You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Timo Walther (Jira)" <ji...@apache.org> on 2022/01/04 12:43:00 UTC

[jira] [Commented] (FLINK-25439) StreamExecCalc collect new StreamRecord to Downstream lost timestamp Attributes

    [ https://issues.apache.org/jira/browse/FLINK-25439?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17468611#comment-17468611 ] 

Timo Walther commented on FLINK-25439:
--------------------------------------

At which location would you like to access the StreamRecord timestamp? By default, StreamRecord do not contain timestamps within the table engine. There are reinserted at some locations when leaving the table engine e.g. to DataStream API. Flink 1.14.3 will fix an issue where the timestamp was not inserted into some kind of sinks (FLINK-24608). 

> StreamExecCalc collect new StreamRecord to Downstream lost timestamp Attributes
> -------------------------------------------------------------------------------
>
>                 Key: FLINK-25439
>                 URL: https://issues.apache.org/jira/browse/FLINK-25439
>             Project: Flink
>          Issue Type: Bug
>          Components: Table SQL / Planner
>    Affects Versions: 1.14.2
>            Reporter: HunterHunter
>            Priority: Major
>         Attachments: image-2021-12-24-10-57-01-479.png, image-2021-12-24-10-58-25-304.png
>
>
> When we read data from kafkasource, we can see in SourceReaderBase.pollNext that the record has a timestamp attribute, but after StreamExecCalc the downstream receives the streamrecord and the timestamp attribute is missing
>  



--
This message was sent by Atlassian Jira
(v8.20.1#820001)