[
https://issues.apache.org/jira/browse/MAPREDUCE-3451?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13159619#comment-13159619
]
Patrick Wendell commented on MAPREDUCE-3451:
--------------------------------------------
I have a few questions about the way that schedulers are implemented in MR2.
It's not clear to me whether the Scheduler class used by the Resource Manager
is responsible for only inter-application scheduling or both inter- ant
intra-application scheduling. It looks to me like only the former is supported,
because the resource requests (via allocate()) do not give any meta-data
regarding the user initiating the request. So if we wanted to enforce Fair
Sharing for jobs requested from a single AM, this doesn't seem possible. When I
look at the Capacity Scheduler, for instance, it seems like allocation requests
are assigned to queues based only on the application from which they are made,
not the actual user launching the job.
Is the idea that, in a shared cluster, each administrative unit has its own MR
Application Master running?
I have some follow-up but first want to see if this is actually a correct
interpretation of the design spec.
> Port Fair Scheduler to MR2
> --------------------------
>
> Key: MAPREDUCE-3451
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3451
> Project: Hadoop Map/Reduce
> Issue Type: New Feature
> Components: scheduler
> Reporter: Patrick Wendell
> Assignee: Patrick Wendell
>
> The Fair Scheduler is in widespread use today in MR1 clusters, but not yet
> ported to MR2. This is to track the porting of the Fair Scheduler to MR2 and
> will be updated to include design considerations and progress.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
