[
https://issues.apache.org/jira/browse/HADOOP-8326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13263944#comment-13263944
]
Robert Joseph Evans commented on HADOOP-8326:
---------------------------------------------
Sadly there is no easy way to detect these processes. This is because when we
do a fork/exec through the default container executor most of the time we will
also do a setsid on them, so that the session id is different from the
original. This also makes it so that the process group id is different. We
have effectively made these processes daemons. They even end up having init as
their PPID when we are done.
I think the best way to handle this is to be conservative about what we kill
off so that we kill off everything in our own process group, and then also any
daemon java processes owned by the current user, and that in their arguments
they have a reference to the current working directory in some way. This
matches all of the errant processes that I have seen on the build machines.
> test-patch can leak processes in some cases
> -------------------------------------------
>
> Key: HADOOP-8326
> URL: https://issues.apache.org/jira/browse/HADOOP-8326
> Project: Hadoop Common
> Issue Type: Bug
> Components: test
> Affects Versions: 3.0.0
> Reporter: Robert Joseph Evans
> Assignee: Robert Joseph Evans
>
> test-patch.sh can leak processes in some cases. These leaked processes can
> cause subsequent tests to fail because they are holding resources, like ports
> that the others may need to execute correctly.
--
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
