[
https://issues.apache.org/jira/browse/DRILL-1431?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14259753#comment-14259753
]
Jacques Nadeau commented on DRILL-1431:
---------------------------------------
Ideally this should be handled. However, fixing out of memory is more
important.
> Drillbit continue to report positive status after hitting out of memory
> condition
> ---------------------------------------------------------------------------------
>
> Key: DRILL-1431
> URL: https://issues.apache.org/jira/browse/DRILL-1431
> Project: Apache Drill
> Issue Type: Bug
> Components: Execution - Flow
> Reporter: Norris Lee
> Priority: Minor
> Fix For: Future
>
>
> Over time, a drillbit tends to hang. Eg. If a drillbit runs out of memory,
> the drillbit will still be in the list of drillbits returned by zookeeper.
> Since Drill Client randomly selects a drillbit from that list, the drillbit
> it connects to may be unhealthy. Drill Client should do a quick connection
> check to determine whether drillbit is actually healthy. If not, it should
> select another drillbit and test again until it has found a healthy drillbit.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
