[
https://issues.apache.org/jira/browse/YARN-4759?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15373071#comment-15373071
]
Shane Kumpf commented on YARN-4759:
-----------------------------------
In my dev environment, I find it very useful to be able to keep containers
around after they have completed for debugging purposes. To achieve this, I've
commented out the "docker rm" logic in container-executor in my dev
environment. We ultimately need to make the "docker rm" configurable, however,
it would be good to get this patch in ASAP as NM restarts will kill all running
containers, which is quite impactful.
I'd like to keep the logic previous laid out for now (call docker stop for any
non-null signal and throw the benign exception if the container has already
been removed by container-executor) so we don't have to rework this patch once
we make "docker rm" configurable. I will open another issue to add a job level
configuration toggle for the "docker rm". The "docker rm" should likely be
moved into the runtime versus handled in container-executor, but that can be
discussed when the time comes.
Working on final testing of the patch now and will submit it for review today.
> Revisit signalContainer() for docker containers
> -----------------------------------------------
>
> Key: YARN-4759
> URL: https://issues.apache.org/jira/browse/YARN-4759
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: yarn
> Reporter: Sidharta Seethana
> Assignee: Shane Kumpf
>
> The current signal handling (in the DockerContainerRuntime) needs to be
> revisited for docker containers. For example, container reacquisition on NM
> restart might not work, depending on which user the process in the container
> runs as.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
