org.apache.beam.sdk.io.kafka

Class KafkaIO.ReadSourceDescriptors<K,V>

java.lang.Object

org.apache.beam.sdk.transforms.PTransform<PCollection<KafkaSourceDescriptor>,PCollection<KafkaRecord<K,V>>>

org.apache.beam.sdk.io.kafka.KafkaIO.ReadSourceDescriptors<K,V>

All Implemented Interfaces:

java.io.Serializable, HasDisplayData

Enclosing class:

KafkaIO

public abstract static class KafkaIO.ReadSourceDescriptors<K,V>
extends PTransform<PCollection<KafkaSourceDescriptor>,PCollection<KafkaRecord<K,V>>>

A PTransform to read from KafkaSourceDescriptor. See KafkaIO for more information on usage and configuration. See ReadFromKafkaDoFn for more implementation details.

During expansion, if isCommitOffsetEnabled() is true, the transform will expand to:

 PCollection<KafkaSourceDescriptor> --> ParDo(ReadFromKafkaDoFn<KafkaSourceDescriptor, KV<KafkaSourceDescriptor, KafkaRecord>>) --> Reshuffle() --> Map(output KafkaRecord)
                                                                                                                                         |
                                                                                                                                         --> KafkaCommitOffset
 

. Note that this expansion is not supported when running with x-lang on Dataflow.

See Also:

Serialized Form

Field Summary

Fields inherited from class org.apache.beam.sdk.transforms.PTransform

name, resourceHints

Constructor Summary

Constructors

Constructor and Description
ReadSourceDescriptors()

Method Summary

Modifier and Type	Method and Description
KafkaIO.ReadSourceDescriptors<K,V>	commitOffsets() Enable committing record offset.
PCollection<KafkaRecord<K,V>>	expand(PCollection<KafkaSourceDescriptor> input) Override this method to specify how this PTransform should be expanded on the given InputT.
static <K,V> KafkaIO.ReadSourceDescriptors<K,V>	read()
KafkaIO.ReadSourceDescriptors<K,V>	withBootstrapServers(java.lang.String bootstrapServers) Sets the bootstrap servers to use for the Kafka consumer if unspecified via KafkaSourceDescriptor#getBootStrapServers()}.
KafkaIO.ReadSourceDescriptors<K,V>	withCheckStopReadingFn(@Nullable SerializableFunction<TopicPartition,java.lang.Boolean> checkStopReadingFn) A custom SerializableFunction that determines whether the ReadFromKafkaDoFn should stop reading from the given TopicPartition.
KafkaIO.ReadSourceDescriptors<K,V>	withConsumerConfigOverrides(java.util.Map<java.lang.String,java.lang.Object> consumerConfig) Replaces the configuration for the main consumer.
KafkaIO.ReadSourceDescriptors<K,V>	withConsumerConfigUpdates(java.util.Map<java.lang.String,java.lang.Object> configUpdates) Updates configuration for the main consumer.
KafkaIO.ReadSourceDescriptors<K,V>	withConsumerFactoryFn(SerializableFunction<java.util.Map<java.lang.String,java.lang.Object>,Consumer<byte[],byte[]>> consumerFactoryFn) A factory to create Kafka Consumer from consumer configuration.
KafkaIO.ReadSourceDescriptors<K,V>	withCreateTime() Use the creation time of KafkaRecord as the output timestamp.
KafkaIO.ReadSourceDescriptors<K,V>	withCreatWatermarkEstimatorFn(SerializableFunction<Instant,WatermarkEstimator<Instant>> fn) A function to create a WatermarkEstimator.
KafkaIO.ReadSourceDescriptors<K,V>	withExtractOutputTimestampFn(SerializableFunction<KafkaRecord<K,V>,Instant> fn) A function to calculate output timestamp for a given KafkaRecord.
KafkaIO.ReadSourceDescriptors<K,V>	withKeyDeserializer(java.lang.Class<? extends Deserializer<K>> keyDeserializer) Sets a Kafka Deserializer to interpret key bytes read from Kafka.
KafkaIO.ReadSourceDescriptors<K,V>	withKeyDeserializerAndCoder(java.lang.Class<? extends Deserializer<K>> keyDeserializer, Coder<K> keyCoder) Sets a Kafka Deserializer for interpreting key bytes read from Kafka along with a Coder for helping the Beam runner materialize key objects at runtime if necessary.
KafkaIO.ReadSourceDescriptors<K,V>	withKeyDeserializerProvider(@Nullable DeserializerProvider<K> deserializerProvider)
KafkaIO.ReadSourceDescriptors<K,V>	withLogAppendTime() Use the log append time as the output timestamp.
KafkaIO.ReadSourceDescriptors<K,V>	withManualWatermarkEstimator() Use the WatermarkEstimators.Manual as the watermark estimator.
KafkaIO.ReadSourceDescriptors<K,V>	withMonotonicallyIncreasingWatermarkEstimator() Use the WatermarkEstimators.MonotonicallyIncreasing as the watermark estimator.
KafkaIO.ReadSourceDescriptors<K,V>	withOffsetConsumerConfigOverrides(@Nullable java.util.Map<java.lang.String,java.lang.Object> offsetConsumerConfig) Set additional configuration for the offset consumer.
KafkaIO.ReadSourceDescriptors<K,V>	withProcessingTime() Use the processing time as the output timestamp.
KafkaIO.ReadSourceDescriptors<K,V>	withReadCommitted() Sets "isolation_level" to "read_committed" in Kafka consumer configuration.
KafkaIO.ReadSourceDescriptors<K,V>	withValueDeserializer(java.lang.Class<? extends Deserializer<V>> valueDeserializer) Sets a Kafka Deserializer to interpret value bytes read from Kafka.
KafkaIO.ReadSourceDescriptors<K,V>	withValueDeserializerAndCoder(java.lang.Class<? extends Deserializer<V>> valueDeserializer, Coder<V> valueCoder) Sets a Kafka Deserializer for interpreting value bytes read from Kafka along with a Coder for helping the Beam runner materialize value objects at runtime if necessary.
KafkaIO.ReadSourceDescriptors<K,V>	withValueDeserializerProvider(@Nullable DeserializerProvider<V> deserializerProvider)
KafkaIO.ReadSourceDescriptors<K,V>	withWallTimeWatermarkEstimator() Use the WatermarkEstimators.WallTime as the watermark estimator.

Methods inherited from class org.apache.beam.sdk.transforms.PTransform

compose, compose, getAdditionalInputs, getDefaultOutputCoder, getDefaultOutputCoder, getDefaultOutputCoder, getKindString, getName, getResourceHints, populateDisplayData, setResourceHints, toString, validate, validate

Methods inherited from class java.lang.Object

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

Constructor Detail

ReadSourceDescriptors

public ReadSourceDescriptors()

Method Detail

read

public static <K,V> KafkaIO.ReadSourceDescriptors<K,V> read()

withBootstrapServers

public KafkaIO.ReadSourceDescriptors<K,V> withBootstrapServers(java.lang.String bootstrapServers)

Sets the bootstrap servers to use for the Kafka consumer if unspecified via KafkaSourceDescriptor#getBootStrapServers()}.

withKeyDeserializerProvider

public KafkaIO.ReadSourceDescriptors<K,V> withKeyDeserializerProvider(@Nullable DeserializerProvider<K> deserializerProvider)

withValueDeserializerProvider

public KafkaIO.ReadSourceDescriptors<K,V> withValueDeserializerProvider(@Nullable DeserializerProvider<V> deserializerProvider)

withKeyDeserializer

public KafkaIO.ReadSourceDescriptors<K,V> withKeyDeserializer(java.lang.Class<? extends Deserializer<K>> keyDeserializer)

Sets a Kafka Deserializer to interpret key bytes read from Kafka.

In addition, Beam also needs a Coder to serialize and deserialize key objects at runtime. KafkaIO tries to infer a coder for the key based on the Deserializer class, however in case that fails, you can use withKeyDeserializerAndCoder(Class, Coder) to provide the key coder explicitly.

withValueDeserializer

public KafkaIO.ReadSourceDescriptors<K,V> withValueDeserializer(java.lang.Class<? extends Deserializer<V>> valueDeserializer)

Sets a Kafka Deserializer to interpret value bytes read from Kafka.

In addition, Beam also needs a Coder to serialize and deserialize value objects at runtime. KafkaIO tries to infer a coder for the value based on the Deserializer class, however in case that fails, you can use withValueDeserializerAndCoder(Class, Coder) to provide the value coder explicitly.

withKeyDeserializerAndCoder

public KafkaIO.ReadSourceDescriptors<K,V> withKeyDeserializerAndCoder(java.lang.Class<? extends Deserializer<K>> keyDeserializer,
                                                                      Coder<K> keyCoder)

Sets a Kafka Deserializer for interpreting key bytes read from Kafka along with a Coder for helping the Beam runner materialize key objects at runtime if necessary.

Use this method to override the coder inference performed within withKeyDeserializer(Class).

withValueDeserializerAndCoder

public KafkaIO.ReadSourceDescriptors<K,V> withValueDeserializerAndCoder(java.lang.Class<? extends Deserializer<V>> valueDeserializer,
                                                                        Coder<V> valueCoder)

Sets a Kafka Deserializer for interpreting value bytes read from Kafka along with a Coder for helping the Beam runner materialize value objects at runtime if necessary.

Use this method to override the coder inference performed within withValueDeserializer(Class).

withConsumerFactoryFn

public KafkaIO.ReadSourceDescriptors<K,V> withConsumerFactoryFn(SerializableFunction<java.util.Map<java.lang.String,java.lang.Object>,Consumer<byte[],byte[]>> consumerFactoryFn)

A factory to create Kafka Consumer from consumer configuration. This is useful for supporting another version of Kafka consumer. Default is KafkaConsumer.

withCheckStopReadingFn

public KafkaIO.ReadSourceDescriptors<K,V> withCheckStopReadingFn(@Nullable SerializableFunction<TopicPartition,java.lang.Boolean> checkStopReadingFn)

A custom SerializableFunction that determines whether the ReadFromKafkaDoFn should stop reading from the given TopicPartition.

withConsumerConfigUpdates

public KafkaIO.ReadSourceDescriptors<K,V> withConsumerConfigUpdates(java.util.Map<java.lang.String,java.lang.Object> configUpdates)

Updates configuration for the main consumer. This method merges updates from the provided map with any prior updates using KafkaIOUtils.DEFAULT_CONSUMER_PROPERTIES as the starting configuration.

In ReadFromKafkaDoFn, there're two consumers running in the backend:

1. the main consumer which reads data from kafka.
2. the secondary offset consumer which is used to estimate the backlog by fetching the latest offset.

See withConsumerConfigOverrides(java.util.Map<java.lang.String, java.lang.Object>) for overriding the configuration instead of updating it.

See withOffsetConsumerConfigOverrides(java.util.Map<java.lang.String, java.lang.Object>) for configuring the secondary offset consumer.

withExtractOutputTimestampFn

public KafkaIO.ReadSourceDescriptors<K,V> withExtractOutputTimestampFn(SerializableFunction<KafkaRecord<K,V>,Instant> fn)

A function to calculate output timestamp for a given KafkaRecord. The default value is withProcessingTime().

withCreatWatermarkEstimatorFn

public KafkaIO.ReadSourceDescriptors<K,V> withCreatWatermarkEstimatorFn(SerializableFunction<Instant,WatermarkEstimator<Instant>> fn)

A function to create a WatermarkEstimator. The default value is WatermarkEstimators.MonotonicallyIncreasing.

withLogAppendTime

public KafkaIO.ReadSourceDescriptors<K,V> withLogAppendTime()

Use the log append time as the output timestamp.

withProcessingTime

public KafkaIO.ReadSourceDescriptors<K,V> withProcessingTime()

Use the processing time as the output timestamp.

withCreateTime

public KafkaIO.ReadSourceDescriptors<K,V> withCreateTime()

Use the creation time of KafkaRecord as the output timestamp.

withWallTimeWatermarkEstimator

public KafkaIO.ReadSourceDescriptors<K,V> withWallTimeWatermarkEstimator()

Use the WatermarkEstimators.WallTime as the watermark estimator.

withMonotonicallyIncreasingWatermarkEstimator

public KafkaIO.ReadSourceDescriptors<K,V> withMonotonicallyIncreasingWatermarkEstimator()

Use the WatermarkEstimators.MonotonicallyIncreasing as the watermark estimator.

withManualWatermarkEstimator

public KafkaIO.ReadSourceDescriptors<K,V> withManualWatermarkEstimator()

Use the WatermarkEstimators.Manual as the watermark estimator.

withReadCommitted

public KafkaIO.ReadSourceDescriptors<K,V> withReadCommitted()

Sets "isolation_level" to "read_committed" in Kafka consumer configuration. This ensures that the consumer does not read uncommitted messages. Kafka version 0.11 introduced transactional writes. Applications requiring end-to-end exactly-once semantics should only read committed messages. See JavaDoc for KafkaConsumer for more description.

commitOffsets

public KafkaIO.ReadSourceDescriptors<K,V> commitOffsets()

Enable committing record offset. If ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG is set together with commitOffsets(), commitOffsets() will be ignored.

withOffsetConsumerConfigOverrides

public KafkaIO.ReadSourceDescriptors<K,V> withOffsetConsumerConfigOverrides(@Nullable java.util.Map<java.lang.String,java.lang.Object> offsetConsumerConfig)

Set additional configuration for the offset consumer. It may be required for a secured Kafka cluster, especially when you see similar WARN log message exception while fetching latest offset for partition {}. will be retried.

In ReadFromKafkaDoFn, there are two consumers running in the backend:

1. the main consumer which reads data from kafka.
2. the secondary offset consumer which is used to estimate the backlog by fetching the latest offset.

By default, offset consumer inherits the configuration from main consumer, with an auto-generated ConsumerConfig.GROUP_ID_CONFIG. This may not work in a secured Kafka which requires additional configuration.

See withConsumerConfigUpdates(java.util.Map<java.lang.String, java.lang.Object>) for configuring the main consumer.

withConsumerConfigOverrides

public KafkaIO.ReadSourceDescriptors<K,V> withConsumerConfigOverrides(java.util.Map<java.lang.String,java.lang.Object> consumerConfig)

Replaces the configuration for the main consumer.

In ReadFromKafkaDoFn, there are two consumers running in the backend:

1. the main consumer which reads data from kafka.
2. the secondary offset consumer which is used to estimate the backlog by fetching the latest offset.

By default, main consumer uses the configuration from KafkaIOUtils.DEFAULT_CONSUMER_PROPERTIES.

See withConsumerConfigUpdates(java.util.Map<java.lang.String, java.lang.Object>) for updating the configuration instead of overriding it.

expand

public PCollection<KafkaRecord<K,V>> expand(PCollection<KafkaSourceDescriptor> input)

Description copied from class: PTransform

Override this method to specify how this PTransform should be expanded on the given InputT.

NOTE: This method should not be called directly. Instead apply the PTransform should be applied to the InputT using the apply method.

Composite transforms, which are defined in terms of other transforms, should return the output of one of the composed transforms. Non-composite transforms, which do not apply any transforms internally, should return a new unbound output and register evaluators (via backend-specific registration methods).

Specified by:

expand in class PTransform<PCollection<KafkaSourceDescriptor>,PCollection<KafkaRecord<K,V>>>