drill-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jacques Nadeau (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (DRILL-2901) In some situations, a limit query causes fragment executors to get into bad state
Date Wed, 29 Apr 2015 16:08:09 GMT

    [ https://issues.apache.org/jira/browse/DRILL-2901?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14519632#comment-14519632
] 

Jacques Nadeau commented on DRILL-2901:
---------------------------------------

[~rkins], unfortunately, it isn't something that is easily reproducible except at scale. 
These were discovered while working with a large customer with proprietary data.  To repro,
you need to run a large cluster with a large number of files and then use a limit 0 or limit
<small number> query to experience these issues.  They all have to do with a distributed
race between starting a fragment and getting a cancellation too early in that process.

> In some situations, a limit query causes fragment executors to get into bad state
> ---------------------------------------------------------------------------------
>
>                 Key: DRILL-2901
>                 URL: https://issues.apache.org/jira/browse/DRILL-2901
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Execution - Flow
>            Reporter: Jacques Nadeau
>            Assignee: Jacques Nadeau
>             Fix For: 0.9.0
>
>         Attachments: DRILL-2901.patch
>
>




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message