After a node upgrade from the provider the node isn't booting

[dlukt]

I am undergoing node size upgrades.
I have started with the 1st worker.
I have drained and cordoned the node and done talosctl shutdown.
Disk layout is
vda rook ceph
vdb talos system
vdc swap

I have then used the provider's (Netcup) UI to upgrade the node with more disk and RAM and vCPU.
After the upgrade, the node isn't booting anymore.

And now I'd like to boot from ISO but I don't know the dowload link for

factory.talos.dev/metal-installer/b117087a5a826d2a186b4feb7326569a03b10cdb8cd3ba449118bf7208539af7:v1.11.2

https://factory.talos.dev/image/b117087a5a826d2a186b4feb7326569a03b10cdb8cd3ba449118bf7208539af7/v1.11.2/metal-amd64.iso

Ok I downloaded the iso and booted from DVD.

It says Talos is already installed to disk but booted from another media and talos.halt_if_installed kernel parameter is set. Please reboot from the disk.

Well I'm trying but idk, the provider has no way to set the boot disk.
What can I do?

[smira]

I don't know what exactly happens for you, but you need to be able to boot from disk.

For debugging, you need to grab console logs for the boot from disk.

If that is not possible/doesn't help, you can boot from an ISO removing talos.halt_if_installed command line argument, this way Talos will boot, find existing disk installation, and continue troubleshooting.

[smira]

Sorry, a typo. in the bootloader UI - when you boot from ISO/disk, it presents a bootloader UI where you can edit a command line.

[dlukt]

No it boots directly, there is no bootloader UI.

[smira]

It might be that your provider doesn't allow to access it, but the UI is there.

[dlukt]

Look, I mount the ISO.
I boot from DVD.
I see the VNC.
There is no UI.
It boots directly into Talos ISO.
"Booting Talos ISO"

[dlukt]

NOW I sent ctrl+alt+del and there is the GRUB UI, but it's not there initially

[dlukt]

Now I have booted from ISO by setting the talos.halt_if_installed=0,
But this will not last forever, the ISO will get deleted after 48h.
I have grabbed the machineconfig now.
How do I do maintenance?
Do I simply apply-config again?
Can I manually fix it through a rescue system? Install grub into vda?
What exactly is install doing?

talos/internal/app/machined/pkg/runtime/v1alpha1/bootloader/dual/dual.go

Line 41 in 77d8cc7

func (c *Config) Install(opts options.InstallOptions) (*options.InstallResult, error) {

install grub
install boot
install extra
does each node have grub?

I don't want to apply-config without your confirmation, it's what ChatGPT suggested, but those machines are careless.

talosctl -n w3 apply-config --mode=interactive -f w3.yaml

doesn't allow the node type to be set to worker, only control plane is selectable

talosctl -n w3 get securitystate -o yaml
node: w3
metadata:
    namespace: runtime
    type: SecurityStates.talos.dev
    id: securitystate
    version: 1
    owner: runtime.SecurityStateController
    phase: running
    created: 2025-10-16T12:41:44Z
    updated: 2025-10-16T12:41:44Z
spec:
    secureBoot: false
    selinuxState: enabled, permissive

[NPastorale]

You dropped a horribly formatted issue at 10:50 UTC. A Sidero engineer replied to your post at 11:25 UTC. Thirtyfive minutes. That is insane speed.

You posted a useless report with no logs, zero details, and no clear reproduction steps.

Before you demand priority attention, you should have a support plan and, most importantly, a testing environment before breaking what sounds to me like a prod setup.

Try and keep it reasonable