[
https://issues.apache.org/jira/browse/HADOOP-11630?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14633688#comment-14633688
]
Allen Wittenauer commented on HADOOP-11630:
-------------------------------------------
bq. Hadoop source will need fixing for IPv6; supporting the scripts is step
one.
I disagree. I'd rather have working Java code first, given that (as pointed out
above) it's trivial to enable IPv6 in the scripts. They were intentionally
written this way in trunk to prevent someone from accidentally turning it on
and wondering why everything broke while still giving any developers wanted to
work on that functionality a way out.
Given that the last I heard, we weren't sure if the changes required to make
IPv6 work were going to be incompatible or not, it seems to be premature to do
a script change for branch-2.
> Allow hadoop.sh to bind to ipv6 conditionally
> ---------------------------------------------
>
> Key: HADOOP-11630
> URL: https://issues.apache.org/jira/browse/HADOOP-11630
> Project: Hadoop Common
> Issue Type: Improvement
> Components: scripts
> Affects Versions: 2.6.0
> Reporter: Elliott Clark
> Assignee: Elliott Clark
> Labels: ipv6
> Attachments: HDFS-7834-branch-2-0.patch, HDFS-7834-trunk-0.patch
>
>
> Currently the bash scripts unconditionally add -Djava.net.preferIPv4Stack=true
> While this was needed a while ago. IPV6 on java works much better now and
> there should be a way to allow it to bind dual stack if needed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
