The horror is not how long booting takes, but rather if it’ll work
I’ve been patching and bouncing hundreds of machines automatically. The first decade, I was concerned and then observant. It’s been so reliable that I just stopped being concerned for the second decade. The last 5 years have been very occasionally (1%) unreliable, thanks to Lennart’s cancer, but not enough that I need to give it more than a glance. EL10 is a bit of a shitshow, so maybe the slow trend since el7 is continuing.
Every day.
The horror of rebooting every day.
Linux doesn’t need reboots for regular stuff. Proper packages can update everything from sendmail to syslog and not need a bounce.
The only time you need a bounce is
That’s almost it.
Once a week and only when I have time to potentially fix something that breaks. Usually because of something I did. Nobody told me I’d have to deal with config file changes in etc. Pacnew in my case. Had to discover that on my own.
Using diff with meld makes it pretty simple. I missed a line in one file recently and SDDM stopped automatically starting on boot. Growing pains. Plus I threw myself in the deep end to learn faster with a rolling release but that’s just how I operate.
Part of me welcomes breakage because that’s how I got gud at bending Windows to my will as a kid.
Exceptions of course if there’s a new feature I want to try out ASAP.
When I think of it. Every few days on average, sometimes weeks though.
I’ve blindly updated a year+ old Arch install without introducing problems. Not saying they don’t ever happen, but it isn’t that common.