hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vinod K V (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-5396) Queue ACLs should be refreshed without requiring a restart of the job tracker
Date Fri, 13 Mar 2009 14:21:50 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-5396?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12681728#action_12681728
] 

Vinod K V commented on HADOOP-5396:
-----------------------------------

Going into the details:
 - JT implements a AdminOperationsProtocol which will encompass all admin related RPCs. Currently
it only has a refreshQueues RPC. RefreshAuthoroziationPolicy cannot be used for this as it
is a public interface that serves an independant purpose and used both in JT as well as NN.
 {code}
     public class jobTracker implements AdminOperationsProtocol ... {
         .....
         .....
        public void refreshQueues() {
            // reload the mapred-queues.xml
        }
     }
 {code}
 - Adding another option to MRAdmin to refresh queue properties.
 - Moving mapred.queue.names, mapred.acls, all queue acl related properties moved to mapred-queues.xml

One issue with the above: We want the AdminOperationsProtocol to be a package private interface
so that it can evolve over time. But MRAdmin is in mapred.tools subpackage and so cannot access
this protocl unless it is is public. One way to resolve this is by moving MRAdmin tool into
the mapred package and deprecating the current mapred.tools.MRAdmin. Thoughts?

> Queue ACLs should be refreshed without requiring a restart of the job tracker
> -----------------------------------------------------------------------------
>
>                 Key: HADOOP-5396
>                 URL: https://issues.apache.org/jira/browse/HADOOP-5396
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: mapred
>            Reporter: Hemanth Yamijala
>            Assignee: Vinod K V
>
> In large shared deployments of the M/R clusters, it is normal that new users will periodically
want to get access to some queues on the M/R framework. Requiring a JT restart for each such
change is operationally inconvenient and seems an overkill. There should be a way for updating
ACLs with new users without requiring a JT restart.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message