org.apache.flink.runtime.scheduler.strategy

类 VertexwiseSchedulingStrategy

java.lang.Object

org.apache.flink.runtime.scheduler.strategy.VertexwiseSchedulingStrategy

所有已实现的接口:

SchedulingTopologyListener, SchedulingStrategy

public class VertexwiseSchedulingStrategy
extends Object
implements SchedulingStrategy, SchedulingTopologyListener

SchedulingStrategy instance which schedules tasks in granularity of vertex (which indicates this strategy only supports ALL_EDGES_BLOCKING batch jobs). Note that this strategy implements SchedulingTopologyListener, so it can handle the updates of scheduling topology.

嵌套类概要

嵌套类

限定符和类型	类和说明
static class	VertexwiseSchedulingStrategy.Factory The factory for creating VertexwiseSchedulingStrategy.

构造器概要

构造器

构造器和说明
VertexwiseSchedulingStrategy(SchedulerOperations schedulerOperations, SchedulingTopology schedulingTopology)

方法概要

限定符和类型	方法和说明
void	notifySchedulingTopologyUpdated(SchedulingTopology schedulingTopology, List<ExecutionVertexID> newExecutionVertices) Notifies that the scheduling topology is just updated.
void	onExecutionStateChange(ExecutionVertexID executionVertexId, ExecutionState executionState) Called whenever an Execution changes its state.
void	onPartitionConsumable(IntermediateResultPartitionID resultPartitionId) Called whenever an IntermediateResultPartition becomes consumable.
void	restartTasks(Set<ExecutionVertexID> verticesToRestart) Called whenever vertices need to be restarted (due to task failure).
void	startScheduling() Called when the scheduling is started (initial scheduling operation).

从类继承的方法 java.lang.Object

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

构造器详细资料

VertexwiseSchedulingStrategy

public VertexwiseSchedulingStrategy(SchedulerOperations schedulerOperations,
                                    SchedulingTopology schedulingTopology)

方法详细资料

startScheduling

public void startScheduling()

从接口复制的说明: SchedulingStrategy

Called when the scheduling is started (initial scheduling operation).

指定者:

startScheduling 在接口中 SchedulingStrategy

restartTasks

public void restartTasks(Set<ExecutionVertexID> verticesToRestart)

从接口复制的说明: SchedulingStrategy

Called whenever vertices need to be restarted (due to task failure).

指定者:

restartTasks 在接口中 SchedulingStrategy

参数:

verticesToRestart - The tasks need to be restarted

onExecutionStateChange

public void onExecutionStateChange(ExecutionVertexID executionVertexId,
                                   ExecutionState executionState)

从接口复制的说明: SchedulingStrategy

Called whenever an Execution changes its state.

指定者:

onExecutionStateChange 在接口中 SchedulingStrategy

参数:

executionVertexId - The id of the task

executionState - The new state of the execution

onPartitionConsumable

public void onPartitionConsumable(IntermediateResultPartitionID resultPartitionId)

从接口复制的说明: SchedulingStrategy

Called whenever an IntermediateResultPartition becomes consumable.

指定者:

onPartitionConsumable 在接口中 SchedulingStrategy

参数:

resultPartitionId - The id of the result partition

notifySchedulingTopologyUpdated

public void notifySchedulingTopologyUpdated(SchedulingTopology schedulingTopology,
                                            List<ExecutionVertexID> newExecutionVertices)

从接口复制的说明: SchedulingTopologyListener

Notifies that the scheduling topology is just updated.

指定者:

notifySchedulingTopologyUpdated 在接口中 SchedulingTopologyListener

参数:

schedulingTopology - the scheduling topology which is just updated

newExecutionVertices - the newly added execution vertices.