The goal is to make Server Manager reachable again without knowing the right commands. A possible solution:
Say in /etc/motd that network-recovery is the command for you if you’re in trouble with the network connection
Implement the network-recovery command that asks for a temporary IP, netmask, gateway, dns
It clears up firewall and IP configuration on all ethernet cards
It creates a bridge brdef with all ethernet cards
It applies the temporary IP configuration.
By default it runs signal-event interface-update && signal-event firewall-adjust on exit. We could allow to quit the tool by keeping the temporary configuration
It can be abused by applying the temporary configuration even if not needed: a working network configuration could be wiped out.
We could ensure the tool is executed from a physical console (not SSH) to avoid lockouts.
Another problem could be if the “temporary config” is never fixed. At next reboot the issue is there again: server is unreachable.
What do you think to always run signal-event interface-update when the tool quits? This ensures the running configuration is consistent with Server Manager.
Another solution can be: the tool cannot quit until Server Manager applies a working configuration.
Something like that would have saved me a lot of time today. Could be very useful indeed.
If you log in on a 
the procedure does not set DNS and IP routing. It is designed for direct-link access only.