AnsweredAssumed Answered

Preparing for M3 CLDB move

Question asked by mandoskippy on Oct 23, 2012
Latest reply on Oct 31, 2012 by mandoskippy
I have read http://answers.mapr.com/questions/115/spof-ha-with-m3-what-are-recovery-steps-if-someone-spills-coke-on-the-cldb-node and things look pretty straight forward. 
I ran maprcli dump cldbnodes -zkconnect hadoopnode2:5181 -json
and found that cldb data does not reside on the node I want to be the CLDB. 

So since I am doing this controlled (i.e. not due to a failure) what are my optimum steps?

 1. Is there a way to ask mapr to put a CLDB copy on the node I want to be CLDB? It's already a tasktracker and fileserver node.
 2. Should I install the cldb service on the new node first?
 3. Should I take down the cluster? (I have the luxury of doing this) Add the service to the new node, remove the service from the old node, run the configure script on all nodes, and restart?  What order is optimum?
 4. I have the luxary of time. So like decommishining a node, I could wait for replication etc, what would the recommended steps be?

Outcomes