com.atlassian.beehive.db

Class DatabaseClusterLockService

java.lang.Object

com.atlassian.beehive.db.DatabaseClusterLockService

All Implemented Interfaces:

ClusterLockService, ManagedClusterLockService

public class DatabaseClusterLockService
extends Object
implements ManagedClusterLockService

This is the database backed implementation of ClusterLockService that can be used by clustered applications.

This implementation relies on the host application providing SPI implementations for Data Access objects.

Constructor Summary

Constructors

Constructor and Description
DatabaseClusterLockService(ClusterLockDao clusterLockDao)

Method Summary

Modifier and Type	Method and Description
Collection<com.atlassian.beehive.db.DatabaseClusterLock>	getAllKnownClusterLocks() Returns all the cluster locks that are currently known to this cluster lock service.
ClusterLock	getLockForName(String lockName) Returns the cluster-wide lock for the given name.
Collection<ClusterLockStatus>	getStatusesOfAllHeldClusterLocks() Returns lock status of every cluster lock being held.
void	onDbErased() Notify the service about removal of any records in tables used by SPI dao implementations during runtime.
void	shutdown()

Methods inherited from class java.lang.Object

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

Constructor Detail

DatabaseClusterLockService

public DatabaseClusterLockService(ClusterLockDao clusterLockDao)

Method Detail

getAllKnownClusterLocks

public Collection<com.atlassian.beehive.db.DatabaseClusterLock> getAllKnownClusterLocks()

Description copied from interface: ManagedClusterLockService

Returns all the cluster locks that are currently known to this cluster lock service.

The specifics are mostly left up to the implementation, with the following caveats:

• It MUST include all locks that are currently locked by this node.
• It SHOULD include all locks that are still strongly referenced outside of the cluster lock service, if possible. Implementations that are unable to comply with this requirement MUST clearly document that limitation.
• It MAY include locks that were previously accessed but are no longer in use, even those that are no longer strongly referenced outside of the cluster lock service.
• It MAY include locks that have not been accessed by this node but are still known to the service through the activity of other nodes within the cluster.

Consumers MUST NOT rely on this method for the correctness of their locking behaviour. This information is intended for instrumentation and other similar management concerns, only.

Specified by:

getAllKnownClusterLocks in interface ManagedClusterLockService

Returns:

all cluster locks that are currently known to this service, in no particular order

getStatusesOfAllHeldClusterLocks

@Nonnull
public Collection<ClusterLockStatus> getStatusesOfAllHeldClusterLocks()

Description copied from interface: ManagedClusterLockService

Returns lock status of every cluster lock being held.

This is intended to provide additional information about misbehaving locks and thus this method should return the most up-to-date and non-cached data.

Consumers must be aware that not all implementations can return a valid update time of the locks.

Specified by:

getStatusesOfAllHeldClusterLocks in interface ManagedClusterLockService

Returns:

a collection of cluster lock statuses, in no particular order

getLockForName

public ClusterLock getLockForName(@Nonnull
                                  String lockName)

Description copied from interface: ClusterLockService

Returns the cluster-wide lock for the given name.

Specified by:

getLockForName in interface ClusterLockService

Parameters:

lockName - the globally unique lock identifier

Returns:

the cluster-wide lock for the given name.

shutdown

@PreDestroy
public void shutdown()

onDbErased

public void onDbErased()

Notify the service about removal of any records in tables used by SPI dao implementations during runtime. Warning: Failinig to notify immediately after the db changes affecting lock tables or notyfying in any other situation than that one may result in unexpected locks' behavior.