Sangjin Lee commented on YARN-3051:

bq. To plot graphs based on timeseries data, we may need to provide a time 
window for metrics too. This would be useful in case of getEntity() API. So do 
we specify this time window separately for each metric to be retrieved or same 
for all metrics ?

My sense is that it should be fine to use the same time window for all metrics. 
[~gtCarrera9]? [~zjshen]?

bq. Queries based on relations i.e. queries such as get all containers for an 
app. We can return relatesto field while querying for an app. And then client 
can use this result to fetch detailed info about related entities. Is that fine 
? Or we have to be handle it as part of a single query ?

For now, let's assume 2 queries from the client side. My thinking was that this 
is an optimization. If the storage can return two levels of entities 
efficiently, we could potentially exploit it. But maybe that's nice to have at 
the moment.

bq. Some understanding on how flow id, flow run id will be stored is required.

Li just posted the schema design in YARN-3134. That should be helpful.

> [Storage abstraction] Create backing storage read interface for ATS readers
> ---------------------------------------------------------------------------
>                 Key: YARN-3051
>                 URL: https://issues.apache.org/jira/browse/YARN-3051
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: timelineserver
>            Reporter: Sangjin Lee
>            Assignee: Varun Saxena
>         Attachments: YARN-3051_temp.patch
> Per design in YARN-2928, create backing storage read interface that can be 
> implemented by multiple backing storage implementations.

This message was sent by Atlassian JIRA

Reply via email to