[
https://issues.apache.org/jira/browse/HUDI-9309?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yinsheng Wang updated HUDI-9309:
--------------------------------
Description:
[https://github.com/apache/hudi/pull/13044]
these two commit both use sparkRowSerDe to convert RDD[InternalRow] to
RDD[Row], and then use sparkSession.createDataFrame to convert to DataFrame.
this introduce perf regression for insert
fix step
use sparkSession.internalCreateDataFrame instead
was:
[https://github.com/apache/hudi/pull/13044]
these two commit both use sparkRowSerDe to convert RDD[InternalRow] to
RDD[Row], and then use sparkSession.createDataFrame to convert to DataFrame.
this introduce perf regression for insert
fix step
use sparkSession.internalCreateDataFrame
> Fix perf regression from supporting display dag of insert/update statment in
> spark ui
> -------------------------------------------------------------------------------------
>
> Key: HUDI-9309
> URL: https://issues.apache.org/jira/browse/HUDI-9309
> Project: Apache Hudi
> Issue Type: Improvement
> Reporter: Yinsheng Wang
> Priority: Major
>
> [https://github.com/apache/hudi/pull/13044]
> these two commit both use sparkRowSerDe to convert RDD[InternalRow] to
> RDD[Row], and then use sparkSession.createDataFrame to convert to DataFrame.
> this introduce perf regression for insert
>
> fix step
> use sparkSession.internalCreateDataFrame instead
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
