Some points I forget:
- the backup services will be started by the hand, I won't automate that,
- the normal services, after a crash, will also be started by the hand, so I'll do a manual restore from the backup
The purpose is to have the less out of service delay in case of big server crash, and the less lost of datas. If this crash(*) occures, I must manage it by myself.
(*) crash could also be a big maintenance operation, or whatever will make the HostA unreachable