drill-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (DRILL-1942) Improve off-heap memory usage tracking
Date Fri, 11 Sep 2015 21:58:45 GMT

    [ https://issues.apache.org/jira/browse/DRILL-1942?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14741601#comment-14741601
] 

ASF GitHub Bot commented on DRILL-1942:
---------------------------------------

Github user adeneche commented on a diff in the pull request:

    https://github.com/apache/drill/pull/133#discussion_r39322427
  
    --- Diff: exec/java-exec/src/main/java/org/apache/drill/exec/physical/impl/join/HashJoinBatch.java
---
    @@ -481,30 +466,28 @@ public HashJoinProbe setupHashJoinProbe() throws ClassTransformationException,
I
               outputType = inputType;
             }
     
    -        ValueVector v = container.addOrGet(MaterializedField.create(vv.getField().getPath(),
outputType));
    +        final ValueVector v = container.addOrGet(MaterializedField.create(vv.getField().getPath(),
outputType));
             if (v instanceof AbstractContainerVector) {
               vv.getValueVector().makeTransferPair(v);
               v.clear();
             }
     
    -        JVar inVV = g.declareVectorValueSetupAndMember("probeBatch", new TypedFieldId(inputType,
false, fieldId));
    -        JVar outVV = g.declareVectorValueSetupAndMember("outgoing", new TypedFieldId(outputType,
false, outputFieldId));
    +        final JVar inVV = g.declareVectorValueSetupAndMember("probeBatch", new TypedFieldId(inputType,
false, fieldId));
    +        final JVar outVV = g.declareVectorValueSetupAndMember("outgoing", new TypedFieldId(outputType,
false, outputFieldId));
     
             g.getEvalBlock().add(outVV.invoke("copyFromSafe").arg(probeIndex).arg(outIndex).arg(inVV));
     
             fieldId++;
             outputFieldId++;
           }
    -      recordCount = left.getRecordCount();
         }
     
    -    HashJoinProbe hj = context.getImplementationClass(cg);
    -
    +    final HashJoinProbe hj = context.getImplementationClass(cg);
    --- End diff --
    
    you can also change this to:
    ```
    return context.getImplementationClass(cg);
    ```


> Improve off-heap memory usage tracking
> --------------------------------------
>
>                 Key: DRILL-1942
>                 URL: https://issues.apache.org/jira/browse/DRILL-1942
>             Project: Apache Drill
>          Issue Type: Improvement
>          Components: Execution - Relational Operators
>            Reporter: Chris Westin
>            Assignee: Chris Westin
>             Fix For: 1.2.0
>
>         Attachments: DRILL-1942.1.patch.txt, DRILL-1942.2.patch.txt, DRILL-1942.3.patch.txt
>
>
> We're using a lot more memory than we think we should. We may be leaking it, or not releasing
it as soon as we could. 
> This is a call to come up with some improved tracking so that we can get statistics out
about exactly where we're using it, and whether or not we can release it earlier.



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

Mime
View raw message