[
https://issues.apache.org/jira/browse/HDFS-6980?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14162578#comment-14162578
]
Andrew Wang commented on HDFS-6980:
-----------------------------------
Hi Tsuyoshi, thanks for working on this. I had a few review comments:
* I agree it's generally good practice to wrap all resource usage with
try/finally like this. However, I think in this particular case, a bunch of
leaks are caused by not cleaning up after expected errors. Changing just these,
and avoiding places where there is no leak as long the test passes, would make
the patch a lot smaller.
* While we're in this file, it'd be good to convert the static initializer into
a @BeforeClass method, and cleanup in an @AfterClass.
Besides that though, +1. Getting the tests in shape would be great.
> TestWebHdfsFileSystemContract fails in trunk
> --------------------------------------------
>
> Key: HDFS-6980
> URL: https://issues.apache.org/jira/browse/HDFS-6980
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: test
> Reporter: Akira AJISAKA
> Assignee: Tsuyoshi OZAWA
> Attachments: HDFS-6980.1-2.patch, HDFS-6980.1.patch
>
>
> Many tests in TestWebHdfsFileSystemContract fail by "too many open files"
> error.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
