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
22 * @author Aaron Schulz
24 namespace Wikimedia\Rdbms
;
27 use InvalidArgumentException
;
30 * Database cluster connection, tracking, load balancing, and transaction manager interface
32 * A "cluster" is considered to be one master database and zero or more replica databases.
33 * Typically, the replica DBs replicate from the master asynchronously. The first node in the
34 * "servers" configuration array is always considered the "master". However, this class can still
35 * be used when all or some of the "replica" DBs are multi-master peers of the master or even
36 * when all the DBs are non-replicating clones of each other holding read-only data. Thus, the
37 * role of "master" is in some cases merely nominal.
39 * By default, each DB server uses DBO_DEFAULT for its 'flags' setting, unless explicitly set
40 * otherwise in configuration. DBO_DEFAULT behavior depends on whether 'cliMode' is set:
41 * - In CLI mode, the flag has no effect with regards to LoadBalancer.
42 * - In non-CLI mode, the flag causes implicit transactions to be used; the first query on
43 * a database starts a transaction on that database. The transactions are meant to remain
44 * pending until either commitMasterChanges() or rollbackMasterChanges() is called. The
45 * application must have some point where it calls commitMasterChanges() near the end of
47 * Every iteration of beginMasterChanges()/commitMasterChanges() is called a "transaction round".
48 * Rounds are useful on the master DB connections because they make single-DB (and by and large
49 * multi-DB) updates in web requests all-or-nothing. Also, transactions on replica DBs are useful
50 * when REPEATABLE-READ or SERIALIZABLE isolation is used because all foriegn keys and constraints
51 * hold across separate queries in the DB transaction since the data appears within a consistent
52 * point-in-time snapshot.
54 * The typical caller will use LoadBalancer::getConnection( DB_* ) to yield a live database
55 * connection handle. The choice of which DB server to use is based on pre-defined loads for
56 * weighted random selection, adjustments thereof by LoadMonitor, and the amount of replication
57 * lag on each DB server. Lag checks might cause problems in certain setups, so they should be
58 * tuned in the server configuration maps as follows:
59 * - Master + N Replica(s): set 'max lag' to an appropriate threshold for avoiding any database
60 * lagged by this much or more. If all DBs are this lagged, then the load balancer considers
61 * the cluster to be read-only.
62 * - Galera Cluster: Seconds_Behind_Master will be 0, so there probably is nothing to tune.
63 * Note that lag is still possible depending on how wsrep-sync-wait is set server-side.
64 * - Read-only archive clones: set 'is static' in the server configuration maps. This will
65 * treat all such DBs as having 0 lag.
66 * - SQL load balancing proxy: any proxy should handle lag checks on its own, so the 'max lag'
67 * parameter should probably be set to INF in the server configuration maps. This will make
68 * the load balancer ignore whatever it detects as the lag of the logical replica is (which
69 * would probably just randomly bounce around).
71 * If using a SQL proxy service, it would probably be best to have two proxy hosts for the
72 * load balancer to talk to. One would be the 'host' of the master server entry and another for
73 * the (logical) replica server entry. The proxy could map the load balancer's "replica" DB to
74 * any number of physical replica DBs.
79 interface ILoadBalancer
{
80 /** @var integer Request a replica DB connection */
81 const DB_REPLICA
= -1;
82 /** @var integer Request a master DB connection */
85 /** @var string Domain specifier when no specific database needs to be selected */
86 const DOMAIN_ANY
= '';
89 * Construct a manager of IDatabase connection objects
91 * @param array $params Parameter map with keys:
92 * - servers : Required. Array of server info structures.
93 * - localDomain: A DatabaseDomain or domain ID string.
94 * - loadMonitor : Name of a class used to fetch server lag and load.
95 * - readOnlyReason : Reason the master DB is read-only if so [optional]
96 * - waitTimeout : Maximum time to wait for replicas for consistency [optional]
97 * - srvCache : BagOStuff object for server cache [optional]
98 * - memCache : BagOStuff object for cluster memory cache [optional]
99 * - wanCache : WANObjectCache object [optional]
100 * - chronologyProtector: ChronologyProtector object [optional]
101 * - hostname : The name of the current server [optional]
102 * - cliMode: Whether the execution context is a CLI script. [optional]
103 * - profiler : Class name or instance with profileIn()/profileOut() methods. [optional]
104 * - trxProfiler: TransactionProfiler instance. [optional]
105 * - replLogger: PSR-3 logger instance. [optional]
106 * - connLogger: PSR-3 logger instance. [optional]
107 * - queryLogger: PSR-3 logger instance. [optional]
108 * - perfLogger: PSR-3 logger instance. [optional]
109 * - errorLogger : Callback that takes an Exception and logs it. [optional]
110 * @throws InvalidArgumentException
112 public function __construct( array $params );
115 * Get the index of the reader connection, which may be a replica DB
117 * This takes into account load ratios and lag times. It should
118 * always return a consistent index during a given invocation.
120 * Side effect: opens connections to databases
121 * @param string|bool $group Query group, or false for the generic reader
122 * @param string|bool $domain Domain ID, or false for the current domain
124 * @return bool|int|string
126 public function getReaderIndex( $group = false, $domain = false );
129 * Set the master wait position
131 * If a DB_REPLICA connection has been opened already, then wait immediately.
132 * Otherwise sets a variable telling it to wait if such a connection is opened.
134 * This only applies to connections to the generic replica DB for this request.
135 * If a timeout happens when waiting, then getLaggedReplicaMode()/laggedReplicaUsed()
138 * @param DBMasterPos|bool $pos Master position or false
140 public function waitFor( $pos );
143 * Set the master wait position and wait for a "generic" replica DB to catch up to it
145 * This can be used a faster proxy for waitForAll()
147 * @param DBMasterPos|bool $pos Master position or false
148 * @param int $timeout Max seconds to wait; default is mWaitTimeout
149 * @return bool Success (able to connect and no timeouts reached)
151 public function waitForOne( $pos, $timeout = null );
154 * Set the master wait position and wait for ALL replica DBs to catch up to it
156 * @param DBMasterPos|bool $pos Master position or false
157 * @param int $timeout Max seconds to wait; default is mWaitTimeout
158 * @return bool Success (able to connect and no timeouts reached)
160 public function waitForAll( $pos, $timeout = null );
163 * Get any open connection to a given server index, local or foreign
165 * @param int $i Server index or DB_MASTER/DB_REPLICA
166 * @return Database|bool False if no such connection is open
168 public function getAnyOpenConnection( $i );
171 * Get a connection by index
173 * @param int $i Server index or DB_MASTER/DB_REPLICA
174 * @param array|string|bool $groups Query group(s), or false for the generic reader
175 * @param string|bool $domain Domain ID, or false for the current domain
180 public function getConnection( $i, $groups = [], $domain = false );
183 * Mark a foreign connection as being available for reuse under a different DB domain
185 * This mechanism is reference-counted, and must be called the same number of times
186 * as getConnection() to work.
188 * @param IDatabase $conn
189 * @throws InvalidArgumentException
191 public function reuseConnection( $conn );
194 * Get a database connection handle reference
196 * The handle's methods simply wrap those of a Database handle
198 * @see ILoadBalancer::getConnection() for parameter information
200 * @param int $i Server index or DB_MASTER/DB_REPLICA
201 * @param array|string|bool $groups Query group(s), or false for the generic reader
202 * @param string|bool $domain Domain ID, or false for the current domain
205 public function getConnectionRef( $i, $groups = [], $domain = false );
208 * Get a database connection handle reference without connecting yet
210 * The handle's methods simply wrap those of a Database handle
212 * @see ILoadBalancer::getConnection() for parameter information
214 * @param int $i Server index or DB_MASTER/DB_REPLICA
215 * @param array|string|bool $groups Query group(s), or false for the generic reader
216 * @param string|bool $domain Domain ID, or false for the current domain
219 public function getLazyConnectionRef( $i, $groups = [], $domain = false );
222 * Get a maintenance database connection handle reference for migrations and schema changes
224 * The handle's methods simply wrap those of a Database handle
226 * @see ILoadBalancer::getConnection() for parameter information
228 * @param int $db Server index or DB_MASTER/DB_REPLICA
229 * @param array|string|bool $groups Query group(s), or false for the generic reader
230 * @param string|bool $domain Domain ID, or false for the current domain
231 * @return MaintainableDBConnRef
233 public function getMaintenanceConnectionRef( $db, $groups = [], $domain = false );
236 * Open a connection to the server given by the specified index
237 * Index must be an actual index into the array.
238 * If the server is already open, returns it.
240 * @note If disable() was called on this LoadBalancer, this method will throw a DBAccessError.
242 * @param int $i Server index or DB_MASTER/DB_REPLICA
243 * @param string|bool $domain Domain ID, or false for the current domain
244 * @return Database|bool Returns false on errors
245 * @throws DBAccessError
247 public function openConnection( $i, $domain = false );
252 public function getWriterIndex();
255 * Returns true if the specified index is a valid server index
260 public function haveIndex( $i );
263 * Returns true if the specified index is valid and has non-zero load
268 public function isNonZeroLoad( $i );
271 * Get the number of defined servers (not the number of open connections)
275 public function getServerCount();
278 * Get the host name or IP address of the server with the specified index
279 * Prefer a readable name if available.
283 public function getServerName( $i );
286 * Return the server info structure for a given index, or false if the index is invalid.
290 public function getServerInfo( $i );
293 * Sets the server info structure for the given index. Entry at index $i
294 * is created if it doesn't exist
296 * @param array $serverInfo
298 public function setServerInfo( $i, array $serverInfo );
301 * Get the current master position for chronology control purposes
302 * @return DBMasterPos|bool Returns false if not applicable
304 public function getMasterPos();
307 * Disable this load balancer. All connections are closed, and any attempt to
308 * open a new connection will result in a DBAccessError.
310 public function disable();
313 * Close all open connections
315 public function closeAll();
320 * Using this function makes sure the LoadBalancer knows the connection is closed.
321 * If you use $conn->close() directly, the load balancer won't update its state.
323 * @param IDatabase $conn
325 public function closeConnection( IDatabase
$conn );
328 * Commit transactions on all open connections
329 * @param string $fname Caller name
330 * @throws DBExpectedError
332 public function commitAll( $fname = __METHOD__
);
335 * Perform all pre-commit callbacks that remain part of the atomic transactions
336 * and disable any post-commit callbacks until runMasterPostTrxCallbacks()
338 * Use this only for mutli-database commits
340 public function finalizeMasterChanges();
343 * Perform all pre-commit checks for things like replication safety
345 * Use this only for mutli-database commits
347 * @param array $options Includes:
348 * - maxWriteDuration : max write query duration time in seconds
349 * @throws DBTransactionError
351 public function approveMasterChanges( array $options );
354 * Flush any master transaction snapshots and set DBO_TRX (if DBO_DEFAULT is set)
356 * The DBO_TRX setting will be reverted to the default in each of these methods:
357 * - commitMasterChanges()
358 * - rollbackMasterChanges()
360 * This allows for custom transaction rounds from any outer transaction scope.
362 * @param string $fname
363 * @throws DBExpectedError
365 public function beginMasterChanges( $fname = __METHOD__
);
368 * Issue COMMIT on all master connections where writes where done
369 * @param string $fname Caller name
370 * @throws DBExpectedError
372 public function commitMasterChanges( $fname = __METHOD__
);
375 * Issue all pending post-COMMIT/ROLLBACK callbacks
377 * Use this only for mutli-database commits
379 * @param int $type IDatabase::TRIGGER_* constant
380 * @return Exception|null The first exception or null if there were none
382 public function runMasterPostTrxCallbacks( $type );
385 * Issue ROLLBACK only on master, only if queries were done on connection
386 * @param string $fname Caller name
387 * @throws DBExpectedError
389 public function rollbackMasterChanges( $fname = __METHOD__
);
392 * Suppress all pending post-COMMIT/ROLLBACK callbacks
394 * Use this only for mutli-database commits
396 * @return Exception|null The first exception or null if there were none
398 public function suppressTransactionEndCallbacks();
401 * Commit all replica DB transactions so as to flush any REPEATABLE-READ or SSI snapshot
403 * @param string $fname Caller name
405 public function flushReplicaSnapshots( $fname = __METHOD__
);
408 * @return bool Whether a master connection is already open
410 public function hasMasterConnection();
413 * Determine if there are pending changes in a transaction by this thread
416 public function hasMasterChanges();
419 * Get the timestamp of the latest write query done by this thread
420 * @return float|bool UNIX timestamp or false
422 public function lastMasterChangeTimestamp();
425 * Check if this load balancer object had any recent or still
426 * pending writes issued against it by this PHP thread
428 * @param float $age How many seconds ago is "recent" [defaults to mWaitTimeout]
431 public function hasOrMadeRecentMasterChanges( $age = null );
434 * Get the list of callers that have pending master changes
436 * @return string[] List of method names
438 public function pendingMasterChangeCallers();
441 * @note This method will trigger a DB connection if not yet done
442 * @param string|bool $domain Domain ID, or false for the current domain
443 * @return bool Whether the database for generic connections this request is highly "lagged"
445 public function getLaggedReplicaMode( $domain = false );
448 * Checks whether the database for generic connections this request was both:
449 * - a) Already choosen due to a prior connection attempt
450 * - b) Considered highly "lagged"
452 * @note This method will never cause a new DB connection
455 public function laggedReplicaUsed();
458 * @note This method may trigger a DB connection if not yet done
459 * @param string|bool $domain Domain ID, or false for the current domain
460 * @param IDatabase|null $conn DB master connection; used to avoid loops [optional]
461 * @return string|bool Reason the master is read-only or false if it is not
463 public function getReadOnlyReason( $domain = false, IDatabase
$conn = null );
466 * Disables/enables lag checks
467 * @param null|bool $mode
470 public function allowLagged( $mode = null );
475 public function pingAll();
478 * Call a function with each open connection object
479 * @param callable $callback
480 * @param array $params
482 public function forEachOpenConnection( $callback, array $params = [] );
485 * Call a function with each open connection object to a master
486 * @param callable $callback
487 * @param array $params
489 public function forEachOpenMasterConnection( $callback, array $params = [] );
492 * Call a function with each open replica DB connection object
493 * @param callable $callback
494 * @param array $params
496 public function forEachOpenReplicaConnection( $callback, array $params = [] );
499 * Get the hostname and lag time of the most-lagged replica DB
501 * This is useful for maintenance scripts that need to throttle their updates.
502 * May attempt to open connections to replica DBs on the default DB. If there is
503 * no lag, the maximum lag will be reported as -1.
505 * @param bool|string $domain Domain ID, or false for the default database
506 * @return array ( host, max lag, index of max lagged host )
508 public function getMaxLag( $domain = false );
511 * Get an estimate of replication lag (in seconds) for each server
513 * Results are cached for a short time in memcached/process cache
515 * Values may be "false" if replication is too broken to estimate
517 * @param string|bool $domain
518 * @return int[] Map of (server index => float|int|bool)
520 public function getLagTimes( $domain = false );
523 * Get the lag in seconds for a given connection, or zero if this load
524 * balancer does not have replication enabled.
526 * This should be used in preference to Database::getLag() in cases where
527 * replication may not be in use, since there is no way to determine if
528 * replication is in use at the connection level without running
529 * potentially restricted queries such as SHOW SLAVE STATUS. Using this
530 * function instead of Database::getLag() avoids a fatal error in this
531 * case on many installations.
533 * @param IDatabase $conn
534 * @return int|bool Returns false on error
536 public function safeGetLag( IDatabase
$conn );
539 * Wait for a replica DB to reach a specified master position
541 * This will connect to the master to get an accurate position if $pos is not given
543 * @param IDatabase $conn Replica DB
544 * @param DBMasterPos|bool $pos Master position; default: current position
545 * @param int $timeout Timeout in seconds [optional]
546 * @return bool Success
548 public function safeWaitForMasterPos( IDatabase
$conn, $pos = false, $timeout = 10 );
551 * Set a callback via IDatabase::setTransactionListener() on
552 * all current and future master connections of this load balancer
554 * @param string $name Callback name
555 * @param callable|null $callback
557 public function setTransactionListener( $name, callable
$callback = null );
560 * Set a new table prefix for the existing local domain ID for testing
562 * @param string $prefix
564 public function setDomainPrefix( $prefix );
567 * Make certain table names use their own database, schema, and table prefix
568 * when passed into SQL queries pre-escaped and without a qualified database name
570 * For example, "user" can be converted to "myschema.mydbname.user" for convenience.
571 * Appearances like `user`, somedb.user, somedb.someschema.user will used literally.
573 * Calling this twice will completely clear any old table aliases. Also, note that
574 * callers are responsible for making sure the schemas and databases actually exist.
576 * @param array[] $aliases Map of (table => (dbname, schema, prefix) map)
578 public function setTableAliases( array $aliases );