[
https://issues.apache.org/jira/browse/FLINK-18453?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17154497#comment-17154497
]
Jark Wu edited comment on FLINK-18453 at 7/9/20, 12:27 PM:
-----------------------------------------------------------
It seems that the generated code is too large. What about to reduce the number
of aggregates and use a smaller {{TableConfig#setMaxGeneratedCodeLength(?)}}
(but not {{1}} length)? [~libenchao]
was (Author: jark):
It seems that the generated code is too large. What about to reduce the number
of aggregates and use a smaller {{TableConfig#setMaxGeneratedCodeLength(??)}}
(but not {{1}} length)[~libenchao]?
> Stack overflow of AggregateITCase#testAggregationCodeSplit on Azure
> --------------------------------------------------------------------
>
> Key: FLINK-18453
> URL: https://issues.apache.org/jira/browse/FLINK-18453
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner, Table SQL / Runtime
> Reporter: Yingjie Cao
> Priority: Major
> Labels: test-stability
>
> Here is some log:
> ```
> [ERROR] testAggregationCodeSplit[LocalGlobal=OFF, MiniBatch=OFF,
> StateBackend=HEAP](org.apache.flink.table.planner.runtime.stream.sql.AggregateITCase)
> Time elapsed: 8.167 s <<< ERROR!
> Caused by: java.lang.StackOverflowError
> at org.codehaus.janino.CodeContext.extract16BitValue(CodeContext.java:720)
> at org.codehaus.janino.CodeContext.flowAnalysis(CodeContext.java:561)
> at org.codehaus.janino.CodeContext.flowAnalysis(CodeContext.java:557)
> ```
> The whole log:
> [https://dev.azure.com/kevin-flink/flink/_build/results?buildId=30&view=logs&j=a1590513-d0ea-59c3-3c7b-aad756c48f25&t=5129dea2-618b-5c74-1b8f-9ec63a37a8a6]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
