[
https://issues.apache.org/jira/browse/HDFS-155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12803881#action_12803881
]
Steve Loughran commented on HDFS-155:
-------------------------------------
This is a very old defect. Has anyone seen it for a while?
The webapp location, {{getWebappsPath()}} is now in
{{org.apache.hadoop.http.HttpServer}}. I would suspect the problem might be in
how Jetty's {{org.mortbay.jetty.webapp.WebAppContext.setWar())}} takes either a
file path or a URL; if it gets confused I could see this problem happening.
> StatusHttpServer is failing if trying to use hadoop as a standalone jar
> -----------------------------------------------------------------------
>
> Key: HDFS-155
> URL: https://issues.apache.org/jira/browse/HDFS-155
> Project: Hadoop HDFS
> Issue Type: Bug
> Environment: WinXP
> Reporter: Paul Malolepsy
> Assignee: Sameer Paranjpye
> Priority: Minor
>
> I just recently tried to upgrade from hadoop .2 and I am getting an error
> during the startup of StatusHttpServer. I did set my hadoop.log.dir, but
> there are still errros. Here is my stack below.
> java.io.FileNotFoundException:
> E:\projects\marzen\file:\E:\projects\marzen\web\WEB-INF\lib\hadoop.jar!\webapps\datanode
> at
> org.mortbay.jetty.servlet.WebApplicationContext.resolveWebApp(WebApplicationContext.java:266)
> at
> org.mortbay.jetty.servlet.WebApplicationContext.doStart(WebApplicationContext.java:449)
> at org.mortbay.util.Container.start(Container.java:72)
> at org.mortbay.http.HttpServer.doStart(HttpServer.java:753)
> at org.mortbay.util.Container.start(Container.java:72)
> at
> org.apache.hadoop.mapred.StatusHttpServer.start(StatusHttpServer.java:177)
> at org.apache.hadoop.dfs.DataNode.<init>(DataNode.java:167)
> at org.apache.hadoop.dfs.DataNode.makeInstance(DataNode.java:1069)
> at org.apache.hadoop.dfs.DataNode.run(DataNode.java:1003)
> ...rest of stack removed...
> The root of the issues seems to be in StatusHttpServer.getWebAppsPath(). Its
> possible that I am using hadoop differently than most people (as a standalone
> jar in another web app), but I'm not sure how this function could ever work
> if you were just using hadoop packaged as a standalone jar. Its trying to
> get the webapps path using this line:
> URL url = StatusHttpServer.class.getClassLoader().getResource("webapps");
> Which is returning the following path. This is a path within the hadoop jar:
> file:\E:\projects\marzen\web\WEB-INF\lib\hadoop.jar!\webapps
> It then tries to return the canonical path of that wich returns this path:
> E:\projects\marzen\file:\E:\projects\marzen\web\WEB-INF\lib\hadoop.jar!\webapps
> I don't even think that is a valid path, and the Jetty server certainly
> doesn't like it. Is there something I'm doing wrong? Is it possible nobody
> else is using hadoop as a standalone jar? Is this a windows specific bug?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
