I have a 10 node cluster and just experienced a disk failure which called for a Ubuntu restart. The server wouldn't come back online so IT reinstalled the OS. I'm now running through the install using the Install manager web interface. I'm moving all of the processes from the node with the faulty disk to a newer machine.
The problem is that I had a TON of data on DFS. The Install manager is asking me to select disks available to MapR on each node. I'm worried that by selecting these they will be totally wiped.
The manager is also complaining that Hadoop and Yarn are already installed on all nodes and I can't move forward without removing them. Also worried this will cause me to lose data.
I saw another question asking something similar and the accepted answer says "By no means, this should be taken as a general procedure for recovering from such scenarios, especially in production environments".
How can I ensure that this data will not be touched?