If the story is true that they needed to bring in a welder who used an angle grinder to get them in the server cage I am amused.
The big narrative seems to be that a pretty bad but not completely catastrophic outage was made catastrophic by Facebook's decision to run everything internally so essentially when Facebook went down Facebook couldn't get on Facebook to fix Facebook.
Running a process that internally for like your day to day stuff... still a bad idea in my opinion but you could at least argue there's some upsides to it.
But not having at least one engineer with a big keyring so they can physically access the building and servers and like a simple phone/e-mail tree for notifications and communications as part of your disaster recovery plan is like massively stupid.
This is "I lock to the keys in the safe to keep them safe" level of "Okay how does this go in your head?"