[
https://issues.apache.org/jira/browse/TEZ-3066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15111081#comment-15111081
]
Jeff Zhang commented on TEZ-3066:
---------------------------------
This issue may happen when task attempt is in the process of kill
(TaskAttemptStateInternal.KILL_IN_PROGRESS), in this case we have logged the
TaskAttemptFinishedEvent, but the task attempt may still alive and will
heartbeat with AM. which cause the ConcurrentModificationException here.
Although it might be better to log TaskAttemptFinishedEvent in the last state,
but it require much change on the TaskAttempt state machine, the easier way is
to copy a new list for dataEvents to avoid the ConcurrentModificationException.
Also check taGeneratedEvents in TaskAttemptFinishedEvent, it won't be updated
after TaskAttemptFinishedEvent is created, otherwise it is some bug of
TaskAttempt's state machine. [~hitesh] [~bikassaha] Please help review it.
> TaskAttemptFinishedEvent ConcurrentModificationException in recovery or
> history logging services
> ------------------------------------------------------------------------------------------------
>
> Key: TEZ-3066
> URL: https://issues.apache.org/jira/browse/TEZ-3066
> Project: Apache Tez
> Issue Type: Bug
> Affects Versions: 0.7.0
> Reporter: Jason Lowe
> Assignee: Jeff Zhang
> Attachments: TEZ-3066-1.patch
>
>
> A ConcurrentModificationException can occur if a TaskAttemptFinishedEvent is
> processed by the recovery service or another history logging service. Sample
> stacktraces to follow.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
