Mailing-List: contact common-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: common-issues@hadoop.apache.org
Date: Wed, 21 Oct 2015 21:29:27 +0000 (UTC)
From: "Chris Nauroth (JIRA)" <jira@apache.org>
To: common-issues@hadoop.apache.org
Message-ID: <JIRA.12824296.1430146335000.22210.1445462967974@Atlassian.JIRA>
In-Reply-To: <JIRA.12824296.1430146335000@Atlassian.JIRA>
References: <JIRA.12824296.1430146335000@Atlassian.JIRA>
 <JIRA.12824296.1430146335238@arcas>
Subject: [jira] [Commented] (HADOOP-11880) aw jira sub-task testing, ignore
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HADOOP-11880?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14967975#comment-14967975 ] 

Chris Nauroth commented on HADOOP-11880:
----------------------------------------

Got it.  Thanks!

bq. Either revert with a hard-coded path and expect parallel tests to fail, or extend the dfs builder to add an operation to set the subdir, which would be retained over a test case/test suite.

I just commented in HDFS-9263 that the Maven build, when running tests in parallel mode, automatically creates a separate test directory for each concurrent test process to use.  Thus, the isolation is achieved at the Maven build level, and there is no requirement for {{MiniDFSCluster}} itself to use a unique workspace.

Of course, that only holds true if all tests are playing nice and using the properties passed down by Maven to determine their test path.  I'd like to explore fixing this by backtracing from the bad paths shown in the RAT report to try to find the offending tests.  Given that I think {{MiniDFSCluster#getBaseDirectory}} is correct already, I suspect we'll find the problems are in tests that don't use {{MiniDFSCluster}}, or in tests that explicitly configure their own data directories instead of relying on {{MiniDFSCluster#getBaseDirectory}}.

I'm reluctant to revert the test parallelization.  If it's more urgent to complete this cutover first though, then I understand.

> aw jira sub-task testing, ignore
> --------------------------------
>
>                 Key: HADOOP-11880
>                 URL: https://issues.apache.org/jira/browse/HADOOP-11880
>             Project: Hadoop Common
>          Issue Type: Sub-task
>            Reporter: Allen Wittenauer
>         Attachments: HDFS-9263-001.patch
>
>


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)