Just had an interesting issue with Talos Linux. The network interface names changed after I created the initial configuration. During an OS upgrade, the floating API IP was not assigned to the new `etcd` leader, resulting in a broken cluster.
Spun up a quick rescue box so I could work from within the VPC to reapply the corrected `MachineConfig`.
Fortunately, the worker nodes remained unaffected and continued to operate normally.