[
https://issues.apache.org/jira/browse/HDFS-7533?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14271049#comment-14271049
]
Kihwal Lee commented on HDFS-7533:
----------------------------------
We can check whether a responder is running, but it may be in the process of
shutting down. Therefore, a proper check requires additional locking.
Alternatively, we can simply catch any {{Throwable}} and ignore, so that the
shutdown thread can complete the rest of the process. Since the out of band
messaging is advisory, this is acceptable, IMO.
> Datanode sometimes does not shutdown on receiving upgrade shutdown command
> --------------------------------------------------------------------------
>
> Key: HDFS-7533
> URL: https://issues.apache.org/jira/browse/HDFS-7533
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Kihwal Lee
> Assignee: Eric Payne
>
> When datanode is told to shutdown via the dfsadmin command during rolling
> upgrade, it may not shutdown. This is because not all writers have responder
> running, but sendOOB() tries anyway. This causes NPE and the shutdown thread
> dies, halting the shutdown after only shutting down DataXceiverServer.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
