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: "Under Discussion"

Discussion thread: : https://lists.apache.org/thread/0f20kvfqkmhdqrwcb8vqgqn80szcrcdd here [Change the link from the KIP proposal email archive to your own email thread]

JIRA: KAFKA-15187 

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.

Current partition method only accepts key and value. Wanted to provide support for selecting partition based on headers too. This lets users pass values to partitioner method without tampering the key and value. 

Sample use case : 

...

Few of the use cases where this can be helpful :

1. The user has a header which provides priority of the message [lets say 0 to 9]. You have 10 partitions which you want to use for each priority. With headers available in partition method the partition can be

...

1. chosen based on the header value. 

   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
   1. 1. Currently, there is an implementation which achieve the functionality by appending the header value to the key and stripping it off in the keySerializer when creating the keyBytes. This is a hacky solution.
   2. The user wants to have a decorator partitioner which they want to use to select from a different set of partitioners based some params but do not want to tamper with the key and value. The user can pass the params in headers, evaluate it in the headers and call partitioner of interest.

   Public Interfaces

   Partitioner interface in Partitioner method in org/apache/kafka/clients/producer will have the following update :

   • A new overloaded method partition(String topic, Object key, byte[] keyBytes, Object value, byte[] valueBytes, Cluster cluster, Headers headers) with default implementation to call the existing method.
   • The object passed as headers is readonly and cannot be updated.

   • Details :

   Proposed Changes

   ...

   • Code Block

     /** * Compute the partition for the given record. Not overriding this method in the Partitioner interface has the same behaviour as using the existing method. The object passed as headers is readonly and cannot be updated. * * @param topic The topic name * @param key The key to partition on (or null if no key) * @param keyBytes The serialized key to partition on( or null if no key) * @param value The value to partition on or null * @param valueBytes The serialized value to partition on or null * @param cluster The current cluster metadata * @param headers The headers to partition on or null */ default int partition(String topic, Object key, byte[] keyBytes, Object value, byte[] valueBytes, Cluster cluster, Headers headers) { return partition(topic, key, keyBytes, value, valueBytes, cluster); };

     
     

   Proposed Changes

   • Add a new overloaded method partition(String topic, Object key, byte[] keyBytes, Object value, byte[] valueBytes, Cluster cluster, Headers headers) with default implementation to call the existing method in the Partitioner

   ...

   • interface in org/apache/kafka/clients/producer.
     • This lets the users to implement the partitioner with ability to choose partition based on header values.
     • If a partitioner implements both the methods, the override of new method (one which accepts headers) will be in effect.
   • Make the call to partition method in KafkaProducer to use the new method.
     • Since it is defaulted to use the existing method, this should not cause any behaviour change.

   Total changes : https://github.com/apache/kafka/pull/13981

   Compatibility, Deprecation, and Migration Plan

   • What impact (if any) will there be on existing users?
     • NoneAn implementation of the new interface will be incompatible with old clients.
   • If we are changing behavior how will we phase out the older behavior?
     • NA
   • If we need special migration tools, describe them here.
     • NA
   • When will we remove the existing behavior?
     • NA

   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

   If 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.

   ...

   This is a backward compatible change as mentioned in the proposed changes section. The new default method calls the existing method unless it is manually overridden by the user. 

   Test Plan

   This has an addition of a default method in the interface. The changes made in Kafka Producer should be covered in the existing tests and should not have any impact.

   Rejected Alternatives

   None