On NixOS we get one new grub menu entry per “deployment”, be it when you deploy to another host or the local one. They both then get a new grub entry. Of course you can still shoot yourself in the foot by making your system inaccessible remotely, by screwing up ssh config for example. There’s also a solution for that called deploy-rs which has a magic rollback feature which rolls back the system if it can’t connect back to the computer initiating the deployment https://github.com/serokell/deploy-rs?tab=readme-ov-file#mag...

I’m using these features since multiple years now and can vouch for them.

Not exactly related to your initial question but I also have my system build in CI and do some playwright tests which become more and more comprehensive as time passes. This all gives me quite a lot of confidence I’ll find an issue early or be able to revert back.