I think there is still an underlying problem in
unattending-upgrades messing up Ansible's
apt module, but by using a new k3s template I created based on the current master branch, at least we shouldn't run into #844 (closed) anymore.
After I solved the problem, I got regular failures with a cert-manager hook that did not exist (yet). I added
healthChecks to the
infrastructure Kustomization to make sure that the kustomization is only ready when the helm charts it installs are ready.
Closes #844 (closed)