Re: Labelling production
I was doing Unix/storage consulting at a place that was implementing SAP. They had just gone live, and were still busily adding functionality to production while they brought different business units into the fold. So the test environment was very active, as one might expect.
The way they had laid it out was that one server in the test environment was the configuration master, and a previous sysadmin had set it up so that it was trusted by everything else in the environment for passwordless root login (leave aside the security concerns, this was 25 years ago when that sort of thing was normal) The configuration master was very active, since any changes to the environment had to be made on the CM server first, copied to a test instance for testing, then to QA, and finally when properly blessed to production.
Luckily not everything had the same name, so the server's hostname showed up in the prompt (which makes me wonder if every "server" in Emily's story had the same root prompt) You'd think that would be enough of a hint, but I guess not as not once but twice one of the client's sysadmins meant to reboot another system but rebooted the CM server because it was in every login chain.
I made a simple alias in the root shell startup across the whole environment, so you had to type the hostname as part of the command to reboot it. So if you were on server foo and issued the reboot command it would say "use reboot-foo to reboot this server" and if you typed reboot-foo it would execute the reboot command.
I don't think I'd ever seen anyone color code environments back then, but since their problem wasn't rebooting production systems they'd have needed to color code a server in the test environment. Which was at that time almost as important as production!