[
https://issues.apache.org/jira/browse/HADOOP-6364?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12779489#action_12779489
]
Steve Loughran commented on HADOOP-6364:
----------------------------------------
# It might make sense to have options for the namenode and job tracker,
something with different config names for each, so a single configuration can
define both.
# It could still be important for some people to say "come up on all
interfaces, localhost included", which could imply it should be a list.
# There's also the problem that the namenode has historically been very fussy
about what hostname was used to refer to it
> add a 'hostname' or 'identity' parameter
> ----------------------------------------
>
> Key: HADOOP-6364
> URL: https://issues.apache.org/jira/browse/HADOOP-6364
> Project: Hadoop Common
> Issue Type: Improvement
> Reporter: Allen Wittenauer
>
> There was a bit of talk about HDFS-34 last night at Apachecon. One of the
> points brought up was the difficulty in assuming that the 'hostname' was
> externally available. Perhaps what needs to happen instead is to follow what
> httpd and a few other apps do. That is provide a way for the service to know
> what hostname to advertise itself as and use in all communications. This
> could also help solve the multi-nic problems (HADOOP-6210) faced when using
> hadoop in a HA environment.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
