[
https://issues.apache.org/jira/browse/FLINK-5114?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15722887#comment-15722887
]
ASF GitHub Bot commented on FLINK-5114:
---------------------------------------
Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/2913
Having a quick look at this: I think this breaks with a fundamental design
in the ExecutionGraph:
The `findExecutionAttemptWithId(...)` method searches the prior execution
attempts.
Why is that necessary? Can we not just assume that if the attempt is not
equal to the current execution attempt, then the status is some form of
"disposed".
If the produced result is finished, the execution will still not be in the
"prior execution attempts". That can only happen once the task restarts, in
which case you should not try and fetch the partition any more.
> PartitionState update with finished execution fails
> ---------------------------------------------------
>
> Key: FLINK-5114
> URL: https://issues.apache.org/jira/browse/FLINK-5114
> Project: Flink
> Issue Type: Bug
> Components: Network
> Reporter: Ufuk Celebi
> Assignee: Ufuk Celebi
>
> If a partition state request is triggered for a producer that finishes before
> the request arrives, the execution is unregistered and the producer cannot be
> found. In this case the PartitionState returns null and the job fails.
> We need to check the producer location via the intermediate result partition
> in this case.
> See here: https://api.travis-ci.org/jobs/177668505/log.txt?deansi=true
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
