org.mariadb.jdbc.internal.failover.impl

Class AuroraListener

java.lang.Object

org.mariadb.jdbc.internal.failover.AbstractMastersListener

org.mariadb.jdbc.internal.failover.AbstractMastersSlavesListener

org.mariadb.jdbc.internal.failover.impl.MastersSlavesListener

org.mariadb.jdbc.internal.failover.impl.AuroraListener

All Implemented Interfaces:

Listener

public class AuroraListener
extends MastersSlavesListener

Field Summary

Fields

Modifier and Type	Field and Description
private java.util.regex.Pattern	auroraDnsPattern
private java.lang.String	clusterDnsSuffix
private HostAddress	clusterHostAddress
private static java.util.logging.Logger	logger

Fields inherited from class org.mariadb.jdbc.internal.failover.impl.MastersSlavesListener

masterProtocol, secondaryProtocol

Fields inherited from class org.mariadb.jdbc.internal.failover.AbstractMastersSlavesListener

waitNewMasterProtocol, waitNewSecondaryProtocol

Fields inherited from class org.mariadb.jdbc.internal.failover.AbstractMastersListener

currentConnectionAttempts, currentProtocol, currentReadOnlyAsked, explicitClosed, globalInfo, lastQueryNanos, lastRetry, proxy, urlParser

Constructor Summary

Constructors

Constructor and Description
AuroraListener(UrlParser urlParser, GlobalStateInfo globalInfo) Constructor for Aurora.

Method Summary

Modifier and Type	Method and Description
boolean	checkMasterStatus(SearchFilter searchFilter) Check master status.
private HostAddress	findClusterHostAddress(UrlParser urlParser) Retrieves the cluster host address from the UrlParser instance.
HostAddress	getClusterHostAddress()
private java.util.List<java.lang.String>	getCurrentEndpointIdentifiers(Protocol protocol) Retrieves all endpoints of a cluster from the appropriate database table.
void	reconnectFailedConnection(SearchFilter initialSearchFilter) Search a valid connection for failed one.
void	retrieveAllEndpointsAndSet(Protocol protocol) Retrieves the information necessary to add a new endpoint.
HostAddress	searchByStartName(Protocol secondaryProtocol, java.util.List<HostAddress> loopAddress) Looks for the current master/writer instance via the secondary protocol if it is found within 3 attempts.
private HostAddress	searchForMasterHostAddress(Protocol protocol, java.util.List<HostAddress> loopAddress) Aurora replica doesn't have the master endpoint but the master instance name.
private void	setUrlParserFromEndpoints(java.util.List<java.lang.String> endpoints, int port) Sets urlParser accordingly to discovered hosts.

Methods inherited from class org.mariadb.jdbc.internal.failover.impl.MastersSlavesListener

checkInitialConnection, checkWaitingConnection, connectedHosts, foundActiveMaster, foundActiveSecondary, getCatalog, getMajorServerVersion, getServerThreadId, getTimeout, handleFailLoop, initializeConnection, invoke, isClosed, isMasterConnected, isMasterConnection, isValid, lockAndSwitchMaster, lockAndSwitchSecondary, noBackslashEscapes, preAbort, preClose, preExecute, primaryFail, prolog, reconnect, removeListenerFromSchedulers, rePrepareOnSlave, reset, secondaryFail, sessionStateAware, switchReadOnlyConnection, versionGreaterOrEqual

Methods inherited from class org.mariadb.jdbc.internal.failover.AbstractMastersSlavesListener

getFilterForFailedHost, getSecondaryHostFailNanos, handleFailover, hasHostFail, isMasterHostFailReconnect, isSecondaryHostFail, isSecondaryHostFailReconnect, resetMasterFailoverData, resetSecondaryFailoverData, setSecondaryHostFail

Methods inherited from class org.mariadb.jdbc.internal.failover.AbstractMastersListener

abortConnection, addToBlacklist, canRetryFailLoop, clearBlacklist, closeConnection, getBlacklistKeys, getCurrentProtocol, getLastQueryNanos, getMasterHostFailNanos, getProxy, getRetriesAllDown, getUrlParser, invoke, isAutoReconnect, isExplicitClosed, isMasterHostFail, isQueryRelaunchable, isReadOnly, pingMasterProtocol, preAutoReconnect, relaunchOperation, removeFromBlacklist, resetOldsBlackListHosts, setMasterHostFail, setProxy, setSessionReadOnly, syncConnection, throwFailoverMessage

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

logger

private static final java.util.logging.Logger logger

auroraDnsPattern

private final java.util.regex.Pattern auroraDnsPattern

clusterHostAddress

private final HostAddress clusterHostAddress

clusterDnsSuffix

private java.lang.String clusterDnsSuffix

Constructor Detail

AuroraListener

public AuroraListener(UrlParser urlParser,
                      GlobalStateInfo globalInfo)
               throws java.sql.SQLException

Constructor for Aurora. This differ from standard failover because : - we don't know current master, we must check that after initial connection - master can change after he has a failover

Parameters:

urlParser - connection information

globalInfo - server global variables information

Throws:

java.sql.SQLException - when connection string contain host with different cluster

Method Detail

findClusterHostAddress

private HostAddress findClusterHostAddress(UrlParser urlParser)
                                    throws java.sql.SQLException

Retrieves the cluster host address from the UrlParser instance.

Parameters:

urlParser - object that holds the connection information

Returns:

cluster host address

Throws:

java.sql.SQLException

getClusterHostAddress

public HostAddress getClusterHostAddress()

reconnectFailedConnection

public void reconnectFailedConnection(SearchFilter initialSearchFilter)
                               throws java.sql.SQLException

Search a valid connection for failed one. A Node can be a master or a replica depending on the cluster state. so search for each host until found all the failed connection. By default, search for the host not down, and recheck the down one after if not found valid connections.

Specified by:

reconnectFailedConnection in interface Listener

Overrides:

reconnectFailedConnection in class MastersSlavesListener

Parameters:

initialSearchFilter - initial search filter

Throws:

java.sql.SQLException - if a connection asked is not found

retrieveAllEndpointsAndSet

public void retrieveAllEndpointsAndSet(Protocol protocol)
                                throws java.sql.SQLException

Retrieves the information necessary to add a new endpoint. Calls the methods that retrieves the instance identifiers and sets urlParser accordingly.

Parameters:

protocol - current protocol connected to

Throws:

java.sql.SQLException - if connection error occur

getCurrentEndpointIdentifiers

private java.util.List<java.lang.String> getCurrentEndpointIdentifiers(Protocol protocol)
                                                                throws java.sql.SQLException

Retrieves all endpoints of a cluster from the appropriate database table.

Parameters:

protocol - current protocol connected to

Returns:

instance endpoints of the cluster

Throws:

java.sql.SQLException - if connection error occur

setUrlParserFromEndpoints

private void setUrlParserFromEndpoints(java.util.List<java.lang.String> endpoints,
                                       int port)

Sets urlParser accordingly to discovered hosts.

Parameters:

endpoints - instance identifiers

port - port that is common to all endpoints

searchByStartName

public HostAddress searchByStartName(Protocol secondaryProtocol,
                                     java.util.List<HostAddress> loopAddress)

Looks for the current master/writer instance via the secondary protocol if it is found within 3 attempts. Should it not be able to connect, the host is blacklisted and null is returned. Otherwise, it will open a new connection to the cluster endpoint and retrieve the data from there.

Parameters:

secondaryProtocol - the current secondary protocol

loopAddress - list of possible hosts

Returns:

the probable master address or null if not found

searchForMasterHostAddress

private HostAddress searchForMasterHostAddress(Protocol protocol,
                                               java.util.List<HostAddress> loopAddress)
                                        throws java.sql.SQLException

Aurora replica doesn't have the master endpoint but the master instance name. since the end point normally use the instance name like "instance-name.some_unique_string.region.rds.amazonaws.com", if an endpoint start with this instance name, it will be checked first. Otherwise, the endpoint ending string is extracted and used since the writer was newly created.

Parameters:

protocol - current protocol

loopAddress - list of possible hosts

Returns:

the probable host address or null if no valid endpoint found

Throws:

java.sql.SQLException - if any connection error occur

checkMasterStatus

public boolean checkMasterStatus(SearchFilter searchFilter)

Description copied from class: MastersSlavesListener

Check master status.

Specified by:

checkMasterStatus in interface Listener

Overrides:

checkMasterStatus in class MastersSlavesListener

Parameters:

searchFilter - search filter

Returns:

has some status changed