[
https://issues.apache.org/jira/browse/YARN-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13755334#comment-13755334
]
Bikas Saha commented on YARN-1065:
----------------------------------
We need the service data provided by the Auxillary Service and not the service
data sent by the AM in the container launch context. The AM provides
application part of service data. In response, the aux service provides the
service part of the service data. This service part ==
ContainerManagerImpl.auxiliaryServices.getMetaData(). That is what we need to
set into the env of the container via sanitizeEnv.
It would be great if we use matching names like getServiceDataFromEnv() and
setServiceDataIntoEnv().
Would be good to prepend something NM specific to the service name before
setting in the env. This just helps identify that this was set by the NM.
Something like NM_AUX_SERVICE_. So for service abcd, we will add
NM_AUX_SERVICE_abcd into the env.
> NM should provide AuxillaryService data to the container
> --------------------------------------------------------
>
> Key: YARN-1065
> URL: https://issues.apache.org/jira/browse/YARN-1065
> Project: Hadoop YARN
> Issue Type: Task
> Reporter: Bikas Saha
> Assignee: Xuan Gong
> Attachments: YARN-1065.1.patch, YARN-1065.2.patch, YARN-1065.3.patch,
> YARN-1065.4.patch
>
>
> Start container returns auxillary service data to the AM but does not provide
> the same information to the task itself. It could add that information to the
> container env with key=service_name and value=service_data. This allows the
> container to start using the service without having to depend on the AM to
> send the info to it indirectly.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
