[
https://issues.apache.org/jira/browse/HDFS-6169?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13962057#comment-13962057
]
Haohui Mai commented on HDFS-6169:
----------------------------------
The overall control flow is more clear in the following way:
{code}
public void messageReceived(ChannelHandlerContext ctx, MessageEvent e) throws
Exception {
String op = getOp(...);
try {
String path = getPath(...);
handleOperation(op, e);
} finally {
e.getFuture().addListener(ChannelFutureListener.CLOSE);
}
}
{code}
Considering the following code in the patch:
{code}
...
+
+ private String trimPath(String path, MessageEvent e) {
+ if (path.startsWith("/webhdfs/v1/")) {
+ return path.replaceFirst("/webhdfs/v1", "");
+ } else {
+ // invalid path, close the channel
+ LOG.warn("Path: " + path + " should start with \"/webhdfs/v1/\"");
+ HttpResponse response = new DefaultHttpResponse(
+ HttpVersion.HTTP_1_1, HttpResponseStatus.NOT_FOUND);
+ LOG.info("404 method=GET target=" + path);
+ e.getChannel().write(response).addListener(ChannelFutureListener.CLOSE);
+ return "";
+ }
+ }
{code}
1. How to differentiate the url "/webhdfs/v1?op=LISTSTATUS" and "/wrongurl"?
You probably need to return null in the invalid case.
2. It is a bad design to have code of reclaiming resources in different places.
I suggest following the control flow described on the top of this comment.
> Move the address in WebImageViewer
> ----------------------------------
>
> Key: HDFS-6169
> URL: https://issues.apache.org/jira/browse/HDFS-6169
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: tools
> Affects Versions: 2.5.0
> Reporter: Akira AJISAKA
> Assignee: Akira AJISAKA
> Attachments: HDFS-6169.2.patch, HDFS-6169.3.patch, HDFS-6169.4.patch,
> HDFS-6169.5.patch, HDFS-6169.patch
>
>
> Move the endpoint of WebImageViewer from http://hostname:port/ to
> http://hostname:port/webhdfs/v1/ to support {{hdfs dfs -ls}} to
> WebImageViewer.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
