[ 
https://issues.apache.org/jira/browse/HADOOP-1822?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Owen O'Malley updated HADOOP-1822:
----------------------------------

    Status: Open  (was: Patch Available)

This is getting close,  but I think it could be a little cleaner on the 
interface:

1. I think that RPC.getProxy should do the lookup internally so that all of the 
rpc clients get the benefits of this.

2. I'd think the configuration should just be a class and we might as well use 
the normal SocketFactory as the default and configure the socks server 
separately. That provides a nice separation of the functionality. 

{code}
<property>
  <name>hadoop.rpc.socket.factory.class.default</name>
  <value>javax.net.SocketFactory</value>
  <description> Default SocketFactory to use for rpc clients. Can change to 
                            org.apache.hadoop.net.ProxySocksFactory to use 
socks. </description>
</property>

<property>
  <name>hadoop.socks.server</name>
  <value>host:port</value>
  <description> Used by ProxySocksFactory to locate the socks server. 
</description>
</property>
{code}

3. You should use the ReflectionUtils.newInstance because we handled lots of 
the little details about creating objects.

4. Can you give me a little more insight for the need for the map from 
SocketFactory to Client in RPC? 



> Allow SOCKS proxy configuration to remotely access the DFS and submit Jobs
> --------------------------------------------------------------------------
>
>                 Key: HADOOP-1822
>                 URL: https://issues.apache.org/jira/browse/HADOOP-1822
>             Project: Hadoop
>          Issue Type: New Feature
>          Components: dfs, ipc
>            Reporter: Christophe Taton
>            Assignee: Christophe Taton
>            Priority: Minor
>             Fix For: 0.15.0
>
>         Attachments: 1822-2007-09-11a.patch, 1822_2007-09-22_5.patch
>
>
> The purpose of this issue is to introduce a new configuration entry to setup 
> SOCKS proxy for DFS and JobTracker clients.
> This enable users to remotely access the DFS and submit Jobs as if they were 
> directly connected to the cluster Hadoop runs on.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to