[
https://issues.apache.org/jira/browse/HADOOP-4470?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12641349#action_12641349
]
Steve Loughran commented on HADOOP-4470:
----------------------------------------
That's interesting. Certainly I can create similar scenarios in my tests, but
that's because I'm doing my test runs all in the same process, so until I have
shutdown working properly across all services (and their wrapper classes) I can
see similar events. But in the Hadoop core junit classes, every test class is
running in its own process. If the port is staying open, it means that the test
case isn't stopping.
1. You can use netstat -a -p to list the processes that have open ports; link
that using jps -v to the java process at fault.
2. do you have test.junit.fork.mode set to something other than perTest ?
> Tests fail on windows due to BindException.
> -------------------------------------------
>
> Key: HADOOP-4470
> URL: https://issues.apache.org/jira/browse/HADOOP-4470
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs, test
> Affects Versions: 0.19.0
> Environment: Windows, cygwin.
> Reporter: Konstantin Shvachko
> Fix For: 0.19.0
>
>
> On Windows a series of tests fail starting from {{TestAbandonBlock}}.
> The reason is that name-nodes or data-nodes cannot start because of the
> "BindException: Address already in use: connect".
> Seems like some of the tests that run before {{TestAbandonBlock}} are not
> releasing resources or not closing sockets.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
