Naganarasimha G R commented on YARN-4494:

bq. One reason why the app recovery is synchronous is that asynchronous 
recovery can cause the RM to tell a client that a job doesn't exist, when it 
really just hasn't been recovered yet, which is an issue even with completed 
jobs. How are you planning to handle that?
Well, yes this will be the limitation of the solution but when timelineservice 
is enabled then if the data is not there in RM then it tries to mitigate the 
issue. Also when we compare the 2 evils *RM fail over being slower* and *RM to 
tell a client that a job doesn't exist momentarily* i feel former is more 
serious and later can be avoided using ATS. 

> Recover completed apps asynchronously
> -------------------------------------
>                 Key: YARN-4494
>                 URL: https://issues.apache.org/jira/browse/YARN-4494
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: resourcemanager
>            Reporter: Jun Gong
>            Assignee: Jun Gong
> With RM HA enabled, when recovering apps, recover completed apps 
> asynchronously.

This message was sent by Atlassian JIRA

Reply via email to