org.apache.ignite.tensorflow.cluster

Class TensorFlowClusterManager

java.lang.Object

org.apache.ignite.tensorflow.cluster.TensorFlowClusterManager

public class TensorFlowClusterManager
extends Object

TensorFlow cluster manager that allows to start, maintain and stop TensorFlow cluster.

Constructor Summary

Constructors

Constructor and Description
TensorFlowClusterManager(Ignite ignite) Constructs a new instance of TensorFlow cluster manager.

Method Summary

Modifier and Type	Method and Description
TensorFlowCluster	createCluster(UUID clusterId, TensorFlowJobArchive jobArchive, Consumer<String> userScriptOut, Consumer<String> userScriptErr) Creates and starts a new TensorFlow cluster for the specified cache.
void	destroy() Destroys TensorFlow cluster manager and related caches.
Map<UUID,TensorFlowCluster>	getAllClusters() Returns list of maintained TensorFlow clusters.
Exception	getChiefException(UUID clusterId) Returns an exception that happened during execution or null if there is no exception.
TensorFlowCluster	getCluster(UUID clusterId) Returns cluster by identifier.
TensorFlowServerManager	getSrvProcMgr()
Exception	getUserScriptException(UUID clusterId) Returns an exception that happened during execution or null if there is no exception.
boolean	isChiefCompleted(UUID clusterId) Checks if chief completed and returns result.
boolean	isUserScriptCompleted(UUID clusterId) Checks if user script completed and returns result.
void	stopClusterIfExists(UUID clusterId) Stops TensorFlow cluster.

Methods inherited from class java.lang.Object

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

Constructor Detail

TensorFlowClusterManager

public TensorFlowClusterManager(Ignite ignite)

Constructs a new instance of TensorFlow cluster manager.

Parameters:

ignite - Ignite instance.

Method Detail

getCluster

public TensorFlowCluster getCluster(UUID clusterId)

Returns cluster by identifier.

Parameters:

clusterId - Cluster identifier.

Returns:

TensorFlow cluster.

createCluster

public TensorFlowCluster createCluster(UUID clusterId,
                                       TensorFlowJobArchive jobArchive,
                                       Consumer<String> userScriptOut,
                                       Consumer<String> userScriptErr)

Creates and starts a new TensorFlow cluster for the specified cache.

Parameters:

clusterId - Cluster identifier.

jobArchive - Job archive.

Returns:

TensorFlow cluster metadata.

stopClusterIfExists

public void stopClusterIfExists(UUID clusterId)

Stops TensorFlow cluster.

Parameters:

clusterId - TensorFlow cluster identifier.

destroy

public void destroy()

Destroys TensorFlow cluster manager and related caches.

isChiefCompleted

public boolean isChiefCompleted(UUID clusterId)

Checks if chief completed and returns result.

Parameters:

clusterId - Cluster identifier.

Returns:

true if chief completed, otherwise false.

getChiefException

public Exception getChiefException(UUID clusterId)

Returns an exception that happened during execution or null if there is no exception.

Parameters:

clusterId - Cluster identifier.

Returns:

Exception that happened during execution or null if there is no exception.

isUserScriptCompleted

public boolean isUserScriptCompleted(UUID clusterId)

Checks if user script completed and returns result.

Parameters:

clusterId - Cluster identifier.

Returns:

true if user script completed, otherwise false.

getUserScriptException

public Exception getUserScriptException(UUID clusterId)

Returns an exception that happened during execution or null if there is no exception.

Parameters:

clusterId - Cluster identifier.

Returns:

Exception that happened during execution or null if there is no exception.

getAllClusters

public Map<UUID,TensorFlowCluster> getAllClusters()

Returns list of maintained TensorFlow clusters.

Returns:

List of maintained TensorFlow clusters.

getSrvProcMgr

public TensorFlowServerManager getSrvProcMgr()