This page is meant as a template for writing a KIP. To create a KIP choose Tools->Copy on this page and modify with your content and replace the heading with the next KIP number and a description of your issue. Replace anything in italics with your own description.

Status

Current state:   [One of "Under Discussion", "Accepted", "Rejected"]

Discussion thread: here

JIRA: here 

Released: <Kafka Version>1.2

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

Motivation

Describe the problems you are trying to solve.

Public Interfaces

Briefly list any new interfaces that will be introduced as part of this proposal or any existing interfaces that will be removed or changed. The purpose of this section is to concisely call out the public contract that will come along with this feature.

A public interface is any change to the following:

• Binary log format

• The network protocol and api behavior

• Any class in the public packages under clientsConfiguration, especially client configuration

  • org/apache/kafka/common/serialization

  • org/apache/kafka/common

  • org/apache/kafka/common/errors

  • org/apache/kafka/clients/producer

  • org/apache/kafka/clients/consumer (eventually, once stable)

• Monitoring

• Command line tools and arguments

• Anything else that will likely break existing users in some way when they upgrade

Proposed Changes

Kafka Streams has a defined "contract" about timestamp propagation at the Processor API level: all processors within a sub-topology, see the timestamp from the input topic record that is currently processed and this timestamp will be used for all result records when writing them to a topic, too. For the DSL and also for custom operators, it would be desirable to allow timestamp manipulation at Processor API level for individual records that are forwarded. This allows to support a larger scope of possible semantics for stream processing.

Public Interfaces

Using Processor API, users use ProcessorContext to forward record to downstream operators by calling forward(). Thus, we need to add overloads for forward() to allow user to pass in a timestamp for the output record.

package org.apache.kafka.streams.processor;


public interface ProcessorContext {
  // existing overloads of forward()
  <K, V> void forward(K key, V value);
  <K, V> void forward(K key, V value, int childIndex);
  <K, V> void forward(K key, V value, String childName);


  // new overloads of forward()
  <K, V> void forward(K key, V value, long timestamp);
  <K, V> void forward(K key, V value, long timestamp, int childIndex);
  <K, V> void forward(K key, V value, long timestamp, String childName);

  // other existing methods omitted for brevity
}

Proposed Changes

We add three new overloads of ProcessorContext#forward() that take an additional timestamp parameter of type long. If users call those new overloads, the output record gets the specified timestamp assigned. For the existing methods, the default contract using the input record timestamp for the output record will be usedDescribe the new thing you want to do in appropriate detail. This may be fairly extensive and have large subsections of its own. Or it may be a few sentences. Use judgement based on the scope of the change.

Compatibility, Deprecation, and Migration Plan

• What impact (if any) will there be on existing users?
• If we are changing behavior how will we phase out the older behavior?
• If we need special migration tools, describe them here.
• When will we remove the existing behavior?

Test Plan

Describe in few sentences how the KIP will be tested. We are mostly interested in system tests (since unit-tests are specific to implementation details). How will we know that the implementation works as expected? How will we know nothing broke?

Rejected Alternatives

This change is backward compatible, as we don't alter the contract of existing methods, and only add new method with new functionality.

Test Plan

We can test this feature with unit tests, by implementing test Processors that manipulate the output record timestamp using the new API, and by verifying that the output record have the assigned timestamps. Existing test ensure, that the old behavior is preserved.

Rejected Alternatives

NoneIf there are alternative ways of accomplishing the same thing, what were they? The purpose of this section is to motivate why the design is the way it is and not some other way.