You're viewing a single thread.
https://www.abc.net.au/news/2024-07-19/what-is-crowdstrike-outage-explained/104120260
This has happened and taken a bunch of services down around the world.
What a garbage.
Just use Linux, SELinux, strong sandboxing, repositories, nonexecutable home directories, strong access control, offline backups.
How about a testing environment separate from production
and phased rollouts …
Best I can do is push it worldwide on a Friday morning
And my axe
I watched a ocean of computers go dead on the floor because I couldn't convince the sysadmin to do exactly that when pushing a major change.
Any more details?
This sounds like the setup to a fun story.
Does that cost money?
Yes. And time.
We make a lot more money by testing in production, and let the users tell us what's wrong. It's much faster.
We've successfully replaced the entire support team with an HTML form creating tickets for the one developer.
Surefire way to receive that efficiency performance bonus.
But how do I integrate everything into Microsoft 365 with that snazzy OneDrive feature? /s
You will escort us to sector zero zero one.
Crowd strike did this to Linux in April.
https://access.redhat.com/solutions/7068083
interesting it uses eBPF.
Damn
Pretty sure it's happened in Linux before, but because it's much less users, obviously it won't have same global outage like what happens now
I mean, I run Fedora and ran many others and had multiple crashes.
Fedora Atomic Desktops not anymore, but still not perfect.
And log monitoring with off machine collections