[
https://issues.apache.org/jira/browse/FLINK-4430?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16542635#comment-16542635
]
Till Rohrmann commented on FLINK-4430:
--------------------------------------
I think at the moment, the solution is that the {{JobManager}} should not
release any slots which have state/resources assigned.
> Prevent BLOCKING partition data loss when task is finished
> ----------------------------------------------------------
>
> Key: FLINK-4430
> URL: https://issues.apache.org/jira/browse/FLINK-4430
> Project: Flink
> Issue Type: Sub-task
> Components: Cluster Management
> Reporter: Kurt Young
> Assignee: zhuhaifeng
> Priority: Major
> Labels: flip-6
>
> When we have a BLOCKING result partition type, the data is actually holded by
> TaskManager, not the slot. When we finish the produce task, we will mark this
> task finished and try to release the slot. Under the new architecture, with
> yarn or mesos mode, releasing slot may trigger releasing the container, so
> the TaskManager will be terminated, thus the result data is lost. We should
> introduce some mechanism to prevent this from happening.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
