[
https://issues.apache.org/jira/browse/HADOOP-12081?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14633408#comment-14633408
]
Steve Loughran commented on HADOOP-12081:
-----------------------------------------
I'm always worried about going near UGI, but this looks low risk. it extends
the 64 bit string comparison with one for the z390 tag, which would only cause
confusion with an OS with the same tag on a different width architecture.
+1
> Support Hadoop on zLinux - fix JAAS authentication issue
> --------------------------------------------------------
>
> Key: HADOOP-12081
> URL: https://issues.apache.org/jira/browse/HADOOP-12081
> Project: Hadoop Common
> Issue Type: Bug
> Components: security
> Affects Versions: 2.6.0
> Environment: zLinux
> Reporter: Adam Roberts
> Assignee: Akira AJISAKA
> Labels: zlinux
> Attachments: HADOOP-12081.001.patch
>
>
> Currently the 64 bit check in security/UserGroupInformation.java uses os.arch
> and checks for "64". s390x is returned on IBM's z platform: s390x is 64 bit.
> Without this change, if we try to use HDFS with Spark, we get a fatal error
> (unable to login as we can't find a login class).
> This address fixes said issue by identifying s390x as a 64 bit platform and
> thus allowing Spark to run on zLinux. A simple fix with very big implications!
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
