[jira] [Updated] (AMBARI-14559) Hive View Visual explain caches and doesn't account for settings changes

["Pallav Kulshreshtha (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/AMBARI-14559?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Pallav Kulshreshtha updated AMBARI-14559:
-----------------------------------------
    Resolution: Fixed
        Status: Resolved  (was: Patch Available)

committed to trunk.

> Hive View Visual explain caches and doesn't account for settings changes
> ------------------------------------------------------------------------
>
>                 Key: AMBARI-14559
>                 URL: https://issues.apache.org/jira/browse/AMBARI-14559
>             Project: Ambari
>          Issue Type: Bug
>          Components: ambari-views
>            Reporter: Pallav Kulshreshtha
>            Assignee: Pallav Kulshreshtha
>             Fix For: 2.4.0
>
>         Attachments: AMBARI-14559_branch-2.2.patch, Map Join.png, Shuffle Join.png
>
>
> A setting like hive.auto.convert.join will often radically change a Hive explain plan. In particular the query below will be a map join or shuffle join depending on this setting being true or false, respectively.
> What I see in the Hive View is that when I open visual explain, I get the correct explain plan based on the setting at that time. If I go and change the setting, then go back to the visual explain, the explain is not updated. If I reload the tab, the correct explain plan will show up when I go back to the explain view.
> See attached shots.
> Query I use is this:
> {code}
> select time_by_day.the_year as c0, time_by_day.quarter as c1,
> count(sales_fact_1997.product_id) as m0,
> sum(sales_fact_1997.unit_sales) as m1,
> sum(sales_fact_1997.store_sales) as m2,
> sum(sales_fact_1997.store_cost) as m3,
> count(distinct sales_fact_1997.customer_id) as m4
> from time_by_day as time_by_day,
> sales_fact_1997 as sales_fact_1997 where
> sales_fact_1997.time_id = time_by_day.time_id and
> time_by_day.the_year = 1997 group by time_by_day.the_year,
> time_by_day.quarter ;
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)