| inventories | ||
| playbooks | ||
| .ansible-lint | ||
| .gitignore | ||
| ansible-docker.sh | ||
| ansible.cfg | ||
| Dockerfile | ||
| README.md | ||
| requirements.txt | ||
| requirements.yml | ||
ansible
Ansible repository to manage the homelab (../homelab-session-summary-*.md).
Everything is reached over WireGuard, so bring the tunnel up first.
Run it from Docker
Ansible runs in a container so it never touches the laptop's own Ansible, Python or collections (ViaRézo, kubic…).
source ansible-docker.sh
ansible-docker build # once, and after changing requirements*
ansible-docker run # shell in the repo, then ansible-playbook …
ansible-docker run ansible-playbook playbooks/utils.yml -CD
The container uses the host network (for wg0) and your SSH agent, reads
~/.ssh/known_hosts without writing to it, and mounts this repo, so edits are
live and kube/config lands here.
Without Docker: uv venv env && uv pip install -p env/bin/python -r requirements.txt,
then env/bin/ansible-galaxy collection install -r requirements.yml -p .ansible/collections.
Inventory (inventories/hosts.yml)
| Group | What | Connects as |
|---|---|---|
hypervisors |
Proxmox host 10.0.0.1 |
root |
lxc |
100–106, 108–110, 250 | root (110 through pct exec on the host) |
vm |
107 pelican-panel + the k3s VMs |
root, k3s nodes as un-oeil + sudo |
k3s |
k3s_masters 131–133, k3s_workers 141–143 |
un-oeil + sudo |
external |
viarezo, the ViaRézo VM 10.10.10.3 |
root |
host_type (lxc, vm, proxmox, external) is what the playbooks test to
treat containers and VMs differently.
Playbooks
Always dry-run first: -CD (--check --diff).
| Playbook | Does |
|---|---|
utils.yml |
Settings in group_vars/all/utils.yml. apt sources → prism.viarezo.fr; CLI tools (ViaRézo's list + a few); en_US/fr_FR locales; Europe/Paris; journald capped (200M, host 1G); sshd keys-only (sshd -t before reload); hostname = inventory name (LXC through pct set, never k3s/host); motd. VMs only: guest agent, NTP, fstrim. Host: pve-no-subscription repo, udp/443 → WireGuard saved in rules.v4. Docker hosts: weekly dangling-image prune. Tags let you run parts (--tags ssh). --tags upgrade -l <hosts>: full-upgrade one host at a time (never the Proxmox host); -e utils_reboot=true reboots VMs that need it, never LXCs. |
add_user.yml |
Accounts from inventories/group_vars/all/users.yml: create, keys (exact list), passwordless sudo, remove. Skips system accounts of the same name (Forgejo's un-oeil). |
deploy_kube.yml |
Longhorn prerequisites on workers, k3s install/upgrade/config one node at a time (init master → masters → workers), writes kube/config. Bump k3s_version in group_vars/k3s.yml to upgrade. |
Secrets
vault/password (gitignored) holds the vault password. Nothing is vaulted yet;
the k3s join token is read from the first master. kube/config is gitignored.