I agree.

Not exactly that we need to praise them or anything, but I think sarcastic tone in some replies ("if only they did something BASIC") aren't needed.

It's a different tradeoff. Systems that fails and *executes recovery path* from time to time tend to be more resilient to catastrophic failures.

Ideally you would want the combination of both: system doesn't usually crash/fail, but you force it to do so regularly. See https://en.wikipedia.org/wiki/Chaos_engineering

Uptime is kinda moot measure.

If it had uptime of 2 days but only went down for a minute at a time that would not be a problem