can you confirm this scenario is supported and throughout all the steps, service is available from the current master, except during the time it takes to perform the promotion operation.
1. I have a master cluster A and replica cluster B, with DR running.
2. Cluster A fails, and i promote B to be the master.
3. I can now restart A as a replica and run the DR the other direction.
4. Once A is resynced, I can promote A to master, bringing me back to my original configuration.