You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Luke Chen (Jira)" <ji...@apache.org> on 2021/10/18 07:29:00 UTC

[jira] [Assigned] (KAFKA-12606) Some improvements for produce record validation

     [ https://issues.apache.org/jira/browse/KAFKA-12606?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Luke Chen reassigned KAFKA-12606:
---------------------------------

    Assignee: Luke Chen

> Some improvements for produce record validation
> -----------------------------------------------
>
>                 Key: KAFKA-12606
>                 URL: https://issues.apache.org/jira/browse/KAFKA-12606
>             Project: Kafka
>          Issue Type: Improvement
>            Reporter: Jason Gustafson
>            Assignee: Luke Chen
>            Priority: Major
>              Labels: needs-kip
>
> KIP-467 introduces an extension to the produce response to let the producer identify the specific records that had failed validation. While implementing KAFKA-12548, we realized a few shortcomings in this proposal:
> 1. Some records may fail only because they are part of a batch which had a record failing validation. In this case, we can return a better exception to the user so that they know the record is safe to be retried. For example, `RecordNotAppendedException` or something like that.
> 2. Records in the same batch may fail validation for different reasons. For example, one record may fail because of an invalid timestamp; another may fail because it uses a null key and the topic is compacted. However, the schema only allows a single error for each partition, so we cannot distinguish these cases in order to throw more specific exception types. We should consider allowing a record-level error code as well.



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