hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Allen Wittenauer (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HADOOP-7984) Add hadoop command verbose and debug options
Date Wed, 24 Sep 2014 20:25:35 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-7984?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14146837#comment-14146837

Allen Wittenauer commented on HADOOP-7984:

Hopefully [~cnauroth] won't mind me adding him to this JIRA so that he can look over the batch
code. ;)

It looks good, with only one issue that I saw.  We should avoid adding more of the crazy project
split variables. So please drop the HADOOP_MAPRED_LOGLEVEL and YARN_LOGLEVEL bits and just
use HADOOP_LOGLEVEL everywhere.  Those per-project vars are a source of continual problems
out in the real world, and we definitely don't want to propagate more of them!  (Easy example:
 'hadoop jar' and 'yarn jar' work differently depending upon what is in the various *-env.sh,
despite doing essentially the same thing...)

I saw that you changed the log level for the security audit log, but I don't think it has
anything but INFO coded.  It's a coin toss whether or not we should actually change that.

Also, one of the stated goals of this JIRA was to change the default from INFO up to WARN.
 Is this something we still want to do?

Comments from the crowd would be good... :)



> Add hadoop command verbose and debug options
> --------------------------------------------
>                 Key: HADOOP-7984
>                 URL: https://issues.apache.org/jira/browse/HADOOP-7984
>             Project: Hadoop Common
>          Issue Type: New Feature
>          Components: scripts
>            Reporter: Eli Collins
>            Assignee: Akira AJISAKA
>            Priority: Minor
>              Labels: newbie
>         Attachments: HADOOP-7984.patch, HADOOP-7984.patch
> It would be helpful if bin/hadoop had verbose and debug flags. Currently users need to
set an env variable or prefix the command (eg "HADOOP_ROOT_LOGGER=DEBUG,console hadoop distcp")
which isn't very user friendly. We currently log INFO by default. How about we only log ERROR
and WARN by default, then -verbose triggers INFO and -debug triggers DEBUG?

This message was sent by Atlassian JIRA

View raw message