Best Laid Plans: Replacing nixos2's Failing Hardware — Part 19 of Building a Resilient Home Server Series

Where We Left Off Part 18 was a Friday-evening update to Vaultwarden's version, making sure I was ahead of a Bitwarden app update before it could cause me any issues — one service, one box, a couple hours. This one is not that. This is the post where I found out what "resilient" actually costs when you have to cash it in for real, on hardware that's dying, against a deadline I didn't pick. There isn't much config to show along the way this time — most of the actual fixes were two or three lines: add one path to a restic job, add a file to .gitignore, swap a hostname. The time didn't go into writing those lines. It went into staring at something that flat-out wouldn't load, or wasn't there, and trying to figure out why. If you've been following along, you know the pitch of this whole series: build it so that if a box dies, you restic restore, remap DNS, and you're back up fast. I've said some version of that more than once. Turns out that claim was half right — DNS and the basics…