[
https://issues.apache.org/jira/browse/FLINK-22897?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
zhengjiewen updated FLINK-22897:
--------------------------------
Description:
I try to use flink sql in batch mode, to sink data into hive partition table,
here is the sql:
{code:java}
//代码占位符
INSERT OVERWRTITE 【targetTable】SELECT 【field】FROM 【sourceTable】;
{code}
And I found that when the parallelism of the sink operator is different from
that of the operator before it, a large number of small files will be
generated. But this is not the case when the parallelism is the same.
was:
I try to use flink sql in batch mode, to sink data into hive partition table,
here is the sql:
{code:java}
//代码占位符
{code}
INSERT OVERWRTITE 【targetTable】SELECT 【field】FROM 【sourceTable】;
And I found that when the parallelism of the sink operator is different from
that of the operator before it, a large number of small files will be
generated. But this is not the case when the parallelism is the same.
> FlinkSQL1.12 Sink to Hive with diffrent parallelism will due to produce many
> small files
> ----------------------------------------------------------------------------------------
>
> Key: FLINK-22897
> URL: https://issues.apache.org/jira/browse/FLINK-22897
> Project: Flink
> Issue Type: Improvement
> Affects Versions: 1.12.1
> Reporter: zhengjiewen
> Priority: Major
>
> I try to use flink sql in batch mode, to sink data into hive partition table,
> here is the sql:
>
> {code:java}
> //代码占位符
> INSERT OVERWRTITE 【targetTable】SELECT 【field】FROM 【sourceTable】;
> {code}
>
>
>
> And I found that when the parallelism of the sink operator is different from
> that of the operator before it, a large number of small files will be
> generated. But this is not the case when the parallelism is the same.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
