how to define/setup a edge node

[#1]

Let's say I have 10 nodes to install hadoop, and 4 nodes are for edge node to install all ML or ETL thrid party tools,

Can I include all nodes in installation, only basic hadoop installation on these 4 nodes only, then install third party tools on them?

0. when start cluster installation, I can list all of 10 server for this cluster installation, right? these 4 nodes supposed to be part of hadoop cluster.

1. During the cluster installation, is any where I can assign a role 'client' to these 4 nodes?

2. if yes, please show me a link, need to see exact step to setup 'role', role as client? got to be a role for these 4 nodes.

3. if not, may I just hadoop on these nodes and simply install third party tools. I already have Ambari installed on a separated server.

thank you very much for your help.

Robin

[#2 · Accepted solution]

@Robin Dong

Response to your questions:

Can I include all nodes in installation, only basic hadoop installation on these 4 nodes only, then install third party tools on them?

Yes

0. when start cluster installation, I can list all of 10 server for this cluster installation, right? these 4 nodes supposed to be part of hadoop cluster.

Yes, you can list all of them and assign clients on 4 nodes which you would like to make them client nodes.

1. During the cluster installation, is any where I can assign a role 'client' to these 4 nodes?

Yes

2. if yes, please show me a link, need to see exact step to setup 'role', role as client? got to be a role for these 4 nodes.

http://docs.hortonworks.com/HDPDocuments/Ambari-2.4.2.0/bk_ambari-installation/content/assign_slaves...

3. if not, may I just hadoop on these nodes and simply install third party tools. I already have Ambari installed on a separated server.

If you want to manage those 4 nodes by Ambari make them part of cluster and install only clients and install your 3rd part tools. If you don't want to manage those 4 nodes by Ambari even in future, then install only clients from command line and copy the Hadoop configuration from existing cluster to these machines.

Hope this helps you.

[#3]

Thank you, it answered all.

thank you so much.