[
https://issues.apache.org/jira/browse/HADOOP-11989?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14550363#comment-14550363
]
Steve Loughran commented on HADOOP-11989:
-----------------------------------------
Allen -thanks for that.
Looking in to why it was failing for me, it's because my shell is "fish", not
"bash". Switch to bash and all works, so it will work in any bash shell spawned
by the NM.
We don't care about cygwin any more, do we?
> Kill command for process group id throws ExitCodeException
> ----------------------------------------------------------
>
> Key: HADOOP-11989
> URL: https://issues.apache.org/jira/browse/HADOOP-11989
> Project: Hadoop Common
> Issue Type: Bug
> Components: util
> Affects Versions: 2.6.0
> Environment: debian 7
> Reporter: Chackaravarthy
> Assignee: Chackaravarthy
> Attachments: HADOOP-11989.patch
>
>
> In the environment where setsid available, kill command is constructed with
> single dash, which throws ExitCodeException (improper usage of command)
> {noformat}
> kill -signal_num -<process_group_id>
> {noformat}
> Encountered this issue due to YARN-3561
> Kill command need to be constructed with double dash (--) which means "end of
> command options" i.e. it tells ssh command not to try to parse what comes
> after command line options. It should be constructed as follows :
> {noformat}
> kill -signal_num -- -<process_group_id>
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
