[jira] Commented: (HADOOP-2896) Using transient jetty servers as guis is a bad idea

["Amareshwari Sriramadasu (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-2896?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12574437#action_12574437 ] 

Amareshwari Sriramadasu commented on HADOOP-2896:
-------------------------------------------------

The textual summary for command hadoop job -history can be described as follows:

bin/hadoop job -history <outputdir> can print the useful data for the user. i.e.
    1. print Job Details
    2. print Task Summary - consisting of number of total, successful, failed and killed map/reduce tasks.
    3. print Job Analysis - similar data as in analyle this job link in job history.
    4. print Failed and Killed map/reduce tasks.
    5. print failed/killed attempts on nodes - consisting of host name, comma seperated task attempt list failed/killed on that node.

Since job history jsps has data about all the tasks and task attempts, and that data is very huge, we can take an option from user to print all the data or not. So, 
bin/hadoop job -history all <outputdir> can print the following
    1. print Job Details
    2. print Task Summary - consisting of number of total, successful, failed and killed map/reduce tasks. 
    3. print Job Analysis - similar data as in analyle this job link in job history.
    4. print Failed and Killed map/reduce tasks.
    5. print Successful map/reduce tasks.
    6. print all map/reduce task attempts - along with hostname on which it ran.
    7. print failed/killed attempts on nodes - consisting of host name, comma seperated task attempt list failed/killed on that node.


Thoughts?

> Using transient jetty servers as guis is a bad idea
> ---------------------------------------------------
>
>                 Key: HADOOP-2896
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2896
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: mapred
>    Affects Versions: 0.17.0
>            Reporter: Owen O'Malley
>            Assignee: Amareshwari Sriramadasu
>             Fix For: 0.17.0
>
>
> Using transient jetty servers (ie. one that last 30 minutes) is a very poor replacement for a gui. I would much rather have bin/hadoop job -history out-dir print a textual summary rather than start a jetty server on the client machine that needs to be queried by the user.

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