Why is RAP whitelist-db getting modified when we add new controller to the cluster?
When we add a new controller with non-empty RAP whitelist-db to an existing Master-Local setup, it will push its whitelist-db contents to the master which will push it to the whole cluster. This can cause issues if the new controller has entries for same RAPs which are already present on whitelist-db on the master.
Issue: The new controller will push its whitelist-db to the cluster. If the new controller already has an entry for a RAP present in master whitelist, but with a different ap-group, it will be over-written and the RAP will be assigned to a new AP-group.
1. RAP white-list is synchronized between all controllers in a cluster automatically. This cannot be disabled. We need not execute "write memory".
2. RAP white-list can be changed from any controller in the cluster. The change will be pushed to all other controllers automatically.
3. Before adding a new controller to the cluster, make sure that it doesn't have any conflicting RAP entries.
4. As a best practice, we can purge all the entries on the new controller before adding it to the cluster:
#whitelist-db rap purge