[
https://issues.apache.org/jira/browse/HADOOP-13801?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15691576#comment-15691576
]
Mingliang Liu commented on HADOOP-13801:
----------------------------------------
{{ContractTestUtils#readBytesToString()}} calls the {{toChar()}} for each char
of the read buffer, which escapes chars like tabs and lines using
{{String.format("%02x", b);}}. Unfortunately the
{{ITestS3AMiniYarnCluster#getResultAsMap}} relies on the newline and tab chars
for splitting tokens, which will only see 09 for tab, 0a for newline in the
escaped string. Actually we don't need to escape; so a simply {{new
String(buf)}} will work just fine.
> regression: ITestS3AMiniYarnCluster failing
> -------------------------------------------
>
> Key: HADOOP-13801
> URL: https://issues.apache.org/jira/browse/HADOOP-13801
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 2.8.0
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Fix For: 2.8.0, 3.0.0-alpha2
>
> Attachments: HADOOP-13801-branch-2-001.patch
>
>
> {{ITestS3AMiniYarnCluster}} is failing with an
> {{ArrayIndexOutOfBoundsException}} splitting up the output of the wordcount
> app. It's not getting word\t\count lines back.
> * git bisect points at commit 67e01f7; HADOOP-13614. Purge some
> superfluous/obsolete S3 FS tests that are slowing test runs down
> * problem exists when running standalone as well as in mvn parallel tests;
> this implies it's not parallelisation
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
