org.apache.flink.runtime.checkpoint

Class RoundRobinOperatorStateRepartitioner

java.lang.Object

org.apache.flink.runtime.checkpoint.RoundRobinOperatorStateRepartitioner

All Implemented Interfaces:

OperatorStateRepartitioner<OperatorStateHandle>

@Internal
public class RoundRobinOperatorStateRepartitioner
extends Object
implements OperatorStateRepartitioner<OperatorStateHandle>

Current default implementation of OperatorStateRepartitioner that redistributes state in round robin fashion.

Field Summary

Fields

Modifier and Type	Field and Description
static OperatorStateRepartitioner<OperatorStateHandle>	INSTANCE

Constructor Summary

Constructors

Constructor and Description
RoundRobinOperatorStateRepartitioner()

Method Summary

Modifier and Type	Method and Description
List<List<OperatorStateHandle>>	repartitionState(List<List<OperatorStateHandle>> previousParallelSubtaskStates, int oldParallelism, int newParallelism)

Methods inherited from class java.lang.Object

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

Field Detail

INSTANCE

public static final OperatorStateRepartitioner<OperatorStateHandle> INSTANCE

Constructor Detail

RoundRobinOperatorStateRepartitioner

public RoundRobinOperatorStateRepartitioner()

Method Detail

repartitionState

public List<List<OperatorStateHandle>> repartitionState(List<List<OperatorStateHandle>> previousParallelSubtaskStates,
                                                        int oldParallelism,
                                                        int newParallelism)

Specified by:

repartitionState in interface OperatorStateRepartitioner<OperatorStateHandle>

Parameters:

previousParallelSubtaskStates - List with one entry of state handles per parallel subtask of an operator, as they have been checkpointed.

oldParallelism - The parallelism before we start redistribution.

newParallelism - The parallelism that we consider for the state redistribution. Determines the size of the returned list.

Returns:

List with one entry per parallel subtask. Each subtask receives now one collection of states that build of the new total state for this subtask.