Hi, I have a small production cluster running and would like to add a node to it. But the node has hdp installed already(this is a test node which was installed before even starting the cluster). All the versions, jdk, ambari, hdp and hadoop versions are exactly same. Will i be able to add this node to the cluster, if yes then what checks i have to make? is it recommended? or do i just do a format or reimage of the node and start over confguring again? Answers are highly appreciated.
It is OK to add this node. I don't see any problem. There is no need to format. Just make sure that there are no HDP process running on this new node.
While adding the node, you may get an warning saying that users and packages already exists on the node and it is safe to ignore it and proceed.
While adding the components like Datanode, ambari will try to install the packages but as the packages are already installed, Ambari will not attempt to reinstall the packages and will proceed with adding the node and components.
Just make sure ambari agent version is same as ambari server version and /etc/ambari-agent/conf/ambari-agent.ini file has correct ambari server host name.
But this node has all services(NN, RM, and many more) installed. The second node has datanode and other services. I think i have to clean up all service directories right??