[
https://issues.apache.org/jira/browse/TEZ-1191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14026739#comment-14026739
]
Siddharth Seth commented on TEZ-1191:
-------------------------------------
We should proceed with this, but the run-away process issue is the biggest
concern - can't have DAGAppMasters lingering on clusters. Is it possible to
make the exit optional, and skipped only when using the in-process work that
you're doing. That's a stop gap until more work is done to ensure proper
shutdown.
> DAGAppMaster shuts down JVM resulting in shutdown propagation up the stack.
> ---------------------------------------------------------------------------
>
> Key: TEZ-1191
> URL: https://issues.apache.org/jira/browse/TEZ-1191
> Project: Apache Tez
> Issue Type: Bug
> Affects Versions: 0.4.0
> Reporter: Oleg Zhurakousky
> Assignee: Oleg Zhurakousky
> Labels: patch
> Attachments: TEZ-1191.patch
>
>
> Depending on the implementation of container executor identified via
> _yarn.nodemanager.container-executor.class_ property, DAGAppMaster may bring
> down YARN cluster if invoked within the same JVM as RM.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
