[
https://issues.apache.org/jira/browse/HIVE-17906?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16238695#comment-16238695
]
Sergey Shelukhin commented on HIVE-17906:
-----------------------------------------
Actually thjere's an issue with the patch that if query kill succeeds, we don't
want the "irrelevant" flag to be set; but if we restart, we do. Meanwhile the
session can come back concurrently (i.e. query finishes just before we want to
kill, I'm not actually sure if kill will succeed or fail). Seems like falling
back to restart will have to go thru the master thread too, so it can compare
whatever happened while kill was failing.
> use kill query mechanics to kill queries in WM
> ----------------------------------------------
>
> Key: HIVE-17906
> URL: https://issues.apache.org/jira/browse/HIVE-17906
> Project: Hive
> Issue Type: Sub-task
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Priority: Major
> Attachments: HIVE-17906.patch
>
>
> Right now it just closes the session (see HIVE-17841). The sessions would
> need to be reused after the kill, or closed after the kill if the total QP
> has decreased
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
