[
https://issues.apache.org/jira/browse/HADOOP-3619?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14066506#comment-14066506
]
Allen Wittenauer commented on HADOOP-3619:
------------------------------------------
A few comments:
* I'm not a big fan of this error message:
bq. + throw new NamingException("hostIp is not
Inet4Address or Inet6Address. Other Addresses are not supported actually");
Perhaps it should be "Unrecognized IP address format: xyz", where xyz is the
address we couldn't parse?
* It might be a good idea to see if this code throws an exception when given a
broken address format in the test. (I'm a big fan of negative tests!)
* What happens on a machine with IPv6 disabled? Do the tests still pass?
* I'm a little uncomfortable doing DNS tests against an actual host. It looks
like the test code for IPv4 is looking at localhost. We should probably do the
same thing for IPv6.
> DNS.getHosts triggers an ArrayIndexOutOfBoundsException in reverseDNS if one
> of the interfaces is IPv6
> ------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-3619
> URL: https://issues.apache.org/jira/browse/HADOOP-3619
> Project: Hadoop Common
> Issue Type: Bug
> Reporter: Steve Loughran
> Labels: patch
> Attachments: HADOOP-3619-v2.patch
>
>
> reverseDNS tries to split a host address string by ".", and so fails if ":"
> is the separator, as it is in IPv6. When it tries to access the parts of the
> address, a stack trace is seen.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
