Mailing-List: contact core-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: core-dev@hadoop.apache.org
Message-ID: <128816665.1226614904414.JavaMail.jira@brutus>
Date: Thu, 13 Nov 2008 14:21:44 -0800 (PST)
From: "Chris Douglas (JIRA)" <jira@apache.org>
To: core-dev@hadoop.apache.org
Subject: [jira] Updated: (HADOOP-3770) improve composition, submission and
 result collection of gridmix
In-Reply-To: <346340062.1216171891653.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/HADOOP-3770?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Chris Douglas updated HADOOP-3770:
----------------------------------

    Status: Open  (was: Patch Available)

I haven't been over the details, but had a few general suggestions after a first pass:
* Most of this doesn't conform to the coding guidelines. Converting tabs to spaces, removing commented-out code, putting constants and defaults in a reasonable place, etc. should be done before this can be committed.
* Temporary directories should be configurable and default off of a single, configurable temp dir rather than being hard-coded off /tmp
* If independent from the original, the configuration and drivers of gridmix2 should not be in the same place. If this is intended as a replacement for gridmix, it should modify the existing benchmark rather than creating _file_, _file2_ pairs. If it's a new benchmark, it should be in src/benchmarks/gridmix2.
* Would it be possible to split the pig benchmarks into a separate JIRA? This is simply too large to review well.
* GridMixRunner is unnecessarily enormous. Most of the methods are setting defaults and performing work best encapsulated in the \*Creator classes that currently do trivial work. This class would also benefit from utility methods converting results from Configuration::getStrings to int[] (instead of subclassing Configuration), abstracting out the creation of unique Strings for runs (the use of Calendar/Date may not be the correct choice), javadoc, and general cleanup
* Exceptions are almost always ignored; most probably should not be.

> improve composition, submission and result collection of gridmix
> ----------------------------------------------------------------
>
>                 Key: HADOOP-3770
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3770
>             Project: Hadoop Core
>          Issue Type: Improvement
>            Reporter: Lingyun Yang
>            Assignee: Runping Qi
>         Attachments: patch-3770.txt, patch-3770.txt, patch-3770.v2.txt
>
>
> Current gridmix submits jobs using a set of scripts, which is inconvenient and the results are difficult to collect.  To improve the gridmix submission and results collection, we implemented a new program  using JobControl to submit and collect the results of jobs 
> Also the new gridmix allows to have more different types of jobs such as, pig jobs, jobs with combiners etc. 

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.