[
https://issues.apache.org/jira/browse/HDFS-2913?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13208683#comment-13208683
]
Bikas Saha commented on HDFS-2913:
----------------------------------
There is currently no clean way to shutdown NN and we are using runtime.exit().
A clean shutdown might be useful in cases such as automatic failover when the
active NN could pro-actively give up control and then terminate. In general,
each component could ideally report its failure to NN and after that NN could
decide whether to stop being active, go into safe mode or shut down completely.
This JIRA tracks adding that framework.
This need not be under HA.
This is not critical to do, specially after runtime.exit solves HDFS-2912.
> HA: Need a way to shutdown the Name Node
> ----------------------------------------
>
> Key: HDFS-2913
> URL: https://issues.apache.org/jira/browse/HDFS-2913
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: ha, name-node
> Affects Versions: HA branch (HDFS-1623)
> Reporter: Bikas Saha
> Assignee: Bikas Saha
>
> Ideally, NameNode.stop() needs to be called because it will change the HA
> state and shutdown all services. NameNode reference is not available
> anywhere. Hence it is not possible to shutdown the name node gracefully.
> A possible solution could be to have a Service interface that gets passed
> down to components like FSNameSystem, via which they can inform the NameNode
> about irrecoverable errors. NameNode could then decide to shutdown.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
