org.apache.flink.runtime.io.network.partition

Class SortMergeResultPartition

java.lang.Object

org.apache.flink.runtime.io.network.partition.ResultPartition

org.apache.flink.runtime.io.network.partition.SortMergeResultPartition

All Implemented Interfaces:

AutoCloseable, AvailabilityProvider, ResultPartitionWriter

@NotThreadSafe
public class SortMergeResultPartition
extends ResultPartition

SortMergeResultPartition appends records and events to DataBuffer and after the DataBuffer is full, all data in the DataBuffer will be copied and spilled to a PartitionedFile in subpartition index order sequentially. Large records that can not be appended to an empty DataBuffer will be spilled to the result PartitionedFile separately.

Nested Class Summary

Nested classes/interfaces inherited from interface org.apache.flink.runtime.io.AvailabilityProvider

AvailabilityProvider.AvailabilityHelper

Field Summary

Fields inherited from class org.apache.flink.runtime.io.network.partition.ResultPartition

bufferCompressor, bufferPool, LOG, numBuffersOut, numBytesOut, numBytesProduced, numSubpartitions, partitionId, partitionManager, partitionType

Fields inherited from interface org.apache.flink.runtime.io.AvailabilityProvider

AVAILABLE

Constructor Summary

Constructors

Constructor and Description

SortMergeResultPartition(String owningTaskName, int partitionIndex, ResultPartitionID partitionId, ResultPartitionType partitionType, int numSubpartitions, int numTargetKeyGroups, BatchShuffleReadBufferPool readBufferPool, Executor readIOExecutor, ResultPartitionManager partitionManager, String resultFileBasePath, BufferCompressor bufferCompressor, org.apache.flink.util.function.SupplierWithException<BufferPool,IOException> bufferPoolFactory)

Method Summary

Modifier and Type	Method and Description
void	broadcastEvent(AbstractEvent event, boolean isPriorityEvent) Writes the given AbstractEvent to all channels.
void	broadcastRecord(ByteBuffer record) Writes the given serialized record to all subpartitions.
void	close() Closes the partition writer which releases the allocated resource, for example the buffer pool.
ResultSubpartitionView	createSubpartitionView(int subpartitionIndex, BufferAvailabilityListener availabilityListener) Returns a reader for the subpartition with the given index.
void	emitRecord(ByteBuffer record, int targetSubpartition) Writes the given serialized record to the target subpartition.
void	finish() Finishes the result partition.
void	flush(int subpartitionIndex) Manually trigger the consumption of data from the given subpartitions.
void	flushAll() Manually trigger the consumption of data from all subpartitions.
CompletableFuture<?>	getAvailableFuture()
int	getNumberOfQueuedBuffers() Returns the total number of queued buffers of all subpartitions.
int	getNumberOfQueuedBuffers(int targetSubpartition) Returns the number of queued buffers of the given target subpartition.
long	getSizeOfQueuedBuffersUnsafe() Returns the total size in bytes of queued buffers of all subpartitions.
void	notifyEndOfData(StopMode mode) Notifies the downstream tasks that this ResultPartitionWriter have emitted all the user records.
protected void	releaseInternal() Releases all produced data including both those stored in memory and persisted on disk.
void	setup() Registers a buffer pool with this result partition.

Methods inherited from class org.apache.flink.runtime.io.network.partition.ResultPartition

canBeCompressed, checkInProduceState, fail, getAllDataProcessedFuture, getBufferPool, getFailureCause, getNumberOfSubpartitions, getNumTargetKeyGroups, getOwningTaskName, getPartitionId, getPartitionIndex, getPartitionManager, getPartitionType, isFinished, isReleased, onSubpartitionAllDataProcessed, release, release, setMetricGroup, toString

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.flink.runtime.io.AvailabilityProvider

and, isApproximatelyAvailable, isAvailable, or

Constructor Detail

SortMergeResultPartition

public SortMergeResultPartition(String owningTaskName,
                                int partitionIndex,
                                ResultPartitionID partitionId,
                                ResultPartitionType partitionType,
                                int numSubpartitions,
                                int numTargetKeyGroups,
                                BatchShuffleReadBufferPool readBufferPool,
                                Executor readIOExecutor,
                                ResultPartitionManager partitionManager,
                                String resultFileBasePath,
                                @Nullable
                                BufferCompressor bufferCompressor,
                                org.apache.flink.util.function.SupplierWithException<BufferPool,IOException> bufferPoolFactory)

Method Detail

setup

public void setup()
           throws IOException

Description copied from class: ResultPartition

Registers a buffer pool with this result partition.

There is one pool for each result partition, which is shared by all its sub partitions.

The pool is registered with the partition *after* it as been constructed in order to conform to the life-cycle of task registrations in the TaskExecutor.

Specified by:

setup in interface ResultPartitionWriter

Overrides:

setup in class ResultPartition

Throws:

IOException

releaseInternal

protected void releaseInternal()

Description copied from class: ResultPartition

Releases all produced data including both those stored in memory and persisted on disk.

Specified by:

releaseInternal in class ResultPartition

emitRecord

public void emitRecord(ByteBuffer record,
                       int targetSubpartition)
                throws IOException

Description copied from interface: ResultPartitionWriter

Writes the given serialized record to the target subpartition.

Throws:

IOException

broadcastRecord

public void broadcastRecord(ByteBuffer record)
                     throws IOException

Description copied from interface: ResultPartitionWriter

Writes the given serialized record to all subpartitions. One can also achieve the same effect by emitting the same record to all subpartitions one by one, however, this method can have better performance for the underlying implementation can do some optimizations, for example coping the given serialized record only once to a shared channel which can be consumed by all subpartitions.

Throws:

IOException

broadcastEvent

public void broadcastEvent(AbstractEvent event,
                           boolean isPriorityEvent)
                    throws IOException

Description copied from interface: ResultPartitionWriter

Writes the given AbstractEvent to all channels.

Throws:

IOException

notifyEndOfData

public void notifyEndOfData(StopMode mode)
                     throws IOException

Description copied from interface: ResultPartitionWriter

Notifies the downstream tasks that this ResultPartitionWriter have emitted all the user records.

Specified by:

notifyEndOfData in interface ResultPartitionWriter

Overrides:

notifyEndOfData in class ResultPartition

Parameters:

mode - tells if we should flush all records or not (it is false in case of stop-with-savepoint (--no-drain))

Throws:

IOException

finish

public void finish()
            throws IOException

Description copied from class: ResultPartition

Finishes the result partition.

After this operation, it is not possible to add further data to the result partition.

For BLOCKING results, this will trigger the deployment of consuming tasks.

Specified by:

finish in interface ResultPartitionWriter

Overrides:

finish in class ResultPartition

Throws:

IOException

close

public void close()

Description copied from interface: ResultPartitionWriter

Closes the partition writer which releases the allocated resource, for example the buffer pool.

Specified by:

close in interface AutoCloseable

Specified by:

close in interface ResultPartitionWriter

Overrides:

close in class ResultPartition

createSubpartitionView

public ResultSubpartitionView createSubpartitionView(int subpartitionIndex,
                                                     BufferAvailabilityListener availabilityListener)
                                              throws IOException

Description copied from interface: ResultPartitionWriter

Returns a reader for the subpartition with the given index.

Throws:

IOException

flushAll

public void flushAll()

Description copied from interface: ResultPartitionWriter

Manually trigger the consumption of data from all subpartitions.

flush

public void flush(int subpartitionIndex)

Description copied from interface: ResultPartitionWriter

Manually trigger the consumption of data from the given subpartitions.

getAvailableFuture

public CompletableFuture<?> getAvailableFuture()

Specified by:

getAvailableFuture in interface AvailabilityProvider

Overrides:

getAvailableFuture in class ResultPartition

Returns:

a future that is completed if the respective provider is available.

getNumberOfQueuedBuffers

public int getNumberOfQueuedBuffers()

Description copied from class: ResultPartition

Returns the total number of queued buffers of all subpartitions.

Specified by:

getNumberOfQueuedBuffers in class ResultPartition

getSizeOfQueuedBuffersUnsafe

public long getSizeOfQueuedBuffersUnsafe()

Description copied from class: ResultPartition

Returns the total size in bytes of queued buffers of all subpartitions.

Specified by:

getSizeOfQueuedBuffersUnsafe in class ResultPartition

getNumberOfQueuedBuffers

public int getNumberOfQueuedBuffers(int targetSubpartition)

Description copied from class: ResultPartition

Returns the number of queued buffers of the given target subpartition.

Specified by:

getNumberOfQueuedBuffers in class ResultPartition