[
https://issues.apache.org/jira/browse/MAPREDUCE-6424?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15264756#comment-15264756
]
Naganarasimha G R commented on MAPREDUCE-6424:
----------------------------------------------
I have uploaded a patch without consideration of above points as i felt though
{{HistoryEvent.getTimelineEntitiesToPublish}} would be ideal but has far too
many changes and need to consider how to handle configs too. So thought better
to postpone this decision to [~varun_saxena]'s patch. Thoughts?
bq. So in general, the penalty for making fine-grained putEntitiesAsync() calls
shouldn't be too big.
What i was considering was if we go for
{{HistoryEvent.getTimelineEntitiesToPublish}}/ {{HistoryEvent.addDataToEntity}}
approach and if config size is too big then we will not be able to address it
by splitting into multiple entities because :
* All entities passed to {{timelineClient.putEntities/putEntitiesAsync}} is
published as a single REST call, currently not considering number of
entities/size.
* *HistoryEvent.addDataToEntity* approach doesn't model multiple entities which
might be required for publishing ApplicationEntity's metrics and configs.
* Based on security concerns we might take different approach in publishing
metrics and configs as part of ApplicationEntity in YARN-3401
TestMRTimelineEventHandling.testMRNewTimelineServiceEventHandling is passing
locally
{code}
-------------------------------------------------------
T E S T S
-------------------------------------------------------
Running org.apache.hadoop.mapred.TestMRTimelineEventHandling
Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 69.195 sec - in
org.apache.hadoop.mapred.TestMRTimelineEventHandling
Results :
Tests run: 1, Failures: 0, Errors: 0, Skipped: 0
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
{code}
would be helpful if someone validates it for me
> Store MR counters as timeline metrics instead of event
> ------------------------------------------------------
>
> Key: MAPREDUCE-6424
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6424
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Reporter: Junping Du
> Assignee: Naganarasimha G R
> Labels: yarn-2928-1st-milestone
> Attachments: MAPREDUCE-6424-YARN-2928.v1.001.patch,
> MAPREDUCE-6424-YARN-2928.v1.002.patch, MAPREDUCE-6424-YARN-2928.v1.003.patch,
> MAPREDUCE-6424-YARN-2928.v1.004.patch
>
>
> In MAPREDUCE-6327, we make map/reduce counters get encoded from
> JobFinishedEvent as timeline events with counters details in JSON format.
> We need to store framework specific counters as metrics in timeline service
> to support query, aggregation, etc.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
