[
https://issues.apache.org/jira/browse/APEXCORE-146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15299215#comment-15299215
]
Ilya Ganelin commented on APEXCORE-146:
---------------------------------------
I've been looking into the code. It seems that a lot of the logic within the
StatelessPartitioner would go away in a ClonePartitioner. Is there any issue
with assigning an identical set of PartitionKeys to each input port / partition
combination? Does there even need to be more than one key for other logic to
work?
Repartitioning logic becomes somewhat undefined. It doesn't make sense to
partition based on load, I think the only function needed is the
definePartitions() function which simply assigns an arbitrary key assignment to
each partition, e.g. a 0 key.
Is my understanding correct?
> Add A Clone Partitioner Which Sends The Same Data To Each Partition
> -------------------------------------------------------------------
>
> Key: APEXCORE-146
> URL: https://issues.apache.org/jira/browse/APEXCORE-146
> Project: Apache Apex Core
> Issue Type: New Feature
> Reporter: Timothy Farkas
> Assignee: Ilya Ganelin
>
> This should go into com.datatorrent.common.partitioner and would be very
> similar to the StatelessPartitioner
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
