When you run REINDEX VERBOSE INDEX <index-name>;.
, is it possible to do this on one AZ at a time, or some other way of maintaining up time throughout the process? Would you need a multi-master cluster for this to be possible?
This would suggest so:
With a multi-master cluster, your application can redirect write operations from the failed DB instance to any other DB instance in the cluster.
I wasn't sure by reading this maintenance doc.