Varun Saxena commented on YARN-4053:

bq. What kind of metrics do you have in mind that will have floating point 
numbers ?
There was some plan for reporting some cluster level metrics in future too, few 
of them would be floating point as well. Refer to json in YARN-3881
Also I remember some discussion during aggregation design regarding storing 
averages. Are we planning to calculate them on the fly instead ?
Moreover,TimelineMetric stores metric value as a {{java.lang.Number}}. This 
means we are saying metric can store a floating point value as well. As we have 
no control over systems outside YARN(say Tez), if they use ATS and publish a 
metric of floating type, I guess we should be able to handle it.

Thoughts ?

If it has been decided that metrics can only be integral values, then its fine. 
Wont have to take care of it then. Let me know.
Also, another key point we need to decide is that do we only support values 
till signed longs(8 bytes) ? 

> Change the way metric values are stored in HBase Storage
> --------------------------------------------------------
>                 Key: YARN-4053
>                 URL: https://issues.apache.org/jira/browse/YARN-4053
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: timelineserver
>    Affects Versions: YARN-2928
>            Reporter: Varun Saxena
>            Assignee: Varun Saxena
> Currently HBase implementation uses GenericObjectMapper to convert and store 
> values in backend HBase storage. This converts everything into a string 
> representation(ASCII/UTF-8 encoded byte array).
> While this is fine in most cases, it does not quite serve our use case for 
> metrics. 
> So we need to decide how are we going to encode and decode metric values and 
> store them in HBase.

This message was sent by Atlassian JIRA

Reply via email to