[jira] [Commented] (FLINK-4165) Event-Time CEP Job Sometimes Fails After Restart

["Aljoscha Krettek (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-4165?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15366019#comment-15366019 ] 

Aljoscha Krettek commented on FLINK-4165:
-----------------------------------------

I found out that this happens if the event type, i.e. the type of elements in your event stream does not have proper implementations of `equals()`/`hashCode()`. I will add a warning to the doc that events must fulfill this requirement.

> Event-Time CEP Job Sometimes Fails After Restart
> ------------------------------------------------
>
>                 Key: FLINK-4165
>                 URL: https://issues.apache.org/jira/browse/FLINK-4165
>             Project: Flink
>          Issue Type: Bug
>          Components: CEP
>    Affects Versions: 1.0.0, 1.1.0, 1.0.1, 1.0.2, 1.0.3
>            Reporter: Aljoscha Krettek
>            Assignee: Aljoscha Krettek
>            Priority: Blocker
>
> This occurs sometimes after FLINK-4162 is fixed. About 40 % of the time, when restarting a job it fails with this exception:
> {code}
> java.lang.RuntimeException: Failure happened in filter function.
> 	at org.apache.flink.cep.nfa.NFA.computeNextStates(NFA.java:316)
> 	at org.apache.flink.cep.nfa.NFA.process(NFA.java:160)
> 	at org.apache.flink.cep.operator.KeyedCEPPatternOperator.processEvent(KeyedCEPPatternOperator.java:48)
> 	at org.apache.flink.cep.operator.AbstractKeyedCEPPatternOperator.processWatermark(AbstractKeyedCEPPatternOperator.java:167)
> 	at org.apache.flink.streaming.runtime.io.StreamInputProcessor.processInput(StreamInputProcessor.java:165)
> 	at org.apache.flink.streaming.runtime.tasks.OneInputStreamTask.run(OneInputStreamTask.java:66)
> 	at org.apache.flink.streaming.runtime.tasks.StreamTask.invoke(StreamTask.java:265)
> 	at org.apache.flink.runtime.taskmanager.Task.run(Task.java:588)
> 	at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.IllegalStateException: Could not find previous shared buffer entry with key: State(start, Normal, [
> StateTransition(TAKE, warning, with filter),
> ]), value: Event{id='three', eventType='start', timestamp=25045188001} and timestamp: 25045188001. This can indicate that the element belonging to the previous relation has been already pruned, even though you expect it to be still there.
> 	at org.apache.flink.cep.nfa.SharedBuffer.put(SharedBuffer.java:104)
> 	at org.apache.flink.cep.nfa.NFA.computeNextStates(NFA.java:293)
> 	... 8 more
> {code}



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