[jira] [Updated] (AMBARI-4223) Cluster topology is hard-coded

["Nate Cole (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/AMBARI-4223?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Nate Cole updated AMBARI-4223:
------------------------------

    Attachment: AMBARI-4223.patch

> Cluster topology is hard-coded
> ------------------------------
>
>                 Key: AMBARI-4223
>                 URL: https://issues.apache.org/jira/browse/AMBARI-4223
>             Project: Ambari
>          Issue Type: Task
>          Components: controller
>            Reporter: Nate Cole
>            Assignee: Nate Cole
>             Fix For: 1.5.0
>
>         Attachments: AMBARI-4223.patch
>
>
> The cluster topology is passed to agents during setup and configuration.  This happens via {{StageUtils}} using a static map like so:
> {code}
>     componentToClusterInfoKeyMap.put("NAMENODE", "namenode_host");
>     componentToClusterInfoKeyMap.put("JOBTRACKER", "jtnode_host");
>     componentToClusterInfoKeyMap.put("SNAMENODE", "snamenode_host");
>     componentToClusterInfoKeyMap.put("RESOURCEMANAGER", "rm_host");
> {code}
> At runtime, we determine the hosts for each role, and pass them down.  When wiring up new services, any role missing from this map will not be sent to the agent, when that information may be required.
> Proposal is to maintain this map for backwards compatibility, and pass any undefined master/slave components as "<rolename>_hosts"  Client topology need not be passed.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)