[
https://issues.apache.org/jira/browse/KAFKA-10686?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17226725#comment-17226725
]
Bruno Cadonna commented on KAFKA-10686:
---------------------------------------
Hi [~lkokhreidze], I guess this ticket makes sense. Looking forward to your KIP!
You might want to look into KIP-441 to see how a Kafka Streams client passes
its task lags to the assignor. I can imagine that you want to use a similar
mechanism to pass information about a Kafka Streams client to the assignor.
> Pluggable standby tasks assignor for Kafka Streams
> --------------------------------------------------
>
> Key: KAFKA-10686
> URL: https://issues.apache.org/jira/browse/KAFKA-10686
> Project: Kafka
> Issue Type: Improvement
> Components: streams
> Reporter: Levani Kokhreidze
> Priority: Major
>
> In production, Kafka Streams instances often run across different clusters
> and availability zones. In order to guarantee high availability of the Kafka
> Streams deployments, users would need more granular control over which on
> instances standby tasks can be created.
> Idea of this ticket is to expose interface for Kafka Streams which can be
> implemented by users to control where standby tasks can be created.
> Kafka Streams can have RackAware assignment as a default implementation that
> will take into account `rack.id` of the application and make sure that
> standby tasks are created on different racks.
> Point of this ticket though is to give more flexibility to users on standby
> task creation, in cases where just rack awareness is not enough.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
