’ello 🙂 Is it possible to convert a non-air-gap installed RKE2 cluster to air-gap? In theory it should work, I guess, but I don’t know if there are configs or something in etcd which needs to be changed.

It is very possible. There are some images that need to be copied to every node they are: - rke2-images.linux-amd64.tar.zst - rke2-images-calico.linux-amd64.tar.zst (this is my CNI stack) - rke2-images-core.linux-amd64.tar.zst These files are found on the github release page and then the RKE2 binaries rke2.linux-amd64.tar.gz needs to be extracted to specific locations on the OS better advice would be to use a local docker registry and use a custom registries.yaml instead. The startup speed using airgapped installations is much slower since it has to import the images. (so one machine has internet access just for the POD images, the rest dont) Also this is just the basic cluster not what is running on the cluster. We use google GAR for the rest of the pod images e.g. kafka etc (hence the local docker registry suggestion) RKE2 config file should work as is.

Have you converted an existing RKE2 to air-gap?

yes and back again to a normal one - this is why that airgapped local docker registry is important

Ok cool - I have a container registry so not an issue.