[
https://issues.apache.org/jira/browse/FLINK-1362?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14321906#comment-14321906
]
Till Rohrmann commented on FLINK-1362:
--------------------------------------
Does this still occur? I think I fixed it. I think the problem was that the
method {{TaskManager.notifyExecutionStateChange}} accessed the internal state
of the TaskManager actor and once it is stopped, then a
{{NullPointerException}} was thrown.
With PR #319, this was refactored. Now the {{Task}} sends a message to the
TaskManager which forwards it to the JobManager. In case that the TaskManager
was already shut down, the message will be sent to the dead letter mailbox. But
the JobManager should be notified about shut down TaskManager.
> Spurious errors in execution state messaging between TaskManager and
> JobManager
> -------------------------------------------------------------------------------
>
> Key: FLINK-1362
> URL: https://issues.apache.org/jira/browse/FLINK-1362
> Project: Flink
> Issue Type: Bug
> Components: Distributed Runtime
> Affects Versions: 0.9
> Reporter: Stephan Ewen
>
> In various tasks, it is possible that a notification from the TaskManager to
> the JobManager fails with the following exception:
> {code}
> Exception in thread "<task thread>" java.lang.NullPointerException
> at
> org.apache.flink.runtime.taskmanager.TaskManager.notifyExecutionStateChange(TaskManager.scala:281)
> at org.apache.flink.runtime.taskmanager.Task.markFailed(Task.java:199)
> at
> org.apache.flink.runtime.execution.RuntimeEnvironment.run(RuntimeEnvironment.java:307)
> at java.lang.Thread.run(Thread.java:745)
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
