org.apache.flink.runtime.taskexecutor

Class TaskExecutor

java.lang.Object

org.apache.flink.runtime.rpc.RpcEndpoint

org.apache.flink.runtime.taskexecutor.TaskExecutor

All Implemented Interfaces:

AutoCloseable, RpcGateway, TaskExecutorGateway, org.apache.flink.util.AutoCloseableAsync

public class TaskExecutor
extends RpcEndpoint
implements TaskExecutorGateway

TaskExecutor implementation. The task executor is responsible for the execution of multiple Task.

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.flink.runtime.rpc.RpcEndpoint

RpcEndpoint.MainThreadExecutor

Field Summary

Fields

Modifier and Type	Field and Description
static String	TASK_MANAGER_NAME

Fields inherited from class org.apache.flink.runtime.rpc.RpcEndpoint

log, rpcServer

Constructor Summary

Constructors

Constructor and Description

TaskExecutor(RpcService rpcService, TaskManagerConfiguration taskManagerConfiguration, HighAvailabilityServices haServices, TaskManagerServices taskExecutorServices, HeartbeatServices heartbeatServices, TaskManagerMetricGroup taskManagerMetricGroup, String metricQueryServiceAddress, BlobCacheService blobCacheService, FatalErrorHandler fatalErrorHandler, TaskExecutorPartitionTracker partitionTracker, BackPressureSampleService backPressureSampleService)

Method Summary

Modifier and Type	Method and Description
CompletableFuture<Boolean>	canBeReleased() Checks whether the task executor can be released.
CompletableFuture<Acknowledge>	cancelTask(ExecutionAttemptID executionAttemptID, org.apache.flink.api.common.time.Time timeout) Cancel the given task.
CompletableFuture<Acknowledge>	confirmCheckpoint(ExecutionAttemptID executionAttemptID, long checkpointId, long checkpointTimestamp) Confirm a checkpoint for the given task.
void	disconnectJobManager(org.apache.flink.api.common.JobID jobId, Exception cause) Disconnects the given JobManager from the TaskManager.
void	disconnectResourceManager(Exception cause) Disconnects the ResourceManager from the TaskManager.
CompletableFuture<Acknowledge>	freeSlot(AllocationID allocationId, Throwable cause, org.apache.flink.api.common.time.Time timeout) Frees the slot with the given allocation ID.
ResourceID	getResourceID()
void	heartbeatFromJobManager(ResourceID resourceID, AllocatedSlotReport allocatedSlotReport) Heartbeat request from the job manager.
void	heartbeatFromResourceManager(ResourceID resourceID) Heartbeat request from the resource manager.
void	onStart() User overridable callback which is called from RpcEndpoint.internalCallOnStart().
CompletableFuture<Void>	onStop() Called to shut down the TaskManager.
void	releaseOrPromotePartitions(org.apache.flink.api.common.JobID jobId, Set<ResultPartitionID> partitionToRelease, Set<ResultPartitionID> partitionsToPromote) Batch release/promote intermediate result partitions.
CompletableFuture<TransientBlobKey>	requestFileUpload(FileType fileType, org.apache.flink.api.common.time.Time timeout) Requests the file upload of the specified type to the cluster's BlobServer.
CompletableFuture<org.apache.flink.types.SerializableOptional<String>>	requestMetricQueryServiceAddress(org.apache.flink.api.common.time.Time timeout) Returns the gateway of Metric Query Service on the TaskManager.
CompletableFuture<Acknowledge>	requestSlot(SlotID slotId, org.apache.flink.api.common.JobID jobId, AllocationID allocationId, ResourceProfile resourceProfile, String targetAddress, ResourceManagerId resourceManagerId, org.apache.flink.api.common.time.Time timeout) Requests a slot from the TaskManager.
CompletableFuture<TaskBackPressureResponse>	requestTaskBackPressure(ExecutionAttemptID executionAttemptId, int requestId, org.apache.flink.api.common.time.Time timeout)
CompletableFuture<Acknowledge>	submitTask(TaskDeploymentDescriptor tdd, JobMasterId jobMasterId, org.apache.flink.api.common.time.Time timeout) Submit a Task to the TaskExecutor.
CompletableFuture<Acknowledge>	triggerCheckpoint(ExecutionAttemptID executionAttemptID, long checkpointId, long checkpointTimestamp, CheckpointOptions checkpointOptions, boolean advanceToEndOfEventTime) Trigger the checkpoint for the given task.
CompletableFuture<Acknowledge>	updatePartitions(ExecutionAttemptID executionAttemptID, Iterable<PartitionInfo> partitionInfos, org.apache.flink.api.common.time.Time timeout) Update the task where the given partitions can be found.

Methods inherited from class org.apache.flink.runtime.rpc.RpcEndpoint

callAsync, closeAsync, getAddress, getEndpointId, getHostname, getMainThreadExecutor, getRpcService, getSelfGateway, getTerminationFuture, internalCallOnStart, internalCallOnStop, isRunning, runAsync, scheduleRunAsync, scheduleRunAsync, start, stop, validateRunsInMainThread

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.flink.runtime.rpc.RpcGateway

getAddress, getHostname

Methods inherited from interface org.apache.flink.util.AutoCloseableAsync

close

Field Detail

TASK_MANAGER_NAME

public static final String TASK_MANAGER_NAME

See Also:

Constant Field Values

Constructor Detail

TaskExecutor

public TaskExecutor(RpcService rpcService,
                    TaskManagerConfiguration taskManagerConfiguration,
                    HighAvailabilityServices haServices,
                    TaskManagerServices taskExecutorServices,
                    HeartbeatServices heartbeatServices,
                    TaskManagerMetricGroup taskManagerMetricGroup,
                    String metricQueryServiceAddress,
                    BlobCacheService blobCacheService,
                    FatalErrorHandler fatalErrorHandler,
                    TaskExecutorPartitionTracker partitionTracker,
                    BackPressureSampleService backPressureSampleService)

Method Detail

canBeReleased

public CompletableFuture<Boolean> canBeReleased()

Description copied from interface: TaskExecutorGateway

Checks whether the task executor can be released. It cannot be released if there're unconsumed result partitions.

Specified by:

canBeReleased in interface TaskExecutorGateway

Returns:

Future flag indicating whether the task executor can be released.

onStart

public void onStart()
             throws Exception

Description copied from class: RpcEndpoint

User overridable callback which is called from RpcEndpoint.internalCallOnStart().

This method is called when the RpcEndpoint is being started. The method is guaranteed to be executed in the main thread context and can be used to start the rpc endpoint in the context of the rpc endpoint's main thread.

IMPORTANT: This method should never be called directly by the user.

Overrides:

onStart in class RpcEndpoint

Throws:

Exception - indicating that the rpc endpoint could not be started. If an exception occurs, then the rpc endpoint will automatically terminate.

onStop

public CompletableFuture<Void> onStop()

Called to shut down the TaskManager. The method closes all TaskManager services.

Overrides:

onStop in class RpcEndpoint

Returns:

Future which is completed once all post stop actions are completed. If an error occurs this future is completed exceptionally

requestTaskBackPressure

public CompletableFuture<TaskBackPressureResponse> requestTaskBackPressure(ExecutionAttemptID executionAttemptId,
                                                                           int requestId,
                                                                           org.apache.flink.api.common.time.Time timeout)

Specified by:

requestTaskBackPressure in interface TaskExecutorGateway

submitTask

public CompletableFuture<Acknowledge> submitTask(TaskDeploymentDescriptor tdd,
                                                 JobMasterId jobMasterId,
                                                 org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Submit a Task to the TaskExecutor.

Specified by:

submitTask in interface TaskExecutorGateway

Parameters:

tdd - describing the task to submit

jobMasterId - identifying the submitting JobMaster

timeout - of the submit operation

Returns:

Future acknowledge of the successful operation

cancelTask

public CompletableFuture<Acknowledge> cancelTask(ExecutionAttemptID executionAttemptID,
                                                 org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Cancel the given task.

Specified by:

cancelTask in interface TaskExecutorGateway

Parameters:

executionAttemptID - identifying the task

timeout - for the cancel operation

Returns:

Future acknowledge if the task is successfully canceled

updatePartitions

public CompletableFuture<Acknowledge> updatePartitions(ExecutionAttemptID executionAttemptID,
                                                       Iterable<PartitionInfo> partitionInfos,
                                                       org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Update the task where the given partitions can be found.

Specified by:

updatePartitions in interface TaskExecutorGateway

Parameters:

executionAttemptID - identifying the task

partitionInfos - telling where the partition can be retrieved from

timeout - for the update partitions operation

Returns:

Future acknowledge if the partitions have been successfully updated

releaseOrPromotePartitions

public void releaseOrPromotePartitions(org.apache.flink.api.common.JobID jobId,
                                       Set<ResultPartitionID> partitionToRelease,
                                       Set<ResultPartitionID> partitionsToPromote)

Description copied from interface: TaskExecutorGateway

Batch release/promote intermediate result partitions.

Specified by:

releaseOrPromotePartitions in interface TaskExecutorGateway

Parameters:

jobId - id of the job that the partitions belong to

partitionToRelease - partition ids to release

partitionsToPromote - partitions ids to promote

heartbeatFromJobManager

public void heartbeatFromJobManager(ResourceID resourceID,
                                    AllocatedSlotReport allocatedSlotReport)

Description copied from interface: TaskExecutorGateway

Heartbeat request from the job manager.

Specified by:

heartbeatFromJobManager in interface TaskExecutorGateway

Parameters:

resourceID - unique id of the job manager

heartbeatFromResourceManager

public void heartbeatFromResourceManager(ResourceID resourceID)

Description copied from interface: TaskExecutorGateway

Heartbeat request from the resource manager.

Specified by:

heartbeatFromResourceManager in interface TaskExecutorGateway

Parameters:

resourceID - unique id of the resource manager

triggerCheckpoint

public CompletableFuture<Acknowledge> triggerCheckpoint(ExecutionAttemptID executionAttemptID,
                                                        long checkpointId,
                                                        long checkpointTimestamp,
                                                        CheckpointOptions checkpointOptions,
                                                        boolean advanceToEndOfEventTime)

Description copied from interface: TaskExecutorGateway

Trigger the checkpoint for the given task. The checkpoint is identified by the checkpoint ID and the checkpoint timestamp.

Specified by:

triggerCheckpoint in interface TaskExecutorGateway

Parameters:

executionAttemptID - identifying the task

checkpointId - unique id for the checkpoint

checkpointTimestamp - is the timestamp when the checkpoint has been initiated

checkpointOptions - for performing the checkpoint

advanceToEndOfEventTime - Flag indicating if the source should inject a MAX_WATERMARK in the pipeline to fire any registered event-time timers

Returns:

Future acknowledge if the checkpoint has been successfully triggered

confirmCheckpoint

public CompletableFuture<Acknowledge> confirmCheckpoint(ExecutionAttemptID executionAttemptID,
                                                        long checkpointId,
                                                        long checkpointTimestamp)

Description copied from interface: TaskExecutorGateway

Confirm a checkpoint for the given task. The checkpoint is identified by the checkpoint ID and the checkpoint timestamp.

Specified by:

confirmCheckpoint in interface TaskExecutorGateway

Parameters:

executionAttemptID - identifying the task

checkpointId - unique id for the checkpoint

checkpointTimestamp - is the timestamp when the checkpoint has been initiated

Returns:

Future acknowledge if the checkpoint has been successfully confirmed

requestSlot

public CompletableFuture<Acknowledge> requestSlot(SlotID slotId,
                                                  org.apache.flink.api.common.JobID jobId,
                                                  AllocationID allocationId,
                                                  ResourceProfile resourceProfile,
                                                  String targetAddress,
                                                  ResourceManagerId resourceManagerId,
                                                  org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Requests a slot from the TaskManager.

Specified by:

requestSlot in interface TaskExecutorGateway

Parameters:

slotId - slot id for the request

jobId - for which to request a slot

allocationId - id for the request

resourceProfile - of requested slot, used only for dynamic slot allocation and will be ignored otherwise

targetAddress - to which to offer the requested slots

resourceManagerId - current leader id of the ResourceManager

timeout - for the operation

Returns:

answer to the slot request

freeSlot

public CompletableFuture<Acknowledge> freeSlot(AllocationID allocationId,
                                               Throwable cause,
                                               org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Frees the slot with the given allocation ID.

Specified by:

freeSlot in interface TaskExecutorGateway

Parameters:

allocationId - identifying the slot to free

cause - of the freeing operation

timeout - for the operation

Returns:

Future acknowledge which is returned once the slot has been freed

requestFileUpload

public CompletableFuture<TransientBlobKey> requestFileUpload(FileType fileType,
                                                             org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Requests the file upload of the specified type to the cluster's BlobServer.

Specified by:

requestFileUpload in interface TaskExecutorGateway

Parameters:

fileType - to upload

timeout - for the asynchronous operation

Returns:

Future which is completed with the TransientBlobKey of the uploaded file.

requestMetricQueryServiceAddress

public CompletableFuture<org.apache.flink.types.SerializableOptional<String>> requestMetricQueryServiceAddress(org.apache.flink.api.common.time.Time timeout)

Description copied from interface: TaskExecutorGateway

Returns the gateway of Metric Query Service on the TaskManager.

Specified by:

requestMetricQueryServiceAddress in interface TaskExecutorGateway

Returns:

Future gateway of Metric Query Service on the TaskManager.

disconnectJobManager

public void disconnectJobManager(org.apache.flink.api.common.JobID jobId,
                                 Exception cause)

Description copied from interface: TaskExecutorGateway

Disconnects the given JobManager from the TaskManager.

Specified by:

disconnectJobManager in interface TaskExecutorGateway

Parameters:

jobId - JobID for which the JobManager was the leader

cause - for the disconnection from the JobManager

disconnectResourceManager

public void disconnectResourceManager(Exception cause)

Description copied from interface: TaskExecutorGateway

Disconnects the ResourceManager from the TaskManager.

Specified by:

disconnectResourceManager in interface TaskExecutorGateway

Parameters:

cause - for the disconnection from the ResourceManager

getResourceID

public ResourceID getResourceID()