[ 
https://issues.apache.org/jira/browse/FLINK-8360?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16313141#comment-16313141
 ] 

ASF GitHub Bot commented on FLINK-8360:
---------------------------------------

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

    https://github.com/apache/flink/pull/5239#discussion_r159836477
  
    --- Diff: 
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/Execution.java
 ---
    @@ -145,8 +145,8 @@
     
        private volatile Throwable failureCause;          // once assigned, 
never changes
     
    -   /** The handle to the state that the task gets on restore */
    -   private volatile TaskStateSnapshot taskState;
    +   /** Information to restore the task onr recovery, such as checkpoint id 
and task state snapshot */
    +   private volatile TaskRestore taskRestore;
    --- End diff --
    
    I would add `SerializableOptional` class, make this field 
`SerializableOptional<TaskRestore>` and pass it down to `Task` to avoid nulls. 
    
    If you do not like this, please at least add `@Nullable` annotations to 
this field and subsequent usages.
    
    Or maybe embed `isEmpty` logic into `TaskRestore`.
      


> Implement task-local state recovery
> -----------------------------------
>
>                 Key: FLINK-8360
>                 URL: https://issues.apache.org/jira/browse/FLINK-8360
>             Project: Flink
>          Issue Type: New Feature
>          Components: State Backends, Checkpointing
>            Reporter: Stefan Richter
>            Assignee: Stefan Richter
>             Fix For: 1.5.0
>
>
> This issue tracks the development of recovery from task-local state. The main 
> idea is to have a secondary, local copy of the checkpointed state, while 
> there is still a primary copy in DFS that we report to the checkpoint 
> coordinator.
> Recovery can attempt to restore from the secondary local copy, if available, 
> to save network bandwidth. This requires that the assignment from tasks to 
> slots is as sticky is possible.
> For starters, we will implement this feature for all managed keyed states and 
> can easily enhance it to all other state types (e.g. operator state) later.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to