Placing data into HDFS..!

[Sugandha Naolekar <su...@gmail.com>]

Hello!

I have a 7 node cluster. But there is one remote node(8th machine) within
the same LAN which holds some kind of data. Now, I need to place this data
into HDFS. This 8th machine is not a part of the hadoop
cluster(master/slave) config file.

So, what I have thought is::
-> Will get the Filesystem instance by using FileSystem api
-> Will get the local file's(remote machine's) instance by using the same
api by passing a different config file which simply states a tag of fs,
default.name

-> And then will simply use all the methods to copy and get the data back
from HDFS...
-> During the complete episode, I will have to take care of the proxy issues
for remote node to get connceted to Namenode.

Is this procedure correct?

Thanking You,

-- 
Regards!
Sugandha

Re: Placing data into HDFS..!

[Usman Waheed <us...@opera.com>]

If you are going to be using this 8th machine as a client only then 
ensure that it is running the same version of hadoop as your cluster.
In the config file hadoop-site.xml point fs.default.name to the namenode.

-Usman
> Hello!
>
> I have a 7 node cluster. But there is one remote node(8th machine) within
> the same LAN which holds some kind of data. Now, I need to place this data
> into HDFS. This 8th machine is not a part of the hadoop
> cluster(master/slave) config file.
>
> So, what I have thought is::
> -> Will get the Filesystem instance by using FileSystem api
> -> Will get the local file's(remote machine's) instance by using the same
> api by passing a different config file which simply states a tag of fs,
> default.name
>
> -> And then will simply use all the methods to copy and get the data back
> from HDFS...
> -> During the complete episode, I will have to take care of the proxy issues
> for remote node to get connceted to Namenode.
>
> Is this procedure correct?
>
> Thanking You,
>
>