[
https://issues.apache.org/jira/browse/CRUNCH-24?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13421709#comment-13421709
]
Gabriel Reid commented on CRUNCH-24:
------------------------------------
It turns out that the issue with the MapsideJoinIT not running on OS X is due
to a combination of DistributedCache not really being supported in local mode,
and the default temporary directory of all things Hadoop being "/tmp" (in
HDFS), while the default temporary directory on Linux is also "/tmp" and on OS
X it's something else.
I've attached an updated version of CRUNCH-24-josh.patch which undoes the
changes to MapsideJoin, and sets the default temporary directory in
MapsideJoinIT. This appears to rectify the issue. I've tested it on OS X, but I
don't have a Linux machine handy at the moment to test it there; however, I'm
confident that it will work there as well.
@Rahul, I took a look at your patch, and I don't think that it will work when
running in distributed mode. There is a call to FileSystem#copyFromLocalFile
with a path that is on HDFS when running in distributed mode.
> Make test suite suitable for continuous integration
> ---------------------------------------------------
>
> Key: CRUNCH-24
> URL: https://issues.apache.org/jira/browse/CRUNCH-24
> Project: Crunch
> Issue Type: Task
> Affects Versions: 0.3.0
> Reporter: Matthias Friedrich
> Assignee: Matthias Friedrich
> Fix For: 0.3.0
>
> Attachments: 0001-CRUNCH-24-Clean-up-test-suite-output.patch,
> 0001-CRUNCH-24-Distributed-cache.patch,
> 0001-CRUNCH-24-make-testsuite-sutable-for-CI.patch,
> 0001-CRUNCH-24-make-testsuite-sutable-for-CI.patch, CRUNCH-24-gabriel.patch,
> CRUNCH-24-josh.patch
>
>
> Right now the integration test suite leaves about 80 files behind on /tmp
> making it unsuitable for a shared continuous integration environment.
> Examples for these files are test case output ("output*") and Crunch's own
> temporary files ("crunch*", see CRUNCH-21).
> We have to delete these files or make sure they aren't created on /tmp in the
> first place.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
