hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Siddharth Seth (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-14572) Investigate jenkins test report timings
Date Thu, 18 Aug 2016 20:27:20 GMT

    [ https://issues.apache.org/jira/browse/HIVE-14572?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15427087#comment-15427087
] 

Siddharth Seth commented on HIVE-14572:
---------------------------------------

Thanks [~kgyrtkirk]. 2.16 was definitely not considering the time spent in setting up and
tearing down a testClass. That's where HIVE-13496 came from.

Will need to look at how the jenkins plugin works ... I don't think anything runs there as
part of ptest, does it?

> Investigate jenkins test report timings
> ---------------------------------------
>
>                 Key: HIVE-14572
>                 URL: https://issues.apache.org/jira/browse/HIVE-14572
>             Project: Hive
>          Issue Type: Sub-task
>          Components: Tests
>            Reporter: Zoltan Haindrich
>
> [~sseth] have noticed some odd timings in the jenkins reports
> I've created a sample project, to emulate a clidriver run during qtest:
> the testclass:
> * 1 sec beforeclass
> * 3x 0.2s test
> created using junit4 parameterized.
> Double checkout; second project runs different tests...or at least they have different
name.
> here are my preliminary findings:
> || thing || expected || 2.16 || 2.19.1
> | total time | ~3.4s | 1.2s | 3.4s 
> | package time | ~3.4s | 0.61s | 1.7s
> | class time | ~3.4s | 0.61s | 1.7s
> | testcase times | ~.2s | ~.2s | ~.2s 
> notes:
> * using 2.16 beforeclass timngs are totally hidden or lost
> * 2.19.1 does account for beforeclass but still fails to correctly aggregate the two
runs of the similary named testclasses
> it might worth a try to look at the bleeding edge of this jenkins plugin...



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

Mime
View raw message