Consider migrating to k3os
Our Debian based server setup is becoming more and more complicated (and with that, a pain to maintain). We could decide that we want the prerequisite for openappstack to be that you already have a Kubernetes cluster.
For single node setups, k3os then becomes a logical candidate. On local setups, you could also make due with Minikube or k3d then, probably. Because some/many (research needed) Kubernetes providers (at least k3os) won't allow you to run Python (or maybe even anything) on the VPS behind Kubernetes, installing OAS should become possible by only applying Kubernetes manifests.
This means we should stop using Ansible, or move Ansible usage to somewhere within Kubernetes.
The great advantage of moving away from Ansible and towards k3os is that we won't have to maintain binaries running on the operating system. k3os basically allows you to upgrade the whole operating system through Kubernetes. Also, because of its very limited set of binaries on the OS level, the attack surface of the OS would probably be reduced in comparison with our current setup. Also we won't be maintaining what is starting to look more and more like an RKE-based Kubespray.
We would loose a lot of things by moving to k3os too: We currently install some tools to the server for easy debugging, like helm. Access to the VPS for "traditional debugging" might be very limited in this possible future. Configuring the VPSs resolv.conf and using it for debugging might also not be very easy