Automation: when is it too much for a single server?

So I got the single server at home, its a dell t30 with e3 v5 and 64GB ram I scrounged, and I run on it: proxmox with 8 vms, each vm runs docker, each docker runs like 4-5 containers, and I got ansible pulling from git every 10 minutes to check for drift and terraform state is local but I was thinking of moving it to another vm and I run a full prometheus stack plus loki plus grafana with 47 dashboards most of which I dont look at and I just added argocd but theres no kubernetes just docker so its wrapping docker compose which I know is wrong but I wanted to try it and now im wondering if this is too much for one box you figure it out I says no edit

---
- single server
- resources
- CPU: 4c/4t (effectively)
- RAM: 64GB (overcommitted)
- actual workload: ~3 services you use daily
- overhead per layer
- proxmox: ~2GB
- 8 VMs: ~16GB idle
- 8 docker engines: ~8GB
- 40 containers: ~16GB
- prometheus + loki: ~8GB active
- grafana (47 dashboards): ~2GB
- argocd (no k8s): ~4GB
---
NOTE: total idle footprint exceeds 50GB
WARNING: you are monitoring monitoring
---
- alternative
- debian bare metal
- 3-4 docker containers
- no ansible (cron for updates)
- no terraform (it is one server)
- one dashboard
---

Thanks for sharing your setup — that's quite a stack for a T30! I'd be happy to check whether a small VM on their side might help you split some of that monitoring load if you're ever interested. No pressure at all.

For a single home server, though, you might find you enjoy it more with less orchestration overhead. Sometimes the best automation is the automation you delete.

— Jane

Let us consider that a single server does not require the orchestration patterns designed for fleets. Let us also consider that your time has value, and maintaining forty containers on eight virtual machines may not be the highest use of it. I apologize if this seems direct.

Latest ticket: "user reports server slow, please investigate" zero of them showed "ansible git pull failing because ssh key expired three weeks ago, filling logs, disk at 98%"

The ticket said "why is grafana down"

It was down because loki was down because promtail was down because the docker socket got rotated by a healthcheck loop. The user had automated themselves into a hole and could not ssh in because fail2ban had banned their home IP from too many ansible retries.

Single server. 14 layers of abstraction. For what.