[
https://issues.apache.org/jira/browse/HDFS-8063?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14484520#comment-14484520
]
Colin Patrick McCabe commented on HDFS-8063:
--------------------------------------------
Thanks for looking at this, [~iwasakims]. It's nice that you refactored the
common code into the {{writeTestFile}} function.
I do wonder if we could avoid shutting down the MiniDFSCluster on every test.
Did you try just doing the hsync by itself? That would cause the background
write thread to have to fire and complete. Shutting down the whole cluster
shouldn't be necessary to ensure that the write completed, surely? The stuff
the NameNode is doing is not really done in the background, so I'm not sure it
needs a shutdown. Maybe I am missing something?
> Fix test failure in TestTracing
> -------------------------------
>
> Key: HDFS-8063
> URL: https://issues.apache.org/jira/browse/HDFS-8063
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: test
> Reporter: Masatake Iwasaki
> Assignee: Masatake Iwasaki
> Priority: Minor
> Attachments: HDFS-8063.001.patch
>
>
> Tests in TestTracing sometimes fails, especially on slow machine. The cause
> is that spans is possible to arrive at receiver after
> {{assertSpanNamesFound}} passed and
> {{SetSpanReceiver.SetHolder.spans.clear()}} is called for next test case.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
