[ https://issues.apache.org/jira/browse/HDFS-4471?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13573888#comment-13573888 ]
Daryn Sharp commented on HDFS-4471: ----------------------------------- Agreed, using the nameservice id is probably the cleanest solution. In the meantime, I'd be more comfortable with the snippet I posted above (assuming it works). A Eli points out, the current patch will break anyone that listens on one interface for http, but another for rpc. We've got NNs listening for http on the wildcard, but a specific ip for rpc. > Namenode WebUI file browsing does not work with wildcard addresses configured > ----------------------------------------------------------------------------- > > Key: HDFS-4471 > URL: https://issues.apache.org/jira/browse/HDFS-4471 > Project: Hadoop HDFS > Issue Type: Bug > Components: namenode > Affects Versions: 3.0.0 > Reporter: Andrew Wang > Assignee: Andrew Wang > Attachments: hdfs-4471-1.patch, hdfs-4471-2.patch > > > The namenode web UI currently passes around the NN's configured RPC address > in the URL. One of the places this is used is for file browsing; the NN > redirects to a random DN, and the DN then uses the NN address from the URL to > contact the NN. However, this fails when the NN RPC address is configured to > be the wildcard (0.0.0.0). > The fix is the same as in HDFS-3932: pass the configured address through > {{NetUtils#getConnectAddress}} first. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira