[
https://issues.apache.org/jira/browse/YARN-8131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16430934#comment-16430934
]
Vrushali C commented on YARN-8131:
----------------------------------
I agree that the end user should not be concerned about data going to specific
tables. The framework should be handling this, like the Tez AM.
In the distributed shell example, we should figure out if there is any data
that is equivalent to a sub-app use case. If not, we should write a different
one to test querying/writing out subapp data. It should not be an CLI option.
The flow name, version and flow run id are to CLI options and that is different
from a sub-app query. If we set the wrong example in DS, then it is likely to
confuse frameworks about using subapp data. Let's have a good example for
sub-application data.
One thing is for sure, the data should be written "logically" by the AM to
timeline service without caring/knowing where exactly the data ends up in the
backend. Meaning, if it is a flow level value, it's stored at the flow level.
If it's an application metric, it's at the app level. The AM need not be
concerned that there are two tables at the backend, flow & application. All it
should care for is, that this particular value belongs to flow level, that
particular value makes sense at the app level and some other third value makes
sense at the sub-app level.
> Provice CLI option to DS for publishing entities into sub application
> ---------------------------------------------------------------------
>
> Key: YARN-8131
> URL: https://issues.apache.org/jira/browse/YARN-8131
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Rohith Sharma K S
> Priority: Major
>
> Post YARN-6936, TimelineV2Client exposes API to publish entities into sub
> application table. We should add this CLI option in DS so that API can be
> tested.
--
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
