You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 16 Next »

Status

Current state: Voting in progress

Discussion thread: here

JIRA: here

Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).

Motivation

KIP-478, a strongly-typed Processor API, brings the option to reconsider the abstractions for custom processing in the Kafka Streams DSL.

Currently, multiple Transformers and a final Processor operation are the options available to implement custom processing with access to Record context; including record metadata (incl. topic, partition, offset), timestamp, and headers; and reference to state stores.

There have been discussions on how to refactor these APIs to minimize/simplify into a single processor:

Transformers are currently limited to the old ProcessorContext. With the addition of the new Processor API, KStream can be extended to offer access to the new, chainable API that can replace most of the operations transformers offer at the moment, with more open control to forward records.

This KIP is aimed to extend the current `KStream#process` API to return output values that could be chained across the topology, as well as introduce a new KStream#processValues  to use processor while validating keys haven't changed and repartition is not required.

This KIP could be considered as a step towards deprecating Transforms, though that should be discussed in a follow-up KIP.

Public Interfaces

New methods

Processors without forcing repartitioning:

  • KStream<K,VOut> KStream#processValues(ProcessorSupplier<K, V, K, VOut> processorSupplier, String... stateStoreNames)
  • KStream<K,VOut> KStream#processValues(ProcessorSupplier<K, V, K, VOut> processorSupplier, Named named, String... stateStoreNames)

Modified methods


  • KStream<KOut,VOut> KStream#process(ProcessorSupplier<K, V, KOut, VOut> processorSupplier, String... stateStoreNames)
    • from void KStream#process(ProcessorSupplier<K, V, Void, Void> processorSupplier, ...)
  • KStream<KOut,VOut> KStream#process(ProcessorSupplier<K, V, KOut, VOut> processorSupplier, Named named, String... stateStoreNames)
    • from void KStream#process(ProcessorSupplier<K, V, Void, Void> processorSupplier, ...)

Deprecated methods

  • KStream#transform 
  • KStream#transformValues 
  • KStream#flatTransform 
  • KStream#flatTransformValues 

Proposed Changes

New Processor APIs replacing most Transformers

With the ability to manage forward calls as part of the Processor itself, flatTransform can be replaced by a Processor:

  private Topology topology() {
    final var builder = new StreamsBuilder();
    final var input = builder.stream("words", Consumed.with(Serdes.String(), Serdes.String()));

    input.processValues(() -> new ContextualProcessor<String, String, String, String>() {
      @Override
      public void process(Record<String, String> record) {
        for (var s : record.value().split(",")) {
          context().forward(record.withValue("Hello " + s));
        }
      }
    }, Named.as("test")) // doesn't flag repartitioning 

    input.process(() -> new ContextualProcessor<String, String, Integer, String>() {
      @Override
      public void process(Record<String, String> record) {
        for (var s : record.value().split(",")) {
          context().forward(record.withKey(record.value().lenght()).withValue("Hello " + s));
        }
      }
    }, Named.as("test")) // flags repartitioning

    .to("output", Produced.with(Serdes.String(), Serdes.String()));
    return builder.build();
  }


New ValueProcessorContext to enforce only Value changes

Processors that require enforcing the same key to avoid repartitioning at the moment trust internal implementations. As a result, the KStream API added *Values methods to signal when the key has not changed, and repartitioning can be avoided.

At the moment record keys can be changed, or record recreated, between the record entering Processor#process(Record) and ProcessorContext#forward(Record).

As an alternative to enforcing the same key, ProcessorContext wrapper could be implemented to define an initial key:


class ValueProcessorContext<KForward, VForward> implements ProcessorContext<KForward, VForward> {
//...
    private KForward key;
    
    public void setRecordKey(KForward initialKey) {
        this.key = initialKey;
    }
    
    public void clearRecordKey() {
        this.key = null;
    }

    @Override
    public <K extends KForward, V extends VForward> void forward(Record<K, V> record) {
        if (key != null) {
            if (record.key() != key) {  // referencial equility is enough to enforce that key is not updated
                 throw new IllegalArgumentException("Key has changed while processing the record and requires enforcing repartitioning.");
             }
        }
        delegate.forward(record.withKey(key));
     }
//...
}

This ValueProcessorContext is to be used internally at the Processor level, and the method setRecordKey(K) must be used before calling user-defined Processor, and the clearRecordKey()  right after.

NOTE: This API is not meant to be a public interface. Only used internally by the processValues DSL operation to enforce at runtime that the key hasn't changed.


class KStreamValueProcessorSupplier<KIn, VIn, VOut> implements ProcessorSupplier<KIn, VIn, KIn, VOut> {
// get ...

    static class KStreamValueProcessor<KIn, VIn, VOut> extends ContextualProcessor<KIn, VIn, KIn, VOut> {

        private final Processor<KIn, VIn, KIn, VOut> processor;

        private ValueProcessorContext<KIn, VOut> processorContext;

        public KStreamValueProcessor(Processor<KIn, VIn, KIn, VOut> transformer) {
            this.processor = transformer;
        }

        @Override
        public void init(final ProcessorContext<KIn, VOut> context) {
            super.init(context);
            this.processorContext = new ValueProcessorContext<>(context);
            processor.init(processorContext);
        }

        @Override
        public void process(final Record<KIn, VIn> record) {
            processorContext.setRecordKey(record.key());
            processor.process(record);
            processorContext.clearRecordKey();
        }
    }

}

Based on the previous example, if the key changes a runtime exception will be thrown:

  private Topology topology() {
    final var builder = new StreamsBuilder();
    final var input = builder.stream("words", Consumed.with(Serdes.String(), Serdes.String()));
    input.processValues(() -> new ContextualProcessor<String, String, String, String>() {
      @Override
      public void process(Record<String, String> record) {
        for (var s : record.value().split(",")) {
          context().forward(record.withKey("FAIL!").withValue("Hello " + s)); // will throw a Runtime Exception because key has changed on `processValues`
        }
      }
    }, Named.as("test"))
    .to("output", Produced.with(Serdes.String(), Serdes.String()));
    return builder.build();
  }


NOTE: The key validation can be defined when processing the message. Though, with punctuations, it won't be possible to define the key for validation before forwarding, therefore it won't be possible to forward from punctuation. This is similar behavior to how ValueTransformers behave at the moment.

Compatibility, Deprecation, and Migration Plan

Modified operation KStream#process 

Modifications to method KStream#process are source compatible with the previous version, though not binary compatible. Therefore will require users to recompile their applications with the latest version. Though users will not require to change their code as the current returned value is void, and the input processor supplier types will include Void  that is the current type for output key and value.

Transform  methods deprecation

Transform APIs won't be deprecated as part of this API, but transform* operations are. This and Transform API use of old ProcessorContext should warn users and drive adoption of the methods proposed in this KIP.

Even though, the new Processor API could potentially replace most Transformer implementations, it is not yet possible to deprecate Transformer APIs:

  • Transform API is broadly adopted for custom processing. Even though most functionality is possible to implement with new Processor API, this migration is not straightforward and will require reimplementing custom processors.
  • KTable#transformValues is an interesting use-case for ValueTransformerWithKey as the internal value is Change<V> — an internal type to handle record changes on KTable — not exposed as public API. An approach would be to deprecate this method in favor of .toStream().processValues().toTable().

A new KIP should be proposed to continue the deprecation of Transformer  APIs.

Rejected Alternatives

Create new Transform APIs with the latest ProcessContext

This would involve extending KStream/KTable APIs even more with specific Transform variants. If a new Processor can support most Transform in the long-term becoming the adopted option to transform values, flat transform, etc.; then we should start getting the new Processor properly adopted in the DSL and let the usage drive the next steps.

Runtime key validation on KStream#processValues vs. key output Void type

On one hand, with Void type for key output, we force the users to cast to Void and change the key to null, though this can be documented on the API, so the users are aware of the peculiarity of forwarding within processValues. 

On the other hand, keeping the key type as output doesn't require to do any change of keys, but this could lead to key-checking runtime exceptions when changing keys.

This KIP proposes the second option and does runtime key validation to avoid allocation of two Record with withKey() (one for setting to Void null, and another for setting the previous key). Users will need to be aware that using processValues  will lead to runtime exceptions when changing the keys when not supposed to.

References

  • No labels