[
https://issues.apache.org/jira/browse/YARN-7973?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16380322#comment-16380322
]
Shane Kumpf commented on YARN-7973:
-----------------------------------
Attaching a patch that adds a new {{relaunchContainer}} method to the
ContainerExecutors and ContainerRuntimes. For all but the
{{DockerLinuxContainerRuntime}}, {{relaunchContainer}} simply calls
{{launchContainer}}, to mimic the existing behavior. In the case of
{{DockerLinuxContainerRuntime}}, relaunch will instead call {{docker start}} on
the existing container. For {{docker start}}, we require the same general flow
as {{docker run}} where it is necessary to get the PID and wait for the process
to exit. As a result, these two paths are the same through c-e, which appears
to work well.
I've tested this against distributed shell, MR PI, MR sleep and several YARN
Native Services apps - both process based and Docker, and tried to inject
failures where appropriate. The testing looks good. I think we have opportunity
to clean up some exception logging from the privileged executor, but I'll open
a new issue to look into that clean up.
> Support ContainerRelaunch for Docker containers
> -----------------------------------------------
>
> Key: YARN-7973
> URL: https://issues.apache.org/jira/browse/YARN-7973
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Shane Kumpf
> Assignee: Shane Kumpf
> Priority: Major
> Attachments: YARN-7973.001.patch
>
>
> Prior to YARN-5366, {{container-executor}} would remove the Docker container
> when it exited. The removal is now handled by the
> {{DockerLinuxContainerRuntime}}. {{ContainerRelaunch}} is intended to reuse
> the workdir from the previous attempt, and does not call {{cleanupContainer}}
> prior to {{launchContainer}}. The container ID is reused as well. As a
> result, the previous Docker container still exists, resulting in an error
> from Docker indicating the a container by that name already exists.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
