I work with Kubernetes tools. My job is L3 support and I troubleshoot Red Hat-based systems and Kubernetes. Those are my two areas and that is all I do. When a client's system breaks down, it is my job to fix it as much as possible.
In the last 12 months, I have been troubleshooting systems and training in Kubernetes.
I deploy applications atop it. I mostly use it as a server for various DevOps concerns. For example, I have a Kubernetes server running on Red Hat Enterprise Linux and Ansible server running on Red Hat Enterprise Linux. It is a DevOps pipeline that is fed by these separate servers.
I just duplicate installations of my clients' machines in order to troubleshoot. The idea is that I am presented with a problem, a broken system. If I can clone it, I do and then I try to fix it locally on my own machine before I present the solution back to the client. It varies slightly, depending on what the clients are using it for. In my very last case, about 2 or 3 weeks ago, there were etcd clusters running on an Ubuntu machine managing a Patroni installation. I tried to set that up on my own systems and started troubleshooting from there.