[
https://issues.apache.org/jira/browse/HDFS-2967?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13437119#comment-13437119
]
Eli Collins commented on HDFS-2967:
-----------------------------------
I've seen cases where this is not true, eg as Colin mentioned re-running after
a Ctrl-C and some tests (eg when run via eclipse) that put data in directories
that doesn't get cleared out.
> unit tests which alter persistent state should call cleanup before running
> --------------------------------------------------------------------------
>
> Key: HDFS-2967
> URL: https://issues.apache.org/jira/browse/HDFS-2967
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Colin Patrick McCabe
> Priority: Minor
>
> I did a little test, and found that aborting unit tests halfway through (for
> example, with the stop button in Eclipse) will NOT invoke the @After
> functions.
> Needless to say, this could have bad consequences for test users. A
> developer might press control-C in the middle of a unit test, and then find
> that he was unable to successfully re-run that test again because of the
> lingering state. A Jenkins build machine might experience a JVM crash
> because of an out-of-memory condition, and then find that all builds turned
> red because a unit test failed to clean up the state which it created.
> I think that in HDFS tests which alter persistent state, we probably should
> invoke the @After functions from a @Before function, just to make sure that
> the necessary cleanup is done. Otherwise, we might be risking unrepeatable
> test failures.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
