[
https://issues.apache.org/jira/browse/YARN-7323?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16206787#comment-16206787
]
Gour Saha commented on YARN-7323:
---------------------------------
{quote}
Data structure for Yarnfile organized launched_command and state at Service
level, but the grouping definition was following Hadoop terminology. For
clarity, we will change data structure to match Hadoop terminology. State and
launch commands are moved to component level.
{quote}
[~eyang], launch command was always there at component-level. It was there at
the service-level as well to support simple no-component services. This patch
is killing the concept of simple no-component service and forcing the user to
create a component even for the simplest case. Also, state is still there at
the service-level along with component-level. Hence, the above description does
not match with what the assignee of this Jira and the author of the patch
intends to do. Next time, please consult with the assignee/patch-author before
making changes to the description OR make your suggestions as comments rather
than re-writing description OR request for ownership change of the jira and
then make changes to the description and provide your patch.
> Data structure update in service REST API
> -----------------------------------------
>
> Key: YARN-7323
> URL: https://issues.apache.org/jira/browse/YARN-7323
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Jian He
> Assignee: Jian He
> Attachments: YARN-7323.yarn-native-services.01.patch,
> YARN-7323.yarn-native-services.02.patch,
> YARN-7323.yarn-native-services.03.patch
>
>
> Unix terminology implies daemon process, such as httpd as a service. In
> Hadoop terminology, hdfs is a file system service, and hdfs has various
> components such as namenode, journal nodes, and datanodes. YARN service data
> model was more closely describing daemon process as a service. This JIRA
> propose to remap the YARN service data model to align with Hadoop
> terminology. Service is composed of a collection of components. Component
> is either a unix process or docker container.
>
> Data structure for Yarnfile organized launched_command and state at Service
> level, but the grouping definition was following Hadoop terminology. For
> clarity, we will change data structure to match Hadoop terminology. State
> and launch commands are moved to component level.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
