[jira] [Commented] (REEF-484) Implement a LocalRuntimeStartHandler to enable restarts on LocalRuntime

["Andrew Chung (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/REEF-484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14634242#comment-14634242 ] 

Andrew Chung commented on REEF-484:
-----------------------------------

This is a work item of larger scope than AM HA on YARN. Currently it is not possible to test restarts on LocalRuntime for the following reason:
LocalRuntime starts the driver process first, which instantiates the local ResourceManager, which in turn starts Evaluator processes as child processes of the driver process. From here, there are two scenarios that can happen: if the driver is force-terminated (e.g. via kill command), child processes may be orphaned, but the ResourceManager dies with it, so no context is kept. If the driver terminates "regularly" by throwing an exception, child processes will be killed by local runtime cleanup. Either way, context is lost.

> Implement a LocalRuntimeStartHandler to enable restarts on LocalRuntime
> -----------------------------------------------------------------------
>
>                 Key: REEF-484
>                 URL: https://issues.apache.org/jira/browse/REEF-484
>             Project: REEF
>          Issue Type: Sub-task
>          Components: REEF Driver, REEF.NET Driver
>            Reporter: Andrew Chung
>
> Currently the LocalRuntime does not have a driver runtime start handler that handles the restart logic. We will need this in order to test out the driver restart feature on local runtime.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)