[
https://issues.apache.org/jira/browse/FLINK-6379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16066083#comment-16066083
]
ASF GitHub Bot commented on FLINK-6379:
---------------------------------------
Github user tillrohrmann commented on the issue:
https://github.com/apache/flink/pull/3942
Really good work @EronWright. I've completed the review and rebased it on
the latest master. While doing the review I applied one fix for a race
condition in the tests, another fix to address the newly introduced checkstyle
rules and one small change which removes the `MesosResourceManager's` internal
RPC calls from the public interface, because they should only be needed by the
RM's internal components. All small things. I really liked how you wrote the
tests :-)
Once Travis gives green light, I will merge your PR.
As a follow up, I noticed two things:
1. Proper shutdown of internal components
2. Getting rid (at some point) of Akka dependency
> Implement FLIP-6 Mesos Resource Manager
> ---------------------------------------
>
> Key: FLINK-6379
> URL: https://issues.apache.org/jira/browse/FLINK-6379
> Project: Flink
> Issue Type: Sub-task
> Components: Mesos
> Reporter: Eron Wright
> Assignee: Eron Wright
>
> Given the new ResourceManager of FLIP-6, implement a new
> MesosResourceManager.
> The minimal effort would be to implement a new resource manager while
> continuing to use the various local actors (launch coordinator, task monitor,
> etc.) which implement the various FSMs associated with Mesos scheduling.
> The Fenzo library would continue to solve the packing problem of matching
> resource offers to slot requests.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
