hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Naveen Gangam (Jira)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-22853) Beeline should use HS2 server defaults for fetchSize
Date Mon, 24 Feb 2020 22:54:00 GMT

    [ https://issues.apache.org/jira/browse/HIVE-22853?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17043944#comment-17043944
] 

Naveen Gangam commented on HIVE-22853:
--------------------------------------

[~belugabehr] Thanks for the review.
Fair point regarding the default. But this needs more thought, I think. Having a no-limit
read when dealing with large scale data, worries me. It could potentially crash both services.
We can fail the set command if value is less than 0. The question still remains if ZERO should
be permitted and honored. [~ychena] Do you have any thoughts on having a no-limit fetch for
0?

>From the hive documentation, 
{noformat}
set <key>=<value>

Sets the value of a particular configuration variable (key).
Note: If you misspell the variable name, the CLI will not show an error.
{noformat}

I do not know if the following syntax is legal or works currently. I was going by what has
been stated in the documentation. But if you know that users have been using alternate syntax
such as these, it should be easy enough to incorporate this change.
{noformat}
set     hive.server2.thrift.resultset.default.fetch.size=1;
set hive.server2.thrift.resultset.default.fetch.size =1
set hive.server2.thrift.resultset.default.fetch.size = 1;
{noformat}



> Beeline should use HS2 server defaults for fetchSize
> ----------------------------------------------------
>
>                 Key: HIVE-22853
>                 URL: https://issues.apache.org/jira/browse/HIVE-22853
>             Project: Hive
>          Issue Type: Bug
>          Components: Beeline
>    Affects Versions: 4.0.0
>            Reporter: Naveen Gangam
>            Assignee: Naveen Gangam
>            Priority: Major
>         Attachments: HIVE-22853.2.patch, HIVE-22853.3.patch, HIVE-22853.patch
>
>
> Currently beeline uses a hard coded default of 1000 rows for fetchSize. This default
value is different from what the server has set. While the beeline user can reset the value
via set command, its cumbersome to change the workloads.
> Rather it should default to the server-side value and set should be used to override
within the session.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message