The machine gets powered off before Omni is able to collect it's status

[Unix4ever]

We should let Omni collect the MachineStatus before powering off the machine.
Otherwise it ends up in the state when it can't be allocated into a cluster, because it's Talos version is not known.

Maybe the controller can be changed to wait until the machine status is populated and then switch the desired state to powered off.

[smira]

I think in general there should be a generous time period before powering off to optimize for machines being allocated after being discovered.