[ 
https://issues.apache.org/jira/browse/HDFS-1805?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13049800#comment-13049800
 ] 

ramkrishna.s.vasudevan commented on HDFS-1805:
----------------------------------------------

Hi Daryn,
Thanks for the comments.

Few points to be clarified,
Making the cluster and conf as final:

@BeforeClass
        public static void startCluster() throws Exception {
                conf = new HdfsConfiguration();
                cluster = new 
MiniDFSCluster.Builder(conf).numDataNodes(2).build();
        }

'conf' can be made final by moving out of this @BeforeClass api.
But 'cluster' if we try to move outside the @BeforeClass then the build() api 
throws exception.  So it can be moved out like 
final static MiniDFSCluster cluster = new 
MiniDFSCluster.Builder(conf).numDataNodes(2).build();


The FileUtil.fullyDelete(fs, new Path("/"))
is deprecated.

If we use the FileUtil.fullyDeleteContents(File dir)  accepts specific file 
name or directory name.  So can we proceed with the api deleteFromDFS().
Based on the comments will prepare patch. Thanks.

> Some Tests in TestDFSShell can not shutdown the MiniDFSCluster on any 
> exception/assertion failure. This will leads to fail other testcases.
> -------------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: HDFS-1805
>                 URL: https://issues.apache.org/jira/browse/HDFS-1805
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: test
>    Affects Versions: 0.23.0
>            Reporter: Uma Maheswara Rao G
>            Assignee: Uma Maheswara Rao G
>            Priority: Minor
>             Fix For: 0.23.0
>
>         Attachments: HDFS-1805-1.patch, HDFS-1805-2.patch, HDFS-1805-3.patch, 
> HDFS-1805.patch
>
>
> Some test cases in TestDFSShell are not shutting down the MiniDFSCluster in 
> finally.
> If any test assertion failure or exception can result in not shutting down 
> this cluster. Because of this other testcases will fail. This will create 
> difficulty in finding the actual testcase failures.
> So, better to shutdown the cluster in finally. 

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Reply via email to