[
https://issues.apache.org/jira/browse/DRILL-7490?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17165473#comment-17165473
]
Yusuf commented on DRILL-7490:
------------------------------
hi,
please,
limit and offset fields should all be pushdown in RDBMS.
Regardless of the number of columns selected, regardless of the number of joins
and tables, pushdown should be done in RDBMS
> limit is not pushed to JDBC storage plugin
> ------------------------------------------
>
> Key: DRILL-7490
> URL: https://issues.apache.org/jira/browse/DRILL-7490
> Project: Apache Drill
> Issue Type: Bug
> Components: Query Planning & Optimization, Storage - JDBC
> Affects Versions: 1.11.0
> Reporter: Vova Vysotskyi
> Assignee: Vova Vysotskyi
> Priority: Major
> Labels: ready-to-commit
> Fix For: 1.17.0
>
> Attachments: image-2020-07-24-17-31-02-589.png,
> image-2020-07-24-17-31-48-831.png, image-2020-07-24-17-35-28-912.png,
> image-2020-07-24-17-45-27-357.png, image-2020-07-24-17-46-05-219.png,
> image-2020-07-24-17-46-59-584.png, image-2020-07-24-17-48-53-740.png
>
>
> The limit is not pushed to JDBC storage plugin.
> The following query:
> {code:sql}
> select person_id from mysql.`drill_mysql_test`.person limit 10
> {code}
> Returns plan:
> {noformat}
> 00-00 Screen
> 00-01 Project(person_id=[$0])
> 00-02 SelectionVectorRemover
> 00-03 Limit(fetch=[10])
> 00-04 Limit(fetch=[10])
> 00-05 Jdbc(sql=[SELECT `person_id` FROM
> `drill_mysql_test`.`person` ])
> {noformat}
> This issue happens since incorrect row count is calculated for {{JdbcSort}}.
> Drill enforces to use {{RelNode.estimateRowCount(mq)}} for calculating row
> count for all inheritants of {{Sort}} rel node (see {{DrillRelMdRowCount}}),
> but {{JdbcSort}} doesn't override this method and uses parent one which
> returns row count of input. So planner choses {{DrillLimitRel}} since it has
> less row count.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
