Workaround for installing Windows 11 without the online account requirement

At the first setup screen instead of answering any of the questions press Shift + F10

CMD will open

Type these commands line by line and hit enter after each line.
Replace Username with your desired username.
Beware case sensibility. defaultUser0 must be written exactly like this

This opens registry editor, navigate to

Delete “DefaultAccountAction”, “DefaultAccountSAMName”, and “DefaultAccountSID”

Right-click on
“LaunchUserOOBE” and rename it to “SkipMachineOOBE”
and make sure the value is set to “1”.

Close registry editor and type “shutdown /r /t 0”

Reklamirajte svoj transport servis besplatno

Na https://social.exyu.eu/transport imate mogućnost da listajte svoj transport servis, i nije potrebno da se ljudi, koji traže putovanja ili transport robe, registriraju.
Nego mogu se i pretražiti ponude.
Svaki račun može imati 1 servis, a svaki servis bezbroj ruta.
Imate i mogućnost dati povratne informacije,
Svaki servis može dodati do 3 slika i tekst opis u markdown formatu, i svaka ruta može isto imati opis.
Da biste listali Vaš servis potrebna je registracija, t.j. otvaranje računa, što je besplatno.
Trenutno radim na mobilnoj aplikaciji, ali će još proći nekog vremena dok bude gotova.
A ako tražite vozače, i za to imate mogućnost na toj platformi.

Does role-playing AI make sense? You’re an expert xyz

ChatGPT’s answer to that is:

That’s a really good question — deeper than it looks.

The short answer is: it’s partly an illusion, but a very useful one.

Here’s what’s actually going on under the hood:

I don’t “become” a master system administrator or a psychologist or a Yugoslav historian. I don’t switch into some separate personality or load a different model.

But when you tell me “you’re a master sysadmin,” you’re really giving me a context constraint — a framing instruction. It tells me:

> “Filter your knowledge and reasoning through the mindset, vocabulary, and priorities of someone who does that professionally.”

That instruction reshapes how I select and weigh information. It narrows the scope of possible completions — for example, emphasizing practical commands and risk mitigation instead of academic theory.

So it’s not that I suddenly become that expert. It’s that the language model adjusts its style, assumptions, and relevance weighting to simulate what an expert would focus on.

That’s why it feels like “role-playing works”: it anchors the response space around the domain you care about.

If you just said “tell me how to secure a Linux server,” I’d answer accurately but generically.
If you said “you’re a 20-year veteran sysadmin who just got called in at 3 a.m. to fix this,” I’d switch tone and prioritization — triage, logs, isolation, rollback — because that’s what a real one would do.

So:
🧠 Cognitively, it’s an illusion.
⚙️ Practically, it’s an interface mechanism.

It’s a way for you to steer the model’s attention — not to change what I am, but to change what slice of my training you’re drawing from.

Talos auf Netcup, Node Upgrade (VPS upgrade) mindfuck

Ich habe heute meinen Kubernetes Cluster upgegradet, weil er ist so speicherhungrig.
von VPS 2000 G11 auf VPS 3000 G11
2 vCPU 8GB RAM und 256GB SSH mehr

Netcups UI ist so schlecht.

Das System kam nicht mehr hoch, weil vda war die rook ceph Datendisk. Talos auf vdb.
Das Problem ist, nachdem Talos von der ISO installiert macht es kein vollständigen reboot, sondern ein kexec.
Ob es die Bootreihenfolge im BIOS ändert weis ich nicht, schätze nicht.
Jedenfalls kann man per VNC nicht ins traditionelle BIOS.
Man muss UEFI installieren.
Seit 1.11.0 werden beide bootloader nicht mehr parallel installiert.

Also Lösung:
UEFI an
ISO booten,
kernel Flag entfernen,
talosctl -n nodename upgrade
talosctl -n nodename shutdown

Prozedur


NS=rook-ceph
TOOLPOD=$(kubectl -n $NS get pod -l app=rook-ceph-tools -o jsonpath='{.items[0].metadata.name}')
CEPH="kubectl -n $NS exec -it $TOOLPOD -- ceph"

$CEPH -s                 # HEALTH_OK
kubectl get nodes -o wide

NODENAME=w1

$CEPH osd set noout
kubectl cordon $NODENAME
kubectl drain $NODENAME --ignore-daemonsets --delete-emptydir-data --grace-period=60 --timeout=15m

dann


talosctl -n nodename shutdown

Upgrade im CustomerControlpanel holen.
ISO mounten und UEFI an


talosctl -n $NODENAME upgrade
talosctl -n $NODENAME shutdown

Im VNC im UEFI das Bootlaufwerk wechseln


$CEPH osd unset noout
kubectl uncordon $NODENAME

Die Talos Leute waren übrigens sehr feindseelig und richtiggehend arrogant.
Ich kann keinem Talos empfehlen. Und ehrlich gesagt seit ich Kubernetes nutze ist der Stresslevel unerhört gestiegen.
Besser coolify oder uncloud nehmen als Kubernetes und ruhig schlafen können.
Spanier und Russe. Als wären sie irgendwelche Hoheiten, die man erbetteln muss und wo man sich glücklich schätzen muss dass einem geantwortet wird.