Do we need to Install Hadoop on Edge Node

[#1]

I have just started to learn about the hadoop cluster. I have few queries about the Edge Node. I have gone through many posts regarding Edge Node. I understand that we have to install all the clients in it.

1. Do we need to install Hadoop on Edge Node?

2. Do we need to install clients on hadoop cluster as well?

Best Regards

[#2 · Accepted solution]

You will need to install the hadoop binaries and configs on the edge node. That will give you the command line clients that come out of the box with HDP, like pig and beekeeper.

See the following on edge nodes:

http://www.dummies.com/programming/big-data/hadoop/edge-nodes-in-hadoop-clusters/

https://dwbi.org/etl/bigdata/187-set-up-client-node-gateway-node-in-hadoop-cluster

See this post making the good point of installing hadoop binaries via Ambari so they are always up to date with the rest of the cluster.

https://community.hortonworks.com/questions/39568/how-to-create-edge-node-for-kerberized-cluster.htm...

[#3]

HDFS client, Pig, Sqoop, Spark, Tez, Falcon are all useful clients to have there.

As well as JDK 8.

Hadoop Configuration files.

Java 8 and Hadoop Configuration files are useful for NiFi servers to have as well.