[ 
https://issues.apache.org/jira/browse/YARN-1994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14078272#comment-14078272
 ] 

Craig Welch commented on YARN-1994:
-----------------------------------

Milan
At present multi-home only works if all interfaces on a box have the same 
names, managing which address a client will connect to has to be managed by 
controlling what address this will resolve to for a particular class of 
clients.  This is necessary because there are cases where links and redirects 
are generated based on names, and for this to operate for all clients on all 
networks the names for the hadoop hosts must be the same everywhere.  For it to 
work in any other way would require logic to use a particular name depending on 
the source network of the client when generating links, and that is not in the 
current scope (there would be other complexity around managing multiple names 
for the same host/service as well, which would be problematic).  Since the only 
way for multi-home to work properly at this point is for the host to have the 
same name on all networks it can be accessed from the additional logic is 
unnecessary - when properly configured, it will always return the same name.  
The same is true for the container manager impl, mrclient service, task 
attempt, etc

> Expose YARN/MR endpoints on multiple interfaces
> -----------------------------------------------
>
>                 Key: YARN-1994
>                 URL: https://issues.apache.org/jira/browse/YARN-1994
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: nodemanager, resourcemanager, webapp
>    Affects Versions: 2.4.0
>            Reporter: Arpit Agarwal
>            Assignee: Craig Welch
>         Attachments: YARN-1994.0.patch, YARN-1994.1.patch, 
> YARN-1994.11.patch, YARN-1994.11.patch, YARN-1994.12.patch, 
> YARN-1994.2.patch, YARN-1994.3.patch, YARN-1994.4.patch, YARN-1994.5.patch, 
> YARN-1994.6.patch, YARN-1994.7.patch
>
>
> YARN and MapReduce daemons currently do not support specifying a wildcard 
> address for the server endpoints. This prevents the endpoints from being 
> accessible from all interfaces on a multihomed machine.
> Note that if we do specify INADDR_ANY for any of the options, it will break 
> clients as they will attempt to connect to 0.0.0.0. We need a solution that 
> allows specifying a hostname or IP-address for clients while requesting 
> wildcard bind for the servers.
> (List of endpoints is in a comment below)



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to