The concept of activating the failover procedure on a system that was working perfectly may seem odd, but it is better to discover bugs and other problems in a controlled situation than during an emergency.
This is a shift in cloud native thinking: accepting that there are errors in production and forcing those errors to happen so you can fix them. It’s like avoiding that psychological trap we all get into: if we don’t acknowledge a problem it must not exist and fixes the problem. Computers don’t act that way – you can’t bury a problem – and humans don’t either, really. So, just force the issue, but in small, controlled batches to limit the blast range. Better you control it than something else making the negative effect larger and worse, most likely.