3 * Database load balancing interface
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
23 namespace Wikimedia\Rdbms
;
26 use InvalidArgumentException
;
29 * Database cluster connection, tracking, load balancing, and transaction manager interface
31 * A "cluster" is considered to be one master database and zero or more replica databases.
32 * Typically, the replica DBs replicate from the master asynchronously. The first node in the
33 * "servers" configuration array is always considered the "master". However, this class can still
34 * be used when all or some of the "replica" DBs are multi-master peers of the master or even
35 * when all the DBs are non-replicating clones of each other holding read-only data. Thus, the
36 * role of "master" is in some cases merely nominal.
38 * By default, each DB server uses DBO_DEFAULT for its 'flags' setting, unless explicitly set
39 * otherwise in configuration. DBO_DEFAULT behavior depends on whether 'cliMode' is set:
40 * - In CLI mode, the flag has no effect with regards to LoadBalancer.
41 * - In non-CLI mode, the flag causes implicit transactions to be used; the first query on
42 * a database starts a transaction on that database. The transactions are meant to remain
43 * pending until either commitMasterChanges() or rollbackMasterChanges() is called. The
44 * application must have some point where it calls commitMasterChanges() near the end of
46 * Every iteration of beginMasterChanges()/commitMasterChanges() is called a "transaction round".
47 * Rounds are useful on the master DB connections because they make single-DB (and by and large
48 * multi-DB) updates in web requests all-or-nothing. Also, transactions on replica DBs are useful
49 * when REPEATABLE-READ or SERIALIZABLE isolation is used because all foriegn keys and constraints
50 * hold across separate queries in the DB transaction since the data appears within a consistent
51 * point-in-time snapshot.
53 * The typical caller will use LoadBalancer::getConnection( DB_* ) to yield a live database
54 * connection handle. The choice of which DB server to use is based on pre-defined loads for
55 * weighted random selection, adjustments thereof by LoadMonitor, and the amount of replication
56 * lag on each DB server. Lag checks might cause problems in certain setups, so they should be
57 * tuned in the server configuration maps as follows:
58 * - Master + N Replica(s): set 'max lag' to an appropriate threshold for avoiding any database
59 * lagged by this much or more. If all DBs are this lagged, then the load balancer considers
60 * the cluster to be read-only.
61 * - Galera Cluster: Seconds_Behind_Master will be 0, so there probably is nothing to tune.
62 * Note that lag is still possible depending on how wsrep-sync-wait is set server-side.
63 * - Read-only archive clones: set 'is static' in the server configuration maps. This will
64 * treat all such DBs as having 0 lag.
65 * - SQL load balancing proxy: any proxy should handle lag checks on its own, so the 'max lag'
66 * parameter should probably be set to INF in the server configuration maps. This will make
67 * the load balancer ignore whatever it detects as the lag of the logical replica is (which
68 * would probably just randomly bounce around).
70 * If using a SQL proxy service, it would probably be best to have two proxy hosts for the
71 * load balancer to talk to. One would be the 'host' of the master server entry and another for
72 * the (logical) replica server entry. The proxy could map the load balancer's "replica" DB to
73 * any number of physical replica DBs.
78 interface ILoadBalancer
{
79 /** @var int Request a replica DB connection */
80 const DB_REPLICA
= -1;
81 /** @var int Request a master DB connection */
84 /** @var string Domain specifier when no specific database needs to be selected */
85 const DOMAIN_ANY
= '';
87 /** @var int DB handle should have DBO_TRX disabled and the caller will leave it as such */
88 const CONN_TRX_AUTOCOMMIT
= 1;
89 /** @var int Return null on connection failure instead of throwing an exception */
90 const CONN_SILENCE_ERRORS
= 2;
92 /** @var string Manager of ILoadBalancer instances is running post-commit callbacks */
93 const STAGE_POSTCOMMIT_CALLBACKS
= 'stage-postcommit-callbacks';
94 /** @var string Manager of ILoadBalancer instances is running post-rollback callbacks */
95 const STAGE_POSTROLLBACK_CALLBACKS
= 'stage-postrollback-callbacks';
98 * Construct a manager of IDatabase connection objects
100 * @param array $params Parameter map with keys:
101 * - servers : Required. Array of server info structures.
102 * - localDomain: A DatabaseDomain or domain ID string.
103 * - loadMonitor : Name of a class used to fetch server lag and load.
104 * - readOnlyReason : Reason the master DB is read-only if so [optional]
105 * - waitTimeout : Maximum time to wait for replicas for consistency [optional]
106 * - maxLag: Try to avoid DB replicas with lag above this many seconds [optional]
107 * - srvCache : BagOStuff object for server cache [optional]
108 * - wanCache : WANObjectCache object [optional]
109 * - chronologyCallback: Callback to run before the first connection attempt [optional]
110 * - hostname : The name of the current server [optional]
111 * - cliMode: Whether the execution context is a CLI script. [optional]
112 * - profiler : Class name or instance with profileIn()/profileOut() methods. [optional]
113 * - trxProfiler: TransactionProfiler instance. [optional]
114 * - replLogger: PSR-3 logger instance. [optional]
115 * - connLogger: PSR-3 logger instance. [optional]
116 * - queryLogger: PSR-3 logger instance. [optional]
117 * - perfLogger: PSR-3 logger instance. [optional]
118 * - errorLogger : Callback that takes an Exception and logs it. [optional]
119 * - deprecationLogger: Callback to log a deprecation warning. [optional]
120 * - roundStage: STAGE_POSTCOMMIT_* class constant; for internal use [optional]
121 * - ownerId: integer ID of an LBFactory instance that manages this instance [optional]
122 * @throws InvalidArgumentException
124 public function __construct( array $params );
127 * Get the local (and default) database domain ID of connection handles
129 * @see DatabaseDomain
130 * @return string Database domain ID; this specifies DB name, schema, and table prefix
133 public function getLocalDomainID();
136 * @param DatabaseDomain|string|bool $domain Database domain
137 * @return string Value of $domain if it is foreign or the local domain otherwise
140 public function resolveDomainID( $domain );
143 * Close all connection and redefine the local domain for testing or schema creation
145 * @param DatabaseDomain|string $domain
148 public function redefineLocalDomain( $domain );
151 * Get the index of the reader connection, which may be a replica DB
153 * This takes into account load ratios and lag times. It should return a consistent
154 * index during the life time of the load balancer. This initially checks replica DBs
155 * for connectivity to avoid returning an unusable server. This means that connections
156 * might be attempted by calling this method (usally one at the most but possibly more).
157 * Subsequent calls with the same $group will not need to make new connection attempts
158 * since the acquired connection for each group is preserved.
160 * @param string|bool $group Query group, or false for the generic group
161 * @param string|bool $domain Domain ID, or false for the current domain
163 * @return bool|int|string
165 public function getReaderIndex( $group = false, $domain = false );
168 * Set the master position to reach before the next generic group DB handle query
170 * If a generic replica DB connection is already open then this immediately waits
171 * for that DB to catch up to the specified replication position. Otherwise, it will
172 * do so once such a connection is opened.
174 * If a timeout happens when waiting, then getLaggedReplicaMode()/laggedReplicaUsed()
177 * @param DBMasterPos|bool $pos Master position or false
179 public function waitFor( $pos );
182 * Set the master wait position and wait for a generic replica DB to catch up to it
184 * This can be used a faster proxy for waitForAll()
186 * @param DBMasterPos|bool $pos Master position or false
187 * @param int|null $timeout Max seconds to wait; default is mWaitTimeout
188 * @return bool Success (able to connect and no timeouts reached)
190 public function waitForOne( $pos, $timeout = null );
193 * Set the master wait position and wait for ALL replica DBs to catch up to it
195 * @param DBMasterPos|bool $pos Master position or false
196 * @param int|null $timeout Max seconds to wait; default is mWaitTimeout
197 * @return bool Success (able to connect and no timeouts reached)
199 public function waitForAll( $pos, $timeout = null );
202 * Get any open connection to a given server index, local or foreign
204 * Use CONN_TRX_AUTOCOMMIT to only look for connections opened with that flag.
205 * Avoid the use of begin() or startAtomic() on any such connections.
207 * @param int $i Server index or DB_MASTER/DB_REPLICA
208 * @param int $flags Bitfield of CONN_* class constants
209 * @return Database|bool False if no such connection is open
211 public function getAnyOpenConnection( $i, $flags = 0 );
214 * Get a connection handle by server index
216 * The CONN_TRX_AUTOCOMMIT flag is ignored for databases with ATTR_DB_LEVEL_LOCKING
217 * (e.g. sqlite) in order to avoid deadlocks. ILoadBalancer::getServerAttributes()
218 * can be used to check such flags beforehand.
220 * If the caller uses $domain or sets CONN_TRX_AUTOCOMMIT in $flags, then it must
221 * also call ILoadBalancer::reuseConnection() on the handle when finished using it.
222 * In all other cases, this is not necessary, though not harmful either.
223 * Avoid the use of begin() or startAtomic() on any such connections.
225 * @param int $i Server index (overrides $groups) or DB_MASTER/DB_REPLICA
226 * @param array|string|bool $groups Query group(s), or false for the generic reader
227 * @param string|bool $domain Domain ID, or false for the current domain
228 * @param int $flags Bitfield of CONN_* class constants
230 * @note This method throws DBAccessError if ILoadBalancer::disable() was called
232 * @throws DBError If any error occurs that prevents the yielding of a (live) IDatabase
233 * @return IDatabase|bool This returns false on failure if CONN_SILENCE_ERRORS is set
235 public function getConnection( $i, $groups = [], $domain = false, $flags = 0 );
238 * Mark a foreign connection as being available for reuse under a different DB domain
240 * This mechanism is reference-counted, and must be called the same number of times
241 * as getConnection() to work.
243 * @param IDatabase $conn
244 * @throws InvalidArgumentException
246 public function reuseConnection( IDatabase
$conn );
249 * Get a database connection handle reference
251 * The handle's methods simply wrap those of a Database handle
253 * The CONN_TRX_AUTOCOMMIT flag is ignored for databases with ATTR_DB_LEVEL_LOCKING
254 * (e.g. sqlite) in order to avoid deadlocks. ILoadBalancer::getServerAttributes()
255 * can be used to check such flags beforehand. Avoid the use of begin() or startAtomic()
256 * on any CONN_TRX_AUTOCOMMIT connections.
258 * @see ILoadBalancer::getConnection() for parameter information
260 * @param int $i Server index or DB_MASTER/DB_REPLICA
261 * @param array|string|bool $groups Query group(s), or false for the generic reader
262 * @param string|bool $domain Domain ID, or false for the current domain
263 * @param int $flags Bitfield of CONN_* class constants (e.g. CONN_TRX_AUTOCOMMIT)
266 public function getConnectionRef( $i, $groups = [], $domain = false, $flags = 0 );
269 * Get a database connection handle reference without connecting yet
271 * The handle's methods simply wrap those of a Database handle
273 * The CONN_TRX_AUTOCOMMIT flag is ignored for databases with ATTR_DB_LEVEL_LOCKING
274 * (e.g. sqlite) in order to avoid deadlocks. ILoadBalancer::getServerAttributes()
275 * can be used to check such flags beforehand. Avoid the use of begin() or startAtomic()
276 * on any CONN_TRX_AUTOCOMMIT connections.
278 * @see ILoadBalancer::getConnection() for parameter information
280 * @param int $i Server index or DB_MASTER/DB_REPLICA
281 * @param array|string|bool $groups Query group(s), or false for the generic reader
282 * @param string|bool $domain Domain ID, or false for the current domain
283 * @param int $flags Bitfield of CONN_* class constants (e.g. CONN_TRX_AUTOCOMMIT)
286 public function getLazyConnectionRef( $i, $groups = [], $domain = false, $flags = 0 );
289 * Get a maintenance database connection handle reference for migrations and schema changes
291 * The handle's methods simply wrap those of a Database handle
293 * The CONN_TRX_AUTOCOMMIT flag is ignored for databases with ATTR_DB_LEVEL_LOCKING
294 * (e.g. sqlite) in order to avoid deadlocks. ILoadBalancer::getServerAttributes()
295 * can be used to check such flags beforehand. Avoid the use of begin() or startAtomic()
296 * on any CONN_TRX_AUTOCOMMIT connections.
298 * @see ILoadBalancer::getConnection() for parameter information
300 * @param int $i Server index or DB_MASTER/DB_REPLICA
301 * @param array|string|bool $groups Query group(s), or false for the generic reader
302 * @param string|bool $domain Domain ID, or false for the current domain
303 * @param int $flags Bitfield of CONN_* class constants (e.g. CONN_TRX_AUTOCOMMIT)
304 * @return MaintainableDBConnRef
306 public function getMaintenanceConnectionRef( $i, $groups = [], $domain = false, $flags = 0 );
311 public function getWriterIndex();
314 * Returns true if the specified index is a valid server index
319 public function haveIndex( $i );
322 * Returns true if the specified index is valid and has non-zero load
327 public function isNonZeroLoad( $i );
330 * Get the number of defined servers (not the number of open connections)
334 public function getServerCount();
337 * Get the host name or IP address of the server with the specified index
340 * @return string Readable name if available or IP/host otherwise
342 public function getServerName( $i );
345 * Return the server info structure for a given index, or false if the index is invalid.
350 public function getServerInfo( $i );
353 * Get DB type of the server with the specified index
356 * @return string One of (mysql,postgres,sqlite,...) or "unknown" for bad indexes
359 public function getServerType( $i );
362 * @param int $i Server index
363 * @return array (Database::ATTRIBUTE_* constant => value) for all such constants
366 public function getServerAttributes( $i );
369 * Get the current master position for chronology control purposes
370 * @return DBMasterPos|bool Returns false if not applicable
372 public function getMasterPos();
375 * Disable this load balancer. All connections are closed, and any attempt to
376 * open a new connection will result in a DBAccessError.
378 public function disable();
381 * Close all open connections
383 public function closeAll();
388 * Using this function makes sure the LoadBalancer knows the connection is closed.
389 * If you use $conn->close() directly, the load balancer won't update its state.
391 * @param IDatabase $conn
393 public function closeConnection( IDatabase
$conn );
396 * Commit transactions on all open connections
397 * @param string $fname Caller name
398 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
399 * @throws DBExpectedError
401 public function commitAll( $fname = __METHOD__
, $owner = null );
404 * Run pre-commit callbacks and defer execution of post-commit callbacks
406 * Use this only for mutli-database commits
408 * @param string $fname Caller name
409 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
410 * @return int Number of pre-commit callbacks run (since 1.32)
412 public function finalizeMasterChanges( $fname = __METHOD__
, $owner = null );
415 * Perform all pre-commit checks for things like replication safety
417 * Use this only for mutli-database commits
419 * @param array $options Includes:
420 * - maxWriteDuration : max write query duration time in seconds
421 * @param string $fname Caller name
422 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
423 * @throws DBTransactionError
425 public function approveMasterChanges( array $options, $fname, $owner = null );
428 * Flush any master transaction snapshots and set DBO_TRX (if DBO_DEFAULT is set)
430 * The DBO_TRX setting will be reverted to the default in each of these methods:
431 * - commitMasterChanges()
432 * - rollbackMasterChanges()
434 * This allows for custom transaction rounds from any outer transaction scope.
436 * @param string $fname Caller name
437 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
438 * @throws DBExpectedError
440 public function beginMasterChanges( $fname = __METHOD__
, $owner = null );
443 * Issue COMMIT on all open master connections to flush changes and view snapshots
444 * @param string $fname Caller name
445 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
446 * @throws DBExpectedError
448 public function commitMasterChanges( $fname = __METHOD__
, $owner = null );
451 * Consume and run all pending post-COMMIT/ROLLBACK callbacks and commit dangling transactions
453 * @param string $fname Caller name
454 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
455 * @return Exception|null The first exception or null if there were none
457 public function runMasterTransactionIdleCallbacks( $fname = __METHOD__
, $owner = null );
460 * Run all recurring post-COMMIT/ROLLBACK listener callbacks
462 * @param string $fname Caller name
463 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
464 * @return Exception|null The first exception or null if there were none
466 public function runMasterTransactionListenerCallbacks( $fname = __METHOD__
, $owner = null );
469 * Issue ROLLBACK only on master, only if queries were done on connection
470 * @param string $fname Caller name
471 * @param int|null $owner ID of the calling instance (e.g. the LBFactory ID)
472 * @throws DBExpectedError
474 public function rollbackMasterChanges( $fname = __METHOD__
, $owner = null );
477 * Commit all replica DB transactions so as to flush any REPEATABLE-READ or SSI snapshots
479 * @param string $fname Caller name
481 public function flushReplicaSnapshots( $fname = __METHOD__
);
484 * Commit all master DB transactions so as to flush any REPEATABLE-READ or SSI snapshots
486 * An error will be thrown if a connection has pending writes or callbacks
488 * @param string $fname Caller name
490 public function flushMasterSnapshots( $fname = __METHOD__
);
493 * @return bool Whether a master connection is already open
495 public function hasMasterConnection();
498 * Whether there are pending changes or callbacks in a transaction by this thread
501 public function hasMasterChanges();
504 * Get the timestamp of the latest write query done by this thread
505 * @return float|bool UNIX timestamp or false
507 public function lastMasterChangeTimestamp();
510 * Check if this load balancer object had any recent or still
511 * pending writes issued against it by this PHP thread
513 * @param float|null $age How many seconds ago is "recent" [defaults to mWaitTimeout]
516 public function hasOrMadeRecentMasterChanges( $age = null );
519 * Get the list of callers that have pending master changes
521 * @return string[] List of method names
523 public function pendingMasterChangeCallers();
526 * @note This method will trigger a DB connection if not yet done
527 * @param string|bool $domain Domain ID, or false for the current domain
528 * @return bool Whether the database for generic connections this request is highly "lagged"
530 public function getLaggedReplicaMode( $domain = false );
533 * Checks whether the database for generic connections this request was both:
534 * - a) Already choosen due to a prior connection attempt
535 * - b) Considered highly "lagged"
537 * @note This method will never cause a new DB connection
540 public function laggedReplicaUsed();
543 * @note This method may trigger a DB connection if not yet done
544 * @param string|bool $domain Domain ID, or false for the current domain
545 * @param IDatabase|null $conn DB master connection; used to avoid loops [optional]
546 * @return string|bool Reason the master is read-only or false if it is not
548 public function getReadOnlyReason( $domain = false, IDatabase
$conn = null );
551 * Disables/enables lag checks
552 * @param null|bool $mode
555 public function allowLagged( $mode = null );
560 public function pingAll();
563 * Call a function with each open connection object
564 * @param callable $callback
565 * @param array $params
567 public function forEachOpenConnection( $callback, array $params = [] );
570 * Call a function with each open connection object to a master
571 * @param callable $callback
572 * @param array $params
574 public function forEachOpenMasterConnection( $callback, array $params = [] );
577 * Call a function with each open replica DB connection object
578 * @param callable $callback
579 * @param array $params
581 public function forEachOpenReplicaConnection( $callback, array $params = [] );
584 * Get the hostname and lag time of the most-lagged replica DB
586 * This is useful for maintenance scripts that need to throttle their updates.
587 * May attempt to open connections to replica DBs on the default DB. If there is
588 * no lag, the maximum lag will be reported as -1.
590 * @param bool|string $domain Domain ID, or false for the default database
591 * @return array ( host, max lag, index of max lagged host )
593 public function getMaxLag( $domain = false );
596 * Get an estimate of replication lag (in seconds) for each server
598 * Results are cached for a short time in memcached/process cache
600 * Values may be "false" if replication is too broken to estimate
602 * @param string|bool $domain
603 * @return int[] Map of (server index => float|int|bool)
605 public function getLagTimes( $domain = false );
608 * Wait for a replica DB to reach a specified master position
610 * This will connect to the master to get an accurate position if $pos is not given
612 * @param IDatabase $conn Replica DB
613 * @param DBMasterPos|bool $pos Master position; default: current position
614 * @param int $timeout Timeout in seconds [optional]
615 * @return bool Success
617 public function safeWaitForMasterPos( IDatabase
$conn, $pos = false, $timeout = 10 );
620 * Set a callback via IDatabase::setTransactionListener() on
621 * all current and future master connections of this load balancer
623 * @param string $name Callback name
624 * @param callable|null $callback
626 public function setTransactionListener( $name, callable
$callback = null );
629 * Set a new table prefix for the existing local domain ID for testing
631 * @param string $prefix
634 public function setLocalDomainPrefix( $prefix );
637 * Make certain table names use their own database, schema, and table prefix
638 * when passed into SQL queries pre-escaped and without a qualified database name
640 * For example, "user" can be converted to "myschema.mydbname.user" for convenience.
641 * Appearances like `user`, somedb.user, somedb.someschema.user will used literally.
643 * Calling this twice will completely clear any old table aliases. Also, note that
644 * callers are responsible for making sure the schemas and databases actually exist.
646 * @param array[] $aliases Map of (table => (dbname, schema, prefix) map)
648 public function setTableAliases( array $aliases );
651 * Convert certain index names to alternative names before querying the DB
653 * Note that this applies to indexes regardless of the table they belong to.
655 * This can be employed when an index was renamed X => Y in code, but the new Y-named
656 * indexes were not yet built on all DBs. After all the Y-named ones are added by the DBA,
657 * the aliases can be removed, and then the old X-named indexes dropped.
659 * @param string[] $aliases
662 public function setIndexAliases( array $aliases );