This message was deleted.

An option would be to use velero for running workloads that are using the anonymous volumes. If you build a new rke2 cluster, you install velero on the rke1, back up to an s3 bucket and then restore on the rke2 cluster. https://velero.io/docs/v1.8/how-velero-works/

Yep, this was an idea, but velero backups pvc trought kubeapi, in this case the kubelet doesnt start to make the backup