Carolus, do you have HA enabled? If so, read below.
Hue needs httpfs for a cluster with HA enabled.
I had the same problem because although I had enabled High Availability, I had not installed and configured httpfs (a webhdfs proxy). When my active/standby name nodes switched their status, Hue still was pointing at the original active node.
(note – I’m running HDP 2.2 on Centos 6.6, and the post-install RPM scripts for hadoop-httpfs RPM don’t properly work. If you have the same behavior, you’ll need to install the init.d script and the conf files manually)
To configure Hue to work with Namenode HA, we need to use httpfs instead of webhdfs to communicate with the namenode.
Below is the list of steps after the whole cluster is setup correctly with Namenode HA enabled.
1. Install httpfs server in any node within the cluster by running: yum install hadoop-httpfs
2. Configure httpfs-site.xml and ensure the following properties are correct:
3. Add the following properties to core-site.xml for httpfs via Ambari Web UI, which requires a restart of HDFS service:
4. Start hadoop-httpfs service by running: service hadoop-httpfs restart
5. Configure hue.ini by modify the below variable to:
webhdfs_url=http://<fqdn of httpfs server>:14000/webhdfs/v1/
6 Restart Hue.