[ 
https://issues.apache.org/jira/browse/APEXCORE-619?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15933473#comment-15933473
 ] 

ASF GitHub Bot commented on APEXCORE-619:
-----------------------------------------

GitHub user tushargosavi opened a pull request:

    https://github.com/apache/apex-core/pull/490

    APEXCORE-619 Checkpoint leaf stateless operator for correct recovery after 
application restart.

    

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/tushargosavi/apex-core 
APEXCORE-619_wrong_recovery_id

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/apex-core/pull/490.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #490
    
----
commit 853405c653f73f32cf708dab9a9959993796ad3f
Author: Tushar R. Gosavi <[email protected]>
Date:   2017-03-20T18:21:12Z

    APEXCORE-619 Checkpoint leaf stateless operator for correct recovery after 
application restart.

----


> recovery window id in future for terminating state less operators during 
> relaunch.
> ----------------------------------------------------------------------------------
>
>                 Key: APEXCORE-619
>                 URL: https://issues.apache.org/jira/browse/APEXCORE-619
>             Project: Apache Apex Core
>          Issue Type: Bug
>            Reporter: Tushar Gosavi
>            Assignee: Tushar Gosavi
>            Priority: Minor
>
> With following DAG
> A -> B -> C
> C is a stateless operator. If this application is killed and restarted after 
> long time between kill and restart, then recovery window id of C is too high 
> compare to A and B. This is because recovery windowid is computed from 
> current timestamp for stateless operators in updateRecoveryCheckpoints.
> The problem this causes 
> - Operator C does not process any data till windowId of B reached to recovery 
> window id of C.
> - If other operators are not able to keep up then C gets killed because it is 
> detected as blocked operator.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to