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
;
29 use DBTransactionError
;
32 use InvalidArgumentException
;
35 * Database cluster connection, tracking, load balancing, and transaction manager interface
37 * A "cluster" is considered to be one master database and zero or more replica databases.
38 * Typically, the replica DBs replicate from the master asynchronously. The first node in the
39 * "servers" configuration array is always considered the "master". However, this class can still
40 * be used when all or some of the "replica" DBs are multi-master peers of the master or even
41 * when all the DBs are non-replicating clones of each other holding read-only data. Thus, the
42 * role of "master" is in some cases merely nominal.
44 * By default, each DB server uses DBO_DEFAULT for its 'flags' setting, unless explicitly set
45 * otherwise in configuration. DBO_DEFAULT behavior depends on whether 'cliMode' is set:
46 * - In CLI mode, the flag has no effect with regards to LoadBalancer.
47 * - In non-CLI mode, the flag causes implicit transactions to be used; the first query on
48 * a database starts a transaction on that database. The transactions are meant to remain
49 * pending until either commitMasterChanges() or rollbackMasterChanges() is called. The
50 * application must have some point where it calls commitMasterChanges() near the end of
52 * Every iteration of beginMasterChanges()/commitMasterChanges() is called a "transaction round".
53 * Rounds are useful on the master DB connections because they make single-DB (and by and large
54 * multi-DB) updates in web requests all-or-nothing. Also, transactions on replica DBs are useful
55 * when REPEATABLE-READ or SERIALIZABLE isolation is used because all foriegn keys and constraints
56 * hold across separate queries in the DB transaction since the data appears within a consistent
57 * point-in-time snapshot.
59 * The typical caller will use LoadBalancer::getConnection( DB_* ) to yield a live database
60 * connection handle. The choice of which DB server to use is based on pre-defined loads for
61 * weighted random selection, adjustments thereof by LoadMonitor, and the amount of replication
62 * lag on each DB server. Lag checks might cause problems in certain setups, so they should be
63 * tuned in the server configuration maps as follows:
64 * - Master + N Replica(s): set 'max lag' to an appropriate threshold for avoiding any database
65 * lagged by this much or more. If all DBs are this lagged, then the load balancer considers
66 * the cluster to be read-only.
67 * - Galera Cluster: Seconds_Behind_Master will be 0, so there probably is nothing to tune.
68 * Note that lag is still possible depending on how wsrep-sync-wait is set server-side.
69 * - Read-only archive clones: set 'is static' in the server configuration maps. This will
70 * treat all such DBs as having 0 lag.
71 * - SQL load balancing proxy: any proxy should handle lag checks on its own, so the 'max lag'
72 * parameter should probably be set to INF in the server configuration maps. This will make
73 * the load balancer ignore whatever it detects as the lag of the logical replica is (which
74 * would probably just randomly bounce around).
76 * If using a SQL proxy service, it would probably be best to have two proxy hosts for the
77 * load balancer to talk to. One would be the 'host' of the master server entry and another for
78 * the (logical) replica server entry. The proxy could map the load balancer's "replica" DB to
79 * any number of physical replica DBs.
84 interface ILoadBalancer
{
85 /** @var integer Request a replica DB connection */
86 const DB_REPLICA
= -1;
87 /** @var integer Request a master DB connection */
90 /** @var string Domain specifier when no specific database needs to be selected */
91 const DOMAIN_ANY
= '';
94 * Construct a manager of IDatabase connection objects
96 * @param array $params Parameter map with keys:
97 * - servers : Required. Array of server info structures.
98 * - localDomain: A DatabaseDomain or domain ID string.
99 * - loadMonitor : Name of a class used to fetch server lag and load.
100 * - readOnlyReason : Reason the master DB is read-only if so [optional]
101 * - waitTimeout : Maximum time to wait for replicas for consistency [optional]
102 * - srvCache : BagOStuff object for server cache [optional]
103 * - memCache : BagOStuff object for cluster memory cache [optional]
104 * - wanCache : WANObjectCache object [optional]
105 * - chronologyProtector: ChronologyProtector object [optional]
106 * - hostname : The name of the current server [optional]
107 * - cliMode: Whether the execution context is a CLI script. [optional]
108 * - profiler : Class name or instance with profileIn()/profileOut() methods. [optional]
109 * - trxProfiler: TransactionProfiler instance. [optional]
110 * - replLogger: PSR-3 logger instance. [optional]
111 * - connLogger: PSR-3 logger instance. [optional]
112 * - queryLogger: PSR-3 logger instance. [optional]
113 * - perfLogger: PSR-3 logger instance. [optional]
114 * - errorLogger : Callback that takes an Exception and logs it. [optional]
115 * @throws InvalidArgumentException
117 public function __construct( array $params );
120 * Get the index of the reader connection, which may be a replica DB
122 * This takes into account load ratios and lag times. It should
123 * always return a consistent index during a given invocation.
125 * Side effect: opens connections to databases
126 * @param string|bool $group Query group, or false for the generic reader
127 * @param string|bool $domain Domain ID, or false for the current domain
129 * @return bool|int|string
131 public function getReaderIndex( $group = false, $domain = false );
134 * Set the master wait position
136 * If a DB_REPLICA connection has been opened already, then wait immediately.
137 * Otherwise sets a variable telling it to wait if such a connection is opened.
139 * @param DBMasterPos|bool $pos Master position or false
141 public function waitFor( $pos );
144 * Set the master wait position and wait for a "generic" replica DB to catch up to it
146 * This can be used a faster proxy for waitForAll()
148 * @param DBMasterPos|bool $pos Master position or false
149 * @param int $timeout Max seconds to wait; default is mWaitTimeout
150 * @return bool Success (able to connect and no timeouts reached)
152 public function waitForOne( $pos, $timeout = null );
155 * Set the master wait position and wait for ALL replica DBs to catch up to it
157 * @param DBMasterPos|bool $pos Master position or false
158 * @param int $timeout Max seconds to wait; default is mWaitTimeout
159 * @return bool Success (able to connect and no timeouts reached)
161 public function waitForAll( $pos, $timeout = null );
164 * Get any open connection to a given server index, local or foreign
166 * @param int $i Server index or DB_MASTER/DB_REPLICA
167 * @return Database|bool False if no such connection is open
169 public function getAnyOpenConnection( $i );
172 * Get a connection by index
174 * @param int $i Server index or DB_MASTER/DB_REPLICA
175 * @param array|string|bool $groups Query group(s), or false for the generic reader
176 * @param string|bool $domain Domain ID, or false for the current domain
181 public function getConnection( $i, $groups = [], $domain = false );
184 * Mark a foreign connection as being available for reuse under a different DB domain
186 * This mechanism is reference-counted, and must be called the same number of times
187 * as getConnection() to work.
189 * @param IDatabase $conn
190 * @throws InvalidArgumentException
192 public function reuseConnection( $conn );
195 * Get a database connection handle reference
197 * The handle's methods simply wrap those of a Database handle
199 * @see ILoadBalancer::getConnection() for parameter information
201 * @param int $i Server index or DB_MASTER/DB_REPLICA
202 * @param array|string|bool $groups Query group(s), or false for the generic reader
203 * @param string|bool $domain Domain ID, or false for the current domain
206 public function getConnectionRef( $i, $groups = [], $domain = false );
209 * Get a database connection handle reference without connecting yet
211 * The handle's methods simply wrap those of a Database handle
213 * @see ILoadBalancer::getConnection() for parameter information
215 * @param int $i Server index or DB_MASTER/DB_REPLICA
216 * @param array|string|bool $groups Query group(s), or false for the generic reader
217 * @param string|bool $domain Domain ID, or false for the current domain
220 public function getLazyConnectionRef( $i, $groups = [], $domain = false );
223 * Get a maintenance database connection handle reference for migrations and schema changes
225 * The handle's methods simply wrap those of a Database handle
227 * @see ILoadBalancer::getConnection() for parameter information
229 * @param int $db Server index or DB_MASTER/DB_REPLICA
230 * @param array|string|bool $groups Query group(s), or false for the generic reader
231 * @param string|bool $domain Domain ID, or false for the current domain
232 * @return MaintainableDBConnRef
234 public function getMaintenanceConnectionRef( $db, $groups = [], $domain = false );
237 * Open a connection to the server given by the specified index
238 * Index must be an actual index into the array.
239 * If the server is already open, returns it.
241 * @note If disable() was called on this LoadBalancer, this method will throw a DBAccessError.
243 * @param int $i Server index or DB_MASTER/DB_REPLICA
244 * @param string|bool $domain Domain ID, or false for the current domain
245 * @return Database|bool Returns false on errors
246 * @throws DBAccessError
248 public function openConnection( $i, $domain = false );
253 public function getWriterIndex();
256 * Returns true if the specified index is a valid server index
261 public function haveIndex( $i );
264 * Returns true if the specified index is valid and has non-zero load
269 public function isNonZeroLoad( $i );
272 * Get the number of defined servers (not the number of open connections)
276 public function getServerCount();
279 * Get the host name or IP address of the server with the specified index
280 * Prefer a readable name if available.
284 public function getServerName( $i );
287 * Return the server info structure for a given index, or false if the index is invalid.
291 public function getServerInfo( $i );
294 * Sets the server info structure for the given index. Entry at index $i
295 * is created if it doesn't exist
297 * @param array $serverInfo
299 public function setServerInfo( $i, array $serverInfo );
302 * Get the current master position for chronology control purposes
303 * @return DBMasterPos|bool Returns false if not applicable
305 public function getMasterPos();
308 * Disable this load balancer. All connections are closed, and any attempt to
309 * open a new connection will result in a DBAccessError.
311 public function disable();
314 * Close all open connections
316 public function closeAll();
321 * Using this function makes sure the LoadBalancer knows the connection is closed.
322 * If you use $conn->close() directly, the load balancer won't update its state.
324 * @param IDatabase $conn
326 public function closeConnection( IDatabase
$conn );
329 * Commit transactions on all open connections
330 * @param string $fname Caller name
331 * @throws DBExpectedError
333 public function commitAll( $fname = __METHOD__
);
336 * Perform all pre-commit callbacks that remain part of the atomic transactions
337 * and disable any post-commit callbacks until runMasterPostTrxCallbacks()
339 * Use this only for mutli-database commits
341 public function finalizeMasterChanges();
344 * Perform all pre-commit checks for things like replication safety
346 * Use this only for mutli-database commits
348 * @param array $options Includes:
349 * - maxWriteDuration : max write query duration time in seconds
350 * @throws DBTransactionError
352 public function approveMasterChanges( array $options );
355 * Flush any master transaction snapshots and set DBO_TRX (if DBO_DEFAULT is set)
357 * The DBO_TRX setting will be reverted to the default in each of these methods:
358 * - commitMasterChanges()
359 * - rollbackMasterChanges()
361 * This allows for custom transaction rounds from any outer transaction scope.
363 * @param string $fname
364 * @throws DBExpectedError
366 public function beginMasterChanges( $fname = __METHOD__
);
369 * Issue COMMIT on all master connections where writes where done
370 * @param string $fname Caller name
371 * @throws DBExpectedError
373 public function commitMasterChanges( $fname = __METHOD__
);
376 * Issue all pending post-COMMIT/ROLLBACK callbacks
378 * Use this only for mutli-database commits
380 * @param int $type IDatabase::TRIGGER_* constant
381 * @return Exception|null The first exception or null if there were none
383 public function runMasterPostTrxCallbacks( $type );
386 * Issue ROLLBACK only on master, only if queries were done on connection
387 * @param string $fname Caller name
388 * @throws DBExpectedError
390 public function rollbackMasterChanges( $fname = __METHOD__
);
393 * Suppress all pending post-COMMIT/ROLLBACK callbacks
395 * Use this only for mutli-database commits
397 * @return Exception|null The first exception or null if there were none
399 public function suppressTransactionEndCallbacks();
402 * Commit all replica DB transactions so as to flush any REPEATABLE-READ or SSI snapshot
404 * @param string $fname Caller name
406 public function flushReplicaSnapshots( $fname = __METHOD__
);
409 * @return bool Whether a master connection is already open
411 public function hasMasterConnection();
414 * Determine if there are pending changes in a transaction by this thread
417 public function hasMasterChanges();
420 * Get the timestamp of the latest write query done by this thread
421 * @return float|bool UNIX timestamp or false
423 public function lastMasterChangeTimestamp();
426 * Check if this load balancer object had any recent or still
427 * pending writes issued against it by this PHP thread
429 * @param float $age How many seconds ago is "recent" [defaults to mWaitTimeout]
432 public function hasOrMadeRecentMasterChanges( $age = null );
435 * Get the list of callers that have pending master changes
437 * @return string[] List of method names
439 public function pendingMasterChangeCallers();
442 * @note This method will trigger a DB connection if not yet done
443 * @param string|bool $domain Domain ID, or false for the current domain
444 * @return bool Whether the generic connection for reads is highly "lagged"
446 public function getLaggedReplicaMode( $domain = false );
449 * @note This method will never cause a new DB connection
450 * @return bool Whether any generic connection used for reads was highly "lagged"
452 public function laggedReplicaUsed();
455 * @note This method may trigger a DB connection if not yet done
456 * @param string|bool $domain Domain ID, or false for the current domain
457 * @param IDatabase|null $conn DB master connection; used to avoid loops [optional]
458 * @return string|bool Reason the master is read-only or false if it is not
460 public function getReadOnlyReason( $domain = false, IDatabase
$conn = null );
463 * Disables/enables lag checks
464 * @param null|bool $mode
467 public function allowLagged( $mode = null );
472 public function pingAll();
475 * Call a function with each open connection object
476 * @param callable $callback
477 * @param array $params
479 public function forEachOpenConnection( $callback, array $params = [] );
482 * Call a function with each open connection object to a master
483 * @param callable $callback
484 * @param array $params
486 public function forEachOpenMasterConnection( $callback, array $params = [] );
489 * Call a function with each open replica DB connection object
490 * @param callable $callback
491 * @param array $params
493 public function forEachOpenReplicaConnection( $callback, array $params = [] );
496 * Get the hostname and lag time of the most-lagged replica DB
498 * This is useful for maintenance scripts that need to throttle their updates.
499 * May attempt to open connections to replica DBs on the default DB. If there is
500 * no lag, the maximum lag will be reported as -1.
502 * @param bool|string $domain Domain ID, or false for the default database
503 * @return array ( host, max lag, index of max lagged host )
505 public function getMaxLag( $domain = false );
508 * Get an estimate of replication lag (in seconds) for each server
510 * Results are cached for a short time in memcached/process cache
512 * Values may be "false" if replication is too broken to estimate
514 * @param string|bool $domain
515 * @return int[] Map of (server index => float|int|bool)
517 public function getLagTimes( $domain = false );
520 * Get the lag in seconds for a given connection, or zero if this load
521 * balancer does not have replication enabled.
523 * This should be used in preference to Database::getLag() in cases where
524 * replication may not be in use, since there is no way to determine if
525 * replication is in use at the connection level without running
526 * potentially restricted queries such as SHOW SLAVE STATUS. Using this
527 * function instead of Database::getLag() avoids a fatal error in this
528 * case on many installations.
530 * @param IDatabase $conn
531 * @return int|bool Returns false on error
533 public function safeGetLag( IDatabase
$conn );
536 * Wait for a replica DB to reach a specified master position
538 * This will connect to the master to get an accurate position if $pos is not given
540 * @param IDatabase $conn Replica DB
541 * @param DBMasterPos|bool $pos Master position; default: current position
542 * @param int $timeout Timeout in seconds [optional]
543 * @return bool Success
545 public function safeWaitForMasterPos( IDatabase
$conn, $pos = false, $timeout = 10 );
548 * Set a callback via IDatabase::setTransactionListener() on
549 * all current and future master connections of this load balancer
551 * @param string $name Callback name
552 * @param callable|null $callback
554 public function setTransactionListener( $name, callable
$callback = null );
557 * Set a new table prefix for the existing local domain ID for testing
559 * @param string $prefix
561 public function setDomainPrefix( $prefix );
564 * Make certain table names use their own database, schema, and table prefix
565 * when passed into SQL queries pre-escaped and without a qualified database name
567 * For example, "user" can be converted to "myschema.mydbname.user" for convenience.
568 * Appearances like `user`, somedb.user, somedb.someschema.user will used literally.
570 * Calling this twice will completely clear any old table aliases. Also, note that
571 * callers are responsible for making sure the schemas and databases actually exist.
573 * @param array[] $aliases Map of (table => (dbname, schema, prefix) map)
575 public function setTableAliases( array $aliases );