[ 
https://issues.apache.org/jira/browse/FLINK-15839?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

ASF GitHub Bot updated FLINK-15839:
-----------------------------------
    Labels: pull-request-available  (was: )

> Revisit Stateful Functions KafkaIngressBuilder properties resolution logic
> --------------------------------------------------------------------------
>
>                 Key: FLINK-15839
>                 URL: https://issues.apache.org/jira/browse/FLINK-15839
>             Project: Flink
>          Issue Type: Bug
>          Components: Stateful Functions
>    Affects Versions: statefun-1.1
>            Reporter: Tzu-Li (Gordon) Tai
>            Assignee: Tzu-Li (Gordon) Tai
>            Priority: Major
>              Labels: pull-request-available
>
> Currently, the properties resolution logic in {{KafkaIngressBuilder}} is a 
> bit inconsistent for different configurations, and some actually incorrect.
> The problem is around the fact that we allow users to directly pass in 
> {{Properties}} to configure the Kafka client, but also support named methods 
> to set some important configs like Kafka address / auto offset reset position.
> For example, we always overwrite {{auto.offset.reset}} set in the properties 
> with the {{autoOffsetResetPosition}} value in the builder. This is correct 
> when the user had actually passed in a value via the named method 
> {{withAutoOffsetResetPosition}}, but incorrect otherwise.
> The same goes for the Kafka address configuration.
> This should be revisited, so that we have a common strategy with dealing with 
> named configurations v.s. properties, with an end goal that:
> * Configs passed via named methods should always overwrite the value set via 
> properties
> * Any default values for named configuration methods should be defined in the 
> builder
> * If no config was passed via its named method, then we use the default value 
> (if any) to overwrite the properties IFF the user also did not provide a 
> value for it there.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to