[jira] [Commented] (DRILL-1253) Inserting key into hash table and updating workspace variable in aggregate container should be an atomic operation

["Jacques Nadeau (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14088795#comment-14088795 ] 

Jacques Nadeau commented on DRILL-1253:
---------------------------------------

Resolved by 8c66525 or earlier.

> Inserting key into hash table and updating workspace variable in aggregate container should be an atomic operation
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: DRILL-1253
>                 URL: https://issues.apache.org/jira/browse/DRILL-1253
>             Project: Apache Drill
>          Issue Type: Bug
>            Reporter: Mehant Baid
>             Fix For: 0.5.0
>
>         Attachments: DRILL-1253.patch
>
>
> Currently the logic in HashAggregate and HashTable assumes that if the insertion of the key into the hash table is successful then updating the workspace variables in the aggregate container will also be successful. 
> If the insertion of the key into the hash table fails we add a new batch in the hash table and retry inserting the key. However we don't check for failures while updating the workspace variables. 
> Problem arises when the insertion of the key into the hash table succeeds but inserting the workspace variable fails.



--
This message was sent by Atlassian JIRA
(v6.2#6252)