[jira] [Resolved] (IMPALA-4660) KuduScanNode may not terminate scanners when parent nodes reach limit

["Sailesh Mukil (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IMPALA-4660?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sailesh Mukil resolved IMPALA-4660.
-----------------------------------
    Resolution: Cannot Reproduce

> KuduScanNode may not terminate scanners when parent nodes reach limit
> ---------------------------------------------------------------------
>
>                 Key: IMPALA-4660
>                 URL: https://issues.apache.org/jira/browse/IMPALA-4660
>             Project: IMPALA
>          Issue Type: Bug
>          Components: Backend
>    Affects Versions: Impala 2.8.0
>            Reporter: Matthew Jacobs
>            Assignee: Sailesh Mukil
>              Labels: kudu
>
> Related to IMPALA-4654 where the limit on a KuduScanNode was not respected and scanner threads would continue to completion.
> There may be an additional bug if a limit is on another node. We do not currently have a repro for this but believe there could be an issue because the KuduScanner doesn't check the scan node's {{done_}} flag, or have some other way of being notified of cancellation except by checking the runtime state's status which may not be set in this case.
> This is based on a discussion in this CR, see for context:
> https://gerrit.cloudera.org/#/c/5493/



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