[ 
https://issues.apache.org/jira/browse/FLINK-35240?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17841117#comment-17841117
 ] 

Alexander Fedulov commented on FLINK-35240:
-------------------------------------------

I don't think touching any of the flush-specific properties should be 
necessary. You can see in the FlameGraph that flush calls are due to close 
being called and, as [~robyoung] mentioned, this is what 
JsonGenerator.Feature#AUTO_CLOSE_TARGET is there for. 

> Disable FLUSH_AFTER_WRITE_VALUE to avoid flush per record
> ---------------------------------------------------------
>
>                 Key: FLINK-35240
>                 URL: https://issues.apache.org/jira/browse/FLINK-35240
>             Project: Flink
>          Issue Type: Improvement
>          Components: Connectors / FileSystem
>            Reporter: Zhongqiang Gong
>            Priority: Minor
>         Attachments: image-2024-04-26-00-23-29-975.png, screenshot-1.png
>
>
> *Reproduce:*
> * According to user email: 
> https://lists.apache.org/thread/9j5z8hv4vjkd54dkzqy1ryyvm0l5rxhc
> *  !image-2024-04-26-00-23-29-975.png! 
> *Analysis:*
> * `org.apache.flink.formats.csv.CsvBulkWriter#addElement` will flush per 
> record.
> *Solution:*
> * I think maybe we can disable `FLUSH_AFTER_WRITE_VALUE` to avoid flush when 
> a record added.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to