[
https://issues.apache.org/jira/browse/KAFKA-3333?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15488853#comment-15488853
]
Jason Gustafson commented on KAFKA-3333:
----------------------------------------
Admittedly, the line may not be totally clear, but we have asked for KIPs for
similar components on the consumer (see KIP-54 for example). In general, it's a
good way to publicize the feature in order to gauge community interest and
collect feedback. We've hesitated in the past to include additional
partitioners because they can always be made available to users through a
separate project. In this case, there might be some concern over the fact that
a key can be sent to multiple partitions which breaks the usual binding of key
and partition, but perhaps it makes sense for some use cases.
> Client Partitioner - Round Robin
> --------------------------------
>
> Key: KAFKA-3333
> URL: https://issues.apache.org/jira/browse/KAFKA-3333
> Project: Kafka
> Issue Type: New Feature
> Components: clients
> Affects Versions: 0.10.0.0
> Reporter: Stephen Powis
> Fix For: 0.10.2.0
>
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> The
> [DefaultPartitioner|https://github.com/apache/kafka/blob/trunk/clients/src/main/java/org/apache/kafka/clients/producer/internals/DefaultPartitioner.java]
> typically distributes using the hash of the keybytes, and falls back to
> round robin if there is no key. But there is currently no way to do Round
> Robin partitioning if you have keys on your messages without writing your own
> partitioning implementation.
> I think it'd be helpful to have an implementation of straight Round Robin
> partitioning included with the library.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
