[
https://issues.apache.org/jira/browse/FLINK-36779?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17900353#comment-17900353
]
Roman Boyko commented on FLINK-36779:
-------------------------------------
It can be easily fixed. Please assign this bug to me.
> topn.cache.size metric is incorrect and non-changing during the time
> ---------------------------------------------------------------------
>
> Key: FLINK-36779
> URL: https://issues.apache.org/jira/browse/FLINK-36779
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Runtime
> Reporter: Roman Boyko
> Priority: Minor
>
> AbstractTopNFunction#registerMetric uses 'long heapSize' to create the metric
> 'topn.cache.size' and set its value. But the heapSize is passed by value so
> it remains the same during the operator life.
> It's resulting to non-changing metric value for UpdatableTopNFunction and 0
> metric value for AppendOnlyTopNFunction, FastTop1Function and
> AppendOnlyFirstNFunction.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
