2 master node and 3 data node setting up HA clusters, request for step-wise document

[#1]

first time installing Hortwonworks cluster with 2 master node and 3 data node with High Availability, Need help on this.. do we have any step-wise document to follow..

[#2]

@Snoops

I think you will get a kick in trying it out by yourself. The most common HA components are NaneNode,RM and of course zookeeper minimum 3 instances to avoid the split brain decision.

When using Ambari it's quite straightforward. The HA components shouldn't reside on the same rack and each rack should have network redundancy Core Switch HA and further even have 2 ISP's, but goes beyond the responsibility of an Hadoop admin and should be the work of your network team though you have to be aware of the consequences when setting up you HA cluster.

So first questions first what components do you intend to have in HA?

Please revert !!

[#3]

@ Geoffrey Shelton

Thanks for the quick response, planning to go HA for these components Hive, oozie, hbase as of now. Can you please suggest me is this the right way to go with. I already setup Ambari server setup with 1 Master node and 2 Data nodes, it's working fine. First time will be going with HA. Is there step-wise documents will be great help.

[#4]

@Snoops

First, ensure you have ZooKeeper ensemble(3) running. then

1. NameNode HA
2. ResourceManager(YARN)
   

Once all is fine, if you are using VMware take a snapshot and proceed

1. MySQL HA
2. Oozie

You will need a Loadbalancer, Virtual IP, or Round-Robin DNS. The load balancer should be configured for round-robin between the Oozie servers to distribute the requests

[#5]

@Geoffrey

Thank you. How is Zookeeper and YARN different in HA, can you please let me know, how exactly it works..