[
https://issues.apache.org/jira/browse/YARN-934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhijie Shen updated YARN-934:
-----------------------------
Attachment: YARN-934.2.patch
Updated the patch:
1. The abstract methods have the protobuf classes as the parameters. Therefore,
the implementations of them just need to persist the protobuf objects.
HistoryWriter already has the logic to extract the required information from
RMApp, RMAppAttempt and RMContainer to fill the fields of these protobuf
objects. Please note that RMContainer now doesn't have all the required
information to persist. We'll work on it a separate ticket.
2. Change from multi-thread model to using the dispatcher on a single thread.
Thought about it again. IMHO, users know the problem best. If users think the
single dispatcher thread is the bottleneck of persisting history data, they can
fork more worker threads in the implementations of the the abstract methods.
3. Add the test file for HistoryWriter
> Defining Writing Interface of HistoryStorage
> --------------------------------------------
>
> Key: YARN-934
> URL: https://issues.apache.org/jira/browse/YARN-934
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Zhijie Shen
> Assignee: Zhijie Shen
> Attachments: YARN-934.1.patch, YARN-934.2.patch
>
>
--
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
