How recover cluster when voter notes failure

[fean5959a]

Hello
I'm currently evaluating disaster recovery. I have a cluster of 6 nodes spread across 2 GCP regions. All nodes in leader group are voter, and the second group has non-voter nodes.

The promote/demote mechanism works correctly for a switch, promote second group, demote first group and restart leader.

However, if I completely shutdown the leader's cluster (the disaster), I can't promote the other nodes (the second group with non-voter nodes). I had to force one node to become the leader (with peers.json), clear the data on the other nodes, and restart.
Is this normal?
Have I missed something?

[cipherboy]

I think you can run the peers.json recovery on all nodes and avoid having to wipe data is my understanding, but based on the conversation in #38, I do think this is the current approach, yes.

[fean5959a]

Thank you very much for your reply.