Spent 4 days setting up a cluster for ONE person, is this ok timewise, my boss says no..

> to slow at it How much automation are you using? There's a lot of ways to do stuff. I connected Ansible to a vmware cluster (later a proxmox cluster) and K8S. So I have automated setting up VMs and and tossing objects into K8S (kubernetes.core.k8s with plenty of variable substitutions) and I automated monitoring (well, used someone elses setup). I have not automated actually installing K8S on the cluster VMs (I guess I could?) and I have not automated proxmox-backup-server although I could. And I haven't gotten the NAS automated yet thats still tediously by hand and i hate it. For a variety of reasons, mostly backup and software defined networking, we're not running K8S on bare metal, which I guess is "weird" but whatever. The general point is the more stuff you put into roles in Ansible the more times you just set a variable in a playbook and run it. Probably the biggest savings for me was automating monitoring, monitoring platforms are hours of point and click if you do it by hand, and automating metallb and certmanager operations, ain't nobody got time to do TLS certs by hand. Automating IPAM by having ansible configure netbox and keep it all in sync and properly documented saved a lot of manual operation time. When I stopped working on that project for unrelated reasons I was trying to get proxmox SDN, physical switch hardware, and netbox IPAM all being automated by Ansible. It starts sounding like a database problem, you have an atomicity issue like the switch hardware updater barfs how do you catch that and do you delete it from proxmox SDN and netbox or just try your best or hope the guy running the Ansible playbook notices or ? You just know you're going to get a trouble ticket on activation if you don't catch all the errors so the next step after automation is you need a test load to run and verify before releasing it to production, which I/we were just making baby steps at. You need more than the hello-world docker container to prove out everything. Its nice to have a goal like all IT infra will be provisioned and configured by Ansible, but it'll never be reached in practice. There's always something like someone has to manually add new employees to pagerduty or whatever you use. I was not involved at automating the enduser stuff, that's someone elses problem. Maybe its yours, maybe its not. I had to set up their load balancer and networking and storageclass provider but their configmaps and pvcs are their problem not mine. Every company has a different demarcation point of course.