[
https://issues.apache.org/jira/browse/HDFS-5755?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13867724#comment-13867724
]
Steve Loughran commented on HDFS-5755:
--------------------------------------
Regarding fixes, the obvious one is to have all the{{RetryInvocationHandler}}
to move from a {{while(true)} to a {{while(clientOpen.get()}}, and with
{{AtomicBoolean clientOpen}} be something {{DfsClient.close()}} sets to true
before it starts waiting for its threads.
Of course, it may be better to put this into the IPC {{RPC.stopProxy}} class,
as it is more at the RPC level -it just surfaces in HDFS as a side effect of
the shutdown hooks
> Filesystem close blocks SystemExit if a DFS client is trying to talk to a
> nonexistent NN
> ----------------------------------------------------------------------------------------
>
> Key: HDFS-5755
> URL: https://issues.apache.org/jira/browse/HDFS-5755
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 2.2.0
> Reporter: Steve Loughran
> Priority: Minor
>
> If a DFS client instance is spinning in a connection refused cycle, and you
> try to stop the process via System.exit(), the exit call does not complete
> until the client has actually given up.
> That is: you can't exit a process with a standard kill while a DFS client
> operation is blocking.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
