org.apache.flink.runtime.executiongraph

Class Execution

java.lang.Object

org.apache.flink.runtime.executiongraph.Execution

All Implemented Interfaces:

Serializable

public class Execution
extends Object
implements Serializable

A single execution of a vertex. While an ExecutionVertex can be executed multiple times (for recovery, or other re-computation), this class tracks the state of a single execution of that vertex and the resources. NOTE ABOUT THE DESIGN RATIONAL: In several points of the code, we need to deal with possible concurrent state changes and actions. For example, while the call to deploy a task (send it to the TaskManager) happens, the task gets cancelled. We could lock the entire portion of the code (decision to deploy, deploy, set state to running) such that it is guaranteed that any "cancel command" will only pick up after deployment is done and that the "cancel command" call will never overtake the deploying call. This blocks the threads big time, because the remote calls may take long. Depending of their locking behavior, it may even result in distributed deadlocks (unless carefully avoided). We therefore use atomic state updates and occasional double-checking to ensure that the state after a completed call is as expected, and trigger correcting actions if it is not. Many actions are also idempotent (like canceling).

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Execution(scala.concurrent.ExecutionContext executionContext, ExecutionVertex vertex, int attemptNumber, long startTimestamp, scala.concurrent.duration.FiniteDuration timeout)

Method Summary

Modifier and Type	Method and Description
void	cancel()
void	deployToSlot(SimpleSlot slot)
void	fail(Throwable t) This method fails the vertex due to an external condition.
SimpleSlot	getAssignedResource()
InstanceConnectionInfo	getAssignedResourceLocation()
ExecutionAttemptID	getAttemptId()
int	getAttemptNumber()
Throwable	getFailureCause()
Map<AccumulatorRegistry.Metric,Accumulator<?,?>>	getFlinkAccumulators()
ExecutionState	getState()
long	getStateTimestamp(ExecutionState state)
long[]	getStateTimestamps()
Map<String,Accumulator<?,?>>	getUserAccumulators()
StringifiedAccumulatorResult[]	getUserAccumulatorsStringified()
ExecutionVertex	getVertex()
String	getVertexWithAttempt()
boolean	isFinished()
void	prepareForArchiving() This method cleans fields that are irrelevant for the archived execution attempt.
boolean	scheduleForExecution(Scheduler scheduler, boolean queued) NOTE: This method only throws exceptions if it is in an illegal state to be scheduled, or if the tasks needs to be scheduled immediately and no resource is available.
void	setAccumulators(Map<AccumulatorRegistry.Metric,Accumulator<?,?>> flinkAccumulators, Map<String,Accumulator<?,?>> userAccumulators) Update accumulators (discarded when the Execution has already been terminated).
void	setInitialState(SerializedValue<StateHandle<?>> initialState, long recoveryTimestamp)
void	stop() Sends stop RPC call.
String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

Execution

public Execution(scala.concurrent.ExecutionContext executionContext,
                 ExecutionVertex vertex,
                 int attemptNumber,
                 long startTimestamp,
                 scala.concurrent.duration.FiniteDuration timeout)

Method Detail

getVertex

public ExecutionVertex getVertex()

getAttemptId

public ExecutionAttemptID getAttemptId()

getAttemptNumber

public int getAttemptNumber()

getState

public ExecutionState getState()

getAssignedResource

public SimpleSlot getAssignedResource()

getAssignedResourceLocation

public InstanceConnectionInfo getAssignedResourceLocation()

getFailureCause

public Throwable getFailureCause()

getStateTimestamps

public long[] getStateTimestamps()

getStateTimestamp

public long getStateTimestamp(ExecutionState state)

isFinished

public boolean isFinished()

prepareForArchiving

public void prepareForArchiving()

This method cleans fields that are irrelevant for the archived execution attempt.

setInitialState

public void setInitialState(SerializedValue<StateHandle<?>> initialState,
                            long recoveryTimestamp)

scheduleForExecution

public boolean scheduleForExecution(Scheduler scheduler,
                                    boolean queued)
                             throws NoResourceAvailableException

NOTE: This method only throws exceptions if it is in an illegal state to be scheduled, or if the tasks needs to be scheduled immediately and no resource is available. If the task is accepted by the schedule, any error sets the vertex state to failed and triggers the recovery logic.

Parameters:

scheduler - The scheduler to use to schedule this execution attempt.

queued - Flag to indicate whether the scheduler may queue this task if it cannot immediately deploy it.

Throws:

IllegalStateException - Thrown, if the vertex is not in CREATED state, which is the only state that permits scheduling.

NoResourceAvailableException - Thrown is no queued scheduling is allowed and no resources are currently available.

deployToSlot

public void deployToSlot(SimpleSlot slot)
                  throws JobException

Throws:

JobException

stop

public void stop()

Sends stop RPC call.

cancel

public void cancel()

fail

public void fail(Throwable t)

This method fails the vertex due to an external condition. The task will move to state FAILED. If the task was in state RUNNING or DEPLOYING before, it will send a cancel call to the TaskManager.

Parameters:

t - The exception that caused the task to fail.

getVertexWithAttempt

public String getVertexWithAttempt()

setAccumulators

public void setAccumulators(Map<AccumulatorRegistry.Metric,Accumulator<?,?>> flinkAccumulators,
                            Map<String,Accumulator<?,?>> userAccumulators)

Update accumulators (discarded when the Execution has already been terminated).

Parameters:

flinkAccumulators - the flink internal accumulators

userAccumulators - the user accumulators

getUserAccumulators

public Map<String,Accumulator<?,?>> getUserAccumulators()

getUserAccumulatorsStringified

public StringifiedAccumulatorResult[] getUserAccumulatorsStringified()

getFlinkAccumulators

public Map<AccumulatorRegistry.Metric,Accumulator<?,?>> getFlinkAccumulators()

toString

public String toString()

Overrides:

toString in class Object