This message was deleted.

Scratch that, this does look like it was due to

additionalUserData.data

. This means that

additionalUserData.data

is not usable with 1-node control planes, because it's always rendered differently between the

RKE2ControlPlane

and the machine's

RKE2Config

. I'm guessing this happens with multi-node control planes too and it seems like it would manifest there as continuous rollouts due to the persistent drift.

There should be no problem running with a single control plane node. The Machine deletion seems to be a bug connected to additionalUserData usage possibly. May you open an issue describing your setup? Thank you.

Thank you I opened https://github.com/rancher/cluster-api-provider-rke2/issues/711