[ 
https://issues.apache.org/jira/browse/FLINK-9634?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16524789#comment-16524789
 ] 

ASF GitHub Bot commented on FLINK-9634:
---------------------------------------

Github user tillrohrmann commented on a diff in the pull request:

    https://github.com/apache/flink/pull/6208#discussion_r198420806
  
    --- Diff: 
flink-runtime/src/main/java/org/apache/flink/runtime/jobmaster/JobMaster.java 
---
    @@ -280,12 +281,7 @@ public JobMaster(
     
                resourceManagerLeaderRetriever = 
highAvailabilityServices.getResourceManagerLeaderRetriever();
     
    -           this.slotPool = new SlotPool(
    -                   rpcService,
    -                   jobGraph.getJobID(),
    -                   SystemClock.getInstance(),
    -                   rpcTimeout,
    -                   jobMasterConfiguration.getSlotIdleTimeout());
    +           this.slotPool = 
checkNotNull(slotPoolFactory).createSlotPool(jobGraph.getJobID());
    --- End diff --
    
    The reason why I introduced the factory is that we want to create a 
`SlotPool` for the `JobID` of the given `JobGraph`. That way we enforce that 
the `SlotPool` is strictly instantiated with the right `JobID`. If we move the 
instantiation of the `SlotPool` out of the `JobMaster` we would have to rely on 
the outside code. Therefore, I would like to keep it like it is.


> Deactivate previous location based scheduling if local recovery is disabled
> ---------------------------------------------------------------------------
>
>                 Key: FLINK-9634
>                 URL: https://issues.apache.org/jira/browse/FLINK-9634
>             Project: Flink
>          Issue Type: Bug
>          Components: Distributed Coordination
>    Affects Versions: 1.5.0
>            Reporter: Till Rohrmann
>            Assignee: Till Rohrmann
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.6.0, 1.5.1
>
>
> With Flink 1.5.0 we introduced local recovery. In order to make local 
> recovery work we had to change the scheduling to be aware of the previous 
> location of the {{Execution}}. This scheduling strategy is also active if 
> local recovery is deactivated. I suggest to also disable the scheduling 
> strategy if local recovery is not enabled.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to