Ansible
Playbooks and roles that bootstrap the non-Talos hosts in the homelab — the Proxmox hosts, the NetBird connector LXCs, and the operator workstation; any Linux box that doesn't ship with an immutable OS.
Playbooks and roles that bootstrap the non-Talos hosts in the homelab — the Proxmox hosts, the NetBird connector LXCs, and the operator workstation; any Linux box that doesn't ship with an immutable OS.
Ansible role that installs the upstream Docker engine on a Fedora host — the docker-ce repository and package, the service, and docker-group membership for the Ansible user.
Standalone Docker host runtimes used outside Kubernetes — the Compose stacks on the Synology DS723+ (Maresa) and ad-hoc k3d clusters for local development.
Single source of truth for the edge Kubernetes cluster on Hetzner Cloud — single-node Talos cluster fronted by a Hetzner VPC and floating IP, used as the public ingress point.
GitOps reconciler running in both clusters — pulls manifests from Gitea (with Codeberg as the documented migration target) and applies them via Kustomize + SOPS decryption.
The IaC + cluster-bootstrap layer that turns hardware and the network fabric into a running Kubernetes cluster — Proxmox, Hetzner, Talos, Flux, OpenTofu, Ansible.
Single-node Hetzner Cloud edge cluster — one CX33 control-plane in nbg1, fronted by a Hetzner-managed VPC and label-selected firewalls, with a floating IP for stable inbound traffic.
Legacy lightweight Kubernetes distribution that ran on the AMD Athlon mini-server cluster. Retired and replaced by the Talos cluster on the Intel NUCs. Kept here as historical reference.
OpenTofu workflow that provisions everything outside Kubernetes — Hetzner edge VMs, Proxmox production VMs, UniFi home network, and the NetBird overlay.
Four-node Proxmox VE cluster (proxmox1–4) hosting the production Talos VMs, NetBird connector LXCs, and the storage-backed worker pool with GPU passthrough.
Immutable, API-driven Linux distribution purpose-built for Kubernetes — running on every node of the production and edge clusters, configured declaratively via topf.
Single source of truth for the on-prem production Kubernetes cluster — hardware, hypervisor, OS, controllers, apps, network, storage, backups, lifecycle.