[ 
https://issues.apache.org/jira/browse/HDFS-3804?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13485123#comment-13485123
 ] 

Daryn Sharp commented on HDFS-3804:
-----------------------------------

Unless the tests were conflicting, I'm a bit bothered about starting and 
stopping the mini-cluster for every test case when most of them don't even need 
the cluster running.  I think you can probably leave the cluster 
startup/shutdown in @BeforeClass/@AfterClass.

It's a shame that cluster.getNamesystem().clear() isn't exposed for testing...  
If you feel ambitious, you could tag it with @VisibleForTesting and then call 
it in a @Before to wipe out the namenode.  If you try it, and it works, we 
should file another jira to change more tests since it might drastically cut 
the runtime of the test suite.
                
> TestHftpFileSystem fails intermittently with JDK7
> -------------------------------------------------
>
>                 Key: HDFS-3804
>                 URL: https://issues.apache.org/jira/browse/HDFS-3804
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: test
>         Environment: Apache Maven 3.0.4
> Maven home: /usr/share/maven
> Java version: 1.7.0_04, vendor: Oracle Corporation
> Java home: /usr/lib/jvm/jdk1.7.0_04/jre
> Default locale: en_US, platform encoding: ISO-8859-1
> OS name: "linux", version: "3.2.0-25-generic", arch: "amd64", family: "unix"
>            Reporter: Trevor Robinson
>            Assignee: Trevor Robinson
>              Labels: java7
>         Attachments: HDFS-3804-2.patch, HDFS-3804.patch
>
>
> For example:
>   testFileNameEncoding(org.apache.hadoop.hdfs.TestHftpFileSystem): Filesystem 
> closed
>   testDataNodeRedirect(org.apache.hadoop.hdfs.TestHftpFileSystem): Filesystem 
> closed
> This test case sets up a filesystem that is used by the first half of the 
> test methods (in declaration order), but the second half of the tests start 
> by calling {{FileSystem.closeAll}}. With JDK7, test methods are run in an 
> arbitrary order, so if any first half methods run after any second half 
> methods, they fail.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to