[
https://issues.apache.org/jira/browse/MAHOUT-578?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12977992#action_12977992
]
Shige Takeda commented on MAHOUT-578:
-------------------------------------
I'm checking github's trunk... it seems this fix is not there yet.
https://github.com/apache/mahout/blob/trunk/core/src/main/java/org/apache/mahout/clustering/canopy/CanopyDriver.java
In sequential mode, config is used like:
FileSystem fs = FileSystem.get(input.toUri(), conf);
Let me try checking how job config impacts to sequential ones... I'm not sure
if -D can accept other hadoop parameters than mapred.*, though...
http://hadoop.apache.org/common/docs/current/api/org/apache/hadoop
> canopy clustering fails if --Dmapred.job.queue.name=unfunded is specified to
> mahout driver command line
> -------------------------------------------------------------------------------------------------------
>
> Key: MAHOUT-578
> URL: https://issues.apache.org/jira/browse/MAHOUT-578
> Project: Mahout
> Issue Type: Bug
> Components: Clustering
> Affects Versions: 0.4
> Environment: Linux 2.6.18-164.el5 x86_64 x86_64 x86_64 GNU/Linux
> Reporter: Shige Takeda
> Attachments:
> 0001-CanopyDriver-calls-ToolRunner.run-without-new-Config.patch
>
>
> Hi, I would like to demonstrate -D option issues by showing one concrete
> example, and would like to propose the fix.
> When I want to run canopy clustering using mahout driver, the command line is
> something like this. NOTE: -Dmapred.job.queue.name is required Job config in
> the company's environment.
> $MAHOUT_HOME/bin/mahout canopy \
> -i input \
> -o output \
> -dm org.apache.mahout.common.distance.EuclideanDistanceMeasure \
> -t1 2.0 \
> -t2 0.05 \
> -cl \
> -Dmapred.job.queue.name=unfunded
> and I get the error:
> Running on hadoop, using HADOOP_HOME=/grid/0/gs/hadoop/current
> HADOOP_CONF_DIR=/grid/0/gs/conf/current
> 11/01/05 20:19:15 ERROR common.AbstractJob: Unexpected
> -Dmapred.job.queue.name=unfunded while processing Job-Specific Options:
> This is because -D parameter is NOT parsed properly by ToolRunner.run but
> passed through to CanopyDriver's command line option parsers.
> ToolRunner.run(Tool,String[]) should be used rather than
> ToolRunner.run(Configuraiton,Tool,String[]) to get -D parameter processed.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
