I could be wrong on this since I've never actually attempted what you are asking. Based on my understanding of how replica assignment is done, I don't think that just changing the rack on an existing node is a good idea.
Changing racks for a node that already contains data would result in that data not being readable, since the replica assignment and ownership would not coincide with the data that physically exists on the node. Repair could help by streaming data from other replicas. You would also need to do a clean up to remove data that the node is no longer a replica for. This would essentially be the same as boot strapping a new node after a catastrophic failure. If you need to move a node's rack you should remove it from the cluster properly and add it again as a new /clean node with the new rack assignment. That way you are not in a degraded state while the data replicates Clint On Mar 23, 2016 11:07 AM, "Anubhav Kale" <anubhav.k...@microsoft.com> wrote: > Hello, > > Suppose we change the racks on VMs on a running cluster. (We need to do > this while running on Azure, because sometimes when the VM gets moved its > rack changes). > > In this situation, new writes will be laid out based on new rack info on > appropriate replicas. What happens for existing data ? Is that data moved > around as well and does it happen if we run repair or on its own ? > > > Thanks ! >
