did you have machines in your cluster that are no longer there?

the is an internal crd created to reconcile the underlying nodes and is specific to rancher

(⎈|default:N/A)➜  ~ k get machines.cluster -A
NAMESPACE     NAME                  CLUSTER   NODENAME         PROVIDERID              PHASE          AGE   VERSION
fleet-local   custom-470732b293dc   local     machine02-trmc   <rke2://machine02-trmc>   Running        17d
fleet-local   custom-71bdd896549b   local     machine03-trmc   <rke2://machine03-trmc>   Running        14d
fleet-local   custom-a05964d96735   local     machine01-trmc   <rke2://machine01-trmc>   Running        7d
fleet-local   custom-caed21cf57b6   local                                              Provisioning   18d

there is a machine which rancher thinks its trying to provision

you could delete this?

since all the nodes in your cluster are already accounted for

I am not a kubernetes expert

kubectl delete machines.cluster custom-caed21cf57b6 -n fleet-local machine.cluster.x-k8s.io "custom-caed21cf57b6" deleted

Boom, upgrade running now, thanks!