[
https://issues.apache.org/jira/browse/HDFS-14162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16726395#comment-16726395
]
Erik Krogen commented on HDFS-14162:
------------------------------------
I attached a quick-and-dirty demonstration patch
([^HDFS-14162-HDFS-12943.wip0.patch]) showing how we can solve this by making
{{NameNodeHAProxyFactory}} aware of {{AlignmentContext}}, and making
{{ObserverReadProxyProvider}} keep track of a {{ClientProtocol}} proxy in
addition to its main proxy. In the normal case of ORPP with {{ClientProtocol}},
the same protocol object is used; in the case of a different protocol (e.g.
{{NameNodeProtocol}}), it instantiates an extra client proxy to use for
{{getHAServiceState()}}. It's not the cleanest way of solving the problem -- I
don't really like needing to directly instantiate the {{ClientProtocol}}
instead of being able to use the factory -- but it works.
I also had to modify the test slightly -- the way the DataNodes were created
didn't seem to work properly.
> Balancer should work with ObserverNode
> --------------------------------------
>
> Key: HDFS-14162
> URL: https://issues.apache.org/jira/browse/HDFS-14162
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Konstantin Shvachko
> Priority: Major
> Attachments: HDFS-14162-HDFS-12943.wip0.patch,
> testBalancerWithObserver.patch
>
>
> Balancer provides a substantial RPC load on NameNode. It would be good to
> divert Balancer RPCs {{getBlocks()}}, etc. to ObserverNode. The main problem
> is that Balancer uses {{NamenodeProtocol}}, while ORPP currently supports
> only {{ClientProtocol}}.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
