[
https://issues.apache.org/jira/browse/FLINK-21820?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17302498#comment-17302498
]
Nicholas Jiang edited comment on FLINK-21820 at 3/17/21, 1:42 AM:
------------------------------------------------------------------
[~Leonard Xu], by default, [most JDBC drivers use a fetch size of
10|https://docs.oracle.com/middleware/1212/toplink/TLJPA/q_jdbc_fetch_size.htm#TLJPA647].
This could set the 'scan.fetch-size' to 10 as default value.
[~jark], [~Leonard Xu], what do you think about the default value?
was (Author: nicholasjiang):
[~Leonard Xu], by default, [most JDBC drivers use a fetch size of
10|https://docs.oracle.com/middleware/1212/toplink/TLJPA/q_jdbc_fetch_size.htm#TLJPA647].
We could set the 'scan.fetch-size' to 10 as default value.
> JDBC connector shouldn't read all rows in per statement by default
> ------------------------------------------------------------------
>
> Key: FLINK-21820
> URL: https://issues.apache.org/jira/browse/FLINK-21820
> Project: Flink
> Issue Type: Improvement
> Components: Connectors / JDBC, Table SQL / Ecosystem
> Reporter: Leonard Xu
> Priority: Major
>
> The default value for JDBC option 'scan.fetch-size' is 0 which means read all
> rows in statement, this may lead to OOM or IO timeout.
> We'd better set a reasonable value as default value.
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
