[
https://issues.apache.org/jira/browse/KNOX-353?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13981520#comment-13981520
]
Larry McCay commented on KNOX-353:
----------------------------------
Initial support will be done using redirection.
Adding a specialized topology with a name that matches the configured (or
default name) will result in a simple web application with only a web.xml that
maps all requests to the GatewayRedirectServlet. Within the web.xml servlet
definition an init param will reflect the configured (or default) context path
to add to the incoming request URL and redirect to.
The client should follow the redirect and all further requests will reflect the
URLs expected by Knox - including the redirects that are a result of various
multi-step Hadoop operations.
Rewrites will continue to behave the same way.
We can consider whether or how this should evolve in a later jira.
In the mean time, this approach should satisfy usecases where mapreduce jobs
would like to access resources through Knox from within the cluster.
> Support Hadoop java client WebHdfs URLs
> ---------------------------------------
>
> Key: KNOX-353
> URL: https://issues.apache.org/jira/browse/KNOX-353
> Project: Apache Knox
> Issue Type: Improvement
> Components: Server
> Reporter: Larry McCay
> Assignee: Larry McCay
> Fix For: 0.5.0
>
>
> In order for the gateway to be useful from within the Hadoop cluster, we need
> to support the URLs used by the Hadoop java client.
> The WebHdfs URLs from the java client do not and cannot be configured to
> interject the gateway_path and cluster_name context path values. We therefore
> need to support a way to deploy a default application that will serve the
> WebHdfs requests or that will redirect to an application that will.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
