At the moment, the only "automation" of the process of restoring an etcd cluster from backups is

https://phabricator.wikimedia.org/P3855

which is a python script that generates the commands to execute and has no way to check the state of the etcd cluster after each step. We should definitely do better and automate this very dangerous and complicated process.

The new script should also take into account the differences between etcd 2 and etcd3.