[
https://issues.apache.org/jira/browse/HDFS-7847?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14354259#comment-14354259
]
stack commented on HDFS-7847:
-----------------------------
Mighty [~clamb] I forgot about that proxy stuff. It was never pretty....
So idea is that you add proxies for protocols' currently hardwired? That sounds
needed if we want to fake out DFSClient that there are thousands of DNs with
millions of blocks each. How comes I don't see your added code replacing
current hard-wiring of DatanodeProtocol, etc?
On NNThroughputBenchmark, I think you should just steal the good bits and leave
the rest behind you. I say this because I don't think you will be able to put
up a load on a NN in a unit test that emulates thousands of nodes making
mega-reports about millions upon millions of blocks while simulating thousands
of clients. I'd think you'll want to stand up a NN and then batter it for hours
on end from many processes distributed over a cluster. Just a suggestion.
> Modify NNThroughputBenchmark to be able to operate on a remote NameNode
> -----------------------------------------------------------------------
>
> Key: HDFS-7847
> URL: https://issues.apache.org/jira/browse/HDFS-7847
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Affects Versions: HDFS-7836
> Reporter: Colin Patrick McCabe
> Assignee: Charles Lamb
> Attachments: HDFS-7847.000.patch, make_blocks.tar.gz
>
>
> Modify NNThroughputBenchmark to be able to operate on a NN that is not in
> process. A followon Jira will modify it some more to allow quantifying native
> and java heap sizes, and some latency numbers.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
