLoadsBySection[$section][ILoadBalancer::GROUP_GENERIC] ) ) {
177 throw new UnexpectedValueException( "Section '$section' has no hosts defined." );
180 return $this->newLoadBalancer(
181 $section,
182 array_merge(
183 $this->serverTemplate,
184 $this->templateOverridesBySection[$section] ?? []
186 $this->groupLoadsBySection[$section],
187 // Use the LB-specific read-only reason if everything isn't already read-only
188 is_string( $this->readOnlyReason )
189 ? $this->readOnlyReason
190 : ( $this->readOnlyBySection[$section] ?? false )
195 * @param DatabaseDomain|string|false $domain
196 * @return DatabaseDomain
198 private function resolveDomainInstance( $domain ) {
199 if ( $domain instanceof DatabaseDomain ) {
200 return $domain; // already a domain instance
201 } elseif ( $domain === false || $domain === $this->localDomain->getId() ) {
202 return $this->localDomain;
203 } elseif ( isset( $this->domainAliases[$domain] ) ) {
204 // This array acts as both the original map and as instance cache.
205 // Instances pass-through DatabaseDomain::newFromId as-is.
206 $this->domainAliases[$domain] =
207 DatabaseDomain::newFromId( $this->domainAliases[$domain] );
209 return $this->domainAliases[$domain];
212 $cachedDomain = $this->nonLocalDomainCache[$domain] ?? null;
213 if ( $cachedDomain === null ) {
214 $cachedDomain = DatabaseDomain::newFromId( $domain );
215 $this->nonLocalDomainCache = [ $domain => $cachedDomain ];
218 return $cachedDomain;
221 public function getMainLB( $domain = false ): ILoadBalancer {
222 $domainInstance = $this->resolveDomainInstance( $domain );
223 $section = $this->getSectionFromDatabase( $domainInstance->getDatabase() );
225 if ( !isset( $this->mainLBs[$section] ) ) {
226 $this->mainLBs[$section] = $this->newMainLB( $domain );
229 return $this->mainLBs[$section];
232 public function newExternalLB( $cluster ): ILoadBalancerForOwner {
233 if ( !isset( $this->externalLoadsByCluster[$cluster] ) ) {
234 throw new InvalidArgumentException( "Unknown cluster '$cluster'" );
236 return $this->newLoadBalancer(
237 $cluster,
238 array_merge(
239 $this->serverTemplate,
240 $this->externalTemplateOverrides,
241 $this->templateOverridesByCluster[$cluster] ?? []
243 [ ILoadBalancer::GROUP_GENERIC => $this->externalLoadsByCluster[$cluster] ],
244 $this->readOnlyReason
248 public function getExternalLB( $cluster ): ILoadBalancer {
249 if ( !isset( $this->externalLBs[$cluster] ) ) {
250 $this->externalLBs[$cluster] = $this->newExternalLB(
251 $cluster
255 return $this->externalLBs[$cluster];
258 public function getAllMainLBs(): array {
259 $lbs = [];
260 foreach ( $this->sectionsByDB as $db => $section ) {
261 if ( !isset( $lbs[$section] ) ) {
262 $lbs[$section] = $this->getMainLB( $db );
266 return $lbs;
269 public function getAllExternalLBs(): array {
270 $lbs = [];
271 foreach ( $this->externalLoadsByCluster as $cluster => $unused ) {
272 $lbs[$cluster] = $this->getExternalLB( $cluster );
275 return $lbs;
278 protected function getLBsForOwner() {
279 foreach ( $this->mainLBs as $lb ) {
280 yield $lb;
282 foreach ( $this->externalLBs as $lb ) {
283 yield $lb;
288 * Make a new load balancer object based on template and load array
290 * @param string $clusterName
291 * @param array $serverTemplate
292 * @param array $groupLoads
293 * @param string|false $readOnlyReason
294 * @return LoadBalancer
296 private function newLoadBalancer(
297 string $clusterName,
298 array $serverTemplate,
299 array $groupLoads,
300 $readOnlyReason
302 $lb = new LoadBalancer( array_merge(
303 $this->baseLoadBalancerParams(),
305 'servers' => $this->makeServerConfigArrays( $serverTemplate, $groupLoads ),
306 'loadMonitor' => $this->loadMonitorConfig,
307 'readOnlyReason' => $readOnlyReason,
308 'clusterName' => $clusterName
310 ) );
311 $this->initLoadBalancer( $lb );
313 return $lb;
317 * Make a server array as expected by LoadBalancer::__construct()
319 * @param array $serverTemplate Server config map
320 * @param int[][] $groupLoads Map of (group => server name => load)
321 * @return array[] List of server config maps
323 private function makeServerConfigArrays( array $serverTemplate, array $groupLoads ) {
324 // The primary DB server is the first host explicitly listed in the generic load group
325 if ( !$groupLoads[ILoadBalancer::GROUP_GENERIC] ) {
326 throw new UnexpectedValueException( "Empty generic load array; no primary DB defined." );
328 $groupLoadsByServerName = [];
329 foreach ( $groupLoads as $group => $loadByServerName ) {
330 foreach ( $loadByServerName as $serverName => $load ) {
331 $groupLoadsByServerName[$serverName][$group] = $load;
335 // Get the ordered map of (server name => load); the primary DB server is first
336 $genericLoads = $groupLoads[ILoadBalancer::GROUP_GENERIC];
337 // Implicitly append any hosts that only appear in custom load groups
338 $genericLoads += array_fill_keys( array_keys( $groupLoadsByServerName ), 0 );
339 $servers = [];
340 foreach ( $genericLoads as $serverName => $load ) {
341 $servers[] = array_merge(
342 $serverTemplate,
343 $servers ? [] : $this->masterTemplateOverrides,
344 $this->templateOverridesByServer[$serverName] ?? [],
346 'host' => $this->hostsByServerName[$serverName] ?? $serverName,
347 'serverName' => $serverName,
348 'load' => $load,
349 'groupLoads' => $groupLoadsByServerName[$serverName] ?? []
354 return $servers;
358 * @param string $database
359 * @return string Main section name
361 private function getSectionFromDatabase( $database ) {
362 return $this->sectionsByDB[$database]
363 ?? $this->sectionsByDB[self::CLUSTER_MAIN_DEFAULT]
364 ?? self::CLUSTER_MAIN_DEFAULT;
367 public function reconfigure( array $conf ): void {
368 if ( !$conf ) {
369 return;
372 foreach ( $this->mainLBs as $lb ) {
373 // Approximate what LBFactoryMulti::__construct does (T346365)
374 $groupLoads = $conf['groupLoadsBySection'][$lb->getClusterName()] ?? [];
375 $groupLoads[ILoadBalancer::GROUP_GENERIC] = $conf['sectionLoads'][$lb->getClusterName()];
376 $config = [
377 'servers' => $this->makeServerConfigArrays( $conf['serverTemplate'] ?? [], $groupLoads )
379 $lb->reconfigure( $config );
382 foreach ( $this->externalLBs as $lb ) {
383 $groupLoads = [
384 ILoadBalancer::GROUP_GENERIC => $conf['externalLoads'][$lb->getClusterName()]
386 $config = [
387 'servers' => $this->makeServerConfigArrays( $conf['serverTemplate'] ?? [], $groupLoads )
389 $lb->reconfigure( $config );