[
https://issues.apache.org/jira/browse/MAHOUT-1794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15201004#comment-15201004
]
ASF GitHub Bot commented on MAHOUT-1794:
----------------------------------------
Github user andrewmusselman commented on the pull request:
https://github.com/apache/mahout/pull/178#issuecomment-198204388
Thanks for the addition, Al. The way this is written you will have to
create a similar path on your local file system that will mirror the path you
have in HDFS.
So for example, if you create a path in HDFS for your data, say
'/user/al/mahout-tmp', and export it into MAHOUT_WORK_DIR, you will need to
make the same path on the local file system at '/user/al/mahout-tmp'. Does that
work for people do you think?
> Support alternate temporary directories in example scripts.
> -----------------------------------------------------------
>
> Key: MAHOUT-1794
> URL: https://issues.apache.org/jira/browse/MAHOUT-1794
> Project: Mahout
> Issue Type: Improvement
> Components: Examples
> Affects Versions: 0.11.1
> Reporter: Albert Chu
> Assignee: Andrew Musselman
> Priority: Minor
> Fix For: 0.12.0
>
>
> In many of the example scripts, a directory in /tmp (e.g.
> /tmp/mahout-work-dir) is created for use as temporary scratch space, such as
> to store data files that are downloaded.
> In a number of HPC environments, /tmp may not exist or /tmp may be very small
> b/c local disk drives don't exist. It'd be convenient to be able to specify
> an alternate directory to use as scratch space.
> Pull request via Github to be sent shortly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
