hadoop-hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Johan Oskarsson (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HIVE-88) hadoop doesn't use conf/hive-log4j.properties
Date Thu, 02 Apr 2009 15:04:13 GMT

     [ https://issues.apache.org/jira/browse/HIVE-88?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Johan Oskarsson updated HIVE-88:
--------------------------------

    Fix Version/s:     (was: 0.2.0)
                   0.4.0

> hadoop doesn't use conf/hive-log4j.properties
> ---------------------------------------------
>
>                 Key: HIVE-88
>                 URL: https://issues.apache.org/jira/browse/HIVE-88
>             Project: Hadoop Hive
>          Issue Type: Improvement
>          Components: Configuration
>            Reporter: Michi Mutsuzaki
>            Assignee: Joydeep Sen Sarma
>            Priority: Minor
>             Fix For: 0.4.0
>
>
> hadoop-0.20.0-dev-core.jar contains log4j.properties file, and I think that's the one
hadoop is picking up. I modified both conf/hive-log4j.properties and hadoopcore/conf/log4j.properties,
but hadoop still printed INFO messages to stderr.
> Pasting relevant posts from the mailing list below:
> Michi Mutsuzaki <michi@cs.stanford.edu> 	Fri, Nov 28, 2008 at 7:14 PM
> To: hive-users@publists.facebook.com
> Hello,
> When I do "ant test" under ql directory, I get many log messages to stderr.
>    [junit] 08/11/28 19:04:14 INFO exec.MapOperator: Got partitions: null
>    [junit] 08/11/28 19:04:14 INFO exec.ReduceSinkOperator: Initializing Self
>    [junit] 08/11/28 19:04:14 INFO exec.ReduceSinkOperator: Using tag = -1
>    [junit] 08/11/28 19:04:14 INFO thrift.TBinarySortableProtocol:
> Sort order is ""
>    [junit] 08/11/28 19:04:14 INFO thrift.TBinarySortableProtocol:
> Sort order is ""
>    ....
> I tried setting log level to ERROR in conf/hive-log4j.properties, but these info lines
still show up. How can I get rid of them?
> Thanks!
> --Michi
> Joydeep Sen Sarma <jssarma@facebook.com> 	Fri, Nov 28, 2008 at 10:49 PM
> To: "michi@cs.stanford.edu" <michi@cs.stanford.edu>, "hive-users@publists.facebook.com"
<hive-users@publists.facebook.com>
> When we run the tests - we run in hadoop 'local' mode - and in this mode, we run map-reduce
jobs by invoking 'hadoop jar ... ExecDriver' cmd line. this was done because we had some issues
submitting map-reduce jobs directly (from same jvm) in local mode that we could not resolve.
> The issue is that when we invoke 'hadoop jar ... ExecDriver' - we don't control log4j
via hive-log4j. one thing u can try is changing the hadoop's log4j.properties that hive is
picking up (probably hadoopcore/conf/log4j.properties).
> Revisiting this after a long time - I think this can be fixed with some changes to MapRedTask.java
(need to add hive-log4j.properties to hadoop classpath here and then reset log4j using this
in execdriver). Feel free to file a jira if this is too irritating ..

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


Mime
View raw message