You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@drill.apache.org by "Paul Rogers (JIRA)" <ji...@apache.org> on 2017/07/11 20:31:00 UTC

[jira] [Updated] (DRILL-5310) Memory leak in managed sort if OOM during sv2 allocation

     [ https://issues.apache.org/jira/browse/DRILL-5310?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Paul Rogers updated DRILL-5310:
-------------------------------
    Fix Version/s:     (was: 1.11.0)
                   1.12.0

> Memory leak in managed sort if OOM during sv2 allocation
> --------------------------------------------------------
>
>                 Key: DRILL-5310
>                 URL: https://issues.apache.org/jira/browse/DRILL-5310
>             Project: Apache Drill
>          Issue Type: Sub-task
>    Affects Versions: 1.10.0
>            Reporter: Paul Rogers
>            Assignee: Paul Rogers
>             Fix For: 1.12.0
>
>
> See the "identical1" test case in DRILL-5266. Due to misconfiguration, the sort was given too little memory to make progress. An OOM error occurred when allocating an SV2.
> In this scenario, the "converted" record batch is leaked.
> Normally, a converted batch is added to the list of in-memory batches, then released on {{close()}}. But, in this case, the batch is only a local variable, and so leaks.
> The code must release this batch in this condition.



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