Package org.apache.flink.runtime.executiongraph

Class ExecutionJobVertex

java.lang.Object

org.apache.flink.runtime.executiongraph.ExecutionJobVertex

All Implemented Interfaces:

org.apache.flink.api.common.Archiveable<ArchivedExecutionJobVertex>, AccessExecutionJobVertex

Direct Known Subclasses:

SpeculativeExecutionJobVertex

public class ExecutionJobVertex
extends Object
implements AccessExecutionJobVertex, org.apache.flink.api.common.Archiveable<ArchivedExecutionJobVertex>

An ExecutionJobVertex is part of the ExecutionGraph, and the peer to the JobVertex.

The ExecutionJobVertex corresponds to a parallelized operation. It contains an ExecutionVertex for each parallel instance of that operation.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	ExecutionJobVertex.Factory	Factory to create ExecutionJobVertex.

Constructor Summary

Constructors

Constructor	Description
ExecutionJobVertex(InternalExecutionGraphAccessor graph, JobVertex jobVertex, VertexParallelismInformation parallelismInfo, CoordinatorStore coordinatorStore, JobManagerJobMetricGroup jobManagerJobMetricGroup)

Method Summary

Modifier and Type	Method	Description
ArchivedExecutionJobVertex	archive()
void	cancel()	Cancels all currently running vertex executions.
CompletableFuture<Void>	cancelWithFuture()	Cancels all currently running vertex executions.
boolean	canRescaleMaxParallelism(int desiredMaxParallelism)
void	connectToPredecessors(Map<IntermediateDataSetID,IntermediateResult> intermediateDataSets)
protected ExecutionVertex	createExecutionVertex(ExecutionJobVertex jobVertex, int subTaskIndex, IntermediateResult[] producedDataSets, Duration timeout, long createTimestamp, int executionHistorySizeLimit, int initialAttemptCount)
protected OperatorCoordinatorHolder	createOperatorCoordinatorHolder(org.apache.flink.util.SerializedValue<OperatorCoordinator.Provider> provider, ClassLoader classLoader, CoordinatorStore coordinatorStore, JobManagerJobMetricGroup jobManagerJobMetricGroup)
void	fail(Throwable t)
StringifiedAccumulatorResult[]	getAggregatedUserAccumulatorsStringified()	Returns the aggregated user-defined accumulators as strings.
static ExecutionState	getAggregateJobVertexState(int[] verticesPerState, int parallelism)	A utility function that computes an "aggregated" state for the vertex.
ExecutionState	getAggregateState()	Returns the aggregated ExecutionState for this job vertex.
CoLocationGroup	getCoLocationGroup()
InternalExecutionGraphAccessor	getGraph()
List<IntermediateResult>	getInputs()
org.apache.flink.api.common.JobID	getJobId()
JobVertex	getJobVertex()
JobVertexID	getJobVertexId()	Returns the JobVertexID for this job vertex.
int	getMaxParallelism()	Returns the max parallelism for this job vertex.
String	getName()	Returns the name for this job vertex.
Collection<OperatorCoordinatorHolder>	getOperatorCoordinators()
List<OperatorIDPair>	getOperatorIDs()	Returns a list containing the ID pairs of all operators contained in this execution job vertex.
int	getParallelism()	Returns the parallelism for this job vertex.
IntermediateResult[]	getProducedDataSets()
ResourceProfile	getResourceProfile()	Returns the resource profile for this job vertex.
SlotSharingGroup	getSlotSharingGroup()	Returns the slot sharing group for this job vertex.
List<SourceCoordinator<?,?>>	getSourceCoordinators()
org.apache.flink.core.io.InputSplitAssigner	getSplitAssigner()
TaskInformation	getTaskInformation()
org.apache.flink.types.Either<org.apache.flink.util.SerializedValue<TaskInformation>,PermanentBlobKey>	getTaskInformationOrBlobKey()
ExecutionVertex[]	getTaskVertices()	Returns all execution vertices for this job vertex.
protected void	initialize(int executionHistorySizeLimit, Duration timeout, long createTimestamp, SubtaskAttemptNumberStore initialAttemptCounts)
protected void	initialize(int executionHistorySizeLimit, Duration timeout, long createTimestamp, SubtaskAttemptNumberStore initialAttemptCounts, ExecutionPlanSchedulingContext executionPlanSchedulingContext)
boolean	isFinished()
boolean	isInitialized()
boolean	isParallelismDecided()
void	setMaxParallelism(int maxParallelism)
void	setParallelism(int parallelism)
CompletableFuture<Void>	suspend()

Methods inherited from class java.lang.Object

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

Constructor Details

ExecutionJobVertex

@VisibleForTesting
public ExecutionJobVertex(InternalExecutionGraphAccessor graph,
 JobVertex jobVertex,
 VertexParallelismInformation parallelismInfo,
 CoordinatorStore coordinatorStore,
 JobManagerJobMetricGroup jobManagerJobMetricGroup)
                   throws JobException

Throws:

JobException

Method Details

initialize

@VisibleForTesting
protected void initialize(int executionHistorySizeLimit,
 Duration timeout,
 long createTimestamp,
 SubtaskAttemptNumberStore initialAttemptCounts)
                   throws JobException

Throws:

JobException

initialize

protected void initialize(int executionHistorySizeLimit,
 Duration timeout,
 long createTimestamp,
 SubtaskAttemptNumberStore initialAttemptCounts,
 ExecutionPlanSchedulingContext executionPlanSchedulingContext)
                   throws JobException

Throws:

JobException

createExecutionVertex

protected ExecutionVertex createExecutionVertex(ExecutionJobVertex jobVertex,
 int subTaskIndex,
 IntermediateResult[] producedDataSets,
 Duration timeout,
 long createTimestamp,
 int executionHistorySizeLimit,
 int initialAttemptCount)

createOperatorCoordinatorHolder

protected OperatorCoordinatorHolder createOperatorCoordinatorHolder(org.apache.flink.util.SerializedValue<OperatorCoordinator.Provider> provider,
 ClassLoader classLoader,
 CoordinatorStore coordinatorStore,
 JobManagerJobMetricGroup jobManagerJobMetricGroup)
                                                             throws Exception

Throws:

Exception

isInitialized

public boolean isInitialized()

isParallelismDecided

public boolean isParallelismDecided()

getOperatorIDs

public List<OperatorIDPair> getOperatorIDs()

Returns a list containing the ID pairs of all operators contained in this execution job vertex.

Returns:

list containing the ID pairs of all contained operators

setMaxParallelism

public void setMaxParallelism(int maxParallelism)

getGraph

public InternalExecutionGraphAccessor getGraph()

setParallelism

public void setParallelism(int parallelism)

getJobVertex

public JobVertex getJobVertex()

getName

public String getName()

Description copied from interface: AccessExecutionJobVertex

Returns the name for this job vertex.

Specified by:

getName in interface AccessExecutionJobVertex

Returns:

name for this job vertex.

getParallelism

public int getParallelism()

Description copied from interface: AccessExecutionJobVertex

Returns the parallelism for this job vertex.

Specified by:

getParallelism in interface AccessExecutionJobVertex

Returns:

parallelism for this job vertex.

getMaxParallelism

public int getMaxParallelism()

Description copied from interface: AccessExecutionJobVertex

Returns the max parallelism for this job vertex.

Specified by:

getMaxParallelism in interface AccessExecutionJobVertex

Returns:

max parallelism for this job vertex.

getResourceProfile

public ResourceProfile getResourceProfile()

Description copied from interface: AccessExecutionJobVertex

Returns the resource profile for this job vertex.

Specified by:

getResourceProfile in interface AccessExecutionJobVertex

Returns:

resource profile for this job vertex.

canRescaleMaxParallelism

public boolean canRescaleMaxParallelism(int desiredMaxParallelism)

getJobId

public org.apache.flink.api.common.JobID getJobId()

getJobVertexId

public JobVertexID getJobVertexId()

Description copied from interface: AccessExecutionJobVertex

Returns the JobVertexID for this job vertex.

Specified by:

getJobVertexId in interface AccessExecutionJobVertex

Returns:

JobVertexID for this job vertex.

getTaskVertices

public ExecutionVertex[] getTaskVertices()

Description copied from interface: AccessExecutionJobVertex

Returns all execution vertices for this job vertex.

Specified by:

getTaskVertices in interface AccessExecutionJobVertex

Returns:

all execution vertices for this job vertex

getProducedDataSets

public IntermediateResult[] getProducedDataSets()

getSplitAssigner

public org.apache.flink.core.io.InputSplitAssigner getSplitAssigner()

getSlotSharingGroup

public SlotSharingGroup getSlotSharingGroup()

Description copied from interface: AccessExecutionJobVertex

Returns the slot sharing group for this job vertex.

Specified by:

getSlotSharingGroup in interface AccessExecutionJobVertex

Returns:

slot sharing group for this job vertex.

getCoLocationGroup

@Nullable
public CoLocationGroup getCoLocationGroup()

getInputs

public List<IntermediateResult> getInputs()

getOperatorCoordinators

public Collection<OperatorCoordinatorHolder> getOperatorCoordinators()

getSourceCoordinators

public List<SourceCoordinator<?,?>> getSourceCoordinators()

getTaskInformationOrBlobKey

public org.apache.flink.types.Either<org.apache.flink.util.SerializedValue<TaskInformation>,PermanentBlobKey> getTaskInformationOrBlobKey()
                                                                                                                                   throws IOException

Throws:

IOException

getTaskInformation

public TaskInformation getTaskInformation()

getAggregateState

public ExecutionState getAggregateState()

Description copied from interface: AccessExecutionJobVertex

Returns the aggregated ExecutionState for this job vertex.

Specified by:

getAggregateState in interface AccessExecutionJobVertex

Returns:

aggregated state for this job vertex

connectToPredecessors

public void connectToPredecessors(Map<IntermediateDataSetID,IntermediateResult> intermediateDataSets)
                           throws JobException

Throws:

JobException

cancel

public void cancel()

Cancels all currently running vertex executions.

cancelWithFuture

public CompletableFuture<Void> cancelWithFuture()

Cancels all currently running vertex executions.

Returns:

A future that is complete once all tasks have canceled.

suspend

public CompletableFuture<Void> suspend()

fail

public void fail(Throwable t)

isFinished

public boolean isFinished()

getAggregatedUserAccumulatorsStringified

public StringifiedAccumulatorResult[] getAggregatedUserAccumulatorsStringified()

Description copied from interface: AccessExecutionJobVertex

Returns the aggregated user-defined accumulators as strings.

Specified by:

getAggregatedUserAccumulatorsStringified in interface AccessExecutionJobVertex

Returns:

aggregated user-defined accumulators as strings.

archive

public ArchivedExecutionJobVertex archive()

Specified by:

archive in interface org.apache.flink.api.common.Archiveable<ArchivedExecutionJobVertex>

getAggregateJobVertexState

public static ExecutionState getAggregateJobVertexState(int[] verticesPerState,
 int parallelism)

A utility function that computes an "aggregated" state for the vertex.

This state is not used anywhere in the coordination, but can be used for display in dashboards to as a summary for how the particular parallel operation represented by this ExecutionJobVertex is currently behaving.

For example, if at least one parallel task is failed, the aggregate state is failed. If not, and at least one parallel task is cancelling (or cancelled), the aggregate state is cancelling (or cancelled). If all tasks are finished, the aggregate state is finished, and so on.

Parameters:

verticesPerState - The number of vertices in each state (indexed by the ordinal of the ExecutionState values).

parallelism - The parallelism of the ExecutionJobVertex

Returns:

The aggregate state of this ExecutionJobVertex.