add scc-air-gapped option
This MR introduces the "
A typical example, and the motivation for this MR, is to check that the containerd configuration that sylva-capi-cluster generates makes use of registry_mirrors as desired, by preventing any fallback of the Internet registries which would be reachable via an HTTP proxy.
Tested in sylva-projects/sylva-core!8135 (closed)
With this MR, the currently broken support that we have for registry_mirrors in ~kubeadm setups, now trigger a direct error instead of silently ignoring registry_mirrors:
Jun 8 13:01:12 mgmt-2584895130-kubeadm-capo-control-plane-724v8 containerd[1529]: time="2026-06-08T13:01:12.491518553Z" level=info msg="fetch failed" error="failed to do request:
Head \"https://registry.k8s.io/v2/kube-apiserver/manifests/v1.35.3\":smiley:
dial tcp 34.96.108.209:443: i/o timeout" host=registry.k8s.io
Jun 8 13:01:12 mgmt-2584895130-kubeadm-capo-control-plane-724v8 containerd[1529]: time="2026-06-08T13:01:12.508034530Z" level=info msg="stop pulling image registry.k8s.io/kube-apiserver:v1.35.3: active requests=0, bytes read=0"
Jun 8 13:01:12 mgmt-2584895130-kubeadm-capo-control-plane-724v8 containerd[1529]:
time="2026-06-08T13:01:12.508112737Z" level=error msg="PullImage \"registry.k8s.io/kube-apiserver:v1.35.3\" failed"
error="rpc error: code = DeadlineExceeded desc = failed to pull and unpack image \"registry.k8s.io/kube-apiserver:v1.35.3\":
failed to resolve image: failed to do request:
Head \"https://registry.k8s.io/v2/kube-apiserver/manifests/v1.35.3\":
dial tcp 34.96.108.209:443: i/o timeout"We see containerd failing to download the very first images that are needed.
The "