org.apache.flink.streaming.api.functions.sink.filesystem

Class Buckets<IN,BucketID>

java.lang.Object

org.apache.flink.streaming.api.functions.sink.filesystem.Buckets<IN,BucketID>

Type Parameters:

IN - The type of input elements.

BucketID - The type of ids for the buckets, as returned by the BucketAssigner.

@Internal
public class Buckets<IN,BucketID>
extends Object

The manager of the different active buckets in the StreamingFileSink.

This class is responsible for all bucket-related operations and the actual StreamingFileSink is just plugging in the functionality offered by this class to the lifecycle of the operator.

Constructor Summary

Constructors

Constructor and Description

Buckets(org.apache.flink.core.fs.Path basePath, org.apache.flink.streaming.api.functions.sink.filesystem.BucketAssigner<IN,BucketID> bucketAssigner, BucketFactory<IN,BucketID> bucketFactory, org.apache.flink.streaming.api.functions.sink.filesystem.BucketWriter<IN,BucketID> bucketWriter, org.apache.flink.streaming.api.functions.sink.filesystem.RollingPolicy<IN,BucketID> rollingPolicy, int subtaskIndex, org.apache.flink.streaming.api.functions.sink.filesystem.OutputFileConfig outputFileConfig) A constructor creating a new empty bucket manager.

Method Summary

Modifier and Type	Method and Description
void	close()
void	commitUpToCheckpoint(long checkpointId)
long	getMaxPartCounter()
void	initializeState(org.apache.flink.api.common.state.ListState<byte[]> bucketStates, org.apache.flink.api.common.state.ListState<Long> partCounterState) Initializes the state after recovery from a failure.
Bucket<IN,BucketID>	onElement(IN value, long currentProcessingTime, Long elementTimestamp, long currentWatermark)
Bucket<IN,BucketID>	onElement(IN value, SinkFunction.Context context)
void	onProcessingTime(long timestamp)
void	setBucketLifeCycleListener(BucketLifeCycleListener<IN,BucketID> bucketLifeCycleListener)
void	setFileLifeCycleListener(FileLifeCycleListener<BucketID> fileLifeCycleListener)
void	snapshotState(long checkpointId, org.apache.flink.api.common.state.ListState<byte[]> bucketStatesContainer, org.apache.flink.api.common.state.ListState<Long> partCounterStateContainer)

Methods inherited from class java.lang.Object

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

Constructor Detail

Buckets

public Buckets(org.apache.flink.core.fs.Path basePath,
               org.apache.flink.streaming.api.functions.sink.filesystem.BucketAssigner<IN,BucketID> bucketAssigner,
               BucketFactory<IN,BucketID> bucketFactory,
               org.apache.flink.streaming.api.functions.sink.filesystem.BucketWriter<IN,BucketID> bucketWriter,
               org.apache.flink.streaming.api.functions.sink.filesystem.RollingPolicy<IN,BucketID> rollingPolicy,
               int subtaskIndex,
               org.apache.flink.streaming.api.functions.sink.filesystem.OutputFileConfig outputFileConfig)

A constructor creating a new empty bucket manager.

Parameters:

basePath - The base path for our buckets.

bucketAssigner - The BucketAssigner provided by the user.

bucketFactory - The BucketFactory to be used to create buckets.

bucketWriter - The BucketWriter to be used when writing data.

rollingPolicy - The RollingPolicy as specified by the user.

Method Detail

setBucketLifeCycleListener

public void setBucketLifeCycleListener(BucketLifeCycleListener<IN,BucketID> bucketLifeCycleListener)

setFileLifeCycleListener

public void setFileLifeCycleListener(FileLifeCycleListener<BucketID> fileLifeCycleListener)

initializeState

public void initializeState(org.apache.flink.api.common.state.ListState<byte[]> bucketStates,
                            org.apache.flink.api.common.state.ListState<Long> partCounterState)
                     throws Exception

Initializes the state after recovery from a failure.

During this process:

1. we set the initial value for part counter to the maximum value used before across all tasks and buckets. This guarantees that we do not overwrite valid data,
2. we commit any pending files for previous checkpoints (previous to the last successful one from which we restore),
3. we resume writing to the previous in-progress file of each bucket, and
4. if we receive multiple states for the same bucket, we merge them.

Parameters:

bucketStates - the state holding recovered state about active buckets.

partCounterState - the state holding the max previously used part counters.

Throws:

Exception - if anything goes wrong during retrieving the state or restoring/committing of any in-progress/pending part files

commitUpToCheckpoint

public void commitUpToCheckpoint(long checkpointId)
                          throws IOException

Throws:

IOException

snapshotState

public void snapshotState(long checkpointId,
                          org.apache.flink.api.common.state.ListState<byte[]> bucketStatesContainer,
                          org.apache.flink.api.common.state.ListState<Long> partCounterStateContainer)
                   throws Exception

Throws:

Exception

onElement

@VisibleForTesting
public Bucket<IN,BucketID> onElement(IN value,
                                                        SinkFunction.Context context)
                                                 throws Exception

Throws:

Exception

onElement

public Bucket<IN,BucketID> onElement(IN value,
                                     long currentProcessingTime,
                                     @Nullable
                                     Long elementTimestamp,
                                     long currentWatermark)
                              throws Exception

Throws:

Exception

onProcessingTime

public void onProcessingTime(long timestamp)
                      throws Exception

Throws:

Exception

close

public void close()

getMaxPartCounter

@VisibleForTesting
public long getMaxPartCounter()