[
https://issues.apache.org/jira/browse/FLINK-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15936909#comment-15936909
]
ASF GitHub Bot commented on FLINK-5890:
---------------------------------------
Github user greghogan commented on the issue:
https://github.com/apache/flink/pull/3515
Not an issue of spilling memory but at least three elements are required to
trigger two reduces and the error condition depends on which value is returned.
@StephanEwen is this a +1?
> GatherSumApply broken when object reuse enabled
> -----------------------------------------------
>
> Key: FLINK-5890
> URL: https://issues.apache.org/jira/browse/FLINK-5890
> Project: Flink
> Issue Type: Bug
> Components: Gelly
> Affects Versions: 1.3.0
> Reporter: Greg Hogan
> Assignee: Greg Hogan
> Fix For: 1.3.0
>
>
> {{GatherSumApplyIteration.SumUdf.reduce}} can store a value from {{arg1}} in
> the new {{Tuple2}} which can be overwritten in {{ReduceDriver}}. We need to
> swap {{arg0.f1}} and {{arg1.f1}} when this happens (as done in
> {{ReduceDriver}} for the returned results).
> {code}
> @Override
> public Tuple2<K, M> reduce(Tuple2<K, M> arg0, Tuple2<K, M> arg1) throws
> Exception {
> K key = arg0.f0;
> M result = this.sumFunction.sum(arg0.f1, arg1.f1);
> return new Tuple2<>(key, result);
> }
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
