[
https://issues.apache.org/jira/browse/MAPREDUCE-4168?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14261268#comment-14261268
]
Steve Loughran commented on MAPREDUCE-4168:
-------------------------------------------
bq. The biggest problem that we have in Hadoop is that we use the server-side
network configuration as the same information we pass to or expect clients to
use for connection information. In multi-home situations, these may not be (and
usually aren't) the same. So if a service is configured to use 0.0.0.0 (to bind
to all interfaces), the client can't use that to connect to it since that
doesn't resolve to anything usable.
The issue here is the expectation that clients will magically get valid
-site.xml files, where the definition of valid is : valid at time {{t}} for
client {{c}} to submit jobs to cluster {{h}}. The configurations used within
cluster {{h}} may not be valid for client {{c}} even for the same time value
{{t}}. This is why, although you can point your browser at
http://resourcemanager:8090/config, you have no guarantees of getting a usable
config.
Allen, if you look closely at YARN-896 you'll see that the registry can live
standalone, and has special space reserved {{/services}} for us to publish
service records for system-wide services, service records that could include
the URL to an API which lets you enum configs and retrieve them. Which is
precisely what we offer in Slider for pushing out dynamically generated
hbase-site.xml, accumulo-site.xml, etc.
The fact that there is effectively a placeholder for publishing core-site XML
&c is not a coincidence. It's something we need.
> Support multiple network interfaces
> -----------------------------------
>
> Key: MAPREDUCE-4168
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-4168
> Project: Hadoop Map/Reduce
> Issue Type: New Feature
> Reporter: Tom White
>
> Umbrella jira to track the MapReduce side of HADOOP-8198.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
