org.jivesoftware.util.cache
Interface CacheFactoryStrategy

All Known Implementing Classes:

DefaultLocalCacheStrategy

public interface CacheFactoryStrategy

Implementation of CacheFactory that relies on the specific clustering solution.

Author:

Gaston Dombiak

Method Summary

Cache	createCache(String name) Creates a new cache for the cache name specified.
void	destroyCache(Cache cache) Destroys the supplied cache.
void	doClusterTask(ClusterTask task) Invokes a task on other cluster members in an asynchronous fashion.
boolean	doClusterTask(ClusterTask task, byte[] nodeID) Invokes a task on other the specified cluster member in an asynchronous fashion.
Collection<Object>	doSynchronousClusterTask(ClusterTask task, boolean includeLocalMember) Invokes a task on other cluster members synchronously and returns the result as a Collection (method will not return until the task has been executed on each cluster member).
Object	doSynchronousClusterTask(ClusterTask task, byte[] nodeID) Invokes a task on a given cluster member synchronously and returns the result of the remote operation.
byte[]	getClusterMemberID() Returns a byte[] that uniquely identifies this member within the cluster or null when not in a cluster.
ClusterNodeInfo	getClusterNodeInfo(byte[] nodeID) Returns the node info for the given cluster node
Collection<ClusterNodeInfo>	getClusterNodesInfo() Returns basic information about the current members of the cluster or an empty collection if not running in a cluster.
long	getClusterTime() Gets the pseudo-synchronized time from the cluster.
Lock	getLock(Object key, Cache cache) Returns an existing lock on the specified key or creates a new one if none was found.
int	getMaxClusterNodes() Returns the maximum number of cluster members allowed.
String	getPluginName() Get the plugin name corresponding to this clustering implementation
byte[]	getSeniorClusterMemberID() Returns a byte[] that uniquely identifies this senior cluster member or null when not in a cluster.
boolean	isSeniorClusterMember() Returns true if this node is the maste node of the cluster.
boolean	startCluster() Returns true if the cluster has been started.
void	stopCluster() Stops the cluster.
void	updateCacheStats(Map<String,Cache> caches) Updates the statistics of the specified caches and publishes them into a cache for statistics.

Method Detail

startCluster

boolean startCluster()

Returns true if the cluster has been started. When running in local mode a true value should be returned.

An error should be logged when the cluster fails to be started.

Returns:

true if the cluster has been started.

stopCluster

void stopCluster()

Stops the cluster. When not running in a cluster this request will be ignored.

createCache

Cache createCache(String name)

Creates a new cache for the cache name specified. The created cache is already configured. Different implementations could store the cache configuration in different ways. It is recommended to store the cache configuration in an external file so it is easier for customers to change the default configuration.

Parameters:

name - name of the cache to create.

Returns:

newly created and configured cache.

destroyCache

void destroyCache(Cache cache)

Destroys the supplied cache.

Parameters:

cache - the cache to destroy.

isSeniorClusterMember

boolean isSeniorClusterMember()

Returns true if this node is the maste node of the cluster. When not running in cluster mode a value of true should be returned.

Returns:

true if this node is the maste node of the cluster.

getClusterNodesInfo

Collection<ClusterNodeInfo> getClusterNodesInfo()

Returns basic information about the current members of the cluster or an empty collection if not running in a cluster.

Returns:

information about the current members of the cluster or an empty collection if not running in a cluster.

getMaxClusterNodes

int getMaxClusterNodes()

Returns the maximum number of cluster members allowed. A value of 0 will be returned when clustering is not allowed.

Returns:

the maximum number of cluster members allowed or 0 if clustering is not allowed.

getSeniorClusterMemberID

byte[] getSeniorClusterMemberID()

Returns a byte[] that uniquely identifies this senior cluster member or null when not in a cluster.

Returns:

a byte[] that uniquely identifies this senior cluster member or null when not in a cluster.

getClusterMemberID

byte[] getClusterMemberID()

Returns a byte[] that uniquely identifies this member within the cluster or null when not in a cluster.

Returns:

a byte[] that uniquely identifies this member within the cluster or null when not in a cluster.

getClusterTime

long getClusterTime()

Gets the pseudo-synchronized time from the cluster. While the cluster members may have varying system times, this method is expected to return a timestamp that is synchronized (or nearly so; best effort) across the cluster.

Returns:

Synchronized time for all cluster members

doClusterTask

void doClusterTask(ClusterTask task)

Invokes a task on other cluster members in an asynchronous fashion. The task will not be executed on the local cluster member. If clustering is not enabled, this method will do nothing.

Parameters:

task - the task to be invoked on all other cluster members.

doClusterTask

boolean doClusterTask(ClusterTask task,
                      byte[] nodeID)

Invokes a task on other the specified cluster member in an asynchronous fashion. If clustering is not enabled, this method will do nothing.

Parameters:

task - the task to be invoked on the specified cluster member.

nodeID - the byte array that identifies the target cluster member.

Returns:

false if not in a cluster or specified cluster node was not found.

doSynchronousClusterTask

Collection<Object> doSynchronousClusterTask(ClusterTask task,
                                            boolean includeLocalMember)

Invokes a task on other cluster members synchronously and returns the result as a Collection (method will not return until the task has been executed on each cluster member). The task will not be executed on the local cluster member. If clustering is not enabled, this method will return an empty collection.

Parameters:

task - the ClusterTask object to be invoked on all other cluster members.

includeLocalMember - true to run the task on the local member, false otherwise

Returns:

collection with the result of the execution.

doSynchronousClusterTask

Object doSynchronousClusterTask(ClusterTask task,
                                byte[] nodeID)

Invokes a task on a given cluster member synchronously and returns the result of the remote operation. If clustering is not enabled, this method will return null.

Parameters:

task - the ClusterTask object to be invoked on a given cluster member.

nodeID - the byte array that identifies the target cluster member.

Returns:

result of remote operation or null if operation failed or operation returned null.

Throws:

IllegalStateException - if requested node was not found.

updateCacheStats

void updateCacheStats(Map<String,Cache> caches)

Updates the statistics of the specified caches and publishes them into a cache for statistics. The statistics cache is already known to the application but this could change in the future (?). When not in cluster mode then do nothing.

The statistics cache must contain a long array of 5 positions for each cache with the following content:

1. cache.getCacheSize()
2. cache.getMaxCacheSize()
3. cache.size()
4. cache.getCacheHits()
5. cache.getCacheMisses()

Parameters:

caches - caches to get their stats and publish them in a statistics cache.

getLock

Lock getLock(Object key,
             Cache cache)

Returns an existing lock on the specified key or creates a new one if none was found. This operation is thread safe. The supplied cache may or may not be used depending whether the server is running on cluster mode or not. When not running as part of a cluster then the lock will be unrelated to the cache and will only be visible in this JVM.

Parameters:

key - the object that defines the visibility or scope of the lock.

cache - the cache used for holding the lock.

Returns:

an existing lock on the specified key or creates a new one if none was found.

getPluginName

String getPluginName()

Get the plugin name corresponding to this clustering implementation

Returns:

the plugin name for this clustering implementation

getClusterNodeInfo

ClusterNodeInfo getClusterNodeInfo(byte[] nodeID)

Returns the node info for the given cluster node

Parameters:

nodeID - The target cluster node

Returns:

The info for the cluster node, or null if not found