You should be able to setup a client only machine and assign spark and hive
clients on that.
On Fri, Dec 4, 2015 at 1:15 PM -0800, "Hoc Phan" <quang...@yahoo.com> wrote:
When I setup Cloudera, there is no /hive dir in management node. I guess I had
to add that role in Cloudera Manager first?
As a best practice, I rather have Zeppelin separately from the rest.
On Friday, December 4, 2015 12:56 PM, Abhi Basu <9000r...@gmail.com> wrote:
Management node, is where we have it installed.
Our management node is setup as a hadoop client.
On Fri, Dec 4, 2015 at 12:50 PM, Hoc Phan <quang...@yahoo.com> wrote:
Hi
I have a Cloudera Express cluster consisting of 1 management node, 1 master and
6 worker nodes. The instruction seems to assume Zeppelin must be in the same
node with Hadoop and Spark as the below environment variables pointing to local
directories:
# ./conf/zeppelin-env.sh
export SPARK_HOME=...
export HADOOP_HOME=...
I installed Zeppelin inside the master node and it works. BUT I need
flexibility of where to put Zeppelin because I may have Hive directory in
separate node from Spark directory.
What is your recommendation?
--
Abhi Basu
