Control Game (potential spoilers)
I wonder if Control 2 will talk about how the FBC had to replace more than 1000 employees and how no one noticed that so many people died or went missing.
Also: Given that the Oldest House reveals itself to those who know to seek for it, how many partners of employees who got told, despite all classification, will show up there?
All supernatural can be told away with "Magic", but these are real Office business problems. 👀
@shortwavesurfer2009 As I just read your post regarding accessible in Linux I saw your profile picture and I wondered: How do you, as a blind person, select a profile picture? What is important for you in a profile picture?
Since I haven't talked about it in a while, I might also want to mention, that I run a blogging room in matrix:
So if you are into blogging, maybe it's worth a shot. We also embed each other's RSS feeds to give feedback to articles :)
Pro tip: If you don't know something, just ask :)
I definitely learned something today!
Expectation: "Oh, this deployment is not properly configured yet, we have to solve this before going in production."
Reality: "Well, it's does what the requirement said, we have no time at hand and I would need to learn how to configure that properly, so let's just go in production and solve this later on." [Later on never happens]
This is way to common. Entire setups "YOLOed" because people have no idea how to do things properly. It highlights how important well automated (self-)services are.
Every time I see someone in a demo for "automatic mitigations for container security issues" suggest to remove an image from a registry or prevent it's pulling, I like to scream. It's 99.9% a stupid idea. It doesn't prevent any compromise but will most likely result in downtime for applications.
Instead of having one problem, you create two, great!
@sheogorath the 50% is explained in the article itself. It is a conservative assumption on cloud data centers being twice as effective utilized as local small DCs. Therefore comes the conclusion that a cloud server needs half the hardware as a dedicated local one. But still not sure if this assumption would hold up.
Mhm, every time I come around these carbon footprint calculators and I dig into their numbers, I find arbitrary stuff.
Today: cloud servers only produce 50% of their CO2 emissions during production "since 50% is manufactured for use in cloud".
I would love when someone can be a know-it-all and explain me how that works.
To expand of this a bit: Failure is part of production. So having stuff failing is part of your infrastructure work.
At the same time, pratice shows that a lot of "test setups" turn into production without ever being setup from scratch, so treating everything like production really helps.
This doesn't mean you shouldn't have any test setups, but unless you have a separate production environment, there is only production.
I don't really like the term homelab. I guess some people would call my infrastructure a homelab, but I think it misrepresents the traits it has. It's production, it's not really a lab environment and choices and tests have to be done carefully.
However I notice that this really helps as you work on stuff, as it requires your to be very considered about your choices and don't cut corners as you would maybe do with something you call a lab.
Math at Lenovo is special…
They sell you a 2129€ notebook for "just" 1532.88€ and then tell you that the overall sum of all articles your purchase is 1703,20€ . Misleading here is that the "My savings" and the "eCoupon-Code" are above the total sum, indicated they would be deducted.
The reality is: The sum of the eCoupon code, was added to the supposed price of all articles just to be removed again. And then, again included in the "My savings" sum, so make it completely arbitrary.
rant / calmed
After quite some annoyance, I wrote various parts of it down, which might becomes an own blog article, for now it's a README in my new os directory:
First tests make it look quite okay. Some further tweaks required, but it's on a good way.
After an evening of fighting with osbuild and osbuild-composer, which is already much easier than coreos-assembler, and some fights with Fedora modularity, and a whole new way to define RPM repositories and their keys, I managed to build an insecure ostree image. Progress!
Maybe at the end of the weekend I'll manage to setup a functioning install and move my infrastructure to it. Or it all ends up horrible and I'll throw stuff into the bin. Who knows… 🤷
I'm a professional relationship therapist for programs and their users.
This is my personal microblog. It's filled with my fun, joy and silliness.