kafka-jira mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "siva santhalingam (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (KAFKA-6451) Simplify KStreamReduce
Date Tue, 16 Jan 2018 20:22:01 GMT

    [ https://issues.apache.org/jira/browse/KAFKA-6451?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16327727#comment-16327727
] 

siva santhalingam commented on KAFKA-6451:
------------------------------------------

Picking this up for now. Please feel free to reassign.

> Simplify KStreamReduce
> ----------------------
>
>                 Key: KAFKA-6451
>                 URL: https://issues.apache.org/jira/browse/KAFKA-6451
>             Project: Kafka
>          Issue Type: Bug
>          Components: streams
>    Affects Versions: 1.0.0
>            Reporter: Matthias J. Sax
>            Assignee: siva santhalingam
>            Priority: Minor
>              Labels: beginner, newbie
>
> If we do aggregations, we drop records with {{key=null}} or {{value=null}}. However,
in {{KStreamReduce}} we only early exit if {{key=null}} and process {{value=null}} – even
if we only update the state with it's old value and also only send the old value downstream
(ie, we still compute the correct result), it's undesired and wasteful and we should early
exit on {{value=null}}, too.
> This problem might occur for {{KStreamAggregate}} or other processors, too, and we need
to double check those to make sure we implement consistent behavior.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message