Package org.apache.activemq.artemis.core.server.cluster.qourum

Class SharedNothingBackupQuorum

java.lang.Object

org.apache.activemq.artemis.core.server.cluster.qourum.SharedNothingBackupQuorum

All Implemented Interfaces:

SessionFailureListener, FailureListener, Quorum, LiveNodeLocator.BackupRegistrationListener

public class SharedNothingBackupQuorum
extends Object
implements Quorum, SessionFailureListener, LiveNodeLocator.BackupRegistrationListener

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	SharedNothingBackupQuorum.BACKUP_ACTIVATION

Field Summary

Fields

Modifier and Type	Field	Description
static int	WAIT_TIME_AFTER_FIRST_LIVE_STOPPING_MSG	This is a safety net in case the live sends the first ReplicationLiveIsStoppingMessage with code ReplicationLiveIsStoppingMessage.LiveStopping.STOP_CALLED and crashes before sending the second with ReplicationLiveIsStoppingMessage.LiveStopping.FAIL_OVER.

Constructor Summary

Constructors

Constructor	Description
SharedNothingBackupQuorum(NodeManager nodeManager, ScheduledExecutorService scheduledPool, NetworkHealthCheck networkHealthCheck, int quorumSize, int voteRetries, long voteRetryWait, int quorumVoteWait, boolean failback)

Method Summary

Modifier and Type	Method	Description
void	beforeReconnect(ActiveMQException exception)
void	causeExit(SharedNothingBackupQuorum.BACKUP_ACTIVATION explicitSignal)	Cause the Activation thread to exit and the server to be stopped.
void	close()	called if the quorum manager is stopping so we can clean up
void	connectionFailed(ActiveMQException exception, boolean failedOver)	if the connection to our replicated live goes down then decide on an action
void	connectionFailed(ActiveMQException me, boolean failedOver, String scaleDownTargetNodeID)
void	failOver(ReplicationLiveIsStoppingMessage.LiveStopping finalMessage)	Releases the latch, causing the backup activation thread to fail-over.
String	getName()	the name of the Quorum.
void	liveIDSet(String liveID)
void	nodeDown(Topology topology, long eventUID, String nodeID)	if the node going down is the node we are replicating from then decide on an action.
void	nodeUp(Topology topology)	called by the quorum when a node in the quorum goes up
void	onBackupRegistrationFailed(boolean alreadyReplicating)
void	reset()
void	setQuorumManager(QuorumManager quorumManager)	called by the quorum manager when a quorum is registered
void	setSessionFactory(ClientSessionFactoryInternal sessionFactory)
SharedNothingBackupQuorum.BACKUP_ACTIVATION	waitForStatusChange()	Called by the replicating backup (i.e.

Methods inherited from class java.lang.Object

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

Field Detail

WAIT_TIME_AFTER_FIRST_LIVE_STOPPING_MSG

public static final int WAIT_TIME_AFTER_FIRST_LIVE_STOPPING_MSG

This is a safety net in case the live sends the first ReplicationLiveIsStoppingMessage with code ReplicationLiveIsStoppingMessage.LiveStopping.STOP_CALLED and crashes before sending the second with ReplicationLiveIsStoppingMessage.LiveStopping.FAIL_OVER.

If the second message does come within this dead line, we fail over anyway.

See Also:

Constant Field Values

Constructor Detail

SharedNothingBackupQuorum

public SharedNothingBackupQuorum(NodeManager nodeManager,
                                 ScheduledExecutorService scheduledPool,
                                 NetworkHealthCheck networkHealthCheck,
                                 int quorumSize,
                                 int voteRetries,
                                 long voteRetryWait,
                                 int quorumVoteWait,
                                 boolean failback)

Method Detail

getName

public String getName()

Description copied from interface: Quorum

the name of the Quorum. this should be unique and is used to locate the correct quorum to use for voting

Specified by:

getName in interface Quorum

liveIDSet

public void liveIDSet(String liveID)

setQuorumManager

public void setQuorumManager(QuorumManager quorumManager)

Description copied from interface: Quorum

called by the quorum manager when a quorum is registered

Specified by:

setQuorumManager in interface Quorum

nodeDown

public void nodeDown(Topology topology,
                     long eventUID,
                     String nodeID)

if the node going down is the node we are replicating from then decide on an action.

Specified by:

nodeDown in interface Quorum

Parameters:

topology -

eventUID -

nodeID -

nodeUp

public void nodeUp(Topology topology)

Description copied from interface: Quorum

called by the quorum when a node in the quorum goes up

Specified by:

nodeUp in interface Quorum

connectionFailed

public void connectionFailed(ActiveMQException exception,
                             boolean failedOver)

if the connection to our replicated live goes down then decide on an action

Specified by:

connectionFailed in interface FailureListener

connectionFailed

public void connectionFailed(ActiveMQException me,
                             boolean failedOver,
                             String scaleDownTargetNodeID)

Specified by:

connectionFailed in interface FailureListener

beforeReconnect

public void beforeReconnect(ActiveMQException exception)

Specified by:

beforeReconnect in interface SessionFailureListener

close

public void close()

Description copied from interface: Quorum

called if the quorum manager is stopping so we can clean up

Specified by:

close in interface Quorum

setSessionFactory

public void setSessionFactory(ClientSessionFactoryInternal sessionFactory)

Parameters:

sessionFactory - the session factory used to connect to the live server

failOver

public void failOver(ReplicationLiveIsStoppingMessage.LiveStopping finalMessage)

Releases the latch, causing the backup activation thread to fail-over.

The use case is for when the 'live' has an orderly shutdown, in which case it informs the backup that it should fail-over.

onBackupRegistrationFailed

public void onBackupRegistrationFailed(boolean alreadyReplicating)

Specified by:

onBackupRegistrationFailed in interface LiveNodeLocator.BackupRegistrationListener

waitForStatusChange

public SharedNothingBackupQuorum.BACKUP_ACTIVATION waitForStatusChange()

Called by the replicating backup (i.e. "SharedNothing" backup) to wait for the signal to fail-over or to stop.

Returns:

signal, indicating whether to stop or to fail-over

causeExit

public void causeExit(SharedNothingBackupQuorum.BACKUP_ACTIVATION explicitSignal)

Cause the Activation thread to exit and the server to be stopped.

Parameters:

explicitSignal - the state we want to set the quorum manager to return

reset

public void reset()